Abstract
Gain-of-function mutations activating JAK/STAT signaling are seen in the majority of patients with myeloproliferative neoplasms (MPN), most commonly JAK2V617F. Although clinically approved JAK inhibitors improve symptoms and outcomes in MPNs, remissions are rare, and mutant allele burden does not substantively change with chronic therapy. We hypothesized this is due to limitations of current JAK inhibitors to potently and specifically abrogate mutant JAK2 signaling. We therefore developed a conditionally inducible mouse model allowing for sequential activation, and then inactivation, of Jak2V617F from its endogenous locus using a combined Dre-rox/Cre-lox dual-recombinase system. Jak2V617F deletion abrogates MPN features, induces depletion of mutant-specific hematopoietic stem/progenitor cells, and extends overall survival to an extent not observed with pharmacologic JAK inhibition, including when cooccurring with somatic Tet2 loss. Our data suggest JAK2V617F represents the best therapeutic target in MPNs and demonstrate the therapeutic relevance of a dual-recombinase system to assess mutant-specific oncogenic dependencies in vivo.
Current JAK inhibitors to treat myeloproliferative neoplasms are ineffective at eradicating mutant cells. We developed an endogenously expressed Jak2V617F dual-recombinase knock-in/knock-out model to investigate Jak2V617F oncogenic reversion in vivo. Jak2V617F deletion abrogates MPN features and depletes disease-sustaining MPN stem cells, suggesting improved Jak2V617F targeting offers the potential for greater therapeutic efficacy.
See related commentary by Celik and Challen, p. 701.
This article is featured in Selected Articles from This Issue, p. 695
INTRODUCTION
Somatic mutations that constitutively activate JAK2 signaling are seen in the majority of patients with myeloproliferative neoplasm (MPN) (1), most commonly the recurrent JAK2V617F alteration, and murine models suggest a critical role for JAK/STAT pathway mutations in promoting the MPN phenotype in vivo (2–6). In contrast to ABL1 kinase inhibition in BCR-ABL1–driven chronic myelogenous leukemia (7), current JAK inhibitors fail to reduce mutant clonal fraction and do not induce pathologic regression of key disease features including myeloproliferation and bone marrow fibrosis, and most patients lose their response over time (8, 9). To date, second-site JAK2 mutations have not been observed as a mechanism of acquired resistance (10), and different mechanisms have been postulated to mediate the inadequate efficacy of JAK inhibition, including incomplete dependency on JAK2 signaling and the presence of cooccurring mutant disease alleles (11). We hypothesized that the limited potency of JAK inhibition relates to insufficient mutant kinase inhibition at achievable therapeutic doses (4, 12), and we and others have elucidated mechanisms by which mutant JAK2 can signal in the presence of type I JAK inhibitors (12–14). Previous model systems evaluating doxycycline-inducible Jak2V617F expression highlight the importance of oncogenic JAK2V617F signaling in sustaining the MPN phenotype (6); however, these systems were limited by the inability to accurately recapitulate reversal of endogenous mutant expression or allow for assessment of oncogenic dependency on MPN hematopoietic stem cell (HSC) fitness alone or in context of comutations acquired during clonal evolution and myeloid transformation. Given this, we developed a system that would more definitively assess JAK2V617F dependency in MPN.
RESULTS
A Conditional Knock-in, Knock-Out Model of Jak2V617F MPN
To assess the requirement for JAK2V617F oncogenic signaling in MPN disease maintenance, we generated a Dre-rox (15), Cre-lox (16) dual-recombinase Jak2V617F knock-in/knock-out mouse model (Jak2Rox/Lox/Jak2RL) by gene targeting in mouse embryonic stem cells (Fig. 1A). The close proximity of the lox sites (82 base pairs) prevents Cre-mediated deletion prior to Dre-mediated recombination and Jak2V617F induction. Once the mutant allele is activated, the lox sites separate allowing for subsequent Cre-mediated deletion of Jak2V617F, including models in which cooperating alleles are induced by antecedent Cre-mediated activation/deletion. A similar strategy, which we have termed GOLDI-Lox for governing oncogenic loci by Dre inversion and lox deletion, was also used to target Flt3ITD (17). Given previous literature demonstrating that Jak2 expression is essential for hematopoiesis (18, 19), all Jak2RL mice used for experiments were heterozygous, with one maintained copy of the wild-type (WT) Jak2 allele (Supplementary Fig. S1A). In the absence of Dre recombination, Jak2RL/+ heterozygous mice displayed no observable phenotype, consistent with previous studies (not shown; refs. 2, 18–21). Sequencing of the Jak2RL locus on sorted Cre reporter cells (22) after Cre recombinase exposure confirmed retainment of the nonrecombined Jak2RL locus (Supplementary Fig. S1B). We transiently expressed Dre recombinase by mRNA electroporation ex vivo in primary lineage-negative bone marrow cells, efficiently inducing Jak2V617F activation and separation of lox sites by inversion (Supplementary Fig. S1C). Single-colony genotyping of these cells cultured in methylcellulose for 7 days revealed evidence of knock-in in 28%–55% of assayed colonies (n = 33/replicate). Efficient Jak2V617F-mutant induction was also observed in lineage-negative bone marrow harvested from primary transplant donors 6 weeks following electroporation and transplant (Supplementary Fig. S1D and S1E). By 3 weeks posttransplant, lethally irradiated mice transplanted with Dre-inducible Jak2RL knock-in bone marrow developed a highly penetrant and fully transplantable MPN characterized by leukocytosis with myeloid preponderance, elevated hematocrit with erythroid progenitor expansion in bone marrow, hepatosplenomegaly, and megakaryocytic hyperplasia consistent with prior Jak2V617F conditional knock-in mouse models of MPN (Supplementary Fig. S1F–S1J; ref. 3). Variable bone marrow fibrosis was observed across primary and secondary transplant recipient cohorts. Although there was minimal evidence of fibrosis in primary recipient mice, in secondarily transplanted mice, by >16 weeks, we observed 0–2+ reticulin fibrosis in 14 of 23 (61%) mice across multiple independent noncompetitive and competitive transplant studies (n = 5; Supplementary Fig. S1K).
Jak2V617F deletion abolishes JAK/STAT signaling and abrogates the MPN phenotype. A, Schematic representation of the dual-recombinase Jak2V617F conditional knock-in/knock-out allele (Jak2RL), the Jak2RL knock-in allele following Dre recombination, and the null recombined allele following Cre-mediated deletion. Semicircles indicate Rox sequences; triangles indicate loxP sequences. B, Representative Western blot depicting phospho-STAT5 abundance of Dre-mediated Jak2V617F knock-in (+Dre) vs. Jak2V617F-deleted (+Dre +Cre) states from isolated splenocytes 7 days following tamoxifen (TAM) administration in comparison with unrecombined (Unrec.) Jak2RL cells (n = 2 biological replicates each; representative of n = 2 independent experiments). C, Peripheral blood count trends (weeks 0–24) of MPN vs. tamoxifen (Jak2V617F-deleted) treated mice: WBCs (left), Hct (right; n ≥ 10 per arm; mean ± SEM). Gray bar represents duration of tamoxifen pulse/chow administration. Representative of n = 2 independent transplants. **, P ≤ 0.01; ****, P ≤ 0.0001. D, Kaplan–Meier survival analysis of MPN vs. tamoxifen (Jak2V617F-deleted) treated mice (n ≥ 12 per arm; log-rank test). Gray bar represents duration of tamoxifen pulse/chow administration. ****, P ≤ 0.0001. E, Spleen weights of MPN vs. tamoxifen (Jak2V617F-deleted) treated mice at timed sacrifice (24 weeks) in comparison with WT control mice (mean ± SEM). Representative of n = 2 independent transplants. ****, P ≤ 0.0001. F, Heat map scaled using Z-scores of serum cytokine/chemokine concentrations of MPN vs. tamoxifen (Jak2V617F-deleted) treated mice harvested at time of sacrifice 18–24 weeks posttransplant in comparison with WT control mice (n = 4–7 biological replicates per arm pooled from n = 3 transplants). Asterisks denote cytokines with FDR ≤ 0.05. Kruskal–Wallis test with FDR correction. G, Representative hematoxylin and eosin (H&E) and reticulin stains of bone marrow of MPN (Control) vs. tamoxifen (Jak2V617F-deleted) treated mice from timed sacrifice at 24 weeks. Representative micrographs of n = 6 individual mouse replicates per arm. All images represented at 400× magnification. Scale bar, 20 μm.
Jak2V617F deletion abolishes JAK/STAT signaling and abrogates the MPN phenotype. A, Schematic representation of the dual-recombinase Jak2V617F conditional knock-in/knock-out allele (Jak2RL), the Jak2RL knock-in allele following Dre recombination, and the null recombined allele following Cre-mediated deletion. Semicircles indicate Rox sequences; triangles indicate loxP sequences. B, Representative Western blot depicting phospho-STAT5 abundance of Dre-mediated Jak2V617F knock-in (+Dre) vs. Jak2V617F-deleted (+Dre +Cre) states from isolated splenocytes 7 days following tamoxifen (TAM) administration in comparison with unrecombined (Unrec.) Jak2RL cells (n = 2 biological replicates each; representative of n = 2 independent experiments). C, Peripheral blood count trends (weeks 0–24) of MPN vs. tamoxifen (Jak2V617F-deleted) treated mice: WBCs (left), Hct (right; n ≥ 10 per arm; mean ± SEM). Gray bar represents duration of tamoxifen pulse/chow administration. Representative of n = 2 independent transplants. **, P ≤ 0.01; ****, P ≤ 0.0001. D, Kaplan–Meier survival analysis of MPN vs. tamoxifen (Jak2V617F-deleted) treated mice (n ≥ 12 per arm; log-rank test). Gray bar represents duration of tamoxifen pulse/chow administration. ****, P ≤ 0.0001. E, Spleen weights of MPN vs. tamoxifen (Jak2V617F-deleted) treated mice at timed sacrifice (24 weeks) in comparison with WT control mice (mean ± SEM). Representative of n = 2 independent transplants. ****, P ≤ 0.0001. F, Heat map scaled using Z-scores of serum cytokine/chemokine concentrations of MPN vs. tamoxifen (Jak2V617F-deleted) treated mice harvested at time of sacrifice 18–24 weeks posttransplant in comparison with WT control mice (n = 4–7 biological replicates per arm pooled from n = 3 transplants). Asterisks denote cytokines with FDR ≤ 0.05. Kruskal–Wallis test with FDR correction. G, Representative hematoxylin and eosin (H&E) and reticulin stains of bone marrow of MPN (Control) vs. tamoxifen (Jak2V617F-deleted) treated mice from timed sacrifice at 24 weeks. Representative micrographs of n = 6 individual mouse replicates per arm. All images represented at 400× magnification. Scale bar, 20 μm.
