Engineering Vascularized Bone: Osteogenic and Proangiogenic Potential of Murine Periosteal Cells§


  • Nick van Gastel,

    1. Laboratory of Clinical and Experimental EndocrinologyDivision of Skeletal Tissue Engineering, KU Leuven, Leuven, Belgium
    2. Prometheus, Division of Skeletal Tissue Engineering, KU Leuven, Leuven, Belgium
    Search for more papers by this author
  • Sophie Torrekens,

    1. Laboratory of Clinical and Experimental EndocrinologyDivision of Skeletal Tissue Engineering, KU Leuven, Leuven, Belgium
    Search for more papers by this author
  • Scott J. Roberts,

    1. Prometheus, Division of Skeletal Tissue Engineering, KU Leuven, Leuven, Belgium
    2. Laboratory for Skeletal Development and Joint DisordersVesalius Research Center, KU Leuven, Leuven, Belgium
    Search for more papers by this author
  • Karen Moermans,

    1. Laboratory of Clinical and Experimental EndocrinologyDivision of Skeletal Tissue Engineering, KU Leuven, Leuven, Belgium
    Search for more papers by this author
  • Jan Schrooten,

    1. Prometheus, Division of Skeletal Tissue Engineering, KU Leuven, Leuven, Belgium
    2. Department of Metallurgy and Materials EngineeringVesalius Research Center, KU Leuven, Leuven, Belgium
    Search for more papers by this author
  • Peter Carmeliet,

    1. Laboratory of Angiogenesis and Neurovascular Link, Vesalius Research Center, VIB, Leuven, Belgium
    2. Laboratory of Angiogenesis and Neurovascular Link, Vesalius Research Center, Department of OncologyVesalius Research Center, KU Leuven, Leuven, Belgium
    Search for more papers by this author
  • Aernout Luttun,

    1. Department of Cardiovascular Sciences, Molecular and Vascular Biology Research UnitVesalius Research Center, KU Leuven, Leuven, Belgium
    Search for more papers by this author
  • Frank P. Luyten,

    1. Prometheus, Division of Skeletal Tissue Engineering, KU Leuven, Leuven, Belgium
    2. Laboratory for Skeletal Development and Joint DisordersVesalius Research Center, KU Leuven, Leuven, Belgium
    Search for more papers by this author
  • Geert Carmeliet

    Corresponding author
    1. Laboratory of Clinical and Experimental EndocrinologyDivision of Skeletal Tissue Engineering, KU Leuven, Leuven, Belgium
    2. Prometheus, Division of Skeletal Tissue Engineering, KU Leuven, Leuven, Belgium
    • Clinical and Experimental Endocrinology, KU Leuven, O&N1 Herestraat 49 bus 902, B-3000 Leuven, Belgium
    Search for more papers by this author
    • Telephone: +32-16-330-731; Fax: +32-16-330-718

  • Author contributions: N.v.G.: conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing; S.T., S.J.R., and K.M.: collection and assembly of data; J.S.: conception and design; P.C. and A.L.: provision of study material; F.P.L.: conception and design, data analysis and interpretation, and provision of study material; G.C.: conception and design, data analysis and interpretation, manuscript writing, and final approval of manuscript.

  • Disclosure of potential conflicts of interest is found at the end of this article.

  • §

    First published online in STEM CELLSEXPRESS August 22, 2012.


One of the key challenges in bone tissue engineering is the timely formation of blood vessels that promote the survival of the implanted cells in the construct. Fracture healing largely depends on the presence of an intact periosteum but it is still unknown whether periosteum-derived cells (PDC) are critical for bone repair only by promoting bone formation or also by inducing neovascularization. We first established a protocol to specifically isolate murine PDC (mPDC) from long bones of adult mice. Mesenchymal stem cells were abundantly present in this cell population as more than 50% of the mPDC expressed mesenchymal markers (CD73, CD90, CD105, and stem cell antigen-1) and the cells exhibited trilineage differentiation potential (chondrogenic, osteogenic, and adipogenic). When transplanted on a collagen-calcium phosphate scaffold in vivo, mPDC attracted numerous blood vessels and formed mature bone which comprises a hematopoiesis-supportive stroma. We explored the proangiogenic properties of mPDC using in vitro culture systems and showed that mPDC promote the survival and proliferation of endothelial cells through the production of vascular endothelial growth factor. Coimplantation with endothelial cells demonstrated that mPDC can enhance vasculogenesis by adapting a pericyte-like phenotype, in addition to their ability to stimulate blood vessel ingrowth from the host. In conclusion, these findings demonstrate that periosteal cells contribute to fracture repair, not only through their strong osteogenic potential but also through their proangiogenic features and thus provide an ideal cell source for bone regeneration therapies. STEM CELLS2012;30:2412–2422


To date, autologous bone transplantation remains the therapy of choice to treat nonhealing skeletal defects caused by traumatic injury, osteo-degenerative diseases, or cancer [1]. Nevertheless, the use of autografts is restricted by major drawbacks, including limited availability and donor site morbidity. In response to these limitations, tissue engineering strategies have emerged as a potential alternative.

A promising concept in the repair of bone defects is the use of osteogenic cells seeded on a degradable carrier [2]. Of great interest are mesenchymal stem cells (MSC), which can be isolated from several adult tissues and are capable of giving rise to cells of different lineages [3, 4]. MSC intended for bone tissue engineering applications have typically been isolated from bone marrow, but the periosteum might represent a more relevant cell source as it is intrinsically closer related to bone formation and repair [5, 6]. As known from orthopedic practice, destruction of the periosteum will lead to delayed bone healing or nonunion [7]. Several animal studies have further demonstrated that periosteal cells contribute to bone repair by recapitulating specific features of the bone development process [8–10]. Following the early inflammatory stage, periosteal cells undergo a strong proliferative burst, after which these cells differentiate toward chondrocytes and osteoblasts, initiating endochondral bone formation. The essential function of the periosteum is further underscored by the finding that periosteum-derived cells (PDC) contain a high number of MSC in comparison to bone marrow stromal cells (BMSC) [11, 12]. Furthermore, a number of animal studies have provided proof of concept using PDC for the repair of critically sized bone defects [13–15].

