Laboratoire de Physiopathologie de la Résorption Osseuse et Thérapie des Tumeurs Osseuses Primitives, INSERM, ERI 7, Nantes, France
Université de Nantes, Nantes Atlantique Universités, Laboratoire de Physiopathologie de la Résorption Osseuse et Thérapie des Tumeurs Osseuses Primitives, EA3822, Nantes, France
INSERM, ERI 7 Laboratoire de Physiopathologie de la Résorption Osseuse et Thérapie des Tumeurs Osseuses Primitives, EA3822, 1 rue Gaston Veil, 44035 Nantes Cedex 1. Telephone: 011-33-2-40-41-28-46; Fax: 011-33-2-40-41-28-60
Bone remodeling involves synthesis of organic matrix by osteoblasts and bone resorption by osteoclasts. A tight collaboration between these two cell types is essential to maintain a physiological bone homeostasis. Thus, osteoblasts control bone-resorbing activities and are also involved in osteoclast differentiation. Any disturbance between these effectors leads to the development of skeletal abnormalities and/or bone diseases. In this context, the determination of key genes involved in bone cell differentiation is a new challenge to treat any skeletal disorders. Different models are used to study the differentiation process of these cells, but all of them use pre-engaged progenitor cells, allowing us to study only the latest stages of the differentiation. Embryonic stem (ES) cells come from the inner mass of the blastocyst prior its implantation to the uterine wall. Because of their capacity to differentiate into all germ layers, and so into all tissues of the body, ES cells represent the best model by which to study earliest stages of bone cell differentiation. Osteoblasts are generated by two methods, one including the generation of embryoid body, the other not. Mineralizing cells are obtained after 2 weeks of culture and express all the specific osteoblastic markers (alkaline phosphatase, type I collagen, osteocalcin, and others). Osteoclasts are generated from a single-cell suspension of ES cells seeded on a feeder monolayer, and bone-resorbing cells expressing osteoclastic markers such as tartrate-resistant alkaline phosphatase or receptor activator of nuclear factor κB are obtained within 11 days. The aim of this review is to present recent discoveries and advances in the differentiation of both osteoblasts and osteoclasts from ES cells.
Bone is a specialized connective tissue with elastic and strength properties. Bone tissue has various functions, including mechanical properties (offering the site and support for insertion of skeletal muscle tissue), protective properties for vital organs (heart, brain, bone marrow, etc.), and a pivotal metabolic function. Indeed, bone mineral matrix represents the main reserve of mineral ions, especially calcium and phosphate. Bone is composed of two main cell populations (osteoblast and osteoclast lineages) and extracellular matrix-associating organic (noncollagen proteins, such as fibronectin, and growth factors, collagen types I and III, etc.) and mineral (hydroxyapatite crystals) phases. Bone undergoes cyclic modifications, named bone remodeling, that allow for adaptation to mechanical constraints and maintain homeostasis of phosphorus and calcium. These cyclic modifications correspond to coordinated phases of bone formation by osteoblasts  and bone resorption by osteoclasts . Physiologically, osteoblast lineage differentiates from bone marrow MSC through a series of progenitor stages to form mature matrix-secreting osteoblasts, progressively transformed into osteocytes . The presence of an extensive endoplasmic reticulum and numerous free ribosomes in the cytoplasms is the main cytologic characteristic of osteoblasts; it gives evidence in favor of a strong activity of protein synthesis. Osteoblasts and osteocytes connected by gap junctions constitute a cellular network allowing the communications between bone surface and mineral matrix . Bone resorption is controlled by osteoclast activity. Osteoclasts differentiate from hematopoietic precursors located in bone marrow and are closely related to macrophages [4, 5]. They possess several cytological features (multinucleation, highly polarized morphology, and numerous mitochondria). Osteoclasts resorb mineralized matrix by attaching to the surface and then secreting protons into an extracellular compartment formed under their ruffled border. This secretion is necessary for bone mineral solubilization and the digestion of the organic matrix by acid proteases . In 1965, Epker and Frost were among the first to show that osteoblasts and osteoclasts are closely associated in time and space . Numerous factors contribute to these coordinated activities and include gap junctions , numerous cytokines, growth factors, vitamins, and hormones [8, 9]. Furthermore, any disruption in this balance results in pathological states leading to increase (i.e., osteopetrosis) or loss (osteolytic diseases and osteoporosis) of bone mass.