To assess the reversibility of the Jak2RL construct, we cultured Dre-electroporated, lineage-negative, tamoxifen-inducible Ubc:CreER-Jak2RL cells isolated from donor mice with active MPN ex vivo with increasing doses of 4-hydroxy-tamoxifen (4-OHT) over bone marrow endothelial cells (BMEC; Supplementary Fig. S2A; ref. 23). Treatment with 4-OHT resulted in deletion of the Jak2V617F allele, which was confirmed by excision PCR (Supplementary Fig. S2B). Loss of Jak2V617F significantly reduced cell numbers ex vivo (mean 4-OHT 0.18 × 106/mL vs. vehicle 2.19 × 106/mL, P ≤ 0.0001), including within immunophenotypically defined hematopoietic stem/progenitor cell (HSPC) compartments, a phenotypic change not observed with vehicle-treated Jak2RL, Cre-inducible Jak2V617F (Jak2Crelox; P ≤ 0.228; ref. 2), or Cre-inducible WT cells (P ≤ 0.114; Supplementary Fig. S2C–S2G). Loss of Jak2V617F also abrogated erythropoietin-independent erythroid differentiation (24) in vitro (P ≤ 0.01; Supplementary Fig. S2H). The cell loss observed was associated with enhanced apoptosis, which was most apparent in Mac1+ mature myeloid cells (mean 4-OHT 35% vs. vehicle 9.3%, P ≤ 0.005; Supplementary Fig. S2I).
We next evaluated the impact of reversible Jak2V617F expression in vivo. Twelve weeks posttransplant, secondary recipient mice transplanted with Dre-electroporated Ubc:CreER-Jak2RL whole bone marrow and exhibiting MPN were administered tamoxifen to delete Jak2V617F (Supplementary Fig. S3A). A sequential rox-stop-rox, lox-TdTomato-stop-lox-eGFP dual-recombinase reporter system (25), in which TdTomato is expressed following Dre and then TdTomato deletion with concomitant GFP+ induction is expressed following Cre, was used to validate Jak2V617F deletion within Cd45.2 reporter–positive cell populations (Supplementary Fig. S3B). Deletion of Jak2V617F was also validated in vivo at the transcriptional level (P ≤ 0.0001; Supplementary Fig. S3C) and was associated with loss of constitutive JAK/STAT signaling (Fig. 1B). Consistent with our in vitro data, we observed normalization of white blood cell (WBC; mean tamoxifen 6.18 K/μL vs. MPN 17.5 K/μL, P ≤ 0.0001), hematocrit (Hct; mean 52.6% vs. 79.9%, P ≤ 0.01), and platelet (mean 786 K/μL vs. 2146 K/μL, P ≤ 0.0004) parameters within 4 weeks following tamoxifen treatment that persisted until timed sacrifice at 24 weeks (Fig. 1C; Supplementary Fig. S3D). As early as 7 days post-tamoxifen, we observed an increase in Annexin V+ cells (mean tamoxifen 34.1% vs. MPN 8.4%, P ≤ 0.01) in HSPC fractions consistent with an acute induction of apoptosis and concomitant reduction in the percentage of cycling HSPCs by flow (G2–M phase tamoxifen 9.4% vs. MPN 14.9%, P ≤ 0.01; Supplementary Fig. S3E and S3F). Two of 12 mice demonstrated reemergence/persistence of the MPN phenotype, both of which showed incomplete excision of the Jak2RL allele highlighting the necessity of Jak2V617F in disease maintenance (Supplementary Fig. S3G). WT Jak2 mRNA levels were increased at 7 days following oncogenic reversion, an effect that was sustained at the protein level until timed sacrifice at 24 weeks, as evidenced by Western blot analysis of harvested splenocytes, suggesting a potential compensatory mechanism in response to oncogenic reversion (Supplementary Fig. S3H and S3I). Genetic reversal of Jak2V617F significantly prolonged overall survival (median not defined vs. 187 days, P ≤ 0.0012) and led to loss of disease-defining MPN features in the majority of mice (9/12; Fig. 1D). Spleen weights (mean 108.9 mg vs. 542.7 mg, P ≤ 0.0001) were reduced, and we observed an overall trend in reduction of multiple inflammatory cytokines with Jak2V617F reversal (Fig. 1E and F). Significant cytokine reductions, although similar to what has previously been seen in patient samples receiving ruxolitinib therapy (26, 27), including IL6 (FDR ≤ 0.015) and MIP-1β (FDR ≤ 0.018), also showed reductions in serum Eotaxin (FDR ≤ 0.024) at time of sacrifice and a trend toward reduction with IP-10 (FDR ≤ 0.067; Supplementary Fig. S3J). Histopathologic analysis of bone marrow and spleen revealed reductions in megakaryocytic hyperplasia, splenic infiltration, reduced overall cellularity, and absence of bone marrow and spleen fibrosis in 8 of 9 assayed Jak2V617F-deleted mice that persisted until timed sacrifice at 24 weeks (Fig. 1G; Supplementary Fig. S3K). The phenotypes observed with Ubc:CreER-Jak2V617F deletion in vivo, including the histologic effects, were not observed with tamoxifen administration in the absence of Jak2V617F reversal (Supplementary Fig. S4A–S4G). We conclude that the MPN phenotype requires maintenance of oncogenic signaling through Jak2V617F.
Jak2V617F Reversal Impairs the Fitness of MPN Cells, Including MPN HSCs
We next evaluated Dre-electroporated Jak2RL bone marrow from Cd45.2 MPN donors in competition with Cd45.1 competitor cells to explore effects of Jak2V617F deletion on peripheral blood and bone marrow–mutant cell fitness (Supplementary Fig. S5A). Both early (3 weeks posttransplant) or late (12 weeks) administration of tamoxifen resulted in abrupt, durable reductions in Cd45.2-mutant cell fraction in the peripheral blood (mean 24.5% vs. 63.9%, P ≤ 0.001), coinciding with normalization of hematologic parameters that persisted until the time of sacrifice (Fig. 2A; Supplementary Fig. S5B). Consistent with the in vitro data, this effect was most pronounced in Mac1+ myeloid cell fractions (P ≤ 0.0001; Supplementary Fig. S5C). In bone marrow at timed sacrifice (24 weeks), the reductions in mutant cell fraction among the different HSPC compartments were more significant than those observed in peripheral blood, including within megakaryocytic-erythroid progenitor (MEP; Lineage−cKit+Sca1−Cd34−Fcg−; P ≤ 0.0001) and granulocytic-monocytic progenitor (GMP; Lineage−cKit+Sca1−Cd34+Fcg+; P ≤ 0.0001) populations and most importantly the LSK (Lineage−cKit+Sca1+; P ≤ 0.0096) stem cell compartment, including the SLAM-positive LSK population enriched for long-term HSCs (LT-HSC; Lineage−Sca1+cKit+Cd150+Cd48−; P ≤ 0.01; Fig. 2B; Supplementary Fig. S5D–S5F). Similar reductions in mutant cell fraction as well as reductions in Ter119+Cd71+ erythroid precursors were also observed in whole spleen (P ≤ 0.05) in both early- and late-tamoxifen cohorts consistent with an attenuation of extramedullary hematopoiesis (Supplementary Fig. S5G and S5H). Recurrent MPN, as was seen in the noncompetitive setting, was observed in 3 of 14 mice across both early- and late-treatment arms and corresponded with residual mutant Jak2V617F expression and sustained mutant chimerism at sacrifice. We next queried mice without MPN, but persistent Cd45.2+ cells from this transplant. In 8 of 8 mice assayed, we observed neither Jak2V617F knock-in nor Jak2V617F excision bands by PCR on sorted Cd45.2+ LSK cells (n = 3 early tamoxifen, n = 5 late tamoxifen), suggesting residual Cd45.2+ cells in these mice represent a non-Dre recombined Jak2RL WT bystander cell population. Similar results were observed in a separate competitive transplant study; however, in 1 of 8 late tamoxifen-treated mice, we also observed a faint knock-in band despite no phenotypic evidence of MPN, suggesting that in a minority of mice, residual mutant cells can remain and not necessarily give rise to disease (Supplementary Fig. S5I). Transplant of unfractionated Jak2RL-deleted bone marrow failed to form phenotypic disease in 4 of 5 secondary transplant recipient mice consistent with depletion of disease-propagating MPN HSCs (Supplementary Fig. S5J–S5L).
Jak2V617F reversal impairs the fitness of MPN cells, including MPN stem cells. A, Peripheral blood (PB) mutant Cd45.2 percent chimerism trend (weeks 0–24) of early (3 weeks posttransplant) tamoxifen (TAM; Jak2V617F-deleted) treated (gold bar) and late (12 weeks posttransplant) tamoxifen-treated (maroon bar) mice (n = 8 each) in comparison with MPN (dark gray bar; n = 6) mice (mean ± SEM). Gray bars represent duration of tamoxifen pulse/chow administration. Representative of n = 2 independent transplants. **, P ≤ 0.01; ***, P ≤ 0.001. B, Bone marrow–mutant cell fraction within LSK (Lineage−Sca1+cKit+), granulocytic-monocytic progenitor (GMP; Lineage−cKit+Sca1−Cd34+Fcg+), and megakaryocytic-erythroid progenitor (MEP; Lineage−cKit+Sca1−Cd34−Fcg−) compartments of early (3 weeks posttransplant) tamoxifen (Jak2V617F-deleted) treated and late (12 weeks posttransplant) tamoxifen-treated mice in comparison with MPN mice at timed sacrifice of 24 weeks (n = 6–8 individual biological replicates per arm; mean ± SEM). Representative of n = 2 independent transplants. **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001. C, Gene-set enrichment analysis (GSEA) of significant Hallmark gene sets of MPN vs. tamoxifen (Jak2V617F-deleted) treated LSKs isolated 7 days after initiation of tamoxifen (n = 3–4 biological replicates per arm). D, Volcano plot demonstrating differential gene expression of MPN vs. tamoxifen (Jak2V617F-deleted) treated LSKs 7 days following initiation of tamoxifen (n = 3–4 biological replicates per arm). E, GMP and MEP stem cell frequencies of MPN vs. tamoxifen (Jak2V617F-deleted) treated mice 7 days following initiation of tamoxifen (n = 8 biological replicates per arm across two independent transplants; mean ± SEM). F, Row normalized heat map of RNA-seq data of key erythroid differentiation factor genes from harvested MEPs at baseline (MPN), day 3 (D3), and day 7 (D7) following initiation of tamoxifen (Jak2V617F deletion). G, HOMER motif analysis from ATAC-seq data demonstrating decreased accessibility of Gata motif signatures with concomitant increased accessibility of Cebp motif signatures of tamoxifen-treated (Jak2V617F-deleted) cKit+ bone marrow cells isolated 7 days following initiation of treatment in comparison with MPN cells (n = 3 biological replicates per arm). Non-Sig., nonsignificant.
