Identification and characterization of an injury-induced skeletal progenitor

Significance

Here, we characterize the injury-induced activation of a specific, highly purified population of multipotent skeletal progenitor cells. These activated progenitors show increased cell frequency, increased viability, and enhanced osteogenic potential. They also possess a unique transcriptional profile that distinguishes them from progenitors found in uninjured bone. We report that these features improve regenerative capacity, suggesting that activated progenitors play a principal role in bone healing. We hope that a better understanding of stem and progenitor activation will inspire novel therapies that restore impaired skeletal regeneration.

Abstract

The postnatal skeleton undergoes growth, remodeling, and repair. We hypothesized that skeletal progenitor cells active during these disparate phases are genetically and phenotypically distinct. We identified a highly potent regenerative cell type that we term the fracture-induced bone, cartilage, stromal progenitor (f-BCSP) in the fracture callus of adult mice. The f-BCSP possesses significantly enhanced skeletogenic potential compared with BCSPs harvested from uninjured bone. It also recapitulates many gene expression patterns involved in perinatal skeletogenesis. Our results indicate that the skeletal progenitor population is functionally stratified, containing distinct subsets responsible for growth, regeneration, and repair. Furthermore, our findings suggest that injury-induced changes to the skeletal stem and progenitor microenvironments could activate these cells and enhance their regenerative potential.

Stem and progenitor cells mediate homeostasis and regeneration in postnatal tissue (1–6), but how these cells contribute to tissue repair, specifically fracture healing, is not well understood. Alterations in tissue microenvironments are known to influence stem cell behavior. Injury, infection, and aging have been shown to induce lineage skewing of hematopoietic stem cell (HSC) progeny (7–9). It is now known that HSCs are a diverse population of cells, including HSCs with lineage biases (10), that could model heterogeneity in other tissue stem cells (1, 11–13). Thus, we propose that fracture activates a distinct subset of skeletal stem and progenitor cells that mediate tissue regeneration and repair. If so, it is conceivable that specific stem or progenitor cells contribute to postnatal regeneration and, thus, could be implicated in cases of skeletal degeneration, such as osteoporosis, osteopenia, and impaired fracture healing.

Previously, our laboratory identified a pure clonal progenitor of bone, cartilage, and hematopoietic niche stromal cells (BCSP) in the appendicular skeleton of mice. We defined the BCSP by the immunophenotype (CD45⁻TER119⁻Tie2⁻AlphaV⁺Thy⁻6C3⁻CD105⁺) and showed that it forms bone, cartilage, and at least three subsets of stromal cells (14). We also observed that the BCSP forms bone via endochondral ossification. Thus, we used a relative stability model of long-bone fracture healing to investigate how the BCSP contributes to bone repair. Recent studies report that bone injury induces progenitor expansion (15, 16), but they do not consider how phenotypic or transcriptional changes within these populations affect tissue regeneration. Our investigative approach allows us to characterize these changes in a specific, highly purified skeletal progenitor population. These findings advance the understanding of postnatal skeletal regeneration and repair on a cellular and molecular level.

Results

BCSP Expansion Precedes Ossified Callus Formation.

Transverse, middiaphyseal, femoral fractures were made in 8-wk-old, C57BL/6 male mice (Fig. 1A). Intramedullary fixation enabled relative stability to facilitate endochondral ossification (17). The callus index (CI), defined as the maximum diameter of callus divided by the diameter of adjacent diaphysis, measured callus formation over time (Fig. 1B) (18). CI peaked significantly at postfracture day 14 and marked maximal callus formation (Fig. 1C, ****P < 0.0001). To determine whether BCSP expansion accompanies callus growth, individual uninjured femora and dissected fracture calluses were analyzed separately. Femoral calluses were dissected at postfracture days 3, 7, 14, 21, and 28. BCSPs were isolated using mechanical and enzymatic tissue dissociation and were stained for fluorescence-activated cell sorting (FACS). FACS analysis showed a significant increase in BCSP frequency after fracture, which peaked at postfracture day 7 (D7) (Fig. 2 A and B, ****P < 0.0001).

Fig. 1.