The search for a satisfactory cell source for bone tissue engineering has frequently been focused on the in vitro osteogenic properties of the cell types of interest. In vivo osteogenesis is however inherently coupled to angiogenesis [16] and successful bone regeneration thus requires the timely formation of blood vessels. During uncomplicated bone fracture healing, blood vessels present in the periosteum and adjacent tissues will guarantee the survival of the periosteal cells and allow their rapid proliferation, preventing the formation of fibrous tissue [17, 18]. The importance of adequate vascularization is underscored by the limited survival of cells in tissue engineering constructs upon implantation [19, 20]. Multiple approaches have been proposed to ensure sufficient vascularization of implants, including the use of angiogenic growth factors, microsurgery, or the addition of endothelial (progenitor) cells to the scaffold [21]. This last strategy is particularly interesting, given the close interplay between bone-forming osteoblasts and endothelial cells during bone development and repair [22–24]. Moreover, different studies have indicated that in order to engineer mature, long-lasting vascular structures, coimplantation of endothelial cells with perivascular cells is necessary [25, 26]. The choice of a cell source for bone tissue engineering applications should thus not only take into account their osteogenic properties but also other cues such as proangiogenic characteristics of the implanted cells [16].

In this study, we explored the proangiogenic potential of murine PDC (mPDC). We first developed a technique to isolate mPDC from long bones of adult mice and confirmed the presence of MSC. Subcutaneous implantation of the mPDC on a collagen-calcium phosphate carrier resulted in the formation of vascularized bone accompanied by the establishment of bone marrow. Further in vitro and in vivo analyses revealed that mPDC can contribute to multiple aspects of neovascularization by producing growth factors that enhance angiogenesis and by providing structural support as mural cells to the vascular networks formed by endothelial cells.


Animals and Human Tissues

C57BL/6 mice and NMRI nu/nu nude mice were purchased from the R. Janvier Breeding Center (Le Genest St. Isle, France). Transgenic mice ubiquitously expressing the enhanced green fluorescent protein (eGFP) under the control of the chicken β-actin promoter (C57BL/6-ACTb-eGFP) were obtained from A. Nagy (Samuel Lunenfeld Research Institute, Toronto, Canada). Human periosteal biopsies were obtained from the proximal medial tibia during orthopedic surgery with patient informed consent. A human fibular bone sample was obtained as discarded tissue during the course of orthopedic surgery. All experiments were conducted in accordance with the Declaration of Helsinki and with approval of the Medical and the Animal Ethics Committee of the KU Leuven.

Isolation of mPDC

To obtain mPDC, femurs and tibias isolated from 7–9-week-old male C57BL/6J or C57BL/6-ACTb-eGFP mice were dissected free of muscle and connective tissue under sterile conditions. Subsequently, the epiphyses were protected from digestion by submerging them in 5% low melting point agarose (SeaPlaque, Lonza, Verviers, Belgium, After solidification of the agarose, mPDC were isolated by a collagenase-dispase digest (3 mg/ml collagenase and 4 mg/ml dispase in α-minimal essential medium [α-MEM] with 2 mM glutaMAX-I; Gibco, Invitrogen, Carlsbad, CA, Cells from the first digest (10 minutes) were discarded as they contained cells from remaining muscle and connective tissue. mPDC were released by a subsequent 1 hour digest. The cell suspension was passed through a 70 μm nylon mesh (BD Falcon, BD Biosciences, Erembodegem, Belgium,, washed twice, and plated at 1 × 104 cells per square centimeter in growth medium (α-MEM with 2 mM glutaMAX-I, containing 1% penicillin/streptomycin and 10% fetal bovine serum; all from Gibco, Invitrogen). mPDC were replated at a 1:3 ratio when reaching 80%–90% confluence.

All other primary murine and human cells were isolated and cultured as previously described [11, 27-31]. Detailed protocols can be found in Supporting Information. Human umbilical vein endothelial cells (HUVEC) were purchased from Lonza and maintained in endothelial growth medium-2 (EGM-2; Lonza).

In Vitro Cell Cultures

Growth curve analysis, in vitro differentiation of mPDC toward osteoblasts, chondrocytes and adipocytes, flow cytometry, fluorescence-activated cell sorting, and colony-forming unit fibroblast (CFU-F) assays were performed using standard techniques and are described in detail in Supporting Information.

The support of osteoclast formation by mPDC, mBMSC, or murine trabecular osteoblasts (mTOB) was determined by coculturing these cells with freshly isolated murine bone marrow cells (mBMC) in a 1:10 ratio in the presence of 20 nM 1,25-dihydroxyvitamin D3 as previously described [31]. After 6 days, cultures were stained for tartrate-resistant acid phosphatase (TRAP) to visualize osteoclasts.

When indicated mPDC, seeded at 3 × 104 cells per square centimeter, were cultured under hypoxia (1% oxygen) or normoxia (21% oxygen) for 24 hours, after which growth medium was collected to determine the secretion of vascular endothelial growth factor (VEGF; also known as VEGF-A) and total RNA or nuclear proteins were extracted from the cultured cells (described in detail in Supporting Information). To functionally assess whether mPDC secrete angiogenic factors, conditioned media from mPDC, grown under normoxia or hypoxia, or α-MEM was added to HUVEC plated on collagen type 1 (BD Biosciences). After 24 hours, proliferation was determined by BrdU incorporation (Biotrak cell proliferation ELISA system, GE Healthcare, Diegem, Belgium,

