Department of Biochemistry, School of Dentistry, Kyungpook National University, Taegu, Korea
Biomolecular Engineering Center, Kyungpook National University, Taegu, Korea
Address reprint requests to: Hyun-Mo Ryoo, D.D.S., Ph.D., Department of Biochemistry, School of Dentistry, Kyungpook National University, 101 Dong In-Dong, Jung-Gu, Taegu 700-422, Korea, or, Hyun-Jung Kim, D.D.S., Ph.D., Department of Pediatric Dentistry, School of Dentistry, Kyungpook National University, 101 Dong In-Dong, Jung-Gu, Taegu 700-422, Korea
Runx2 (previously known as Cbfa1/Pebp2αA/AML3), a key transcription factor in osteoblast differentiation, has at least two different isoforms using alternative promoters, which suggests that the isoforms might be expressed differentially. Haploinsufficiency of the Runx2 gene is associated with cleidocranial dysplasia (CCD), the main phenotype of which is inadequate development of calvaria. In spite of the biological relevance, Runx2 gene expression patterns in developing calvaria has not been explored previously, and toward this aim we developed three probes: pRunx2, which comprises the common coding sequence of Runx2 and hybridizes with all isoforms; pPebp2αA, which specifically hybridizes with the isoform transcribed with the proximal promoter; and pOsf2, which hybridizes with the isoform transcribed with the distal promoter. These probes were hybridized with tissue sections of mouse calvaria taken at various time points in development. Runx2 expression was localized to the critical area of cranial suture closure, being found in parietal bones, osteogenic fronts, and sutural mesenchyme. Pebp2αA and Osf2 showed tissue-specific expression patterns. The sites of Pebp2αA expression were almost identical to that of pRunx2 hybridization but expression was most intense in the sutural mesenchyme, where undifferentiated mesenchymal cells reside. The Osf2 isoform was strongly expressed in the osteogenic fronts, as well as in developing parietal bones, where osteopontin (OP) and osteocalcin (OC) also were expressed. However, in contrast to Pebp2αA, Osf2 expression did not occur in sutural mesenchyme. Pebp2αA also was expressed prominently in primordial cartilage that is found under the sutural mesenchyme and is not destined to be mineralized. Thus, Osf2 isoforms contribute to events later in osteoblast differentiation whereas the Pebp2αA isoform participates in a wide variety of cellular activities ranging from early stages of osteoblast differentiation to the final differentiation of osteoblasts.
BONE FORMATION occurs by either a direct intramembraneous or an indirect endochondral process. Both processes require that mesenchymal cells condense so that the differentiation of stem cells and synthesis and mineralization of extracellular matrix can occur. Intramembraneous bone formation occurs during embryonic development by the direct transformation of mesenchymal stem cells into osteoblasts. This type of bone formation is restricted to the development of the cranial vault, some facial bones, and parts of the mandible and clavicles. Endochondral bone formation occurs through the differentiation of condensed mesenchyme into cartilage, which then is transformed into bone. This type of bone formation occurs in the weight-bearing long bones and a group of bones located in the cranial base.
Runx2, previously referred to as Pebp2αA/Cbfa1/AML3 and now authorized by a nomenclature committee to be denoted as Runx2, has been revealed to be a transcription factor that controls osteogenesis of both the intramembraneous and the endochondral types. Its DNA binding domain is homologous to that of runt, the Drosophila pair-rule gene.(1) Homozygous disruption of the gene results in the complete absence of both intramembraneous and endochondral bone formation,(2) whereas haploinsufficiency of the gene results in cleidocranial dysplasia (CCD), which is characterized by hypoplastic clavicles and delayed suture closure in humans(3, 4) and mice.(5) In addition, it regulates the transcription of many osteoblast genes such as osteocalcin (OC),(6, 7) osteopontin (OP),(8) bone sialoprotein,(9) collagenase 3,(10, 11) and osteoprotegerin.(12) Moreover, transgenic mice overexpressing the dominant negative form of Runx2 in differentiated osteoblasts after birth developed an osteopenic phenotype.(13) These observations support the widely held view that Runx2 is a master transcription factor that controls osteoblast differentiation as well as the maintenance of differentiated osteoblasts.