Jak2V617F reversal impairs the fitness of MPN cells, including MPN stem cells. A, Peripheral blood (PB) mutant Cd45.2 percent chimerism trend (weeks 0–24) of early (3 weeks posttransplant) tamoxifen (TAM; Jak2V617F-deleted) treated (gold bar) and late (12 weeks posttransplant) tamoxifen-treated (maroon bar) mice (n = 8 each) in comparison with MPN (dark gray bar; n = 6) mice (mean ± SEM). Gray bars represent duration of tamoxifen pulse/chow administration. Representative of n = 2 independent transplants. **, P ≤ 0.01; ***, P ≤ 0.001. B, Bone marrow–mutant cell fraction within LSK (Lineage−Sca1+cKit+), granulocytic-monocytic progenitor (GMP; Lineage−cKit+Sca1−Cd34+Fcg+), and megakaryocytic-erythroid progenitor (MEP; Lineage−cKit+Sca1−Cd34−Fcg−) compartments of early (3 weeks posttransplant) tamoxifen (Jak2V617F-deleted) treated and late (12 weeks posttransplant) tamoxifen-treated mice in comparison with MPN mice at timed sacrifice of 24 weeks (n = 6–8 individual biological replicates per arm; mean ± SEM). Representative of n = 2 independent transplants. **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001. C, Gene-set enrichment analysis (GSEA) of significant Hallmark gene sets of MPN vs. tamoxifen (Jak2V617F-deleted) treated LSKs isolated 7 days after initiation of tamoxifen (n = 3–4 biological replicates per arm). D, Volcano plot demonstrating differential gene expression of MPN vs. tamoxifen (Jak2V617F-deleted) treated LSKs 7 days following initiation of tamoxifen (n = 3–4 biological replicates per arm). E, GMP and MEP stem cell frequencies of MPN vs. tamoxifen (Jak2V617F-deleted) treated mice 7 days following initiation of tamoxifen (n = 8 biological replicates per arm across two independent transplants; mean ± SEM). F, Row normalized heat map of RNA-seq data of key erythroid differentiation factor genes from harvested MEPs at baseline (MPN), day 3 (D3), and day 7 (D7) following initiation of tamoxifen (Jak2V617F deletion). G, HOMER motif analysis from ATAC-seq data demonstrating decreased accessibility of Gata motif signatures with concomitant increased accessibility of Cebp motif signatures of tamoxifen-treated (Jak2V617F-deleted) cKit+ bone marrow cells isolated 7 days following initiation of treatment in comparison with MPN cells (n = 3 biological replicates per arm). Non-Sig., nonsignificant.
We sought to characterize transcriptional changes following acute Jak2V617F reversal. We performed RNA sequencing (RNA-seq) analysis of purified HSPCs 3 and 7 days following Jak2V617F deletion (n = 3–4) compared with MPN controls (n = 3–4). Transcriptional analysis of sorted, Jak2V617F-deleted LSK and MEP populations revealed near-complete loss of expression of STAT5 target genes as early as 3 days postdeletion [LSK: normalized enrichment score (NES) = −1.77, FDR ≤ 0.002; MEP: NES = −1.53, FDR ≤ 0.0065] indicating immediate disengagement from disease-defining pathway signaling (Supplementary Fig. S6A). By 7 days, we observed significant negative enrichment in IFNγ (NES = −1.61, FDR ≤ 0.0005), TGFβ (NES = −1.45, FDR ≤ 0.071), and TNFα via NFκB (NES = −1.54, FDR ≤ 0.0017) Hallmark proinflammatory response pathways as well as downregulation of MAPK (NES = −1.52, FDR ≤ 0.0052) and MTORC1 (NES = −1.46, FDR ≤ 0.0071) targets in LSKs, suggesting abrupt reduction in proinflammatory and proliferative signaling in the setting of Jak2V617F deletion (Fig. 2C; Supplementary Fig. S6B; Supplementary Table S1). A flux toward increased expression of myeloid genes sets compared to erythroid gene sets was also observed at 7 days post-tamoxifen initiation, characterized by increased S100a8, S100a9, Mpo, and Hdc expression in LSKs; increases in GMP (mean 14.5% vs. 7.8%, P ≤ 0.018) versus MEP (mean 13.8% vs. 32%, P ≤ 0.0025) frequencies within the HSPC compartment; and enrichment in bone marrow Mac1+ myeloid cells (mean 41.7% vs. 27.8%, P ≤ 0.0084; Fig. 2D and E; Supplementary Fig. S6C). In line with reduced erythroid output, we also observed a marked decrease in heme metabolism in MEPs (NES = −2.07, FDR ≤ 4.71 × 10−5) with associated reductions in critical erythroid/megakaryocytic transcription factors and signaling mediators, including Nfe2 (28), Plek2 (29), and EpoR (30), which coincided with concomitant reductions in total erythroid progenitor cell numbers (P ≤ 0.021) and significantly reduced burst-forming unit-erythroid (BFU-E) colony output of Jak2V617F-deleted cells (P ≤ 0.001; Fig. 2F; Supplementary Fig. S6D–S6F). Assay for Transposase Accessible Chromatin with high-throughput sequencing (ATAC-Seq) on Jak2V617F-deleted cKit+ cells demonstrated an increase in open chromatin with Cebp motifs (P ≤ 1 × 10−10) and reduced accessibility at Gata motifs (p ≤ 1 × 10−620), including at critical erythroid loci (e.g., EpoR; log2FC = 1.49, FDR ≤ 0.00135), further consistent with an erythroid-to-myeloid lineage switch (Fig. 2G; Supplementary Fig. S6G; Supplementary Table S2). Lineage deconvolution (31) further suggested priming of cKit+ cells toward a monocyte-to-granulocyte maturation switch in the setting of oncogenic reversion, consistent with our flow cytometric data showing changes in lineage output before and after mutational reversion (Supplementary Fig. S6H). Although reduced accessibility at putative Gata target sites was observed, we did not observe differential expression of either Gata1 (P ≤ 1.0) or Gata2 (P ≤ 0.82) in Jak2V617F-deleted LSKs or MEPs compared with controls. These data suggest the transcriptional networks regulating the MPN phenotype are not obligated to be achieved through transcription factor expression dysregulation but through differential transcription factor–mediated output.
Differential Efficacy of Jak2V617F Deletion Compared with JAK Inhibitor Therapy
Given the limited ability of current JAK inhibitors to achieve disease modification and/or clonal remissions in polycythemia vera and myelofibrosis, we next compared the phenotypic and transcriptional effects of JAK inhibitor therapy with ruxolitinib to the effects of Jak2V617F reversal. We first performed RNA-seq on Jak2V617F-mutant LSKs and MEPs following 7 days of ruxolitinib treatment (n = 3) and compared this to the effects of Jak2V617F deletion (n = 3). JAK-STAT target gene expression and erythroid pathway gene expression were much less potently inhibited with ruxolitinib than with Jak2V617F deletion. Specifically, Jak2V617F deletion resulted in a significant reduction in JAK/STAT signaling (NES = −1.51, P ≤ 0.003) and expression of negative regulators including Socs2 (32), Pim2 (33), and Cish (34). In contrast, ruxolitinib treatment was associated with a muted reduction in the same targets, with no significant changes in STAT5 target gene expression identified by GSEA (NES = −0.913, P = 0.84) at this time point (Fig. 3A; Supplementary Fig. S7A; Supplementary Table S3). Furthermore, the alterations in erythroid pathway gene expression in MEPs (NES = 1.45, P ≤ 0.012 vs. NES = −1.82, P ≤ 0.0005) and skewing of GMP and MEP frequencies observed with Jak2V617F deletion were not observed with ruxolitinib (mean GMP: vehicle 6.93% vs. ruxolitinib 6.66% vs. tamoxifen 20.1%, P = 0.91 vs. P ≤ 0.0001, MEP: vehicle 27.1% vs. ruxolitinib 35.3% vs. tamoxifen 14.2%, P = 0.25 vs. P = 0.014; Fig. 3B and C; Supplementary Fig. S7B and S7C). Expression of the gene sets associated with TGFβ (P = 0.65) and TNFα/NFκB (P = 0.90) inflammatory signaling pathways also displayed minimal changes with ruxolitinib and were more potently downregulated with Jak2V617F deletion. Consistent with this lack of change, genotype-aware single-cell ATAC-seq (scATAC-seq) on myelofibrosis patient samples (Supplementary Table S4) demonstrated unaltered NFκB accessibility in JAK2V617F-mutant HSPCs following JAK inhibitor treatment (Fig. 3D; Supplementary Fig. S7D; see Myers and colleagues (35) supporting the notion of insufficient mitigation of inflammatory signaling by JAK inhibition on MPN-sustaining stem cells.
Differential efficacy of Jak2V617F deletion compared with JAK inhibitor therapy. A, Scatter plot depicting −log10(Padj)*sign(log2 Fold Change) of ruxolitinib (RUX) treated vs. tamoxifen (TAM; Jak2V617F-deleted) treated LSKs (Lineage−Sca1+cKit+) in comparison with MPN control LSKs isolated after 7 days of treatment (n = 2–3 biological replicates per arm); differentially expressed genes as indicated by color (see Supplementary Tables S1 and S3). B, Gene-set enrichment analysis (GSEA) depicting a positive enrichment in heme metabolism in ruxolitinib-treated (n = 3) vs. negative enrichment in tamoxifen (Jak2V617F-deleted) treated (n = 3) LSKs isolated after 7 days of treatment. C, Box plot of the top leading edge genes in the Hallmark heme metabolism gene set of ruxolitinib-treated (blue) or tamoxifen (Jak2V617F-deleted) treated (red) megakaryocytic-erythroid progenitor (MEP; Lineage−cKit+Sca1−Cd34−Fcg−) cells as compared with untreated MPN cohorts. D, Box plots of scATAC-seq motif accessibility for either NFKB1 or REL transcription factors for untreated human JAK2 WT (n = 188 cells from 4 patients; gray), untreated JAK2V617F-mutant (n = 105 cells from 4 patients; gray), and ruxolitinib-treated JAK2V617F-mutant (n = 87 cells from 3 patients; blue) HSPCs (35). P values indicated are from linear mixture model explicitly modeling patient identity as random effect to account for patient-specific effects, followed by likelihood ratio test. ****, P ≤ 0.0001. E, Peripheral blood counts of vehicle (VEH), ruxolitinib (RUX), the type II JAK2 inhibitor CHZ868 (CHZ), or tamoxifen (Jak2V617F-deleted) treated mice at the conclusion of a 6-week in vivo trial: WBCs (left), Hct (right; n ≥ 4 each; mean ± SEM). **, P ≤ 0.01; ***, P ≤ 0.001; ****, p ≤ 0.0001. F, Peripheral blood (PB) mutant Cd45.2 percent chimerism trend (0–6 weeks) of vehicle, ruxolitinib, CHZ868, or tamoxifen (Jak2V617F-deleted) treated mice (n ≥ 4 each; mean ± SEM). *, P ≤ 0.05. G, Bone marrow–mutant cell fraction of LSK (Lineage−Sca1+cKit+), granulocytic-monocytic progenitor (GMP; Lineage−cKit+Sca1−Cd34+Fcg+), and megakaryocytic-erythroid progenitor (MEP; Lineage−cKit+Sca1−Cd34−Fcg−) compartments of vehicle, ruxolitinib, CHZ868, or tamoxifen (Jak2V617F-deleted) treated mice at the conclusion of the 6-week in vivo trial (n ≥ 4 each; mean ± SEM). *, P ≤ 0.05; ****, P ≤ 0.0001. H, GSEA depicting a negative enrichment in downregulation of KRAS signaling targets in ruxolitinib-treated (n = 3) vs. positive enrichment in tamoxifen (Jak2V617F-deleted) treated (n = 3) MEPs isolated after 7 days of respective treatment. I, IHC of phospho-ERK on sectioned bone marrow of vehicle, ruxolitinib, or tamoxifen (Jak2V617F-deleted) treated mice following 7 days of treatment (n = 3 individual biological replicates per arm). All images represented at 400× magnification. Scale bar, 20 μm. J, Quantitative PCR demonstrating relative Ybx1 expression levels from isolated cKit+ bone marrow of vehicle vs. ruxolitinib vs. tamoxifen (Jak2V617F-deleted) treated mice after 7 days of treatment (n = 2–4 individual biological replicates per arm; mean ± SEM). *, P ≤ 0.05; **, P ≤ 0.01. E–G, Representative of n = 3 independent experiments.