Two in vitro coculture models were used to study the interactions between mPDC and endothelial cells. In the first assay, mPDC were seeded at 5 × 104 cells per square centimeter and after 24 hours human blood outgrowth endothelial cells (hBOEC) were added at 1 × 104 cells per square centimeter. Cells were cultured for 7 days in growth medium with or without the addition of anti-murine VEGF-164 antibody (anti-mVEGF164; 200 ng/ml; R&D systems, Minneapolis, MN, or recombinant mVEGF receptor (mVEGFR)-1/fms-related tyrosine kinase(Flt)-1 Fc Chimera (smFlt-1; 100 ng/ml; R&D systems, Minneapolis, MN, After culture, the endothelial cells were stained with a mouse-anti-human CD31 primary antibody (Supporting Information Table S1) and visualized using the TSA Cyanine 3 System (NEN, PerkinElmer, Boston, MA, For the Matrigel in vitro angiogenesis assay, mPDC and HUVEC/hBOEC were first labeled with, respectively, CellTracker Green 5-chloromethylfluorescein diacetate (CMFDA) and CellTracker CM-DiI (Molecular Probes, Invitrogen) according to the manufacturer's instructions. Subsequently, mPDC (1 × 104 cells per square centimeter) and HUVEC or hBOEC (2 × 104 cells per square centimeter) were resuspended alone or together in EGM-2 and seeded on top of growth factor-reduced Matrigel (BD Biosciences) and cultured for 24 hours.

In Vivo Bone Formation

To investigate the in vivo osteogenic potential, 1 × 106 mPDC or human PDC (hPDC) (passage 3) were seeded onto Collagraft (NeuColl Campbell, CA, scaffolds (3 × 3 × 3 mm3), incubated overnight, and implanted subcutaneously on the back of female NMRI nu/nu mice, as described previously [32]. Scaffolds without cells were used as negative controls. Eight weeks after implantation, scaffolds were retrieved and processed for histology. For cell tracing, mPDC were labeled with CellTracker CM-DiI prior to seeding.

In Vivo Vasculogenesis/Angiogenesis Assay

The in vivo coimplantation of mPDC and HUVEC was adapted from Allen et al. [33]. Cells (mPDC, HUVEC or mPDC and HUVEC in a 60:40 ratio) were resuspended in collagen type 1 (5 mg/ml in EGM-2) at a density of 1 × 106 cells per milliliter and 200 μl was injected subcutaneously on the back of female NMRI nu/nu mice. After 7 days, the gels were retrieved and processed for histology.

(Immuno)Histochemistry and Histomorphometry

Isolated bones and scaffolds were fixed in 2% paraformaldehyde overnight and decalcified in EDTA for 14 days at 4°C prior to dehydration, embedding in paraffin, and sectioning at 4 μm (bone samples) or 10 μm (Collagraft). Collagen gels were fixed in 2% paraformaldehyde overnight, embedded in NEG-50 frozen section medium (Richard-Allen Scientific, Kalamazoo, MI), and sectioned with a cryostat at 7 μm.

All histochemical stainings (hematoxylin and eosin [H&E], Safranin O, Sirius Red, and TRAP) and immunohistochemical stainings (CD31, CD34, CD45, osteocalcin, Ulex europaeus agglutinin [UEA]-lectin, and α-smooth muscle actin [α-SMA]) have been reported previously [23, 28, 30, 34, 35] and were performed as described in Supporting Information. Images were taken on a Zeiss Axioplan 2 light microscope or a Zeiss LSM510-META NLO multiphoton confocal microscope. Collagen fibers were imaged by polarized light microscopy following Sirius Red staining as previously described [35]. Histomorphometry was performed using the ImageJ software (National Institutes of Health). Quantification of newly formed bone in Collagraft scaffolds was performed by manual outlining areas of bone and total tissue. Quantification of blood vessels was performed by counting or outlining CD31/CD34/UEA lectin-positive vessels with a clear lumen.

Statistical Analysis

Data are presented as means ± SE. Data were analyzed by two-sided two-sample Student's t test or one-way ANOVA using the NCSS statistical software. Differences were considered statistically significant at p < .05.


Generation of Primary Cell Cultures from Adult Mouse Periosteum

The isolation of human PDC is typically achieved by mechanically stripping off the periosteum from the diaphyseal bone, followed by an enzymatic digestion [36]. The periosteum from adult mice is however only a few cell layers thick and thus difficult to isolate by dissection (Fig. 1A, 1B). A possible solution is to perform an enzymatic digest directly on the long bones, but without releasing cells from the growth plate, articular cartilage, tendon, and connective tissue, thus avoiding the attainment of a heterogeneous cell population. To prevent the release of nonperiosteal cells, we embedded the epiphyses of the bones in low melting point agarose (Fig. 1C), prior to the enzymatic digest. Histological analysis showed that the isolation of the mPDC was not only efficient but also selective since joint and connective tissues surrounding the epiphysis were not removed by the enzymatic digest (Supporting Information Fig. S1). The isolated cells exhibited a fibroblast-like morphology (Fig. 1D) and possessed a high proliferative capacity.compared to mBMSC and mTOB, the growth curve of mPDC was characterized by a shorter lag time and the population doubling time was shorter compared to mTOB (Fig. 1E).

Figure 1.

Isolation and characterization of mPDC. (A, B): H&E staining of bone sections comparing human (A) and murine (B) periosteum (b: bone; c: cambium layer; f: fibrous layer; m: muscle; p: periosteum). (C): Murine femur and tibia with epiphyses embedded in 5% low melting point agarose. (D): In vitro culture of mPDC showing a fibroblast-like morphology. (E): Growth curves (left) and doubling times (right) of mPDC, mBMSC, and mTOB (n = 3). (F, G): qRT-PCR analysis of osteogenic (F) and chondrogenic (G) gene expression by mPDC, compared to, respectively, mTOB and mGCH (n = 3–6). *, p < .05; ***, p < .001. Scale bars = 50 μm in (B); 100 μm in (A) and (D). Abbreviations: Acan, aggrecan; Col, collagen; H&E, hematoxylin and eosin; mBMSC, murine bone marrow stromal cells; mGCH, murine growth plate chondrocytes; mPDC, murine periosteum-derived cells; mTOB, murine trabecular osteoblasts; Ocn, osteocalcin; Osx, osterix; Runx2, runt-related transcription factor 2; Sox9, SRY (sex-determining region Y)-box containing gene 9.

