Contents - Previous - Next

The cell biology of bone growth

1. Introduction
2. The structure and function of bone
3. Skeletal morphogenesis and growth
4. Structure of the growth plate
5. Bone cells
6. Models for the study of skeletal development
7. Regulation of growth plate chondrocytes and bone cells
8. Regulation and mechanisms of cytokine action
Summary
References

J.S. Price, B.O. Oyajobi and R.G.G. Russell

Department of Human Metabolism and Clinical Biochemistry, Sheffield University Medical School, Beech Hill Road, Sheffield S10 1RX, UK

Correspondence to: J.S. Price.

1. Introduction

Growth takes place at the epiphyseal growth plate of long bones by a finely balanced cycle of cartilage growth, matrix formation and calcification of cartilage that acts as a scaffold for bone formation. This sequence of cellular events constitutes endochondral ossification. Another feature of bone growth is a process of modelling, where bone is being continuously resorbed and replaced by new bone. Modelling is most active during childhood and adolescence, and enables long bones to increase in diameter, to change shape and develop a marrow cavity. Modelling continues throughout adult life with bone resorption equally balanced by bone formation in a healthy skeleton, although in the adult the process is referred to as remodelling. An individual's skeletal growth rate and adult limb bone length have an important genetic determinant, but are influenced by many factors including circulating hormones, nutritional intake, mechanical influences and disease. Growth disturbances result when there is disruption of the normal cellular activity of growth plate chondrocytes and/or the cells of bone.

The hormonal regulation of skeletal tissues has been extensively studied. However, there is an increasing body of evidence which demonstrates that factors produced locally in bone and cartilage, or trapped within hard tissue matrix, may play a critical role in regulating normal and pathological skeletal growth and remodelling. Important local mediators are cytokines and growth factors. These are soluble peptides produced by cells that can act in an autocrine, paracrine or endocrine manner. In recent years, a large number of cytokines have been discovered and their molecular structures and biological activities described. A surprisingly large number of cytokines have been shown to affect skeletal tissues, at least experimentally, making this a complex topic, but one which is furthering knowledge of how the cellular events of bone growth may be so precisely regulated.

This paper introduces the mechanisms of skeletal growth, the origin and function of the cells of growth plate and bone, models that can be used to study growth, and ways in which intercellular communication may be regulated by local factors.

2. The structure and function of bone

Bone is an organ composed of cortical and trabecular bone, cartilage, haemopoetic and connective tissues (Ham, 1974). These tissues enable the skeleton to serve its main functions that include the protection of internal organs, movement of parts of the body, and the provision of a site for haematopoesis. The skeleton is also of central importance in mineral homeostasis, bone being the principal reservoir of calcium, phosphorous, sodium, magnesium and carbonate.

Cortical, or compact bone, makes up around 80% of total bone mass, and is most abundant in the shafts of long bones. It has a high mineral content (approximately 70%), and its function is principally mechanical. Spongy or trabecular bone is composed of a lattice of fine bone plates filled with haemopoetic marrow, fat containing marrow, or blood vessels. Located in vertebral bodies, flat bones and in the epiphyses of adult long bones, trabecular bone serves to reduce skeletal weight without compromising strength, and its multiple surfaces are important sites of bone remodelling.

Microscopically, cortical bone tissue is made up of a number of cylindrical units, the osteons, at the centre of which is a Haversian canal containing blood vessels and nerves. This canal is surrounded by up to half a dozen layers of bone, described as lamellae. In cortical bone, osteons have a well defined longitudinal arrangement. Most bone in the adult is described as lamellar bone since the collagen fibres assume an ordered arrangement in thin sheets. Where bone is formed very rapidly, such as occurs pre-natally, in the rapidly growing child, or during fracture repair, the collagen fibrils often assume a very irregular orientation. This is mechanically weak, non-lamellar or woven bone. Woven bone is gradually replaced by mature bone or included within its fabric.

