Analysis of αSMA-Labeled Progenitor Cell Commitment Identifies Notch Signaling as an Important Pathway in Fracture Healing



Fracture healing is a regenerative process that involves coordinated responses of many cell types, but characterization of the roles of specific cell populations in this process has been limited. We have identified alpha smooth muscle actin (αSMA) as a marker of a population of mesenchymal progenitor cells in the periosteum that contributes to osteochondral elements during fracture healing. Using a lineage tracing approach, we labeled αSMA-expressing cells, and characterized changes in the periosteal population during the early stages of fracture healing by histology, flow cytometry, and gene expression profiling. In response to fracture, the αSMA-labeled population expanded and began to differentiate toward the osteogenic and chondrogenic lineages. The frequency of mesenchymal progenitor cell markers such as Sca1 and PDGFRα increased after fracture. By 6 days after fracture, genes involved in matrix production and remodeling were elevated. In contrast, genes associated with muscle contraction and Notch signaling were downregulated after fracture. We confirmed that activating Notch signaling in αSMA-labeled cells inhibited differentiation into osteogenic and adipogenic lineages in vitro and ectopic bone formation in vivo. By characterizing changes in a selected αSMA-labeled progenitor cell population during fracture callus formation, we have shown that modulation of Notch signaling may determine osteogenic potential of αSMA-expressing progenitor cells during bone healing. © 2014 American Society for Bone and Mineral Research.


Fracture healing is a complex process that usually results in the generation of new bone and connective tissue with similar anatomy and functionality to the pre-injury site. Immediately after injury, a hematoma forms, followed by an inflammatory response at the fracture site mediated by cells such as platelets, macrophages, and neutrophils, initiating the repair process.[1] Within days of fracture, the periosteum, if intact, thickens because of extensive cell proliferation, and callus formation begins.[2, 3] Fractures that are mechanically stable heal via a process similar to intramembranous bone formation; however, if there is some instability, a fibrocartilage-based callus will form and undergo endochondral ossification.[4] Bone formation begins proximal and distal to the fracture, whereas cartilage is prominent near the fracture site.[1] After cartilage mineralization, a hard callus forms and is ultimately remodeled to regain the original bone anatomy. Although many fractures heal well with standard clinical treatment, those with excess trauma, or in osteoporotic patients, may exhibit excessively long healing times, resulting in significant morbidity.[5] Developing an understanding of the mechanisms of normal fracture healing by defining the cell populations involved and identifying how they are recruited and expanded is important for identifying approaches to improve fracture healing outcomes.

Multiple cell lineages are important in the fracture healing process. Initially, hematopoietic cells are necessary for hematoma formation and the inflammatory response. Minimizing hematoma formation by removing bone marrow delays the periosteal response.[6] Other important hematopoietic lineage cells include macrophages, which appear to modulate bone formation during bone healing, and osteoclasts, which are required for the remodeling phase.[7] Endothelial cells and the development of vasculature are also important for healing, and vascular invasion of the cartilaginous callus is critical for osteogenesis and remodeling to occur.[8] Mesenchymal progenitors form the osteochondral elements of the callus. The source of these cells is likely the local environment, and the periosteum is generally considered the primary source of cells.[9, 10] Mesenchymal progenitors from the bone marrow and muscle can also contribute to fracture healing, particularly in severe fractures where there is extensive soft tissue injury and the periosteum is compromised.[10, 11]

There has been a concerted effort in recent years to identify markers of mesenchymal stem cells and characterize their location in vivo. Mesenchymal lineage cells are generally defined as cells that do not express CD45 and other hematopoietic lineage markers such as Ter119 and CD11b, or endothelial markers such as CD31.[12] Markers that have been utilized to positively identify mouse mesenchymal stem/progenitor cells from bone marrow and other tissues include Sca1/Ly6a,[13] PDGFRα,[14] CD105,[15, 16] leptin receptor,[17] nestin,[18] osterix,[19] and combinations thereof. Additional markers in human cells include CD146[20] and STRO1.[21] Mesenchymal stem/progenitor cells appear to reside in a perivascular niche in many tissues in vivo.[22, 23] Alpha smooth muscle actin (αSMA) is a marker of pericytes and myofibroblastic cells that identifies mesenchymal progenitor cells with proliferative and multi-lineage differentiation potential in bone marrow stromal cell cultures in vitro,[24] and osteoblast precursors in the bone marrow and periodontium in vivo.[25, 26]

In this study, we have utilized an αSMACreERT2 mouse combined with a Cre-activated fluorescent reporter to isolate a specific cell population involved in fracture repair, and have examined cell surface markers and gene expression profiles in this population during the early stages of fracture healing. The results indicate numerous changes in gene expression involved in cell proliferation, upregulation of inflammatory markers, and alterations in cell signaling pathways. We have identified downregulation of Notch signaling pathway components during the early commitment of periosteal progenitors to osteochondral elements after fracture. Our results confirm that Notch signaling alters the differentiation capacity of αSMA-expressing mesenchymal progenitor cells.

