Enhanced expression of TGF-β and c-fos mRNAs in the growth plates of developing human long bones

The longitudinal growth of skeleton occurs mainly at growth plates of long bones. This process involves in sequence, chondrocyte proliferation, deposition of cartilaginous matrix (containing type II collagen), hypertrophy and degeneration of chondrocytes, calcification of cartilage, and finally replacement of this matrix by capillaries and by osteoblasts producing a matrix of type I collagen fibres which subsequently becomes calcified (von der Mark & von der Mark, 1977). The growth plate provides an interesting developmental model since the location of individual cells determines their stage of development and all the stages can be studied in the same sample (Sandberg & Vuorio, 1987). Examination of this model of programmed cell growth and differentiation for the spatial and temporal expression of genes for various growth factors and proto-oncogenes should help to establish relationships between such factors and production of type I and type II collagens, as well as bone and cartilage differentiation. For this study, we chose the transforming growth factor-β (TGF-β) and protooncogene c-fos which both have been implicated to have a role in osseochondral development.

Transforming growth factor β (TGF-β) is a 25×10³M_r polypeptide growth factor composed of two identical disulphide-linked chains of 112 amino acids (Sporn et al. 1986). The ordinary TGF-β has been renamed TGF-β1 since another homologous TGF-β2 chain has recently been discovered (Cheifetz et al. 1987). The complete sequence of human TGF-is known from the corresponding cDNA (Derynck et al. 1985). The mRNA transcripts for TGF-β have been detected in a variety of tissues. Several lines of investigation have proposed important functions for TGF-β in connective tissue. These include stimulation of fibroblast proliferation (Roberts et al. 1985; Leof et al. 1986) and collagen production (Sporn et al. 1983; Ignotz & Massagué, 1986; Roberts et al. 1986; Raghow et al. 1987; Ignotz et al. 1987), as well as regulation of pericellular proteolysis (Laiho et al.1986). Bone has been shown to contain high levels of TGF-β (Seyedin et al. 1985), although the cell responsible for its production remains unknown. Recently, the cartilage-inducing factor-A (CIF-A) from bovine demineralized bone has been shown to be identical with TGF-β1 (Seyedin et al. 1986). Available sequence data suggests a similar relationship for CIF-B (Seyedin et al. 1987) and TGF-β2 (Cheifetz et al.1987). Antibodies capable of recognizing the N- terminal portion of TGF-β have been used to localize this factor in osteocytes, chondrocytes and some bone marrow cells of developing bovine bones (Ellings-worth et al. 1986). All this data suggests that TGF-β could have an important role in the metabolism of chondrocytes and osteoblasts.

The cellular proto-oncogene c-fos has also been suggested to play a role in osseochondral development (Rüther et al. 1987). The transient expression of the c-fos gene by various growth factors suggests that its protein product has a function in growth control and differentiation (Verma et al. 1986; Müller, 1986; Adamson, 1987). Two lines of evidence suggest that c-fos expression has a role in bone development. Deregulated expression of c-fos under the metallo-thionein promoter in transgenic mice was found to specifically interfere with the development of long bones (Rüther et al. 1987). Two mouse sarcoma viruses, capable of inducing osteosarcoma-like tumours in mice, have been found to carry structurally different v-fos genes (Verma et al. 1986). The v-fos gene product can also transcriptionally frarax-activate promoters of other genes such as the gene for type III collagen (Setoyama et al. 1986).

In the present study, we attempted to determine the roles of TGF-β and c-fos in normal osseochondral development by localizing their expression by Northern and in situ hybridization in the various zones of developing human growth plates.

Tissue samples from human fetal long bones and calvaria were obtained from prostaglandin-induced abortuses of 16 weeks of gestation. For isolation of total RNA, the long bones were dissected free from surrounding tissues and divided into three fractions: epiphyseal (resting) cartilage, growth cartilage and diaphyseal bone. The methods used for RNA extraction and analyses have been described before (Elima et al. 1985; Vuorio et al. 1987). RNA was also extracted from calvarial bone, which represents intramem-branous ossification. For Northern blotting, samples of total RNA were electrophoresed under denaturing conditions and transferred to Pall Biodyne or GeneScreen Plus membranes as suggested by the suppliers. The filters were prehybridized and hybridized with nick-translated ³²P-dCTP-labelled probes. After washing, the hybridization signals were detected using X-ray films and intensifying screens.

Details of the techniques have been published elsewhere (Sandberg & Vuorio, 1987). Tissue samples from radial heads of 16-week human fetuses were fixed with formaline, embedded in paraffin and sectioned. For in situ hybridization, the sections were pretreated with proteinase K and HC1, and acetylated. The hybridizations were performed for 50 h using probes labelled by nick-translation with ³³S-deoxy(thio)ATP, washed and autoradiographed for 7–29 days, followed by staining of the sections with haematoxylin.