BCSPs were isolated from mouse femora. (A) Schematic showing our experimental model, including fracture induction, tissue dissection (fracture calluses were dissected at postfracture days 3, 7, 14, 21, and 28), and BCSP isolation via FACS. (B) Callus index (CI) is a ratio of fracture callus diameter to adjacent diaphysis diameter (red, callus diameter; blue, diaphysis diameter). (C) CI peaks at postfracture day 14 (****P < 0.0001, n = 9 femora; D14 CI, 2.06 ± 0.08; Student’s t test).

Fig. 2.

BCSP expansion precedes callus formation. (A) FACS analysis shows BCSP expansion after fracture. (B) Bar graph showing a significant increase in the frequency of BCSPs at postfracture day 7 versus uninjured (UI) control (****P < 0.0001; UI, 307 ± 61, n = 8; D7, 18,444 ± 3,314, n = 9; Student’s t test). Data represent mean cell frequencies of individual samples. Whole UI femora were used whereas calluses were dissected. (C) CI is significantly reduced in irradiated calluses versus nonirradiated controls (*P < 0.05, n = 4–6, Student’s t test). (D) Bar graph showing a significant reduction in the frequency of BCSPs at postfracture day 7 (D7 f-BCSP) in irradiated calluses versus nonirradiated controls (**P < 0.01, n = 5 calluses; nonirradiated, 25,442 ± 3,758; irradiated, 9,739 ± 562; Student’s t test). (E) Bar graph showing D7 f-BCSP frequency remains depleted in irradiated calluses versus nonirradiated controls when fracture is induced 3 mo postirradiation (**P < 0.01, n = 3–5 calluses; nonirradiated, 27,478 ± 12; irradiated, 11,267 ± 6; Student’s t test).

Hindlimb Irradiation Reduces Fracture-Induced BCSP Expansion.

Radiotherapy is an important modality for the treatment of many malignancies; however, it often induces significant osseous side effects (19). We hypothesized that bone dysfunction correlates with reduced BCSP frequency. After hindlimb irradiation with 800 rad 12 h before fracture, delayed callus formation and prolonged healing was observed in irradiated femora (Fig. 2C; day 14, *P < 0.05; day 21, *P < 0.05). Notably, BCSP expansion was reduced significantly in irradiated versus nonirradiated D7 calluses (Fig. 2D, **P < 0.01). Both callus development and BCSP expansion remained impaired 3 mo postirradiation (Fig. 2E, **P < 0.01). These results suggest that BCSPs are necessary for functional bone regeneration.

Injury Induces Phenotypic Changes in BCSPs.

We next investigated injury-induced phenotypic changes in BCSPs. FACS-isolated BCSPs from uninjured femora (uninjured-BCSP, u-BCSP) and D7 fracture calluses (fracture-BCSP, f-BCSP) were pooled separately for in vitro and in vivo assays (Fig. 3A). f-BCSPs demonstrated significantly increased plating efficiency, as determined by colony number (Fig. 3B, *P < 0.05), significantly greater viability (Fig. 3C, *P < 0.05), and markedly reduced apoptotic activity (Fig. 3D). Furthermore, in vitro immunofluorescence staining showed that f-BCSP–derived colonies form bone via a collagen II intermediate, as previously shown by the u-BCSP (1, 14) (Fig. 3E).

Fig. 3.

Injury induces phenotypic changes in BCSPs. (A) Schematic showing experimental model for in vitro and in vivo assays (ODM, osteogenic differentiation media; XTT, viability assay; CFU, colony-forming units). (B) Bar graph showing D7 f-BCSPs form significantly more colonies than u-BCSP (*P < 0.05, n = 3; f-BCSP, 22 ± 4.40; u-BCSP, 10.33 ± 1.88; Student’s t test). (C) XTT cell viability curves show that D7 f-BCSPs are significantly more viable than adult u-BCSPs within 36 h of plating (*P < 0.05, n = 3; Student’s t test). (D) FACS analysis showing u-BCSPs (Top Left, black) have greater apoptotic activity than f-BCSPs (Bottom Left, red). The percentage of Annexin V-positive cells is listed in each plot on the Left. A merged plot of u-BCSPs (black) and f-BCSPs (red) is shown on the Right. (E) Immunofluorescence staining shows that f-BCSPs form bone (osteocalcin, green) through a cartilaginous intermediate (collagen II, purple) as seen previously in the u-BCSP (14). DAPI, blue. (Scale bar: 1 mm; Magnification, Inset: 2.5×.)