Embryonic stem (ES) cells come from the inner mass of the blastocyst prior to implantation into the uterine wall. ES cells have important characteristics: they are pluripotent, meaning that a single cell has the capability of developing cells of all germ layers (endoderm, ectoderm, and mesoderm) and can give rise to all tissues of the body. From the ectoderm, ES cells can generate cells of skin, brain, eyes, and neural tissue. From the mesoderm, they can generate bone, cartilage, muscle, heart, or kidneys, and from the endoderm, they can generate liver, pancreas, thymus, thyroid, lung, and so on. ES cells have the potential of long-self renewal in vitro and can be induced to differentiate with appropriate culture conditions and specific factors. In addition to their potential interest in regenerating bone tissue by transplantation, ES cells offer a strong advantage in fundamental research by permitting the study of osteoblast and osteoclast differentiations from the earliest differentiation state of the cell. Indeed, the other models in the literature (and described below) use cells already pre-engaged, and so early steps of the differentiation could not be studied. Furthermore, ES cells are the best model to study the differentiation into mesenchymal cells that give rise to osteoblasts. ES cells are thought to be a powerful system to decipher the molecular mechanisms involved in both osteoblast and osteoclast differentiation, and so from the earlier to the later stages of the differentiation. The present review focuses on the recent advances concerning osteoblast and osteoclast differentiation from ES cells.
From ES Cells to Osteoblasts
Osteoblasts, which derive from the mesoderm, are cells that specialize in the production of extracellular matrix (which is mainly composed of type I collagen) and the mineralization process . They also play an important but indirect role in bone resorption by closed contacts with osteoclasts and by secreting various factors, such as interleukin-6 , receptor activator of nuclear factor κB ligand (RANKL) , and prostaglandin E2 , which in turn stimulate osteoclasts and so bone resorption [13, 14]. A direct role of osteoblasts in bone degradation could also be suggested, as osteoblasts release matrix metalloproteases [15, 16]. Osteoblasts and osteoclasts also interact with cell-to cell contact. After a bone-forming phase, approximately 10%–20% of the osteoblasts become embedded in the bone and differentiate progressively into osteocytes . Study of the differentiation of stem cells into osteoblasts is important for two main reasons. First, large amounts of active osteoblasts are necessary for bioengineering, such as in transplantation used in regenerative medicine. Second, understanding the differentiation program of a cell into osteoblast is of major interest for finding new key genes involved in this process, genes that could be deregulated during pathological processes (bone tumors, osteoporosis, etc.). Studies of osteoblast differentiation are usually performed with preosteoblastic cells, such as the MC3T3 cell line [18, , , , –23], or primary cells isolated from bone [24, , , –28] or from mesenchymal stem cells isolated from bone marrow , adipose tissue [29, 30], placenta , cartilage , and others. All these different models permit the study of osteoblast differentiation and testing of the effects of different factors that may improve this process (Table 1). Thanks to their capacity for differentiation and their presence in adult tissues, MSC represent a good tool to regenerate bone in an autogenous bone graft. In this context, Bruder et al. showed the capacity of human MSC to increase bone formation when implanted onto ceramic carrier into rat femurs . The same kinds of results have been obtained in a model of osteoporotic rabbit, in which implanted MSC were able to regenerate bone . Others studies also demonstrated the capacity of MSC to generate bone, both in vitro  and in vivo, when implanted under the kidney capsule . However, MSC are rare cells in adult tissues and need to be expanded in vitro prior to being used in therapy; the expansion can lead to cellular alterations. Furthermore, the number of divisions in vitro is not infinite. Indeed, Bonab et al. demonstrated that human MSC enter senescence after nine passages and consequently lose their stem cell characteristics and so their osteogenic potential . Furthermore, MSC, as well as the calvaria model or MC3T3, can be considered osteoblastic progenitor cells or pre-engaged cells and so do not allow study of the early steps of osteoblastic differentiation. A new strategy is the use of ES cells, which are also considered new potential cells for transplantation.