Both the RNA filters and tissue sections were hybridized with the following cDNA probes: pHCALl (Vuorio et al. 1987) for human proαl (I) collagen mRNA, pHCAR3 (Sandberg & Vuorio, 1987; Elima et al. 1987) for human proαl (II) collagen mRNA, and pβCl (Derynck et al. 1985) for human TGF-β1 mRNA. To detect c-fos mRNAs, genomic subclone p21Al, containing a 3·1 kb XhoI-NcoI fragment corresponding to most of the coding sequence, was used (van Straaten et al. 1983). For Northern hybridizations, whole plasmids were nick-translated to specific activities of approximately 10⁸cts min^-1 using ³²P-dCTP. For in situ hybridization, specific (300–400bp) restriction fragments were isolated from inserts of the plasmidsand labelled by nick-translation with ³⁵S-deoxy(th-io)ATP. For in situ hybridization, 100–700bp fragments of phage lambda DNA, generated with BglI, were collected, nick-translated and used for negative control probes.

Northern hybridization was first employed to compare the relative amounts of mRNAs for type I and type II collagen, TGF-β and c-fos in total RNA extracted from various parts of developing bones (Fig. 1). The levels of type I collagen mRNAs were high both in growth plates (lane lb) and calvarial bone (lane Id), but low in diaphyseal bone (lane 1c), and almost undetectable in epiphyseal cartilage (lane la). The hybridization signal for type II collagen mRNA was strongest in the growth plates (lane 2b). Only a low level of type II collagen mRNA was seen in epiphyseal cartilage (lane 2a) and none was detected in bony tissues. Hybridization of the same filters with the cDNA probe for TGF-β revealed a 2·5 kb band in all four RNAs. The highest levels of TGF-β mRNA were seen in growth plates (lane 3b); in calvaria (3d), diaphyses (3c) and epiphyseal cartilages (3a), the levels were considerably lower. When the genomic clone for c-fos was used as a probe, the strongest hybridization was observed to the epiphyseal cartilage RNA (lane 4a). Two mRNA bands with approximate sizes of 2·2 and 3·5 kb were detected, the smaller one corresponding in size to the mature mRNA and the larger one to the unspliced primary transcript (Verma et al. 1985). Lower levels of these mRNAs were seen in the other samples.

Identification of the cells responsible for production of these mRNAs was accomplished by in situ hybridization of sections prepared from fixed tissues of the same fetuses used for RNA isolation. The low-power views of two typical sections and hybridization patterns (dark-field images), illustrated in Fig. 2, demonstrate the various structures studied and the specificity of the hybridization patterns obtained with the different probes.

A remarkably sharp line of demarcation was observed between the bone-specific type I (Figs 2B, 3A,B) and the cartilage-specific type II collagen mRNAs (Figs 2C, 3C,D): there is no overlap between the hybridization patterns obtained with cDNA probes for the two collagen types. When TGF-β cDNA was used as the probe, hybridization was mainly detected in cells containing high levels of type I procollagen mRNAs in developing bone and soft tissues, but only in small amounts in some (but not all) chondrocytes of the growth plate and epiphyses (Figs 2D, 3E,F). Osteoclasts and occasional large unidentified cells associated with bone marrow (further away from the growth plate) were also found to contain high levels of grains when hybridized with the TGF-β probe (Fig. 4C,D). Osteoclasts were identified by their location adjacent to bone trabeculae, morphological characteristics and by lack of reactivity with antibodies against factor VIII (which distinguished them from megakaryocytes).

In situ hybridization of growth plate sections with the c-fos probe revealed weak signals to individual cells both on the bony and on the cartilaginous side of the growth plate (Figs 2E, 3G,H). Closer examination of osteogenic zone revealed that the cells containing high levels of c-fos transcripts were osteoclasts (Fig. 4E,F). Examination of the epiphyseal cartilage, where high mRNA levels were seen in Northern hybridization, showed that c-fos expression was particularly prominent in the cells of the peri-chondrium/periosteum and in the chondrocyte layers nearest to the synovial cavity (Figs 2E, 5C,D). Conversely, the levels of type II collagen mRNA in the chondrocytes of the articular surface were very low (Fig. 5A,B).

The growth plates of long bones were studied for relationships between production of growth factors, proto-oncogenes and extracellular matrix components. Northern hybridizations (Fig. 1) revealed that growth plates contain the highest levels of TGF-β mRNA, considerably more than bone which has been suggested to be a major reservoir of the factor in the body. The concentration of TGF-β in bone (about 200 μg kg^-1 tissue) is about 100-fold higher than in other tissues (Seyedin et al. 1985, 1986). The finding that growth plates contain higher levels of TGF-β mRNA than diaphyseal bone probably has a physiological explanation: metabolically active osteoblasts are predominantly located at growth plates. In addition, the RNA from diaphyseal osteoblasts and osteoclasts is ‘diluted’ with bone marrow-derived RNA. Coexpression of TGF-β and type I collagen genes in the same cells (Figs 2–4) is in agreement with earlier findings of the high content of TGF-β in bone (Seyedin et al. 1985, 1986; Ellingsworth et al. 1986), including calvaria (Centrella & Canalis, 1985), and of the capacity of TGF-β to induce synthesis of (type I) collagen (Sporn et al. 1983; Ignotz & Massague, 1986; Roberts et al. 1986; Raghow et al. 1987; Ignotz et al. 1987). Recently, high levels of TGF-β mRNA were detected in bovine osteoblasts, but they were not associated with increased collagen production (Robey et al. 1987). Through its ability to induce the expression of platelet-derived growth factor and the c-fos proto-oncogene in fibroblastic cells (Leof et al. 1986; Mäkelä et al. 1987) TGF-β could stimulate osteoblast proliferation and differentiation, too.