Three different messenger RNA (mRNA) sequences of Runx2 have been reported, and are caused by different promoter usage or alternative splicing.(14–16) One of these isoforms, denoted as Pebp2αA, uses the proximal promoter (P2) and was reported originally as a T cell-specific factor.(17) The other two isoforms, Osf2/Cbfa1(7) and Til-1 G1,(14) use the distal promoter (P1) and generally are referred to as Osf2. Although the latter two isoforms have different mRNA sequences because of alternative splicing in the 5′-untranslated region (UTR),(16) their translation products are identical.(15, 18) Consequently, we refer to the proximal promoter product as Pebp2αA and the distal promoter product as Osf2. These products also have been denoted as types I and II,(19) respectively, as well as p56 and p57 (according to the molecular weight of the translation products).(15) Because the two isoforms use different promoters it is strongly implied that their expression pattern may differ, and indeed there have been reports suggesting that they may play different roles in endochondral bone formation.(19) However, their expression pattern in intramembraneous bone formation has not been investigated. In particular, whether any of the isoforms are specifically responsible for the impaired intramembraneous bone formation that leads to the CCD phenotype is not known. In addition, whether the isoforms vary functionally is not clear.
The expression of bone-related genes during bone formation was studied by using both in vivo and in vitro systems. The in vivo systems consist of developing rat bones and mouse embryos and have been studied for the expression of bone-marker genes. The in vitro models consist of primary explant cultures of fetal rat/mouse calvaria or bone marrow cells, and these have been studied extensively for the expression of bone-related proteins during osteoblast differentiation. Studies using these in vivo and in vitro systems have shown, among other things, that bone-marker genes are expressed in a temporal pattern.(20, 21) In the study reported here, we have used developing calvaria as our model of typical intramembraneous bone formation. In ontogeny, the early development of calvaria is coordinated with the growth of the brain through a series of interactions between the developing brain, the growing cranial bones, and the sutures that unite the bones. The first steps in cranial suturing involve the proliferation and condensation of sutural mesenchymal cells at the periphery of the extending bone field. This field is called the osteogenic front and the cells contained in the front undergo differentiation into osteoblastic lineage cells that ultimately synthesize the bone matrix and lay down mineral in the matrix, resulting in formation of the parietal bones.(22, 23) Here, we show that early cranial suture morphogenesis is a good experimental model system for the study of osteoblast differentiation from undifferentiated mesenchymal cells to osteoblasts that are fully differentiated. This model has the major advantage of the cells in distinct stages of differentiation separating anatomically from each other. This has been shown in this study by cellular features and the tissue-specific expression patterns of two osteoblast marker genes OP and OC. We used this model to examine the normal expression patterns of the Runx2 isoforms during cranial bone formation. Based on these expression patterns, we attempted to determine if the Runx2 isoforms vary in their functional properties in intramembraneous bone formation.
MATERIALS AND METHODS
Preparation of probes
To assess the expression patterns of the Runx2 isoforms, we subcloned three different mouse Runx2 complementary DNA (cDNA) constructs (Fig. 1) into pBluscript KS + vector. pRunx2 includes 1.6 kilobases (kb) of the common coding region of Runx2 and hybridizes with all Runx2 transcripts. It was subcloned from Pebp2aA expression vector(24) that was digested with BamHI and XbaI, and subcloned into the same restriction site in pBluscript KS + vector. pPebp2αA is a 556-base pair (bp) Pebp2αA-specific sequence from the 5′-UTR of mouse Pebp2αA(17) using XbaI and BsaWI and then subcloned into XbaI and XbaI site of pBluscript KS + vector. pOsf2 is specific for transcripts regulated by the distal promoter and was subcloned from the 5′-UTR and part of the 5′ coding sequence of Til-1 G1(14) using BamHI (412 bp). The orientation of each cloned cDNA construct was checked by restriction digestion and sequencing.
Calvariae of ICR mice aged between embryonic day 15 (E15) and postnatal day 3 (P3) were prepared as previously described.(25) Briefly, the dissected calvariae were fixed overnight at 4°C in 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS). Coronal sections of tissue consisting of parietal bones and the interposed sagittal suture were dehydrated by a series of ethanol steps and embedded in paraffin. Five-micrometer sections were mounted on silanized slides, dried overnight at 37°C, and stored at 4°C. Postnatal material for tissue sections was decalcified at 4°C in 12.5% EDTA/2.5% PFA in PBS for about 10 days, with a solution change every fourth day. Sections were stained with hematoxylin and eosin to assess the histology of the developing calvaria.