Table Table 1.. Main in vitro models described in the literature and used to study osteoblastic differentiation
Many markers are used to estimate the differentiation level of the cells: core binding factor α1 (Cbfa1/Runx2), which is the main specific transcription factor involved in osteogenesis; type I collagen; osteopontin (OP); osteonectin; osteocalcin (OC); bone sialoprotein (BSP); and alkaline phosphatase (ALP), which may affect the mineralization. The capacity of osteoblasts to mineralize in vitro can be visualized by von Kossa or Alizarin red staining. In vitro differentiation of ES cells into osteoblasts needs the addition of several factors, which in vivo are released by cells from the microenvironment of osteoblasts. Among these factors, ascorbic acid and 1α,25-OH vitamin D3 are mandatory for matrix deposition and mineralization, which are visualized as hydroxyapatite-containing mineral nodules when a source of organic phosphate is added to the culture (for example, β-glycerophosphate). This osteogenic medium has already been shown to induce osteoblast formation from human MSC [38, , –41]. Dexamethasone is also used as differentiation factor [42, 43] and potentiates mineralization .
Different mouse ES cells line are used to generate osteoblasts: D3 [45, 46]; CGR8 ; E14TG2a and its subclones, such as BK4 and HM1 ; EFC1 ; and R1 (current studies) . The scheme shown in Figure 1 represents the protocol used in our laboratory to generate osteoblasts from R1 mouse ES cells. This protocol is the most commonly used one, and it is based on the “hanging drop” culture method. This culture system is composed of three steps. The first step is the hanging drop culture: R1 ES cells are seeded onto the inner side of the lid of a Petri dish and condense by gravitational force for 2 days. The cell aggregates, called EBs (Fig. 2A), are then transferred for 3 days into a new Petri dish, where they grow in suspension (second step). The third and last step consists of an adherence culture: the EBs are transferred into gelatin-coated tissue culture plate with or without differentiation factors (50 μg/ml ascorbic acid, 10 mM β-glycerophosphate, 5 × 10−8 M 1α,25-OH vitamin D3, and 10−8 M dexamethasone). The complete program of differentiation usually takes place in 25–30 days. At the end of the differentiation time, mineralization is observed by Alizarin red staining, as shown in Figure 2B. In addition, analysis of the expression of osteoblastic markers (as described below) demonstrates the successful differentiation of ES cells into active osteoblasts. A second protocol to generate EBs had been established by Bronson et al. just by seeding ES cells directly into a Petri dish, without hanging drop . In some studies, authors describe the necessity of culturing EBs with retinoic acid during the 3 days of growth [43, 47]. Retinoic acid is also known to influence cells to undergo to neurectoderm cells  and adipogenesis  but inhibits cardiogenesis, indicating that retinoic acid alters mesoderm formation in EBs. Thus, the role of retinoic acid appears to be crucial during the first stages of the culture, after EB formation but before differentiation.
zur Nieden et al. studied the expression pattern of osteoblastic markers and their apparition during the differentiation of ES cells . Three periods can be defined: (a) a proliferative phase, followed by (b) the period of matrix deposition and (c) the mineralization phase. They showed that type I collagen mRNA is mainly expressed at the end of the proliferation and during the matrix deposition phase. The same patterns of expression were observed for osteonectin and ALP mRNA. At the end of the matrix deposition phase and the beginning of the mineralization phase, osteopontin mRNA is expressed. BSP, followed by Cbfa1/Runx2, is expressed during the mineralization phase, corresponding to the presence of mature osteoblasts. Finally, osteocalcin mRNA is expressed at very high level and so is designed as the essential marker of the mineralization state. Similar patterns of kinetics of expression are obtained with the other models. For example, after a proliferation phase (from day 0 to day 4), MC3T3 cells enter a phase of bone matrix formation from day 10 to day 16, with ALP as the first gene expressed, followed by type 1 collagen and osteonectin expression. Thereafter, the mineralization phase starts, with a maximum expression of osteocalcin mRNA at day 28 [21, 51].
