Bone marrow is a reservoir of pluripotent stem/progenitor cells for mesenchymal tissues. Upon in vitro expansion, in vivo bone-forming efficiency of bone marrow stromal cells (BMSCs) is dramatically lower in comparison with fresh bone marrow, and their in vitro multidifferentiation potentials are gradually lost. Nevertheless, when BMSCs are isolated and expanded in the presence of fibroblast growth factor 2, the percentage of cells able to differentiate into the osteogenic, chondrogenic, and adipogenic lineages is greater. Osteogenic progenitors are not exclusive to skeletal tissues. We could also think of cells in different adult tissues as potentially capable of following an osteochondrogenic differentiation pathway, but, under normal physiological conditions, they are inhibited in this process by the environment and/or the adjacent cell populations. When, for some reason such as pathology, the environment changes dramatically and the inhibiting condition is removed, these cells could become osteoblasts. Bone is repaired via local delivery of cells within a scaffold. Bone formation was first assessed in small animal models. Large animal models were successively developed to prove the feasibility of the tissue engineering approach in a model closer to a real clinical situation. Eventually, pilot clinical studies were performed. Extremely appealing is the possibility of using mesenchymal progenitors in the therapy of genetic bone diseases via systemic infusion. There is experimental evidence to suggest that mesenchymal progenitors delivered by this route engraft with a very low efficiency and do not produce relevant and durable clinical effects. Under some conditions, where the local microenvironment is either altered (i.e., injury) or under important remodeling processes (i.e., fetal growth), engraftment of stem and progenitor cells seems to be enhanced. A better understanding of their engraftment mechanisms will, hopefully, extend the field of therapeutic applications of mesenchymal progenitors.
Stem and progenitor cells from adult tissues represent an important promise in the therapy of several pathological conditions. Stem cells have the ability to self-replicate for long periods or, in the case of adult stem cells, maintain their differentiation potential throughout the life of the organism. Progenitor cells are derived from stem cells; they retain the differentiation potential and high proliferation capability, but they have lost the self-replication property. Adult stem cells were first identified in tissues characterized by a high rate of cell turnover, such as marrow. Recently, stem cells also have been isolated from adult organs previously thought to be capable of limited self-repair, offering the prospect of new therapeutic strategies for repair of damaged tissues [1, 2].
From a classical view, adult stem cells are long-lived cells restricted to differentiating along the lineage pathways of their own tissue. Similarly, progenitor cells have been considered committed to the cell phenotypes of their tissues of origin. This concept of lineage restriction has been challenged by experimental evidence over the past few years. Indeed, most stem and progenitor cell types display an amazing plasticity, which is the property of cells to differentiate into phenotypes not restricted to the tissues and, in some cases, to the germ layers from which they are derived.
Bone marrow (BM) transplantation made it possible to follow the fates of transplanted cells in vivo by designing experiments in which donor cells were distinguishable from those of the recipient. In a similar experimental setting, although often at a very low frequency, BM-derived cells of adult donors were found integrated in the parenchyma of a variety of nonhematopoietic tissues, such as liver, brain, vascular wall, skeletal, and cardiac muscle, and even in skin, lung, and digestive tract epithelium .
In this review we discuss the prospect of using stem/ progenitor cells in the cell therapy of bone, taking into consideration the properties of osteoprogenitor cells from different sources and the influence of the microenvironment on cell delivery and differentiation.
Osteogenic Progenitors from BM
BM is not only the site where hematopoiesis physiologically occurs in postnatal life, but it is also a reservoir of pluripotent stem cells for mesenchymal tissues. Hematopoietic and mesenchymal progenitors reside in close contact in the postnatal BM cavity. Hematopoietic stem cells (HSCs) in this cavity are functionally and structurally supported in their blood-forming abilities by a complex structure that, as a whole, forms the BM stroma network [4, 5]. This stroma is composed of a heterogeneous variety of cell types, among which BM stromal cells (BMSCs) are essential elements of the hematopoietic microenvironment through both direct contact with cell surfaces and supply of soluble mediators [6-9].