Materials and Methods

Mouse strains

All animal procedures were approved by an institutional animal care and use committee. The αSMACreERT2[25] and Col2.3GFP mice[27] were previously described. Ai9 reporter mice (B6;129S6-Gt(ROSA)26Sortm9(CAG-tdTomato)Hze/J, stock # 007905), Rosa-NICD mice (Gt(ROSA)26Sortm1(Notch1)Dam/J, stock # 008159), and immunodeficient NSG mice (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ, stock # 005557) were obtained from Jackson Laboratory (Bar Harbor, ME, USA). αSMACreERT2 animals were bred with Ai9 to generate SMACre/Ai9 mice (termed SMA9). Genotyping for Cre and NICD (using GFP primers) was performed with the PCR primers shown in Supplemental Table S1. Detection of recombination of the NICD construct was performed as previously described using primers in Supplemental Table S1.[28]


To label αSMA-expressing cells, tamoxifen was administered at a dose of 75 µg/g of body weight the day before and day of fracture. Closed transverse diaphyseal fractures of either the right tibia or both tibias were created in 3- to 5-month-old mice as described.[25] Briefly, before fracture, a 0.38-mm-diameter stainless steel pin was inserted into the medullary canal. Fractures were created 1 to 2 mm proximal to the distal tibia-fibula junction using a drop-weight blunt guillotine device. X-ray (Faxitron LX-60) was used to confirm pin and fracture placement.


Bones were fixed in 10% formalin for 5 days at 4°C, the intramedullary pins were removed, and bones were decalcified in 14% EDTA for 7 days, incubated overnight in 30% sucrose/PBS, and embedded in Cryomatrix (Thermo Fisher Scientific, Waltham, MA, USA). Sections (7 µm) were obtained on a cryostat (Leica, Wetzler, Germany) using a tape transfer system (Section-lab, Hiroshima, Japan). Imaging was performed using an Observer.Z1 microscope (Carl Zeiss, Thornwood, NY, USA) as previously described.[25]

Flow cytometry

Periosteum or callus tissue pooled from 4 to 9 sex-matched mice was used to sort SMA9+ cells for RNA extraction or perform cell surface marker analysis. Unfractured controls were collected 2 days after tamoxifen injections, whereas fractured samples were obtained 2 and 6 days after fracture. Tibias were dissected, muscle and tendon tissue was removed, and bone marrow flushed. The periosteum or periosteum/fracture callus was scraped from the diaphyseal cortex, then digested in PBS containing 0.125% trypsin (Lonza, Walkersville, MD, USA), 0.1% collagenase A (Roche, Mannheim, Germany), 0.1% hyaluronadase (Sigma, St. Louis, MO, USA) for 1 hour at 37°C with agitation. Cell surface marker analysis was performed using antibodies listed in Supplemental Table S2. Staining was performed in 1× Hank's balanced salt solution, 10 mM HEPES, 2% fetal bovine serum (FBS), and, after secondary antibody incubation (where applicable), cells were stained with Live/Dead Fixable Near-IR Dead Cell Stain (Life Technologies, Carlsbad, CA, USA). Cell sorting and analysis was performed using a FACSAria II (BD Biosciences, San Jose, CA, USA). Voltages and gates were set based on unstained periosteum samples from Cre-negative animals. At least 104 cells per sample were sorted for RNA extraction.

RNA extraction and gene expression analysis

Sorted cells were resuspended in Trizol reagent (Life Technologies), and after addition of 20 µg glycogen (Roche), RNA was purified. RNA quality was verified on a Bioanalyzer RNA Nano Chip (Agilent Technologies, Santa Clara, CA, USA), and biotinylated cRNA was generated from 10 to 250 ng RNA using the MessageAmp Premier RNA Amplification Kit (Life Technologies), according to the manufacturer's instructions. Biotinylated cRNA was hybridized to MouseRef-8 BeadChips (Illumina, San Diego, CA, USA) for 16 hours at 58°C. BeadChips were then washed, stained with streptavidin-Cy3, and scanned on the Illumina BeadArray Reader. Microarray data were analyzed using Illumina GenomeStudio and the PBC pattern-based clustering program[29] as described in the Supplemental Methods. Sorted cells from three independent experiments were used for microarray analysis. Real-time PCR is described in the Supplemental Methods.

