This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Yamaguchi, A. Articles by Suda, T. Search for Related Content PubMed PubMed Citation Articles by Yamaguchi, A. Articles by Suda, T.

Regulation of Osteoblast Differentiation Mediated by Bone Morphogenetic Proteins, Hedgehogs, and Cbfa1

Department of Oral Pathology (A.Y.), Nagasaki University School of Dentistry, 1–7-1 Sakamoto, Nagasaki 852, Department of Molecular Medicine (T.K.), School of Medicine, Osaka University, 2–2 Yamada-oka, Suita, Osaka 565 and "Form and Function", PRESTO, Japan Science and Technology Corporation, and Department of Biochemistry (T.S.), School of Dentistry, Showa University, 1–5-8 Hatanodai, Shinagawa-ku, Tokyo 142, Japan

	Abstract

Top Abstract I. Introduction II. Origin of Osteoblasts III. Regulation of Osteoblast... IV. Transcription Factors That... V. Summary References

 
Osteoblasts arise from common progenitors with chondrocytes, muscle and adipocytes, and various hormones and local factors regulate their differentiation. We review here regulation of osteoblast differentiation mediated by the local factors such as bone morphogenetic proteins (BMPs) and hedgehogs and the transcription factor, core-binding factor -1 (Cbfa1). BMPs are the most potent regulators of osteoblast differentiation among the local factors. Sonic and Indian hedgehogs are involved in osteoblast differentiation by interacting with BMPs. Cbfa1, a member of the runt domain gene family, plays a major role in the processes of a determination of osteoblast cell lineage and maturation of osteoblasts. Cbfa1 is an essential transcription factor for osteoblast differentiation and bone formation, because Cbfa1-deficient mice completely lacked bone formation due to maturation arrest of osteoblasts. Although the regulatory mechanism of Cbfa1 expression has not been fully clarified, BMPs are an important local factor that up-regulates Cbfa1 expression. Thus, the intimate interaction between local factors such as BMPs and hedgehogs and the transcription factor, Cbfa1, is important to osteoblast differentiation and bone formation.

I. Introduction

II. Origin of Osteoblasts

III. Regulation of Osteoblast Differentiation by Bone Morphogenetic Proteins (BMPs) and Hedgehogs

A. BMPs

B. Sonic and Indian hedgehogs

IV. Transcription Factors That Regulate Osteoblast Differentiation and Bone Formation

A. Transcription factors involved in osteoblast differentiation

B. Cbfa1 is an important transcription factor regulating osteoblast differentiation

C. Absence of ossification in Cbfa1-deficient mice

D. Role of Cbfa family transcription factors in osteoblast differentiation

E. Cbfa1 is involved in chondrocyte maturation

F. Cbfa1 is involved in osteoclastogenesis

G. Heterozygous mutations of Cbfa1 locus cause cleidocranial dysplasia

V. Summary

	I. Introduction

Top Abstract I. Introduction II. Origin of Osteoblasts III. Regulation of Osteoblast... IV. Transcription Factors That... V. Summary References

 
THE skeletal tissue is composed of various types of mesenchymal cells such as osteoblasts, chondrocytes, myoblasts and bone marrow stromal cells including adipocytes. These cell lineages are believed to originate from common mesenchymal progenitors (1, 2, 3, 4, 5, 6, 7) called pluripotent mesenchymal stem cells (5, 6, 7). These progenitors acquire specific phenotypes depending on their maturation during differentiation. Osteoblasts express various phenotypic markers such as high alkaline phosphatase (ALP) activity and synthesize collagenous and noncollagenous bone matrix proteins including osteocalcin (4). The most important function of osteoblasts is to form mineralized bones. Osteoblasts express receptors for various hormones including PTH (8, 9), 1,25-dihydroxyvitamin D₃ [1,25(OH)₂D₃] (10), estrogen (11, 12), and glucocorticoids (13, 14), which are involved in the regulation of osteoblast differentiation. Osteoblast differentiation is also regulated by various local factors in a paracrine and/or an autocrine fashion (4). To investigate the roles of these hormones and local factors in osteoblast differentiation, various osteoblastic cell lines have been successfully established (15, 16, 17). It is also possible to use multipotent mesenchymal progenitors to examine the differentiation process of osteoblasts in vitro (18, 19). Experiments using these in vitro assay systems have yielded a great deal of information concerning local factors that regulate osteoblast differentiation. Indeed, several research groups, including ourselves, using various culture systems have demonstrated that bone morphogenetic proteins (BMPs) are potent local factors that regulate osteoblast differentiation (18, 19, 20, 21, 22).

The hedgehog signaling pathway mediates inductive events during development in invertebrates and vertebrates (23). In higher vertebrates, the Hedgehog gene family consists of at least three members, Sonic, Indian, and Desert hedgehog (Shh, Ihh, and Dhh, respectively) (23, 24). Among these, Shh and Ihh are involved in the skeletal formation during development (25, 26, 27) and in skeletal repair (27, 28, 29). In Drosophila, hedgehog signaling induces expression of decapentaplegic (dpp), which is a homolog of vertebrate BMP, in adjacent cells, in which dpp acts as a secondary signaling molecule to control the fate of these cells (23). Similar interactions between hedgehogs and BMPs were also demonstrated in several organs during vertebrate development (30). Thus, the hedgehog-BMP interaction is highly conserved in the patterning process of various organs including skeletons in higher vertebrates.

Transcription factors that determine the differentiation pathways of specific cell types have been identified in several cell lineages. In the case of skeletal muscles, the muscle-specific transcription factors of the MyoD family, which belong to the basic helix-loop-helix (HLH) family, are necessary for determining the pathway of differentiation into the muscle lineage and are required for the differentiation of committed myoblasts to fully differentiated myotubes (31). In addition, peroxisome proliferator-activated receptor 2 (PPAR2) has been reported to play an important role in determining the differentiation pathway of adipocyte lineage cells (32). The specific transcription factors that determine osteoblast differentiation remained unclear. Recently, several research groups independently reported that Cbfa1 [Core binding factor 1, also called Pebp2A (Polyomavirus enhancer binding protein 2 A), AML3 (acute myelocytic leukemia 3) and OSF2 (osteoblast specific factor 2)], which belongs to the runt-domain gene family, is an important transcription factor for osteoblast differentiation and bone formation (33, 34, 35, 36). Since it has been reported that BMP up-regulates expression of Cbfa1 during osteoblast differentiation (33), Cbfa1 seems to be a downstream factor controlled by BMP. In contrast to MyoD and PPAR2, Cbfa1 is necessary but not sufficient to support differentiation to the mature osteoblast phenotype (37).

Although osteoblast differentiation is regulated by many factors, the three molecules described above, BMP, hedgehogs, and Cbfa1, play important roles in the differentiation process with intimate interaction between them. In this article we review recent advances regarding the regulation of osteoblast differentiation mediated by BMP, hedgehogs, and Cbfa1.

	II. Origin of Osteoblasts

Top Abstract I. Introduction II. Origin of Osteoblasts III. Regulation of Osteoblast... IV. Transcription Factors That... V. Summary References

 
Several lines of evidence from classical embryology have established that two different embryonic lineages, neural crest and mesoderm, form the early skeleton (4, 38). The branchial arch derivatives of the craniofacial skeleton originate from neural crest, whereas the axial skeleton, ribs, appendicular skeletons, and the skull base arise from mesoderm. Among the skeletal tissues formed by the mesoderm, the axial skeleton originates from the sclerotome, and the appendicular skeleton arises from the lateral plate mesoderm.

