Circulating osteogenic cells: Implications for injury, repair, and regeneration

Abstract

The aim of this review is to provide a critical reading of recent literature pertaining to the presence of circulating, fluid-phase osteoblastic cells and their possible contribution to bone formation. We have termed this group of cells collectively as circulating osteogenic precursor (COP) cells. We present evidence for their existence, methods used for their isolation and identification, possible physiological and pathophysiological roles, cellular origins, and possible mechanisms for their migration to target tissues. © 2011 American Society for Bone and Mineral Research

Introduction

Bone remodeling presupposes a continuous recruitment of osteoclastic and osteoblastic cells to newly established bone-remodeling sites where bone resorption and bone formation are taking place and coupled in space and time. Although osteoclastic cells originate from hematopoietic precursors that reach bone surfaces through the circulation, the nature of the osteoblastic cells recruited to bone formation surfaces is not completely understood. The classic description of osteoblastic cells is that they are fibroblast-like, derived from stem cells present primarily in bone marrow stroma termed marrow stromal stem cells (MSCs) (also known as skeletal or mesenchymal stem cells). Recent evidence suggests that MSCs are present in the perivascular niche in abluminal surfaces of blood vessels.1 MSCs are extensively studied in ex vivo systems, and they are isolated from low-density mononuclear cells through plastic adherence and described as CD105+, CD73+, CD90+, CD14−, CD34−, CD45−, CD79−, and CD19− cells.2 In addition, they have been characterized as CD44+, CD63+, CD146+, and Stro-1+ cells.3 In systemic infusion studies, MSCs exhibit limited capacity for crossing the endothelial barrier and poor homing to noninjured skeletal tissues,4 and thus contribute to the common notion that they are not circulating (i.e., solid-phase cells, in contrast to hematopoietic stem cells that are fluid phase).5

However, several investigators have revisited an old idea that the osteoblastic cells recruited to bone formation surfaces are fluid-phase cells that access bone formation sites through blood vessels. This concept has been supported by recent histomorphometric studies demonstrating that bone resorption and bone formation occur in specialized “bone-remodeling compartments” that are lined by a layer of flat cells connected to a capillary that provides a conduit for the osteoclastic and osteoblastic cells to reach bone surfaces.6

The aim of this review is to provide a critical reading of recent literature pertaining to the presence of circulating, fluid-phase osteoblastic cells and their possible contribution to bone formation. We have termed this group of cells collectively circulating osteogentic precursor (COP) cells.

Evidence for the Presence of COP Cells

Several types of experimental studies have provided evidence for the presence of circulating osteoprogenitor cells. These studies are grouped into the following categories.

Evidence from parabiosis experiments

Parabiosis experiments are based on creating a conjoint pair of mice that share a common circulatory system. Using a parabiosis model, Kumagai et al. created a conjoint pair of a mouse constitutively overexpressing green fluorescence protein (GFP) and a wild-type (WT) syngenic mouse.7 After fibular fracture in the conjoined WT partner, GFP+ alkaline phosphatase (ALP)+ cells were found localized to the fracture callus, suggesting that fracture (i.e., tissue injury and inflammation) induced a stimulus for recruitment of circulating ALP+ cells. However, the nature of these cells was not further characterized. In an another experiment, Boban et al. examined for the existence of circulating osteoprogentior cells in transgenic collagen I 2.3-GFP (Col I 2.3-GFP) or osteocalcin-GFP mice.8 These mice were joined to transgenic parabionts overexpressing thymidine kinase (TK) under the control of the Col I 2.3 promoter (Col I 2.3ΔTK). When Col I 2.3ΔTK transgenic mice are given ganciclovir, osteoblasts are destroyed, but on removal of the drug the bone recovers uneventfully. In addition, no effect of ganciclovir on bone marrow is observed. After ganciclovir-induced osteoblast ablation, the authors found no evidence for the presence of GFP-marked cells in the Col I 2.3ΔTK mice, suggesting that late-stage osteoblastic cells and osteocytes expressing collagen type I (Col I) or OC do not circulate.

