Introduction
Normal bone development and linear growth depends on the co-ordinated contributions of various genetic, environmental, endocrine and nutritional factors and continues until closure of the epiphyseal growth plate following puberty.

Thyroid hormone (T3) is essential for normal skeletal development. Childhood hypothyroidism results in growth arrest, delayed bone age, epiphyseal dysgenesis and short stature (1-3). Thyroid hormone replacement induces catch-up growth, though maximum predicted height may not be reached and any height deficit correlates with the duration of untreated hypothyroidism (3;4).

Untreated childhood thyrotoxicosis causes accelerated growth and advanced bone age with premature closure of the growth plate and short stature (5). Resistance to thyroid hormone (RTH) is associated with a variable phenotype and, although only a few patients have been investigated in detail with regard to skeletal phenotype, features including advanced bone age, delayed bone age, short stature, craniofacial abnormalities and fractures have all been described (6;7).

The molecular mechanisms of action of thyroid hormone within developing bone are currently incompletely understood. This short paper will describe the normal epiphyseal growth plate and briefly summarize available information on the role of thyroid hormone in ensuring normal skeletal development.

Bone formation and the normal growth plate
Bone formation begins when mesenchymal cells form condensations or clusters of cells (8). These cells either differentiate directly into bone forming osteoblasts (intramembranous ossification of flat bones cf. skull, scapula, ileum) or into chondrocytes that lay down a cartilage mould that is subsequently replaced by ossified bone (endochondral ossification of long bones cf. tibia, humerus, femur).

The epiphyses and metaphyses of long bones originate from separate ossification centres that are separated by a growth plate (Figure).

Figure: View the figure

Within the normal growth plate, chondrocytes are organised into layers. At the epiphyseal end chondroblast progenitor cells occur singly and in small clusters to form the reserve zone. Progressing towards the metaphysis flattened chondroblasts undergo clonal expansion forming discrete columns that constitute the proliferative zone. These proliferative cells then lose their ability to divide and differentiate to form hypertrophic chondrocytes, which secrete type X collagen. Chondrocytes of the hypertrophic zone enlarge and finally undergo apoptosis to leave a cartilaginous framework that forms a scaffold for invading osteoblasts to lay down newly mineralised bone within the primary spongiosum. These osteoblasts are derived form bone-marrow stromal cells that enter the primary spongiosum via capillaries that advance from the adjacent bone marrow cavity.

Thyroid hormone receptors in the developing bone growth plate
Thyroid hormones act via two thyroid hormone receptors (TRs), TR and TR, which act as hormone inducible transcription factors (9). Each TR is expressed as multiple isoforms. TRs are normally expressed by chondrocytes of the reserve, proliferative and prehypertrophic zones, as well as by osteoblasts of the primary spongiosum (Figure). Differentiated hypertrophic chondrocytes do not express T3 receptors suggesting that reserve zone and proliferating chondrocytes are primary T3 target cells but that differentiated chondrocytes are unresponsive to T3 (10;11).

The hypothyroid growth plate
The tibial growth plate of thyroparathyroidectomized rats is disorganized with reductions in the numbers of reserve zone chondrocytes, proliferating chondrocytes and bone marrow stromal cells (11;12). Proliferating chondrocytes fail to form discrete columns and the hypertrophic zone is diminished in width and morphologically indistinct. Expression of collagen X, a specific marker of hypertrophic chondrocyte differentiation, is undetectable in the hypothyroid growth plate, indicating that hypertrophic chondrocyte differentiation is severely impaired. The growth plate is separated from the primary spongiosum by a mineralised interface, essentially sealing off the growth plate from vascular invasion and preventing further bone lengthening, leading to growth retardation. There is disruption of the normal functional continuity between maturing chondrocytes and mineralizing osteoblasts with markedly reduced osteoblast invasion and fewer, thinner bone trabeculae. An increased concentration of TR-expressing mast cells is also found in the bone marrow adjacent to this abnormal growth plate (13).