Osteogenic potential was examined in vitro and in vivo. Alkaline phosphatase activity, an intermediate marker of bone formation, and alizarin red stain, an extracellular matrix mineralization marker, were used to measure osteogenic potential in vitro. f-BCSPs showed significantly greater alkaline phosphatase activity and alizarin red stain relative to u-BCSPs (Fig. 4 A and B, alizarin quantification, ****P < 0.0001). A model of ectopic bone formation previously described by our laboratory was used to determine osteogenic activity in vivo (20). In short, 20,000 GFP-labeled, FACS-isolated cells from each BCSP population were transplanted separately under the renal capsules of immunodeficient mice (Fig. 3A). Grafts were explanted after 4 wk. Both populations formed ectopic bone and marrow (20), but engrafted f-BCSPs formed significantly larger bone specimens (Fig. 4C, Top and Middle and Fig. 4D, *P < 0.05). These phenotypic differences could describe progenitor activation or characterize a more regenerative progenitor subpopulation.

Fig. 4.

f-BCSPs have enhanced osteogenic potential. (A) Staining for alkaline phosphatase activity (Top) and alizarin red (Bottom) shows enhanced osteogenic activity of f-BCSPs versus u-BCSPs in vitro. (Scale bars: 200 μm.) (B) Bar graph of alizarin red absorbance shows significantly greater staining in f-BCSPs versus u-BCSPs (****P < 0.0001, n = 3; f-BCSP, 1.907 ± 0.046; u-BCSP, 0.228 ± 0.009; Student’s t test). (C) GFP-labeled BCSPs transplanted under renal capsules show ectopic bone formation in vivo. f-BCSP grafts (Middle) are larger than those of adult u-BCSPs (Top). f-BCSP grafts more closely resemble perinatal u-BCSP (P3 u-BCSP) grafts in size (Bottom). (Far Left) Brightfield; (Left) brightfield. (Scale bars: 1 mm.) (Right) Fluorescent microscopy. (Scale bars: 1 mm.) (Far Right) Movat’s Pentachrome stain of explanted tissue (yellow, bone; blue, cartilage; brown, marrow; red, renal tissue), brightfield. (Scale bars: 200 μm.) (D) Bar graph showing significantly greater in vivo ectopic bone formation by f-BCSPs versus u-BCSPs (*P < 0.05; f-BCSP, 130,391.29 ± 17,444.52, n = 7; u-BCSP, 67,729.63 ± 19,077.15, n = 8; Student’s t test).

Differential Gene Expression Characterizes BCSP Response to Injury.

We investigated the transcriptional mechanisms underlying these phenotypic differences using microarray analysis. Representative genes involved in cell survival, osteogenesis, chondrogenesis, and angiogenesis were analyzed (Fig. 5A) using the Gene Expression Commons (Materials and Methods) (21). Analysis included BCSPs isolated from uninjured femora at postnatal days 3 (P3, u-BCSP) and 56 (P56, u-BCSP) and from calluses at postfracture days 3 (D3) and 7 (D7).

Fig. 5.

BCSPs show differential gene expression during disparate phases of skeletal growth. (A) Heatmap showing differential gene expression between FACS-isolated BCSPs using microarray analysis. Genes representative of skeletogenesis (Left) are clustered according to function (Right). Scale: pink, high expression; blue, low expression. (B) Schematic showing that u-BCSPs predominate skeletal homeostasis (Left). Injury induces BCSP expansion at the injury site (Middle, postfracture day 3) to mediate healing (Right, postfracture day 7). Gray cell, u-BCSP; red cell, f-BCSP.

Cell survival factor analysis showed differential apoptotic gene expression between groups (Fig. 5A, fifth cluster from top). Notably, D7 f-BCSPs down-regulated the following apoptotic factors: B-cell leukemia/lymphoma 10, Caspase3, Dido1, Mitochondrial ubiquitin ligase activator 1, and Cell division cycle and apoptosis regulator 1 (22, 23). These data support reduced f-BCSP apoptotic activity seen in Fig. 3D.