Evidence from bone marrow transplantation (BMT) experiments

Olmsted-Davis et al. provided evidence for the presence of a common circulating hematopoietic and osteoblast progenitor cell. They found that this cell population of common progenitors may belong to the “side population” of marrow stem cells, defined by their ability to expel a DNA binding dye and regenerate hematopoietic lineages.9 Similar results were obtained by Dominici et al.: GFP-marked, plastic-nonadherent bone marrow cells generated osteoblasts, osteocytes, and hematopoietic cells after transplantation into lethally irratiated mice.10 The evidence of a common osteoblast and hematpoietic precursor was suggested by the demonstration of a common retroviral integration site in clonogenic hematopoietic and osteoblastic cells and the absence of cell fusion. Further evidence for the presence of a common circulating hematopoietic/mesenchymal stem cell is provided by Hayakawa et al., who reported that bone marrow cells, including MSCs, can be reconstituted in lethally irradiated WT mice with GFP+ bone marrow cells obtained from GFP-transgenic mice.11

Evidence from ectopic bone formation experiments

Otsuru et al. examined the ability of circulating osteogenic cells to contribute to bone morphogenetic protein 2 (BMP-2)-induced ectopic bone formation in a mouse model.12 After lethal dose irradiation and subsequent GFP-transgenic BMT, GFP+ osteocalcin+ osteoblastic cells were found in the newly formed ectopic bone. Also, transplantation of GFP+ peripheral blood mononuclear cells isolated from BMP-2-implanted GFP-mice to BMP-2-implanted WT nude mice led to accumulation of GFP+ osteocalcin+ cells in the ectopic bone, suggesting that these cells are circulating cells in peripheral blood. Using a parabiosis model, these circulating osteogenic cells have been characterized as CD45− CD44+ CXCR4+, a phenotype similar to that of plastic-adherent bone marrow MSCs, suggesting that BMP-2 and tissue injury can mobilize MSCs from marrow to peripheral blood.13

Evidence from human diseases

Few studies have examined the presence of donor MSCs in bones of patients who have received successful BMT. Koc et al. examined MSCs were transferred to allogenic hematopoietic stem cell transplant recipients for treatment of lysosomal or peroxisomal storage diseases.14 Bone marrow MSCs were cultured from 13 patients 1 to 14 years after transplantation. Despite successful donor type hematopoietic engraftment, there was no evidence for the presence of donor MSCs, based on fluorescent in situ hybridization analysis using probes for X and Y chromosomes in gender-mismatched transplantations or radiolabeled polymerase chain reaction amplification of polymorphic simple sequence repeats.14 This suggests that hematopoietic stem cells do not give rise to MSC-derived osteoblastic cell populations, at least in recipients without primary bone disease. In contrast, BMT was shown to improve lethal osteogenesis imperfecta (OI), and transplantation of plastic-adherent bone marrow MSCs resulted in engraftment and clinical improvement in 5 of 6 patients with severe OI.15 In another study, Suda et al. reported the presence of donor-derived COP cells in patients that received gender-mismatched hematopoietic stem cell transplantation.16 However, examination of the bone or bone marrow-derived MSC population for the presence of donor cells was not carried out in this study. Taken together, human studies using current technologies may suggest the possibility of a common hematopoietic/skeletal (stromal) stem cell.

Evidence from MSC mobilization studies using pharmacological agents

Hong et al. were able to isolate circulating osteogenic stem cells from peripheral blood after injection with substance P in mice and rats, and they were characterized as present in the peripheral blood mononuclear fraction CD45− CD29+ and capable of mesoderm-type cell differentiation (osteoblast, adipocyte, and chondrocyte).17 It is assumed that substance P is capable of mobilizing bone marrow resident stem cells to peripheral circulation. Pitchford et al.18 reported peripheral blood enrichment of MSC-like cells (plastic adherent, CD29+, CD105+, CD34−, CD45−, vascular endothelial [VE]-cadherin, von Willebrand factor [vWF]) after treatment with vascular endothelial growth factor and the CXCR4 antagonist AMD3100, suggesting mobilization from bone marrow to peripheral blood. Interestingly, treating plastic-adherent human bone marrow MSCs with 1,3-fucosyltransferase, which converted the native CD44 glycoform into the hematopoietic cell E-selectin/L-selectin ligand, enhanced MSCs' ability to home into bone marrow after intravenous infusion.19

Also, Tondreau et al. demonstrated that granulocyte–colony-stimulating factor (G-CSF)-mobilized CD133+ cells in peripheral blood contain plastic-adherent MSC-like cells capable of osteoblast differentiation.20 In this study, the number of colony-forming unit fibroblastics (surrogate markers for the number of clonal MSCs) was very low in the CD133+ fraction, suggesting that G-CSF did not mobilize MSCs from bone marrow, but the presence of other hematopoietic cells in the CD133+ cell fraction enhanced the growth of MSCs.21

In conclusion, these studies suggest that two populations of COP cells are present. One is related to hematopoietic stem cells, and the other to the plastic-adherent MSCs. Both populations by experimental inference originate from bone marrow. Also, some cells can be recruited to sites of new bone formation induced by tissue injury and/or BMPs. The contribution of these cell populations in experimental models of noninjured tissues is limited.

