Thyroid Hormone Signaling in the Development of the Endochondral Skeleton

Abstract

Thyroid hormone (TH) is an established regulator of skeletal growth and maintenance both in clinical studies and in laboratory models. The clinical consequences of altered thyroid status on the skeleton during development and in adulthood are well known, and genetic mouse models in which elements of the TH signaling axis have been manipulated illuminate the mechanisms which underlie TH regulation of the skeleton. TH is involved in the regulation of the balance between proliferation and differentiation in several skeletal cell types including chondrocytes, osteoblasts, and osteoclasts. The effects of TH are mediated primarily via the thyroid hormone receptors (TRs) α and β, ligand-inducible nuclear receptors which act as transcription factors to regulate target gene expression. Both TRα and TRβ signaling are important for different stages of skeletal development. The molecular mechanisms of TH action in bone are complex and include interaction with a number of growth factor signaling pathways. This review provides an overview of the regulation and mechanisms of TH action in bone, focusing particularly on the role of TH in endochondral bone formation during postnatal growth.

1. INTRODUCTION

Thyroid hormones (TH) exert specific effects on growth, development, and metabolism in a variety of tissues in humans as well as experimental animal models. The presence of TH is required during development of the skeleton, linear bone growth, maintenance of bone mass, and efficient healing of fractures (Bassett & Williams, 2016). Congenital hypothyroidism, the most common congenital endocrine disorder, has an increasing incidence which has recently risen as high as 1 in 1400 (Albert et al., 2012; Bassett & Williams, 2016; Corbetta et al., 2009; Deladoëy, Ruel, Giguère, & Van Vliet, 2011; Mengreli et al., 2010; Mitchell, Hsu, Sahai, & Massachusetts Pediatric Endocrine Work Group, 2011; Wassner & Brown, 2015). Congenital and acquired juvenile hypothyroidism are known causes of short stature due to arrested bone growth, development, and mineralization; left untreated, these patients can develop a skeletal dysplasia which includes coarse facial features, truncal shortening, scoliosis, patent fontanelles, vertebral deformities, and absent femoral ossification centers (Hüffmeier, Tietze, & Rauch, 2007). However, therapeutic TH replacement in these cases can prevent dysplasia and result in corrected growth and final height (Delvecchio et al., 2015; Rivkees, Bode, & Crawford, 1988). Furthermore, juvenile thyrotoxicosis, most commonly caused by Graves’ disease, leads to accelerated bone formation and early closure of the growth plates, thus resulting in short stature (Bassett & Williams, 2016; Segni, Leonardi, Mazzoncini, Pucarelli, & Pasquino, 1999). Evidence for the importance of TH for skeletal development is strong, but the molecular mechanisms underlying TH action in bone are not fully understood. This review discusses the regulation and mechanisms of TH action in bone, focusing especially on recently uncovered developments in our understanding of the cellular and molecular mechanisms of TH regulation of endochondral bone formation during postnatal growth.

2. TH PHYSIOLOGY

2.1. Regulation and Production

The hypothalamus–pituitary–thyroid (HPT) axis is responsible for maintaining systemic serum TH levels via a classical negative feedback loop (Fig. 1). The paraventricular nucleus of the hypothalamus produces thyrotropin-releasing hormone (TRH). Through the hypothalamic–pituitary portal circulation, TRH travels to the anterior pituitary gland and triggers thyrotrophs to release thyroid-stimulating hormone (TSH). The TSH receptor (TSHR) on thyroid follicular cells responds to TSH, stimulating cellular proliferation and the synthesis and secretion of 3,5,3′,5′-l-tetraiodothyronine (thyroxine, T4), the primary hormone secreted by the thyroid gland, and 3,5,3′-l-triiodothyronine (T3), the more active form of the hormone which is primarily produced by peripheral conversion of T4 to T3 (Fig. 2; Bassett & Williams, 2016; Kopp, 2001). Once secreted into the circulation, the vast majority of T3 and T4 are bound to transport proteins such as thyroid hormone-binding globulin, transthyretin, and albumin; only ~0.2% of serum T3 and ~0.02% of serum T4 are free and unbound in the circulation, available to act on tissues. Serum T3 feeds back upon the hypothalamus and pituitary gland, signaling via TH receptor β (TRβ) to inhibit TRH and TSH synthesis and release in order to maintain serum TH levels within a physiological range (Abel, Ahima, Boers, Elmquist, & Wondisford, 2001; Forrest et al., 1996; Macchia et al., 2001; Nikrodhanond et al., 2006). Thus, the inverse relationship between TSH and TH levels which defines the HPT axis set point is maintained by a negative feedback loop (Andersen, Bruun, Pedersen, & Laurberg, 2003; Bassett & Williams, 2008).

