Animal Experiments
C57BL/6 mice were purchased from Harlan UK. TPM4−/− mice were generated as we have previously described16. All animal studies were approved by local research ethics committee (AWERB) and licensed under UK Home Office project license PPL 30/3445.
Mouse ex vivo heart-lung preparation
Mice aged at 8–12 weeks were sacrificed by a Schedule 1 process, by exposure to rising CO2. A tracheostomy was then performed, and lungs ventilated with room air or pure nitrogen via a rodent ventilator (Minivent type 845, Harvard Apparatus, USA). Respiratory rate was maintained at 200 breaths/min and tidal volume was 10 mL/kg (~ 250 µL). The inferior vena cava was then exposed by laparoscopy, and the chest opened by median sternotomy and fat tissue carefully removed.
A 21-gauge catheter was passed into the right ventricle and approx. 100 mL warm (36–38°C) perfusion buffer (Krebs-Hepes buffer (KHB): 140 mM NaCl, 3.6 mM KCl, 0.5 mM NaH2PO4,0.2 mM MgSO4, 1.5 mM CaCl2, 10 mM Hepes (pH 7.4), 2 mM NaHCO3) containing 10 U/mL heparin was perfused at 3–4 mL/min into the lung vasculature. Once the colour of mouse lung turned pale, suggesting most of blood in the lung circulation was flushed, the catheter was disconnected.
The left and right superior venae cavae, ascending aorta, inferior vena cava and descending aorta were ligated using 6 − 0 silk, and a small incision was made in the left ventricle for collection of the perfusate. A suspension of mouse MKs (usually labelled with a fluorescently tagged anti-CD41 Ab) was pumped into the right ventricle, whilst the lungs were either ventilated with air or N2 or without ventilation, and the flow rate was maintained at 0.35 mL/min by a SyringeOne Programmable Syringe Pump (Product SKU: NE-1000-ES). The perfusate was collected and re-pumped into the right ventricle after samples were collected for imaging and FACS analysis. The perfusate was recirculated through the pulmonary vasculature in this manner for a total of 18 passages.
After the final passage, 1 mL perfusion buffer was injected into the system to remove some of the remaining cells in the lung vasculature, and all perfusate collected for evaluation of platelet function. The mouse lung was immediately removed and fixed in 4% PFA/PBS. The volume of the lung was measured by fluid displacement.
Microfluidic chamber design, fabrication, and experimental protocol
A set of channels mimicking a physiological vascular system was constructed using standard PDMS approach. The design shows a branching structure such that as branching progressing, the channel diameter decreases by half. All channels are rectangular in cross-section and 10 µm deep, with the largest channels being 100 µm across, decreasing to the smallest channels which are 12.5 µm across. From each larger channel, 16 smaller channels branch off, allowing for maintenance of flow resistance due to the r4 power relationship between resistance and channel diameter. Fluid then flows from larger diameter channels to smaller diameter channels, and in reverse on the way out of the system. Cells are passed through the system, repeatedly. The system is scaled up, through multiplexing in parallel, to allow greater cell volumes to be used, as shown in Fig. 2c.
The mask and SU-8 master mold were fabricated by NuNano (Bristol, UK). The microfluidic channels were fabricated by soft lithography. The mixture of PDMS in a 10:1 ratio was poured over the SU-8 master mold after being degassed in a vacuum desiccator. The PDMS mixture was cured at 80°C for 2 h and incubated in the oven overnight. The PDMS mold was removed from the SU-8 master. Input and output holes were punched using a 0.5 mm OD biopsy puncher (Elveflow, Paris, France). Finally, the PDMS microchannels were irreversibly bonded to glass slides using oxygen plasma for 3 min in a plasma device (Diener Plasma Systems, Ebhausen, Germany).
A suspension of mouse MKs pre-labelled with CD41-PE was pumped into the microfluidic chamber at 0.30 mL/min flow rate. Suspensions were collected from the output hole and re-pumped into the microfluidic chamber, after 25 µL samples were taken for FACS analysis. Suspensions were then recirculated through the microfluidic chamber in this manner for a total of 18 passages.