During skeletogenesis, bone is formed in two different manners, intramembranous ossification and endochondral ossification, regardless of the embryonic lineage. In the case of intramembranous ossification, osteogenesis occurs directly in the condensed mesenchymal cells. Ossification generated in this fashion is responsible for forming the flat bones of the skull, part of the clavicle, and the additional bone on the periosteal surface of long bones. In the process of endochondral ossification, mesenchymal cells first condense to form a cartilage model, and then bone formation occurs replacing this cartilage. This type of ossification forms most of the bones including the axial and appendicular skeletons.

Several in vivo and in vitro experiments have demonstrated the presence of osteoprogenitors in both bones and extraskeletal tissues in the postnatal state. In bone tissues, osteoprogenitors are present in bone marrow and the periosteum. Friedenstein and colleagues (39, 40, 41, 42) have proved that osteoprogenitors are present in bone marrow. They showed that bone marrow cells harvested from confluent in vitro cultures of marrow cells retained the ability to form osteogenic tissues when cultured in vivo within diffusion chambers. Then they demonstrated by various in vivo and in vitro experiments that the single cell-derived fibroblastic colonies, termed CFU-F (colony forming units-fibroblastic) (41), retained osteogenic potential (42). Other groups also demonstrated that bone marrow cells, including that harvested from human marrow, contained mesenchymal progenitors, which differentiated into osteogenic, chondrogenic, and adipogenic lineage cells (4, 5, 6, 7, 43). Further characterization of human mesenchymal stem cells is important to develop new therapeutic drugs for bone diseases such as osteoporosis. The osteogenic potential of the periosteum was also shown by several experiments. In vivo experiments using [³H]-thymidine as a tracer demonstrated that the cells located in the outer layer of the periosteum differentiated into mature osteoblasts and osteocytes (44). Periosteum or periosteum- derived cells generate bone nodules in in vitro cultures (45). These osteoprogenitors in the periosteum contribute to formation of bone callus during fracture repair.

Transplantation of BMPs into muscle or subcutaneous sites induces ectopic bone formation (46, 47), indicating that osteoprogenitors, which respond to BMPs, are also present at extraskeletal sites. These osteoprogenitors may have BMP receptors, but other characteristics of these cells have not been analyzed in detail. Further characterization of these cells is important to develop effective cell therapy for bone repair by transplantation of such extraskeletal osteoprogenitors after appropriate in vitro culture (48).

	III. Regulation of Osteoblast Differentiation by BMPs and Hedgehogs

Top Abstract I. Introduction II. Origin of Osteoblasts III. Regulation of Osteoblast... IV. Transcription Factors That... V. Summary References

 
Osteoblast differentiation and bone formation are regulated by many local factors. Among these, BMPs are one of the most potent factors. Recent studies revealed that specific signaling systems through receptors and the specific inhibitors such as noggin and chordin regulate BMP function at the extracellular level. In addition, several lines of evidence indicate that hedgehogs modulate BMP function during pattern formation including skeletal formation in vertebrates.

To investigate the roles of local factors involved in osteoblast differentiation and bone formation, it is important to establish in vitro culture systems that reflect the different stages of maturation during osteogenesis. Although many cell lines are available for such investigations, we have used several cell lines that are useful for investigating osteoblast differentiation in vitro (Table 1). Using these cell lines, we can explore the roles of BMPs and hedgehogs that regulate osteoblast differentiation and bone formation. In this section, we describe details of the roles of two important local factors, BMPs and hedgehogs, in osteoblast differentiation.

Since various members of the BMP family are expressed during skeletogenesis, localization of such BMPs provides important information to understand the role of each BMP in skeletal development. For example, BMP-5 mRNA is localized to mesenchymal condensations before cartilage development (57), whereas mRNAs for BMP-2, BMP-4, and BMP-7 (OP-1:osteogenic protein 1) are present in the mesenchyme surrounding cartilaginous anlage (53, 58). The expression of mRNAs for these BMPs continue to be present at perichondrium and periosteum at later stages (53, 57, 58). Growth/differentiation factor5 (GDF5), a member of the BMP family, is weakly expressed at perichondrium, whereas its expression is strong at the interface between cartilage anlages where joints will later form (59). This expression pattern correlates closely with joint patterning defects in GDF5 mutant (brachypodism) mice (59, 60). Skeletal analysis in null mutation mice by targeted disruption of BMP genes will enable us to understand the role of each BMP in skeletogenesis. Although studies using such mutant mice revealed important functions of BMPs in mesodermal induction (61, 62, 63) and organogenesis (63, 64, 65, 66), they failed to provide much information on the role of each BMP in skeletogenesis. BMP-2 and BMP-4 knockout mice die during early gastrulation due to failure of mesoderm induction (61, 67, 68). BMP-7 null mutation mice die shortly after birth due to severe renal failure and eye defects, and they exhibited mild skeletal changes such as polydactyly and occasional abnormalities of ribs (65). Normal skeletons formed in these mice might be rescued by the redundant function of other BMPs, which were expressed cooperatively with BMP-7. The generation of conditional knockout mice for each BMP will provide more important information concerning the roles of each BMP in skeletogenesis.

1. BMP receptors and signal transduction systems. Similarly to TGF-ß, BMPs bind to two types of serine-threonine receptors, termed BMP type I and type II receptors (69, 70). Both types of receptors are necessary to transduce BMP signals. BMP type I receptors (BMPR-I) also bind BMPs directly in the absence of BMP type II receptor (BMPR-II), whereas the TGF-ß type I receptor does not bind ligands in the absence of the TGF-ß-type II receptor (71, 72, 73). Two kinds of BMPR-Is, BMPR-IA and BMPR-IB, have been cloned in mammals (74, 75, 76). During embryogenesis, BMPR-IA is more widely expressed than BMPR-IB in various tissues (77, 78). BMPR-IA is also expressed in various types of cultured cells including MC3T3-E1 cells, C2C12 cells (79), C3H10T1/2 cells, and primary osteoblasts isolated from newborn rat calvariae (78), but BMPR-IB is highly expressed in a limited number of osteoblastic cells such as ROB-C26 (55) and primary osteoblasts isolated from calvariae. More importantly, immunohistochemical and in situ hybridization analyses demonstrated that osteoblasts express BMPs and their receptors in the process of bone formation during skeletal development and fracture repair (53, 57, 58, 59, 77, 78, 80, 81, 82, 83, 84, 85), suggesting that BMPs are involved in the differentiation process of osteoblasts from osteoprogenitors to mature osteoblasts.

Genetically engineered mutant forms of BMPRs are useful for exploring the functions of these proteins. A soluble form of BMPR-IA lacking the transmembrane and cytoplasmic domains can bind ligands and antagonize the action of BMP (85). Truncated or kinase-inactivated forms of BMPR-Is are also capable of blocking the activity of BMPs (76, 79, 86). In contrast, constitutively active forms of BMPRs can induce the action of BMP in the absence of ligand (87). These mutant BMPRs have been successfully used to investigate the signal transduction pathways during osteoblast differentiation.

