Human studies
All centres had appropriate research and ethical approval and patients gave their written informed consent. Patients’ samples for histological studies were obtained from patients with myeloproliferative neoplasms (MPN). Some samples were derived from patients attending outpatient clinics at Addenbrooke’s Hospital in the UK, under the clauses of the ‘Causes of Clonal Haematological Disorders Project’ which had regional ethical approval from the Eastern Multi-region Ethics Committee (MREC 02/5/22 and 07/MRE05/44) and local research and ethical approval at participating UK hospitals. Other samples were derived from patients enrolled in the MAJIC Phase-2 clinical study (ISRCTN61925716), which is a randomized trial of ruxolitinib (JAK1/2 inhibitor) vs best available therapy (BAT) in ET and PV patients resistant or intolerant to hydroxycarbamide25. Samples were collected from patients before or 12 months after treatment. For xenotransplantation, bone marrow (BM) or peripheral blood (PB) was collected from three ET and three PV patients after obtaining informed consent. The study was approved by the local ethics committee (KEK-ZH-NR: 2009-0062/1 and BASEC-NR: 2018-00539).
Mouse strains
Mice were housed in specific pathogen free facilities. All experiments using mice followed protocols approved by the Animal Welfare Ethical Committee (AWERB), according to United Kingdom Home Office regulations (PPL P0242B783). Nes-gfp82, FVB/N-Adrb3tm1Lowl/J (JAX stock number 006402), B6.129S(Cg)-Stat1tm1Dlv/J (JAX stock number 012606), B6.129S6-Stat5btm1Mam Stat5atm2Mam/Mmjax (JAX stock number 032053), B6.FVB-Tg(Acta2-DsRed)1Rkl/J (JAX stock number 031159) (Jackson Laboratories), Vwf-TdTomato55, Vav-Cre;JAK2-V617F41, Mx1-Cre;JAK2-V617F41, JAK2:V617F45, Scl-tTA;JAK2-V617F43, MISTRG mice47 and congenic B6.SJL-Ptprca Pepcb/BoyJ (CD45.1), CD45.2 C57BL/6 mice (Charles River Laboratories), were used in this study. Vav-Cre;JAK2-V617F mice were used as ET-like model. Vav-Cre;JAK2-V617F mice express active Cre in fetal and adult HSCs. Mx1-Cre;JAK2-V617F41, JAK2R/R 45 and Scl-tTA;JAK2-V617F43 mice were used as independent PV-like models. In double transgenic Mx1-cre;JAK2-V617F mice, JAK2(V617F) mutation can be conditionally expressed in haematopoietic cells upon Myxovirus resistance-1 (Mx1)-driven Cre recombinase activation by polyinosinepolycytosine (pIpC)41. Cre activation was induced through intraperitoneal (i.p.) injection of pIpC (Sigma, P1530). In Scl-tTA;JAK2-V617Fmice, JAK2V617F expression was induced by intraperitoneal injection of 143 mg/kg tamoxifen (i.p., 3 times on alternate days)43. ET-like and PV-like models and WT mice transplanted with BM cells from MPN mice develop progressive symptoms of MPN. Animals were sacrificed and analyzed at 13-20 weeks of age when they evidenced signs of MPN disease in peripheral blood, which was monitored over time using an automated blood counter (Horiba, UK).
BM transplantation
Age-matched, CD45.1 or CD45.2 C57BL/6J mice (8-12-week-old) were used as recipients in BM transplantation assays. Recipients were subjected to lethal irradiation (12 Gy whole body irradiation, split dose 6.0 + 6.0 Gy, 3h apart) before injection. For competitive transplantation, 106 BM cells from a Mx1-cre;JAK2-V617F;DsRed+ CD45.2 donor and 106 BM cells from a Vav-Cre;JAK2-V617F CD45.2 donor were injected in the tail vein of CD45.1 mice. For non-competitive transplant, 2x106 BM cells were injected.