Cell culture

Periosteal cells were isolated as described above, and the cells from two mice were seeded in a 100-mm dish in basal medium (αMEM 10% FBS, 1% penicillin/streptomycin) and cultured in 5% oxygen. Cre activity was induced by 1 µM 4-OH-Tamoxifen administered on day 2 and day 4 of culture. Once cells reached confluence, SMA9+ cells were sorted and replated for differentiation assays as described in the Supplemental Methods. Bone marrow stromal cell (BMSC) cultures were seeded from flushed bone marrow at a density of 6 × 107 cells/100 mm dish in basal medium. Cells were labeled and sorted as described above before replating for differentiation assays. Mineralization of osteogenic cultures was evaluated by von Kossa staining, adipogenesis by oil red O staining, and chondrogenesis by alcian blue staining as previously described.[30]

Subcutaneous implantation assay

BMSCs were cultured as described, and 2 × 106 cells were suspended in a 100-µL gel composed of 3 mg/mL rat type I collagen (BD Biosciences).[31, 32] Gels were implanted into NSG mice subcutaneously (2 to 4 gels containing cells of the same experimental group). Implants were harvested after 3 weeks, fixed in 4% paraformaldehyde for 4 hours, scanned by X-ray and dual-energy X-ray absorptiometry (DXA; PIXImus, GE-Lunar, Madison, WI, USA), then histology performed without decalcification. After fluorescence imaging, sections were stained with von Kossa and counterstained with hematoxylin, then reimaged. Counts of tdTomato+ cells and mineralized area were determined using ImageJ (NIH, Bethesda, MD, USA).


αSMA-labeled periosteal cells contribute to fracture callus formation

The SMA9 mice used in this study combine a tamoxifen-inducible Cre transgene driven by the αSMA promoter and a Cre-activated fluorescent reporter. Tamoxifen permanently labels cells expressing Cre and their progeny. To evaluate the molecular events occurring early in the fracture healing process in this cell population, we first examined fracture histology at early time points and cell distribution in the periosteum of unfractured bones. In intact tibias of adult SMA9 mice, tamoxifen labeled a population of cells with fibroblast morphology in the periosteum (Fig. 1A–C). The periosteum also contains mature osteoblasts that express the Col2.3GFP transgene. The SMA9+ cells were separate from this population in unfractured bone (Fig. 1C), confirming that they were immature progenitor cells. During the inflammatory phase of repair, 2 days after fracture, there was evidence of periosteal thickening, which included expansion of the SMA9+ cell population (Fig. 1D, E). SMA9+ cells were apparent in the periosteum both proximal and distal to the fracture site. Six days after fracture, a soft callus has formed, and SMA9+ cells comprise a large proportion of cells with chondrocyte morphology in the callus (Fig. 1F, H). Bone formation has also begun farther from the fracture site, and SMA9+ osteocytes are apparent (Fig. 1G). At later time points, SMA9+/Col2.3GFP+ osteoblasts and osteocytes are present in fracture calluses.[25]

Figure 1.

Fracture histology of SMA9 mice. The distribution of SMA9+ cells in the tibia was determined histologically. (A) Histology of unfractured bone 2 days after tamoxifen induction indicates that SMA9+ cells (red) are present in the periosteal layer, particularly in the proximal end of the bone (left side). The bone structure is indicated by green autofluorescence and DIC. Higher magnification of the boxed area (B) indicates labeled elongated cells in the periosteal layer (arrowheads). DAPI-labeled nuclei are blue. (C) SMA9+ periosteal cells (arrowheads) are distinct from mature osteoblast lineage cells as indicated by separation from Col2.3GFP-labeled green cells on the bone surface in SMA9/Col2.3GFP mice. (D) Histology of a day 2 fracture indicates that SMA9+ cells are present on the periosteal surface proximal and distal to the fracture site, and expansion of these cells can be observed in areas of periosteal thickening (boxed area, E). (F) Histology of a day 6 fracture indicates marked expansion of the SMA9+ population and contribution to multiple skeletal elements within the fracture callus including osteoblasts and osteocytes (arrows, G) and chondrocytes (H). A, D, and F are composites of scanned images (scale bar = 500 µm). The scale bar on B represents 50 µm on all other images. cb = cortical bone; m = muscle.