Isolated for the first time from a BM cell suspension by Friedenstein and coworkers as adherent, fibroblast-like, and clonogenic cells , BMSCs were demonstrated to be multipotential in that they not only acted as myelosupportive stroma, but could differentiate into osteoblasts, chondrocytes, and adipocytes when implanted in vivo [11, 12]. Their ability to generate almost all the mesenchymal lineages of connective tissues has strengthened the idea that they represent, or at least contain, a population of mesenchymal stem cells from which all mesenchymal lineages originate under the influence of different microenvironments [12-14].
Plated at low densities, single precursor cells derived from BM, and referred to as colony-forming units-fibroblastic, give rise to distinct and heterogeneous colonies. These colonies have been shown to undergo osteogenic, chondrogenic, and adipogenic differentiation. Analysis at the single colony level depicts the heterogeneity of the original BMSC population, from which clones can be obtained . Indeed, the heterogeneity of BMSCs at the clonal level is also reflected in the in vivo assay, where only 60% of clones form bone . Nevertheless, when clones are isolated and expanded in the presence of fibroblast growth factor 2 (FGF-2), the frequency of clones able to differentiate into the osteogenic, chondrogenic, and adipogenic lineages is greater, even though “tripotential” clones progressively lose their differentiation potentials . Furthermore, in the in vitro hierarchy observed, the adipogenic lineage diverges and becomes independent early, whereas the osteogenic and chondrogenic lineages proceed together, possibly diverging later; the in vitro hierarchy can be reconciled with a model proposing lineage hierarchy in vivo . According to that model of differentiation, the osteogenic pathway is the default lineage of this population. Experiments have been performed where the reciprocal control of osteogenic-chondrogenic and osteogenicadipogenic differentiation has been demonstrated in an immortalized cell line of human stromal cells .
Since Till and McCulloch first provided physiological evidence of a multipotential marrow cell , HSCs have been isolated and characterized by several authors . HSCs probably represent the best-studied prototype of adult stem cells. Throughout the lifetime of an individual, they generate all mature blood cells. HSCs display both of the properties that define stem cells: self-renewal and multidifferentiation capability. Given the low tissue turnover in most adult connective tissues [12, 20], the above-mentioned criteria for stem cells might not apply entirely to cells of postnatal mesenchymal tissues. It has been proposed that, in the case of mesenchymal tissues, self-renewal is less crucial than cell plasticity and phenotypic flexibility . In keeping with this hypothesis, most relevant is the fact that the commitment of cells of mesenchymal origin is reversible, probably in response to environmental cues, and that the same cells display the ability of interconversion from one cell type to another at a later stage than that of multipotential stem cells [22-24]. Because of the solid phase, these two features are necessary for growth and remodeling of connective tissues in vivo .
The attractiveness of BMSCs as a source of multipotent mesenchymal cells is also based on the ease with which they can be isolated from a patient's BM and expanded many fold in vitro. For clinical applications, it is of critical importance that the cells used have high regenerative potentials in order to maintain a tissue or a function over time, possibly a lifetime. However, extensive in vitro proliferation seems to affect both the differentiation capabilities and the replicative potentials of BMSCs. In fact, it has been demonstrated that after the first confluence, cultured BMSCs slow down their proliferation rate [25, 26]. Upon in vitro expansion, the in vivo bone-forming efficiency of BMSCs is dramatically lower than with fresh BM, and their in vitro multidifferentiation potential is gradually lost [15, 26]. In fact, in our culture conditions, the largest drop in proliferation rate occurs when other components of the marrow microenvironment, which can be responsible for a complex network of soluble and cell-contact signals to the BMSC, are removed. Furthermore, at a clonal level, a sequential loss of multipotency has been described, the osteogenic potential being maintained longer . Such behavior is suggestive of a progressive loss of stem features with increasing population doublings in culture.