The growth plates of hypothyroid rats also have abnormal cartilage matrix deposition. Normal cartilage matrix is composed of proteoglycans containing chondroitin and heparan sulfates and hyaluronic acid residues. In hypothyroid rats, critical electrolyte staining studies using alcian blue have revealed an abnormal increase in sulfation of heparan sulfate proteoglycans in proliferating chondrocytes. This abnormal matrix is deposited in a patchy irregular fashion suggesting that thyroid hormones influence extra-cellular matrix biology as well as cellular activity of the growth plate. Treatment of these rats with thyroid hormone reverses these changes and studies have shown that this is through the direct actions of T3 on bone, and is not growth hormone (GH) mediated (12).

Tibial dyschondroplasia (TD), a disorder of broiler chickens, associated with avascular non-mineralized cartilage extending from the epiphyseal growth plate, results from the inability of proliferating chondrocytes to undergo terminal differentiaton to hypertrophic chondrocytes. This disorder has been shown to be associated with a markedly reduced expression of iodothyronine deiodinase tyoe 2 (DIO2) in the growth plate (14). DIO2, by catalysing the conversion of T4 to T3, is the key enzyme that determines the availability of T3 to target cells, including growth plate chondrocytes. The fact that circulating thyroid hormone levels (primarily determined by activity of the DIO1 enzyme) are normal in these chickens suggests that growth plate specific hypothyroidism, caused by reduced conversion of T4 to T3 by DIO2, is a major aetiological factor in the development of TD and provides further evidence of the essential role of T3 in terminal chondrocyte differentiation. Activity of DIO2 has also been suggested in neonatal rat tibia cultures and a mouse chondrogenic cell line (ATDC5) (15). Higher levels of DIO2 activity at the later stages of development in the organ-cultured tibias again support the hypothesis that local T3 production contributes to hypertrophic chondrocyte differentiation.

Thus T3 is necessary to stimulate resting zone cells to differentiate to proliferating chondrocytes, for chondrocyte hypertrophy and differentiation and for vascular invasion of the growth plate.

The Ihh/PTHrP signalling pathway in the bone growth plate
Indian hedgehog (Ihh) is a member of the hedgehog family of secreted ligands and is a master regulator of bone development. Ihh is synthesized by prehypertrophic and hypertrophic chondrocytes (8;16).

Ihh stimulates production of parathyroid hormone-related peptide (PTHrP) from cells at the periarticular ends of bones (Figure). PTHrP acts on the PTH/PTHrP receptor (PPR) to keep proliferating chondrocytes in the proliferative pool. When the source of PTHrP is sufficiently distant the chondrocytes are no longer stimulated by PTHrP, they stop proliferating and start to synthesize Ihh. In addition, Ihh stimulates chondrocyte proliferation directly and also controls the differentiation of osteoblasts from perichondrial cells during the formation of the bone collar.

Thus interactions between Ihh and PTHrP determine the lengths of proliferating columns of chondrocytes in the growth plate and hence the pace of bone growth.

Changes in the Ihh/PTHrP signalling pathway in hypo- and hyperthyroidism
The distribution of Ihh mRNA within the proliferative, prehypertrophic and hypertrophic zones of the growth plate is similar in euthyroid and thyrotoxic rats. In contrast, in hypothyroid animals Ihh is mainly located within the upper regions of the proliferative zone and the reserve zone (11).

PTHrP mRNA expression is also altered in the hypothyroid growth plate. Levels of expression are increased and include expression by chondrocytes extending throughout the proliferative and reserve zones. In euthyroid, thyrotoxic and hypothyroid-T4 treated animals PTHrP mRNA expression is restricted to a discrete layer of prehypertrophic and hypertrophic chondrocytes.

PTH/PTHrP receptor (PPR) is also altered by thyroid status. It is expressed throughout all zones of the growth plate in euthyroid and hypothyroid animals, but is completely absent in the thyrotoxic growth plate and restricted to proliferative and prehypertrophic chondrocytes in hypothyroid-T4 treated rats.