Signal-transducing molecules of the TGF-ß superfamily, termed Smads, have been identified (69, 70, 88) (Figs. 1 and 2). At present, eight mammalian Smad proteins, Smad1 through Smad8, have been isolated (69, 70, 89). These are classified into three subgroups according to their structures and functions (69, 70). The first subgroup is pathway- restricted Smad (R-Smad). Smads belonging to this subgroup are ligand specific and activated by the binding of ligands to type I receptors. Among these, Smad1, Smad5, and Smad8 are involved in BMP signaling (89, 90, 91, 92, 93, 94, 95), and Smad2 and Smad3 mediate TGF-ß/activin signaling (96, 97). The second subgroup of Smads is the common mediator Smads (C-Smads). Smad4 (also termed DPC-4) belongs to this subgroup (98, 99). R-Smads are phosphorylated by the serine/threonine kinase receptors that interact with C-Smads, forming a heterodigomeric complex. This complex is translocated into the nucleus and regulates the transcription of target genes through direct binding to DNA as well as association with other DNA-binding proteins (70). The third subgroup of Smads is the inhibitory Smads (I-Smads). Smad6 and Smad7 comprise this subgroup (100, 101, 102, 103). These Smads inhibit ligand activity by stably binding to type I receptors. Smad6 binds to the TGF-ß type I receptor, activin type IB receptor, and BMPR-IB (101), while Smad7 binds to TGF-ß type I receptor (100, 102). I-Smads compete with R-Smads for binding to type I receptors. Smad6 also competes for binding of activated Smad1 with Smad4 (103).

View larger version (14K): [in this window] [in a new window]   Figure 1. The mammalian Smad family is comprised of three classes of protein, R-Smad (pathway-restricted Smad), C-Smad (common mediator Smad), and I-Smad (inhibitory Smad). [Adapted from M. Kawabata et al.: Cytokine Growth Factor Rev 9:49–61, 1998 (70 ).]

View larger version (24K): [in this window] [in a new window]   Figure 2. Possible mechanism of BMP signaling through BMP receptors and Smads. BMPs bind to BMP receptors (BMPR-I and BMPR-II) in the target cells, and Smad1 and Smad5 transduce BMP signals interacting with Smad4. Smad6 inhibits BMP activity by binding to BMPR-I. Smads1/4/5 act positively, and Smad6 acts negatively in BMP signaling. [Adapted from M. Kawabata et al.: Cytokine Growth Factor Rev9:49–61, 1998 (70 ).]

b. BMP and differentiation of osteoblast precursor cells.
ROB-C26 is a committed osteoprogenitor cell line, retaining the differentiation potential to form myotubes and adipocytes (3). The developmental potential of this cell line is similar to that of RCJ 3.1, which is one of the osteoblastic cell lines isolated from fetal rat calvariae by Aubin et al. (108) and characterized by Grigoriadis et al. (2). RCJ 3.1 cells are capable of differentiating into chondrocytes in addition to osteoblasts, adipocytes, and myotubes, while ROB-C26 cells lack the potential to differentiate into chondrocytes. Kellermann and colleagues (109, 110) established a mesodermal tripotential progenitor cell line (C1) from mouse teratocarcinoma, which differentiated into three types of cells including osteoblasts, chondroblasts, and adipocytes. These cell lines are also useful for studying the regulatory mechanism of osteoblast differentiation from mesenchymal progenitors. Among these, ROB-C26 cells have been frequently used to investigate the effects of BMPs on osteoblast differentiation. BMP-2 stimulated ALP activity and PTH-dependent cAMP production and induced osteocalcin synthesis in ROB-C26 cells (19). Gitelman et al. (22) reported that overexpression of BMP-6 accelerated osteoblast differentiation in ROB-C26 cells, and this effect was antagonized by the addition of a neutralizing antibody against BMP-6. Nishitoh et al. (55) demonstrated that GDF-5 stimulated ALP activity in ROB-C26, which was mediated by BMPR-IB and BMPR-II. BMP-7 also bound predominantly to BMPR-IB in ROB-C26 cells, and Smad5 was a key component in the intercellular signaling of BMP-7 (111). These results indicated that BMPs are important regulators of osteoblast differentiation from multipotent mesenchymal cells.

There are several osteoblast precursor cell lines, the differentiation potential of which are restricted to the osteoblast lineage. Among these, MC3T3-E1, which is a clonal osteoblastic cell line isolated from calvariae of a late stage mouse embryo (17), is most frequently used to study osteoblast differentiation. This cell line expresses various osteoblast functions including formation of mineralized bone nodules in long-term culture. In MC3T3-E1 cells, BMP-2 and BMP-7 increased ALP activity, PTH responsiveness, and osteocalcin production (112, 113), suggesting that BMPs promote differentiation of osteoblast precursors to more mature osteoblasts. BMP-7 stimulated osteoblast differentiation in ROS17/2.8 cells, a typical osteoblastic cell line isolated from rat osteosarcoma, by increasing synthesis of collagen and osteocalcin, ALP activity, and PTH responsiveness (114). BMP-12, alternatively called GDF7, increased ALP activity within 24 h of treatment in ROS17/2.8 cells (115), whereas it failed to increase ALP activity in this cell line after treatment for 6 days (116). The effects of BMP-12 on osteoblast differentiation should be more extensively studied using other osteoblastic cell lines.

Osteoblastic cells isolated from the calvariae of newborn rats (117) or the bone marrow of adult rats (primary osteoblasts) (118) provide a suitable model in which to explore the bone formation process in vitro, because these cells generate numerous mineralized bone nodules when cultured in the presence of ß-glycerophosphate and ascorbic acid. Since only a limited number of clonal cell lines retain the capacity to form mineralized bones in vitro (17), primary osteoblasts are important tools for analyzing the differentiation process of osteoblasts from osteoprogenitors to bone-forming osteoblasts. To explore the roles of BMPs in formation of bone nodules, we investigated the distributions of BMPs and their receptors in osteoblastic cells isolated from newborn rat calvariae (119). In situ hybridization studies detected strong signals for BMP-2 and BMP-4 mRNAs in bone nodule-forming cells, but not in the cells located in internodular regions. In addition, immunohistochemical analysis using an antibody reactive with both BMP-2 and BMP-4 demonstrated that positive cells first appeared in unmineralized nodules and were then localized preferentially in mineralized nodules at a later stage in culture. BMP receptors such as BMPR-IA, BMPR-IB, and BMPR-II were preferentially expressed at the sites of nodule formation in calvarial culture (119). Harris et al. (120) demonstrated by Northern blotting analysis that not only BMP-2 and BMP-4 but also BMP-6 mRNAs were expressed during bone nodule formation by osteoblasts isolated from fetal rat calvariae. The maximal levels of expression of each BMP mRNA coincided with the formation of mineralized bone nodules. These results suggested that several BMPs are involved in the mechanism of bone nodule formation by osteoblasts in vitro. Hughes et al. (121) compared the effects of BMP-2, BMP-4, and BMP-6 on the formation of bone nodules by rat calvaria-derived osteoblastic cells. BMP-2 was less potent than BMP-4 and BMP-6 in this assay system. Boden et al. (122) reported that glucocorticoid-induced formation of bone nodules in fetal rat calvarial osteoblasts was mediated by BMP-6. Glucocorticoids preferentially increased expression of BMP-6 mRNA, and the antisense oligonucleotide corresponding to BMP-6 strongly inhibited formation of bone nodules. BMP-7 also increased formation of bone nodules by rat calvarial osteoblasts (123, 124). The effects of BMPs on osteoblast differentiation were also investigated using human osteoblastic cells. Lecanda et al. (125) reported that BMP-2 had profound effects on proliferation, expression of most of the bone matrix proteins, and the mineralization of human osteoblastic cells. We also demonstrated that BMP-2 stimulated ALP activity and PTH- dependent cAMP production in primary osteoblastic cells isolated from human bones (126). Taken together, these observations indicated that various BMPs play important roles in the process of osteoblast differentiation in a paracrine and/or autocrine fashion.