Reyes et al. have attempted to improve culture conditions for the in vitro expansion of BMSCs . To ensure the absence of hematopoietic contaminants, they purified BM cell suspensions from CD45+/glycophorin A+ cells and plated the residual mononuclear cells in wells coated with fibronectin in the presence of epidermal growth factor, platelet-derived growth factor BB, and a low concentration fetal calf serum. Those authors reported that those culture conditions allowed the isolation of mesodermal progenitor cells, which, upon extensive in vitro expansion, retained the capability to differentiate along both mesodermal and nonmesodermal lineages without signs of senescence.
The growth factor requirement for the in vitro expansion of BMSCs has been investigated by several groups [28-34]. Gronthos and Simmons demonstrated that, in the absence of serum, growth factors are necessary for the proliferation of BMSCs in culture . Among the different growth factors analyzed, FGF-2 demonstrated the greatest ability to promote the proliferation of these cells . Although the effectiveness of dexamethasone (Dex) to induce an osteogenic phenotype in BMSCs has been established [35-37], FGF-2 appears to enhance the expression of the osteogenic phenotype even in Dex-treated cultures . In our culture conditions, FGF-2-expanded BMSCs yielded a higher in vivo bone formation than cells expanded in the presence of Dex or other growth factors (Fig. 1).
These data support the idea that FGF-2 maintains cells in a more immature state, allowing for the in vitro expansion of osteoprogenitors, possibly by selecting a specific cell subset of the expanding population. Indeed, together with the greater colony size, FGF-2 affects the colony number, which was about 30% lower with respect to the same BMSC primary culture plated in the absence of the factor . Taking into account the very early phase of the culture, where the supposed selection occurs, we suggest that the mechanism involved might be either impaired adhesion or apoptosis, induced by the factor in a subset of cells, or possibly both. Addition of FGF-2 alone to the culture, even though it notably improved the number of multipotential clones, is not sufficient to prevent their gradual senescence . Further efforts need to be undertaken to optimize BMSC culture conditions to maintain their full stem cell potential in vitro.
Osteogenic Progenitors Are not Exclusive to Skeletal Tissues
As early as the 1970s, Friedenstein and Lalykina isolated single cell suspensions from the thymuses of guinea pigs and rabbits and induced bone formation by implanting those cell types in diffusion chambers in combination with an inductor (i.e., transitional epithelium) . In the same way, Chailakhan and Lalykina isolated osteogenic cells from the spleen . Friedenstein referred to those cell types as inducible osteoprogenitor cells (IOPCs) to distinguish them from BMSCs that form bone spontaneously, and are therefore considered determined osteoprogenitor cells. Inducible osteoprogenitor cells are cells not yet assigned to a specific fate; they retain the potential to form bone but require an inductive stimulus to differentiate .
Following the footsteps of Friedenstein, many researchers have recently reported that cells from adult tissues retain the ability to undergo differentiation along pathways other than those they have been determined to follow.
Earlier experiments by Volek-Smith and Urist  and Van de Putte and Urist  showed how implantation of demineralized bone matrix or recombinant human bone morphogenetic protein 2 in the fascia of skeletal muscle leads to the formation of cartilage and bone. Further evidence of the existence of osteochondrogenic progenitors in skeletal muscle tissue was derived from the study of patients affected with fibrodysplasia ossificans progressiva (FOP). FOP is a very rare disorder where ectopic bone is formed within the skeletal muscles [45-47]. These preliminary observations led several groups to demonstrate that the skeletal muscle harbors selective cell types able to form bone under inductive stimuli.
