RETRACTED: In Vivo Monitoring of Polycythemia Vera Development Reveals Carbonic Anhydrase 1 as a Potent Therapeutic Target

A focal model of PV development reveals its invasion of distant organ sites, dependency on the spleen, and carbonic anhydrase 1 as a potential therapy target.

Abstract

Current murine models of myeloproliferative neoplasms (MPNs) cannot examine how MPNs progress from a single bone marrow source to the entire hematopoietic system. Thus, using transplantation of knock-in JAK2^V617F hematopoietic cells into a single irradiated leg, we show development of polycythemia vera (PV) from a single anatomic site in immunocompetent mice. Barcode experiments reveal that grafted JAK2^V617F stem/progenitor cells migrate from the irradiated leg to nonirradiated organs such as the contralateral leg and spleen, which is strictly required for development of PV. Mutant cells colonizing the nonirradiated leg efficiently induce PV in nonconditioned recipient mice and contain JAK2^V617F hematopoietic stem/progenitor cells that express high levels of carbonic anhydrase 1 (CA1), a peculiar feature also found in CD34⁺ cells from patients with PV. Finally, genetic and pharmacologic inhibition of CA1 efficiently suppresses PV development and progression in mice and decreases PV patients’ erythroid progenitors, strengthening CA1 as a potent therapeutic target for PV.

Significance:

Follow-up of hematopoietic malignancies from their initiating anatomic site is crucial for understanding their development and discovering new therapeutic avenues. We developed such an approach, used it to characterize PV progression, and identified CA1 as a promising therapeutic target of PV.

This article is highlighted in the In This Issue feature, p. 265

INTRODUCTION

Hematopoietic malignancies have been modeled using transgenic and knock-in models with or without bone marrow (BM) transplantation of mutated hematopoietic cells (1, 2). These models result in the expression of the mutated gene(s) in all targeted hematopoietic cells and in all organs that contain these hematopoietic cells. Thus, how a hematopoietic malignancy can progressively invade hematopoietic organs from a defined single anatomic site, as it occurs in patients, has not been studied in immunocompetent mice, calling for the development of new experimental approaches.

JAK2^V617F mutation within blood-forming cells is found in more than 90% of patients with polycythemia vera (PV; ref. 3) and in more than 50% of patients with essential thrombocytemia (4) and primary myelofibrosis (5). Mouse models of these myeloproliferative neoplasms (MPNs) have been useful in characterizing JAK2 inhibitors or other drugs that were subsequently employed for MPN treatment (6). Unfortunately, except for use of IFNα (7), none of these treatments efficiently targets the malignant clone and after an initial beneficial response, many patients lose responsiveness to these inhibitors (8), and/or develop adverse effects (9), indicating the urgent need for new therapeutic targets.

We have developed an in vivo approach to study the progression of PV from a single irradiated anatomic site to the entire hematopoietic system and, in doing so, we show an essential role of the spleen and a previously unsuspected major contribution by carbonic anhydrase 1 (CA1) to PV development.

RESULTS

An Approach to Following PV Progression in Immunocompetent Mice

Sequential irradiation (2 × 5 Gy with 17-hour rest) of a single C57BL/6J mouse leg resulted in decreased BM cellularity and in decreased numbers of long-term hematopoietic stem cells (LT-HSCs) only in the irradiated leg 7 hours after the second irradiation (Supplementary Fig. S1A–S1C). Transplantation of wild-type (WT) BM cells led to a stable but largely restricted engraftment of WT hematopoietic cells within the irradiated leg and to a stable blood chimerism during weeks 4 to 20 (Fig. 1A; Supplementary Fig. S1D and S1E), thus precluding an adoptive transfer of WT BM cells. In contrast, transplantation of JAK2^V617F BM cells from knock-in mice resulted in a progressive increase of JAK2^V617F hematopoietic cells in both legs (Fig. 1A) without any significant change in their BM cellularity (Supplementary Fig. S1F). Numbers of JAK2^V617F LT-HSCs, multipotent progenitors (MPP) and granulocyte–monocyte progenitors (GMP; Fig. 1B; Supplementary Fig. S1G and S1H for gating strategies) were increased in the BM of irradiated and nonirradiated legs by 6 weeks after transplantation, indicating proliferation and/or differentiation of JAK2^V617F LT-HSCs and of JAK2^V617F MPP and GMP compartments. In both legs and only after transplantation of JAK2^V617F BM cells, the percentages of endogenous Lin⁻Sca⁺cKit⁺ (LSK) cells and megakaryocyte–erythroid progenitors (MEP) in G₀ phase and of JAK2^V617F MEP in S–G₂–M phase increased (Fig. 1C). This was associated, in both legs and as early as two weeks after transplantation, with an expansion of JAK2^V617F MEP at the expense of endogenous MEP and an almost full contribution of JAK2^V617F cells to the CD71⁺Ter119⁺ erythroid cell compartment without expansion of this compartment (Fig. 1D; Supplementary Fig. S1H and S1I for comparison with CD11b⁺Gr1⁺ myeloid cells). Mice with a single-irradiated-leg transplanted with JAK2^V617F BM cells developed PV syndrome as assessed by progressively increased JAK2^V617F cell burden and red blood parameters (Fig. 1E and F) followed by death of 80% of mice 9 weeks posttransplantation (Fig. 1G). Altogether, these results indicate that this single-irradiated-leg mouse model can be used to study JAK2^V617F hematopoietic cell dynamics associated with PV progression in an immunocompetent mouse model.

Figure 1.