Differential efficacy of Jak2V617F deletion compared with JAK inhibitor therapy. A, Scatter plot depicting −log10(Padj)*sign(log2 Fold Change) of ruxolitinib (RUX) treated vs. tamoxifen (TAM; Jak2V617F-deleted) treated LSKs (Lineage−Sca1+cKit+) in comparison with MPN control LSKs isolated after 7 days of treatment (n = 2–3 biological replicates per arm); differentially expressed genes as indicated by color (see Supplementary Tables S1 and S3). B, Gene-set enrichment analysis (GSEA) depicting a positive enrichment in heme metabolism in ruxolitinib-treated (n = 3) vs. negative enrichment in tamoxifen (Jak2V617F-deleted) treated (n = 3) LSKs isolated after 7 days of treatment. C, Box plot of the top leading edge genes in the Hallmark heme metabolism gene set of ruxolitinib-treated (blue) or tamoxifen (Jak2V617F-deleted) treated (red) megakaryocytic-erythroid progenitor (MEP; Lineage−cKit+Sca1−Cd34−Fcg−) cells as compared with untreated MPN cohorts. D, Box plots of scATAC-seq motif accessibility for either NFKB1 or REL transcription factors for untreated human JAK2 WT (n = 188 cells from 4 patients; gray), untreated JAK2V617F-mutant (n = 105 cells from 4 patients; gray), and ruxolitinib-treated JAK2V617F-mutant (n = 87 cells from 3 patients; blue) HSPCs (35). P values indicated are from linear mixture model explicitly modeling patient identity as random effect to account for patient-specific effects, followed by likelihood ratio test. ****, P ≤ 0.0001. E, Peripheral blood counts of vehicle (VEH), ruxolitinib (RUX), the type II JAK2 inhibitor CHZ868 (CHZ), or tamoxifen (Jak2V617F-deleted) treated mice at the conclusion of a 6-week in vivo trial: WBCs (left), Hct (right; n ≥ 4 each; mean ± SEM). **, P ≤ 0.01; ***, P ≤ 0.001; ****, p ≤ 0.0001. F, Peripheral blood (PB) mutant Cd45.2 percent chimerism trend (0–6 weeks) of vehicle, ruxolitinib, CHZ868, or tamoxifen (Jak2V617F-deleted) treated mice (n ≥ 4 each; mean ± SEM). *, P ≤ 0.05. G, Bone marrow–mutant cell fraction of LSK (Lineage−Sca1+cKit+), granulocytic-monocytic progenitor (GMP; Lineage−cKit+Sca1−Cd34+Fcg+), and megakaryocytic-erythroid progenitor (MEP; Lineage−cKit+Sca1−Cd34−Fcg−) compartments of vehicle, ruxolitinib, CHZ868, or tamoxifen (Jak2V617F-deleted) treated mice at the conclusion of the 6-week in vivo trial (n ≥ 4 each; mean ± SEM). *, P ≤ 0.05; ****, P ≤ 0.0001. H, GSEA depicting a negative enrichment in downregulation of KRAS signaling targets in ruxolitinib-treated (n = 3) vs. positive enrichment in tamoxifen (Jak2V617F-deleted) treated (n = 3) MEPs isolated after 7 days of respective treatment. I, IHC of phospho-ERK on sectioned bone marrow of vehicle, ruxolitinib, or tamoxifen (Jak2V617F-deleted) treated mice following 7 days of treatment (n = 3 individual biological replicates per arm). All images represented at 400× magnification. Scale bar, 20 μm. J, Quantitative PCR demonstrating relative Ybx1 expression levels from isolated cKit+ bone marrow of vehicle vs. ruxolitinib vs. tamoxifen (Jak2V617F-deleted) treated mice after 7 days of treatment (n = 2–4 individual biological replicates per arm; mean ± SEM). *, P ≤ 0.05; **, P ≤ 0.01. E–G, Representative of n = 3 independent experiments.
To evaluate the phenotypic effects of Jak2V617F deletion in direct comparison to JAK kinase inhibition, we performed an in vivo trial lasting 6 weeks comparing ruxolitinib to Jak2V617F deletion (Supplementary Fig. S8A). We saw a greater improvement in hematologic parameters, spleen weights (mean vehicle 457 mg vs. ruxolitinib 235 mg vs. tamoxifen 125 mg, P ≤ 0.0027), restoration of histopathologic morphology in both bone marrow and spleen, and reduced Cd45.2-mutant chimerism in peripheral blood (mean vehicle 40.7% vs. ruxolitinib 37.7% vs. tamoxifen 17.3%, P ≤ 0.0059) of Jak2V617F-deleted mice versus ruxolitinib-treated mice (Fig. 3E and F; Supplementary Fig. S8B–S8D). Reductions in total erythroid progenitors were observed with both ruxolitinib- and tamoxifen-treated mice by the conclusion of the study (mean vehicle 0.55 × 106/mL vs. ruxolitinib 0.28 × 106/mL vs. tamoxifen 0.21 × 106/mL, P ≤ 0.001), with a greater effect on megakaryocytic progenitor (mean vehicle 0.55 × 106/mL vs. ruxolitinib 0.53 × 106/mL vs. tamoxifen 0.23 × 106/mL, P ≤ 0.05) and total megakaryocyte output with Jak2VF deletion specifically (Supplementary Fig. S8E–S8H). Most importantly, the reduction in mutant cell fraction seen with Jak2V617F deletion within hematopoietic progenitor (GMP: P ≤ 0.0001, MEP: P ≤ 0.0001) and LSK stem cell–enriched populations was not observed with pharmacologic type I JAK inhibition (mean vehicle 87.9% vs. ruxolitinib 87.6% vs. tamoxifen 28.7%, P ≤ 0.0001; Fig. 3G).
We previously showed that the type II JAK2 inhibitor CHZ868 showed improved efficacy compared with ruxolitinib in vivo (36). Consistent with these observations, treatment with CHZ868 showed greater efficacy than ruxolitinib with regard to improvement in hematologic parameters (mean Hct: CHZ868 50.3% vs. ruxolitinib 85.8%, P ≤ 0.0001) and spleen volume reduction (mean CHZ868 76 mg vs. ruxolitinib 235 mg, P < 0.0001), at par with Jak2V617F deletion (Fig. 3E and F; Supplementary Fig. S8B and S8C). Significant reductions in MEP, GMP, and LSK-mutant allele burden, as well as in more committed MEP populations, were also observed in CHZ868-treated mice compared with vehicle/ruxolitinib-treated mice (LSK: P ≤ 0.02, GMP: P ≤ 0.013, MEP: P ≤ 0.013), but not to the extent seen with Jak2V617F deletion (Fig. 3G; Supplementary Fig. S8F and S8G). These data confirm that more potent, selective target inhibition, including with type II JAK inhibitors, offers the potential for greater therapeutic efficacy when compared with current type I JAK inhibitors.
Previous studies have suggested that MAPK signaling plays an important role in MPN disease cell survival in the setting of type I JAK inhibitor therapy (13), and recent work has implicated the MAPK-dependent factor YBX1 as a critical mediator of JAK2V617F-mutant cell persistence (14). We observed distinct effects on MAPK activity by RNA-seq with ruxolitinib treatment versus Jak2V617F deletion in comparison with vehicle-treated mice. Negative regulators of KRAS signaling were downregulated with ruxolitinib (NES = −1.64, FDR ≤ 0.0005) and upregulated with Jak2V617F deletion (NES = 1.35, FDR ≤ 0.039) in MEPs, suggesting enhanced MAPK signaling with ruxolitinib and MAPK attenuation with Jak2V617F deletion (Fig. 3H). IHC of bone marrow sections confirmed increased phospho-ERK abundance in ruxolitinib-treated mice that was abrogated with Jak2V617F deletion (Fig. 3I), and genotype-specific scATAC-seq revealed increased accessibility of MAPK-mediated AP-1 factors FOS/JUN (37) within HSPCs of ruxolitinib-treated MF patients in comparison with untreated MF HSPCs consistent with enhanced MAPK activity (Supplementary Fig. S8I). Furthermore, expression of Ybx1 in sorted murine cKit+ cells was increased with ruxolitinib therapy but potently suppressed with Jak2V617F deletion (mean relative expression: vehicle 1.37 vs. ruxolitinib 2.52 vs. tamoxifen 0.42, P ≤ 0.0094; Fig. 3J). These data suggest that potent, mutant-specific Jak2V617F targeting can abrogate pathologic MAPK signaling and YBX1-mediated persistence of Jak2V617F-mutant HSPCs.
JAK2V617F Dependency with Cooperative TET2 Loss
Previous studies of mutational order in primary MPN cells have shown that cooperating mutations in epigenetic regulators, including TET2, can precede the acquisition of JAK2V617F in the clonal evolution of MPN and that antecedent TET2 mutations can alter the in vitro sensitivity to ruxolitinib (38). In addition, TET2 loss is the most frequently cooccurring mutation with JAK2V617F in MPNs, and in vitro and in vivo studies have shown that concurrent TET2 and JAK2V617F mutations promote enhanced mutant HSC fitness and increased risk of MPN disease progression (39–41). Our Jak2RL system allows for the assessment of JAK2V617F dependency in the setting of cooccurring mutant allele activation/inactivation, including in the context of antecedent mutations in epigenetic regulators. We therefore assessed the impact of Jak2V617F activation in concert with preexisting Tet2 loss with the reversible Jak2RL allele (Fig. 4A). Mice transplanted with Dre-electroporated Ubc:CreER-Jak2RL/Tet2−/− cells demonstrated enhanced leukocytosis (mean 13.1 K/μL vs. 26.0 K/μL, P ≤ 0.001) and thrombocytosis, increased spleen volumes (mean 317.6 mg vs. 612.2 mg, P ≤ 0.021), and expanded mutant peripheral blood chimerism (mean 25.9% vs. 39.9%, P ≤ 0.025) compared with Ubc:CreER-Jak2RL and single-mutant Tet2−/−-transplanted mice (Fig. 4B–D; Supplementary Fig. S9A). Tet2−/− and Jak2RL/Tet2−/− HSCs also exhibited improved serial replating capacity in colony-forming assays compared to single-mutant Jak2RL cells (Supplementary Fig. S9B). Ex vivo coculture of Tet2−/− and Jak2RL/Tet2−/− cells over BMECs exhibited a near three-fold increase in hematopoietic cell output (mean 2.92 × 106/mL vs. 2.41 × 106/mL vs. 0.80 × 106/mL, P ≤ 0.005), including among Mac1+ mature myeloid cells (mean 1.04 × 106/mL vs. 0.60 × 106/mL vs. 0.22 × 106/mL, P ≤ 0.023), compared with Jak2RL cells consistent with the known role of TET2 loss of function in enhancing myeloid lineage commitment (Supplementary Fig. S9C; ref. 42). Together, these data are phenotypically consistent with previous Tet2−/− and Jak2VF/Tet2−/− models (39, 40) and highlight the utility of the Dre-Cre dual-recombinase system to model sequential acquisition of mutations in vivo and mimic the evolution of disease from a premalignant, clonally restricted hematopoietic state (i.e., single-mutant Tet2−/− knock-out) to overt MPN.