Several experiments concerning the regulation of BMP activity have been reported. IL-1ß synergistically increased BMP-2-induced ALP activity in MC3T3-E1 cells, but tumor necrosis factor- (TNF) inhibited BMP-2-induced ALP activity in this cell line (113). Insulin-like growth factor I (IGF-I) synergistically enhanced BMP-7-induced osteoblast differentiation in primary culture of fetal rat calvaria (124). These results suggest that the action of BMP is modulated by various local factors. Rickard et al. (127) investigated the effects of estrogen on BMP production using two estrogen-responsive human immortalized osteoblastic cell lines (hFOB/ER3 and hFOB/ER9). Interestingly, estrogen (17ß-estradiol: 10^-10 to 10^-7 M) increased the expression level of BMP-6 mRNA and production of BMP-6 protein, while levels of mRNAs encoding TGF-ß1, TGF-ß2, and BMPs-1 through -5 and -7 were unchanged (127). They suggested that some of the skeletal effects of estrogen on bone might be mediated by increased production of BMP-6 by osteoblasts. However, further experiments are needed to confirm such a role for estrogen, because estrogen suppresses the rate of bone remodeling in vivo. Recently, Mundy et al. (128) searched 30,000 small molecule compounds that activated the promoter of BMP-2 and found that the statins, lovastatin and simvastatin, drugs used for lowering serum cholesterol, had such activity. In addition, they demonstrated by an organ culture system and in vivo subcutaneous injection that statins stimulated new bone formation associated with an increased expression level of BMP-2 mRNA. This suggests a therapeutic application of statins for osteoporosis.

Thus, various BMPs promote osteoprogenitors to differentiate into more mature osteoblasts. However, it has not been established which BMP is the most potent in osteoblast differentiation, because these studies were conducted using different cell types and different culture conditions. More extensive in vitro studies using a standardized culture system are necessary to evaluate the potential of each BMP. It is also important to investigate further the regulation of BMP activity by local and systemic factors.

c. BMP and differentiation of bone marrow stromal cells.
The osteogenic potential of bone marrow stromal cells has been demonstrated by studies using in vivo and in vitro culture systems. Bone marrow-derived clonal cell lines and freshly isolated bone marrow stromal cells have often been used in such studies. Various bone marrow-derived cell lines show the characteristics of preadipocytes. More importantly, several cell lines retain the capacity to support hematopoiesis including osteoclastogenesis (129).

Thies et al. (130) reported that BMP-2 induced the mouse bone marrow-derived cell line W-20–17 to exhibit osteoblast phenotypic markers using an in vitro culture system. We investigated the effects of BMPs on osteoblast differentiation using two mouse bone marrow stromal cell lines (131), ST2 (132) and MC3T3-G2-PA6 (PA6)(133), because the two cell lines had preadipocytic properties and retained the capacity to support hematopoiesis including osteoclastogenesis (129). Neither ST2 nor PA6 cells exhibited features typical of osteoblast phenotype under control culture conditions. BMP-2, BMP-4, and BMP-6 induced ST2 cells to express osteoblast phenotypic markers such as elevated levels of ALP activity, PTH-dependent production of cAMP, and the synthesis of osteocalcin (131). Ascorbic acid also induced osteoblast differentiation in ST2 cells via the action of BMP (134). In contrast, the stimulatory effects of the BMPs on ALP activity and PTH-dependent production of cAMP were weaker in PA6 cells than in ST2 cells, and BMPs failed to induce the synthesis of osteocalcin in PA6 cells (131). These results indicated that the effects of BMPs on osteoblast differentiation of bone marrow stromal cells differ between different cell lines. It will be interesting to explore differences in BMP receptor-signaling systems in these cell lines. Rickard et al. (127) reported that BMP-2 induced osteoblast differentiation in primary cultures of rat bone marrow cells. In this case, BMP-2 exerted synergistic effects on bone nodule formation with dexamethasone (127).

Adipocytes are an important component of bone marrow stromal cells derived from common progenitors with osteoblasts. As described above, BMP-2 and BMP-4 promoted C3H10T1/2 cells to differentiate into not only osteochondrogenic cells but also adipocytes (104, 105). Chen et al. (135) investigated roles of BMPRs in the process of BMP-induced differentiation of osteoblasts and adipocytes using 2T3 cells, which were derived from the calvariae of transgenic mice expressing T antigen driven by the BMP-2 promoter. BMP-2 induced this cell line to differentiate into mature osteoblasts or adipocytes. Overexpression of a kinase domain-truncated BMPR-IB in 2T3 cells completely inhibited osteoblast differentiation in this cell line, and the decreased level of ALP activity in the 2T3 cells with the truncated BMPR-IB was rescued by transfection with wild-type BMPR-IB. In addition, overexpression of constitutively active BMPR-IB induced formation of bone in 2T3 cells in the absence of BMP-2. In contrast, overexpression of a kinase domain-truncated BMPR-IA blocked adipocyte differentiation, whereas transfection of constitutively active BMPR-IA induced adipocyte differentiation, increasing expression levels of adipocyte differentiation-related genes such as adipsin and PPAR2 in 2T3 cells. These results suggested that BMPR-IA and BMPR-IB have different functions in the differentiation of osteoblasts and adipocytes in 2T3 cells: BMPR-IB is the major receptor involved in osteoblast differentiation and BMPR-IA is the major receptor for adipocyte differentiation. As described below, however, an important role of BMPR-IA has been demonstrated in the process of BMP-2-induced osteoblast differentiation in C2C12 myoblasts (79). In contrast to the stimulatory effects of BMP-2 on adipocyte differentiation, Gimble et al. (136) reported that BMP-2 and BMP-4 inhibited adipocyte differentiation of murine bone marrow stromal cells. Inhibitory effects of BMP-2 on adipocyte differentiation were also demonstrated in the immortalized human bone marrow stromal cell line [hMS (2, 3, 4, 5, 6)] (137). Since reciprocal regulation of osteogenesis and adipogenesis in the bone marrow microenvironment has been suggested (136, 137), further investigation of the regulatory mechanism involved in lineage determination of osteoblasts and adipocytes is important to understand the pathogenesis of osteopenic diseases such as osteoporosis. Lecka-Czernik et al. (138) reported interesting findings in this regard. They demonstrated that overexpression of PPAR2 in mouse bone marrow cells stimulated adipocyte differentiation and inhibited osteoblast differentiation by suppressing expression of Cbfa1 mRNA. This suggests that PPAR2 negatively regulates osteoblast differentiation of bone marrow stromal cells by suppressing Cbfa1 expression.

Thus, BMPs play important roles in the process of cell lineage determination of osteoblasts and adipocytes from bone marrow stromal cells. In this process, BMPs promote osteoblastic differentiation, but they exert diverse effects on adipocyte differentiation depending on cell type. The diverse actions of BMPs on adipocyte differentiation might be caused by different usage of the BMP receptor-signaling systems and the transcription factors relating to cell lineage determination such as Cbfa1 and PPAR2.

d. Role of BMPs in osteogenic transdifferentiation of myogenic cells.
Classical transplantation experiments of BMPs into muscular sites demonstrated that BMPs induced ectopic cartilage and bone formation (46, 47). In addition, tissue culture experiments showed that muscle cultured on decalcified bone generated chondrogenic cells (139). These findings suggest that muscles contain osteochondrogenic progenitor cells, and that BMPs divert the differentiation pathway of myogenic cells into osteochondrogenic lineage cells.

We first investigated the effects of BMP-2 on myogenic differentiation in ROB-C26, which is an osteoblast precursor cell line with the capacity to differentiate into myogenic cells (3). BMP-2 inhibited myogenic differentiation with concomitant stimulation of osteoblast differentiation in this cell line (19).