Culture and Differentiation of Murine Megakaryocytes
Briefly, BM from C57BL/6 or TPM4−/− mice was isolated and dispersed prior to centrifugation at 200 x g for 10 minutes. BM was re-suspended in IMDM- Glutamax containing 1% penicillin/streptomycin and 2% serum replacement 1. Cells were cultured for 3 days in the presence of 20 ng/mL rSCF and a further 13 days in the presence of 10 ng/mL rTPO at 37°C and 5% CO2. From day 8 to day 16, cells were transmitted to fresh cell culture dishes every day to reduce megakaryocytes (MKs) to contact with fibroblast cells.
On day 16, MKs were harvested and enriched with a 1.5%-3% albumin gradient for one hour in cell incubator, following stained with either IgG-PE/ or FITC, or CD41-PE or -FITC for 3 hours and further with DNA dye Hoechst 33342 for 20 mins. Then MKs were washed and resuspended in 2 mL medium containing 10 U/mL heparin prior to use.
In vivo assessment of passage of intact mouse MKs through the pulmonary vasulature
C57BL/6 mice were anaesthesized by intraperitoneal injection of a 100 mg/kg ketamine/10 mg/mL xylazine mix. Mouse MKs were stained with CellTracker™ Red CMTPX dye and Hoescht33342 prior to injection into the right external jugular vein of an anaesthetized recipient mouse. Blood was collected from the left common carotid artery. Cells were imaged by confocal fluorescence microscopy after lysis and removal of red blood cells.
Flow cytometry
Platelets derived from IgG-PE stained MKs were set up as negative controls. Washed mouse platelets were used to set gates for generated platelets (P1) at the FSC/SSC density plot by size and granularity. 25 µL of CD41-PE-stained MK suspension, or perfusates from lung or suspensions from microfluidic chambers after 1, 2, 3, 6, 9, 12, 15 or 18 passages, were fixed in 2% PFA/PBS before FACS. DNA content, viability and mitochondrial membrane potential of generated platelets were determined by Draq5, Calcein AM and TMRM staining, respectively.
To detect the viability and mitochondrial membrane potential of pulmonary endothelial cells (ECs), assays for Calcein Deep Red retention in ECs and TMRM accumulation in active mitochondria were conducted. Pulmonary ECs were isolated from perfused lungs under air- or pure-N2-ventilation or unventilated for approximately 2 hours and marked with either anti-CD31/PECAM-1 or anti-CD102/ICAM-2 antibodies, followed by loading with Calcein Deep Red and TMRM dyes, respectively, for 25 mins at room temperature. Pulmonary ECs from fresh lungs served as control.
Platelet surface glycoproteins were measured by incubating with FITC-conjugated anti-mouse CD61, CD42b, CD42d, CD49b, Glycoprotein VI (GPVI) antibodies and isotype-nonspecific IgG for 20 mins at room temperature before fixation.
To investigate the function of generated platelets (platelets derived from CD41 stained MKs), assays for αIIbβ3 integrin activation (JON/A-binding) and P-selectin exposure were performed. Washed platelets were stimulated with 2 U/ml thrombin or 5 µg/mL CRP for 10 mins followed by incubation with PE-JON/A or PE-P selectin antibodies for 20 mins at room temperature before fixation. Tirofiban and PE-IgG were used as negative controls for the measurement of PE-conjugated JON/A or PE-conjugated P-selectin exposure, respectively.
Samples were analysed on a BD Accuri™ C6 Plus flow cytometer (BD) with 50,000 gated events/sample.
Two Photon Imaging and counting of generated platelets in the lung vasculature
After overnight fixation at 4°C, the lung was staged on a microslicer device (Type: DTK-1000N, UK) and cut into small slices with 800 µm thickness and flat surface. Lung slices from different lobes were fixed into a 100 mm cell culture dish and immersed in PBS for two photon imaging.
Imaging was performed using a DeepSee multiphoton laser (Spectra Physics) attached to an upright SP8 confocal microscope (Leica Microsystems). All images were collected using a 25x/0.95NA water dipping lens. For CD41-FITC or CD41-PE, excitation was provided by tuning the multiphoton laser to 927nm, for Hoechst33342, excitation was provided by tuning the multiphoton laser to 750 nm. The resultant fluorescence for both scans passing off a SP500 dichroic beam splitter and through both a SP680 filter and either a 525/50 nm bandpass filter to selected only CD41-FITC signal, a 630/75 nm bandpass filter to select only CD41-PE signal or a 460/50 nm bandpass to select only Hoechst33342 signal. CD41-FITC or CD41-PE fluorescence was detected using non-descanned Hybrid detectors (Leica Microsystems) and Hoechst33342 fluorescence was detected using a non-descanned PMT.