Levy et al. obtained muscle-derived cells after dissection of muscle tissue subsequent to digestion with collagenase and plating on a collagen-coated plastic dish . They were able to isolate a cell population positive for alkaline phosphatase (ALP) that expressed osteocalcin (as revealed by reverse transcription-polymerase chain reaction). Furthermore, Asakura et al. showed that, when stimulated with bone morphogenetic protein-7, muscle satellite cells expressed ALP and osteocalcin and differentiated into mature osteoblasts after culture on matrigel coating .
Given the fact that bone formation is often accompanied by intense vascularization, a hypothesis was raised that cells strictly associated with the capillaries, such as pericytes, might represent an additional source of osteoprogenitor cells provided by the incoming and newly formed microvasculature observed during bone formation. Certainly, physiological similarities exist between pericytes and BMSCs. Indeed, some authors have suggested that marrow pericytes and marrow stromal cells are the same entity .
The pericyte is a cell of mesenchymal origin, with long cellular processes that encircle capillaries. Pericytes have been isolated from the microvasculature of connective tissues; nervous tissue including the cerebral cortex, peripheral nerves, and the retina; muscle tissue; and the lungs [51-57].
Diaz-Florez et al. labeled pericytes and endothelial cells with monastral blue and, consequently, followed the fate of the pericyte in the process of new bone formation . Six days after activation of the periosteum, the dye was found in newly formed osteoblasts. More recently, Doherty et al. demonstrated that pericytes isolated from bovine retina (the most easily accessible source of this cell type) not only expressed specific osteogenic markers, such as osteopontin, osteonectin, and bone sialoprotein, but were also able to form bone and cartilage spontaneously when implanted in diffusion chambers in vivo .
Incoming vasculature carries not only pericytes but also the so-called vascular smooth muscle cells (VSMCs) that contribute to the wall architecture. Bostrom et al. and Watson et al. have isolated and cultured VSMCs from the aorta wall and observed spontaneous mineralization in culture, as shown by the appearance of calcified nodules [58, 59]. In particular, in a cut-through discovery, it was shown that, not only were these cells able to induce mineralization, but they were also actively producing bone-specific matrix proteins such as osteopontin, osteonectin, and matrix Gla protein [60, 61]. Furthermore, Halvorsen et al. isolated stromal cells from human adipose tissue and were able to induce bone formation in vitro by culturing the cells in the presence of Dex and vitamin D .
Where does the truth stand? Are differentiated cells able to undergo “dedifferentiation” and regress toward a more primitive phenotype, thus being able to redirect their fate toward a different one? Alternatively, are we in the presence of a pool of quiescent primitive multipotent cells (IOPCs) that become activated under particular circumstances? We could also think of cells in adult tissues as potentially capable of following an osteochondrogenic differentiation pathway, but, under normal physiological conditions, they are inhibited in this process by the environment and/or the adjacent cell populations. When, for some reason such as pathology, the environment changes dramatically and the inhibiting condition is removed, these cells could become osteoblasts.
In syndromes like FOP , progressive osseous heteroplasia, and osteoma cutis , the formation of ectopic bone is always preceded by a dramatic inflammation event . In interleukin-5 (IL-5) transgenic mice that were developed to study the role of IL-5 as an asthma effector, splenomegaly (the spleens of these mice are 50 times larger than those of the wild type) was observed due to an increase in white blood cells and, particularly, eosinophils. Surprisingly, bone nodules developed in the spleens of those mice .
These observations suggest a relationship between inflammatory agents and ectopic bone formation. Environmental changes, consequent to infiltration of lymphocytes and or eosinophils and to alterations in the nature and amount of cytokines, could affect the quiescent residing pool of multipotent stem cells acting as inducers or removing the blockage of an osteochondrogenic differentiation pathway.
Repair of Large Bone Defects by BMSCs
Several pathological conditions lead to an extensive loss of bone tissue (trauma, inflammation, surgical treatment for neoplasias). The reconstruction of large bone segments remains an important clinical problem. None of the approaches proposed thus far has proven to be ideal. The implantation of engineered bone graft material, where an osteoconductive scaffold is combined with osteogenic-committed cells, could represent a valid alternative.