In vivo treatments
CASIN (Tocris Biosciences, cat. no. 5050) powder was resuspended in DMSO at 100mM and stored at -20°C. The drug was thawed and diluted in PBS before injection to the mice. Ruxolitinib (Jakavi, Novartis) was provided by Dr. Claire Harrison. One ruxolitinib pill (20mg) was crushed using a mortar and pestle with 3ml of Poly(ethylene)glycol (Aldrich – ref. 202371 – 500g) and 1ml of PBS (Sigma), warmed for 20min at 60°C in a water bath, and sonicated until total dissolution (3 rounds of sonication, each time 12cycles of 30s) and kept at 4°C for no more than 1w before given to the mice, and briefly rewarmed at 37°C before use. For the chronic treatment of MPN mice with ruxolitinib or CASIN, the cohort was irradiated as described above and transplanted with 2x106 of Vav-Cre;JAK2-V617F or Mx1-cre;JAK2-V617F BM cells. The mice were randomized depending on the blood counts 1 month later and treated with CASIN (5mg/kg, i.p.), ruxolitinib (70mg/kg, o.d.) or vehicle every other day for 5w, until sacrificed. For intravital imaging, 8-12-week-old Nes-GFP recipient mice were irradiated and transplanted with 1,000 to 5,000 Lin-CD150+CD48- HSCs isolated from b-actin-DsRed donors (B6.FVB-Tg(Acta2-DsRed)1Rkl/J mice intercrossed with Vav-Cre;JAK2-V617F or Mx1-cre;JAK2-V617F mice). Recipient mice were treated with CASIN (10mg/kg, i.p.), ruxolitinib (70mg/kg, o.d.) or vehicle 24h, 48h and 72h after transplantation. The last treatment was performed 2h before the beginning of the surgery for intravital imaging.
Intravital microscopy
Intravital microscopy was performed as described before83 using a Zeiss LSM 780 upright confocal microscope with a motorized stage and the following lasers: Argon, 561, 633 and a tunable infrared multiphoton laser (Spectraphysics Mai Tai DeepSee 690-1040). Signal was visualized with a W Plan-Apochromat ×20 DIC water immersion lens (1.0 N.A). Anesthesia was induced in mice with 4% isoflurane mixed with pure oxygen. This was gradually reduced to approximately 1% as anesthesia stabilized. To ensure steady positioning of mice on the microscope, surgery to attach the headpiece/imaging window was then performed as described 84. Large three-dimensional ‘tile scans’ of the entire BM cavity space were acquired by stitching adjacent, high-resolution z-stack images. Blood vessels were highlighted by i.v. injection of 50 μl of 8 mg/ml 500 kDa Cy5-Dextran (Nanocs, MA). For repeated imaging, protective intrasite gel (Smith & Nephew) was applied to the imaging window to preserve the bone integrity and prevent scar formation. The window was bandaged, and mice were allowed to recover from anesthesia. Owing to the lock-and-key mechanism of the imaging window84 mice could then be re-anaesthetized and accurately repositioned on the microscope stage and the same BM areas re-imaged. After each imaging session, analgesia was administered via oral buprenorphine in raspberry jelly at a dose of approximately 0.8 mg/kg.
Imaging sessions were performed on day 1 and day 3 after injection of the cells on day 0. Each time-lapse session lasted 1 h and positions of the cells were recorded once every 3 minutes to enable a reliable tracking of the cells’ movement. Altogether, 21 images were collected during each imaging session.
Stills from intravital imaging were analyzed with Volocity software (PerkinElmer) to measure in 3D space the distances between the DsRed labelled cells and the bone surface. Minimal distance module within Volocity software was used. To analyze and visualize the movement of the DsRed cells, manual tracking of the cells positions in 3D space was performed using Imaris software (release 9.0, Oxford Instruments). Positions XYZ for every cell were analyzed as described84.