Fig. 1.

Negative feedback establishes the HPT axis set point. Thyrotropin-releasing hormone (TRH) is secreted by the hypothalamus. Via the hypothalamic–pituitary portal circulation, TRH signals anterior pituitary thyrotrophs to release thyroid-stimulating hormone (TSH). TSH stimulates the thyroid gland to produce and release T3 and T4, which travel through the circulation to exert effects on peripheral tissues such as bone. T3 exerts negative feedback via TRβ2 to the hypothalamus and pituitary, inhibiting release of TRH and TSH and stabilizing serum TH levels.

Fig. 2.

T4 (3,5,3′,5′-l-tetraiodothyronine, thyroxine) is the major form of TH found in circulation. In peripheral tissues, T4 is converted to T3 (3,5,3′s-l-triiodothyronine), the TH with higher affinity for TH receptors (TRs), by deiodinases. Two genes, THRA and THRB, encode the TRs TRα and TRβ, each of which produces multiple splice variants. The differential expression of the various forms of TR mediates the tissue-specific effects of TH.

Although T3 has a much higher affinity for the thyroid hormone receptors (TRs) than T4 and is therefore classically considered the biologically active form of TH, the fact that T4 exists in much higher circulating concentrations has prompted investigation into whether T4 itself may have biological actions. Indeed, recent studies of T4, T3, and their metabolites have revealed that, at circulating concentrations and in the absence of deiodinases, T3 and T4 were equally able to induce osteoblast differentiation (Cheng, Xing, Pourteymoor, & Mohan, 2016). Further exploration of possible direct T4 effects on specific cell types such as osteoblasts could provide therapeutic targets that can affect the skeleton without off-target T3 effects on other tissues.

2.2. TH Transport and Metabolism

Conversion of T4 to T3 occurs in peripheral tissues via the action of deiodinases (Bianco & Kim, 2006; Lin, Fukuda, & Cheng, 1990). Type 1 deiodinase (D1) can activate or inactivate TH, catalyzing the removal of 5- or 5′-iodine atoms to form three possible products: T4 can be converted to either active T3 or inactive 3,3′,5′-triiodothyronine (reverse T3, rT3), and both T3 and rT3 can be converted to inactive 3,3′-diiodothyronine (T2). The majority of circulating T3 is produced by conversion of T4 to T3 by D1. Type 2 deiodinase (D2) primarily converts T4 to active T3, providing another source of circulating T3 and regulating intracellular T3 concentrations. Type 3 deiodinase (D3), however, is primarily responsible for inactivation of THs, irreversibly converting T3 and T4 to inactive T2 and rT3, respectively (Gereben et al., 2008; Maia, Kim, Huang, Harney, & Larsen, 2005; St Germain, Galton, & Hernandez, 2009). D1 is not expressed in the skeleton or cartilage (Gouveia et al., 2005; LeBron, Pekary, Mirell, Hahn, & Hershman, 1989; Waung, Bassett, & Williams, 2012; Williams et al., 2008), but D2 is expressed in osteoblasts (Bassett et al., 2010; Williams et al., 2008), and D3 is expressed in cells of all skeletal lineages (Capelo et al., 2008; Williams et al., 2008). The specific roles of deiodinases in cartilage and bone development in vivo remain to be determined.

While THs are relatively lipophilic and thus can passively diffuse across cell membranes, there are also several specific membrane transporter proteins which actively regulate TH transport into and out of cells. These membrane transporters include the monocarboxylate transporters 8 and 10 (MCT8 and MCT10), sodium-dependent organic anion cotransporting polypeptide 1 (OATP1), the sodium taurocholate cotransporting polypeptide (NTCP), and the l-type amino acid transporters 1 and 2 (LAT1 and LAT2) (Friesema, Jansen, Milici, & Visser, 2005; Heuer & Visser, 2009; Visser, 2013; Visser, Friesema, & Visser, 2011). Studies have revealed that MCT8, LAT1, and LAT2 are expressed in mouse skeletal tissues (Capelo et al., 2008), suggesting that the levels of active T3 available for intracellular signaling via the nuclear TRs in bone depend upon both the relative activities of activating D2 and inactivating D3 and the expression and efficiency of TH membrane transport proteins.