The periosteum is known to contain different cell populations including osteoprogenitor and chondroprogenitor cells and we therefore analyzed the expression of early osteogenic and chondrogenic genes in mPDC relative to primary mTOB and murine growth plate chondrocytes, which are more mature cells (Fig. 1F, 1G). mPDC expressed markers of osteogenic progenitor cells, including runt-related transcription factor 2 and osterix (Osx; also known as Sp7), while the expression of the mature osteoblast markers collagen type 1 alpha 1 and osteocalcin (Ocn; also known as Bglap) was low. Similarly, mPDC expressed sex-determining region Y-box containing gene 9, a marker of chondroprogenitor cells, whereas the levels of mature chondrocyte markers collagen type 2 alpha 1 (Col2a1) and aggrecan (Acan) and of the hypertrophic chondrocyte marker collagen type 10 alpha 1 were negligible.

mPDC Can Differentiate Toward the Osteoblastic, Chondrogenic, and Adipogenic Lineage

Next, the differentiation potential of mPDC toward different mesenchymal cell lineages was investigated. Osteoblastic differentiation was induced by culturing confluent mPDC in medium containing ascorbic acid and β-glycerophosphate and was analyzed at days 14 and 21. The positive alkaline phosphatase (ALP) staining at day 14 was indicative of osteogenic differentiation and quantification of ALP activity showed a threefold increase compared to day 0 (Fig. 2A). After 21 days of culture in osteogenic medium, the formation of mineralized nodules was evidently shown by Alizarin Red S staining (Fig. 2B). The marked upregulation of Osx, osteopontin (Opn; also known as Spp1), and Ocn expression confirmed the differentiation of mPDC toward mature osteoblasts (Fig. 2B).

Figure 2.

mPDC can differentiate toward the osteoblastic, chondrogenic, and adipogenic lineage. (A, B): Osteogenic differentiation of mPDC. (A): Left: mPDC cultured for 14 days in osteogenic medium and stained for ALP. Right: Quantification of ALP activity confirming upregulation of ALP levels at day 14 compared to day 0. (B): Left: Alizarin Red S staining of mPDC cultured for 21 days in osteogenic medium showing the presence of mineralized nodules. Right: qRT-PCR analysis of Osx, Opn, and Ocn gene expression. (C): Chondrogenic differentiation of mPDC. Left: Alcian Blue staining of proteoglycans in mPDC micromass cultured for 7 days in chondrogenic medium. Right: qRT-PCR analysis of Col2a1 and Acan gene expression. (D): Adipogenic differentiation of mPDC. Left: Oil Red O staining of mPDC after culture for 14 days in adipogenic medium revealed the presence of oil droplets in the cytoplasm. Right: qRT-PCR analysis of Pparg and Fabp4 gene expression. *, p < .05; **, p < .01; ***, p < .001; n = 4. Scale bars = 200 μm in (A), (B), (D); 1 mm in (C). Abbreviations: Acan, aggrecan; ALP, alkaline phosphatase; Col2a1, collagen type 2 alpha 1; d, day; Fabp4, fatty acid binding protein 4; mPDC, murine periosteum-derived cells; Ocn, osteocalcin; Opn, osteopontin; Osx, osterix; Pparg, peroxisome proliferator-activated receptor-gamma.

To induce chondrogenic differentiation, mPDC were seeded as micromasses and treated for 7 days with transforming growth factor-β1. At the end of culture, the micromasses clearly stained positive with Alcian Blue, indicating that proteoglycans were produced and deposited in the matrix (Fig. 2C). Analysis of gene expression confirmed the differentiation of mPDC to mature chondrocytes, characterized by the increase in Col2a1 and Acan (Fig. 2C).

Finally, we induced adipogenic differentiation by culturing the cells in medium containing insulin, dexamethasone, 3-isobutyl-1-methylxanthine, and indomethacin. After 14 days of culture, the presence of adipocytes was marked by the detection of oil droplets in the cytoplasm (Fig. 2D). The upregulation of peroxisome proliferator-activated receptor-gamma and fatty acid binding protein 4 gene expression confirmed the differentiation of mPDC to adipocytes (Fig. 2D). These data indicate that mPDC have trilineage differentiation potential, which is one of the hallmarks of MSC.

MSC Are Abundantly Present in the mPDC Population

To further confirm the presence of MSC in the mPDC population, we next analyzed the surface marker profile of these cells. Flow cytometric analysis revealed that more than 50% of the mPDC at passage 1 expressed one of the mesenchymal markers CD105, CD90.2, CD73, CD51, and stem cell antigen-1 (Sca-1) (Fig. 3A). Conversely, only very few cells expressed hematopoietic (CD11b and CD45) or endothelial (CD31, CD34, and VEGFR-2) markers. Analysis of mPDC at different passages (freshly isolated, passages 1, 3, and 6) showed that the cells expressing hematopoietic or endothelial markers disappeared after culture (<1% by passage 3), while the number of cells expressing mesenchymal markers further increased (Supporting Information Table S3). We next compared the abundance of putative MSC in the periosteum and the bone marrow stroma using the stem cell marker Sca-1, which has been described to identify MSC in murine bone marrow [37, 38]. To quantify the number of MSC relative to the stromal fraction, mPDC and mBMC cell populations were first gated for the CD45-negative fraction hereby excluding the hematopoietic cells that are abundantly present especially in the bone marrow. Flow cytometric analysis revealed that the percentage of putative MSC was considerably higher in periosteum than in the bone marrow stroma, even when mBMC were isolated by collagenase treatment of bone fragments, which is known to enrich the number of isolated MSC [38]. By performing a CFU-F assay, we showed that in periosteum, as has been described for bone marrow, the Sca-1+CD45 fraction contains the vast majority of cells with colony-forming potential (Fig. 3C, 3D). Taken together, these results confirm that MSC are abundantly present within the mPDC population.

Figure 3.