Xenograft
Primary human HSPCs (CD34+) cells were purified from the BM or PB of patients using Ficoll density gradient centrifugation and further magnetically isolated using the MACS CD34 MicroBead Kit (Miltenyi Biotec). Patient CD34+ cells were cryopreserved and slowly thawed in IMDM 50% FCS at 37°C before xenotransplantation and re-suspended in 25µl PBS for injection. 8–12-week-old MISTRG mice85 were sub-lethally irradiated (181cGy using an X-ray RS-2000 irradiator, Rad Source) and transplanted with a 22-gauge needle (Hamilton Company) intra-femorally with 1.5-3x105 CD34+ hHSPCs previously purified from PB or BM and cryopreserved. Mice were maintained at the University Hospital Zurich animal facility in accordance with the Swiss Federal Veterinary office. Animal experiments were approved by the cantonal veterinary office of Zurich, Switzerland.
Immunohistochemistry of human BM trephines
Conventional immunohistochemistry protocol was performed on paraffin sections. Briefly, the sections were deparaffinized in xylene followed by progressive re-hydration with decreasing concentration of ethanol. Antigen retrieval was performed by proteolytic enzyme digestion using Tris-EDTA Buffer (10 mM Tris; 1 mM EDTA, 20mg/ml Proteinase K, ph8). The endogenous peroxidase was quenched using H2O2 30% in TBS 1X, before being blocked with 10% goat serum/10mg/ml BSA/0.1% Triton X-100/TBS. The sections were incubated with avidin/biotin (Vector) and then with an anti-CD34 antibody (Thermofisher, cat. no. MA1-10202), and then with Horseradish Peroxidase (HRP) coupled 2ary antibody (goat anti-mouse Fab2-IgG, Jackson, cat. no. 115-035-006). The samples were blocked with ABC kit (Vectra) and incubated with the substrate of peroxidase, 3,3’Diaminobenzidine (DAB, SIGMAFAST, Sigma-Aldrich). The sections were stained in hematoxylin, rinsed in water and dipped in 1% acid alcohol, and mounted with aqueous mounting medium (Vector). Slides were imaged using a slide scanning microscope (Zeiss Axioscan) and analyzed manually for the measure of the distance, areas, and quantifications of cells and vessels using the NDP2.view software (Hamamatsu).
Histology of mouse bones
Femurs and tibia were harvested just after sacrifice, cleaned and put in PBS, 2% PFA (Sigma Aldrich) overnight. For cryosectioning and immunostaining, the bones were washed once with PBS, decalcified in 250mM EDTA/PBS for 1.5 weeks at 4°C, put in 15% sucrose/PBS for 24h and then in 30% sucrose/PBS for another 24h, and embedded with OCT (Fisher Scientific Ltd – 12678646) in plastic cryomolds (Sakura, 4557), using chilled methyl-butane (Sigma-Aldrich) for snap freezing. The samples were stored at -80°C and sections (12µm) or whole mount were obtained using a cryostat (Leica Biosystem). Alternatively, after fixation the bones were embedded in paraffin and 5µm-thick sections were stained with hematoxylin and eosin (HE) and Masson's trichome for conventional morphological evaluation.
Immunofluorescence of cryosections
Immunofluorescence staining of cryosections or whole mounts was performed as previously described86 using femurs, with minor modifications. Briefly, the samples were permeabilised with 0.1% Triton X-100 (Sigma) in TNB buffer (0.1 M Tris–HCl, pH 7.5, 0.15 M NaCl, 0.5% blocking reagent, Perkin Elmer) at 4°C. Samples were incubated with primary antibodies: goat anti-CD31 (R&D, AF3628 1:100), rat anti-EMCN (Insight Biotechnology, sc-65495 1:100), rabbit anti-Sp7 (Abcam, ab22552, 1:200), rat anti-CD31 (BD Biosciences, Clone MEC13.3; 1:200), goat-anti-OPN (R&D, AF808, 1:100), goat anti-CD117 (R&D, AF1356, 1:200), rabbit anti-Ki67 (Abcam, ab15580, 