Follow up of PV development in mice. A, Experimental design (top) and kinetics of transplanted WT or JAK2^V617F BM cells contribution to total BM hematopoietic cells in the irradiated and nonirradiated legs (bottom). B, Kinetics of WT or JAK2^V617F cells contribution to the long-term hematopoietic stem cell (LT-HSC). C, Cell-cycle analysis of WT (left, blue) and JAK2^V617F (middle and right, red) Lin⁻cKit⁺Sca⁺ (LSK) cells (top) and megakaryocyte–erythroid progenitors (MEP, bottom) and their endogenous counterparts (grey) in the irradiated and nonirradiated legs 3 weeks after transplantation. D, Kinetics of WT or JAK2^V617F cells contribution to MEP (top) and to CD71⁺Ter119⁺ mature erythroid (bottom) compartments in the irradiated and nonirradiated legs. E, Kinetics of peripheral blood cells after transplantation of JAK2^V617F or WT BM cells in a single irradiated leg. F, Kinetics of blood parameters after transplantation of JAK2^V617F or WT BM cells in a single irradiated leg. G, Survival curve of mice after transplantation of JAK2^V617F or WT BM cells in a single irradiated leg. Data are mean ±SEM from three independent experiments (n = 8–12 mice). Significance was assessed using unpaired two-tailed t test and two-way ANOVA followed by post hoc analysis (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; B and D, asterisks denote P-value comparing chimerism (% of GFP cells as indicated with the number (mean)) between Donor JAK2 vs Donor WT, and # denotes P-value comparing the % of cells studied from BM (irradiated and nonirradiated leg) of mice transplanted with WT or JAK2^V617F cells).

JAK2^V617F LSK Cells from the Irradiated Leg Colonize the Nonirradiated Leg

Chemokine detection assays undertaken one week after BM transplantation revealed increased levels of chemokines such as MMP9, CCL6, soluble V-CAM-1 and I-CAM-1, known to be involved in hematopoietic stem and progenitor cell (HSPC) mobilization (10, 11). These increases were observed only in the irradiated leg BM supernatant and only after JAK2^V617F BM transplantation (Supplementary Fig. S2A). Increased chemokine levels were associated with increased numbers of circulating JAK2^V617F HSPC starting 4 days after transplantation (Fig. 2A) and suggested that JAK2^V617F HSPC grafted into the irradiated leg could migrate to nonirradiated organs such as the nonirradiated leg. To test this hypothesis, JAK2^V617F LSK cells were transduced with a lentiviral barcode library (12) at a MOI of 1 and transplanted into the irradiated leg of recipient mice (Fig. 2B, top scheme; Supplementary Fig. S2B and S2C). Of note, the number of donor WT LSK cells in the nonirradiated leg was too low (chimerism lower than 5%, see Fig. 1A and B) to sort these cells and to perform bar-coded analysis. As expected, PV development in mice transplanted with JAK2^V617F LSK cells was slower than in mice transplanted with JAK2^V617F BM as shown by the delayed increase of blood parameters (compare Fig. 1F with Supplementary Fig. S2D). As the single-irradiated-leg mouse model does not lead to aplasia, analysis of bar-coded JAK2^V617F LSK cells could be performed as early as 11 weeks after their transplantation. This analysis showed that the number of bar-coded clones detected in the irradiated leg was more than two-fold higher than in the nonirradiated leg (Fig. 2B; Supplementary Fig. S2E). Of note, all except two clones (t19 and t185) present in the nonirradiated leg were present in the irradiated leg (Fig. 2B). Interestingly, while the bar-coded JAK2^V617F LSK found in the irradiated leg differed in their number and relative proportion in the 3 mice studied (Fig. 2C, left), only one or two bar-coded JAK2^V617F LSK were dominating in the nonirradiated legs of these mice (Fig. 2C, right). These results suggested different dynamics of JAK2^V617F LSK clone proliferation and/or differentiation in the irradiated and nonirradiated legs. Accordingly, in the nonirradiated leg, 67% bar-coded CD71⁺Ter119⁺ erythroid cells shared barcodes with JAK2^V617F LSK cells (Fig. 2D; Supplementary Fig. S2E), whereas only 50% of bar-coded myeloid cells had corresponding bar-coded JAK2^V617F LSK cells (Fig. 2E; Supplementary Fig. S2E). In addition, in the nonirradiated leg, 75% of bar-coded JAK2^V617F LSK gave rise to bar-coded erythroid and myeloid progeny but with higher frequencies of erythroid cells (Fig. 2F; Supplementary Fig. S2E). Altogether, these results showed that JAK2^V617F LSK cells from the irradiated leg could colonize the nonirradiated leg where they contributed largely to erythroid cells production.

Figure 2.

JAK2^V617F LSK from irradiated leg colonize the nonirradiated leg. A, Kinetics of donor-derived WT or JAK2^V617F colony-forming unit per mL of blood from mice transplanted with WT or JAK2^V617F BM cells in a single irradiated leg. Data are mean ±SEM. Significance was assessed using unpaired two-tailed t test and two-way ANOVA followed by post hoc analysis (**, P < 0.01; ***, P < 0.001). B, Experimental design (top) and frequency of JAK2^V617F LSK bar-coded clones in the nonirradiated (y-axis) and irradiated (x-axis) legs (bottom left), the barcode frequency is log transformed and bar-coded clones discussed in the main text are highlighted. Venn diagram of bar-coded JAK2^V617F LSK clones in the nonirradiated and irradiated legs (n = 3, bottom right). C, Representative distribution of bar-coded JAK2^V617F LSK clones in the nonirradiated and irradiated legs for each mouse (n = 3). D, Venn diagram of bar-coded JAK2^V617F LSK and CD71⁺Ter119⁺ erythroid clones in the nonirradiated and irradiated legs (n = 3). E, Venn diagram of bar-coded JAK2^V617F LSK and CD11b⁺Gr1⁺ myeloid clones in the nonirradiated and irradiated legs (n = 3). F, Venn diagram of bar-coded JAK2^V617F LSK, CD71⁺Ter119⁺ erythroid and CD11b⁺Gr1⁺ myeloid clones in the nonirradiated leg (left) and histograms showing the ratio of distribution of erythroid (E) and myeloid (M) bar-coded clones (right).