The highest level of type II collagen gene expression was detected in the chondrocytes of the lower proliferative and upper hypertrophic zones of cartilage (Figs 2, 3). TGF-β (CIF-A) has also been shown to have a role in induction of the chondrocyte phenotype, including type II collagen synthesis, in culture (Seyedin et al. 1985, 1986). Although TGF-β mRNA could clearly be localized in some chondrocytes of this zone, the levels were considerably lower than in bone. Similarly, immunohistochemical studies have shown that TGF-β is not associated with the chondrocytes of the growth plate (Ellingsworth et al. 1986). If TGF-β has a role in the final stages of chondrocyte differentiation, it could act via a paracrine route, i.e. diffuse from other sources, possibly bone and perichondrium, to exert its effects on chondrocytes.

Localization of TGF-β mRNA in osteoclasts (Fig. 4C,D) is consistent with its suggested role in bone remodelling: TGF-β has been shown to stimulate bone resorption and prostaglandin E₂ production in organ cultures of neonatal mouse calvaria (Tashjian et al. 1985). TGF-β also induces an inhibitor of plasminogen activator, a response which may be important in the regulation of pericellular proteolysis (Laiho et al. 1986). Modulation of bone resorption by osteotropic hormones has recently been associated with release of TGF-β activity in calvarial bone cultures (Pfeilschifter & Mundy, 1987). The fact that osteoclasts can produce both TGF-β and an acidic environment which is needed to activate this growth factor (Seyedin et al. 1985) makes osteoclasts potentially very important in regulation of matrix production by osteoblasts. Interestingly in situ hybridization revealed that osteoclasts also contain high levels of c-fos transcripts (Fig. 4E,F). Since osteoclasts are thought to represent a class of monocytic cells this finding is in accordance with the observations that differentiating macrophages express the c-fos gene (Mitchell et al. 1985; Adamson, 1987). As osteoclasts are responsible for bone remodelling, deregulated c-fos expression in transgenic mice might have interfered with this aspect of normal development of long bones (Rüther et al. 1987). Some, but not all, osteoblasts were also positive for c-fos mRNA, possibly as a sign of high proliferative activity.

In situ hybridization of epiphyseal cartilage, where high c-fos mRNA levels were seen in Northern hybridization, showed that c-fos expression was particularly prominent in the chondrocytes nearest to the articular surface (Figs 2, 5). This suggests that c-fos is constitutively expressed in these cells. This phenomenon could be related to the development of the synovial cavity or to the differentiation of articular cartilage. After this study was submitted, Dony & Gruss (1987) reported a similar distribution of c-fos expression in articular surfaces of mouse embryos. Interestingly the levels of proaT(II) collagen mRNA were very low in the cells where c-fos was being expressed (Figs 2, 5).

The complex regulation of extracellular matrix production during osseochondral development is likely to be controlled through the action of numerous hormones, growth factors and proto-oncogenes. In the present study, several different expression patterns were observed. TGF-β expression was found to be low in chondrocytes that express the type II collagen gene. Near articular surfaces where type II collagen mRNAs were low, c-fos expression was high. In osteoblasts, production of TGF-β mRNA coincided with active type I collagen synthesis. In perichondrium and periosteum, the levels of proαl(I) collagen, TGF-β and c-fos mRNAs were very high. Finally, osteoclasts involved in bone remodelling contained high levels of TGF-β and c-fos transcripts. These findings suggest that TGF-β and c-fos could very well have functions in both skeletal growth and remodelling during embryogenesis and in some diseases of bone and cartilage. Analysis of normal human development by in situ hybridization can thus bring valuable new information about the temporal and spatial expression of such factors at various stages of bone growth and development. These observations also warrant further studies on the roles of TGF-β and c-fos in postnatal development and in diseases affecting the skeletal system.

We are grateful to Dr Rik Derynck for making the cDNA probe for TGF-β, and to Drs Jean-Francois Gaubet and Rolf Müller for making the c-fos probe available to us, to Dr Jorma Keski-Oja for critical reading of the manuscript and to Dr Heikki Aho for valuable advice regarding identification of osteoclasts. The expert technical assistance of Merja Haapanen and Henna Lehtoranta is gratefully acknowledged. This study was supported by grants from the Medical Research Council of the Finnish Academy, the Sigrid Jusélius Foundation, the Finnish Cultural Foundation, the Varsinais-Suomi Cultural Foundation and the Finnish Cancer Association. A preliminary report of this study has been presented at the European Developmental Biology Meeting in Helsinki, Finland on June 15, 1987.