Skeletogenic analysis showed that all groups maintain high expression levels of Runx2 and Sox9, master regulator genes of skeletal growth and development (24, 25), and BMP2, a major postnatal bone-forming factor (26) (Fig. 5A, first cluster from top). Expression of osteogenic signaling pathway (BMP2; canonical Wnt) antagonists was down-regulated in D7 f-BCSP (Fig. 5A, second cluster from top). Down-regulated BMP antagonists include Gremlin1 and Chordin (27, 28), and down-regulated Wnt antagonists include Dikkopf1 (Dkk1), Secreted frizzled-related protein (Sfrp)-4, Sfrp5, and Wnt inhibitory factor 1 (Wif1) (29–31).

Hedgehog pathway analysis showed that Indian hedgehog (Ihh) expression increases whereas Parathyroid hormone-like peptide (Pthlh) expression decreases in D7 f-BCSP (Fig. 5A, third cluster from top). Parathyroid hormone receptor 1 expression remains high. In addition, Ihh pathway antagonist, Hedgehog interacting protein (Hhip), is down-regulated whereas Dispatched-1 (Disp1), a protein involved in Ihh secretion, is up-regulated (32). Interestingly, expression of Collagen 10α1 (Col10α1), a hypertrophic chondrocyte marker, is up-regulated as well (33). These data suggest that fracture activates hedgehog signaling in f-BCSPs.

Angiogenic factor analysis (Fig. 5A, fourth cluster from top) showed that the expression of Vascular endothelial growth factor alpha (Vegfα) is up-regulated in D7 f-BCSP and that Placental growth factor (Plgf), a member of the VEGF family (20), rises in D3 f-BCSP. Transcriptional expression of Dicer1, a factor required for murine embryonic angiogenesis (34), is up-regulated in D7 f-BCSP. Cumulatively, these data mechanistically support f-BCSP activation (Fig. 5B).

CD49f⁺ Expression Identifies Activated f-BCSP.

Because f-BCSPs and P3 u-BCSPs (Fig. 4C, Bottom Row) formed more bone in vivo than u-BCSPs, we used microarray data to identify surface marker expression selective for f-BCSPs and P3 u-BCSPs. Integrin subunit α6 (defined by the CD49f surface marker) expression is up-regulated in both f-BCSPs and P3 u-BCSPs, but not in u-BCSPs (Fig. 6A). Expression of its beta-components, integrin subunit β1 and integrin subunit β4, remains low. Because CD49f expression diminishes gradually with age (Fig. S1), we hypothesized that it could classify a more regenerative f-BCSP subpopulation. Thus, phenotypic differences between CD49f^+/− f-BCSPs were investigated. CD49f^+/− f-BCSPs were isolated as previously described with an additional antibody for CD49f. FACS analysis showed that the proportion of CD49f⁺ f-BCSPs increases relative to CD49f⁻ f-BCSPs after fracture (Fig. 6B and Fig. S2 A and B). In vitro assays demonstrate that CD49f⁺ f-BCSPs form significantly more colonies (Fig. 6C, ***P < 0.001) and have significantly greater osteogenic potential than CD49f⁻ f-BCSPs (Fig. 6D, ***P < 0.001 and Fig. S2 C and D). These data could identify CD49f as an activation marker, or they could demonstrate the selective expansion of a minor subpopulation of CD49f⁺ cells in adult u-BCSP populations during bone repair.

Fig. 6.

CD49f⁺ expression identifies activated f-BCSPs. (A) Microarray analysis of FACS-isolated BCSP populations shows that CD49f expression is recapitulated after injury. (B) FACS analysis showing CD49f⁺ expression increases after fracture (merged FACS plot, Bottom Right). CD49f expression of D7 f-BCSPs (Top Right) was compared with u-BCSPs (Top Middle) and a negative control (FMO-CD49f, Top Left). (C) Bar graph showing CD49f⁺ f-BCSPs form significantly more colonies than CD49f⁻ f-BCSPs in vitro (***P < 0.001, n = 9; CD49f⁺, 16 ± 1.32; CD49f⁻ 8.5 ± 0.96; Student’s t test). (D, Top) The osteogenic potential of FACS-isolated BCSP populations was assayed in vitro using alizarin red staining. (Bottom) Results were quantified using spectrophotometry to measure absorbance. f-BCSPs have significantly enhanced osteogenic potential compared with u-BCSPs (****P < 0.0001, n = 3; Student’s t test), as shown previously. Further subfractionation of the f-BCSP population demonstrates that CD49f⁺ f-BCSPs are significantly more osteogenic than CD49f⁻ f-BCSPs (***P < 0.001, n = 3, Student’s t test).