1:100) diluted in 0.1% Triton X-100 TNB for 3days on horizontal shaking (whole mount) or O/N (cryosection) at 4°C. Samples were rinsed with PBS 4-5times for 24h (whole mount) or 10min (cryosection) and incubated for 24h (whole mount) or 1h (cryosection) with secondary antibody diluted at 1:300 in TNB : donkey anti-goat AF488 (Thermo Fisher, A11055), donkey anti-rabbit AF488 (Thermo Fisher, A21206), donkey anti-rat AF488 (Thermo Fisher, A21208), donkey anti-goat AF546 (Thermo Fisher, A11056), donkey anti-rabbit AF546 (Thermo Fisher, A10040), donkey anti-rat AF555 (Thermo Fisher, A21434), donkey anti-goat AF647 (Thermo Fisher, A21447), donkey anti-rabbit AF647 (Thermo Fisher, A31573), donkey anti-rat Dylight650 (Thermo Fisher SA5-10029). Repetitive washes were performed with PBS for 1day (whole mount) or 5min twice (cyosection). Stained tissue sections were counterstained for 10 min with 5mM DAPI in PBS and rinsed with PBS. For sections, the slides were mounted in mounting medium (DAKO, cat. no. S3023). Images were acquired with a confocal microscope either Leica SP5, SP8, Stellaris or ZEISS 980 using 10x, 20x and 40x objectives and analysed with ImageJ. At least 2 independent and randomly selected BM areas in the diaphysis were imaged and analysed per sample. Arterioles (small calliper, CD31hiEMCN- vessel) number was counted and normalized to the BM area. For the quantification of sinusoid diameter, the diameter of 20 random sinusoids was measured and the average was calculated for each sample. The quantification of different cell populations frequency using DAPI, CD117, KI67 and VWF staining was done using ImageJ and Cell Profiler software. The overall number of nuclei in a defined area was obtained by quantifying the number of DAPI+ objects with Cell Profiler. The same strategy was applied to define the number of cells of different cell populations. The endosteal area of the sample was defined as the area in the 150µm lining the bone, the central marrow was defined by the marrow area localized at more than 150µm away from the bone. The quantification of the Sp7+ or osteopontin+ area over the total bone marrow area was done on ImageJ.
Osteoclast resorption assay
Murine femurs and tibia from Vav-Cre;JAK2-V617F, Mx1-cre;JAK2-V617F or control mice were cleaned and the marrow wash flushed with PBS/2% FCS. The ivory disks were placed in 96-well plates and the surrounding wells were filled with PBS. 50µl of alpha MEM (without red phenol) supplemented with 10% FCS, 1% penicillin&streptomycin was added onto the disk. The cells were resuspended in 50µl alpha MEM (without red phenol) supplemented with 10% FCS, 1% penicillin&streptomycin, and seeded onto the disk and cultured for 5d in a water-jacketed incubator at 37°C, 5% CO2. The slides were fixed with 4% paraformaldehyde for 10min at RT and the Acid Phosphatase Leukocyte kit (TRAP, from Sigma, cat. no. 387A) staining was usd following the manufacturer’s recommendations. The slides were imaged using bright field microscopy and the TRAP+ area was quantified with ImageJ.
RNA isolation and qPCR
RNA isolation was performed using Trizol Reagent (Sigma T9424) on mouse whole BM cells. Reverse transcription was performed using the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems 4368814), following the manufacturer’s recommendations. qPCR was performed using the PowerUp SYBR Green Master Mix (Applied Biosystems A25742) and ABI PRISM® 7900HT Sequence Detection System. The expression level of each gene was determined by using the absolute quantification standard curve method. All values were normalized with Gapdh as endogenous housekeeping gene.