To further investigate the regulatory mechanism of myogenic differentiation by BMP-2, Katagiri et al. (21) used C2C12 myoblasts, which originated from muscular tissue satellite cells. Both BMP-2 and TGF-ß1 inhibited myotube formation completely in C2C12 cells, but only BMP-2 induced them to differentiate into osteoblast lineage cells (21). BMP-2 exerted effects similar to those observed in C2C12 cells in the primary muscle cells isolated from newborn mice (21). In the process of myogenic inhibition in C2C12 cells, both BMP-2 and TGF-ß1 strongly down-regulated the levels of expression of mRNAs encoding MyoD and myogenin (21, 140), which are critical transcription factors regulating myogenic differentiation. Chalaux et al. (141) demonstrated the involvement of JunB in the early steps of inhibition of myogenic differentiation by BMP-2 and TGF-ß1. Thus, BMP-2 and TGF-ß1 have similar inhibitory effects on myotube formation, but only BMP-2 induced osteoblast differentiation, indicating different functional effects on osteoblast differentiation between these two molecules.

To understand the molecular mechanism involved in osteogenic transdifferentiation in C2C12 cells induced by BMP-2, the roles of BMPRs and Smads were investigated. Wild-type C2C12 cells expressed BMPR-IA and BMPR-II mRNAs, but not BMPR-IB mRNA (79). A subclonal cell line of C2C12 stably expressing a kinase domain-truncated BMPR-IA generated numerous myotubes but failed to differentiate into ALP-positive cells after treatment with BMP-2 (79). When wild-type BMPR-IA was transiently transfected into the BMPR-IA mutant cells, BMP-2 inhibited myogenic differentiation and induced ALP-positive cells (79). BMP-2 did not induce ALP-positive cells in BMPR-IA mutant cells transfected with wild-type BMPR-IB (79). These results suggest that BMP-2 signals inhibiting myogenesis and inducing osteoblast differentiation are transduced via BMPR-IA, at least in C2C12 cells. Interestingly, Akiyama et al. (87) demonstrated that C2C12 cells stably transfected with constitutively active BMPR-IB exhibited osteoblast phenotypic markers, but did not express myogenic phenotypic markers. These results suggest that a common signal transducer(s) including Smads is involved in the signal transduction pathway via BMPR-IA and BMPR-IB during differentiation of C2C12 cells. C2C12 cells constitutively expressed Smad1, Smad2, Smad4, and Smad5 mRNAs (94). Yamamoto et al. (94) demonstrated that Smad1 and Smad5, which belong to the R-Smad family and mediate BMP signaling, are involved in the process of myogenic inhibition and induction of osteoblast differentiation in C2C12 cells. Nishimura et al. (95) demonstrated that BMP-2 caused serine phosphorylation of Smad1 and Smad5, unlike TGF-ß. They also showed that the activation of Smad5 and subsequent formation of the complex of Smad5 and Smad4, which is alternatively called DPC4 and belongs to the C-Smad family, were key steps in the process of BMP-2-induced osteoblast differentiation in C2C12 cells (95). Overexpression of I-Smads (Smad6 and Smad7) repressed ALP activity induced by BMP-6 in C2C12 cells (142), whereas BMP-2 or BMP-7 markedly induced mRNA encoding Smad6 in C2C12 cells (143). These results suggest that Smad6 is involved in a feedback loop to regulate the signaling activity of BMPs.

Using primary cells isolated from human muscle, we reported that BMP-2 inhibited myotube formation and stimulated ALP activity, but failed to induce osteocalcin production (126). In addition, transplantation of these myogenic cells with BMP-2 using diffusion chambers into athymic mice induced ALP-positive cells in the chambers but did not induce formation of bone or cartilage. These results suggested that the capacity of human muscular cells to differentiate into the osteoblast lineage is more restricted than that in rodents.

Taken together, the findings obtained from in vitro experiments show that BMPs are important local factors regulating the differentiation pathway of mesenchymal cell lineages into osteoblasts, chondrocytes, adipocytes, and muscles. Furthermore, BMPs promote osteoblastic and chondrocytic differentiation, inhibit myogenic differentiation, and exert diverse actions on adipogenic differentiation.

3. Extracellular regulation of BMP activity. Recent molecular embryological findings have shown that BMPs play crucial roles in the induction and patterning of ventral mesoderm at an early stage of development (63). During gastrulation, the Spemann organizer provides essential patterning information to the adjacent mesoderm and the overlying ectoderm. In 1996, noggin (144) and chordin (145), which are the Spemann organizer signals, were demonstrated to bind BMP-4 with high affinities at an extracellular region and to antagonize the action of BMP (Fig. 3A). Subsequently, two other molecules, gremlin (146) and follistatin (147), were found to antagonize the action of BMP at the extracellular level. Indeed, noggin, chordin, and gremlin inhibited BMP-induced ALP activity in W-20–17 bone marrow stromal cells and C3H10T1/2 cells (144, 145, 146). During mouse embryogenesis, noggin is expressed not only in the node, notochord, and dorsal somite, but also in the condensing cartilage and immature chondrocytes (148, 149). Experiments in noggin null mutant mice indicated that this molecule plays important roles in normal patterning of the neural tube, somites, and cartilage including joint formation (148, 149). These results indicated that BMP activity is also regulated by BMP antagonists such as noggin, chordin, follistatin, and gremlin at the extracellular level.

View larger version (15K): [in this window] [in a new window]   Figure 3. Mechanisms of extracellular regulation of BMP activity. A, An inhibitory mechanism of chordin by competing with BMP ligands for receptor binding at an extracellular level. B, Xolloid, a human BMP-1 homolog in Xenopus, cleaves chordin, and reactivates BMP.

Recently, Engstrand et al. (152) reported that BMP-3 antagonized BMP-2-induced osteoblastic differentiation in W-20–17 cells. They also demonstrated increased bone formation and bone density in BMP-3-deficient mice compared with wild-type controls. These observations suggest that BMP-3 is an inhibitory regulator of bone formation. Further studies of the regulatory mechanism of action of BMP by antagonistic molecules at the extracellular level will provide deeper insight into the mechanism of osteoblast differentiation and bone formation by BMPs.

B. Sonic and Indian hedgehogs
1. Involvement of Sonic and Indian hedgehogs in skeletogenesis. The gene hedgehog is a segment polarity gene regulating embryonic segmentation and patterning in Drosophila and is highly conserved in vertebrates (23). In higher vertebrates, the Hedgehog gene family consists of at least three members, Shh, Ihh, and Dhh (23). Shh has multiple functions during formation of various organs and tissues including formation of skeletal tissues in vertebrae and limbs (25). The phenotypes observed in Shh knockout mice indicated that Shh plays a critical role in patterning of embryonic tissues, including the brain, the spinal cord, the eyes, and the skeleton (25). They completely lacked vertebrae and partly lacked autopods (25). These results suggested that Shh mutations cause some malformations in humans. Indeed, the similarity of forebrain development between Shh mutant mice and cases of human holoprosencephary with SHH mutation is reported (24, 153, 154, 155). In addition, mutation of human PATCHED, which encodes a transmembrane protein that negatively regulates Shh signaling in target cells, causes the human autosomal disease termed nevoid basal cell carcinoma syndrome (156). Developmental skeletal abnormalities and a high risk of various forms of cancers, mainly basal cell carcinoma, characterize this syndrome. Mutations in the human SHH gene and genes that encode components of its downstream intracellular signaling pathway also cause three distinct congenital disorders, Greig syndrome, Pallister-Hall syndrome, and isolated postaxial polydactyly (157). Thus, SHH signaling is involved in the pathogenesis of several diseases including those of skeletal tissues in humans.