The Spleen Is Necessary for PV Development through Amplification of the JAK2^V617F Erythroid Compartment

Although splenomegaly is associated with MPN (13), the role of the spleen in the development of these neoplasms is incompletely defined (14). Bearing in mind that mouse but not human spleen is a full-fledged hematopoietic organ, we examined the spleen in the mouse model used. Spleen weight and size increased after transplantation of JAK2^V617F hematopoietic cells (Fig. 3A) with an increased JAK2^V617F chimerism in splenic white blood cells (Supplementary Fig. S3A). In the spleen, JAK2^V617F cells increased their contribution to the LSK compartment without any overall significant increase of this compartment (Fig. 3B). As in BM, the splenic MEP compartment increased due to high contribution of JAK2^V617F MEP but, in contrast to BM, this increase was associated with specific pronounced expansion of the CD71⁺Ter119⁺ erythroid compartment as early as two weeks after transplantation (compare Fig. 3C with 1D for CD71⁺Ter119⁺ erythroid cells and Supplementary Fig. S3B for CD11b⁺Gr1⁺ myeloid cells). These results suggested that the spleen greatly increased red blood cell production in this mouse model of PV. Barcode analysis showed that more than 90% of LSK barcode clones present in the irradiated and nonirradiated legs were also present in the spleen (Fig. 3D) suggesting that these clones migrated from the irradiated leg into the nonirradiated leg and the spleen. Compared with the nonirradiated leg, the spleen contained 2× more erythroid clones and 1.3× more myeloid clones (Fig. 3E, top schemes) but amplification of erythroid or myeloid clones common to both spleen and nonirradiated leg was much more frequent for myeloid than for erythroid clones (Fig. 3E, histograms), suggesting that the spleen harbored more erythroid clones and preferentially amplified myeloid clones. To unravel the role of the spleen in PV development, mice were splenectomized one week after transplantation of JAK2^V617F hematopoietic cells, that is, when the spleen began to be colonized by JAK2^V617F cells (Supplementary Fig. S3C). The splenectomized mice did not die prematurely, suggesting escape from PV syndrome (Fig. 3F). These mice showed normal red blood parameters (Fig. 3G) and a blood chimerism similar to that in mice transplanted with WT cells when followed up to 16 weeks after transplantation (Fig. 3H). Furthermore, in nonhematopoietic organ of these splenectomized mice such as the liver, a decreased number of JAK2^V617F erythroid cells was found (Supplementary Fig. S3D). These observations suggested that splenectomy rescue these mice from PV. Nevertheless, at 26 weeks, these mice showed a high contribution by JAK2^V617F cells to both myeloid-, erythroid- (Fig. 3I) and LT-HSC (Fig. 3J) compartments associated with a decreased BM cellularity in both legs (Fig. 3K) indicating that these mice survived with an almost completely mutant cell–derived hematopoiesis. Similar results were obtained when splenectomy was performed 2 weeks after transplantation; that is, after detectable mutant cell hematopoiesis in the nonirradiated leg (Supplementary Fig. S3E–S3H). These results indicate that despite the persistence and amplification of JAK2^V617 cells in bones, splenectomy hinders the amplification of the erythroid compartment and the development of PV.

Figure 3.

The spleen is necessary for PV development through amplification of the JAK2^V617F erythroid compartment. A, Kinetics of spleen weight (left) and pictures of spleens 3 weeks after transplantation of WT and JAK2^V617F BM cells (right). B, Kinetics of WT or JAK2^V617F cells contribution to the LSK compartment. C, Kinetics of WT or JAK2^V617F cells contribution to the megakaryocyte–erythroid progenitors (MEP) and the CD71⁺Ter119⁺ erythroid compartments in the spleen. D, Percentage of JAK2^V617F LSK bar-coded clones shared between the irradiated leg, the nonirradiated leg, and the spleen. E, Venn diagrams of bar-coded CD71⁺Ter119⁺ erythroid (left top) and CD11b⁺Gr1⁺ myeloid clones (right top) in the nonirradiated leg and in the spleen and histograms showing the ratio of distribution of bar-coded erythroid (left bottom) and myeloid (right bottom) clones in the nonirradiated leg (NIR E, NIR M) and in the spleen (spleen E, spleen M; n = 3). F, Experimental design (top) and survival curve (bottom) of mice after splenectomy one week after transplantation of JAK2^V617F or WT BM cells in a single irradiated leg. G, Kinetics of blood parameters in mice after splenectomy 1 week after transplantation of JAK2^V617F or WT BM cells in a single irradiated leg. H, Kinetics of blood cell chimerism in mice after splenectomy one week after transplantation of JAK2^V617F or WT BM cells in a single irradiated leg. I, JAK2^V617F, WT and endogenous cells contributions to the irradiated and nonirradiated legs to the CD11b⁺Gr1⁺ myeloid and the CD71⁺Ter119⁺ erythroid compartments. J, JAK2^V617F, WT and endogenous cells contributions to the irradiated and non­irradiated legs to the LT-HSC compartment 26 weeks after transplantation. K, Total hematopoietic cell number in the irradiated and nonirradiated legs 26 weeks after transplantation. Data are mean ±SEM from 8 to 12 mice. Significance was assessed using unpaired two-tailed t test and two-way ANOVA followed by post hoc analysis (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; B and C, asterisks denote chimerism, and # denotes percentage of cells).

