Study design and animal use

The study design was developed under the guidance of a university-wide AAALAC-accredited laboratory animal medicine program directed by veterinarians specialized in laboratory animal medicine. The protocol was reviewed and approved by the University of Michigan Institutional Animal Care and Use Committee (IACUC) and adhered to all applicable federal, state, local, and institutional laws and guidelines. CD47-null mice were developed by Lindberg et al. by replacing exon 2 of the Iap (CD47) genomic DNA with the neomycin resistance gene driven by the TK-promoter.47 and were acquired from Dr. Jeff Isenberg. Knockout mice were back-crossed onto a C57Bl/6J background and compared against WT (WT) mice C57Bl/6J obtained from Jackson Labs (Bar Harbor, ME). All mice were bred in the same vivarium to ensure similar microbiomes and genotyped either in house using a polymerase chain reaction (PCR) with primers specific to CD47-null (forward: GGCATTGCCTCTTTGAAAATGGATA, reverse: TGGCTTCTGAGGCGGAAAG) knockout mice or by a third-party service (Transnetyx, Cordova, TN). Animals were socially housed and allowed ad libitum access to food and water. Euthanasia was performed in accordance with current American Veterinary Medical Association guidelines using CO2 and confirming death secondarily with the generation of a bilateral pneumothorax. For all surgeries, animals were sedated through inhalant anesthetics (4%–5% isoflurane for induction; 1%–2% for maintenance). Adequate sedation was confirmed by a lack of responsiveness to a hind-limb toe pinch. The first dose of postoperative analgesia buprenorphine (0.1 mg/kg subcutaneously) was administered at the time of surgery. Ocular lubricant was applied to prevent corneal drying. Surgeries and quantitative analyses were conducted with investigators blinded from timepoint, genotype, and treatment. For μCT and histology analysis, 6–14 animals were used per time point with a distribution of males and females. In vitro marrow-derived MSC assays were repeated 2-3 separate times with pooled cells from two to three mice. In vitro periosteum assays were performed in three separate experiments, each replicate is one mouse.

Femoral fracture

Fractures were performed at 14 weeks of age during the late skeletal growth phase and just prior to peak bone mass.48,49 The stabilization and limb fracture technique is published.50 Briefly, the femur was stabilized with a 23-gauge needle placed percutaneously into the medullary cavity. Radiographs were acquired to confirm that the pin extended the length of the medullary canal but did not protrude from the proximal or distal end of the femur. The limb was positioned prone in the fracture apparatus using a jig to center the point of impact at the middle third of the femur. A 272-g weight was dropped from a height of 4.40 cm onto the femur and was stopped at an impact depth of 0.11 cm. Post-fracture radiographs were acquired to confirm the fracture location and pattern.

Hindlimb ischemia (HLI)

Hair was removed on one leg from mid-abdomen to mid-tibia, exposing the medial leg and knee joint, using depilatory cream. Mice were placed supine on the surgical field and the operative limb was secured in an abducted, externally rotated position. The surgical area was prepared and cleaned with two swabs of povidone-iodine and 70% EtOH and then protected with sterile drapes. A medial 15 mm incision was made proximal to the knee along the longitudinal axis of the limb. Retractors were placed and blunt dissection was used to visualize the femoral neurovascular bundle. The vein and nerve were dissected from the femoral artery to preserve vascular return and limb sensory-motor function. A 2 mm section of the femoral artery distal to the epigastric artery and proximal to the medial genicular artery was ligated using 8-0 nylon sutures and resected to induce deep distal acute ischemia.51,52 The incision was closed using simple interrupted sutures with 6-0 nylon. Laser Doppler measurements of hindlimb perfusion were used to confirm ischemia.