Jak2V617F dependency with cooperative Tet2 loss. A, Schematic of the experimental setup for the double-mutant Jak2RL/Tet2f/f competitive transplants. Downward arrows represent initial pulse tamoxifen (TAM) administration to genetically inactivate Tet2. B, WBC counts of primary Jak2RL vs. Tet2−/− vs. Jak2RL/Tet2−/− transplanted mice at 16 weeks posttransplant (n = 5–6 each; mean ± SEM). Representative of n = 2 independent transplants. *, P ≤ 0.05; ***, P ≤ 0.001. C, Spleen weights of primary Jak2RL vs. Tet2−/− vs. Jak2RL/Tet2−/− transplanted mice at time of sacrifice (n = 5–6 each; mean ± SEM). Representative of n = 2 independent transplants. *, P ≤ 0.05; **, P ≤ 0.01. D, Peripheral blood Cd45.2-mutant percent chimerism of Jak2RL vs. Tet2−/− vs. Jak2RL/Tet2−/− secondary competitive transplant mice at 9 weeks posttransplant (n ≥ 10 per arm; mean ± SEM). Representative of n = 2 independent transplants. *, P ≤ 0.05. E, Peripheral blood count trends (weeks 0–21) of MPN vs. tamoxifen (Jak2V617F-deleted) treated Jak2RL vs. Jak2RL/Tet2−/− competitive transplant mice: WBCs (left), hematocrit (Hct; right; n = 3–4 per arm; mean ± SEM). Gray bars represent duration of tamoxifen pulse/chow administration. Representative of n = 2 independent transplants. **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001. F, Fold change from baseline (pretreatment) to posttreatment of Cd45.2-mutant peripheral blood chimerism of Jak2RL vs. Tet2−/− vs. Jak2RL/Tet2−/− transplanted mice treated for 6 weeks with either vehicle, ruxolitinib (RUX; 60 mg/kg twice daily), or tamoxifen (Jak2VF deletion; n = 4–5 per arm; mean ± SEM). *, P ≤ 0.05. G, Reticulin stains of bone marrow from MPN vs. tamoxifen (Jak2V617F-deleted) treated Jak2RL vs. Jak2RL/Tet2−/− mice at timed sacrifice (21 weeks). Representative micrographs of n = 3 individual mouse replicates per arm. All images represented at 400× magnification. Scale bar, 20 μm. H, Bone marrow–mutant Cd45.2 percent chimerism within the LSK (Lineage−Sca1+cKit+) compartment of MPN vs. tamoxifen (Jak2V617F-deleted) treated Jak2RL vs. Jak2RL/Tet2−/− mice at timed sacrifice (21 weeks; n ≥ 7 biological replicates per arm across two independent transplants; mean ± SEM). *, P ≤ 0.05; ***, P ≤ 0.001. I, Serial replating assay of plated MPN vs. tamoxifen (Jak2V617F-deleted) treated Jak2RL vs. Jak2RL/Tet2−/− bone marrow cells harvested at timed sacrifice 21 weeks and scored at day 8 after each plating (each sample plated in triplicate, representative of n = 2 independent experiments, mean ± SD). cGy, centigray; KI, knock-in; KO, knock-out; Lin-neg BM, lineage-negative bone marrow; trx, transplant.
Jak2V617F dependency with cooperative Tet2 loss. A, Schematic of the experimental setup for the double-mutant Jak2RL/Tet2f/f competitive transplants. Downward arrows represent initial pulse tamoxifen (TAM) administration to genetically inactivate Tet2. B, WBC counts of primary Jak2RL vs. Tet2−/− vs. Jak2RL/Tet2−/− transplanted mice at 16 weeks posttransplant (n = 5–6 each; mean ± SEM). Representative of n = 2 independent transplants. *, P ≤ 0.05; ***, P ≤ 0.001. C, Spleen weights of primary Jak2RL vs. Tet2−/− vs. Jak2RL/Tet2−/− transplanted mice at time of sacrifice (n = 5–6 each; mean ± SEM). Representative of n = 2 independent transplants. *, P ≤ 0.05; **, P ≤ 0.01. D, Peripheral blood Cd45.2-mutant percent chimerism of Jak2RL vs. Tet2−/− vs. Jak2RL/Tet2−/− secondary competitive transplant mice at 9 weeks posttransplant (n ≥ 10 per arm; mean ± SEM). Representative of n = 2 independent transplants. *, P ≤ 0.05. E, Peripheral blood count trends (weeks 0–21) of MPN vs. tamoxifen (Jak2V617F-deleted) treated Jak2RL vs. Jak2RL/Tet2−/− competitive transplant mice: WBCs (left), hematocrit (Hct; right; n = 3–4 per arm; mean ± SEM). Gray bars represent duration of tamoxifen pulse/chow administration. Representative of n = 2 independent transplants. **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001. F, Fold change from baseline (pretreatment) to posttreatment of Cd45.2-mutant peripheral blood chimerism of Jak2RL vs. Tet2−/− vs. Jak2RL/Tet2−/− transplanted mice treated for 6 weeks with either vehicle, ruxolitinib (RUX; 60 mg/kg twice daily), or tamoxifen (Jak2VF deletion; n = 4–5 per arm; mean ± SEM). *, P ≤ 0.05. G, Reticulin stains of bone marrow from MPN vs. tamoxifen (Jak2V617F-deleted) treated Jak2RL vs. Jak2RL/Tet2−/− mice at timed sacrifice (21 weeks). Representative micrographs of n = 3 individual mouse replicates per arm. All images represented at 400× magnification. Scale bar, 20 μm. H, Bone marrow–mutant Cd45.2 percent chimerism within the LSK (Lineage−Sca1+cKit+) compartment of MPN vs. tamoxifen (Jak2V617F-deleted) treated Jak2RL vs. Jak2RL/Tet2−/− mice at timed sacrifice (21 weeks; n ≥ 7 biological replicates per arm across two independent transplants; mean ± SEM). *, P ≤ 0.05; ***, P ≤ 0.001. I, Serial replating assay of plated MPN vs. tamoxifen (Jak2V617F-deleted) treated Jak2RL vs. Jak2RL/Tet2−/− bone marrow cells harvested at timed sacrifice 21 weeks and scored at day 8 after each plating (each sample plated in triplicate, representative of n = 2 independent experiments, mean ± SD). cGy, centigray; KI, knock-in; KO, knock-out; Lin-neg BM, lineage-negative bone marrow; trx, transplant.
We next evaluated effects of Jak2V617F deletion on Jak2RL/Tet2−/−–mutant cell fitness in vivo in competition with Cd45.1 bone marrow. Treatment with tamoxifen at 9 weeks posttransplant resulted in normalization of hematologic parameters (P ≤ 0.005) and reductions in peripheral blood mutant cell fraction of double-mutant cells to a similar extent observed with Jak2V617F deletion in single-mutant Jak2RL-transplanted mice (Fig. 4E and F). Furthermore, spleen sizes (mean 103 mg vs. 529 mg, P ≤ 0.0001) and total BM cellularity (femur; mean 11.6 × 106/mL vs. 15.7 × 106/mL, P ≤ 0.0035) were similarly normalized with Jak2V617F deletion (Supplementary Fig. S9D and S9E). Although the extent of reticulin fibrosis was increased in Jak2RL/Tet2−/− mice compared with Jak2RL, mutant allele reversal resolved fibrosis in both mutational contexts (Fig. 4G). The reduction in mutant cell fraction, as was observed with single-mutant mice, persisted down to the level of HSPCs in tamoxifen-treated Jak2RL/Tet2−/− mice, including within the LSK stem cell–enriched compartment (mean tamoxifen 28.7% vs. MPN 73.7%, P ≤ 0.001; Fig. 4H; Supplementary Fig. S9F). This decrease in mutant cell fraction appeared, at least in part, to be due to increased apoptosis, as ex vivo treatment with 4-OHT resulted in an increase in Annexin V+ cells in Jak2RL and double-mutant cells but not Tet2−/− cells (Supplementary Fig. S9G). This effect was specific to Jak2V617F deletion, as treatment of Tet2−/− and Jak2RL/Tet2−/− mice with type I JAK inhibition (ruxolitinib) did not alter allelic fraction (Supplementary Fig. S9H). Finally, in a subset of assayed Jak2RL/Tet2−/− mice following Jak2V617F deletion (4/9), we were unable to detect Tet2−/− knock-out bands in whole marrow at time of sacrifice. Cells harvested from Jak2RL/Tet2−/− recipient mice following oncogenic deletion were unable to serially replate, indicating loss of self-renewal capacity in comparison with control double-mutant mice (Fig. 4I; Supplementary Fig. S9I). These data support the notion that cooccurring loss-of-function mutations of TET2 do not dramatically alter reliance on JAK/STAT signaling for disease maintenance and that, despite the fitness advantage engendered by TET2 loss on MPN HSCs, the reductions in HSC fitness in the setting of Jak2V617F reversion suggest a unique dependency on oncogenic JAK2V617F that renders double-mutant cells susceptible to eradication.
DISCUSSION
Mutated kinases occur frequently in cancer and are amenable to targeted inhibition; however, mechanisms mediating acquired resistance have been observed for most targeted therapies (43). In contrast, current JAK inhibitors fail to eliminate JAK2V617F-mutant clones in patients with MPN, suggesting inadequate target inhibition and/or other genetic/nongenetic factors mediate JAK2V617F-mutant cell persistence in the setting of JAK inhibitor therapy (44). We show in preclinical models that there is an absolute requirement for JAK2V617F in MPN cells and that mutant-specific targeting of JAK2V617F abrogates MPN features, reduces mutant cell fraction, and extends overall survival with concomitant depletion of disease-sustaining stem cells within the HSPC compartment. Furthermore, our data suggest that JAK2V617F dependency persists even in the setting of antecedent mutations in epigenetic regulators, specifically TET2. Moreover, we demonstrate the feasibility of our dual-recombinase system to evaluate oncogenic signaling dependencies in vivo, and we believe that a similar approach will allow us to assess oncogenic dependencies and mechanisms of mutant-mediated transformation across a spectrum of malignant contexts.
These data support the notion that improved targeting of aberrant JAK2 signaling and downstream effectors offers greater therapeutic potential than current JAK kinase inhibitors and that JAK2V617F mutant–selective inhibition represents a potential curative strategy for the treatment of patients with MPN. Clinical translation may include more potent JAK kinase inhibitors that inhibit both mutant and WT JAK2, as shown preclinically with the type II JAK inhibitor CHZ868 in MPN models and in B-cell acute lymphoid leukemia (ALL; refs. 36, 45). Recent data highlight the potential for selective targeting of mutant calreticulin (CALR) in MPNs (46), and the elucidation of the first full-length mutant JAK kinase structure (47) provides a path to the development of true mutant-specific JAK2V617F inhibitors. As more potent (type II JAK2 inhibitors) and mutant-selective JAK2V617F inhibitors enter the clinic, we expect that these agents will show increased efficacy, including the ability to substantively reduce mutant allele burden. Our studies suggest that therapeutic agents that more potently inhibit constitutive JAK2 signaling will offer greater benefit to patients with MPN than current therapies, including in the presence of cooperating clonal hematopoiesis disease alleles.