Inhibition of CA1 Suppresses PV Progression

Evolution and/or selection of migrating JAK2^V617F clones were functionally studied by secondary transplantation. Whereas 20% of nonconditioned immunocompetent mice transplanted with BM from the irradiated leg of primary JAK2^V617F recipient mice developed PV, up to 66% of mice transplanted with BM from the nonirradiated leg developed PV (Fig. 4A), indicating that JAK2^V617F LSK cells that have migrated and nested within the nonirradiated leg exhibited increased fitness for a nonirradiated environment. We thus searched for the molecular basis of this increased fitness, with the hypothesis that such selection might mirror observed clonal dominance acquisition of mutated HSCs during the natural history of human disease (15). To obtain further mechanistic insights, we compared transcriptomic profiles of sorted JAK2^V617F HSPC and their reciprocal endogenous WT HSPC in the nonirradiated leg (Supplementary Fig. S4A). Gene-set enrichment analysis revealed the downregulation of genes associated with signaling by tyrosine kinases and activation of genes associated with oxidative phosphorylation and erythrocytes oxygen uptake and carbon dioxide release (Supplementary Fig. S4B and S4C). One of the most differentially expressed genes was the CA1 gene (Car1) gene and, indeed, high Car1 mRNA levels were detected in JAK2^V617F HSPC from both legs with the highest Car1 mRNA levels in the nonirradiated leg (Fig. 4B; Supplementary Fig. S5A). This high mRNA level was associated with a high CAR1 protein level (Fig. 4C). In JAK2^V617F CD34⁺ progenitor cells from patients with PV, RNA-sequencing (RNA-seq) analysis showed that only CA1 and CA2 mRNA levels were increased with CA1 mRNA levels being by far the highest (Fig. 4D). The increase of CA1 mRNA level in CD34⁺ progenitor cells from patients with PV was validated by qRT-PCR (Supplementary Fig. S5B) and associated with a 4-fold increase of CA1 protein level (Fig. 4E). To assess the contribution of CAR1 to PV development in mice, JAK2^V617F LSK cells were transduced with Car1 shRNA or control shRNA lentiviruses and transplanted into the irradiated leg of recipient mice (Supplementary Fig. S5C, top scheme). Mice transplanted with control shRNA JAK2^V617F LSK cells displayed blood parameters associated with PV progression, whereas mice transplanted with Car1 shRNA JAK2^V617F LSK cells showed no sign of the disease (Fig. 4F; Supplementary Fig. S5D and S5C, left and middle panels), except when nontransduced JAK2^V617F LSK cells were the most abundant (Supplementary Fig. S5C, bottom right panels). Car1 shRNA used is a mix of 3 short hairpin RNAs (shRNA). As shRNAs are notorious for off-targets effects, we performed a shRNA rescue experiment to detect any off-target effects. JAK2^V617F Lin⁻c-Kit⁺ cells from BM of single-leg-irradiated mice transplanted with JAK2V617F BM cells were cotransduced with a lentivirus expressing control sh (shCtl) or sh 1 or sh2 or sh3 that target Car1 mRNA and a lentivirus expressing a WT CAR1 protein (silent mutant CAR1 as SmCAR1) from a Car1 cDNA containing silent mutations that impair the action of sh1, sh2, and sh3. Similar increases of JAK2^V617F BFU-E/CFU-E were obtained whatever shRNA expressed together with SmCAR1 (Supplementary Fig. S5E) and were associated with increased levels of CAR1 (Supplementary Fig. S5F) indicating the specificity and efficiency of the three shRNAs used. These findings prompted us to pharmacologically inhibit CAR1. Mice transplanted with JAK2^V617F BM cells received daily injections of furosemide, a specific CAR1 inhibitor currently used as a diuretic (16). Treatment at 20 mg/kg of furosemide resulted in decreased JAK2^V617F cell burden in blood (Fig. 4G), decreased spleen weight (Fig. 4H), and improved blood parameters (Supplementary Fig. S5G) and improved survival rates (Fig. 4I) when compared with untreated littermates transplanted with JAK2^V617F BM cells. Similar results were obtained by treating mice with another CAR1 inhibitor, acetazolamide (Supplementary Fig. S5H and S5I) but, for the remainder of this study, we used furosemide as its K_i for CAR1 is higher than the acetazolamide one (17). To determine whether furosemide might impede PV after its establishment, furosemide was injected daily into mice after disease onset. This treatment resulted in a rapid, stable, and furosemide-dependent decrease of JAK2^V617F cell burden in blood (Fig. 4J, left) associated with a decreased hematocrit (Fig. 4J, right) and spleen weight (Fig. 4K), increased survival of the treated mice and a significant decrease of JAK2^V617F LT-HSC contribution to the LT-HSC compartment (Fig. 4L). As furosemide is also a diuretic, we treated JAK2^V617F mice with spironolactone, a diuretic which displays no inhibition of CAR1 activity (18) and we found that spironolactone had no effect on the survival of mice with PV (Supplementary Fig. S5J) and PV-associated symptoms (Supplementary Fig. S5K). To study the effect of furosemide at a dose within the range used for patients, mice transplanted with JAK2^V617F BM cells were daily injected with furosemide at 5 mg/kg. This treatment resulted in a therapeutic effect similar to the one obtained with 20 mg/kg (Supplementary Fig. S5L). Finally, furosemide treatment of CD34⁺ cells from patients with PV decreased the percentages of BFU-E (Fig. 4M) with preferential effect on JAK2^V617F CD34⁺ cells monitored by the decreased variant allele frequency in the BFU-E (Fig. 4N). Altogether, these results demonstrate that CAR1 is a potential therapeutic target for suppression of PV syndrome in this preclinical model and for PV patient erythroid progenitors through specific targeting of JAK2^V617F cells.

Figure 4.

Genetic and pharmacologic inhibition of CA1 suppresses PV progression. A, Experimental design (top left), mortality trends after the secondary transplantations (bottom left) and kinetics of blood chimerism in secondary recipients transplanted with BM from the irradiated leg or the nonirradiated leg of primary mice transplanted with JAK2^V617F or WT BM cells (right). B, Heat map of all members of the CA family (JAK2^V617F HSPC vs. endogenous HSPCs; left) and relative mRNA levels of CA1 in HSPCs from irradiated and nonirradiated legs 6 weeks after transplantation of JAK2^V617F BM cells (right; n = 3–6). C, CAR1 protein expression level in HSPCs of the nonirradiated leg (left) normalized against GAPDH protein levels (right). D, mRNA level of all CA isoforms in human CD34⁺ cells from control and patients with JAK2^V617F PV (n = 4). E, CA1 protein expression in human CD34⁺ cells from control and patients with JAK2^V617F PV (left) (n = 4) and CA1 expression level normalized against GAPDH expression level (right). F, Blood chimerism (left) and hematocrit (right) in mice transplanted with JAK2^V617F or WT LSK cells transduced with Car1 or Ctl shRNA lentivirus in a single-leg-irradiated mice (n = 5). G, Kinetics of blood chimerism in mice transplanted with JAK2^V617F or WT BM cells and daily treated or not with furosemide at 20 mg/kg starting two weeks after transplantation (n = 8–12). H, Spleen weight 8 weeks after the beginning of furosemide treatment. I, Survival curve after furosemide treatment (n = 8–12). J, Blood chimerism (left) and hematocrit (right) after transplantation of JAK2^V617F BM cells followed by furosemide treatment that began at 10 weeks (wk 10), ceased at 25 weeks (wk 25), and began again at 29 weeks (wk 29; see arrows; n = 8–12). The crosses indicated nontreated mouse mortality. K, Spleen weight before furosemide treatment (10 weeks in J) and at the end of the sequential furosemide treatment (31 weeks in J). L, JAK2^V617F cells contribution to LT-HSCs in nonirradiated and irradiated legs before furosemide treatment (10 weeks in J) and at the end of the sequential furosemide treatment (31 weeks in J; n = 3–5). M, Effects of 5 and 20 μg/mL of furosemide on the percentages of BFU-E from CD34⁺ cells of patients with PV (n = 4). 100% is used as percentage of BFU-E in the absence of furosemide (N.T.). N, Relative percentage of JAK2^V617F VAF in the BFU-E after treatment with 5 and 20 μg/mL of furosemide. 100% is used as percentage of VAF in the BFU-E in the absence of furosemide (N.T.). Data are mean ± SEM. Significance was assessed using unpaired two-tailed t test and two-way ANOVA followed by post hoc analysis (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