Bitgood and McMahon (30) first reported that Ihh is expressed in cartilage during skeletogenesis in mouse embryos. Vortkamp et al. (26) demonstrated that Ihh regulated chondrocyte differentiation through regulation of PTHrP in chicken embryos. They also showed that the hedgehog- responsive genes Patched and Gli (transcription factor) were highly expressed in the perichondrium, where formation of bone collar occurred directly from perichondrial cells (26, 27). These results suggested that the target cells for Ihh are located in the perichondrium, and that Ihh induces adjacent perichondrial cells to differentiate into bone-forming osteoblasts. Since Shh and Ihh have similar functions in chondrocyte differentiation (26, 158), it is likely that these hedgehog proteins are involved in osteoblast differentiation as well as chondrocyte differentiation in vertebrates. Indeed, this was supported by the recent report that Ihh null mutant mice exhibited failure of osteoblast development in endochondral bones as well as markedly reduced chondrocyte proliferation and maturation (159).

The hedgehog family retains structural and functional similarities between Drosophila and vertebrates. In Drosophila, a major role of hedgehog signaling is the activation of additional signals including dpp, which is a homolog of vertebrate BMP, and wingless. Laufer et al. (160) reported that Shh is capable of regulating the expression of BMP-2 in chicken limb buds, because BMP-2 mRNA was expressed adjacent to Shh-expressing cells and the ectopic transplantation of Shh-expressing cells induced BMP-2 expression in the cells around the transplanted cells. By in situ hybridization using serial sections, Bitgood and McMahon (30) showed an intimate correlation between the expression of mouse Shh/Ihh genes and BMPs in various tissues. These findings prompted us to investigate whether Shh and Ihh are involved in osteoblast differentiation by a mechanism involving BMPs.

2. Regulation of osteoblast differentiation by hedgehogs. Several lines of evidence obtained from in vitro experiments indicate that hedgehogs regulate osteoblast differentiation. We first examined the effects of hedgehogs on osteoblast differentiation using the conditioned media collected from Shh- or Ihh-overexpressing chicken embryonic fibroblasts (106, 161). Addition of each conditioned medium increased ALP activity in C3H10T1/2 and MC3T3-E1 cells and increased the level of osteocalcin mRNA expression in MC3T3-E1 cells. Chicken embryonic fibroblasts used for the transfection of Shh or Ihh constitutively expressed substantial levels of mRNAs for BMP-2 and BMP-4. In addition, each conditioned medium induced no apparent increases in BMP-2, BMP-4, or BMP-6 mRNAs in C3H10T1/2 and MC3T3-E1 cells, but the increase in ALP activity induced by the conditioned media was abolished by addition of soluble BMPR-IA (Ref. 161 and T. Yuasa, and A. Yamaguchi, unpublished data), which antagonized the action of BMP on osteoblast differentiation in vitro (85). These results suggest that the stimulatory effects induced by addition of the conditioned media might be synergistically induced with Shh or Ihh and the BMPs produced by chicken fibroblasts themselves. Indeed, recombinant Shh (rShh) synergistically stimulated the BMP-2-induced ALP activity and the expression level of osteocalcin mRNA in C3H10T1/2 cells (T. Yuasa and A. Yamaguchi, unpublished data). Since the cooperative action of Shh and BMP-7 was reported in the induction of forebrain ventral midline cells by prechordal mesoderm (162), a cooperative effect of Shh and BMPs might be important in osteoblast differentiation as well. Murtaugh et al. (163) reported that chondrogenesis of somitic tissues is regulated by intimate interaction between Shh and BMPs. The intimate link between Ihh and the BMP/noggin signaling pathway during chondrocyte differentiation is also suggested by other investigators (164, 165, 166). Therefore, it is likely that Shh and BMPs act cooperatively during differentiation of osteochondrogenic cells, but further studies are necessary to determine the precise interaction between Shh and BMPs in this process.

3. Role of hedgehogs in bone formation. To investigate whether Shh and Ihh induce ectopic bone formation, we transplanted Shh- or Ihh-overexpressing chicken fibroblasts cultured on type I collagen gel into intraperitoneal sites in athymic mice (161). Endochondral bone formation was induced at the site of transplantation (106, 161). Since the transplanted chicken embryonic fibroblasts expressed low levels of mRNAs encoding BMP-2 and BMP-4, it should be elucidated whether such endochondral bone formation is due to the direct effect of hedgehog proteins alone or the synergistic effects of Shh/Ihh and BMPs. Further studies using recombinant proteins of hedgehogs are currently underway in our laboratory.

Important roles of Ihh during bone repair have been suggested by in vivo experiments. Vortkamp et al. (27) investigated the expression patterns of Ihh and BMPs during fracture repair. The fracture site expressed neither type X collagen, which is a marker of hypertrophic chondrocytes, nor Ihh at an early stage (within 3 days after fracture), but both mRNAs were strongly expressed in cartilaginous callus by 7 days after fracture. Ferguson et al. (28) also reported a similar expression pattern of Ihh during fracture repair. When the cartilage was completely replaced by bone at 3 weeks after fracture, expression of both mRNAs encoding type X collagen and Ihh disappeared (27). Interestingly, BMP-2 and BMP-4 were expressed in a number of chondrocytes of the healing callus overlapping the Ihh-expressing cells, suggesting some interaction between Ihh and BMPs during fracture repair (27). Although these observations suggest that Ihh is involved in fracture repair, further investigations are needed to explore a more precise role for Ihh because Ito et al. (29) reported that up-regulation of Ihh mRNA occurred within hours after fracture of mouse ribs.

Thus, hedgehogs, by interacting with BMPs, may play an important role in bone formation, especially at early stages of skeletogenesis and fracture repair.

	IV. Transcription Factors That Regulate Osteoblast Differentiation and Bone Formation

Top Abstract I. Introduction II. Origin of Osteoblasts III. Regulation of Osteoblast... IV. Transcription Factors That... V. Summary References

 
A. Transcription factors involved in osteoblast differentiation
Many transcription factors are involved in the regulatory mechanism of differentiation of different cell types. Among them, cell lineage-specific transcription factors play crucial roles in determining the fate of each cell type. Such transcription factors have been identified in several cell lineages including myoblasts and adipocytes, e.g., the MyoD family in myoblasts (31) and PPAR-2 in adipocytes (32) (Fig. 4).

View larger version (22K): [in this window] [in a new window]   Figure 4. Osteoblasts, chondrocytes, myotubes, and adipocytes arise from a common pluripotent mesenchymal cell. Each differentiation pathway is regulated by the cell lineage-specific transcription factors. The transcription factor(s) that determines chondroblast differentiation has not been clarified.

Recently, Schinke and Karsenty (179) purified osteoblast specific factor 1 (OSF1) as a 40-kDa protein, which specifically bound to distinct DNA sequences designated OSE1 in the OG2 promoter. They also suggested that OSF1 regulated Cbfa1 transcription by binding to the OSE1 sequence in Cbfa1 itself (179).

B. Cbfa1 is an important transcription factor regulating osteoblast differentiation
Two laboratories independently demonstrated that the osteoblast-specific DNA binding activity, designated OSF2 and NMP-2, was identical to Cbfa1 (33, 36). Thereafter, several laboratories including these two showed that Cbfa1 regulated the expression of various genes expressed in osteoblasts (33, 36, 180, 181, 182). Overexpression of Cbfa1 in nonosteogenic cells such as C3H10T1/2 cells and skin fibroblasts induced them to express osteoblast-related genes (33, 182). Cbfa1 was highly expressed in osteoblast lineage cells (33, 34). Antisense oligonucleotides for Cbfa1 down-regulated expression of osteoblast-related mRNAs in ROS17.2/8 osteoblastic cells (33). Using rat primary osteoblasts, Banerjee et al. (36) also demonstrated that antisense oligonucleotides for Cbfa1 inhibited osteoblast differentiation including formation of bone nodules in vitro. These results indicated that Cbfa1 plays a crucial role in osteoblast differentiation.