METHODS
Experimental Animals
All animal studies were performed in accordance with institutional guidelines established by Memorial Sloan Kettering Cancer Center (MSKCC) under the Institutional Animal Care and Use Committee–approved animal protocol (#07-10-016) and the Guide for the Care and Use of Laboratory Animals (National Academy of Sciences 1996). All experimental animals were maintained on a 12-hour light–dark cycle with access to water and standard chow ad libitum. Veterinary staff provided regular monitoring and husbandry care. All mice had intact immune systems, were drug and test naïve, and had not been involved in previous procedures. Animals were monitored daily for signs of disease or morbidity, bleeding, failure to thrive, infection, or fatigue and sacrificed immediately if they exhibited any of the above signs. Mice harboring the Jak2RL allele were generated by Ingenious Targeting Laboratory in a C57BL/6J background. Specifically, a 8.86-kb genomic DNA used to construct the targeting vector was first subcloned from a positively identified C57BL/6J BAC clone (RP23-316C6). The region was designed such that the long homology arm (LA) extends ∼6 kb 5′ to the cluster of Lox2272-Rox-Rox12-Lox2272 sites, and the short homology arm (SA) extends about 2.2 kb 3′ to the Neo cassette and 3′ Rox12 site. The inversion cassette is in between the second set of Lox2272 and Rox sites, and it consists of the mutant exon 14* (V617F) and its flanking genomic sequences for correct splicing (SaE14*Sd). The inversion cassette replaces WT exon 14 and the same flanking genomic sequences included in the cassette. The BAC was subcloned into a ∼2.4 kb pSP72 (Promega) backbone vector containing an ampicillin selection cassette for retransformation of the construct. Ten micrograms of the targeting vector was then linearized and transfected by electroporation of FLP C57BL/6J (B6) embryonic stem cells. After selection with G418 antibiotic, surviving clones were expanded for PCR analysis to identify recombinant ES clones. After successful clone identification, the neomycin cassette was removed with a transient pulse of Cre recombinase and clones were reconfirmed following expansion. Finally, ES cells were injected in C57BL/6J mice via tetraploid complementation (NYU). Tet2f/f conditional knock-out mice, Cre-lox Jak2V617F knock-in mice, RC::RLTG reporter mice, Cre TdTomato reporter mice, and Ubc:CreER mice have been described previously (2, 22, 25, 42, 48). Six- to 8-week-old female and male Jak2RL or Jak2RL/Tet2f/f donor mice were used for Dre electroporation knock-in experiments. Age-matched 6- to 10-week-old female mice were used as donors for all transplant experiments (Ly5.1 Cd45.1 competitive or C57BL/6J noncompetitive). All Jak2RL donor mice used were crossed in a heterozygous fashion so as to retain a WT copy of JAK2.
Mouse Genotyping
DNA was isolated using the DNeasy Blood & Tissue Kit (Qiagen). The presence of the Jak2RL locus was genotyped using the following primers: forward: 5′-CGTGCATAGTGTCTGTGGAAGTC-3′; reverse: 5′-CGTGGAGAGTCTGTAAGGCTCAA-3′. The WT allele gives a band of 246 bp; the mutant allele gives a band of 833 bp. Jak2V617F knock-in genotyping was carried out with the following primers: forward: 5′-GCCATCTTTCCAGCCTAAAATTAG-3′; reverse: 5′-TCCAAAGAGTCTGTAAGTACAGAACT-3′ and with the following reaction conditions: 94°C for 3 minutes followed by 15 cycles of 94°C for 15 seconds, 65°C for 15 seconds, and 72°C for 30 seconds decreasing by 1°C per cycle, which is then followed by an additional 25 cycles of 94°C for 15 seconds, 50°C for 15 seconds, and 72°C for 30 seconds. Jak2V617F knock-out genotyping was carried out using the following primers: forward: 5′-GCCATCTTTCCAGCCTAAAATTAG-3′; reverse: 5′-ACCAGTTGCTCCAGGGTTACACG-3′ and with the following reaction conditions: 94°C for 2 minutes followed by 30 cycles of 94°C for 30 seconds, 53°C for 30 seconds, and 72°C for 30 seconds. Sequencing of the unrecombined Rox-lox locus was carried out using the following primers: forward: 5′-AGGAGCATCGATGACTACATGATGAG-3′; reverse: 5′-AGACTCTCCACGGTCTCATCTACG-3′ and with the following reaction conditions: 98°C for 30 seconds followed by 35 cycles of 98°C for 10 seconds, 65°C for 15 seconds, and 72°C for 30 seconds. Tet2 genotyping were carried out using the following primers/conditions: forward: 5′-AAGAATTGCTACAGGCCTGC-3′; reverse: 5′-TTCTTTAGCCCTTGCTGAGC-3′; ExR: 5′-TAGAGGGAGGGGGCATAAGT-3′ and with the following reaction conditions: 94°C for 2 minutes followed by 39 cycles of 94°C for 35 seconds, 58°C for 45 seconds, and 72°C for 55 seconds. Annotation of PCR genotyping results was carried out on a QIAxcel Advanced System (Qiagen) and analyzed using QIAxcel ScreenGel software (Qiagen). Sanger sequencing was performed by Genewiz and analyzed using Benchling software.
Dre mRNA Electroporation
Dre mRNA was purchased from TriLink Biotechnologies and electroporation carried out using the Neon Transfection System (Thermo Fisher Scientific) as per the manufacturer's protocol. Specifically, bone marrow donor cells were isolated from limb bones into PBS (pH 7.2) containing 2% FCS via centrifugation. After red blood cell (RBC) lysis, single-cell suspensions were depleted of lineage-committed hematopoietic cells using a Lineage Cell Depletion Kit according to manufacturer's protocol (EasySep, StemCell Technologies, Inc.). A total of 2.5–3.0 × 106 lineage-depleted bone marrow was then washed in PBS and then resuspended in 135 μL Buffer T, to which 15 μL of Dre mRNA (at 1 μg/μL) was quickly added and electroporated at the following conditions: 1,700V for 20 ms ×1 pulse. The cells were then pipetted into penicillin–streptomycin free StemSpan SFEM medium with thrombopoietin (TPO; 20 ng/mL; PeproTech) and stem cell factor (SCF; 20 ng/mL; PeproTech), cultured for 2 hours, and then subsequently harvested and washed/resuspended in PBS and transplanted via lateral tail vein injection into lethally irradiated (900 cGy) 6- to 8-week-old C57BL/6J recipient mice at approximately 4 × 105 cells per recipient along with 50,000 unelectroporated WT whole bone marrow support cells. Single-mutant Tet2−/− or double-mutant Jak2RL/Tet2−/− transplants/electroporations were carried out as above, except donor mice were dosed with tamoxifen (100 mg/kg by oral gavage daily ×4; purchased from MedChemExpress) 6–8 weeks prior to harvest and excision confirmed prior to Dre electroporation.
Transplantation Assays and In Vivo Experiments
Jak2RL, Tet2f/f, and Jak2RL/Tet2f/f lines were crossed to Ubc:CreER tamoxifen-inducible Cre lines and RLTG dual-recombinase reporter lines (25, 48). Primary recipient mice transplanted with Dre mRNA-recombined Ubc:CreER-Jak2RL, Ubc:CreER-Tet2−/−, or Ubc:CreER-Jak2RL/Tet2−/− bone marrow cells were bled every 3–4 weeks posttransplant to monitor disease status. Peripheral blood was isolated by submandibular bleeds, and complete blood counts were determined using a ProCyte Dx (IDEXX Laboratories) per manufacturer's instruction. For competitive repopulation assays, 1.2 × 106 whole bone marrow from primary transplant recipient mice exhibiting MPN was harvested 6–8 weeks posttransplant and combined with age-matched 0.8 × 106 Cd45.1 (Jackson Laboratories) whole bone marrow and transplanted into 6- to 8-week-old lethally irradiated Cd45.1 secondary recipient mice. Mice transplanted with Dre-recombined Jak2V617F cells demonstrating low Cd45.2 chimerism at baseline (<15%) and/or evidence of poor MPN cell engraftment were excluded from study cohorts. To induce Cre and delete Jak2V617F, mice were treated with tamoxifen (purchased from MedChemExpress) 100 mg/kg daily (dissolved in corn oil) by oral gavage × 4 followed by 14 days of tamoxifen chow (80 mg/kg daily; ENVIGO). Tamoxifen control studies were carried out using similar dosing schedules on 45.1 mice transplanted in competition with Dre-electroporated, Cre-negative Jak2RL MPN bone marrow cells. For terminal tissue isolation, mice were euthanized with CO2 asphyxiation, and tissues were dissected and fixed with 4% paraformaldehyde for histopathologic analysis. For whole bone marrow isolation, the femurs, hips, and tibias were dissected and cleaned. Cells were then isolated using centrifugation at 8,000 × g for 1 minute followed by RBC lysis (BioLegend) for 10–15 minutes. Bone marrow cell numbers and viability were determined using an automated cell counter (ViCell Blu, Beckman Coulter). Spleen cell suspensions were generated by crushing whole spleen and filtering through a 70-μm filter. RBC lysis (BioLegend) was performed, and cells were prepared for downstream processing or frozen.
In Vivo Drug Studies
For in vivo inhibitor studies, approximately 8 weeks after transplant, secondary transplant cohorts of lethally irradiated mice transplanted with Ubc:CreER-Jak2RL bone marrow in competition with Cd45.1 marrow (as above) and exhibiting active MPN were bled and cohorted on the basis of peripheral blood Cd45.2 chimerism and total WBC count to achieve congruency across treatment arms. Mice were then treated with ruxolitinib (60 mg/kg orally twice daily; dissolved in 20% Captisol in PBS; purchased from MedChemExpress), CHZ868 (30 mg/kg orally daily; dissolved in 0.5% methylcellulose + 0.5% Tween-80 in dH2O; purchased from MedChemExpress), tamoxifen to delete Jak2V617F (as above; purchased from MedChemExpress), or vehicle. Investigators were not blinded to the identity of mice or samples. Mice were treated for a total of 6 weeks before timed sacrifice and marrow/spleen harvested as above.
BMEC Culture
Bone marrow cells were isolated from limb bones into FACS buffer (PBS + 2% FBS) via centrifugation. After RBC lysis, single-cell suspensions were depleted of lineage-committed hematopoietic cells using a Lineage Cell Depletion Kit according to manufacturer's protocol (EasySep, StemCell Technologies, Inc.). Subsequently, 50,000 of the resulting lineage− cells were plated on a confluent monolayer of BMECs in a single well of a 12-well plate. Each well had 1 mL StemSpan SFEM (StemCell Technologies, Inc.) with 20 ng/mL recombinant murine SCF (PeproTech) in addition to the corresponding drug treatment: either 4-OHT (Sigma Aldrich; stock concentration: 13 mmol/L) or its vehicle, appropriately diluted in media to its final concentration [i.e., 0.01% (v/v) of ethanol (EtOH), or 200 nmol/L, 400 nmol/L or 1 μmol/L of 4-OHT; three replicates/condition]. The BMECs were seeded 2 days before plating the lineage− cells at a density of 100,000 cells/well. Cocultures were maintained for a total of 7 days at 37°C and 5% CO2, with media being completely refreshed with the original SCF and drug/vehicle concentrations. 4-OHT or EtOH vehicle was added to the culture on day 1 and again on day 4. On day 7, total cells were harvested with Accutase (BioLegend) and cell numbers were determined via an automatic cell counter (ViCell Blu, Beckman Coulter). Cells were then stained with the desired antibody cocktail and phenotyped by flow cytometry.