Isolation and Identification of COP Cells

A number of methods have been described to isolate cells with osteoblastic potential from peripheral blood. We group these cells according to the phenotypic description reported (Fig. 1).

Figure 1. Characterization of circulating osteogenic precursor cells and their possible lineage derivations. - - - = presumptive relationships; BF = bone formation; diff = differentiation; NA = nonadherent; OB = osteoblast.

Plastic-adherent MSC-like cells

Bone marrow MSCs are traditionally isolated in ex vivo cultures from the low-density mononuclear cell fraction of bone marrow by adherence to plastic surfaces. Thus, when examining for the presence of circulating osteoblastic cells, several authors employed similar methodology. Kuznetsov et al. identified circulating MSCs (termed circulating connective tissue precursors22 or circulating skeletal stem cells23) from human and experimental animal peripheral blood. The authors isolated plastic-adherent osteonectin+ osteopontin+ Col I+, alpha smooth muscle (ASM)+ CD45− and endothelial marker-negative cells. The frequency of these cells was extremely low in humans, but they were more abundant in experimental animals (e.g., guinea pigs).22, 23 Interestingly, Zvaifler et al. modified this method and reported a more successful isolation of MSCs from peripheral blood of healthy volunteers with a similar phenotype.24 Rochefort et al. reported that the frequency of plastic-adherent MSCs present in peripheral blood increased in a rat model for chronic hypoxia, suggesting that pathophysiological conditions including hypoxia can mobilize MSCs from bone marrow to peripheral blood.25 Also, Otsuru et al. isolated plastic-adherent CD45− CD44+ CXCR4+ cells from murine peripheral blood and demonstrated their homing to sites of BMP-2-induced ectopic bone formation.12, 13 In addition to the success of isolating circulating MSCs from peripheral blood of postnatal organisms, circulating MSC-like cells were also isolated from umbilical cord blood.26, 27

Plastic-nonadherent cells

Long et al. reported that bone marrow plastic-nonadherent cells, isolated by osteocalcin (OCN) or osteopontin antibodies, may represent a more primitive osteoprogenitor cell population based on their clonal growth kinetics.28, 29 Eghbali-Fatourechi et al. extended these studies to isolate a plastic-nonadherent cell population from peripheral blood.30, 31 Similarly, these investigators employed osteocalcin and ALP antibodies to show that osteocalcin+ cells exhibit mononuclear cell morphology, and up to ∼50% of these cells express CD34. However, these cells seem to exhibit under standard in vitro culture conditions low growth rates and less robust osteoblast differentiation capacity compared with bone marrow MSCs. 30, 31 Unfortunately, plastic adherence is not a standardized procedure and varies with time or the type of plastic surface employed in different laboratories; thus, it may be more appropriate to base ontological or hierarchical relationships on specific cellular phenotypes.

Cells belonging to the vascular lineage

There is a well-known intimate relationship between vasculature elements and bone formation. Thus, several investigators have examined the idea of differentiation plasticity of progenitor cells from vascular and bone tissue. A common progenitor (termed mesoangioblasts) for endothelial cells and mesodermal cells (osteoblasts, adipocytes, myocytes) has been hypothesized and characterized in cultures of embryonic dorsal aorta.32, 33 Mesoangioblasts are Flk1+ CD34+ VE-cadherin+ ASM+ cells. 32, 33 It is possible that a counterpart of this cell population is present in the bone marrow of postnatal organisms.34, 35