αSMA-labeled periosteal cells express progenitor cell markers after fracture

To characterize the labeled cell population, we performed flow cytometry surface marker analysis of periosteal cell populations. The periosteal cell preparations contained a high percentage of hematopoietic lineage cells (expressing CD45 and/or mature hematopoietic lineage markers, Lin+), possibly as a result of contaminating bone marrow cells. SMA9+ cells comprised more than 20% of nonhematopoietic periosteal cells in the unfractured control samples and 2 days after fracture (Fig. 2A). By 6 days after fracture, cell yields per bone increased, the proportion of nonhematopoietic cells in the analysis was much higher (63.6% compared with 12.9% in unfractured), and the proportion of SMA9+ cells in the nonhematopoietic cell population doubled to 40%, indicating expansion of SMA9+ cells. Further analysis of cell surface markers was performed on SMA9+ and SMA9 populations. The SMA9+ cells were almost all nonhematopoietic cells (Fig. 2B). This is consistent with our previous analysis of SMA9+ periosteal cells and differs from the αSMA-expressing population in the bone marrow, which are mainly myeloid lineage cells.[25, 33] We subsequently analyzed expression of a number of markers of mesenchymal stem/progenitor cells within the CD45- Lin- population (Fig. 2C, Supplemental Fig. S1). Of the markers examined, Sca1 was expressed in the highest percentage of cells at all time points for SMA9+ (9 ± 0.4% for unfractured, 29 ± 1.2% for day 2, 26 ± 0.4% for day 6) and SMA9 (28 ± 3.1% for unfractured, 58 ± 1.7% for day 2, 35 ± 1.1% for day 6) populations. Putative mesenchymal stem cell markers PDGFRα and leptin receptor and pericyte/mesenchymal marker PDGFRβ[14, 17, 34] were only expressed in approximately 2% of SMA9+ cells from unfractured bones, but the proportion of cells positive for these markers increased after fracture to 16% to 18% at day 6. The proportion of cells expressing mesenchymal progenitor cell markers in the SMA9 population was very low and remained fairly constant after fracture (Supplemental Fig. S1). Very few cells in the CD45- Lin- periosteal preparations expressed CD31, suggesting they were mainly mesenchymal and not endothelial lineage. However, a different endothelial marker, MECA32 (Plvap), was present on a subset of SMA9+ cells and was expressed on more cells after fracture (Fig. 2C).

Figure 2.

Flow cytometry analysis of periosteal cells from fractured tibias of SMA9 mice. SMA9 mice treated with tamoxifen had periosteum harvested 2 days later (unfractured control) or underwent fracture and periosteum/fracture callus tissue was harvested 2 or 6 days later. Cells were stained for surface markers, and flow cytometry analysis performed. (A) The frequency of SMA9+ cells within the nonhematopoietic component was determined by gating CD45- and mature hematopoietic lineage marker negative cells (upper panel), then determining the percentage of SMA9+ tdTomato expressing cells (lower panel). (B) Analysis of SMA9+ and SMA9- periosteal cells for CD45 and hematopoietic lineage markers. (C) Expression of mesenchymal stem/progenitor and endothelial markers in SMA9+ CD45- Lin- cells. Data from a representative experiment are presented.