Flow Cytometry, Cell Sorting, and Western Blot Analysis
After single-cell preparation, murine peripheral blood, whole bone marrow, or spleen mononuclear cells were lysed for 10–15 minutes with RBC lysis buffer (BioLegend) and washed twice with FACS buffer. Cells were then resuspended in Fc (Cd16/32) block for 15 minutes and then subsequently stained with a cocktail comprised of antibodies targeting Cd3 (17A2), Cd45R/B220 (RA3-6B2), Gr-1 (RB6-8C5), Cd11b (M1/70), Cd45.2 (104), and Cd45.1 (A20) for 30 minutes. For hematopoietic stem/progenitor cell analysis, lysed bone marrow was stained with a cocktail of lineage markers along with antibodies against cKit (2B8), Sca1 (D7), FcγRII/III (2.4G2), Cd34 (RAM34), Cd150 (9D1), and Cd48 (HM48-1). Erythroid progenitor flow was carried out on unlysed bone marrow or spleen with the addition of the following antibodies: Cd105 (43A3), Cd71 (R17217), Cd41 (MWReg30), and Ter119 (Ter-119). All FACS antibodies were purchased from BD, BioLegend, or eBioscience. Following antibody incubation, cells were washed with FACS buffer and resuspended in a 4′,6-diamidino-2-phenylindole (DAPI)-containing FACS buffer solution for analysis and sorting. Samples were run on an LSRFortessa (Becton Dickinson) using FACSDiva software and analyzed with FlowJo v10.8.1 (Treestar). For sorting of Cd45.2+ lin−Sca1+cKit+ experiments, whole bone marrow samples were stained with antibodies for lineage cocktail, cKit, and Sca1 as well as Cd45.2 and Cd45.1 as above and gated and sorted on Lin−cKit+Sca1+ Cd45.2+ fractions using a FACSAria 3 (Becton Dickinson) instrument. Samples were subsequently spun at 1,500 rpm for 5 minutes, resuspended in Buffer ATL Cell Lysis solution (Qiagen), and DNA extracted using the DNA Micro Kit (Qiagen) as per the manufacturer's instructions. For Western blot analysis, whole-cell protein extracts from harvested splenocytes were prepared using RIPA buffer (Thermo Fisher Scientific) containing a protease/phosphatase inhibitor cocktail (Thermo Fisher Scientific). Protein quantification was performed using the Pierce BCA protein assay kit (Thermo Fisher Scientific) and analyzed on a Cytation 3 plate reader (BioTek). Proteins were separated by NuPAGE 4%–12% Bis-Tris Gel and transferred to a nitrocellulose membrane. The following antibodies were used: β-actin (Cell Signaling Technology, 4970S), STAT5 (Cell Signaling Technology, 94205S), and pSTAT5 (Cell Signaling Technology, 9359S). Images were obtained using the ChemiDoc Imaging System (Bio-Rad) and analyzed using ImageLab software (Bio-Rad).
Histology Staining, IHC, and Photography
Tibia and spleen samples were fixed in 4% paraformaldehyde for more than 24 hours and then embedded in paraffin. Paraffin sections were cut on a rotary microtome (Mikrom International AG), mounted on microscope slides (Thermo Fisher Scientific), and air-dried in an oven at 37°C overnight. After drying, tissue section slides were processed either automatically for hematoxylin and eosin (H&E) staining (COT20 stainer, Medite) or manually for reticulin staining. All samples and slide preparation, including IHC, were carried out at the Tri-Institutional Laboratory of Comparative Pathology (LCP) core facility. The following antibodies were used for IHC: Mac1 (Cedarlane CL8941B, 1:100), Ter119 (BD Biosciences, 550565 1:200), and p-44/42 MAPK (Erk1/2; Cell Signaling Technology, 4376, 1:100). Pictures were taken at 100×, 200×, and 400× (H&E, reticulin, and respective IHC) magnification using an Olympus microscope and analyzed with Olympus Cellsens software. Tissue sections were formally evaluated by a hematopathologist (W. Xiao), including reticulin scoring.
Assessment of Cell Cycle, Apoptosis, and Viability
Apoptosis was measured by flow cytometry on a LSRFortessa (Becton Dickinson) cytometer with Annexin V PerCPCy5.5 antibody (BioLegend) in combination with the antibody cocktail (above) in Annexin binding buffer (BioLegend) at 1:50 dilution in combination with DAPI as live/dead cell stain. For cell-cycle analysis, lineage-negative marrow was surface stained with the LSK antibody cocktail above followed by the Zombie UV Fixable Viability Kit (BioLegend) and then subsequently fixed and permeabilized using the FIX&PERM Cell Permeabilization Kit (Invitrogen) as per the manufacturer's instructions and stored at −20°C until further staining. Cells were then washed twice in FACS buffer, pelleted, and stained with anti–Ki-67 antibody (BioLegend) or isotype control for 30 minutes; washed again; and resuspended in FACS buffer with DAPI. Samples were run on linear for DAPI stain.
Colony-Forming Assays
To assess colony formation and serial replating capacity, 50,000 RBC-lysed whole bone marrow cells were seeded in 1.5 mL MethoCult M3434 (StemCell Technologies) with no additional supplemental cytokines in triplicate on 6-well plates and scored on day 8. For replating, cells were harvested and pooled and then reseeded once more at 50,000 cells/well in 1.5 mL MethoCult M3434 in triplicate. We assessed Dre mRNA-mediated recombination efficiency both pretransplant and posttransplant using either freshly Dre-electroporated Jak2RL lineage-negative bone marrow cells or whole marrow harvested 6 weeks following transplant from primary recipient mice transplanted with Dre-electroporated Jak2VF knock-in marrow. These cells were seeded as above, and after 7 days, individual colonies were plucked into 70 μL of Buffer ATL and DNA extraction was carried out using the DNA Micro Kit (Qiagen) as per the manufacturer's instructions.
Serum Cytokine Profiling
Serum samples were diluted two-fold with PBS (pH 7.2) and stored at −80°C until analysis. Cytokine assays were carried out using the Millipore Mouse Cytokine 32-plex kit and FlexMAP 3D platform (Luminex) per the manufacturer's instructions. xPONENT (Luminex) and Milliplex Analyst Software (Millipore) were used to convert mean fluorescent intensities (MFI) values into molecular concentrations using a standard curve (5-parameter logistic fitting method). Data were then normalized by first transforming concentration values using the log2 function and the mean and SD of the log values calculated across all samples for each analyte. Z-scores were then computed using the formula Z-score = (mean of log2 concentration values for an analyte per condition – mean of average log2 values for an analyte across all conditions)/SD calculated across the three conditions (WT, MPN, tamoxifen) and then used to normalize the cytokine data. The heat map was generated using the R package tidy_heatmap to visualize Z-score normalization for cytokines that displayed differential expression across the groups.
RNA-seq and Data Analysis
For gene expression analysis, secondary cohorts of lethally irradiated C57BL/6J mice transplanted with Ubc:CreER-Jak2RL-RLTG reporter bone marrow 8 weeks posttransplant and exhibiting MPN were treated with ruxolitinib (60 mg/kg orally twice daily), and tamoxifen (100 mg/kg by oral gavage daily × 4 followed by 80 mg/kg of tamoxifen chow × 3 days) ± vehicle (MPN control) for 7 days and then sacrificed. Lineage-depleted bone marrow was isolated and stained with an antibody cocktail containing a combination of lineage markers along with antibodies against cKit (2B8), Sca1 (D7), FcγRII/III (2.4G2), and Cd34 (RAM34) for 30 minutes; washed; and then resuspended in FACS buffer containing DAPI as a live/dead stain. TdTomato+ (Jak2RL knock-in) or GFP+ (Jak2RL knock-out) LSKs and MEPs were then sorted on a FACSAria III directly into TRIzol LS (Invitrogen) and stored at −80°C until processing. RNA was subsequently isolated using the Direct-Zol Microprep Kit (Zymo Research, R2061) according to manufacturer's protocol and quantified using the Agilent High Sensitivity RNA ScreenTape (Agilent 5067–5579) on an Agilent 2200 TapeStation. cDNA was generated from 1 ng of input RNA using the SMART-Seq HT Kit (Takara 634455) at half reaction volume followed by Nextera XT (Illumina FC-131-1024) library preparation. cDNA and tagmented libraries were quantified using High Sensitivity D5000 ScreenTape (5067–5592) and High Sensitivity D1000 ScreenTape, respectively (5067–5584). Libraries were sequenced on a NovaSeq at the Integrated Genomics Operation (IGO) at MSKCC. FASTQ files were mapped and transcript counts were enumerated using STAR (genome version mm10 and transcript version M13). Counts were input into R and RNA-seq analysis using DESeq2. Genes were filtered out prior to modeling in DESeq if they were not detected in all, with MEPs and LSKs modeled separately. Differentially expressed genes were identified with a log2-fold change of 1 and an adjusted P value of 0.05. Gene-set enrichment analysis was performed using the fgsea package at 100,000 permutations with gene sets extracted from the msigdbr package. ssGSEA was performed using the gsva package. To determine the frequency of the Jak2V617F allele and relative mutant expression, the samtools(v1.5)/mpileup variant calling tool was used. A minimum mapping quality of 30 for each read and default minimum base quality of 13 was used. Maximum depth was set to 100,000. Bcftools (v1.8) was used to convert BCF files into VCF files, and the vcfR (v1.14) package in R was used to parse the VCF files of alternative and reference alleles and read depth counts. Statistical differences between the different conditions were calculated using the one-sided Wilcoxon rank sum test. Figures were prepared using the ggplot2, ggsignif, ggrepel, and tidyheatmaps packages in R. Complete scripts can be found on github at https://github.com/bowmanr/goldilox.