2.3. TH Receptors and Actions

The primary mechanism by which THs exert their actions is via the nuclear TRs, ligand-inducible transcription factors which regulate expression of target genes (Fig. 2). Two types of TR have been identified: TRα and TRβ. The mammalian THRA gene produces three splice variants of TRα in addition to two truncated TRα proteins (TRΔα1 and TRΔα2). Of these TRα splice variants, only TRα1 binds both T3 and DNA to regulate target gene expression. TRα2 and TRα3 do not bind T3; they act as antagonists of TH signaling in vivo (Fraichard et al., 1997; Harvey & Williams, 2002; Koenig et al., 1989; Plateroti et al., 2001; Williams, 2000). TRΔα1 and TRΔα2 have been shown to play a role in intestinal development (Chassande et al., 1997; Plateroti et al., 2001), but their role in other tissues including bone remains to be elucidated. The THRB gene produces two splice variants, TRβ1 and TRβ2, both of which bind T3 and act as functional receptors. Transcripts of TRβ3 and TRΔβ3 have been detected, but any possible physiological action is unknown (Harvey, Bassett, Maruvada, Yen, & Williams, 2007; Williams, 2000).

TRα1 and TRβ1 are widely expressed across all tissue types; however, their relative abundance and functions depend upon stage of development and tissue type, with the result that different tissues are more or less responsive to a specific TR (Forrest, Sjöberg, & Vennström, 1990). TRα1, for example, is highly expressed in the heart, the brain, and bone, while TRβ1 exhibits high expression in the liver and the pituitary gland (Cheng, 2005). TRβ2, however, is restricted in expression to a few tissues; in the hypothalamus and pituitary gland, TRβ2 mediates the negative feedback inhibition of TRH and TSH to regulate the HPT axis, and in the retina and cochlea, TRβ2 is responsible for regulating development of sensory organs (Abel et al., 2001, 1999; Jones, Ng, Liu, & Forrest, 2007; Ng, Hurley, et al., 2001).

As with other steroid hormones, recent studies have discovered that T3 can exert nongenomic actions in addition to its canonical genomic actions. These nongenomic effects of T3 act via mechanisms other than direct regulation of nuclear gene transcription; in fact, the nongenomic effects can be measured before changes in nuclear gene transcription would have had time to occur (Lösel & Wehling, 2003; Martin et al., 2014). Nongenomic TH actions can occur via a number of mechanisms. TH can bind directly to the αVβ3 integrin, triggering a downstream signaling cascade. Alternatively, the TRs themselves can mediate nongenomic actions by binding to proteins in other signaling pathways.

3. TH IN THE DEVELOPING SKELETON

3.1. Skeletal Development

In endochondral bone formation, the process by which long bones develop, a cartilage model is replaced by bone tissue. Chondrocytes differentiate from condensations of mesenchymal precursor cells and secrete matrix proteins including type II collagen and proteoglycans to produce a cartilage template. In the traditional model of endochondral bone formation, chondrocytes in the mid-shaft hypertrophy, and blood vessels infiltrate the cartilage model to bring osteoclasts and osteoblast precursors which form the primary ossification center. As osteoclasts degrade and resorb the cartilaginous matrix and osteoblasts replace that cartilage with bone, the primary ossification center expands outward toward the ends of the developing bone. In long bones, secondary ossification centers (SOCs) develop within the epiphyses; during growth, the primary and SOCs are separated by the epiphyseal growth plates, where linear growth due to growth plate chondrocyte proliferation and hypertrophy followed by new bone formation continues until adulthood. When the primary and SOCs meet, obliterating the growth plate, skeletal maturity is achieved (Mackie, Tatarczuch, & Mirams, 2011).

Regulation of endochondral bone formation and linear growth is the result of the endocrine, paracrine, and autocrine actions of many systemic hormones, cytokines, and growth factors including TH, growth hormone (GH), glucocorticoids, sex steroids, insulin-like growth factor I, parathyroid hormone-related peptide, Indian hedgehog (Ihh), bone morphogenetic proteins (BMPs), fibroblast growth factors (FGFs), and vascular endothelial growth factors (Kronenberg, 2003; Lui, Nilsson, & Baron, 2014). Although linear growth ends during puberty with the closure of the growth plates, the processes of mineralization and accumulation of bone mass continue until early adulthood when peak bone mass is reached (Bonjour & Chevalley, 2014; Bonjour, Theintz, Law, Slosman, & Rizzoli, 1994; Karsenty & Wagner, 2002). Indeed, TH is critical for the development of peak bone mass during childhood and the maintenance of bone mass in adults. However, thyroid status is a balancing act; hyperthyroidism in adults causes an increase in bone turnover and fracture incidence (Vestergaard & Mosekilde, 2002; Vestergaard, Rejnmark, & Mosekilde, 2005).