Laser Doppler

Hindlimb perfusion was measured using laser Doppler blood flow monitor (Moor, moorVMS-LDF) equipped with a non-invasive skin probe (Moor, VP2). Animals were sedated as described above and maintained a respiratory rate of 55-100 breaths/min. Animals were placed supine on a heating pad (Hallowell EMC, 2789B hard pad & heat therapy pump) at 37.5 °C and remained undisturbed for 5 minutes to normalize to conditions. The Doppler probe was placed on the plantar surface of the hindlimb with the distal boarder of the probe abutting the digital walking pads. Alternating perfusion measurements were taken three times for each limb and averaged.

Tibia fracture with HLI

An ischemic tibial fracture was created by resecting the femoral artery, reducing distal perfusion as described above. The limb was stabilized using a 30-gauge intramedullary pin and a simple transverse fracture was created. Perfusion was tracked using laser Doppler flowmetry and fracture pattern was confirmed with digital radiographs.

CD47 morpholino treatment

Expression of CD47 was disrupted using a translation-blocking antisense morpholino oligonucleotide. On days 2 and 5 post-fracture, mice were injected intraperitoneally using a 23-gauge needle with 1.0 nmol/g of morpholino in 750 μL of saline. Mice were injected with either a CD47 antisense vivo-morpholino (CGTCACAGGCAGGACCCACTGCCCA) or a vivo-morpholino standard control (CCTCTTACCTCAGTTACAATTTATA) oligonucleotide (Gene Tools, Philomath, OR).

Sample preparation and micro-computed tomography (μCT)

Femurs were harvested at day 10 and 20 post-fracture for non-ischemic injury experiments. Tibia was harvested at day 10, 15 and 20 for knockout ischemic studies and on day 15 post-fracture for morpholino ischemic experiments. Bones were fixed in 4% paraformaldehyde (PFA) for 24 h at 4 °C under gentle agitation. After fixation, the intramedullary pins were carefully removed without disrupting the callus by extracting them from the distal end of the femur using small-nosed needle nose plyers. Femurs were placed in a 4-limb positioning jig and scanned in eXplore Locus scanner (GE Healthcare). The femur containing jig was immersed in water and specimens were scanned at an 18 µm voxel size using the following settings: 80 kV, 80 µA, 1 600 ms and 400 views. Reconstruction and analysis were performed in MicroView (Parallax Innovations). A callus region of interest (ROI) was created by manually sectioning around the perimeter of the callus every 10 slices and then performing a spline interpolation between each individual section. Cortical bone was subtracted from the callus ROI using a similar sectioning and interpolation technique. For day 10 fractures, a 3.00 × 3.00 × 5.00 cylindrical ROI was centered around the fracture to avoid sectioning errors caused by vague delineation of density between callus edge and soft tissue. The callus ROI was analyzed for callus length, callus volume (commonly referred to as total volume (TV)), bone volume (BV), total mineral content (TMC), and bone mineral content (BMC) using a threshold of 1 650 Hounsfield units (HU). The threshold was established by averaging the bone threshold of 10 whole intact femurs using the automatic bone density algorithm MicroView based on thresholding methods.53 Bone volume fraction was calculated by dividing the BV by the TV and represents how much of the callus contains mineralized tissue. Tissue mineral density (TMD) was calculated by dividing the TMC by the BV, and bone mineral density was calculated by dividing BMC by TV. Representative images were created using isosurfaces of cortical bone and callus with a threshold set at 1 200 HU. Isosurfaces were post-processed using smoothing.

Sample preparation for histology and immunofluorescence

Tibiae were harvested at days 4 and 7 post-fracture for knockout experiments and fixed in 4% paraformaldehyde for 24 h under gentle agitation. After fixation, the intramedullary pins were carefully removed without disrupting the callus by extracting them from the distal end of the tibia using small-nosed needle plyers. Tissues were formalin-fixed, paraffin-embedded using a tissue processor (Leica ASP 300S) and sectioned in the sagittal plane at 5 µm for immunofluorescent staining or 10 µm sections for Milligan’s Trichrome and Safranin O analyses. Slides were deparaffinized using a sequence of xylene and ethanol dilutions.