In situ hybridization
The pRunx2 was digested with BamHI or XbaI. Antisense and sense riboprobes were produced by T3 and T7 RNA polymerase, respectively. The pPebp2αA vector was digested with SacI or HindIII and antisense and sense riboprobes were produced by T3 and T7 RNA polymerase, respectively. The pOsf2 was digested with XhoI or SacI and antisense and sense riboprobes were produced by T3 and T7 RNA polymerase, respectively. OP and OC probes were prepared as previously described.(26) In situ hybridization on tissue sections was performed using [35S]uridine triphosphate (UTP)-labeled (Runx2 isoforms and OC probes) or digoxigenin-UTP-labeled (OP) riboprobes as described previously.(25, 27) For [35S]UTP-labeled riboprobes, final probe concentration was adjusted between 50,000 and 60,000 cpm/μl. After 2 minutes of denaturation at 80°C, 20-100 μl of probe solution was placed on each slide and covered in parafilm. After overnight hybridization in a humidified sealed box at 52°C, high stringency washes with 50% formamide and 20 mM dithiothreitol (DTT) at 65°C was carried out. Slides were then prepared for autoradiography. The dehydrated slides were dipped into photographic emulsion (Kodak NTB-2; Eastman Kodak, Rochester, NY, USA), dried, and exposed for between 2 and 3 weeks at 4°C. The slides were then developed (Kodak D-19; Eastman Kodak), fixed (Kodak Unifix; Eastman Kodak), and then briefly counterstained with hematoxylin or methyl green and mounted with DePeX (BDG). For digoxigenin-UTP-labeled OP riboprobe, final concentration of probe was 0.5-1.0 μg/ml. After a similar hybridization and washing step, color development was carried out in the dark for 2 h following the manufacturer's manual (Promega, Madison, WI, USA) and then briefly counterstained with methyl green and mounted with DePeX (BDG, Poole, UK).
The histological appearance of sagittal suture morphogenesis was analyzed by making coronal sections of calvaria taken from E15 to P1 mice and staining them with hematoxylin and eosin (Figs. 2A-2F). The coronal section of developing calvaria includes three histological landmarks, namely, the parietal bones (arrow), the osteogenic fronts (arrowhead), and the intervening sutural mesenchyme (asterisk; Figs. 2A and 2B). Previous reports showed that sutural mesenchyme includes many subdifferentiated mesenchymal cells.(23, 25) During development, osteogenic fronts are formed as the sutural mesenchymal cells begin to proliferate and differentiate into the osteoblastic lineage. The resulting osteoblasts synthesize the parietal bone matrix and lay down mineral in the matrix.(23, 25) Histologically, the cells in osteogenic fronts can be distinguished from the surrounding sutural mesenchymal cells by their close proximity to each other (mesenchymal condensation) and their increased cytoplasmic density (Fig. 2C). At E15, the osteogenic fronts of parietal bones in the presumptive sagittal suture are widely separated by abundant mesenchyme (Fig. 2A) but the amount of the intervening mesenchyme decreases dramatically as the suture space becomes reduced over subsequent days of development (Figs. 2B, 2D, and 2E). Suture closure occurs very rapidly and is shown by the changes displayed by the E16 section relative to the E15 section (Figs. 2A and 2B). Thus, during this stage of development, osteoblast differentiation progresses swiftly.
With regard to the expression of the OP and OC genes, which are markers of fully differentiated osteoblasts, the cells in the sutural mesenchyme did not hybridize with either the OC or the OP probes (Figs. 2G, 2H, 2J, and 2K). The primordial cartilage under the sutural mesenchyme found at later embryonic stages and in the postnatal period (Figs. 2D and 2E) also did not hybridize with either probe (Figs. 2G, 2H, 2J, and 2K). In addition, the cells in the osteogenic fronts did not hybridize with the OP or OC probes (Figs. 2G, 2H, 2J, and 2K). Negative hybridization results were obtained by using all sense probes (Figs. 2I and 2L). OP and OC expression was found only in the lining cells of parietal bones, which are known to contain fully differentiated osteoblasts (Figs. 2G, 2H, 2J, and 2K). Thus, mesenchymal cells first attain the characteristics of preosteoblasts in osteogenic fronts before becoming fully differentiated OP- and OC-positive osteoblasts in parietal bones. There were differences in the OP and OC expression patterns. OP expression occurred early and was found in the parietal bones next to the osteogenic fronts (Figs. 2G and 2H). However, OC expression could only be detected in later stages of bone formation (E17) and was found only in the lining cells of parietal bones, far from the osteogenic fronts (Figs. 2J and 2K). These observations strongly support the idea that OC is a marker of events that occur later in bone formation, whereas OP is an earlier marker.