Bone formation was first assessed in small animal models by implanting BMSCs from various species with porous bioceramics subcutaneously in syngeneic rats  or immunodeficient mice [32, 67-69] and for the repair of small experimentally induced osseous defects [70, 71]. In some cases, transplanted osteoprogenitor cells/ceramic scaffolds were used to create vascularized bone flaps [72, 73] in an attempt to combine a traditional approach with an innovative technique.
Large animal models were successively developed to prove the feasibility of the tissue-engineering approach in a model closer to a real clinical situation [68, 74-76]. In all of these models, the approach was similar: a large segmental defect was created in a long bone and the gap obtained was filled with a cylinder of porous bioceramic carrying autologous in vitro-expanded osteogenic progenitors. These studies differed substantially in the animal model (dog or sheep), anatomical segment (femur or tibia), chemical composition, geometry, and resorbability of the biomaterial used. Still, the results were very similar, indicating an important advantage in bone formation and, therefore, in the healing of the segmental defect when stromal osteoprogenitors were delivered together with the bioceramic scaffolds. Recently, we transposed this cell-based tissue-engineering approach to the clinic to treat patients with large (4-7 cm) bone defects . Autologous osteoprogenitors were isolated from the BM of the patients and expanded ex vivo. Cells were delivered in vivo with macroporous hydroxyapatite scaffolds. External fixation was provided initially for mechanical stability. Abundant callus formation along the implant and good integration at the interface with the host bone are observed by the second month after implantation (Fig. 2). All patients recovered limb function between 6 and 12 months after the procedure. Using a traditional approach, under the most favorable conditions and in the absence of complications, the expected recovery time is much longer.
Systemic Injection of Osteogenic Cells
In principle, systemic infusion of osteogenic progenitor cells represents an important therapeutic strategy for the treatment of genetic bone disorders. To this end, functional engraftment of osteogenic progenitors is critical to ensure successful and enduring treatment of disease. While some animal studies suggest the occurrence of osteogenic progenitor engraftment following BM transplant, results on patients who have undergone BM transplantation for cancer treatment are somewhat discouraging.
Following intravenous infusion, marrow-derived mesenchymal precursor cells have been reported to be capable of homing not only to the BM of immunodeficient  or irradiated mice [79, 80] and irradiated baboons , but also in multiple sites such as bone, cartilage, lung, and spleen . On the other hand, in patients who undergo BM transplantation, the BM stromal component is clearly of host origin. By in situ hybridization, karyotype, or polymerase chain reaction analyses, several authors have shown that marrow-derived stromal cells that proliferate in long-term cultures are the host's genotype [82-88]. The only apparent exception is a study by Cilloni et al. that reported the existence of mixed chimerism at the stromal cell level in 2 of 41 transplanted patients, although, in all 41 cases, a successful donor-type hematopoietic engraftment was observed .
Allogeneic BM transplantation also has been attempted for the treatment of severe osteogenesis imperfecta. In a study by Horwitz et al., three children with this disease were treated. Three months after transplantation, the total body bone mineral content was increased and new dense bone formation was observed in trabecular bone [90, 91]. The authors of that study suggested that the improvement in the clinical conditions was possibly due to the engraftment of functional mesenchymal progenitor cells. The durability of the observed improvements remains in question and may depend on both the amount of engrafted cells and the relative frequencies of stem or early progenitor cells engrafted.
A different scenario can be envisaged if the transplant is performed in a particular microenvironment, such as the fetus. In fact, BMSC engraftment has been observed in the sheep fetal microenvironment. Cotransplantation of human BMSCs in fetal sheep improved the long-term engraftment of HSCs, whereas BMSCs transplanted alone remained functional in the hematopoietic microenvironment  and persisted in multiple tissues undergoing site-specific differentiation for as long as 13 months after transplantation .