The effects of ascorbic acid, 1α,25-OH vitamin D3, β-glycerophosphate, and dexamethasone have been studied in a temporal way. The addition of ascorbic acid, 1α,25-OH vitamin D3, and β-glycerophosphate only during the formation of EBs does not induce the osteoblastic markers, and the best result is observed when these factors are added together after EB formation . The addition of others cofactors, such as compactin or bone morphogenetic protein 2 (BMP-2), results in an increase of mineralization and a stimulatory effect on alkaline phosphatase and osteocalcin mRNA expression with BMP-2 . BMP-2 expression in ES cells is consistent with the data observed on rat osteoblast-like cell lines (ROB-C20 and ROB-C26), in which BMP-2 stimulates differentiation into osteocalcin-producing osteoblasts . However, the pro-osteogenic effect of BMP-2 is controversial; indeed, the real effect of BPM-2 seems to be in favor of chondrocyte differentiation and is crucial at earlier stages of the differentiation (during EB generation) [46, 53]. Other factors shown as strong activators in the other models of osteoblast differentiation, such as fibroblast growth factor-8 (FGF-8) in MSC model , BMP-6 and phenytoin in the calvaria model [24, 26], and adiponectin in MC3T3 model , could also be tested during the differentiation of ES cells. Furthermore, it will be interesting to know whether these factors could potentiate osteoblast differentiation as soon as the beginning of the process from the ES cells or whether they act in the later states of the differentiation, as with the pre-engaged models.
Generation of osteoblast has also been studied in human ES (hES) cells [55, , , –59] and monkey ES cells . Generation of EBs was obtained by aggregation of the ES cells without the hanging drop method. However, in a recent study, Karp et al. omitted the EB step and seeded ES cells directly to the dish . Results obtained with hES cells are the same those for mouse ES cells; indeed, hES cells differentiate into osteoblast upon action of the same osteogenic factors as mouse ES cells. Mineralization and induction of osteoblastic markers were observed. According to Karp et al., without the EB step, the number of bone modules is higher; they were formed in 10 or 12 days instead of the 4 weeks that formation required with EB generation, and osteogenic markers (ALP and osteocalcin) appeared quickly . Ahn et al. cocultured hES cells with primary bone-derived cells and showed that primary bone-derived cells are able to induce differentiation of hES cells into osteoblasts, without the addition of exogenous factors . Indeed, primary bone-derived cells secrete BMP-2 and BMP-4, which are efficient to induce the ES cells differentiation, probably in combination with other pro-osteoblastic factors also secreted by the primary-bone cells but not yet characterized. Furthermore, cell-to-cell contacts with primary bone-derived cells may help the differentiation.
Osteoblasts can also be produced from ES cell-generated MSC. Indeed, two recent studies provided a model to generate MSC from human ES cells [41, 59]. By adding this “MSC step” to the differentiation process, these protocols permit large amounts of pure MSC to be obtained and consequently increase the number of osteoblasts obtained from these MSC and suitable for therapeutic applications. Moreover, these ES cell-derived MSC have the capacity to support the growth of undifferentiated ES cells, which is a real advance in stem cell research because it provides a source of autologous feeder for the culture of ES cells, which considerably increases the possibility of creating secure ES cells banks for therapeutic use.
Differentiation of mouse or human ES cells into osteoblasts is effective. Different approaches are used to achieve osteogenic differentiation: with or without EB generation and with various osteogenic molecules. Further work is needed to elucidate the potential role of other cytokines that could enhance osteogenic differentiation.