More direct evidence for a putative contribution of vascular-derived cells to bone formation is that circulating endothelial cells can acquire an osteoblast-like phenotype in vitro and enhance fracture healing when applied locally.36 Finally, in a lineage-tracing study to identify cells recruited to sites of ectopic bone formation in mice, Lounev et al. demonstrated that Tie2+ cells, but not MyoD+ (muscle cells) or smooth muscle myocin heavy chain+ cells, made significant contributions to the newly formed bone.37 Tie2-expressing cells do not exclusively demonstrate the endothelial lineage, and hematopoietic cells also express Tie2. However, BMT studies have shown that cells of the hematopoietic system do not contribute to the fibroproliferative or chondrogenic stages of BMP-induced heterotopic ossification (HO) in a mouse model and in the case of human BMT.38 Further, vascular endothelial cells may transform into multipotent stem-like cells by an activin-like kinase-2 (ALK2) receptor-dependent mechanism.39 Endothelial markers are expressed in chondrocytes and osteoblasts in lesions from individuals with fibrodyplasia ossificans progressiva (FOP), a disorder of HO due to activating ALK2 mutations, or in lesions from transgenic mice expressing constitutively active ALK2.39 Expression of constitutively active ALK2 in endothelial cells may cause an endothelial-to-mesenchymal transition and acquisition of a stem cell-like phenotype. In chondrogenic lesions from subjects with FOP, coexpression of Tie2 and vWF with SOX9 was demonstrated, whereas in osteogenic lesions cells showed strong coexpression of Tie2 and vWF with osteocalcin in cells lining the calcified tissue.

Fibrocytes, monocytes, and other cells from the hematopoietic lineage

Several cell populations present in peripheral blood with a hematopoietic lineage phenotype have been characterized by several investigators as having osteogenic potential. Matsumoto et al. demonstrated that the human peripheral blood CD34+ cell fraction contains a minor population of osteocalcin+ cells that upon intravenous infusion were identified at bone formation sites in a rat femoral fracture model.40 Several studies have examined the biological characteristics of circulating connective tissue cells termed fibrocytes. The cells were first identified in in vivo assays of wound repair models.41 Fibrocytes are cultured from the mononuclear cell fraction of peripheral blood through adherence to fibronectin-coated plastic and characterized by fibroblast-like morphology and a combined hematopoietic CD34+, CD45+, CD13+ and stromal mesodermal phenotype typified by Col I+. Fibrocytes exhibit variable levels of CXCR4 expression and differentiate into mesoderm-type cells (e.g., osteoblast, adipocyte, chondrocyte), but the efficiency of differentiation is low, and no evidence for in vivo bone formation has been demonstrated.42 Fibrocytes are a relatively abundant cell type and represent 0.1% to 1% of the nucleated cells in peripheral blood of healthy donors, and in a number of studies the cells have been shown to be involved in stimulation of T cells, wound healing, and pathological fibrosis.41, 43–50

A quite related cell type has been described in the literature as monocytic cells. Kuwana et al. identified monocyte-derived mesenchymal progenitors (MOMPs) in the peripheral blood mononuclear cell fraction based on adherence to fibronectin-coated plastic plates.51 The cells exhibited initially monocyte-like morphology and then acquired fibroblast-like morphology in culture. The cells are CD14+ CD45+ CD34+ Col I+ and can differentiate into osteoblastic, adipocytic, and myocytic cells using standard differentiation protocols.51 Suda et al. corroborated these results and described the cells as a CD14+ CD34+ CD45+ Col I+ ALP+ Tie2+ population that formed heterotopic bone in vivo upon implantation in immune-deficient mice.16 Similar approaches have been reported to isolate stem cells with wide differentiation potential from peripheral blood monocytic cells.52, 53 Thus, MOMPs and fibrocytes seem to be identical cell populations. The phenotypic difference in CD14 expression between MOMPs and fibrocytes may be caused by in vitro culture conditions, because in some experiments fibrocytes were shown to be derived from CD14+ CD16− CXCR2+ peripheral blood monocytes.43 Also, freshly isolated fibrocytes are CD14− CXCR4+ CXCR2−54 or CD14+ cells that lose CD14 positivity with time in culture.16

With a BMP-inducible model of ectopic bone formation and a Cre-lox system to generate myeloid-restricted β-galactosidase expression, specific activity was detected in newly formed chondrocytes.55 That early chondrocyte progenitors may be of myeloid origin is supported by other studies in which myeloid precursor cells can transdifferentiate into chondrocyte-like cells or MSC-like adherent cells can be established from the peripheral blood of a patient with acute myeloid leukemia.56, 57