Images were acquired with an additional zoom of 2.5x with 1772.5x1772.5 pixels (XY), with an effective pixel size of 100 nm. Ten Z stacks were captured with a z-step spacing of 2 µm. All images were capture using a scan speed of 400 Hz with a bidirectional scan. Total volume of each analysed lung 2-photon images was therefore (177.25x177.25x2)x10 = 628351.25 µm3, and the number of fluorescent events (2–4 µm diameter) determined by eye using Fiji ImageJ. Total lung volume was determined by a fluid displacement approach, as per details in Supplementary Fig. 1d.
Confocal Microscope Imaging
Confocal images were obtained on an inverted SP8 confocal microscope (Leica Microsystems) attached to a DMI6000 microscope frame (Leica Microsystems). All images were acquired using a 100x/1.44 NA oil immersion objective. Excitation was provided by either a 405 nm laser (Hoechst33324) or 488 nm laser (for CD41-FITC) or 561 nm laser (for CD41-PE) and the resultant fluorescence was detected using a Hybrid detector in the range 410–470 nm for Hoechst 33342, 495–570 nm for CD41-FITC or 571–650 nm for CD41-PE. A total of 50 randomly chosen fields of view were imaged and where z-stacks were acquired, a 2 µm z-step spacing was used for Extended Data Video 6–7, or a 0.5 µm z-step spacing for Extended Data Video 8.
To obtain an estimate of the dimensions of each nucleus in XY, z-stacks were loaded into FIJI and subjected to a maximum intensity projection and a region of interest (ROI) was manually drawn around an individual nucleus. The inbuilt measure function within imageJ was used. To obtain the dimensions of the major and minor axis, the ‘fit ellipse’ option was enabled. To estimate the nuclear depth, analysis was performed in FIJI. Image stacks were loaded, and a cell nucleus was selected using the ROI tool. The average intensity profile of this nucleus was plotted in z and then fitted with a gaussian profile using the built-in plot and fitting tools of FIJI. From the fitted parameters the depth of the nucleus was estimated using the full width at half maximum (FWHM) of the gaussian fit. The FWHM was calculated as 2√(2𝑙𝑛2) 𝜎. This process was manually repeated for multiple cell nuclei.
Transmission Electron Microscope Imaging
To visualize and compare the ultrastructures of generated and control platelets by transmission electron microscopy (TEM), host platelets were first depleted by intraperitoneal administration of anti-GPIbα antibody R300 (2 µg/g bodyweight) prior to MK infusion through the heart-lung preparation. After 18 passages, the generated platelets were pelleted by centrifuging the collected perfusate. The platelet pellet was then resuspended in a cacodylate-buffered glutaraldehyde fixative and fixed at 4–8°C overnight and then post-fixed in osmium ferrocyanide. Fixed cells were then embedded in a solidifying BSA/glutaraldehyde gel. Gel-embedded platelets were stained with uranyl acetate and lead aspartate followed by dehydration with ethanol and infiltrated with Epon resin in a Tissue Processor (Leica EMTP). 70nm sections were cut from blocks with a Reichert Ultracut E and imaged with a Tecnai 12 electron microscope (ThermoFisher UK).
In Vitro Thrombus Formation
In brief, ibidi µ-Slide VI0.1 chips were coated with 50 µg/mL collagen overnight at 4°C before being flushed and blocked with 2% fatty acid–free BSA prepared in HEPES-Tyrode’s buffer. Freshly drawn mouse blood was collected from the inferior vena cava using 10 U/mL heparin and 20 µM PPACK (D-phenylalanyl-prolyl-arginyl chloromethyl ketone) as anticoagulant, following euthanasia by rising CO2. Mouse blood was mixed with 2 mL NaCl (150 mM) and centrifuged to remove PRP. Megakaryocytes stained with 2 µM DiOC6 were passaged through the mouse-lung preparation 18 times, and approx. 0.7 mL perfusate was mixed with 1.3 mL mouse blood lacking PRP and incubated with CellTracker™ Red CMTPX dye (1:1000 dilution) for 10 mins.