Histomorphometry analysis

Sagittal sections (10 µm) through the entire block were cut and every thirtieth slide was stained with Safranin-O/fast green. Adjacent slides were stained with modified Milligan’s Trichrome. An Olympus CAST microscope (Center Valley, PA) and software made by Visiopharm (Hørsholm, Denmark) were used to perform stereology.54 Histomorphometry was used to calculate callus volume and the volume of cartilage, bone, fibrous tissue, and marrow within the comprising callus. Total volumes were estimated using Cavalieri’s principle for a conical frustum, as in the past.55

EdU analysis and quantification

Mice were injected with 10 mmol/L EdU intraperitoneally at 4 h prior to harvest on day 4 and day 7. Tissue samples were harvested and prepared as described above. Sagittal sections (10 µm) were utilized for EdU quantification. Post-deparaffinization, slides were permeabilized with 0.5% Triton X-100 for 10 min. A commercially available EdU Click-it Kit (Fisher, C10339) was used to label EdU-positive cells. Slides were washed three times in PBS and counter-stained with DAPI. Entire sections were imaged at 10× resolution and stitched in both Texas Red and DAPI channels. Photoshop was used to define the callus ROI by outlining the callus to exclude bone, marrow, and muscle, capturing only the callus region. CellProfiler was used to quantify EdU-positive nuclei within the ROI using Texas Red. DAPI was used to quantify nuclei. These were overlayed to determine the percent of EdU-positive nuclei for each callus.

Endothelial cell density quantification

Heat-mediated antigen retrieval was performed, and tissues were permeabilized with 0.5% Triton-X 100 in phosphate-buffered saline (PBS) prior to immunofluorescent staining. Immunofluorescence staining was performed on 5 µm sagittal sections using rabbit anti-mouse CD31(Abcam, ab281583) and rat anti-mouse EMCN primary antibodies to label endothelial cells, followed by AF750 and AF594 secondary antibodies. DAPI stain was used to identify nuclei. Primary, secondary, isotype, and unstained controls were used to detect non-specific staining or autofluorescence. 2–4 images at the central callus and distal callus regions were taken using an automatic microscope (Agilent, BioTek Lionheart FX, Santa Clara, CA). Central callus regions were defined as proximal to the fracture site on either side of the bone, and distal callus was defined as the periphery of the fracture callus. The Vessel Analysis plug-in on ImageJ was used to calculate endothelial cell density in central and distal regions of the fracture callus.

Marrow cell harvest

Femurs and tibias were harvested from adult mice and dissected of all soft tissue under sterile conditions. The distal femur and proximal tibia were separated at the level of the metaphysis to expose the medullary canal. The long bones were placed exposed end down in a 0.5 mL microcentrifuge tube with a hole in the bottom made with an 18-gauge needle. The microcentrifuge tube was nested in a 1.5 mL microcentrifuge tube and centrifuged at 10 000 × g for 15 min. Cells were suspended in culture media and passed through a 70 μm strainer. Cells were then pelleted and washed twice with culture media.

Stromal cell isolation from bone marrow

Bone-marrow-derived cells were counted and plated at 65 million cells/75 cm2 for 8 days with a half media change at 4 days post-harvest and a full media change at 6 days post-harvest. At 8 days post-harvest cells were treated with Trypsin and all cells were collected. Cells were then sorted with CD45 MicroBeads (Miltenyi Biotec, 130-052-301) and CD45- and CD45+ cells were plated separately at 60 000 cells/mL. Cell Counting Kit 8 (CCK8) assay was performed as detailed in section 4.14 at 1, 3, 5, 7 and 9 days post-plating to assess metabolic activity as a surrogate for cell proliferation.

Periosteum cell harvest

Femurs and tibias were harvested from adult CD47-null and C57 Bl/6 mice. The bone was isolated from the muscle and the articular and epiphyseal ends were dipped into pre-warmed 5% agarose gel. After solidification of the agarose, the femurs and tibias were placed in a 10% Collagenase P Digestion buffer at 37 °C with agitation. After 10 min, the buffer was discarded to remove contaminating cells. Bones were transferred to a new tube containing 10% Collagenase P Digestion for 1 h at 37 °C with agitation. Cells were passed through a 70 μm strainer, pelleted, and resuspended in culture media.