Gene expression profile of αSMA-labeled periosteal cells after fracture

To further characterize the SMA9+ periosteal cell population during fracture healing, we isolated SMA9+ cells using flow cytometry for RNA extraction and microarray analysis at the same time points previously characterized (Fig. 3A, B). A summary of the changes in gene expression at different time points is shown in Table 1. The complete data set is available from the GEO repository, accession number GSE45156. A total of 189 genes showed statistically significant changes in expression between the unfractured control and day 2 of fracture; however, the magnitude of these changes tended to be small. Many more changes are apparent by day 6 of fracture, with 782 genes differentially expressed compared with the unfractured control. Expression of a number of lineage marker genes is shown in Fig. 3C. Endogenous αSMA expression was high in unfractured samples but reduced after fracture, consistent with its downregulation during differentiation. This result was confirmed by real-time PCR, which indicated that αSMA expression was sixfold and eightfold lower at day 2 and day 6, respectively, compared with the control (data not shown). Expression of a number of osteoblast lineage markers was elevated by day 6, including osterix, bone sialoprotein, and Col1a2, consistent with the histological appearance of SMA9+ osteoblasts by this time. Chondrocyte markers Col2a1, aggrecan, and Sox9 were highly elevated at day 6, consistent with histological observations, and hypertrophic chondrocyte marker Col10a1 was also elevated, albeit to a lesser degree than the other cartilage matrix proteins. To identify groups of genes that were differentially regulated, gene ontology analysis was performed. Representative categories are shown in Table 2. Analysis indicated that 2 days after fracture, expression of numerous cytokines and chemokines was increased, including TNFα and IL1β (Supplemental Table S3). In addition, a number of genes associated with cell proliferation are upregulated (Supplemental Table S4), consistent with the marked expansion of this cell population between day 2 and day 6 of fracture. By day 6, many of the upregulated genes are associated with matrix deposition and remodeling. It is notable that numerous matrix genes and matrix metalloproteinases and their inhibitors are differentially regulated in these cells after fracture (Supplemental Table S5). Most of these changes occur at day 6, but genes such as matrix gla protein and fibromodulin show transient downregulation at day 2. Groups of genes that were expressed at higher levels in the control samples than at either time point after fracture included genes involved in muscle contraction, blood vessel development, and components of the Notch signaling pathway. Because regulation of Notch signaling is important for bone and cartilage development but its role in fracture healing is poorly defined,[35-37] expression of Notch signaling components was confirmed by real-time PCR (Fig. 4). This indicated that the receptors Notch1–4 were all reduced in expression post fracture, as were two of the canonical target genes Hes1 and Hey1. The pathway ligand Jagged2 was also downregulated, whereas Jagged1 expression did not change consistently. Most of the changes in Notch pathway components were specific to the SMA9+ population, with only Notch1 showing significantly altered expression in unsorted or SMA9 periosteal cell preparations (Supplemental Fig. S2).

Figure 3.

Cell sorting and gene expression analysis of SMA9+ periosteal cells from fractures. (A) Sorting of SMA9+ cells from total unstained periosteal cell preparations was performed based on the gates indicated. (B) Reanalysis of sorted cells indicated significant enrichment of SMA9+ cells at all time points. A representative experiment is shown. (C) Expression of selected skeletal lineage marker genes in sorted SMA9+ cells as determined by microarray. Data are pooled from 3 replicates, and where genes were represented by multiple probes, representative data are shown. *p < 0.05 compared with unfractured control.

Table 1. Summary of Gene Expression Changes in SMA9+ Cells After Fracture
p < 0.05>5-folda>2-foldap < 0.05>5-folda>2-folda
  1. aGenes showing indicated fold change regardless of p value.
Control–day 21375205529347
Control–day 66124336117039441
Day 2–day 6 44281 17249
Table 2. Gene Ontology Analysis of Microarray Data
Gene ontology termp valueOdds ratioExpected gene countActual gene countTotal genes
  1. aAlso represented in the analysis of genes downregulated at day 2.
Upregulated at fracture day 2
Immune response9.7E-138.73.322387
Response to external stimulus3.2E-117.13.922462
Regulation of protein kinase cascade4.0E-048.90.6574
Upregulated at fracture day 6
Phosphate transport1.8E-1114.51.31475
Cell adhesion2.8E-093.99.331556
Antigen processing and presentation of peptide antigen via MHC class II2.2E-0633.60.2514
Cartilage condensation4.6E-0627.50.3516
Skeletal development2.1E-054.82.812166
Collagen fibril organization8.8E-0521.90.3415
Collagen catabolic process8.8E-0521.90.3415
Cellular carbohydrate metabolic process1.8E-043.83.412207
Downregulated at fracture day 6
Muscle contractiona1.6E-1220.50.81372
Regulation of muscle contractiona5.3E-1250.90.3925
Blood vessel morphogenesisa2.0E-087.92.114179
Metal ion transport3.1E-064.24.617393
Activation of protein kinase C activity2.4E-0432.50.1311
Response to wounding5.5E-043.63.411289
Figure 4.

Expression of Notch pathway components in SMA9+ periosteal cells after fracture. Expression of Notch pathway genes was determined using real-time PCR. The average expression of unfractured control samples was normalized to 1. Data are pooled from 2 to 3 replicates. *p < 0.05 compared with control determined by one-way ANOVA with Dunnett's post test.