The mixed sample was then transferred to a 5 mL syringe and perfused using an Aladdin AL-1000 syringe pump (World precision instruments, United Kingdom) through the ibidi slide, at a shear rate of 1000/s for 20 minutes. Platelets were fixed by perfusion of 4% paraformaldehyde through channels for 4 minutes before nonadherent cells were flushed away with HEPES Tyrode buffer.
Thrombus formation was determined by generating confocal z-stacks (1024×1024 pixels, 0.787 µm z stack distance) from 5 randomly chosen fields of view using a Leica SP8 confocal microscope. Images were acquired using a 20x/0.7 NA air immersion objective. Excitation was provided by either a 488 nm (DiOC6) or 561 nm (CellTracker™ Red CMTPX) laser with the resultant fluorescence being detected by Hybrid detectors in the range 498–551 nm (DiOC6) or 571–623 nm (CellTracker Red CMTPX).
Two-photon intravital microscopy of the BM.
Mice were anaesthesized by intraperitoneal injection of medetomidine 0.5 µg/g, midazolam 5 µg/g and fentanyl 0.05 µg/g body weight. A 1 cm incision was made along the midline to expose the frontoparietal skull, without damaging the bone tissue. To immobilize the head while imaging, the mouse was fastened with a stereotactic holder on a heated customized stage. BM vasculature was visualized by intravenous injection of anti-CD105-AlexaFluor 546 antibody (1 µg/g body weight) and 100 µl AlexaFluor 546-labeled BSA. Platelets and MKs were stained intravenously with anti-GPIX-AlexaFluor 488 antibody (1.5 µg/g body weight). Images and time-lapse videos were acquired using a 20x water objective with a numerical aperture of 0.95 (Leica Microsystems CMS) on a confocal TCS SP8 MP (Leica Microsystems CMS) equipped with a tunable broadband laser (Coherent).
Materials
Cell culture reagents including IMDM-Glutamax and penicillin-streptomycin were from Invitrogen (Paisley, UK). Recombinant murine stem cell factor (rSCF) and recombinant murine thrombopoietin (rTPO) were purchased from PeproTech EC (London, UK). PE or FITC -conjugated anti-CD41 and isotype-nonspecific IgG were from BD Biosciences (Berkshire, UK). FITC-conjugated anti-mouse CD31, CD102 antibodies and isotype-nonspecific IgG were from Biolegend (San Diego, USA). SYLGARDTM 184 Silicone Elastomer Kit was from Dow Corning (Michigan, USA). Anti-mouse GPIbα antibody R300, FITC-conjugated anti-mouse CD61, CD42b, CD42d, CD49b, Glycoprotein VI (GPVI) antibodies and isotype-nonspecific IgG, PE-conjugated anti-CD62P antibodies, JON/A anti-integrin αIIbβ3 and isotype-nonspecific IgG were from Emfret Analytics (Würzburg, Germany). Hoechst 33324, DRAQ5, Calcein AM, Calcein Deep Red, tetramethylrhodamine (TMRM) and CellTracker™ Red CMTPX dye were from Molecular Probes™ (Loughborough, UK). DiOC6 iodide was supplied by Enzo Life Sciences (Exeter, UK). Ibidi slides were from Thistle Scientific (Glasgow, UK). Horm fibrillary collagen was from Takeda Pharma (Linz, Austria). Cross-linked collagen-related peptide (CRP-XL) was provided by Richard Farndale (University of Cambridge, UK). Unless stated, all other reagents were from Sigma-Aldrich (Dorset, UK). Reagents for intravital bone marrow (BM) two-photon experiments were as follows: anti-CD105 Alexa Fluor 546 (clone MJ7/19) was purified and labeled in house; rat monoclonal anti-mouse GPIX (CD42a) - AF488 was from Emfret Analytics and labeling was in house; BSA was from AppliChem and labeled in house.
Statistical information
All data were analyzed using GraphPad Prism 7 software (GraphPad Software Inc., San Diego, CA, USA). All image analysis was performed using ImageJ-win 64 software. Quantified data are presented as mean ± SEM from at least 3 independent experiments. A value of *P < 0.05 was considered statistically significant and was determined using either a paired Student's t-test or two-way ANOVA with Dunnett's or Bonferroni post hoc test for multiple comparison.