Fig. S1.

BCSPs gradually lose CD49f expression during postnatal development (Left, merged FACS plot). The frequency of CD49f⁺ BCSPs was determined at postnatal weeks 2 (blue), 4 (yellow), and 8 (green) (Right). The population percentage of CD49f⁺ BCSPs is shown in each plot (Top Left, control FMO-FITC; Top Right, postnatal week 2; Bottom Left, postnatal week 4; Bottom Right, postnatal week 8).

Fig. S2.

Dissection of uninjured and fractured bone suggests regional f-BCSP activation. (A) FACS analysis showing CD49f expression of BCSP populations isolated from distinct femoral regions: femora from uninjured mice (Top Left), uninjured femora contralateral to a fractured femur (Top Right), callus tissue from a unilaterally fractured mouse (Bottom Left), callus tissue from a bilaterally fractured mouse (Bottom Middle), and noncallused tissue from fractured bone (Bottom Right). The population percentage of CD49f⁺ BCSPs is shown on each plot. Results show that CD49f⁺ BCSP expansion occurs around the site of injury. (B) Schematic showing CD49f⁺ f-BCSP expansion at the site of injury. (C) Staining for alkaline phosphatase activity shows enhanced osteogenic potential of CD49f⁺ f-BCSPs versus CD49f⁻ BCSPs in vitro. (Scale bars: 200 μm.) (D) The osteogenic potential of FACS-isolated BCSPs harvested from distinct femoral regions (Fig. S2A) was assayed using alizarin red staining in vitro. Micrographs (Left) are shown with corresponding spectrophotometric absorbance quantification (Right). Osteogenic activity is confined largely to the callus region and is significantly higher in CD49f⁺ BCSPs versus CD49f⁻ BCSPs (****P < 0.0001; *P < 0.05, n = 3, Student’s t test). (Scale bars: 200 μm.)

Discussion

Our experiments show that injury-induced BCSP expansion precedes ossified callus formation shortly after injury, suggesting BCSP mediation of osseous healing. Recent work by Tabin and coworkers suggests that irradiation induces skeletal progenitor dysfunction that disrupts skeletal development (35). Here, we show that hindlimb irradiation before fracture suppresses f-BCSP expansion and delays callus formation. Thus, we propose that postnatal skeletal repair depends on a functional progenitor response at the fracture site.

We also describe injury-induced variations between BCSP phenotypes. Of particular note, the in vivo intrinsic bone-forming capacity of f-BCSPs resembles that of P3 u-BCSPs more closely than that of adult u-BCSPs. These findings suggest that f-BCSPs maintain perinatal phenotypes in the postnatal adult skeleton. This phenomenon could occur via the activation of all tissue-resident BCSPs or the expansion of specific BCSPs with greater regenerative capacity.

Phenotypic variations between BCSPs isolated during different periods of skeletal growth are supported by microarray data. Notably, whereas expression of skeletogenic factors remains high in all groups, f-BCSPs show significantly greater osteogenic capacity than u-BCSPs. Perhaps, differential gene expression, including the down-regulation of osteogenic pathway antagonists, the recapitulation of embryonic skeletogenic factors, and/or the mediation of tissue revascularization, augments the skeletogenic potential of f-BCSPs. Comprehensive transcriptional analyses of BCSP-mediated repair, although beyond the scope of our study, is amenable to future examination.