From ES Cells to Osteoclasts
Osteoclasts are bone resorbing cells that derive from the hematopoietic mesodermal lineage. Different models, presented in Table 2, have been developed to study osteoclast differentiation: bone marrow/spleen cells with or without osteoblastic/stromal cells [61, , –64], calvaria [65, 66], CD14+ fraction of peripheral blood mononuclear cells [67, , , –71], RAW 264.7 cell line , and so on. The main markers used to determine the presence of osteoclasts are tartrate-resistant alkaline phosphatase (TRAP), calcitonin receptor, vitronectin receptor, cathepsin K, and the capacity of the cells to resorb mineralized matrix. The two main factors in osteoclastogenesis are macrophage-colony-stimulating factor (M-CSF) and RANKL, which are both involved in the expansion of osteoclastic precursors and their maturation into osteoclasts [73, 74]. In the model of RAW 264.7, RANKL alone is able to induce differentiation into TRAP-positive osteoclast in 5 days . In the coculture system of bone marrow cells with osteoblast or stromal cells, only dexamethasone  or 1α,25-OH vitamin D3 [28, 64, 75, 76] is used. However, there are strong possibilities that other factors, such as interleukin (IL)-6, prostaglandin E2, or RANKL, are secreted by the stromal cells or the osteoblasts upon the action of dexamethasone or 1α,25-OH vitamin D3 and so are involved in the osteoclastogenesis [9, 11, 12].
Table Table 2.. Main in vitro models developed to study osteoclastic differentiation
A few studies report the generation of osteoclasts from ES cells. The mouse ES cells used are D3 [77, , , –81], J1 [80, 81], CCE , and R1 (current studies). No formation of EBs is necessary, and ES cells are directly seeded on a 24-well plate  or on a monolayer of bone marrow-derived stromal cells (ST2) or newborn calvaria-derived stromal cells (OP9) [77, , –80]. ST2 cells are M-CSF-producing cells and so promote osteoclastogenesis. OP9 cells, which are M-CSF-deficient, have been shown to be an excellent system to generate lymphohematopoietic cells . Yamane et al. compared four different culture methods: (a) culture of ES cells on an ST2 monolayer for 11 days, (b) culture of ES cells on OP9 for 5 days followed by 6 days on a monolayer of ST2 or OP9, (c) culture of ES cells on OP9 for 14 days, and (d) culture of ES cells on OP9 for 5 days followed by 5 days on OP9 and 6 days on ST2  (Fig. 3). During the initial phase (first 5 days of culture), a strong induction of hematopoietic precursors takes place, including osteoclast precursors, followed by an expansion of these precursors and then their maturation into osteoclasts. Multistep culture provides a good tool to study the effect of osteoclastic factors and their different targets (hematopoietic precursors, osteoclast precursors, and mature osteoclasts). In all conditions, osteoclasts were generated in the presence of dexamethasone and 1α,25-OH vitamin D3, and they were visualized by TRAP staining and characterized by reverse transcription-polymerase chain reaction (markers described above). Figure 4B shows the TRAP staining at the periphery of a colony formed from one ES cell, observed after 11 days of culture, with R1 cells cultured on ST2 monolayer, in the presence of dexamethasone and 1α,25-OH vitamin D3 (no dexamethasone or 1α,25-OH vitamin D3 was added in Fig. 4A). This result, obtained in our laboratory, is in agreement with the work of Hemmi et al. . Furthermore, Yamane et al. studied the capacity of resorption on dentin slices, demonstrating the functionality of these cells . Apparition of specific markers has been shown to be essential in the differentiation process (Fig. 5). Indeed, Flk-1 (fetal liver kinase-1) and SLC/tal-1 are crucial to generate hematopoietic cells [80, 81]. GATA-2 is important for the generation of osteoclastic precursors . M-CSF may be added to the initial phase of the culture to stimulate the expansion of hematopoietic precursors, but adding M-CSF later in the culture inhibits the formation of mature osteoclasts . Adding ascorbic acid to the culture medium also promotes osteoclastogenesis by increasing OC precursors . As for osteoblasts, osteoclasts can be generated efficiently from ES cells. The multistep culture is a good model by which to investigate the mechanisms and the chronological apparition of specific markers, as each different step of the culture corresponds to a different stage of the differentiation.