MSC markers are expressed by a large subpopulation of the mPDC. (A): Flow cytometric analysis of the surface marker profile of mPDC at passage 1 revealed abundant presence of cells expressing MSC markers (CD105, CD90.2, CD73, CD51, and Sca-1), while few cells expressed hematopoietic (CD11b and CD45) or endothelial (CD31, CD34, and VEGFR-2) markers. For each marker, the percentage of positive cells is shown as the average ± SEM (n = 4). Full histograms represent marker expression, open histograms represent isotype controls. (B): Quantification of the number of Sca-1-positive cells in the CD45-negative fraction of freshly isolated mPDC, mBMC, and mBMCcol showed a significantly higher number of MSC in the periosteum compared to the bone marrow stroma. (C, D): Methylene Blue staining (C) and quantification (D) of CFU-F assays showing that the Sca-1+CD45 fraction of the periosteum comprises the vast majority of cells with colony-forming potential. *, p < .05; ***, p < .001. Abbreviations: CFU-F, colony-forming unit fibroblast; mBMC, murine bone marrow cells; mBMCcol, murine bone marrow cells isolated by collagenase treatment; mPDC, murine periosteum-derived cells; MSC, mesenchymal stem cell; Sca-1, stem cell antigen-1; VEGFR-2, vascular endothelial growth factor receptor-2.

In Vivo Transplanted mPDC Form Vascularized Bone with a Hematopoietic Compartment

To assess the in vivo osteogenic potential of mPDC, cells were seeded on Collagraft scaffolds and implanted subcutaneously in nude mice. hPDC were used as a positive control. Implants were retrieved 8 weeks postimplantation, and histomorphometrical analysis of H&E stained sections revealed that mPDC formed a substantial amount of bone tissue throughout the scaffolds, comparable to hPDC (9.2% ± 0.9% vs. 7.1% ± 0.5%; Fig. 4A–4C). The newly formed tissue displayed the typical features of bone. Indeed, polarized light microscopy of Sirius Red stained sections showed the presence of highly organized thick collagen fibers in the newly formed bone matrix (Fig. 4D–4F). In addition, osteocalcin staining revealed that differentiated osteoblasts were lining the bone surface (Fig. 4G), further confirming that mature bone was formed. By fluorescent labeling of the mPDC prior to implantation, we could show that several of the osteocytes embedded in the bone matrix were derived from the implanted cells (Fig. 4H). Bone formation was not detected when scaffolds without cells were implanted (data not shown). To assess whether a cartilage intermediate preceded the bone formation in these implants, sections were stained with Safranin O to detect proteoglycans. As seen in Figure 4I, implants were at 8 weeks predominantly devoid of chondrocytes, and only occasionally very small cartilaginous condensations, consisting of a few chondrocytes embedded in the calcium phosphate granules, could be detected (inset in Fig. 4I).

Figure 4.

mPDC form mature bone when transplanted in vivo. (A–C): Comparison of bone formed by mPDC and hPDC in vivo on a Collagraft scaffold. H&E-stained sections showing bone formation by mPDC (A) and hPDC (B), characterized by osteocytes embedded in the newly formed bone (b) and the establishment of bone marrow compartments (white arrow) besides the remains of the Collagraft (§). Quantification of newly formed bone (C) showed equal amounts of bone formed by mPDC and hPDC (n = 4). (D): Polarized light microscopy of Sirius Red stained sections, revealing thick, highly organized collagen fibers (birefringent fibers, orange) in the bone around the scaffold granules and thin fibers (green) in the fibrous compartments. (E): Detail of the boxed region in (D) (white arrow indicates birefringent collagen fibers). (F): Bright-field image of the same region shown in (E). (G): Immunohistochemistry showing osteocalcin-positive osteoblasts (white arrowheads) lining the bone surfaces. (H): Confocal microscopy of implanted CM-DiI-labeled mPDC counterstained with Hoechst, showing a CM-DiI-positive osteocyte (white arrow) embedded in the bone matrix. (I): Safranin O staining revealed the absence of large amounts of cartilage tissue in the scaffolds. Inset: Safranin O staining showing a small group of chondrocytes located within a calcium phosphate granule. (J, K): Bone marrow compartments were characterized by the presence of CD45-positive cells (J) and large CD31-positive blood vessels (K). Insets: adjacent sections stained with H&E. (L): TRAP staining showing osteoclasts resorbing newly formed bone (black arrows). (M): TRAP staining of in vitro cocultures of bone marrow cells with mTOB, mPDC, or mBMSC, stimulated with 1,25-dihydroxyvitamin D3 for 6 days. Scale bars = 20 μm in (H), inset in (I); 50 μm in (G), (J), (K), and (L); 100 μm in (A), (B), (E), and (F); 200 μm in (D) and (I). Abbreviations: BV, bone volume; H&E, hematoxylin and eosin; hPDC, human periosteum-derived cells; mBMSC, murine bone marrow stromal cells; mPDC, murine periosteum-derived cells; mTOB, murine trabecular osteoblasts; N.S., not significant; TRAP, tartrate-resistant acid phosphatase; TV, tissue volume.

The presence of mPDC contributed not only to the formation of bone but also to the development of a hematopoietic compartment. Typical for bone marrow spaces, these compartments were packed with CD45-positive hematopoietic cells in close association with large CD31-positive sinusoid-like blood vessels (Fig. 4J, 4K). In addition, TRAP-positive osteoclasts were detected on the bone surfaces bordering the hematopoietic compartments, indicating active bone remodeling (Fig. 4L). Osteoclasts are indispensable for the remodeling of the newly formed immature bone into mature, lamellar bone [39], and are thus part of the global bone formation process. They are generally formed by close interaction with osteogenic cells and we therefore investigated whether mPDC can support osteoclast formation using an in vitro coculture system. As shown in Figure 4M, mPDC induced the formation of a large number of multinucleated TRAP-positive cells, comparable to cocultures with mTOB, and manifestly more than in conditions using mBMSC, which yielded only a limited number of osteoclasts.