Possible Physiological and Pathological Functions

Several studies have suggested that COP cells participate in a number of physiological processes, including long bone development and fracture healing. For example, there were more than five times as many osteocalcin+ COP (plastic-nonadherent) cells in the circulation of adolescent boys with markers of bone formation consistent with pubertal growth as compared with adults, and their number correlated with serum levels of insulin-like growth factor (IGF) I and IGF-binding protein 3.30 Also, Eghbali-Fatourechi et al. reported an increase in the number of circulating osteocalcin+ COP cells in three men after recent fractures.30 In another study, circulating plastic-adherent COP cells (MSC-like; CD105+, CD73+, CD90+, CD45−, CD14−) were identified in peripheral blood from 22% of hip fracture patients and 46% of younger fracture patients and in none of an age- and sex-matched group of women with hip osteoarthritis.58

Animal studies have provided supportive data for these findings in humans. In the above-mentioned studies of Kumagai et al. in which parabiotic animals were formed by surgically conjoining transgenic mice constitutively expressing GFP and syngeneic WT mice, a transverse fibular fracture was created in the contralateral hind limb of the conjoined WT partner and assessed for the contribution of circulating cells to the fracture callus.7 Based on analysis of GFP+ cells and colocalization of ALP staining, histomorphometric analysis of the fracture callus revealed a significant increase of GFP+ ALP+ cells 2 and 3 weeks after the fracture compared with nonfractured controls. Interestingly, bone healing assessed by biomechanical, radiological, and histological criteria was significantly enhanced by human CD34 cell transplantation in a nonhealing femoral fracture model in nude rats40 and by plastic-adherent murine MSC-like and CXCR4+ cells in a stabilized tibia fracture mouse model.59

Taken together, these data are consistent with the idea that COP cells are mobilized to sites of fracture and may contribute to osteogenesis in the early stages of fracture healing. These studies, however, have some limitations that include a small sample size, incomplete understanding of the kinetics of COP cell recruitment during fracture healing, and inconsistencies in markers used to identify COP cells. Also, evidence for the participation of COP cells in fracture healing is based on their anatomical presence at bone formation sites or on the association between circulating levels with fracture-healing status. In addition, the presence of MSCs in the fracture hematoma60 may represent simple leakage from bone marrow caused by the fracture itself.

There are many pathophysiological processes in which COP cells have also been implicated. A causative role for COP cells in extraskeletal ossification or HO has been actively pursued. HO can be found at soft-tissue sites in pulmonary, vascular, cardiac, and periarticular locations. Pulmonary ossification may be idiopathic or due to fibrosing lung disorders, pulmonary venous hypertension, or conditions that increase calcium-phosphate product levels.61 Given the strong evidence suggesting circulating fibrocytes' involvement in pulmonary fibrosis,49 it is possible that fibrocytes can mediate both fibrosis and ossification in the lung.

There is increasing evidence that COP cells participate in HO after hip arthroplasty, in end-stage aortic valvular disease, and in a genetic syndrome of extraskeletal bone formation.16, 62, 63 Animal models of ectopic bone formation also support the notion of COP cell involvement in lesion formation.12, 13, 16 In a rare genetic disorder of HO, FOP, patients with active episodes of extraskeletal bone formation have higher numbers of clonally derived COP cell colonies than patients with stable disease or unaffected individuals, and these COP cells are present in early fibroproliferative lesions.16

The process of vascular calcification was formerly considered the result of passive calcium deposition but is now recognized to be an active pathophysiological process resulting in de novo bone formation in the late stages.64 Several different cell types with mineralization potential have been isolated from vascular tissue: pericytes in microvessels, calcifying vascular cells in the intima, and myofibroblasts in the adventitia. However, the origin of these cells and the mechanism(s) of ossification are unknown.65–69 Ossification is present in about 13% to 15% of carotid endarterectomy and stenotic aortic valve specimens.70, 71 The existence of COP cells in patients with end-stage aortic valve disease suggests that these cells are involved in the process of vascular ossification where there is preexisting calcification.63 This observation is also supported by reports showing that levels of osteocalcin-positive cells are elevated in patients with peripheral arterial disease and that expression of osteocalcin in endothelial progenitor cells is increased in patients with coronary artery disease.72, 73 Ossified lesions in vascular disease develop in the setting of injury and inflammation,70 suggesting that ectopic ossification, regardless of location, shares initiating events.