Mouse ATAC-seq and Data Analysis
Chromatin accessibility assays utilizing the bacterial Tn5 transposase were performed as described previously (49). Briefly, 5.0 × 104 TdTomato+ (Jak2RL knock-in) or GFP+ (Jak2RL knock-out) cKit+ bone marrow cells from mice treated for 7 days with tamoxifen or an untreated MPN control cohort were sorted on a FACSAria III directly into PBS and subsequently lysed and incubated with transposition reaction mix containing PBS, Tagment DNA buffer, 1% Digitonin, 10% Tween-20, and Transposase (lllumina). Samples were then incubated for 30 minutes at 37°C in a thermomixer at 1,000 rpm. Prior to amplification, samples were concentrated with the DNA Clean and Concentrator Kit-5 (Zymo). Samples were eluted in 20 μL of elution buffer and PCR-amplified using the NEBNext 2X Master Mix (NEB) for 11 cycles and sequenced on a NextSeq 500 (Illumina). All samples were processed at the Center for Epigenetics Research (CER) core facility at MSKCC. Libraries were sequenced on a NovaSeq at the Integrated Genomics Operation (IGO) at MSKCC. Data analysis was completed through in-house scripts at the CER, in brief: Reads were trimmed with “trim_galore” and aligned to mouse genome mm9 using bowtie2 (default parameters). Duplicates were removed with the Picard tool “MarkDuplicates,” and peaks were called with MACS2, merged, and used to create a full peak atlas. Read counts were tabulated over this atlas using featureCounts. Downstream differential enrichment testing was completed in DESeq2 with default normalization scheme. HOMER was used for known motif enrichment amongst the differentially enriched peaks as defined by a fold change of ± 1.5 and an adjusted P value of 0.1. For the lineage deconvolution analysis presented in Supplementary Fig. S6H, we performed a process that uses a reference library from aggregated biological replicates across multiple cell types and selects key lineage-specific loci to deconvolve samples and generate component estimates (31). A nonnegative least-squares regression (NNLS) comparing each unknown sample to the set of normal hematopoietic states is then performed. Deconvolution coefficients are interpreted as proportions to estimate the magnitude per hematopoietic stage.
Human scATAC-seq and Data Analysis
scATAC-seq data were processed using cellranger-ATAC (v2.0.0) mkfastq. ATAC-sequencing reads were then aligned to the hg38 reference genome using cellranger-ATAC count function. Fragment files generated by cellranger-ATAC were used as input for the ArchR (ref. 50; v1.0.0). For initial dimensionality reduction and patient data integration, the cell by genomic bin matrix was used as input for reciprocal latent semantic indexing (LSI) as calculated by the Signac (v1.1.1). Transcription factor motif accessibility z-scores were calculated with ChromVAR (ref. 51; v1.8.0). The earliest HSPCs (cluster HSPC1, ref. 35) were subset for downstream analysis, and statistical comparisons of motif accessibility for NFKB1, REL, FOS, and JUN transcription factors were performed via linear mixture model including patient identity as random effect to account for potential technical confounders arising from sample-specific batch effects. For heat map representation, motif accessibility z-scores were used as input and the pheatmap (v1.0.12) R package was used.
Quantitative Real-Time PCR
Total RNA was extracted from magnetic-bead isolated cKit+ bone marrow (Miltenyi Biotec) using the Direct-zol RNA extraction kit (Zymo) as per the manufacturers’ protocols, respectively. Complementary DNA was then reverse transcribed using the Verso cDNA Synthesis kit (Thermo Fisher Scientific). Ybx1 expression was evaluated by quantitative reverse-transcription (qRT) PCR using TaqMan probes purchased from Thermo Fisher Scientific (Mm00850878_g1) on the RealPlex thermocycler (Thermo Fisher Scientific).
Statistical Analysis
Statistical analyses were performed using Student t test (normal distribution) using GraphPad Prism version 6.0h (GraphPad Software) unless otherwise noted. Kaplan–Meier curves were determined using the log-rank test. P < 0.05 was considered statistically significant. For the cytokine analysis presented in Fig. 1F and Supplementary Fig. S3J, individual cytokines were analyzed using the Kruskal–Wallis test with lower values set at the lower limit of the assay and P values generated by doing multiple comparisons testing across treatment arms (WT vs. MPN vs. tamoxifen) and adjusting based on FDR of ≤ 0.05. The number of animals, cells, and experimental replication can be found in the respective figure legends.
Data Availability
Raw and processed sequencing data are made available at https://github.com/bowmanr/goldilox and via the NCBI Gene-Expression Omnibus (GEO) at GSE203464.
Authors’ Disclosures
A.J. Dunbar reports personal fees from Incyte outside the submitted work; in addition, A.J. Dunbar has a patent for PCT/US2023/066910 pending. R.L. Bowman reports a patent for PCT/US2023/066910 pending. F. Izzo reports grants from American Society of Hematology Fellow-to-Faculty Scholar Award during the conduct of the study. W. Xiao reports grants from Stemline therapeutics outside the submitted work. S.F. Cai reports a consultancy for and previously held equity interest in Imago Biosciences, neither of which is directly related to the content of this article. J.L. Glass reports grants from NIH during the conduct of the study and personal fees from GLG outside the submitted work. A.D. Viny reports other support from Arima Genomics outside the submitted work. R.P. Koche reports personal fees from Econic Biosciences outside the submitted work. S.C. Meyer reports grants from Swiss National Science Foundation, Cancer League Basel, Foundation “Stiftung für krebskranke Kinder Regio Basiliensis,” and grants from Foundation for the Fight against Cancer during the conduct of the study; personal fees from Novartis, Celgene/BMS, GSK, OrphaSwiss GmbH, Ajax Therapeutics Inc.; other support from AbbVie AG; and other support from Amgen outside the submitted work. In addition, S.C. Meyer has a patent for PAT058952-US-PSP pending and a patent for PAT058953-US-PSP pending. D.A.L. is on the Scientific Advisory Board of Mission Bio, Alethiomics, Pangea, Quotient Therapeutics and C2i Genomics and has received prior research funding from BMS, 10X Genomics, Mission Bio, Ultima Genomics, Oxford Nanopore, and Illumina unrelated to the current manuscript. R.L. Levine reports other support from Ajax during the conduct of the study; in addition, R.L. Levine has a patent for JAK2V617F reversible mouse pending; is on the supervisory board of Qiagen; and is a scientific advisor to Imago, Mission Bio, Zentalis, Ajax, Auron, Prelude, C4 Therapeutics, and Isoplexis. R.L. Levine also receives research support from Ajax, Zentalis, and Abbvie; has consulted for Incyte, Janssen, and Astra Zeneca; and has received honoraria from Astra Zeneca for invited lectures. No disclosures were reported by the other authors.
Authors’ Contributions
A.J. Dunbar: Conceptualization, data curation, formal analysis, funding acquisition, investigation, visualization, writing–original draft, writing–review and editing. R.L. Bowman: Conceptualization, resources, data curation, formal analysis, supervision, validation, investigation, visualization, writing–original draft. Y.C. Park: Investigation. K. O'Connor: Validation, investigation, methodology. F. Izzo: Conceptualization, formal analysis, validation, visualization, methodology. R.M. Myers: Conceptualization, formal analysis, validation, investigation, visualization, methodology, writing–original draft. A. Karzai: Investigation. Z. Zaroogian: Investigation. W. Kim: Validation, investigation. I. Fernandez-Maestre: Validation, investigation, visualization, methodology. M.R. Waarts: Investigation, visualization. A. Nazir: Investigation. W. Xiao: Validation, investigation, visualization, writing–original draft. T. Codilupi: Validation, investigation, visualization. M. Brodsky: Investigation. M. Farina: Investigation. L. Cai: Investigation. S.F. Cai: Investigation, visualization. B. Wang: Investigation. W. An: Investigation. J.L. Yang: Investigation, visualization, methodology. S. Mowla: Investigation. S.E. Eisman: Investigation. A. Hanasoge Somasundara: Investigation. J.L. Glass: Investigation. T. Mishra: Investigation. R. Houston: Investigation. E. Guzzardi: Investigation. A.R. Martinez Benitez: Investigation. A.D. Viny: Conceptualization, investigation. R.P. Koche: Conceptualization, data curation, formal analysis, investigation, methodology. S.C. Meyer: Investigation, visualization, methodology. D.A. Landau: Conceptualization, resources, data curation, formal analysis, funding acquisition, validation, investigation, visualization, methodology. R.L. Levine: Conceptualization, resources, data curation, formal analysis, funding acquisition, writing–original draft, project administration, writing–review and editing.
Acknowledgments
We are grateful to members of the Levine Lab for their discussion of the work. We would also like to acknowledge Dr. Alex Joyner (MSKCC), Dr. Patricia Jensen (NIH), and Dr. Tudor Badea (NIH) for discussion on Dre-Rox technology as well as Sime Brkic (University Hospital Basel Switzerland) for their technical advice and support. This work was supported by the NCI award P01 CA108671 (to R.L. Levine). R.L. Levine was supported by a Leukemia and Lymphoma Society Scholar award. A.J. Dunbar is a William Raveis Charitable Fund Physician-Scientist of the Damon Runyon Cancer Research Foundation (PST-24-19). He also has received funding from the NIH (T32CA009207), AACR (17-40-11-DUNB), and the American Association of Clinical Oncology. R.L. Bowman was supported by a Damon Runyon-Sohn Fellowship and the NCI (K99CA248460). R.M. Myers is supported by a Medical Scientist Training Program grant from the National Institute of General Medical Sciences of the NIH under award number T32GM007739 to the Weill Cornell/Rockefeller/Sloan Kettering Tri-Institutional MD-PhD Program and by the Weill Cornell Medicine NYSTEM Training Program under award number C32558GG. F. Izzo is supported by the American Society of Hematology Fellow-to-Faculty Scholar Award. W. Xiao is supported by Alex's Lemonade Stand Foundation and the Runx1 Research Program, a Cycle for Survival's Equinox Innovation Award in Rare Cancers, MSK Leukemia SPORE (Career Enhancement Program, NIH/NCI P50 CA254838-01), and an NCI grant (K08CA267058-01). S.F. Cai is supported by a Career Development Award from the NCI (K08CA241371-01A1). J.L. Glass is supported by a K08 through the NIH (CA230172). A.D. Viny is supported by the National Cancer Institute MERIT award (R37CA286857), an EvansMDS Discovery grant from the Edward P. Evans Foundation, a Clinical Investigator grant from the Damon Runyon Cancer Research Foundation (120-22), a Clinician Scientist Development grant from the Doris Duke Charitable Foundation, and grants from the Columbia University Vagelos College of Physicians & Surgeons (Gerstner Scholar and Early Career Physician Scientist). S.C. Meyer receives funding from the Swiss National Science Foundation (PZ00P3_161145, PCEFP3_181357), the Cancer League Basel and the “Stiftung für krebskranke Kinder Regio Basiliensis” (KLbB-4784-02-2019), and the Foundation for the Fight against Cancer. D.A. Landau is supported by the Burroughs Wellcome Fund Career Award for Medical Scientists, Valle Scholar Award, Leukemia Lymphoma Scholar Award, the MacMillan Family Foundation and the MacMillan Center for the Study of the Non-Coding Cancer Genome at the New York Genome Center, and the Mark Foundation Emerging Leader Award as well as the Tri-Institutional Stem Cell Initiative, the National Heart Lung and Blood Institute (R01HL145283; R01HL157387-01A1), the NCI (R33 CA267219), the National Human Genome Research Institute, Center of Excellence in Genomic Science (RM1HG011014), and the NIH Common Fund Somatic Mosaicism Across Human Tissues (UG3NS132139). Studies supported by MSK core facilities were supported in part by MSKCC Support Grant/Core Grant P30 CA008748 and the Marie-Josée and Henry R. Kravis Center for Molecular Oncology. R.L. Levine is also supported by a Leukemia & Lymphoma Society Specialized Center of Research grant.
Note: Supplementary data for this article are available at Cancer Discovery Online (http://cancerdiscovery.aacrjournals.org/).