All these studies help us better understand the molecular mechanisms of osteoclastogenesis and so elucidate deregulated events happening during a tumoral process. Furthermore, differentiation of ES cells into osteoclasts is promising for therapy. For example, transplantation of ES cell-generated osteoclasts could be envisaged to cure some pathologies, such as osteoclast-poor osteopetrosis, which is a rare genetic disorder characterized by severely reduced bone resorption due to a defect in osteoclast development. Presently, the treatment for this pathology, consisting of engraftment of hematopoietic stem cells, fails .
The treatment of osteoporosis, as well as other osteolytic bone diseases, is mainly based on bisphosphonate drugs, which are inhibitors of osteoclast-mediated bone resorption . Alternative therapies for bone defects are bone transplantation using biomaterials such as alginate hydrogels [85, –87] or calcium phosphate ceramics or cements [88, 89], which can be carriers of stem cells, mesenchymal or embryonic. Both kinds of cells can be induced to differentiate into mesenchymal tissues, such as bone, fat, cartilage, and so on , using different factors. As shown in Table 1, dexamethasone, ascorbic acid, and 1α,25-OH vitamin D3 are the components of the main cocktail used to differentiate both ES cells and mesenchymal stem cells into osteoblasts. Furthermore, BMPs are molecules often used to control osteogenic differentiation. Interestingly, BMPs contribute to the maintenance of self-renewal of mouse ES cells in presence of leukemia inhibitory factor (LIF), whereas in the absence of LIF, BMPs induce mesoderm differentiation . Indeed, BMPs, mainly BMP-2, activate Smad proteins (Smad 1 and 5)  which become phosphorylated and in turn activate the transcription factor Dlx5, followed by Cbfa1/Runx2 and Osterix, which are also transcription factors (reviewed by Ryoo et al. ). Cbfa1/Runx2 has been shown to preferentially initiate two steps of the differentiation process, stem cells into preosteoblasts and preosteoblasts into osteoblasts, whereas Osterix acts only during the preosteoblast/osteoblast step. BMPs are used by these two cellular targets for ES cell [43, 47, 60] or mesenchymal cell differentiations  into osteoblasts, first by inducing mesodermal differentiation and then by inducing the osteoblast lineage specifically.
The differentiation of osteoclasts from ES cells is also effective. It is interesting to note that generation of osteoblasts and osteoclasts could be realized with the same differentiating factors (dexamethasone and 1α,25-OH vitamin D3, with or without ascorbic acid). The main difference between the two differentiations is the generation of EBs for osteoblasts (except in one study), which is not used to generate osteoclasts. For osteoclastogenesis, one possibility to explain this difference is the production of RANKL by ST2 cells upon dexamethasone stimulation and 1α,25-OH vitamin D3 stimulation . Indeed, ST2 cells treated with 1α,25-OH vitamin D3 supported osteoclastogenesis . Furthermore, some cell-to-cell contacts between ES cells and ST2/OP9 cells could also be crucial for the differentiation into osteoclasts. In conclusion, ES cells are the best tool to understand molecular mechanisms involved in osteoblast and osteoclast differentiations, and particularly during the earliest stages of the differentiation. Indeed, other models used to generate bone cells are realized with pre-engaged cells and so omit the earliest phases of the differentiation. ES cells are also a good model to understand the mechanisms involved during mesenchymal cell differentiation that then give rise to osteoblasts or other cell types, such as adipocytes. As osteogenic factors or adipogenic factors are added since the beginning of the culture of ES cells, it could be envisaged that different populations of mesenchymal cells can emerge from ES cells. Furthermore, ES cells are also the best model to uncover the mechanisms involved in mesenchymal cell differentiation or other kind of cells, such as cells of the hematopoietic lineage.
Even if some studies should be performed to decipher all the signals that influence osteoblast differentiation and to control all the necessary steps for a successful transplantation, ES cells, as well as MSC, are promising in regenerative medicine. ES cells also offer the possibility of finding new key genes involved in the differentiation program and consequently those that can be deregulated during a tumoral process. In this way, ES cells represent a positive approach to find new target genes for bone cancer treatment and other bone disease therapies.
The authors indicate no potential conflicts of interest.
L.D. is supported by an Association pour la Recherche sur le Cancer postdoctoral fellowship.