Thyroid hormone has been shown to stimulate terminal differentiation of growth plate chondrocytes by down regulation of Sox9, a transcription factor present in cells of mesenchymal condensations and proliferating chondrocytes but not in hypertrophic chondrocytes (17). This terminal differentiation process is associated with expression of cyclin-dependant kinase inhibitors known to regulate the cell cycle checkpoint (18).

This data strongly supports a role for thyroid hormone in regulating components of the Ihh/PTHrP feedback loop in the growth plate and thus the pace of chondrocyte differentiation and bone growth.

Fibroblast Growth Factor Receptor-1 (FGFR1) as a T3-target gene in bone
FGFRs are membrane tyrosine kinase receptors that are widely expressed during embryogenesis (19;20). Three FGFRs (1-3) are known to be essential for skeletal development. Mutations of all three FGFRs can cause premature fusion of skull sutures and other variable bony abnormalities, while an activating mutation of FGFR3 is the cause of achondroplasia, the most common genetic form of dwarfism (20).

FGFR1 has been identified as a T3-target gene in osteoblasts (21). T3 acting via the thyroid hormone receptor- (TR) enhances FGF stimulation of FGFR1 activity. TR^0/0 mice, that display a hypothyroid phenotype of delayed endochondral ossification, have abnormalities of cartilage matrix similar to those described above, namely an increase in heparan sulfate proteoglycans (22). It is known that heparan sulfate is required for binding of FGF to FGFR and for ligand-induced receptor activity (23). Therefore T3-regulated production of heparan sulfate, or modification of its structure, might be the mechanism by which T3 regulates FGFR1 signalling.

Thyroid status and mast cells in the bone marrow
As mentioned above, an increased concentration of TR-expressing mast cells is also found in the bone marrow adjacent to the abnormal growth plate in hypothyroid rats, suggesting a possible involvement of mast cells in thyroid hormone dependent endochondral bone formation (13). Mast cells and chondrocytes can communicate with one another and mast cells can regulate chondrocyte proteoglycan production and matrix deposition. Mast cells are also implicated in matrix degradation, angiogenesis, the release of bound growth factors from extracellular matrix stores and FGF signalling (24;25). A number of matrix degrading enzymes are induced by T3 and thyroid hormone is necessary for the proteoglycan degradation that occurs before endochondral ossification (26;27). Thus mast cells and thyroid hormones have the opportunity to interact on several important processes to regulate endochondral ossification and skeletal development, though the precise mechanisms for these likely interactions remain poorly understood.

Angiogenesis of the developing growth plate and thyroid hormone
Vascular invasion of the growth plate is essential for normal longitudinal bone growth (28). Angiogenesis is intimately linked to chondrocyte hypertrophy and when hypertrophy is inhibited angiogenesis and subsequent endochondral ossification is blocked (29). Thus hypertrophic, but not resting or proliferating chondrocytes, express vascular endothelial growth factor (VEGF), a key regulator of angiogenesis (28). T3 is also known to be required for normal endochondral angiogenesis, although the molecular mechanisms involved are not yet understood (11;12). This may occur via modulation of the Ihh-PTHrP loop to control the pace of chondrocyte differentiation, via alterations in FGF signalling or by T3-mediated degradation of the cartilage matrix to release sequestered FGFs and VEGFs.

The role of thyroid hormone in skeletal development in genetically modified mice
In order to further our understanding of the role of T3 in skeletal development a number of mouse models in which components of the T3 signalling pathway in bone have been altered have been generated. These include the Pax8^-/- mouse, a genetic model of congenital hypothyroidism due to thyroid gland agenesis, the TR^PV/PV mouse, a model of human RTH, a thyroid stimulating hormone receptor (TSH) knockout mouse and a number of mice with knockouts of one or more TR isoforms (30-35). The Table summarises some of these genetic models of thyroid hormone deficiency, excess and resistance in terms of effects on skeletal development and growth.