DISCUSSION

We have developed an in vivo approach that allows examination of the temporal progress of a mouse model of PV and that can also be used for other mouse models of hematopoietic malignancy. Local irradiation of a single leg has been shown to alter the dynamics of HSC homeostasis by increasing their homing to the site of irradiation with almost no HSC homing to the nonirradiated leg but this approach has never been used to study myeloproliferative or leukemia development. Finally, in contrast to TBI, this approach does not lead to aplasia and thus can be used to study donor hematopoiesis or MPN development at earlier time points.

Increased contribution of JAK2^V617F LT-HSCs to BM of irradiated and non-irradiated legs was not associated with decreased contribution of endogenous LT-HSCs but rather with an increased total number of LT-HSCs suggesting an absence of competition between endogenous and mutant LT-HSCs and/or the use of different niches by mutant and endogenous LT-HSCs (19, 20). In contrast to LT-HSCs, the JAK2^V617F MEP compartment gradually increased at the expense of the endogenous MEP compartment and this increase was associated with a higher proportion of cycling JAK2^V617F MEP and of endogenous WT MEP and LSK in G₀. This increased number of resting, endogenous WT cells was only observed after JAK2^V617F BM cells transplantation and might be related to chronic inflammation promoted by JAK2^V617F mature hematopoietic cells (21, 22). Finally, given that the mature erythroid compartment contained only mutant cells as early as two weeks after transplantation whereas percentages of endogenous and mutant MEP were equal at this time point, suggesting that erythroid differentiation of endogenous MEP might be decreased due to low levels of EPO in PV mouse models (23) and/or JAK2^V617F MEP might have a dramatically increased capacity to differentiate (24).

Barcode experiments showed that mutant cells present in the nonirradiated leg derived from the irradiated leg but the frequency of JAK2^V617F LSK clones present in irradiated and nonirradiated legs was different, thus highlighting a different microenvironment and/or postmigration selection of clones that could expand in the nonirradiated leg. Migration of most JAK2^V617F LSK was associated with their differentiation into erythroid and myeloid cells in the nonirradiated leg whereas one third of JAK2^V617F LSK clones present in the irradiated leg did not differentiate into mature cells, again suggesting different dynamics of JAK2^V617F LSK clones in the irradiated and nonirradiated legs.

As expected, establishment of the PV phenotype was associated with splenomegaly. In contrast to observations in the legs, a high increase of mature mutant erythroid cells was found in the spleen and barcode experiments showed that one third of erythroid clones present in the spleen were not detected in either irradiated or nonirradiated leg. These results are in accordance with previous reports (25) and indicate that the spleen is necessary for the amplification of mutant erythroid cells in this PV mouse model. Indeed, splenectomy one or two weeks after transplantation normalized red blood cells parameters. However, in the bones of the splenectomized mice, hematopoiesis mainly derived from JAK2^V617F LT-HSCs and their differentiating or mature progeny indicating these mice had a JAK2^V617F hematopoiesis despite the extinction of the PV phenotype.

At present, the JAK1/JAK2 inhibitor ruxolitinib has been approved for the treatment of patients with PV resistant/intolerant to hydroxyurea (5, 26), but this drug has no major effect on JAK2^V617F cell burden and some patients develop cytopenia (9) due to its nonselective effect on JAK2^V617F thus underscoring the need for new therapeutic targets in PV. In this study, genetic or pharmacologic inhibition of CAR1, the major CA isoform overexpressed in JAK2^V617F HSPC, resulted in an increased survival of mice associated with normalization of blood parameters of PV and of JAK2^V617F burden without any detrimental effect on normal hematopoiesis. In normal mature erythroid cells, CAR1 acts as a regulator of intracellular pH (27) and is a key protein for favoring O₂ intake in the lungs and its release in tissues. In contrast to many isoforms of carbonic anhydrase which play important role in cancer cells (24, 28, 29), little is known about functions of CAR1 in cancer. One possible role for CAR1 overexpression in JAK2^V617F PV might be to alter JAK2^V617F HSPC differentiation, as has been shown for enzymes that alter the cytosolic pH level to facilitate phosphorylation of proteins that, together with ncRNAs (30), regulate cellular differentiation.

Increased CA1 mRNA and CA1 protein levels were also found in CD34⁺ cells from PV patients with the JAK2^V617F mutation whereas mRNA levels of other CA isoforms were unchanged, with the exception of a slight increase of the CA2 mRNA level. Our study demonstrate that the CAR1 inhibitor furosemide not only impairs PV development and targets JAK2^V617F LT-HSCs in mice but also inhibits PV patients’ BFU-E from their CD34⁺ cells, highlighting CAR1 as a potent and specific therapeutic target to treat human PV.

METHODS

Mice

Conditional floxed Jak2^V617F knock-in (KI) mice (23) were crossed with VavCre transgenic (TG) mice (31) to express the mutant Jak2^V617F gene in HSCs. To quantify allele burden in kinetic experiments, Jak2^V617F KI::VavCre mice were crossed with Ubi-GFP TG mice in the Gustave Roussy animal facilities. All these mice are under C57BL/6 (CD45.2) genetic background. For one-leg transplantation, F1 crossed C57BL/6.SJL/BoyJ (CD45.1) × C57BL/6J (CD45.2) mice were used as recipient mice. All mice were housed in a specific pathogen-free section at an animal facility of CEA/DRF/IBFJ/iRCM (registration number A920322) according to the institutional guidelines and French Ministry of Agriculture's regulations. Animal experiment studies were approved by the French Government (APAFIS#20683) after reviewing by the institutional ethics committee.