3. Skeletal morphogenesis and growth

The embryonic primordiae of the appendicular skeleton are the limb buds, which are mesodermal structures covered by ectoderm. The first visible outline of the embryonic limb follows a condensation of mesenchymal cells which subsequently differentiate into cartilage cells, the chondrocytes. These cells secrete a matrix and so produce cartilaginous models of the future bones. Surrounding this cartilage is the perichondrium, the outer layer of which becomes a connective tissue sheath while the inner cells remain pluripotential. This cartilage rudiment grows by interstitial and appositional growth, and a vascular system develops to invade the perichondrium. A collar of bone is then laid down around the mid-shaft of the bone. This ossification is a result of the inner perichondrial cells differentiating into bone forming cells, the osteoblasts. At the same time the osteoblasts, together with capillaries, invade the centre of the shaft to form a primary, or diaphyseal ossification centre, at a site where the cartilage cells and matrix have begun to disintegrate. Trabecular bone is then deposited on cartilaginous remnants. The embryonic bone increases in width by appositional growth, and the central cancellous bone core gradually becomes resorbed to form a marrow cavity.

In long bones, another secondary centre of ossification appears at the growing cartilaginous ends, the epiphyseal ossification centre (Fig. 1). (This ossification does not replace the cartilage at the articular end of the model; this remains as articular cartilage.) In addition, a transverse plate of cartilage extends across the bone separating the epiphyseal from the diaphyseal ossification centre. This is the epiphyseal growth plate that persists until an individual stops growing. Growth of cartilage in the epiphyseal plate is continuous, but the plate does not become thickened because on its diaphyseal side the cartilage matures, is calcified, resorbed and replaced by bone. This is endochondral ossification, the mechanism responsible for increasing the length of the bone. In the growing child this is a site of many complex cellular events; namely cartilage growth, maturation, resorption and bone formation. Disturbance of any one of these processes may be reflected in growth retardation.

As an individual's height increases, the bone must increase its diameter, and this is achieved by new bone being laid down by the osteogenic layer of the periosteum. This is intramembranous ossification that does not involve prior cartilage formation. However, the shafts of long bones do not increase in width significantly, as this would increase skeletal mass excessively, because there is resorption of bone on the inner (endosteal) surface by bone resorbing cells, the osteoclasts. This leads to an increase in the size of the marrow cavity with age, and means that the cortical bone of an adult's femur, for example, is not the same bone that existed in childhood. The cycle of bone resorption and formation is bone remodelling, and in the growing skeleton this is often described as 'structural modelling'. Remodelling of bone is a dynamic process that continues throughout life with losses from osteoclastic bone resorption made good by bone formation. Histomorphometric studies of bone have shown that in the remodelling cycle osteoclasts resorb bone surfaces to form an erosion cavity. Mononuclear cells then fill in the cavity, differentiate into osteoblasts and begin to lay down matrix (Eriksen, 1986). It has been estimated that this process can take up to three months with mature osteoblasts secreting matrix for up to 100 days. This balanced process is described as coupling (Frost, 1964), and it is the uncoupling of formation from resorption that leads to skeletal diseases such as osteoporosis where net resorption is greater than formation.

Fig. 1. Schematic diagram of the articular region of a growing long bone illustrating the vascular supply.

4. Structure of the growth plate

The epiphyseal growth plate is made up of three tissue types: the cartilage component divided into distinct zones (Fig. 2), the bony tissue of the metaphysis and the fibrous tissue that surrounds the growth plate. The vascular supply to the growth plate is illustrated in Fig. 1. The secondary ossification centre is supplied by the epiphyseal artery, branches of which end in the proliferating cartilage zone. The metaphysis is supplied mainly by the nutrient artery, with the periphery having an additional supply from metaphyseal vessels (Chung, 1976). Terminal branches of these arteries end in capillary loops below intact cartilage septae that delineate the end of the cartilage zone. These capillaries drain into the large central vein of the diaphysis. Since there are no branches from metaphyseal or epiphyseal arteries to the hypertrophic zone, this region of the growth plate is avascular. Only the proliferative zone has an abundant blood supply.

Fig. 2. Schematic diagram of a longitudinal section through the epiphyseal growth plate. B = bone, OB = osteoblast, CC = calcified cartilage, C = cartilage matrix.