Another prerequisite for adequate bone formation is the presence of functional blood vessels. The number of blood vessels as well as the area covered by the vessels was significantly higher in the mPDC-seeded implants compared to scaffolds without cells (Fig. 5A–5C). Detailed analysis revealed numerous blood vessels in close contact with the newly formed bone (Fig. 5D; arrow) or even enclosed by bone tissue (Fig. 5D; arrowhead). Costaining for the pericyte marker α-SMA and the endothelial cell marker CD31 (Fig. 5E, 5F) showed that many of the blood vessels near the bone surface were covered with mural cells (white arrows), which stabilize nascent vessels and promote vessel maturation. Blood vessels close to the hematopoietic sites on the other hand (white arrowhead) did not stain positive for α-SMA. Labeling of the mPDC with CM-DiI prior to implantation furthermore indicated that some of these perivascular cells were of donor origin (Fig. 5G). These results highlight an important role of the mPDC in promoting blood vessel ingrowth into the implants and possibly also in contributing to vessel stabilization and maturation.

Figure 5.

mPDC stimulate neoangiogenesis after in vivo implantation. (A, B): Comparison of the number of blood vessels in empty scaffolds (A) or scaffolds seeded with mPDC (B) visualized by CD34 staining. (C): Quantification of the number of blood vessels per mm2 (left) and the area occupied by the blood vessels (right) in scaffolds seeded with mPDC versus empty scaffolds (*, p < .05; n = 3–4). (D): Detailed view, revealing blood vessels in close contact with the newly formed bone (arrow) or even enclosed by bone (arrowhead). (E): CD31 and α-SMA costaining showing blood vessels surrounded by α-SMA-positive mural cells (white arrows) near the newly formed bone in the scaffolds. Blood vessels located in the bone marrow compartments (white arrowhead) were negative for α-SMA. (F): Bright-field image of same region shown in (E) (§: scaffold; b: bone; bm: bone marrow). (G): Labeling of mPDC with CM-DiI prior to implantation revealed that mPDC can support blood vessels as perivascular cells (α-SMA-positive) in vivo (white arrow). Blood vessel lumens are marked with an asterisk. Immunofluorescence sections were counterstained with Hoechst to visualize cell nuclei. Scale bars = 20 μm in (G); 50 μm in (D), (E), and (F); 100 μm in (A) and (B). Abbreviations: mPDC, murine periosteum-derived cells; α-SMA, α-smooth muscle actin.

Hypoxia Activates Metabolic and Proangiogenic Responses in mPDC

To further elucidate the proangiogenic properties of mPDC, we switched to in vitro cell cultures. Hypoxia is known to accompany the early stages of fracture healing [40] and is a major stimulus for angiogenesis. We therefore assessed the response of mPDC to low oxygen tensions (1% oxygen) by analyzing factors involved in the hypoxia-inducible pathway. Western blot analysis showed that hypoxia-inducible factor (HIF)-1α protein levels were noticeably elevated in nuclear extracts of mPDC cultured under hypoxia, whereas no differences in HIF-2α levels were detected (Fig. 6A). Based on these results, we analyzed the expression of known HIF-1α target genes Glut1 (glucose transporter 1; also known as Slc2a1) and Ldha (lactate dehydrogenase A), which are involved in cellular metabolism, and Vegf, a prime angiogenic factor. Hypoxia increased Glut1 and Ldha gene expression (Fig. 6B), suggesting a shift toward a more anaerobic energy metabolism. In addition, mRNA and protein levels of VEGF were almost threefold upregulated in hypoxia (Fig. 6B, 6C).

Figure 6.

Hypoxia activates metabolic and angiogenic responses in mPDC. (A–C): Analysis of the response of mPDC cultured for 24 hours under normoxic (21% oxygen) or hypoxic (1% oxygen) conditions. Western Blot analysis of nuclear extracts of mPDC (A) showed an increase of HIF-1α, but not HIF-2α, levels in hypoxia. qRT-PCR analysis (B) of Glut1, Ldha, and Vegf mRNA levels (n = 3). ELISA on culture medium (C) indicated an increased secretion of VEGF by mPDC in hypoxia (n = 3). (D, E): Culture of HUVEC for 24 hours in CM of mPDC grown under normoxic or hypoxic conditions. Microscopic analysis of HUVEC (D) revealing a larger number of cells when CM of mPDC grown under hypoxia was added. Quantification of BrdU incorporation (E) showed increased proliferation under this condition (n = 4). (F): mPDC were cocultured with hBOEC in α-MEM with or without the addition of anti-mVEGF164 (200 ng/ml) or smFlt-1 (100 ng/ml). hBOEC were stained with anti-hCD31 and cell nuclei were counterstained with Hoechst. (G): Quantification of hCD31 staining after 7 days (n = 4). *, p < .05; **, p < .01; ***, p < .001. Scale bars = 50 μm in (D); 200 μm in (F). Abbreviations: BrdU, bromodeoxyuridine; CM, conditioned medium; Glut1, glucose transporter 1; hBOEC, human blood outgrowth endothelial cells; HIF, hypoxia-inducible factor; HUVEC, human umbilical vein endothelial cells; Ldha, lactate dehydrogenase A; mPDC, murine periosteum-derived cells; O.D., optical density; VEGF, vascular endothelial growth factor.

To assess whether these changes in VEGF expression were physiologically relevant, we cultured HUVEC, a well-studied endothelial cell model, in medium conditioned by mPDC. When medium from mPDC cultured for 24 hours in hypoxia was added, HUVEC displayed a significantly higher proliferative response in comparison to the addition of normoxia-conditioned medium (Fig. 6D, 6E). Furthermore, even normoxia-conditioned medium increased the proliferation of the endothelial cells in comparison to standard medium. These data indicate that mPDC produce angiogenic factors, like VEGF, and that this production is upregulated when oxygen levels decrease.