With regard to expression of Runx2 and its isoforms, pRunx2 strongly hybridized to parietal bones and osteogenic fronts and also was detected in the sutural mesenchyme (Figs. 3A-3D). Runx2 expression was stronger at the early stages of cranial suture development (Figs. 3A and 3B), at which point osteoblast differentiation was very active. In terms of the expression of the two Runx2 isoforms, although pPebp2αA expression coincided with all the areas where pRunx2 had hybridized (Figs. 3F-3I), it was most strongly expressed in sutural mesenchyme and its expression levels gradually decreased as cell differentiation progressed (Figs. 3H and 3I). Thus, its expression was weakest in the parietal bones. In contrast, Osf2 was expressed only in the parietal bones and osteogenic fronts (Figs. 3K-3N) and not at all in the sutural mesenchyme (Figs. 3F-3G). Negative hybridization results were obtained by using all sense probes (Figs. 3E, 3J, and 3O).
The expression patterns of both isoforms and several bone marker genes (Table 1) thus show that Pebp2αA expression occurs mainly during earlier stages of osteoblast differentiation, such as in undifferentiated mesenchymal cells and preosteoblasts, whereas Osf2 expression occurs during later stages of differentiation. This pattern of stage-specific expression of the two Runx2 isoforms in intramembraneous bone formation is quite similar to that found in endochondral bone formation,(19) where Pebp2αA expression is localized in prehypertrophic chondrocytes whereas Osf2 expression occurs in hypertrophic chondrocytes and osteoblasts.
Table Table 1.. The Expression Pattern of Runx2 Isoforms and Several Bone-Marker Genes During the Mouse Sagittal Suture Morphogenesis
In our experiments, the primordial cartilage that is transiently found under the sutural mesenchyme during osteogenesis and does not go on to transform into bone (Figs. 3P-3R) was found to hybridize only with the Pebp2αA isoform. This observation, together with the differential expression patterns of the two isoforms, strongly suggests that the expression of the isoforms clearly requires different transcriptional environments and the functional differences of the isoforms.
The early development of calvaria is a good model for the study of osteoblast differentiation
Osteogenesis involves the continuous recruitment and proliferation of osteoprogenitor cells and their differentiation into osteoblasts.(20) Many transcription factors, including Runx2, are involved in this complex process.(13, 28–30) Examination of these processes in vivo is hampered because the bone in which the osteoblasts reside is very hard to process and also because the cells in various stages of differentiation are in close proximity to each other. In vitro osteoblast differentiation models also have several problems, namely, it takes a long time before a fully differentiated phenotype is achieved in vitro and, in addition, even if the cells in a culture dish generally have achieved a specific stage of differentiation, the cells continue to be heterogeneous in their degree of differentiation. To overcome some of these problems, we used early cranial suture development as a model to study osteoblast differentiation. This model has several advantages over other model systems used for this purpose. As indicated in a previous study,(25) the cells in osteogenic fronts distinguish themselves from the surrounding mesenchymal cells by their cellular condensation, denser cytoplasm, and extensive protein synthesis machinery. The gene expression patterns of fibroblast growth factor receptor 2 (FGFR-2), bone morphogenetic protein 2 (BMP-2), sonic hedgehog,(25) and bone sialoprotein(31) also indicate that the cells in osteogenic fronts are quite different from surrounding mesenchymal cells. In this study, we show that the cells in osteogenic fronts also can be distinguished clearly from the cells in parietal bones by the bone-specific markers OC and OP, because only cells in parietal bones expressed these genes. These results strongly suggest that the cell population in osteogenic fronts bears the characteristics of just-committed preosteoblasts. This conclusion is supported by gene expression patterns previously found both in vivo and in vitro.(20, 32) Thus, the greatest advantage of this model system is that cells at the same differentiation stage are quite clearly grouped in limited anatomical sites, such that mesenchymal cells, preosteoblasts, and osteoblasts can be found only in sutural mesenchyme, osteogenic fronts, and parietal bones, respectively. These features are summarized schematically in Fig. 4.
Differential temporal and spatial expression of the two Runx2 isoforms suggests they may differ functionally
The Runx2 gene is comprised of at least nine exons.(3, 16, 33) From this gene, three isoforms can be generated through alternative promoter usage and splicing. The Pebp2αA, in which its translation begins with MRIPVD, is regulated by a proximal promoter(3, 17, 34) (it also is variously designated as type I/Cbfa1(19) or p56(15)). The other two isoforms use a distal promoter and are designated as Osf2/Cbfa1(13) and Til-1 G1.(14) Recent studies have shown that the translation products of the latter two isoforms are identical.(15, 18) For this reason, we have grouped these two isoforms and denoted them as Osf2 (they are also designated by other groups as type II/Cbfa1(19) or p57(15)).