The cartilage matrix is primarily composed of collagens and proteoglycans. These macromolecules play a critical role in the development and maintenance of a variety of functions including tissue strength, architecture, and cell to cell interactions. If abnormal molecules are present in the matrix, it can lose its functional integrity, and the organised arrangement of chondrocytes and their closely regulated proliferation and biosynthesis will be disrupted. Such abnormalities are called dyschondroplasias, and affected individuals suffer from dwarfism. Fortunately the understanding of cartilage matrix molecules has progressed significantly in recent years with the development of techniques enabling improved protein characterisation and localisation, together with knowledge of the gene structure of many matrix molecules. It is now known that genetic defects of a single matrix molecule are the cause of some of these dyschondroplasias.

Type II collagen is the most abundant of the collagens in the growth plate, and since it is found almost exclusively in cartilage it is a specific phenotype marker for chondrocytes. Type II collagen is composed of three identical chains that are wound into the characteristic triple helix of the collagen molecule (Burgeson & Nimni, 1991). Type II collagen molecules form banded fibres seen with the electron microscope and are therefore classified as fibre forming (class I) collagen. In the developing limb and in models of endochondral ossification, type II collagen synthesis can be correlated with chondrogenesis (Dessau et al., 1980; Yu et al., 1991). Type II procollagen may be expressed in two forms, IIA or IIB, due to differential splicing of recently transcribed RNA. In embryonic human vertebral column, type IIB mRNA expression is correlated with cartilage matrix synthesis, whereas IIA is expressed in pre-chondrocytes, the cells surrounding the cartilage (Sandell et al., 1991). Type XI collagen, also a class I collagen, is present in cartilage matrix and is integrated into the interior of type II collagen fibrils (Mayne, 1989). Its function is not known. Type IX collagen is also found in cartilage, but is not a fibre forming collagen since it will not form supramolecular aggregates alone. Type IX is associated with the exterior of the type II collagen molecules (Eyre et al., 1987) and, since it has a single glycosaminoglycan side chain, it is also a proteoglycan.

Type X collagen is a short chain, non-fibril forming collagen with a restricted tissue distribution within the hypertrophic calcifying region of growth plates in fetal and developing bone, where it makes up 45% of total collagen (Kirch & Von der Mark, 1992; Reichenberger et al., 1991). It has been proposed that type X collagen may play a role in regulating mineralisation of cartilage calcification, however, this remains to be proven.

The other main structural component of cartilage is proteoglycan. Proteoglycans are proteins with one or more attached glycosaminoglycan side chains, e.g. chondroitin sulphate, heparan sulphate, dermatan sulphate. These sulphated side chains occupy approximately two thirds of the C terminus region of the molecule, while the other third, the carbohydrate-rich portion, binds to hyaluronic acid (Ruoslahti, 1988). The main proteoglycan of cartilage is aggrecan, a large proteoglycan composed of approximately 90% chondroitin sulphate chains. Aggrecan is found as multi-molecular aggregates composed of many proteoglycan monomers (up to 100) bound to hyaluronan. A small link protein helps to stabilize the aggregate. Synthesis of aggrecan is another specific marker of the chondrocyte phenotype (Doege et al., 1990).

Another important matrix component is the enzyme alkaline phosphatase (ALP). ALP is abundant in matrix vesicles and on the plasma membrane of the maturing chondrocytes, and is required in the calcification process although the precise mechanism of action remains unclear (Vaananen, 1980).

Growth plate chondrocytes are organised into different zones (Fig. 2) with each cell population being part of a different stage of maturation in the endochondral sequence (Brighton, 1978; Kember, 1978). Traditionally studies of growth plate have described cell populations in terms of cell size, shape and contents (Brighton, 1978; Brighton et al., 1983; Ham, 1974; Holtrop, 1972a,b). Functions of these cells used to be speculated upon depending on their morphology. However, improvements in cell and molecular biology techniques now enable growth plate chondrocyte function to be studied where morphology is maintained.