Endothelial cell harvest and purification

Endothelial cells were isolated from the lungs of 4-week-old mice. Lungs were removed using sterile scissors and cut into <0.2 cm pieces. Lung tissue was incubated under agitation at 37 °C in a 50 mL conical tube with 10 mL of collagenase A in PBS (5 μg/mL). After 30 min, 20 mL of PBS was added to the conical tube and then vigorously shaken to further dissociate tissue. The cell suspension was filtered through a 70 μm strainer, pelleted, and washed once with endothelial cell media (Lifeline Cell Technology, VascuLife VEGF Endothelial Complete, Frederick, MD). Cells were combined across four mice of the same sex and genotype. The cell suspension was sorted for CD31+ endothelial cells using anti-CD31 MicroBeads (Miltenyi Biotec, 130-097-418; San Diego, CA) and a magnetic cell isolation system (Miltenyi Biotec, QuadroMACS Separator & LS Columns). CD31+ cells were cultured in endothelial cell media on 2% gelatin-coated plates and flasks or used immediately for flow cytometry.

Colony forming unit—fibroblast (CFU-F)

Cells harvested from marrow were plated at a density of 2 × 105 cells/cm2 in 60 mm dishes and grown in 5% CO2 at 37 °C. Half of the culture media (Alpha-MEM, 10% FBS, 1%, L-glutamine, 1% anti-anti) was exchanged 4 days post-culture and then every three days until cell harvest. At day 12 post-plating, bone marrow-derived plates were stained with alkaline phosphatase to mark mesenchymal stem cells. Dishes were washed twice with PBS and then fixed with Citrate-Acetone-36% Formaldehyde (ratio, 25:65:8) for 30 s and then rinsed with deionized water. Naphthol AS-BI Alkaline Solution (Sigma-Aldrich, 861-10) was added to the plates at 25 °C for 15 min and then rinsed with deionized water. Plates were counterstained at 25 °C for 2 minutes with Neutral Red (Sigma-Aldrich, N6264). Finally, dishes were rinsed with tap water and allowed to air dry. Individual dishes were examined using an upright brightfield stereo microscope (Bausch & Lomb) placed over a 1 cm grid. Groups of 25-100 ALP-positive cells were counted as small colonies and confluent groups of >100 ALP-positive cells were counted as large colonies. Counting was performed blinded to cell genotype. Cells harvested from the periosteum were plated at a density of 1 × 105 cells/cm2 in 60 mm dishes and grown in 5% CO2 at 37 °C. At day 10, post-plating alkaline phosphatase staining was performed on periosteal MSC as described above. Individual dishes were imaged in color brightfield (BioTek, Lionheart FX) to analyze and count clusters with alkaline phosphatase staining.

Cellular proliferation (MTT)

Metabolically active cells were quantified using an MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium Bromide) assay which measures formazan produced by cleavage of tetrazolium salts (MTT) by the succinate-tetrazolium reductase system. A commercial cell proliferation kit (Millipore Sigma, Cell Proliferation Kit I (MTT)) was used for mesenchymal stem cells and endothelial cells. Cells were grown in 12-well plates in 5% CO2 at 37 °C as described. Briefly, bone marrow-derived cells were either directly plated (Fig. 3c) or cells were passaged with 2 min of exposure to trypsin-EDTA and cells from 2 to 3 individual animals pooled (equivalent males and females in pools) to increase overall cell number and to decrease mouse to mouse variability (Fig. 3d). On the day of assay, MTT was added to the cells (10 μL MTT per 100 μL of media). Cells were returned to the incubator for 4 h and then a solubilization solution was added to the cells (100 μL solubilization solution per 100 μL of media). Cells were incubated overnight to solubilize the purple formazan crystals. The next day 210 μL from each well was added in triplicate to a 96-well plate. Formazan concentration was quantified on a multi-mode microplate reader (Molecular Devices, SpectraMax M3) using a 575 nm wavelength for formazan spectrophotometric absorbance and 650 nm for reference. Readings were averaged across technical replicates.