Ducy et al. (33) demonstrated that BMP-7 induced expression of Cbfa1 mRNA before induction of osteocalcin mRNA. BMP-2 also increased the level of Cbfa1 mRNA expression in an immortalized human bone marrow stromal cell line [hMC(2, 3, 4, 5, 6)] (137), C2C12 cells (37, 183), and 2T3 cells (135). Nishimura et al. (183) reported that BMP-2 induced Cbfa1 mRNA in C2C12 myoblasts, and this induction was abolished by overexpression of dominant-negative Smad1, Smad4, and Smad5. In addition, Hanai et al. (184) demonstrated that Smad 1 or Smad 5 and Cbfa1 formed complexes, indicating an intimate interaction between these molecules during osteoblast differentiation. These results suggest that Cbfa1 is a nuclear target of BMP signaling in osteoblast differentiation. On the other hand, we found that calvaria-derived cells isolated from Cbfa1-deficient embryos increased production of osteocalcin in response to BMP-2, although it was less than that produced by wild-type embryos (34). This suggests that transcription factors other than Cbfa1 also play some roles in BMP-2-induced osteocalcin synthesis, at least in vitro. Lee et al. (37) demonstrated that both BMP-2 and TGF-ß transiently up-regulated expression of Cbfa1 mRNA in C2C12 cells, but only BMP-2 induced expression of osteoblast differentiation-related mRNAs. Recently, Wang et al. (185) isolated several subclones from the MC3T3-E1 osteoblastic cell line. Characterization of each subclone indicated that the presence of Cbfa1 in a subclone was not sufficient for osteoblast differentiation (185). Taken together, these observations indicated that Cbfa1 plays a crucial role in the differentiation process of osteoblasts, but it is not a sufficient transcription factor for osteoblast differentiation. Isolation of cell lines from Cbfa1-deficient mice may provide useful tools for investigating transcription factors, other than Cbfa1, involved in osteoblast differentiation. Such studies are important to understand the regulatory mechanism of osteoblast differentiation, and they are currently underway in our laboratories.

C. Absence of ossification in Cbfa1-deficient mice
To investigate the precise function of Cbfa1, we disrupted exon 1 of the Cbfa1 gene, which contained the first 41 amino acids of the runt-domain (34). We extensively examined skeletal changes in Cbfa1-deficient mice on embryonic day 18.5 (E18.5) because Cbfa1-deficient mice died soon after birth due to respiratory insufficiency. In Cbfa1-deficient embryos at E18.5, only parts of the tibia, radius, and vertebrae were weakly calcified, and no calcification occurred in the skull, mandible, humerus, or femur, while wild-type embryos at E18.5 exhibited extensive calcification of all the skeletons on soft x-ray examination (Fig. 5). Histological examination revealed that Cbfa1-deficient embryos completely lacked ossification. Interestingly, ALP-positive cells surrounded calcified cartilage such as the tibia and radius in Cbfa1-deficient embryos, whereas no ALP-positive cells appeared around uncalcified cartilage such as the humerus and femur. These findings suggest that calcified cartilage contains some factor(s) inducing early differentiation of osteoblast lineage cells even in Cbfa1-deficient embryos. In E18.5 Cbfa1-deficient embryos, only a thin layer of the fibrous connective tissue was observed between the brain and subcutaneous connective tissue. ALP-positive cells were detected in the fibrous connective tissues, but no calcified bone was observed. Similar skeletal changes in Cbfa1-deficient mice were reported by Otto et al. (35). These morphological changes were confirmed extensively at ultrastructural and histochemical levels by Hoshi et al. (186). These results demonstrated that Cbfa1 is an important transcription factor for bone formation.

View larger version (64K): [in this window] [in a new window]   Figure 5. Soft x-ray of E18.5 wild-type (+/+) and Cbfa1-deficient (-/-) embryos. Wild-type embryo exhibits well calcified skeletons, but mutant embryos had barely calcified skeletons.

Since Cbfa1-deficient mice die soon after birth, it is difficult to explore the exact role of Cbfa1 in growing mice. To investigate the function of Cbfa1 in growing mice, Ducy et al. (187) generated transgenic mice overexpressing the Cbfa1 DNA-binding domain (Cbfa1) driven by the OG2 promoter. Cbfa1 was expressed in differentiated osteoblasts only postnatally and acted in a dominant-negative fashion due to a higher affinity for DNA than Cbfa1 itself. The skeleton of Cbfa1-transgenic mice was normal at birth, but they suffered from osteopenia due to a decrease in bone formation rate 3 weeks after birth. These results indicate that Cbfa1 plays a crucial role in not only osteoblast differentiation but also osteoblast function.

D. Role of Cbfa family transcription factors in osteoblast differentiation
Several isoforms of Cbfa1 produced by the differential promoter usage have been identified. One isoform originally cloned from ras-transformed NIH3T3 cells was named Pebp2A by Ogawa et al. (188, 189) (tentatively referred to as type I isoform, which begins with the N-terminal amino acid sequence MRIPVD). Subsequently, two other isoforms of Cbfa1 have been identified from osteoblasts and lymphoblasts (33, 190). In these two isoforms, two methionine residues were found in the novel N-terminal region: one was a shorter isoform translated from the second methionine residue (tentatively referred to as type II isoform, which begins with the N-terminal amino acid sequence MASNSL), and the other was a longer isoform translated from the first methionine residue (tentatively referred to as type III isoform, which begins with the N-terminal amino acid sequence MLHSPH). Type II isoform was originally reported as til-I by Stewart et al. (190), and type III was first identified as Cbfa1/Osf2 by Ducy et al. (33). The expression pattern of these Cbfa1 isoforms in various cell types has not been fully investigated. We demonstrated by RT-PCR analysis that these three isoforms were expressed in adult mouse bones (182). Xiao et al. (191) extensively analyzed genomic structure and isoform expression of mouse, rat, and human Cbfa1. They demonstrated that type II isoform was expressed in osteoblasts of all species, and type III isoform was recognized in osteoblasts of the mouse and rat but not in human osteoblasts. These expression patterns suggest that type II isoform, rather than type III isoform, plays an important role in osteoblast differentiation. The expression of Cbfa1 mRNA in nonosteogenic cells is still controversial. We demonstrated that C3H10T1/2 cells expressed undetectable levels of mRNAs for three isoforms of Cbfa1 in control culture (182). Ducy et al. (33) also reported that C3H10T1/2 cells expressed an undetectable level of Type III isoform of Cbfa1 in control culture, but another group showed that C3HT101/2 cells as well as NIH3T3 fibroblasts constitutively exhibited a substantial level of mRNA for Type I isoform (180). The discrepant results of Cbaf1 expression among laboratories using the same cells might arise from the different culture conditions employed.