The inflammatory joint fluid and synovium of patients with rheumatoid arthritis contain cells similar to COP cells, suggesting that degenerative bone and joint deformities, perhaps osteophyte formation, may be associated with these cells.74

Recent data also suggest that COP cells may reflect changes in bone remodeling due to metabolic bone disease. Using antibodies against osteocalcin and stem cell markers CD34 and CD146, COP cells were identified by flow cytometry in patients with hypoparathyroidism and in control subjects.75 Osteocalcin+ cell numbers were lower in hypoparathyroid subjects than in controls, but with parathyroid hormone [PTH(1-84)] administration, the number of cells increased threefold. Coexpression of CD34 and CD146 among osteocalcin+ cells was greater in hypoparathyroid subjects, but decreased with PTH treatment. An increased number of COP cells correlated with circulating and iliac crest histomorphometric indices of osteoblast function. COP cells may thus reflect skeletal osteoblast activity, and the anabolic properties of PTH may be related to increases in COP cell number and maturity. Changes in COP cell differentiation and gene expression may also be associated with bone remodeling. For example, circulating MSCs in patients with osteoporosis are increased, but undergo aberrant osteogenic differentiation.76 Similarly, changes in gene expression in circulating ALP+ cells may reflect rates of bone loss in postmenopausal women.77

Bone regeneration, formation, and remodeling after bone marrow ablation may also be mediated by COP cells in response to PTH.78 After ablation, repopulation and regeneration of bone tissue in PTH^−/− mice was delayed relative to WT mice. However, when ablated long bones from WT mice were transplanted into the back muscle of PTH intact β-galactosidase transgenic mice, Lac-Z-positive osteoblastic and fibroblastic cells were found in the newly formed bone, presumably by recruitment of circulating mesenchymal progenitors.

Cellular and Tissue Origins

It is generally considered that bone marrow is the origin of COP cell populations because these cells express hematopoietic cell markers. However, the putative bone marrow stem cells giving rise to COP cells have not yet been isolated prospectively. It is possible to consider COP cells as representing a number of cell populations that span a continuum of phenotypes from hematopoietic cells to plastic-adherent stromal cells with several intermediate forms (Fig. 1). As suggested in our model (Fig. 1), it is possible that these cell populations branch off at different developmental stages from hematopoietic/stromal stem cells in bone marrow. According to our model, bone marrow plastic-adherent MSCs are one of the primary cell populations that are mobilized to peripheral blood under pathophysiological conditions (e.g., hypoxia, tissue injury, fracture) or by direct or indirect pharmacological mobilization (e.g., substance P). On the other hand, hematopoietic stem cells may give rise to vascular cells, fibrocytes, and MOMPs under similarly stressful conditions. For example, Falla et al. showed that bone marrow obtained from mice treated with 5-fluorouracil (5-FU) (a drug that eliminates all bone marrow dividing cells) was enriched with osteogenic, bone nodule-forming cells in vitro, and these cells were present in the nonplastic-adherent, low-density nucleated cell fraction.79 A working model for the possible relationships between bone marrow-MSC and COP cells is illustrated in Fig. 1.

Migration of Circulating Osteogenic Cells

In both physiological and pathophysiological processes in which COP cells may play etiologic or ancillary roles, a common mechanistic link may be homing to sites of injury, inflammation, or relative hypoxia. In individuals undergoing normal physiological growth or repair of bone tissue, COP cells may be recruited to target tissues by inflammatory signals, such as during fracture repair, or by signals released from a hypoxic microenvironment, such as the growth plate during long bone development. For example, possible explanations for elevated levels of COP cells during pubertal growth may be formation of an oxygen gradient associated with intense remodeling and, as suggested by Canalis,80 that circulating osteogenic cells then return to the skeleton where they may function as mature osteoblasts. In individuals predisposed toward pathological ossification, injury is a usual precipitating factor—either obvious injury, as in the case of traumatic HO, or microscopic injury, as in the case of early endothelial dysfunction associated with eventual vascular calcification and ossification.

The basis for homing of COP cells to areas of injury and inflammation is best described for HO. Extraskeletal bone formation can be precipitated by soft-tissue injury in skeletal muscle, causing the presumptive release of inflammatory cytokines and migratory factors (Fig. 2). Inflammatory signals appear to be necessary for BMP-induced HO, and cells of the monocyte lineage appear to be necessary for triggering ectopic bone formation after injury.81

Figure 2. Putative mechanism for COP cell homing in heterotopic bone formation.