Table: Genetically modified mouse models of altered thyroid hormone signalling


Pax8^-/- mice have no thyroid and thus no circulating thyroid hormone (30;31). They have a severely abnormal skeletal phenotype with growth retardation and die at weaning unless rescued by replacement with thyroid hormone. This phenotype is more severe than that seen in double null TR^0/0TR^-/- mice, which have no thyroid hormone receptors, suggesting that a total lack of thyroid hormone, resulting in persistent expression of unliganded or apo-TRs, is more detrimental to skeletal development than a complete deficiency of TRs (22). Interestingly, the skeletal phenotype of Pax8^-/- mice can be partially rescued by crossing them with TR0/0 mice, which lack only TR receptors (but not TR ^-/- mice, which lack only TR receptors), indicating that apo-TR is responsible for the potent detrimental effects on bone development seen in congenital hypothyroidism (30).

The predominant TR mRNA expressed in bone is TR which is present in 12-fold greater concentrations than TRß (32). Although they have normal thyroid hormone, GH and IGF-1 levels, TR^0/0 mice have a skeletal phenotype that includes failure of hypertrophic differentiation, impaired mineralization, delayed endochondral bone formation and growth retardation (22). In contrast, skeletal development in TRß^-/- mice appears to be normal (36;37).

However, TR^PV/PV mutant mice, which have a point mutation in TR and autosomal dominant thyroid hormone resistance with very high circulating thyroid hormone levels have pronounced skeletal abnormalities (32). These include accelerated growth in utero with premature ossification. In the postnatal period, growth rate slows, bone mineralization increases as the skeleton matures early and the growth plate becomes quiescent leading to shortened limb length and ultimately growth retardation. Such a phenotype is similar to that seen in childhood thyrotoxicosis and indicates that the skeleton of TR^PV/PV mice is thyrotoxic. FGFR1 mRNA expression has also been shown to be increased in the perichondrial region, growth plate and osteoblasts of TR^PV/PV mice, further evidence to strengthen the hypotheses that thyroid status regulates FGFR1 signalling in bone and that the skeleton in TR^PV/PV mice is thyrotoxic (32).

The thyrotoxic skeletal phenotype of the TR^PV/PV mice can be explained as follows. Bone is a TR sensitive organ, whereas the pituitary is a TR sensitive tissue. Mutations in TR cause resistance to the normal negative feedback effects of thyroid hormone on the pituitary, resulting in elevated circulating thyroid hormone levels. These act on TR, which is predominant in bone, to generate a TR-mediated thyrotoxic skeletal phenotype. In contrast, TR1^PV/PV mice, which harbour a similar mutation in TR1, have severe growth retardation and a hypothyroid phenotype, providing further support that TR is the predominant functional TR isoform in bone.

Recently a TSHR^-/- mouse has been generated (34). Homozygous mice were hypothyroid, underweight, growth retarded and died by 10 weeks of age. Thyroid hormone replacement at weaning normalized body weight but did not correct bone length, bone weight or bone mineral density. It was suggested therefore that TSH acts as a direct inhibitor of bone turnover. However, two alternative possibilities need to be considered. Firstly, it is possible that the mice in this study, which were analysed at seven weeks of age, had not had sufficient time following hormone replacement for catch-up growth to occur. Alternatively, untreated hypothyroidism in utero or in the first few weeks prior to weaning and hormone replacement may result in permanent irreversible effects on skeletal development and bone mineral density. At this stage it is not clear whether the skeletal effects of TSH are direct and independent of thyroid hormone or whether the influence of hypothyroidism during development irreparably impairs skeletal function in later life and the effects of thyroid hormone on bone predominate over the actions of TSH.

The future: many unanswered questions, scope for further investigation
In summary, T3 effects on the developing skeleton are complex. T3 is essential for normal skeletal development, with roles in chondrocyte differentiation, hypertrophy and angiogenesis. The skeletal response to T3 is accelerated in thyrotoxicosis and decelerated in hypothyroidism, indicating the exquisite sensitivity of the growth plate to T3 and the necessity for euthyroidism to ensure linear growth progresses normally.