3.2. TH in Postnatal Growth

Several different genetic mouse models with mutations affecting TSH and TH levels have been studied to investigate the contribution of TH signaling to skeletal development (Table 1). For example, mice lacking the Pax8 gene, a thyroid-specific transcription factor necessary for development of thyroid follicular cells (Friedrichsen et al., 2003; Mansouri, Chowdhury, & Gruss, 1998), fail to develop a thyroid gland. Pax8^−/− mice have undetectably low TH levels and produce extremely high amounts of TSH in an attempt to compensate. In the absence of TH, these mice exhibit impaired linear growth, delayed endochondral bone formation, defective chondrocyte differentiation, reduced cortical bone, impaired trabecular bone remodeling, and decreased bone mineralization. Additionally, mice with a non-functioning TSHR (Tshr^hyt/hyt mice) have high levels of TSH, extremely low levels of TH, and a similarly impaired skeletal phenotype (Beamer & Cresswell, 1982; Beamer, Eicher, Maltais, & Southard, 1981; Xing et al., 2012). Furthermore, dual oxidase-2 (Duox2) mutant mice with hypothyroidism exhibited reduced bone mass at the end of the prepubertal growth period (Xing et al., 2012). The skeletal phenotypes of these mice appear to be due primarily to the effects of TH deficiency, not to the presence of excess TSH (Bassett et al., 2008). TSHR knockout (TshR^−/−) mice, however, do display a bone phenotype with severe growth retardation, impaired linear growth, reduced bone mineral density, and increased bone resorption (Marians et al., 2002). The fact that this low bone mass could not be rescued by TH supplementation later in life led to the suggestion that TSH may be an independent negative regulator of bone remodeling (Abe et al., 2003). Clinical congenital hypothyroidism in humans due to a loss-of-function TSHR mutation results in prolonged jaundice, constipation, poor feeding, umbilical hernia, macroglossia, a wide-open posterior fontanelle, and edematous and dry skin (Papadimitriou, Papadimitriou, Papadopoulou, Nicolaidou, & Fretzayas, 2007; Persani et al., 2010).

Table 1.

Mutant Mouse Models Showing Phenotypes of Disrupted TH Signaling Pathways in the Developing Skeleton