Notch pathway activation inhibits differentiation of αSMA-labeled cells

To examine the role of Notch signaling in SMA9+ cells, we used a model of Cre-induced forced Notch signaling, the Rosa-NICD mouse. In the presence of Cre activity, a floxed stop cassette is excised and expression of the Notch1 intracellular signaling domain (NICD) occurs under the control of the ubiquitous Rosa26 promoter, resulting in constitutive activation of Notch signaling.[38] These mice were crossed with αSMACreERT2 mice and Ai9 reporter mice, allowing concomitant fluorescent labeling of SMA9 cells where Notch signaling activation should also have occurred (termed SMA9/NICD). We examined the effect of Notch signaling activation on differentiation in the SMA9+ compartment of 4-OH-Tamoxifen–treated periosteal cultures from either SMA9/NICD cultures (activated Notch) or SMA9 littermates (Control). Expression of a Notch target gene, Hey1, was highly elevated in the activated Notch cultures compared with the control, indicating that the NICD expression resulted in Notch pathway activation (Fig. 5A). Recombination of the transgene was also confirmed in the majority of SMA9/NICD+ cells (Fig. 5B). Osteoblast differentiation was blocked in activated Notch periosteal cultures, as indicated by a lack of induction of bone sialoprotein and osteocalcin expression (Fig. 5C), which accompanied culture overgrowth and contraction of the cell layer (data not shown). Adipogenesis occurred in control periosteal cultures; however, activation of Notch signaling completely blocked adipogenesis as indicated by Oil Red O staining and adiponectin and adipsin expression (Fig. 5D, E). These results were confirmed in BMSC cultures. We have shown that SMA9+ BMSCs are capable of differentiation into osteoblasts, adipocytes, and chondrocytes, whereas SMA9 cells have limited differentiation potential.[25] Hey1 expression was elevated in the tdTomato+ activated Notch cultures compared with the control cultures (Supplemental Fig. S3A), although not to the same extent as in periosteal cultures. Equivalent tdTomato- cultures did not show elevated Hey1 expression from activated Notch cultures compared with controls, indicating that the cell sorting enriched for cells where Notch signaling was activated (Supplemental Fig. S3B). Activated Notch cultures showed minimal osteogenic (Supplemental Fig. S3C, D) and chondrogenic differentiation (Supplemental Fig. S3E, F). Adipogenesis was also reduced in activated Notch cultures compared with controls; however, some adipogenesis did occur (Supplemental Fig. S3G, H). To investigate the effect of Notch activation in these cells in vivo, BMSC and periosteal cells were implanted subcutaneously in collagen gels. Periosteal cells did not form ectopic bone under these conditions (data not shown), but BMSCs with activated Notch driven by αSMACreERT2 formed ossicles with different structure, lower bone mineral density, and reduced mineralization compared with control implants (Fig. 6). In addition fewer SMA9+ cells were embedded in mineralized tissue in activated Notch ossicles (Fig. 6D, F), suggesting these cells were less likely to undergo terminal osteogenic differentiation. These results show that differentiation of SMA9+ progenitor cells is inhibited or blocked in the presence of Notch signaling activation.

Figure 5.

Notch signaling activity and differentiation potential in periosteal cultures from SMA9/NICD mice. SMA9+ periosteal cells from SMA9/NICD mice (Activated Notch) and SMA9 littermates (Control) were replated for differentiation assays. Notch pathway activation was evaluated by expression of target gene Hey1 (A). Cre-induced loxP recombination was confirmed by PCR (B), with the intact construct indicated by a 492-bp band, as in untreated animals (lane 1), whereas recombination (296-bp band) is evident in treated unsorted (lane 2) and SMA9+ (lane 3) cultures. Under osteogenic differentiation conditions, tdTomato+ control cultures underwent differentiation as indicated by induction of bone sialoprotein and osteocalcin expression, whereas activated Notch cultures showed minimal increases in expression (C). Adipogenesis was evaluated by morphology (D, upper panel, tdTomato fluorescence overlaid by phase contrast) and oil red O staining (D, lower panel). This was confirmed by expression of adipocyte markers adiponectin and adipsin (E). Data are from an experiment representative of 2 to 3 biological replicates. Real-time PCR data are normalized to expression in the control confluent (Conf.) culture. *p < 0.05 compared with equivalent control sample determined by Student's t test. Scale bar = 50 µm. Conf. = confluent culture; OB = osteogenic culture; AD = adipogenic culture.

Figure 6.

Effect of SMA9-driven Notch signaling on ectopic bone formation. Unsorted SMA9 (Control) and SMA9/NICD (Activated Notch) BMSC cultures were embedded in collagen gels and implanted subcutaneously in NSG mice. Implants were harvested 3 weeks later. (A) X-rays indicate qualitative structural differences in the ossicles formed from the different cultures. (B) Bone mineral density (BMD) was determined by DXA. (C) Mineralized area was determined using sections stained with von Kossa. Representative sections from each group are shown (E). (D, F) tdTomato+ cells embedded in mineralized tissue, identified by DIC imaging, were counted (see green arrowheads) and the number per mm2 of mineralized tissue were calculated. n = 6 implants/group. Image analysis was performed on 3 to 6 sections per implant. *p < 0.05 determined by Student's t test.