Zone I has otherwise been described as the reserve or resting zone. Cells exist singly or in pairs separated by an abundant extracellular matrix, and have low rates of proliferation (Kember, 1978). Proteoglycan synthesis and type IIB collagen synthesis is low (Schmidt, Rodergerds & Buddecke, 1978; Sandell et al., in press). However, these cells have a high lipid body and vacuole content that has led to the suggestion that this zone is involved with storage for later nutritional requirements (Brighton, 1978). The adoption of the term 'reserve zone' to describe this region may be inappropriate because these cells do not transcribe type IIA collagen, the marker of pre-chondrocytic cells (Sandell et al., in press), evidence that the cells have already differentiated into chondrocytes, i.e., this is not a germinal layer of 'mother cartilage cells'.

Zone II is otherwise described as the upper proliferative or columnar region. The function of the proliferative zones is matrix production and cell division that result in longitudinal growth. Chondrocytes assume a flattened appearance and are arranged in longitudinal columns. The zone is the true germinal layer of the growth plate, with cells actively dividing (Kember, 1978). Type II collagen synthesis and mRNA expression increase in this zone (don der Mark, 1980; Kosher, Kulyk & Gray, 1986), as does that of type XI and aggrecan, although in bovine growth plate type IIB collagen levels are relatively higher (Sandell et al., in press).

Cells of zone III, the lower, more mature region of the proliferating zone, are morphologically no different from those of zone II, but have decreased DNA synthesis (Kember, 1978). Type II collagen synthesis remains high; studies of human fetal growth plate report the highest levels of mRNA for type II collagen in these cells (Sandberg et al., 1988).

Zone IV is the upper hypertrophic zone, where cell size abruptly increases and the columnar arrangement is less regular. Although not proliferating, hypertrophic zone cells retain the full complement of cytoplasmic components, and light microscopy reveals increasing vacuolation of the cells. Hypertrophic chondrocytes are metabolically active cells, with overall matrix synthesis per cell increased approximately three-fold, compared to the proliferative zone (Hunziker, Schenk & Cruz-Orive, 1987). The main matrix components synthesised are types II and X collagen and aggrecan.

Zone V is the zone of the terminal chondrocyte. The end of this zone is marked by the last intact transverse cartilage septum (Fig. 2). Matrix calcification occurs in longitudinal septae between the columns of chondrocytes, and this calcified matrix becomes the scaffolding for bone deposition in the metaphysis. The hypertrophic zone contains the highest levels of alkaline phosphatase (Brighton, 1965). The traditional view was that these cells were metabolically very inactive, and that increasing vacuolation indicated death by hypoxia. However, these cells are clearly actively involved in the synthesis of type X and type II collagen (Kirch & Von der Mark, 1992; Von der Mark & Von der Mark, 1977). Improvements in techniques of growth plate fixation that retain chondrocyte morphology have led to the proposal that a terminal chondrocyte spends most of its life as a fully viable cell indistinguishable from hypertrophic chondrocytes positioned further proximally in the growth plate. The cells then die by apoptosis, a distinct biological form of cell death, lasting approximately 18% of a terminal chondrocyte's life span (Farnum & Wilsman, 1989). Apoptosis may be triggered by the metaphyseal vasculature beyond the last intact cartilage septum.

Zone VI is the junction of the growth plate with the metaphysis, the region where the transition from cartilage to bone occurs. Chondrocyte lysis is evident from empty lacunae invaded by vascular endothelial loops (Ham, 1974; Brighton, 1978). The vascular region of calcified cartilage is the primary spongiosum, upon which osteoblasts lay down unmineralised bone, the osteoid. Metaphyseal bone formation is associated with type I procollagen mRNA expression in the empty lacunae, osteoid, bone and perichondrium (Reichenberger et al., 1991). Type I collagen, a marker of the osteoblast phenotype, is immunolocalised to the same areas, while types II and X collagen have restricted immunolocalisation to calcified cartilage trabecular remnants within spongy bone (Kirch & Von der Mark, 1992; Gannon et al., 1991). Newly formed woven metaphyseal bone is gradually replaced by lamellar bone following osteoclastic degradation of bony matrix and chondroclastic removal of remaining cartilage trabeculae. At the same time external reshaping of the bone is brought about by surface osteoclastic bone resorption and appositional bone formation by periosteally derived osteoblasts.