The following primers were used:
Col1a1-Fw : TATTGCTGGACAACGTGGTG
Col1a1-Rv : ACCTTGTTTGCCAGGTTCAC
Gapdh-Fw: GCATGGCCTTCCGTGTTC
Gapdh-Rv: CTGCTTCACCACCTTCTTGAT
Immunofluorescence of sorted HSCs
Mx1-cre;JAK2-V617F mice were injected as described above with pIpC at 8 weeks old. At either 10-13weeks old or 25weeks old, Mx1-cre;JAK2-V617F, Vav-Cre;JAK2-V617F or Cre-negative control mice were sacrificed, BM cells were immunomagnetically depleted of hematopoietic lineage marker-expressing cells using biotin-conjugated lineage cocktail, the magnetic streptavidin-conjugated beads (BD, cat. no. 557812) and the cell separation magnet (BD Imag, cat. no. 552311), following the manufacturer’s recommendations. Lin- cells were stained as described below and Lin-Sca1+cKit+CD150+CD48-DAPI- HSCs were sorted. The day before, some chambered coverslips (Ibidi, cat. No. 81811) were coated with fibronectin (40µg/ml) (Sigma-Aldrich, F1141-1MG) in NaHCO3 overnight at 4°C, and washed with PBS. The sorted HSCs were seeded on the coverslip in IMDM (Termo Fisher, cat. No. 21056-023), supplemented with 10% FCS, 1% penicillin&streptomycin and treated with ruxolitinib (INCB018424, resuspended in DMSO, final concentration 1µM), CASIN (1µM, TOCRIS), AC-4-130 (5µM, STAT5 inhibitor, MedChemExpress), NSC 118218 phosphate (10µM, STAT1 inhibitor, MedChemExpress), BP-1-102 (5µM, STAT3 inhibitor, selleckchem) or DMSO, for 16h at 37°C, 5% CO2. The medium was carefully removed and the cells were fixed with BD Cytofix Fixation Buffer (BD Biosciences, cat. No. BD 554655) for 20 min at RT. The cells were washed gently with PBS and permeabilized with 0.2% Triton X-100 (Sigma, cat. no. 9002-93) in PBS for 20 minutes, washed and blocked for 1h with 20% donkey serum (Merck, cat. no. D9663) in PBS. The cells were incubated with the primary antibody (rabbit-a-mouse CDC42, 1:100, Merck, cat. no. 07-1466) in 5% donkey serum O/N at 4°C, and washed twice with PBS, before incubation with 2ndary antibody (donkey-a-rabbit AF647, 1:300, Thermo Fisher, cat. no. A-31573) for 1h at RT and washed twice with PBS. The cells were stained with DAPI for 2 min at RT, washed, and mounted using fluorescent mounting medium (DAKO, cat. no. S3023). Imaging was performed at 40x or 63x using the super resolution mode with the airyscan on a Zeiss 980 confocal microscope. To define the polarity, the distribution of CDC42 was analysed across the whole z-stack and the HSCs was considered polar when a clear asymmetric distribution of the protein was visible. A minimum of 20 HSCs per condition were analysed. The 3D reconstitution of the HSC was obtained using Imaris software.
µCT
Femurs were collected, and the attached soft tissue was removed thoroughly and fixed in 4% paraformaldehyde. The fixed femurs were scanned using a SkyScan 1174 scanner: 50 kV, 800 μA, 8.3-μm isometric voxel resolution and 0.7-degree rotation step. Images were analyzed using SkyScan CT Analyzer software version 1.9.3.0.
BM cell extraction, flow cytometry and cell sorting
Hematopoietic cell isolation from BM or PB was performed as previously described39. The marrow was flushed and the bones crushed with a mortar and pestle (Fisher, 10656405), in PBS, 2% FCS and filtered through a 40µm strainer (Greiner-Bio-one, cat. no. 542040). PB or BM cell suspension were depleted of red blood cells using ACK lysis buffer for 8min at 4°C (Biolegend, 420301), washed and numerated using trypan blue (Fisher, cat. no. 10593524). Cells were incubated with the appropriate dilution (2-5 mg/ml) of fluorescent antibody conjugates. DAPI (Sigma Aldrich, cat. no. D9542) or 7AAD (BioLegend, Cat. No. 420404) were added to discriminate dead cells. Samples were analysed with a LSR Fortessa flow cytometer (BD Biosciences) or sorted (FACSAria, BD Bioscience) equipped with FACSDiva Software (BD Biosciences). The following antibodies were used to detect the human hematopoietic cells in PDX model: PE-conjugated mouse-anti-human CD45 (BD, cat. No. 555483), APC-conjugated mouse-anti-human CD34 (BD, cat. No. 555824), PE-Cy7-conjugated mouse-anti-human CD38 (BD, cat. No. 560677). The following antibodies were used for the staining of mouse hematopoietic cells in different panels: biotin-conjugated lineage cocktail (BioLegend, cat. no. 133307), APC-Cy7-conjugated anti-Sca1 (BioLegend, cat. no. 108126), FITC-conjugated anti-CD117 (BioLegend, cat. no. 105805), BV711-conjugated anti-CD48 (BioLegend, cat. no. 103439), PE-Cy7-conjugated CD150 (BioLegend, cat. no. 115914), PE-conjugated streptavidin (BioLegend, cat. no. 405207), APC-Cy7-conjugated anti-CD117 (BioLegend, cat. no. 105826), BV421-conjugated anti-Sca1 (BioLegend, cat. no. 108128), BV605-conjugated anti-CD150 (BioLegend cat. no. 115927), PE-CY7-conjugated anti-CD117 (BioLegend, cat. no. 105814), APC-Cy7-conjugated anti-CD45.2 (Insight 25-0454-U100), BV605-conjugated anti-CD41 (BioLegend, cat. no. 133921), BV711-conjugated anti-CD16/CD32 (BioLegend cat. no. 101337), APC-conjugated anti-CD71 (Thermo Fisher, cat. no. 17-0711-80), BV605-conjugated anti-Ter119 (BioLegend, cat. no 116239), FITC-conjugated anti-CD34 (BD, cat. no. 553733), biotin-conjugated anti-CD3e (BD, cat. no BD 553060), PE-conjugated anti-Ly6G (BioLegend, cat. no 108408), BV421-conjugated anti-CD11b (BioLegend, cat. no 101235), BV510-conjugated streptavidin (BioLegend, cat. no. 405234), AF488-conjugated streptavidin (Invitrogen, cat. No. 405235). For the erythoprogenitor analysis, we followed the same protocol but didn’t perform the red blood cell lysis buffer.
For cell cycle analysis, the cells were stained as described above, permeabilized with the cytofix/perm kit (BD, cat. no. BDB554714), stained with the APC-conjugated anti-Ki67-antibody (BioLegend, cat. no. 652405) o/n at 4°C in perm/wash buffer (BD, cat.no. BD554723), stained with Hoechst 33342 (Thermo Fischer, cat. no. 62249) and analysed. Sub-G0 cells were excluded and the percentage of cells in the different phase was calculated among the G0, G1, S, G2 and M phases. For CDC42 staining by flow cytometry, the cells were stained for membrane panel as described above and permeabilized with the cytofix/perm kit (BD, cat. no. BDB554714), stained with the anti-CDC42 antibody (Merck, cat. no. 07-1466) in perm/wash buffer 5% donkey serum O/N at 4°C, washed and stained with donkey anti-rabbit AF647 (Thermo Fisher, A31573) in per/wash buffer, washed and analysed. For stromal cell analysis, the bone marrow was flushed from long bones and the remaining bones were crushed in PBS, and digested in collagenase (Stem Cell Technologies, cat. No. 07902) for 30 min at 37°C in a water bath with agitation. The enzyme was quenched by adding PBS with 2% FCS. Cell suspensions were filtered, pelleted and red blood cell lysis was performed as stated above. Samples were stained with the following antibodies: biotin-conjugated anti-Ter119 (BD, cat. no. 553672), biotin-conjugated anti-CD45 (BD, cat. no. 553078), PE-Cy7-conjugated anti-CD31 (BioLegend, cat. no. 102524), PE-conjugated anti-EMCN (Santa Cruz, cat. no. sc-665495 PE), APC-Cy7-conjugated anti-Sca1 (BioLegend, cat. no. 108126), BV421-conjugated rat-anti-mouse CD51 (BD cat. No. 740062), APC-conjugated anti-CD140a (BioLegend cat. no. 135908), BV710-conjugated streptavidin (BioLegend cat. no.405241).
Statistical analyses
Statistical analyses and graphics were carried out with Graph-Pad Prism software. Data sets were compared using different tests described in the legends. P-values less than 0.05 were considered statistically significant.