One Leg Irradiation and Transplantation

Recipient mice were anesthetized with ketamine–xylazine and their right leg was fixed on a box with the left leg toward the opposite side of their body as shown in Supplementary Fig. S1. The box was put in the radiation field of a GSRD1 radiation equipment. The mouse body was shielded from irradiation with lead plates and the leg was irradiated 2 × 5 Gy with 17-hour rest. BM cells (10⁷) were intravenously transplanted into the irradiated mice 7 hours after the second irradiation. For barcode and shRNA experiments, 10⁵ transduced Lin⁻Sca1⁺cKit⁺ (LSK) hematopoietic cells were transplanted after isoflurane anesthesia.

Blood Analysis

Blood was collected and analyzed using a hemocytometer (CD3700, ABBOTT). The remaining blood was hemolyzed using an ammonium chloride solution (STEMCELL). The hemolyzed blood samples were stained with antibodies against CD45.1 (A20) and CD45.2 (104) and analyzed using flow cytometry to characterize blood chimerism.

Flow Cytometry Analysis

Recipient mice were sacrificed at the indicated time. BM cells (femur and tibia) of irradiated leg and of nonirradiated leg were harvested by flushing with 1 mL of PBS. After centrifugation, BM supernatants were collected and kept at −80°C with antiprotease (ROCHE) for chemokine analyses. BM cells were suspended in PBS; splenocytes or liver cells were also collected in PBS by crushing spleen or liver on a 70-μm filter, respectively. Before hemolysis, cells were stained with Ter119 (TER-119) and CD71 (C2) antibodies to characterize the erythroid compartment. The remaining cells were hemolyzed and stained for analyses of granulocytes, hematopoietic progenitor cells (HPC), and HSCs by using antibodies against CD45.1 (A20), CD45.2 (104), CD11b (M1/70), Gr-1 (RB6–8C5), c-Kit (2B6), Sca-1 (D7 or E13–161.7), CD34 (RAM34 or MEC14.7), CD16/32 (93), CD48 (HM48–1), and CD150 (TC15–12F12.2). For HPCs and HSCs, cells were stained with Lineage Cell Detection Cocktail-Biotin (Miltenyi Biotec) revealed with PE-CF594 streptavidin antibody. For cell-cycle analyses, stained HSPCs were permeabilized with Cytofix/Cytoperm buffer following manufacturer's instructions (BD Biosciences) and labeled with anti-Ki67 antibody (SolA15) and Hoechst 33342. All antibodies were purchased from BioLegend, eBioscience, or BD Biosciences. Flow cytometry analyses were performed through FACSCanto II (BD Biosciences) and LSR II (BD Biosciences), and cell sorting was conducted by using FACS Aria II (BD Biosciences). Analyses were performed with the FlowJo software (TRISTAR) and illustrated with the Prism7 software (GraphPad).

Mouse Colony Assay

One-hundred microliters of blood from recipient mice was collected and hemolyzed using an ammonium chloride solution (STEMCELL Technologies). The hemolyzed cells were plated in 3 mL of MethoCult GF M3434 (STEMCELL Technologies) on 35-mm dishes. The number of total and donor GFP⁺ colonies were counted at 14 days on an EVOS FL Auto Microscope (Thermo Fisher).

15,000 mouse Lineage negative (Lin⁻) cells BM cells from mice transplanted with JAK2^V617F (GFP⁺) BM cells were transduced for 72 hours with shRNA CTL or shRNAs targeting Car1 and seeded in triplicate in 35-mm petri dishes in Methocult M3434 (STEMCELL Technologies). After 8 days at 37°C, the number of total and GFP⁺ BFU-E colonies were scored using an EVOS FL Auto Imaging System (Thermo Fisher).

Patient Cell Purification and Colony Assay

Nine male and female patients (ages 40–90 years) were selected on PV diagnosis according to the 2016 iteration of the World Health Organization classification. Peripheral blood samples were collected from these patients with their written informed consent in accordance with the Declaration of Helsinki, and the study was approved by the Ethics Committee from Centre Hospitalier Universitaire (CHU) Dijon, Saint Louis Hospital, and Gustave Roussy; from Comite de Protection des Personnes Ile de France IV Institutional Review Board (IRB; agreement from the US Department of Health and Human Services IRB 00003835-protocol 2015/59-NICB) and Commission Nationale de l'Informatique et des Libertes (authorization 915663). Control donor samples were obtained from donors after leukapheresis. Patient blood samples were collected between 2015 and 2021. Mononuclear cells were isolated by a Ficoll density gradient and CD34⁺ cells were purified by a double-positive magnetic cell sorting system (AutoMACS; Miltenyi Biotec) and stored at −140°C in DMSO.

Colony assay was performed on purified CD34⁺ cells from patients with JAK2^V617F PV. 1,000 human CD34⁺ cell preincubated or not for 1 hour 30 minutes with furosemide (5 and 20 μg/mL) in H5100 medium supplemented with 1% penicillin/streptomycin were seeded in triplicate in 35-mm petri dishes in Methocult H4435 (STEMCELL Technologies) supplemented or not with furosemide. Total and BFU-E colonies were scored after 12 to 20 days at 37°C. Individual colonies corresponding to the progeny of each progenitor were plucked using PBS 1×, lysed with proteinase K and DNA was genotyped for homozygous, heterozygous, or WT status by Taqman allelic discrimination on the ABI Prism GeneAmp 7500 Sequence Detection System (Applied Biosystem, Thermo Fisher Scientific) as previously described (3).

Chemokines Assay

Chemokines from BM supernatant of irradiated and nonirradiated legs of recipient mice were analyzed with a proteome profiler mouse XL cytokine array following manufacturer's instructions (BIOTECHNE). Signal quantification was done with ImageQuant TL software (GE Healthcare).

Barcode Experiment

10⁵ Donor WT or JAK2^V617F LSK sorted cells were spin occulated for 1 hour 30 minutes followed by 3 hours 30 minutes at 37°C with 10 μL of a vector library of 2,500 barcodes in a StemSpan SFEM medium (STEMCELL Technologies) with mSCF (50 ng/mL; R&D Systems) to obtain a multiplicity of infection (MOI) of 1. After removal of the virus medium, 1 × 10⁵ transduced LSK cells were transplanted into the irradiated leg of recipient mice. The transduction efficiency was in a range of 15% to 20%. After developing PV disease symptoms, mice were sacrificed and BM cells were separately collected from the irradiated leg and the nonirradiated leg, stained, and the donor-derived GFP⁺ cells were sorted from the indicated cell fractions. Genome DNA was extracted by lysing the cells with Proteinase K solution (Invitrogen) at 55°C for 2 hours. After the lysis, the DNA samples were split into technical duplicates. Amplification reaction, sequencing and barcode classification were performed at Curie Institute as described in Perié and colleagues (12).