5. Bone cells

Osteoblasts are the cells responsible for the formation and organization of the extracellular matrix of bone and its subsequent mineralisation. They are derived from mesenchymal precursor cells in marrow that have the potential to differentiate into fat cells, chondrocytes or muscle cells (Owen & Ashton, 1986; Beresford, 1989). The origin of osteoblastic cells in the developing long bones is less well defined. One hypothesis is that osteoblasts are derived from blood-borne elements. This view is supported by evidence that cells in empty lacunae express type I collagen mRNA and are morphologically similar to osteoblasts, but unlike hypertrophic chondrocytes do not express type X collagen mRNA. The alternative view is that osteoblasts are derived from hypertrophic chondrocytes, since type I collagen has been immunolocalised in apparently intact lacunae (don der Mark, 1989). In addition, cultured growth plate chondrocytes will synthesise bone matrix proteins after becoming hypertrophic, although it may be that hypertrophic chondrocytes de-differentiate in vitro.

The principal products of the mature osteoblast are type I collagen (90% of the protein in bone), the bone specific vitamin-K dependent proteins, osteocalcin and matrix Gla protein, the phosphorylated glycoproteins including bone sialoproteins I & II, osteopontin and osteonectin, proteoglycans and alkaline phosphatase. A proportion of osteoblasts become trapped in lacunae within the matrix of bone as osteocytes, connected by a system of canaliculi. These cells may be responsible for intercellular communication; there is evidence that osteocytes regulate the response of bone to the mechanical environment (Skerry et al., 1989). The different cell types present in bone are illustrated in Fig. 3.

Bone resorption. The cell responsible for the resorption of bone matrix is the osteoclast, a large motile, multinucleated cell located on bone surfaces tightly associated with the calcified matrix. There is much evidence supporting the view that osteoclasts are formed by the fusion of mononuclear cells derived from haematopoetic stem cells in marrow. Since these mononuclear cells have some features of macrophages, it is proposed that they and osteoclasts have a common precursor (Hagenaars et al., 1989; Mundy & Roodman, 1987). The erstwhile thesis that osteoclasts and osteoblasts are derived from different precursors has been challenged by a recent report proposing that stromal cells and haemopoetic cells have a common ancestry: a single multi-potential stem cell in marrow (Huang & Tertappen, 1992).

Osteoclasts are polarised cells, having a ruffled border region of the cell membrane that is surrounded by an organelle-free region, or 'clear zone', and they adhere to the bone surface via integrins, which are specialised cell surface receptors (Vaes, 1988). Osteoclastic bone resorption initially involves mineral dissolution, followed by degradation of the organic phase. These processes take place beneath the ruffled border and depend on lysosomal enzyme secretion and an acid microenvironment (Baron, 1989). A pH gradient across the ruffled membrane is the consequence of active transport mechanisms such as Na⁺/H⁺ exchange, ATP-dependent proton pumps, and the enzyme carbonic anhydrase (Baron et al., 1986; Blair et al., 1989; Sly et al., 1983). Osteoclasts actively synthesise lysosomal enzymes, in particular the tartrate resistant isoenzyme of acid phosphatase (TRAP) (used as a marker of the osteoclast phenotype), and cysteine-proteinases such as the cathepsins that are capable of degrading collagen. Lysosomal enzymes are only released at the ruffled border region of the osteoclast cell membrane (Baron, 1989). Other cells in bone, in particular the osteoblast, may be involved in degrading the organic non-mineralised phase from bone surfaces. In vitro studies have shown that removal of non-mineralised organic matrix is necessary before mineralised matrix may be resorbed by isolated osteoclasts (Chambers & Fuller, 1985).