mPDC Promote the Survival of Endothelial Cells Through the Production of VEGF

We next investigated whether the proangiogenic activity of mPDC could be attributed to VEGF, as the survival of endothelial cells is known to depend on the presence of VEGF in the medium [41, 42]. We therefore assessed whether coculture of mPDC could substitute for the addition of exogenous angiogenic factors in cultures of hBOEC. The withdrawal of angiogenic factors induced cell death of hBOEC within 4 days (data not shown). The survival of hBOEC could however be rescued in a coculture with mPDC, as shown by hCD31 staining after 7 days of culture (Fig. 6F, 6G). The importance of mPDC-produced VEGF in the survival of endothelial cells was evidenced by the finding that inhibition of VEGF activity, by the addition of an anti-mVEGF164 antibody or a soluble receptor (smFlt-1), resulted in massive endothelial cell death. Taken together, these experiments illustrate that mPDC promote the survival of endothelial cells through the production of VEGF.

mPDC Possess Perivascular Cell Characteristics and Promote Both Vasculogenesis and Angiogenesis In Vivo

As vessel maturation is critical to the establishment of a long-lasting functional vasculature, we next investigated the pericytic characteristics of mPDC in more detail. Endothelial cells (HUVEC or hBOEC) and mPDC were seeded alone or together in a 2:1 ratio on Matrigel and their distribution, with respect to each other, was analyzed after 24 hours. Microscopic analysis revealed that the cocultures induced more robust networks than when either cell type was cultured alone (Supporting Information Fig. S2). Confocal microscopy showed that mPDC were localized perivascularly with cell-to-cell contact between the endothelial cells and the mPDC (Fig. 7A). Consistent with these observations, flow cytometric analysis revealed that numerous mPDC express the pericyte markers α-SMA (76.4% ± 5.7%) and platelet-derived growth factor receptor-β (PDGFR-β; 43.6% ± 0.2%) (Fig. 7B).

Figure 7.

mPDC possess perivascular cell characteristics and promote both vasculogenesis and angiogenesis in vivo. (A): Confocal microscopy of pseudovascular cords formed in cocultures of CMFDA-labeled mPDC and CM-DiI-labeled HUVEC (stack of 92 images; left: orthogonal Z-sections; right: selected optical sections), showing a perivascular localization of the mPDC, with cell-to-cell contact between mPDC and HUVEC (white arrows). (B): Surface marker analysis of mPDC by flow cytometry revealed expression of the mural cell markers α-SMA and PDGFR-β by a high percentage of cells (n = 3). (C): In vivo implantation of mPDC (left), mPDC together with HUVEC in a 60:40 ratio (middle), or HUVEC alone (right) in collagen gels for 7 days (g: gel, h: host). Immunofluorescence staining for mCD31 and UEA-lectin demonstrated abundant blood vessels derived from the donor and the host in the gels containing both mPDC and HUVEC. (D): Quantification of the number of blood vessels per square millimeter (mCD31: host-derived vessels; UEA-lectin: donor-derived blood vessels; n = 4). (E): Implantation of mPDC derived from C57BL/6-ACTb-eGFP mice (mPDC-GFP) together with HUVEC showed a perivascular localization of mPDC around blood vessels formed by HUVEC (stained with UEA-lectin). Immunofluorescence sections were counterstained with Hoechst to visualize cell nuclei. *, p < .05; **, p < .01; ***, p < .001 Scale bars = 20 μm in (E), 200 μm in (C). Abbreviations: CMFDA, 5-chloromethylfluorescein diacetate; HUVEC, human umbilical vein endothelial cells; mPDC, murine periosteum-derived cells; PDGFR-β, platelet-derived growth factor receptor-β UEA-lectin, Ulex europaeus agglutinin-lectin; α-SMA, α-smooth muscle actin.

When mPDC were implanted together with endothelial cells in a collagen gel in vivo, the total number of formed blood vessels was significantly higher than when either cell type was injected alone (Fig. 7C, 7D). The addition of mPDC to HUVEC increased the number of blood vessels lined by human endothelial cells almost twofold (Fig. 7D), even though less endothelial cells were initially present in the coculture implants. The formed blood vessels were filled with erythrocytes, confirming their connection to the host vasculature (Supporting Information Fig. S3). Implants seeded with mPDC also contained numerous blood vessels derived from the host, confirming that mPDC have a strong proangiogenic potential. The higher number of host-derived blood vessels in gels containing only mPDC compared to cocultures can likely be explained by the higher number of mPDC initially seeded. By implanting mPDC derived from C57BL/6-ACTb-eGFP mice together with HUVEC, we could further demonstrate that numerous mPDC closely encircled the vessels formed by HUVEC (Fig. 7E), supporting our in vitro observations. In conclusion, our results show that mPDC can contribute to new blood vessel formation by the production of angiogenic growth factors and by obtaining a pericyte-like phenotype, contributing structurally to the new vessels.


The progress in bone repair strategies critically depends on adequate in vitro and in vivo models to evaluate new therapies and obtain mechanistic insight in the bone healing process. Due to their unique role during bone fracture repair, PDC are a valuable candidate for tissue engineering strategies. However, periosteal cells and their in vivo mechanisms of action remain poorly characterized, which might be related to the lack of a robust technique to isolate PDC from mice in a manner comparable to human PDC. In this study, we developed a technique to isolate mPDC from long bones of adult mice and showed that these cells possess a strong osteogenic potential and marked proangiogenic features.

We were able to selectively isolate mPDC from the long bones of adult mice through the use of agarose to shield cells of the perichondrium and the articular surface from the enzymatic digest applied to the nonprotected diaphyseal region. The mPDC population contains an abundant number of MSC, as shown by means of trilineage differentiation studies and surface marker analysis. Our results further suggest that considerably more MSC are present in the periosteum in comparison to the bone marrow stroma, confirming earlier studies on mouse, rat, and human cells [11, 12, 43-45]. In addition, we show that the mPDC expressed markers of osteoprogenitor and chondroprogenitor cells, but not of more differentiated osteoblasts or chondrocytes, suggesting the presence of both very early cells and more committed skeletal progenitor cells in the mPDC population. The large number of MSC in PDC cultures emphasizes their suitability as a cell source for bone tissue engineering applications and favors their use over BMSC. Although rich in MSC, the mPDC population is heterogeneous in nature and the question remains whether all elucidated properties (bone formation, establishment of hematopoiesis, attraction of osteoclasts, and support of blood vessels) can be attributed to one single cell type or whether these different functions are carried out by specific subpopulations.