Cellular proliferation (CCK8)

Metabolic activity within periosteal MSC was quantified using a Cell Counting Kit 8 assay (WST-8/CCK8; Abcam, Cambridge, UK). Water-soluble tetrazolium salts are used to quantify cell viability through the production of formazan dye similar to that of an MTT assay. Periosteal MSC was isolated and grown in 12 well plates at 5 × 104 cells/well. Starting at day 1, every other day CCK8 reagent was added to the cells every other day for 9 days. The plates were returned to the incubator and analyzed 2 h later. 100 μL from each well was added to a 96-well plate and read on the spectrophotometric at 460 nm. Readings were averaged across technical replicates.

RNA isolation and gene expression (quantitative RT-qPCR)

mRNA was isolated from 1st passage adherent cells plated as described in Fig. 3d using 4 ml TRIzol and extracted using acid guanidinium thiocyanate-phenol-chloroform with GlycoBlue co-precipitant. mRNA was purified with Qiagen RNeasy Midi spin columns (Qiagen, Redwood City, CA) with on-column DNase digestion per the manufacturer’s protocol. cDNA was reverse transcribed from mRNA with Superscript III (Invitrogen, ThermoFisher), and quantitative RT-PCR reactions were conducted with 20 ng of template using custom primers (Table 1) and SYBR Select master mix (Applied Biosystems, Waltham, MA) for 50 cycles. Gene expression was calculated by normalizing to GAPDH (ΔCT) and then to WT mice (ΔΔCT). Fold change was calculated as 2−ΔΔCT.

Table 1 Sequences of reverse and forward primers used to assess RNA through q-PCR analysisFlow cytometryCell cycle and apoptosis analysis

MSC cultures used for flow cytometry were obtained following culturing pooled 1st passage cells as described in Fig. 3d harvested at day 3. Cells were again trypsinized, and counted, and 5 × 105–1 × 106 cells were placed in individual 12 × 75 mm round-bottom tubes suspended in flow cytometry staining buffer (Invitrogen, 00-4222-26). For apoptosis, a Caspase-3/7 assay kit was used (Invitrogen, C10427). Each tube was brought to a volume of 1 mL staining buffer and 1 μL of CellEvent Caspase-3/7 reagent was added. Cells were incubated at 37 °C in the dark for 25 min and then 1 μL of SYTOX AADvanced dead stain solution was added. Cells were incubated at 37 °C for 5 min and then analyzed. For cell cycle analysis, cells were fixed with 0.5 mL of 100% cold ethanol for 20 min at 4 °C under gentle rotation. Cells were pelleted and the ethanol was decanted. Cells were resuspended in 1 mL of staining buffer and 4 drops (164 μL) of Propidium Iodide Ready Flow (Invitrogen, R37169) was added to the cells and incubated at 25 °C for 20 min. Cells were analyzed on a ZE5 Cell Analyzer (Bio-Rad) and data were processed using FCS Express 6 Flow Cytometry software (De Novo Software, Pasadena, CA). Gating was performed on large cells, to capture the stromal (CD45-) population, The percentage of the cell population in G1, or S/G2 phase of the cell cycle was calculated by fitting Propidium Iodide excitation counts using FCS Express: Multicycle (De Novo Software).