Of particular importance, however, is the differential expression of CD49f. We show that fracture results in local expansion of CD49f⁺ BCSPs, as observed in the callus proper and to a lesser degree in noncallused tissue of the fractured femur. BCSP expansion is not seen in the contralateral uninjured femur (Fig. S2). CD49f expression enables the segregation of pure human HSCs from multipotent progenitors (36); however, our laboratory demonstrated that blocking its function in mouse HSC was inconsequential (37). Perhaps CD49f expression identifies a more regenerative BCSP subpopulation in the postnatal skeleton.

Our findings indicate that a specific skeletal progenitor (termed f-BCSP) plays a key role in postnatal skeletal repair in mice. The f-BCSP is distinct from progenitors isolated from uninjured bone and is characterized by CD49f⁺. CD49f⁺ f-BCSPs have enhanced regenerative potential and are active during acute periods of repair. A better understanding of the requisite stimuli for bone repair offers translational promise because cell-specific activation could enhance osteogenesis amid dysfunctional bone health.

Materials and Methods

Mice.

C57BL/6, C57BL/6-β-actin-EGFP, C57BL6/Rag-2/γ(c)KO mouse strains were derived and maintained in our laboratory. Animals used in this experiment were 8-wk-old male mice. All animals were maintained in Stanford University Laboratory Animal Facilities in accordance with Institutional Animal Care and Use Committee and National Institutes of Health (NIH) guidelines. All procedures in this experiment were performed in accordance with Stanford Administrative Panel on Laboratory Animal Care Protocol 27757.

Mouse Femoral Fracture.

Mice were anesthetized using inhalational isoflurane. Analgesia was given before medial parapatellar incision. The patella was dislocated laterally to expose femoral condyles. The medullary cavity was reamed with a 26.5G needle (BD) before insertion of an intramedullary pin of equal diameter. The femur was fractured at its midpoint using a scissors. The pin remained in situ to stabilize the fracture and allow for postoperative mobility. The patella was relocated, muscles were reapproximated, and the skin was closed using a 6/0 nylon suture.

X-Ray Assessment.

Femoral fractures were followed using a SPECTRAL Ami X Optical Imaging Platform with a 40-kV variable energy X-ray source (Spectral Instruments Imaging, LLC). Mice were anesthetized using inhalational isoflurane and pronated on the scanning platform, with the fractured limb abducted and the knee joint aligned flat at 90°. A high-resolution digital image was acquired. Mice were scanned weekly for 7 wk.

Isolation of Skeletal Progenitor Cells.

Complete, uninjured femora from uninjured mice and dissected fracture calluses from postfracture days 3, 7, 14, 21, and 28 were harvested. Each sample was crushed separately using mortar and pestle and placed in independent conicals with 10 mL of collagenase at room temperature. Each sample underwent three serial enzymatic digestions in collagenase with DNase at 37 °C for 30 min under gentle agitation. Dissociated cells were filtered through a 40-μm nylon mesh, pelleted at 200 × g at 4 °C, resuspended in staining media [2% (vol/vol) FBS in PBS], and layered onto a histopaque gradient before centrifugation at 500 × g for 15 min at room temperature with zero acceleration. The cloudy cellular interphase was pipetted off, washed with staining media, and centrifuged. The pellet was stained with fluorochrome-conjugated antibodies against CD45, Tie2, α_v-integrin, CD105, and Thy 1.1 for purification by flow cytometry (14). An additional stain for CD49f was included as needed. To determine cell frequency, each sample was analyzed separately. Samples were pooled for in vitro and in vivo experiments.

In Vitro Osteogenic Potential.

FACS-isolated BCSPs were harvested from uninjured femora and postfracture day 7 calluses. Then, 10,000 cells were plated in triplicate and were maintained in MEMα, 20% FBS, and 1% Penstrep. Once 80% confluent, cells were transferred to osteogenic differentiation media (ODM) (38). Two stains were used. Alkaline phosphatase activity was measured after 7 d in ODM. Alizarin red staining was done after 14 d in ODM. Results were imaged at 5× magnification. To quantify the alizarin red stain, each well was washed briefly in PBS, followed by a 15-min wash in 20% methanol/10% acetic acid in ddH₂O under gentle agitation to lift the stain. Absorbance was measured via spectrophotometry at 450 nm.

Colony-Forming Units.

Five hundred cells of each population were plated in triplicate and maintained in MEMα, 10% FBS, and 1% Penstrep for 2 wk before colony counting.