The Pebp2αA and Osf2 isoforms use different promoters, strongly implying that they differ in their tissue- and stage-specific expression. However, although there have been reports suggesting that their expression patterns differ during endochondral bone formation,(19) the case for intramembraneous bone formation has not been studied. Specifically, their expression patterns during calvaria development have not been examined. This is of wider relevance as it may be that the CCD syndrome in humans is characterized by hypoplastic clavicle and delayed calvaria bone development.(3, 4) In this article, we show that Pebp2αA and Osf2 are expressed in different tissues and at different stages of differentiation. Although Osf2 expression was prominent in the preosteoblasts contained in osteogenic fronts, it was somewhat weaker in the fully differentiated OP- and OC-positive osteoblasts in parietal bones. These observations strongly support the notion that Osf2 (osteoblast-specific factor 2) really is bone specific. In contrast, Pebp2αA expression was most strongly detected in undifferentiated mesenchymal cells, with only weak expression being found in osteogenic fronts and parietal bones. Thus, Pebp2αA expression occurs during earlier stages in osteoblast differentiation, whereas Osf2 expression occurs later. The sequence of the distal promoter of the Runx2 gene shows that there are six Runx2 binding sites,(35) and thus it is possible that a regulatory cascade between Pebp2αA and Osf2 might exist. For instance, the expression of Pebp2αA during the commitment of mesenchymal cells into osteoblasts might stimulate Osf2 expression so that the final differentiation of the cell can be accomplished.
The question remaining is what are the functional differences between the two isoforms, if any? Furthermore, could the differential expression patterns of the two isoforms during bone formation have any functional consequences? The fact that these two isoforms are functionally different is suggested by their different temporal expression patterns but also by several other lines of evidence. First, the N-terminal amino acids differ between the PEBP2αA and OSF2 proteins in that the N-terminal of PEBP2αA bears five unique amino acids, whereas the N-terminal of OSF2 consists of 19 unique amino acids.(16) The 19 amino acids of OSF2 contain a transcriptional activation domain (AD1)(19) that is not present in PEBP2αA. Second, forced expression of either isoform of Runx2 in pluripotent C3H10T1/2 cells resulted in functional differences, although they were not marked.(36) Third, although studies on endochondral bone formation have shown that Pebp2αA plays an important role during earlier stages of chondrocyte maturation while Osf2 acts in hypertrophic chondrocytes and osteoblasts,(19) we showed in this study that the primordial cartilage (that lies under the developing calvaria and normally does not undergo mineralization) hybridized only with Pebp2αA but not with Osf2. These results consistently indicate that the role of Osf2 is in the final differentiation of the osteoblast either by intramembraneous or endochondral processes whereas that of Pebp2αA is in a much earlier stage. All of these observations strongly support the notion that the two isoforms differ functionally. However, we are still not able to clarify the exact roles that these isoforms play in bone development; therefore, further study is required for clarification of this question.
It should be noted that the C-terminal splice variant of the Osf2 isoform Til-1 G2 may have different functions to the full-length Osf2. Til-1 G2 is expressed naturally in ATDC5, a chondrocytic cell line, and its overexpression in this line suppresses OC promoter activity and chondrocytic differentiation of the cells.(37) Thus, this isoform exhibits dominant negative activity over the full-length Osf2. Regarding the CCD syndrome, CCD has been associated with mutations in the runt or C-terminal domains of Runx2. However, it does not appear that these mutations specifically target one of the Runx2 isoforms(3, 4, 38) and, thus, CCD probably does not result from the dominant expression of either of the Pebp2αA and Osf2 isoforms. From this, we can speculate that both isoforms are required in the process of intramembraneous bone formation of the calvaria.
Taken together, our results strongly suggest that PEBP2αA and OSF2 differ functionally, but that both are crucial in bone development. The fact that Pebp2αA expression is widespread, being found in underdifferentiated sutural mesenchymal cells, preosteoblasts in osteogenic fronts, OP- and OC-positive osteoblasts in parietal bones, and primordial cartilage cells, indicates that Pebp2αA plays an important role in a wide range of cellular differentiation events. In terms of intramembraneous bone formation, it probably is involved in the initial commitment steps and continues to exert its effects to the final differentiation of osteoblasts. Because expression of Osf2 is restricted to later events of cell differentiation, being found only in preosteoblasts and osteoblasts, its contribution probably is more specific in the final step into osteoblast differentiation.
The authors thank Dr. Suk-Chul Bae, Chungbuk National University, Chungju, Korea, for his Runx2 cDNA and personal advices with his unpublished data. This work was supported by a Korea Science and Engineering Foundation general research grant (KOSEF 1999-2-209-008-5).