Can these apparently contradictory sets of results be reconciled? In interpreting animal data reporting isolation of cells of donor origin and expression of reporter markers in tissues and in ex-vivo-cultured cells, some caveats should be taken into consideration.
The presence of donor cells in host tissues may reflect a simple cell survival, either without the acquisition of tissue-specific functions or with the acquisition of a mature cell phenotype as a consequence of cell plasticity rather than a real stem cell engraftment. Indeed, cell plasticity has been well documented in cultured stromal cells that respond to changes in the culture microenvironment (addition of different growth factors and hormones, three-dimensional versus two-dimensional cultures, substrate changes) by expressing differentiation markers specific to different lineages.
A possible alternative explanation might come from recent reports showing that mouse BM cells  and progenitor cells of the central nervous system  spontaneously fuse with embryonic stem cells under certain culture conditions, although at low frequencies (10−5 to 10−6 per cell plated). If BMSCs also fuse in vivo with somatic cells, this could, at least in part, account for donor cells found in recipient tissues and organs.
Tissue Damage and Stem Cell Engraftment
In principle, stem cells could be recruited in particular anatomical sites by microenvironmental cues coming from injured tissues. Results from Ferrari and coworkers corroborate this hypothesis, showing that beta-gal BM-derived cells engrafted in regenerating skeletal muscle fibers after an induced injury . They also observed expression of beta-gal when BM cells were injected directly into the injured muscle. Similar results were obtained by Gussoni and coworkers in a murine model of Duchenne's muscular dystrophy in the absence of external injury . In animal models where liver damage was induced, hepatic progenitors derived from donor BM cells were found integrated in the liver parenchyma [98, 99]. Very interestingly, in a study by Lagasse et al., the efficiency of engraftment was high enough to allow restoration of liver function . Alison and coworkers reported the extrahepatic origin of some of the hepatocytes in patients treated with either liver or BM sex-mismatched transplantation .
Also, in the case of myocardial ischemic damage, engraftment and integration of donor cells at the lesion site have been demonstrated [101, 102]. Engraftment in damaged tissues of different types of stem cells also has been observed by other authors. In myeloablated mice, stem cells derived from skeletal muscle  or brain  engrafted and showed a high level of multilineage differentiation characteristic of hematopoietic stem cells.
Successful engraftment of a population of stem/progenitor cells has been established by demonstrating the long-lasting viability of physically and functionally integrated donor cells. BMSCs have been shown to be capable of functional grafting. Locally delivered together with the proper biomaterials, BMSCs efficiently allow for the healing of bone defects of critical sizes.
Furthermore, the increasing number of studies that provide evidence of unorthodox plasticity, together with the attempt to cure more generalized mesenchymal diseases or to correct genetic defects, has prompted the investigation of the feasibility of systemic injection. Intravenously infused mesenchymal progenitors engraft with very low efficiencies and do not produce relevant and durable clinical effects.
It could be suggested that only recent tissue damage, as in the case of tissue injuries or wounds, or tissue remodeling, as it occurs during fetal development and in some genetic disorders, provides a permissive milieu for the engraftment of adult stem cells. Indeed, in those situations, a common set of proteins is expressed. In this regard, it is to be noted that the same stress proteins are physiologically expressed in tissues where active remodeling is taking place during development and in tissues characterized by an acute phase response due to pathological conditions . During inflammation, specific cell types migrate to the site of damage. These cells interact with resident cells and, as a consequence of those interactions, specific growth factors, soluble proteins, and extracellular matrix macromolecules are released into the area. This may provide the unique microenvironment required for stem cell engraftment.
This work was partially supported by the Italian Space Agency (ASI), the European Space Agency (ESA; ERISTO), and the Italian Ministero Istruzione, Universitá e Ricerca (MIUR). We acknowledge the First International Workshop “Cell Therapy: Filling the Gap Between Basic Science and Clinical Trials” (October 15-17, 2001, Rome).