Fracture healing is a complex process that involves many cell lineages. Numerous studies have been performed analyzing global gene expression patterns in whole bones or total fracture calluses during healing.[39-44] Although these studies have provided insight into genes and pathways important during the healing process, interpretation of the results is difficult because there are massive changes in the types and numbers of cells present in addition to gene regulation events. As a result, one study reported that over the course of 3 weeks of fracture healing in mice, more than 50% of expressed genes in the genome were differentially regulated.[42] In the present study, we were able, for the first time, to analyze gene expression in a subpopulation of cells involved in fracture healing as they expanded and began to differentiate. Given that the periosteum is an important source of cells during fracture healing, we focused on cells from this compartment. We have shown that the cells labeled by αSMACreERT2 in the periosteum are mesenchymal progenitors that give rise to osteoblasts and chondrocytes within the fracture callus. These cells appear to respond to fracture by upregulating genes associated with proliferation as well as numerous cytokines and chemokines. These changes presumably occur in response to hematoma formation, which has been shown to be important for periosteal response,[6] and suggest that SMA9+ cells contribute to the production of cytokines and chemokines after fracture. By day 6 after fracture, SMA9+ cells are clearly contributing to chondrogenic elements within the callus, as well as osteoblasts, and this is illustrated in the microarray data by large increases in cartilage genes and upregulation of a variety of genes involved in skeletal development, matrix deposition, and collagen organization. Many of the genes downregulated after fracture were associated with muscle contraction and blood vessel morphogenesis. Downregulation of muscle contraction-related genes after fracture is likely to be indicative of the vascular smooth muscle/pericyte phenotype associated with αSMA expression that is lost after fracture once the cells differentiate toward osteochondral lineages. SMA9+ cells are also present in the muscle compartment; however, lineage tracing data labeling muscle cells with MyoD-Cre indicates that muscle-derived cells only make a major contribution to fracture callus formation when the periosteum is removed, implying that the vast majority of SMA9+ cells contributing to healing in our model are derived from the periosteum.[11]

The periosteum is a tissue that responds rapidly to injury, makes a major cellular contribution to both cartilage and bone, and promotes angiogenesis during the healing process, such that healing is impaired when the periosteum is absent.[6, 9, 10] Periosteal replacement can heal critical-sized bone defects in animal models.[45, 46] Cells isolated from the periosteum may also be suitable for tissue engineering applications, and under some conditions they show similar or greater levels of bone formation than BMSCs.[47, 48] However, the identity of mesenchymal progenitor cells within the periosteum is not well defined. A number of studies have shown that periosteal cells express markers such as CD105 and CD90 in culture, but there have been very few studies that have characterized freshly isolated cells.[49, 50] The marker used in our study, αSMA, is typically considered to be a marker of pericytes and myofibroblasts; however, given that SMA9+ cells in the periosteum show fibroblastic morphology, are highly prevalent (approximately 20% of nonhematopoietic cells by flow cytometry), and only a small proportion express pericyte marker PDGFRβ, it is unlikely that they are exclusively perivascular. This is consistent with classical studies indicating that osteoprogenitor cells contributing to appositional bone growth in young animals derive from a pool of rapidly dividing cells in the cambium layer of the periosteum.[51] Based on cell surface marker expression, the periosteal SMA9+ cell population is heterogeneous, and in intact bone less than 10% of the SMA9+ cells express Sca1 and other mesenchymal progenitor markers. After fracture, with the expansion of periosteal tissue and callus formation, the proportion of cells expressing all of the examined mesenchymal markers increased, suggesting either a large expansion of certain cell subsets or acquisition of immature markers by SMA9+ cells that did not express them previously. In contrast to the periosteum, one of the functions ascribed to bone marrow mesenchymal stem cells is support of hematopoietic stem cells, and many studies use the ability to recapitulate a hematopoietic microenvironment after transplantation to test the utility of putative markers.[15, 17, 18, 20] For periosteal cells, their osteochondral potential and ability to promote angiogenesis are critical for their contribution to bone healing and hence their potential clinical utility. Further studies are, therefore, required to clarify the cellular properties (for example, self-renewal, multipotency, hematopoietic support) associated with specific markers, and investigate whether markers that have been characterized in bone marrow are suitable for identification of stem/progenitor cell populations in periosteum.