There are many unanswered questions about the signals that influence endochondral bone formation. The full details of how T3 interacts with these multiple pathways and switches from the primarily anabolic role it adopts in utero and childhood to its predominantly resorptive actions seen in adulthood remains unclear. Do the actions of T3 in fetal life, childhood and early adulthood influence the development of peak bone mass? If so, could subtle genetic variations in the components of this pathway predispose to later osteoporosis? Are there sex specific differences in the role of thyroid hormone in fetal life? Newborn boys with congenital hypothyroidism are twice as likely as girls to have absent knee epiphyses at birth, suggesting that the impact of thyroid hormone on fetal skeletal maturation is gender specific (38). What are the potential mechanisms for this?

We do not yet know the physiological function of the apo-TR in bone. Nor do we have a clear understanding of the role of TR in bone – in mice at least TR does not appear to play a primary role in normal skeletal development although there is evidence that it can partially compensate for the absence of TR. We also do not know whether TR and TR are co-expressed in the same cells. The availability of genetically modified mice together with the development of selective TR agonists provides us with tools to investigate these questions further. GC-1 is a synthetic thyroid hormone analogue selective for binding and activation of TR. Compared with the bone loss seen in T3-treated thyrotoxic rats, rats treated with GC-1, exhibit predictable thyrotoxic effects on organs such as the liver and pituitary but preserve their bone mass, further confirmation that TR is the predominant isoform in bone and potentially a very useful mechanism for examining receptor/organ selective effects of T3 (39;40).

We know that liganded-TRs act as hormonal inducible transcription factors. FGFR1 has recently been identified as a T3 target gene in bone. What are the as yet other unidentified target genes? What are the molecular mechanisms that lead to their modulation by T3?

In addition to the thyroid hormone receptors, receptors for growth hormone (GH), insulin like growth factor-1 (IGF-1), and glucocorticoid (GC) are also expressed by growth plate chondrocytes (41-45). T3 influences expression of several components of GH/IGF-1 signalling in bone and can regulate hepatic 11ß-hydroxysteroid dehydrogenase, the enzyme that is responsible for maintaining circulating concentrations of GC (46). In return GC are known to regulate thyroid hormone deiodinase activity in renal tubular cells, and as the deiodinases are known to be expressed in the growth plate, this may be a mechanism whereby GC control local levels of available T3 (15;47). Understanding the mechanisms behind the interactions between the T3 signalling pathway and the systemic and paracrine effects of GH/IGF-1 and GC will be important in teasing out the molecular biology of thyroid hormone-dependent skeletal development.

Our whole outlook on the role of thyroid hormones in bone has recently been challenged by the suggestion that T3 may not be responsible for all the actions of thyroid hormones in bone and that TSH itself may have direct effects on skeletal turnover. Clarification of the relative roles of TSH and T3 in bone will provide a better understanding of bone turnover changes associated with hypo- and hyperthyroidism and the relative contributions of TSH and T3 to skeletal development.

The last few years have seen an increasing interest in the molecular actions of T3 within bone and its influence on skeletal growth and homeostasis, and this remains an emerging and exciting area of clinical and scientific research.

Legend Figure The euthyroid and hypothyroid growth plates, with a summary of T3 actions.





The anatomical features of the upper region of a long bone (a) are shown together with an expanded view of the euthyroid growth plate (b) and an outline of changes which occur in the hypothyroid growth plate (c). Reserve cells undergo clonal expansion to form columns of proliferating chondrocytes. Prehypertrophic chondrocytes differentiate into hypertrophic chondrocytes, which enlarge and finally undergo apoptosis. The resulting lacunae form the scaffold for new trabecular bone formation. Vascular invasion facilitates the migration of osteoblasts.

Also outlined is the Ihh/PTHrP negative feedback loop. PTHrP is secreted from perichondrial cells and cells at the end of long bones (1). PTHrP acts on proliferating chondrocytes to keep them in the proliferative pool and thereby delay the production of Ihh. When the source of PTHrP is sufficiently distant then Ihh is expressed. Ihh acts on chondrocytes to increase their rate of proliferation (2) and stimulate the production of PTHrP (3). Ihh also acts on perichondrial cells to convert them into the osteoblasts of the bone collar (4).

The location of various TR isoforms within bone and a summary of T3 actions in bone are also given.