As described above, Ducy et al. (33) demonstrated that transfection of type III isoform of Cbfa1 into nonosteogenic cells induced gene expression related to osteoblast differentiation, but functional differences between the three isoforms of Cbfa1 have not been clarified. Harada et al. (182) investigated the functional differences in these isoforms of Cbfa1 by transfection of the respective isoforms into C3H10T1/2 cells and transcription assay using Cbfa1 target gene promoter driven-luciferase reporter genes. Both transient and stable transfection with type I and type II Cbfa1 isoforms, but not with type III isoform, induced ALP activity in C3H10T1/2 cells. All of the Cbfa1 isoforms induced or up-regulated expression of osteocalcin, osteopontin, and type I collagen mRNAs in stable transformants, although the cells transfected with type II isoform exhibited the highest level of osteocalcin mRNA expression. Luciferase reporter gene assay using 6XOSE2-SV40 promoter (six tandem binding elements for Cbfa1 ligated in front of the SV40 promoter sequence) and mouse osteocalcin promoter revealed differences in the transcriptional induction of target genes by each Cbfa1 isoform. These findings were supported in a recent similar study by Xiao et al. (191). Although all three Cbfa1 isoforms might be involved in stimulation of osteoblast differentiation, the expression pattern of Cbfa1 isoforms and the transfection experiments of these isoforms suggest that the type III isoform has much less activity than the type II isoform. The lower translational efficiency of type III isoform compared with type II isoform (192) supports this notion.

It has been shown that Cbfa1 and Ets1, which is a nuclear phosphoprotein of the Ets transcription factor family modulating cell proliferation, differentiation, and oncogenic transformation (193), synergistically enhanced promoter activity of osteopontin in skeletal tissue (194). The molecular mechanism of DNA binding of Cbfa2 and Ets-1 has been well investigated as a model system of combinatorial control that utilizes multiple transcription factors (195, 196). Both Cbfa2 and Ets-1 contain a negatively regulatory domain for DNA binding in their sequences, and interaction between each negative regulatory domain is necessary and sufficient for cooperative DNA binding (195). Further investigation is necessary to gain deep insights into the regulatory mechanism of osteoblast differentiation by Cbfa1.

E. Cbfa1 is involved in chondrocyte maturation
Cbfa1 was apparently expressed in hypertrophic chondrocytes (34, 197). In Cbfa1-deficient mice, calcification of cartilage occurred in the distal limbs (tibia, fibula, radius, and ulna), and almost all other cartilage remained uncalcified (34). These observations suggested that Cbfa1 played some roles in chondrocyte differentiation. We investigated expression patterns of cartilage-related mRNAs in Cbfa1-deficient mice by in situ hybridization. In the distal limbs showing calcification, hypertrophic chondrocytes expressed Ihh, type X collagen, and BMP-6, but did not express osteopontin or collagenase 3 (197). In the humerus and femur in Cbfa1-deficient mice, however, chondrocytes expressed no detectable levels of mRNAs encoding PTH/PTHrP receptor, Ihh, type X collagen, or BMP-6, indicating that chondrocyte differentiation was blocked before prehypertrophic chondrocytes in these skeletal structures (197). Similar findings concerning maturational arrest of chondrocytes were reported in other Cbfa1-deficient mice (198) generated by Otto et al. (35). These observations suggest that Cbfa1 plays an important role in chondrocyte maturation.

F. Cbfa1 is involved in osteoclastogenesis
In 1981, Rodan and Martin (199) proposed an important hypothesis concerning the possible involvement of osteoblast lineage cells in the hormonal control of bone resorption. They suggested the potential direct activation of osteoclasts by the products of osteoblast lineage cells in response to bone-resorbing hormones. A series of experiments have confirmed this hypothesis (200, 201, 202), but the precise molecular mechanism involved in the interaction between osteoblast lineage cells and osteoclasts has not been clarified.

Recently, two molecules produced by osteoblast lineage cells, which play important roles in osteoclastogenesis, were identified. One is osteoprotegerin (OPG) (203), which is identical to osteoclastogenesis-inhibitory factor (OCIF) (204, 205). OPG is a secretary protein belonging to the TNF receptor family (203, 204, 205). This protein inhibited not only formation of osteoclast-like cells (OCLs) in culture but also bone resorption both in vitro and in vivo (203, 204, 205). In addition, OPG knockout mice exhibited severe osteopenia due to accelerated bone resorption (206, 207). The other molecule is RANKL (receptor activator of NF-kB ligand) (208), which is identical to OPG ligand (OPGL) (209), TRANCE (TNF- related activation-induced cytokine) (210), and osteoclast differentiation factor (ODF)(211). RANKL belongs to the TNF ligand family and binds to OPG. A soluble form of RANKL (soluble RANKL) together with macrophage colony-stimulating factor induced formation of OCLs from spleen cells in the absence of osteoblast lineage cells in vitro (209, 211). Recently, Kong et al. (212) reported that OPGL- deficient mice exhibited severe osteopetrosis and completely lacked osteoclasts as a result of an inability of osteoblasts to support osteoclastogenesis. The formation of OCLs induced by soluble RANKL was completely abolished by the addition of OPG (209, 211), indicating a specific interaction between RANKL and OPG in osteoclastogenesis.

We reported that osteoclastogenesis was markedly retarded in Cbfa1-deficient mice (34). These results suggested that the maturational arrest of osteoblasts caused by disruption of the Cbfa1 gene might be related to the insufficient osteoclastogenesis in Cbfa1-deficient mice. These observations also allowed us to speculate on the role of Cbfa1 in the regulation of RANKL and OPG, because both are synthesized by osteoblast lineage cells.

We investigated the mechanism involved in retarded osteoclastogenesis in Cbfa1-deficient mice (213). Cocultures of calvarial cells isolated from embryos with three different Cbfa1 genotypes (Cbfa1^+/+, Cbfa1^+/-, and Cbfa1^-/-) and normal spleen cells generated TRAP-positive OCLs in response to 1,25(OH)₂D₃ and dexamethasone, but the number and bone-resorbing activity of OCLs formed in coculture with Cbfa1^-/- calvarial cells were significantly decreased in comparison with those formed in cocultures with Cbfa1^+/+ or Cbfa1^+/- calvarial cells. The expression of RANKL mRNA was increased by treatment with 1,25(OH)₂D₃ and dexamethasone in calvarial cells from Cbfa1^+/+ and Cbfa1^+/- mouse embryos, but not in those from Cbfa1^-/- embryos. In contrast, the expression of OPG mRNA was inhibited by 1,25(OH)₂D₃ and dexamethasone to a similar extent in all three types of calvarial cells. RANKL and OPG mRNAs were highly expressed in the tibia and femur of Cbfa1^+/+ and Cbfa1^+/- embryos. In the tibia and femur of Cbfa1^-/- embryos, however, RANKL mRNA was undetectable, and the expression of OPG mRNA was also decreased compared with those in Cbfa1^+/+ and Cbfa1^+/- embryos. Thus, it is likely that Cbfa1 is involved, at least in part, in osteoclastogenesis by regulating the expression of RANKL. This was supported by recent reports by O’Brien et al. (214) and Kitazawa et al. (215). They identified potential Cbfa1 binding sites in the promoter region of murine RANKL, suggesting that Cbfa1 may directly regulate RANKL expression. More extensive studies on the regulation of RANKL by Cbfa1 will provide insight into the molecular mechanism involved in the classical hypothesis proposed by Rodan and Martin (199) concerning the interaction between osteoblasts and osteoclasts during bone remodeling.

G. Heterozygous mutations of Cbfa1 locus cause cleidocranial dysplasia
Cleidocranial dysplasia (CCD) is an autosomal-dominant disease showing hypoplastic clavicles, open fontanelles, supernumerary teeth, short stature, and other skeletal changes (216, 217). Mice heterozygous for mutation in the Cbfa1 locus (Cbfa1^+/-) (34, 35) exhibited similar skeletal changes to CCD (218, 219). They exhibited hypoplastic frontal, parietal, interparietal, temporal, and supraoccipital bones with open fontanelles and sutures. The