Combining inducible Cre transgenes with a reporter is a powerful technique for tracing the fate of endogenous cells in vivo. Our studies suggest that αSMACreERT2 labels cells in vitro and in vivo without the need to administer tamoxifen over an extended period as has been reported for some inducible Cre transgenes.[16, 52] There is also essentially no leakage of αSMACreERT2 activity in musculoskeletal tissue.[25] A number of other Cre transgenes have been used to evaluate contributions of various cell populations to bone turnover and healing. Mx1-Cre can be used to label a self-renewing mesenchymal population that shows multi-lineage potential in vitro; however, this transgene also labeled other populations such as hematopoietic stem cells, making lineage tracing studies technically challenging. Mx1-Cre also labeled cells in the periosteum, although unlike the cell population labeled in our study, these cells contributed only to osteoblasts and not chondrocytes during healing of a small hole in the femur.[16] Kawanami and colleagues developed a Prx1CreER mouse in an attempt to specifically target periosteal cells.[52] Lineage tracing indicated that cells labeled by this transgene differentiated into osteoblasts and chondrocytes in a fracture callus; however, this followed administration of 10 doses of tamoxifen before and after the fracture, suggesting low expression or activity of Cre and limiting the utility of the model.

One of the notable observations in the microarray analysis was downregulation of multiple components of the Notch signaling pathway. Notch signaling is an important developmental pathway that is involved in the maintenance of stem and progenitor cell populations in many tissues. When Notch receptors interact with membrane-bound ligands on an adjacent cell, proteolytic cleavage results in release of the NICD, which interacts with cofactors in the nucleus to induce transcription of target genes in the Hes/Hey family. In the skeleton, disruption of Notch signaling accelerates chondrocyte differentiation and promotes osteoblast formation in vitro and in vivo but ultimately results in depletion of the mesenchymal progenitor pool.[36, 37] Notch signaling does not appear to be critical for the function of mature osteoblasts, but forced Notch signaling impairs their differentiation and results in osteosclerosis.[28, 37, 53, 54] There has been limited characterization of the role of Notch signaling in fracture repair, although regulation of Notch signaling is important during healing in a number of other tissues.[55-57] Dishowitz and colleagues examined expression of Notch signaling pathway components during fracture and calvarial defect repair, and found that many components of the pathway are upregulated after fracture, peaking around 10 days after injury when total callus RNA was analyzed.[35] A subsequent study showed that genetically inducible systemic inhibition of Notch signaling prolonged the inflammatory phase of fracture healing and altered cartilage formation, but it is unclear which cell types are responsible for this change.[58] In the current study, we noted downregulation of many components of the pathway in SMA9+ cells in the early phases after fracture, including all four receptors and two of the canonical target genes. These changes were specific to the SMA9+ population. We also demonstrated that forced expression of Notch in SMA9+ cells from periosteum and bone marrow blocked differentiation into osteoblast, chondrocyte, and adipocyte lineages in vitro and reduced bone formation in an in vivo implant, consistent with previous reports in osteoblastic cells.[53] Further studies are required to clarify the role of Notch signaling in different lineage populations in vivo at different stages of fracture healing.

In summary, we have utilized the SMA9 model to characterize a population of mesenchymal progenitor cells within the periosteum that are involved in fracture healing. Downregulation of Notch signaling in these cells may be important for their contribution to fracture callus formation. Further characterization of the mechanisms by which these cells and others in the periosteum can be expanded and recruited to injury sites could ultimately lead to development of therapies that improve healing mediated by endogenous cells.


All authors state that they have no conflicts of interest.


This work was supported by NIH/NIAMS AR055607-01 grant to IK. HR was supported by Croatian Science Foundation.

We acknowledge Anupinder Kaur from the UConn Health Center Genomics Core for assistance with microarray processing and analysis, and Stefano Zanotti from Saint Francis Hospital and Medical Center for providing real-time PCR primer sequences.

Authors' roles: Study design: BGM, DG, and IK. Data acquisition: BGM, DG, LW, YH, HR, DJA, and IK. Data analysis: BGM, DG, PJ, and DGS. Data interpretation: BGM, DG, and IK. Manuscript preparation: BGM and IK. Revising manuscript content: BGM, DG, DJA, and IK. Final approval of manuscript: all authors. BGM, DG, and IK take responsibility for the integrity of data analysis.