In this study, we used relatively young mice (7–9 weeks old). It is however known that with age, the thickness and cellularity of the periosteum decreases [6, 46], resulting in diminished periosteal bone formation and delayed fracture repair [47–49]. In addition, the angiogenic response during fracture healing may also decline with age [48, 50]. Conversely, the osteogenic and chondrogenic differentiation potential of human periosteal cells is maintained regardless of donor age [11, 36]. Future studies are thus required to investigate whether the proangiogenic properties of PDC decline with age and provide an explanation for the reduced fracture repair in the elderly.

When transplanted in vivo, mPDC formed not only mature bone but also a hematopoiesis-supportive stroma. As recently highlighted, the ability of transplanted cells to form marrow stroma at an ectopic site can be considered desirable and even essential for a bone tissue engineering cell source as it marks the presence of a very early, uncommitted skeletal stem cell (SSC) [51]. The presence of the SSC has generally been ascribed to BMSC [52–54], but this study and our previous work on hPDC [55] evidently proves that the SSC is also present in the periosteum. The presence of SSC at both locations can likely be explained by the common origin of the bone marrow stroma and periosteum, which both derive from the perichondrium during fetal bone development.

The marrow compartments were further characterized by the presence of TRAP-positive osteoclasts, resorbing the adjacent bone matrix. In vitro cocultures revealed that mPDC were very potent in supporting osteoclastogenesis and even more effective than mBMSC. These data suggest that during fracture repair periosteal cells can attract osteoclast precursors and promote their differentiation and activity, independent of MSC from the bone marrow, and hereby regulate the remodeling of the fracture callus into mature lamellar bone. These findings further underscore and clarify previous studies showing that removal of the bone marrow does not significantly influence fracture callus formation or remodeling [10, 56]. Osteoclasts are also crucial for the turnover of immature into mature bone in a tissue engineering construct [39], and the osteoclastogenic activities of PDC additionally favor their use in bone tissue engineering applications.

An important finding of this study is that mPDC have proangiogenic properties, and to our knowledge, this is the first report to describe this feature of periosteal cells. We demonstrated that mPDC significantly contributed to the attraction of new blood vessels into engineered constructs after in vivo implantation. Indeed, both in calcium phosphate-based scaffolds and in collagen gels together with endothelial cells, the presence of mPDC augmented the angiogenic response from the host. Bone fracture results in the rupture of the vasculature and during the first phases of fracture repair, cells in the periosteum experience hypoxia [40]. HIF-1α levels increased in mPDC exposed to hypoxia and resulted in increased expression of genes involved in anaerobic cell metabolism and angiogenesis. The VEGF produced by the (hypoxic) mPDC enhanced endothelial cell proliferation and was necessary for endothelial cell survival, indicating that VEGF plays a pivotal role in the communication between mPDC and endothelial cells.

In addition to angiogenic growth factor production, PDC may also structurally contribute to neovascularization. Indeed, several blood vessels in the engineered construct were mature, as they were covered with mural cells, and multiple of these perivascular cells were of donor origin. This finding was confirmed by in vitro coculture of mPDC and endothelial cells which resulted in the formation of robust pseudovascular networks, in which the mPDC obtained a pericyte-like phenotype. When coimplanted in vivo with HUVEC, mPDC substantially increased vasculogenesis by the HUVEC, suggesting that the combination of PDC and endothelial cells might be an ideal strategy to promote blood vessel formation in tissue engineering constructs. The capability of the mPDC to act as blood vessel-supporting cells was further emphasized by the perivascular localization of mPDC in the implants and by the expression of the pericyte markers α-SMA and PDGFR-β by a large subpopulation of the mPDC. The presence of these markers has also been described for MSC derived from other sources, such as the bone marrow [54, 57] and the expression of pericyte markers might be related to the perivascular origin of MSC in different tissues [58]. Whether the MSC in the periosteum reside in a perivascular location remains elusive, but evidence exists that periosteal pericytes possess MSC characteristics [59, 60]. As the periosteum is highly vascularized [61], this model could also explain the high number of MSC found in this tissue.


In conclusion, we showed that periosteal cells can be isolated from the long bones of adult mice. mPDC are rich in MSC and fulfill all requirements of an ideal cell source for bone regeneration, including the potential to form mature bone and to stimulate blood vessel ingrowth. The proangiogenic potential of mPDC can be attributed to angiogenic growth factor production and to the structural support of blood vessels as perivascular cells. Finally, the straightforward isolation method of mPDC, their beneficial properties, and their in vivo bone formation potential, which is similar to hPDC, make mPDC a valuable preclinical cell model for bone tissue engineering research. Virtually, every available transgenic mouse strain, including reporter and knockout animals, can be used to isolate mPDC, which could help to further clarify the nature of the cells in the periosteum and understand by which mechanisms these cells orchestrate fracture repair, knowledge that is essential to bring periosteal cell-based tissue engineering constructs into the clinic.


We thank Ingrid Stockmans, Riet Van Looveren, Carla Geeroms, and Petra Vandervoort for excellent technical assistance. We are also grateful to Kjell Laperre for the isolation of growth plate chondrocytes and to Steve Stegen for help with the isolation of bone marrow cells. This work was supported by grants from the Fund for Scientific Research Flanders (FWO; G.0500.08 and G.0982.11; G.C.), the Flemish Government (Long-term Structural Funding—Methusalem; P.C.), the European Commission (FP7-StG-IMAGINED 203291; A.L.), and the KU Leuven (IOF Knowledge Platform “Prometheus”; IOFKP/07/004; S.J.R., J.S., F.P.L. and Stem Cell Institute of Leuven; S.J.R.). N.v.G. is a fellow from the Agency for Innovation by Science and Technology in Flanders (IWT). This work is part of Prometheus, the Leuven Research & Development Division of Skeletal Tissue Engineering of the KU Leuven.


The authors indicate no potential conflicts of interest.