Systemic agents, important in regulating osteoclastic bone resorption, are parathyroid hormone (PTH), 1,25 di-hydroxy vitamin D₃[1,25(OH)₂D₃] and calcitonin. PTH and 1,25(OH)₂D₃ are unable to stimulate osteoclastic bone resorption in vitro in the absence of osteoblastic cells (McSheehy & Chambers, 1986). This gave rise to the idea that these agents stimulate osteoclasts to resorb bone via a 'coupling' factor. Osteoclasts do not have receptors for 1,25(OH)₂D₃ (Merke et al., 1986), and until recently were not believed to have PTH receptors, although the functional significance of PTH receptors on osteoclasts remains to be established (Agarwal & Gay, 1992). Osteoclasts have calcitonin receptors (Lin et al., 1991), and this inhibitor of bone resorption acts directly on the osteoclast to reduce cellular motility, retract cytoplasmic extensions and reduce ruffled border size.

Glucocorticoids (GCs) are another class of systemic agents that cause bone loss. Although this may be due to their inhibition of intestinal calcium absorption and the induction of secondary hyperparathyroidism, GCs also have direct actions on bone cells. These direct effects on bone are believed to be via the local regulation of cytokine and prostaglandin production (Peck et al., 1984). Prostaglandins are locally produced by most cells in the body, and have been shown to have direct effects on osteoclasts and their precursors, inhibiting bone resorption by mature osteoclasts and increasing the formation of their precursors (Chenu et al., 1987; Chambers et al., 1985).

Fig. 3. Schematic diagram of a developing long hone illustrating the cells of bone.

6. Models for the study of skeletal development

Techniques used to study bone growth and development in children and adolescents are restricted to non-invasive methods; however, the study of endochondral ossification and bone remodelling requires the use of appropriate experimental models (Table 1). Studies of human skeletal development are possible in the embryo and young fetus, and provide useful data on gene expression and protein synthesis at this early stage (Kirch & Von der Mark, 1992; Sandberg & Vuorio, 1987). Much information on bone and cartilage cell interactions and the regulation of gene expression has come from in vitro studies. Primary cell lines derived from rodent and human bone may be successfully cultured, as can growth plate chondrocytes from a variety of species (Braidman et al., 1983; Beresford, 1983). Osteoclasts are more difficult cells to study in vitro, since it is difficult to obtain sufficient numbers of pure cells, and there are no available cell lines. However, the development of long-term marrow cultures to study the formation of osteoclasts has proved invaluable in providing information on their origin and development (MacDonald et al., 1987). Widespread use is made of cell lines derived from osteoblastic tumours, although non-transformed cells reflect the way cells function in vivo more closely. One disadvantage of the in vitro approach is that observed responses may not reflect the complex interplay of systemic and local factors that are involved in vivo.

Table 1. Models for the study of skeletal development

The various events of very early limb differentiation in the embryo have been extensively studied in the chick limb bud and in the frog, and in our laboratory we have become interested in a model of naturally occurring bone regeneration in the adult, deer antler growth. The rodent, being the most inexpensive and readily available laboratory animal, has been used most extensively for pre-and post-natal studies of bone growth and remodelling. Results from studies in the rat may, however, be difficult to relate directly to human growth as in the rat growth plate closure does not occur. Likewise, studies of cultured rat bone cells often yield different results to comparable studies on human bone cells, e.g. in the case of the insulin-like growth factors (IGFs) rat osteoblasts synthesise predominantly IGF-I, whereas human osteoblasts synthesise far more IGF-II than IGF-I (Mohan & Baylink, 1991).

Two in vivo models that have yielded much information on the events taking place during endochondral ossification are the subcutaneous implantation of demineralised bone matrix (Urist, 1965), and the formation of bone and cartilage from precursor cells in intraperitoneal diffusion chambers (Owen & Ashton, 1986). Injection of demineralised bone matrix at an ectopic site leads to formation of a marrow containing bony ossicle following the formation of a cartilage matrix. The osteoinductive agents in demineralised bone are now known to be the bone morphogenetic proteins (BMPs, Wozney, 1989). In recent years molecular biology has led to the exciting field of genetic manipulation, where genes believed to be important in regulating skeletal development may be selectively deleted or over-expressed.

Contents - Previous - Next