Using a customized script in R (12), sample data were filtered (quality control and repeat use of barcodes, see extended data Fig. 2), normalized (to 10⁵ reads), and classified by tagging each barcode. Read counts were transformed using the hyperbolic arcsine function that is similar to a logarithmic function but can accommodate barcode reads with a value of 0 as described (12). Resulting data have been graphed with a customized script in R from the R Graph Gallery.

Splenectomy and Secondary Transplantation

Recipient mice were irradiated on one leg, intravenously transplanted with 10⁷ donor WT or JAK2^V617F BM cells and splenectomized 1 or 2 week(s) after grafting by surgical intervention under isoflurane anesthesia. For secondary transplantation experiments, BM cells were separately harvested from the irradiated legs and nonirradiated legs of recipient mice 5 weeks after transplantation, then hemolyzed and transplanted at 10⁷ into secondary nonconditioned immunocompetent mice.

Hematoxylin and Eosin Coloration and Anti-GFP Immunostaining

Spleens were collected at indicated time and fixed in 4% PFA for 24 hours at 4°C, dehydrated in alcohol, parrafin-embedded with Tissue-Tek VIP 6 AI (Vacuum Infiltration Processor, Sakura Finetek), and then cut into 5-μm sections and stained with hematoxylin and eosin (H&E). For anti-GFP staining, tissue sections were dewaxed then rehydrated and boiled in HIER citrate buffer pH 6 (Zytomed, Diagnomics) in an autoclave (Retriever 21000, Proteogenix) for unmasked and stained with anti-GFP antibody (1:200, clone JL-8, #632381, Clontech (R)) following manufacturer's instructions (“Mouse-on-Mouse” detection kit, MOM, BMK-2202, Vector Laboratories). Images were acquired on Olympus BX51 Microscope Lightpath (Olympus).

Transcriptome and qRT-PCR Analysis

10⁵ Donor JAK2^V617F LSK–sorted cells were transplanted into the irradiated leg of recipient mice. After developing PV disease symptoms, JAK2^V617F HSPCs (ST-HSC and LT-HSCs) cells from irradiated and nonirradiated leg BM were sorted and lysed with RLT Plus Buffer (QIAGEN). Total RNA was extracted according to the protocol of RNeasy Plus Micro Kit (QIAGEN). The RNA integrity (RNA Integrity Score ≥7.0) was checked on the Agilent 2100 Bioanalyzer (Agilent) and quantity was determined using Qubit (Invitrogen). For transcriptome analysis, total RNA samples were prepared with the GeneChip Whole-Transcript (WT) Pico Reagent Kit and analyzed with the GeneChip Whole-Transcript Expression array using Clariom S assay following manufacturer's instructions (Thermo Fisher). Data were collected and analyzed with the TAC software recommended by the manufacturer (Thermo Fisher). For quantitative RT-PCR analysis, total RNA was extracted according to the protocol of RNeasy Plus Micro Kit (QIAGEN) and cDNAs were synthesized by the reverse transcription using SuperScript IV First-Strand Synthesis System (Invitrogen). qRT-PCR was performed by using StepOnePlus Real-Time PCR System with TaqMan Fast Advanced Master Mix (Applied Biosystems). Primers used for detection of target CA1 gene (Car1-Mm00486717_m1) were purchased from Applied Biosystems. The gene expression level was calculated on the basis of the 2^–ΔΔC_t method.

Car1 Genetic Inhibition

To perform CAR1 inhibition experiments, shRNA sequences directed against MmCar1 gene (see these sequences below) were cloned in pTRIP-MND-Cherry-H1 plasmid (from CEA/iRCM/CIGEx). Briefly, shRNA sequences were generated by annealing a complementary pair of primers harboring homology sequence to pTRIP-MND-Cherry-H1 plasmid digested by BamHI for further cloning by complementary single-strand annealing–based method (32). Sh CTL sequence was directed against hepatitis C virus (HCV, GTGTTGGGTCGCGAAAGG). All enzymes were from New England Biolabs, antibiotics from Sigma-Aldrich, and primers from Eurofins Genomics (Supplementary Table S1). All constructs were validated by DNA sequencing. 10⁵ Donor WT or JAK2^V617F LSK–sorted cells were transduced with the 3 Car1 shRNAs or Ctl shRNA in BIT 9500 serum with mSCF (100 ng/mL), mFLT3 (100 ng/mL), mIL3 (20 ng/mL), IL6 (10 ng/mL), and TPO (10 nmol/L) at 37°C for 24 hours and transplanted into the irradiated leg of recipient mice. More than 85% of the transduced LSK cells were detected positive for fluorescence expression of mCherry before grafting.

To perform rescue experiments, a mutant silent Car1 cDNA containing silent mutations that impair the effect of the 3 shRNAs was synthetized (bold letters indicate the mutated sequences)

sh1:

• Sequence targeted by sh1: AGAACAGTCAGAGCCTCATTT

• Mutated sequence: CGAACTGTCCGTGCTAGCTTC

• Protein sequence: R T V R A S F

These mutations create a NheI site.

sh2:

• Sequence targeted by sh2: CTGCTTCCTTCATCTCTGGATTAC

• Mutated sequence: CTTCTACCGAGCTCTTTGGACTAC

• Protein sequence: L L P S S L D Y

These mutations create a SacI site.

sh3:

• Sequence targeted by sh3: ACCGTGGATGGAACTAGATATTCT

• Mutated sequence: ACTGTCGACGGCACTCGATACTCT

• Protein sequence: T V D G T R Y S

These mutations create a SalI site.