XTT Assay.

A trypan blue exclusion assay was used to assess cell viability. Viable cells were counted using light microscopy. The [2,3-bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide] (XTT) assay (Cell Proliferation Kit II XTT; Roche Applied Science) was done per the manufacturer's instructions. Briefly, 1,000 cells of each population were plated in triplicate and maintained in culture medium. Proliferation was evaluated every 12 h. The absorbance of each well was determined using a microplate reader at 492 nm and 690 nm (SpectraMAX 384 Plus; Molecular Devices Ltd.).

Transplantation of Skeletal Progenitor Subsets and Analysis of Kidney Grafts.

FACS-isolated BCSPs were pelleted and resuspended in 2 μL of Growth Factor Reduced Matrigel (BD Biosciences). Each population was transplanted beneath independent kidney capsule of anesthetized age- and sex-matched immunocompromised mice. Engrafted kidneys were harvested and imaged after 4 wk. Kidneys were fixed in 2% paraformaldehyde overnight at 4 °C, decalcified for 2 wk at 4 °C using 19% EDTA solution, and embedded in optimal cutting temperature for cryosectioning. Representative sections were stained with Movat’s Pentachrome stain (39). To quantify kidney grafts, analytical software (Adobe Photoshop CS6) was used to measure pixel area of engrafted bone specimens. Images are 35× magnification.

Microarray Analyses of Bone Marrow Stromal Progenitors.

Microarray analyses were done on highly purified, double-sorted cell populations. Each population was sorted into TRIzol (Life Technologies) in three independent sorts. RNA was isolated with RNeasy Micro Kit (Qiagen) per the manufacturer’s instructions. RNA was amplified twice with a RiboAmp RNA amplification kit (Arcturus Engineering). Amplified cRNA was streptavidin-labeled, fragmented, and hybridized to Affymetrix 430–2.0 arrays (Affymetrix). Arrays were scanned with a Genechip Scanner 3000 (Affymetrix) running GCOS 1.1.1. software. Scanned data were exported to DCHIP software for normalization. Raw microarray data were submitted to Gene Expression Commons (https://gexc.stanford.edu/) (21), where data normalization was computed against the Common Reference, a large collection (n = 11,939) of publicly available microarray data from the National Center for Biotechnology Information Gene Expression Omnibus. Metaanalysis of the Common Reference also provides the dynamic range of each probe set on the array. The probe set with the widest dynamic range was used for analysis. The Affymetrix Mouse Genome 430 2.0 Array includes 45,101 probe sets, of which 17,872 annotated genes are measurable. Heat maps representing fold change of gene expression were generated in Gene Expression Commons.

Hindlimb Irradiation.

Eight-week-old C57BL/6 mice were placed in a radiation shield to expose only the hindlimbs. Mice received a single dose of 800 rad (8 Gy). Surgical fracture induction was done 12 h postirradiation.

Immunofluorescence Staining.

BCSP-derived colonies were fixed with paraformaldehyde and labeled using a standard immunofluorescence staining protocol with monoclonal antibody to mouse collagen II (NeoMarkers) and monoclonal antibody to osteocalcin (Abcam). Alexa dye-conjugated secondary antibodies were purchased from Molecular Probes. After three rounds of washing with staining buffer, secondary antibodies were added. Samples were washed with PBS, cover-slipped, and imaged with a Leica DMI6000B epifluorescence-equipped inverted microscope.

FACS Analysis of Annexin V-Positive Cells.

Quantification of Annexin V-positive BCSPs was conducted using an Annexin V: FITC Apoptosis detection kit II (BD). Cells were stained per the manufacturer's instructions and analyzed on a BD FACS ARIA II FACS Sorter.

Statistical Analysis.

All statistical analysis was performed in GraphPad Prism (GraphPad Software). Data analysis was performed using one-way ANOVA and post hoc Tukey correction or Student’s t test assuming two-tailed distribution and unequal variances. The use of an asterisk denotes statistical significance of P < 0.05 between the sample groups described. In graphs with more than two groups, statistical analysis describes the two groups connected by a bar line. All results are shown as mean ± SEM. All samples represent at least n = 3.