In an inflammatory milieu, stromal cell-derived factor 1 (SDF-1) and BMP may serve as important chemoattractant molecules. 49, 82, 83 It is well documented that SDF-1 (CXCL12) is induced by hypoxic tissue injury and forms a gradient that attracts cells expressing its cognate receptor CXCR4. Release of COP cells from bone marrow and their homing to sites of injury may thus occur through CXCR4 (Fig. 2). This homing mechanism is established in fibrocyte localization to lesions of pulmonary fibrosis.49 It is also established in a mouse model of BMP-2-induced HO, where osteoprogenitor cells expressing CXCR4 migrate from the bone marrow to regions of ectopic bone formation by SDF-1 chemoattraction.12, 13 BMP may play roles in both bone formation and attraction of inflammatory cells.84 BMP stimulation of brown adipocytes may also lower tissue oxygen tension,85 thereby promoting local chondrogenesis and inducing SDF-1. The homing of COP cells may then ossify the cartilage template. It is also plausible that COP cells that display a vascular/hematopoietic phenotype play a regulatory role in bone formation through their response to humoral factors.

The SDF-1/CXCR4 axis may be the final common pathway for mobilization of bone marrow-derived progenitor cells by hypoxia, angiogenic peptides, inflammatory cytokines, and injury.86–90 This axis has been implicated in processes as diverse as development, regeneration, and tumorgenesis/metastasis. It is not so surprising, then, that the same homing mechanism may be involved in COP cell mobilization and targeting. Having such an ubiquitous mechanism in place for trafficking of progenitor cells implies that the consequences of COP cell migration and their appearance in target tissues depend not only on the plasticity of the COP cell, but also on the microenvironment in which the COP cells ultimately reside.

Prospectus

Studies of COP cells have revealed the complex nature of the cellular basis of bone formation under physiological and pathophysiological conditions. The skeletal biologist is moving toward phenotypical mapping of the bone-forming cell population that needs to be developed in an analogous manner to cells of the hematopoietic system. Future studies are thus needed to clarify these ontological and hierarchical relationships between different populations of COP cells based on fine mapping of surface markers as well as functional analysis.

At our current state of knowledge, several key concepts appear to be generally true regarding the physiological functions of COP cells. The first is that the bone-forming function of COP cells may not be their primary role, but is an adaptive response in conditions of injury, repair, or abnormal cytokine signaling. The ultimate fate of COP cells may be to participate in tissue regeneration, which under certain circumstances dictates de novo bone formation, depending on the microenvironment that COP cells are drawn to and in which they may otherwise assume different roles. A corollary of this hypothesis is that local MSCs serve as the primary osteochondro-progenitors, whereas COP cells likely play a role in bone formation at nonskeletal sites and during tissue injury (e.g., fracture healing).

The second emerging concept is that COP cell homing may be mediated by the CXCR4/SDF-1 axis that is shared by multiple processes requiring the migration of stem cells. That COP cells may share a common mechanism for progenitor cell migration is not so surprising, especially given that their putative roles are likely precipitated by injury and inflammation. Their ultimate fate then must be dictated not only by where they go, but also by how they get there. However, several details are needed to confirm these hypotheses, based on prospective identification of COP cells using defined criteria and testing their functionality using well-controlled in vivo assays and studies to identify mechanisms underlying their dual nature as bone-forming and bone formation-regulating cells.

In addition, COP cells may potentially be useful in gene and cell therapy protocols to enhance bone formation, remodeling, or regeneration. Although formal studies using COP cells in therapies for musculoskeletal defects and injuries are lacking, one can propose based on their presence in the fracture hematoma and callus that they may play a role amenable to their use as a cell-based adjunctive treatment in fracture healing. Use of COP cells in other types of musculoskeletal injury would be even more speculative at present.

Finally, the burgeoning area of COP cell biology holds promise for development of diagnostic tests based on COP cell levels as a biomarker for disease states.

Disclosures

Both authors state that they have no conflicts of interest.

Acknowledgements

Supported by National Institutes of Health grants R01AG028873 (RJP) and AG025929 (RJP); the Ian Cali Endowment/University of Pennsylvania Center for Research in FOP and Related Disorders Developmental Grant Award (RJP); and grants from the Novo Nordisk Foundation (MK), Lundbeck Foundation (MK), and the region of Southern Denmark (MK).