Silent mutant Car1 cDNA was cloned in pTRIP-dU3-EF1-MND-ECFP plasmid (from CEA/iRCM/CIGEx) and sequenced. JAK2^V617F Lin⁻cKit⁺ cells from single-leg-irradiated mice (n = 9) transplanted with JAK2^V617F BM cells were sorted and cotransduced with a lentivirus expressing control sh (shCtl) or sh1 or sh2 or sh3 that target Car1 mRNA and a lentivirus expressing a WT CAR1 protein (silent mutant CAR1 as SmCAR1) from silent mutant Car1 cDNA in StemSpan medium supplemented with mSCF (100 ng/mL), mFLT3 (100 ng/mL), and IL11 (10 ng/mL) at 37°C for 24 hours. These transduced were maintained with an additional 24 hours in the same medium and cultured for 3 days in erythroid medium conditions (StemSpan with mSCF, 100 ng/mL) and EPO (1 U/mL). The different hematopoietic populations were then analyzed by FACS.

Pharmacologic Treatment

Furosemide (MKCF8657) and acetazolamide (A6011) was purchased from Sigma. Furosemide was dissolved in DMSO for stock solution (50 mg/mL) and diluted in PBS solution prior to use. At indicated time, recipient mice were daily treated by intravenous injection of furosemide at 5 or 20 mg/kg or of acetazolamide (50 mg/kg). Spironolactone (Sigma) was prepared in water bottles at 250 mg/kg to treat recipient mice as described previously (33).

Whole-Transcriptome RNA-seq of CD34⁺ Cells from Patients

The patients were selected for a high JAK2^V617F variant allele frequency (60–100% VAF). Blood samples from patients were obtained from Gustave Roussy (Villejuif, France), Saint Louis Hospital (Paris, France), and Saint Antoine Hospital (Paris, France). Control donor samples were obtained from donors after leukapheresis. Peripheral blood from patients was collected in EDTA tubes. Hematopoietic progenitors (CD34⁺) of 4 patients with PV and of 4 control donors were isolated from mononuclear cells (MNC) by immunomagnetic enrichment (Miltenyi Biotec) and were amplified for 5 days in serum-free medium in the presence of a cocktail of human recombinant cytokines containing EPO (1 U/mL; Amgen), TPO (20 ng/mL; Kirin), SCF (25 ng/mL; Biovitrum AB), IL3 (10 ng/mL), FLT3-L (10 ng/mL), G-CSF (20 ng/mL), and IL6 (100 U/mL; Miltenyi Biotec). Total RNA was isolated using the RNeasy Mini Kit (Qiagen). The RNA integrity (RNA Integrity Score ≥7.0) was checked on an Agilent 2100 Bioanalyzer (Agilent) and RNA quantity was determined using Qubit (Invitrogen). SureSelect Automated Strand Specific RNA Library Preparation Kit was used according to manufacturer's instructions with the Bravo Platform. Briefly, 50 to 200 ng of total RNA sample was used for poly-A mRNA selection using oligo(dT) beads and subjected to thermal mRNA fragmentation. The fragmented mRNA samples were subjected to cDNA synthesis and were further converted into double-stranded DNA using the reagents supplied in the kit, and the resulting dsDNA was used for library preparation. The final libraries were bar-coded, purified, pooled together in equal concentrations, and subjected to paired-end sequencing on the Novaseq-6000 sequencer (Illumina) at Gustave Roussy.

mRNA Expression Analysis of CD34⁺ Cells from Patients

Total RNA of CD34⁺ cells from 4 to 9 patients was isolated using the RNeasy Mini Kit (Qiagen) and cDNA was synthesized by SuperScript IV Reverse Transcriptase (Invitrogen). PCR were carried out in the ABI Prism GeneAmp 7500 Sequence Detection System (Applied Biosystems, Invitrogen), using TaqMan Fast Advanced Master Mix (Applied Biosystems) and with specific primers (CA1-Hs00266139_m1, Thermo Fisher Scientific). HPRT was used as housekeeping gene.

Automated Capillary Immunoassay (WES)

Automated capillary immunoassay (Simple Western) was performed on a Western immunoassay (WES) system (Protein Simple). Mouse HSPC or Lin⁻ cells and human CD34⁺ cells were incubated with a RIPA buffer for 30 minutes, centrifuged for 15 minutes at 14,000 × g, and protein supernatants were stored at −80°C. The protein equivalent of 5,000 cells was used to carry out WES experiments. Car1 (NBP1, BIOTECHNE) and GAPDH (Cell Signaling Technology) antibodies were, respectively, used at a 1:50 and 1:100 dilution for both mouse and human proteins. The analyses were performed on a 12–230 kDa separation module (SM-W004) according to the manufacturer's instructions (Protein Simple).

Statistical Analyses

Experiments were performed on 8 to 12 recipient mice except when indicated and on 4 to 9 patients with PV. Data are provided as mean ± SEM. All data were tested for significance using unpaired two-tailed t test and two-way ANOVA, following by post hoc analysis. Survival rate was investigated by Kaplan–Meier plot and tested with log-rank test statistics. In all tests, P values of less than 0.05 were considered statistically significant (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

Data Availability

All data (Figs. 1–4; Supplementary Figs. S1–S5) are available in the Article or from the corresponding authors on reasonable request. Raw data transcriptome files are available on GEO repository with the accession number GSE165902 and raw data RNA-seq files are available on EGA repository with the accession number EGAD00001007531.

Authors’ Disclosures

S. Murakami reports a patent for EP21165307 pending. V. Barroca reports a patent for EP21165307 pending. P. Romeo reports a patent for EP21165307 pending. D. Lewandowski reports a patent for EP21165307 pending. No disclosures were reported by the other authors.

Authors’ Contributions

S. Murakami: Resources, software, formal analysis, funding acquisition, validation, investigation. V. Barroca: Resources, data curation, validation. L. Perie: Data curation, formal analysis. A. Bravard: Formal analysis. J. Bernardino-Sgherri: Formal analysis. A. Tisserand: Resources, data curation, formal analysis. C. Devanand: Resources, data curation. V. Edmond: Resources, data curation. A. Magniez: Resources, data curation. S. Tenreira Bento: Resources, data curation. C. Torres: Resources, data curation. F. Pasquier: Resources. I. Plo: Resources. W. Vainchenker: Resources. J. Villeval: Resources, funding acquisition. P. Romeo: Conceptualization, supervision, funding acquisition, writing–original draft, writing–review and editing. D. Lewandowski: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.