Endothelial cell immunophenotype

Endothelial cells were harvested from lung tissue and sorted as described. Cells for flow cytometry were counted and 5 × 105–1 × 106 cells were placed in individual 12 × 75 mm round-bottom tubes suspended in flow cytometry staining buffer (Invitrogen, 00-4222-26). A conjugated antibody for CD31 and control (Invitrogen, 12-0311-82, preparation: conjugated PE, host: rat, isotype: IgG2a, clone: 390) were added to the cells at 4 °C for 1 h. Cells were pelleted and washed twice with staining buffer and suspended 1 ml of staining buffer with DAPI for an end concentration of 1 × 106 cells/mL and DAPI (0.2 μg/mL; BD Biosciences, 564907) added. Cells were analyzed on a ZE5 Cell Analyzer (Bio-Rad) and data were analyzed using FCS Express 6 Flow Cytometry (De Novo Software).

Periosteal skeletal stem cell phenotype

Periosteal MSC were harvested as described from C57 Bl/6 and CD47-null mice. Post digestion, cells were resuspended in PBS and stained with a panel to mark skeletal stem cells24,56 using conjugated antibodies for CD90, Sca1, CD140a, CD51, CD200, Ly51, CD105, and a lineage-negative cocktail. Samples were stained with fixable viability dye for 30 min. TrueStain FcX block (Biolegend, San Diego, CA; 156604) was added and cells were pelleted and washed prior to antibody staining at 4 °C for 1 h. Following staining, samples were pelleted, washed, and fixed using commercially available Fixation/Permeabilization Buffer (Biolegend, 426803). Cells were analyzed on an LSRFortessa Cell Analyzer (BD), and data were analyzed using FlowJo (BD).

Osteogenic differentiation

Marrow cells were cultured in a T75 flask for 7 days and then trypsinized, counted, passaged, and plated in osteogenic media (Alpha-MEM, 0.5% β-glycerophosphate, 0.1% ascorbic acid-2-phosphate, 1% L-glutamine, 1% antibiotic-antimycotic) at density of 200 000 cells per well in a six-well plate. Each sample contained cells pooled from two mice. At day 14, cells were stained with ALP as described or Alizarin Red S (ARS) to image mineralization. For ARS staining, cells were washed gently with PBS and fixed with 4% PFA for 60 min under slow rotation. Filtered 1% Alizarin Red S (Millipore, A5533) was added to the cell monolayer for 15 min in the dark at 25 °C. Plates were washed 3–4 times with distilled water until background staining was removed. ALP and ARS plates were imaged using an inverted conventional brightfield microscope (BioTek, Lionheart FX) affixed with RGB imaging cubes. Stitched imaged were acquired using a Phase Plan Fluorite 1.25× air objective, registered using the blue channel, and stitched with 10% overlap. Stitched images were analyzed for staining quantity using Fiji ImageJ2.57 Only the blue channel was used for analysis. A 560 × 560-pixel circular ROI was created to capture the whole plate but exclude edge artifacts. Image thresholds were set to 21/124 and 139/158 for ALP and ARS, respectively. Images were then measured for percent of ROI with positive staining.

Statistical analysis

The μCT, stereology, flow cytometry, and gene expression datasets were grouped by genotype as an independent variable (WT or CD47-null; [M]Ctrl or [M]CD47) and by day as indicated. Data were tested for normal distribution and equal variances before analysis. Mean and standard deviation (SD) were calculated for continuous variables. Sex of the sample is shown where applicable for fracture phenotype analyses using μCT, histomorphometry, and immunofluorescent analysis. Categorical variables were expressed as numbers and percentages. Two-sided t-tests were performed to test for differences between genotypes at each timepoint. A one-way ANOVA was performed to assess differences in perfusion between limbs and across genotype. Perfusion recovery data was fit to a one-phase nonlinear curve (Prism 8, GraphPad, San Diego, CA) and compared using extra sum-of-squares F test. Data were aggregated and analyzed using open-source R.58 and Prism 10. Statistical significance was set at P ≤ 0.05. Points of significance are annotated graphically in figures with *P < 0.05, **P < 0.01, and ***P < 0.001. Data were further separated by sex for μCT, histomorphometry, cell culture, and immunofluorescent analysis were possible and a two-way ANOVA was performed to assess the independent effects of sex and genotype or whether there was an interaction effect (Tables S1–8).