Mouse Model	Genetic Alteration	TH Axis Phenotype	Developing Skeletal Phenotype	References
Pax8−/−	Impaired thyroid follicular cell development due to loss of transcription factor Pax8	Lack of thyroid, ~2000-fold elevation of TSH, undetectable T3/T4, functional TSHR	Compromised linear growth, late endochondral bone formation, decreased cortical bone and bone mineralization	Bassett et al. (2008), Flamant et al. (2002), Friedrichsen et al. (2003), and Mansouri et al. (1998)
Tshrhyt/hyt	Loss of function in TSHR	~2000-Fold elevation of TSH, low T3/T4, no TSHR signaling	Compromised linear growth, late endochondral bone formation, decreased cortical bone and mineralization	Bassett et al. (2008), Beamer and Cresswell (1982), Gu et al. (1995), and Stein et al. (1994)
Tshr−/−	Deleted Tshr	Hypoplastic thyroid, undetectable T3/T4, large elevation in TSH	Severely impaired linear and overall growth, decreased mineral density, increased bone resorption, altered bone formation, death if not treated with TH after weaning	Abe et al. (2003), Baliram et al. (2012), Marians et al. (2002), and Sun et al. (2013)
Dio2−/−	Deleted deiodinase 2	Slight increase in T4 and TSH, normal T3	Normal growth, reduced bone formation, increased mineralization, brittle bones with age	Bassett et al. (2010), Liao et al. (2011), and Schneider et al. (2001)
Dio3−/−	Deleted deiodinase 3	Perinatal thyrotoxicosis, hypothyroidism in adulthood	Reduced body length	Hernandez et al. (2006, 2007)
TRα−/−	Deleted TRα1/2, TRΔα1/2 preserved	Severe reduction in T3/T4, elevated TSH (normal GH)	Severe growth retardation, late endochondral bone formation, impaired chondrocyte differentiation, reduced mineralization	Chassande et al. (1997), Fraichard et al. (1997), and Gauthier et al. (1999)
TRα1−/−	Deleted TRα1/Δα1, TRα2/Δα2 preserved	Mild decreased T4 in males only	No change in growth	Wikström et al. (1998)
TRα2−/−	Deleted TRα2/Δα2, increased TRα1/Δα1	Mild decrease in T3/T4 (low IGF-I)	No change in growth, low BMD/cortical bone by adulthood	Saltó et al. (2001)
TRα0/0	Deleted TRα1/2 and TRΔα1/2	Slight decrease in T4, normal T3 and TSH (normal GH)	Transient growth retardation, late intramembranous and endochondral bone formation, compromised chondrocyte differentiation and mineralization	Bassett, O’Shea, et al. (2007) and Gauthier et al. (2001)
TRαPV/+	Heterozygous dominant-negative TRα	Mild thyroid failure (normal GH, reduced IGF-I)	Transient growth retardation, late endochondral bone formation, reduced chondrocyte differentiation and mineralization	Bassett et al. (2014), Kaneshige et al. (2001), and O’Shea et al. (2005)
TRβ−/−	Deleted TRβ1/2	RTH, goiter, increased T3/T4, TSH	Persistent short stature, accelerated intramembranous and endochondral bone formation, increased mineralization	Bassett, Nordström, et al. (2007), Bassett, O’Shea, et al. (2007), Forrest et al. (1996), Gauthier et al. (2001), and Ng, Hurley, et al. (2001)
TRβ2−/−	Deleted TRβ2	Mild RTH, slight increase in T3/T4, TSH	No change in growth	Abel et al. (1999) and Ng, Rüsch, et al. (2001)
TRβPV/PV	Homozygous dominant-negative TRβ	Severe RTH, goiter, ~400-fold increase in TSH, ~15-fold increase T4, ~9-fold increase in T3 (reduced GH, increased IGF-I)	Accelerated prenatal growth, persistent postnatal growth retardation, accelerated intramembranous and endochondral bone formation, increased mineralization	Bassett, Nordström, et al. (2007), Kaneshige et al. (2000), and O’Shea et al. (2005, 2003)
TRα−/− TRβ−/−	Deleted TRα1/2 and TRβ1/2	RTH, small goiter, increased T3/T4, TSH	Disorganized growth plate, late endochondral bone formation, compromised chondrocyte differentiation and mineralization, death within 5 weeks	Gauthier et al. (1999)
TRα1−/− TRβ−/−	Deleted TRα1/Δα1 and TRβ1/2, TRα2/Δα2 preserved	RTH, large goiter, increased T3/T4, TSH (reduced GH/IGF-I)	Persistent growth retardation, late endochondral bone formation, reduced mineralization	Göthe et al. (1999) and Kindblom et al. (2001, 2005)
TRα2−/− TRβ−/−	Deleted TRα2/Δα2 and TRβ1/2, increased TRα1/Δα1	Mild decrease in T3/T4, normal TSH	Transient growth delay	Ng, Hurley, et al. (2001) and Ng, Rüsch, et al. (2001)
TRα0/0 TRβ−/−	Deleted TRα1/2, TRΔα1/2, and TRβ1/2	RTH, goiter, >10-fold increase in T3/T4, >200-fold increase in TSH (low GH/IGF-I)	Disorganized growth plate, delayed growth, late endochondral bone formation, compromised chondrocyte differentiation and mineralization	Bassett, Swinhoe, Chassande, Samarut, and Williams (2006), Flamant et al. (2002), and Gauthier et al. (2001)

Studies investigating the effect of knocking out deiodinase activity on the developing skeleton have largely found no effect (Berry et al., 1993; Liao et al., 2011; Schneider et al., 2006; Schoenmakers, Pigmans, Poland, & Visser, 1993), although D2-null mice show reduced bone formation, increased mineralization, and brittle bones with age (Bassett et al., 2010; Liao et al., 2011; Schneider et al., 2001), and D3-null mice exhibit thyrotoxicosis when they are very young and have reduced body length (Hernandez, Martinez, Fiering, Galton, & St Germain, 2006; Hernandez et al., 2007). Furthermore, the only reports of an effect on the developing skeleton due to knocking out TH transporters involve combining Mct8 knockout with Oatp1c1, D1, or D2 knockout (Liao et al., 2011; Mayerl et al., 2014; Morte et al., 2010). Clinical cases involving mutant Mct8 genes exhibit normal linear growth (Dumitrescu, Liao, Best, Brockmann, & Refetoff, 2004; Friesema et al., 2004); however, patients with a mutation in the pathway necessary to produce functional deiodinases have presented with short stature and transient growth retardation (Dumitrescu et al., 2005; Refetoff & Dumitrescu, 2007).

3.3. TH in Secondary Ossification

In mice, there is a window from postnatal days 5–14 when serum T3 levels rapidly increase. This window is followed by a shift from longitudinal bone growth to GH-independent bone mass acquisition and increase in BMD (Xing et al., 2012). This shift from linear bone growth to accelerated bone mass acquisition coincides with the initiation of secondary ossification in the epiphyses in mice. In humans, the identification of a proximal humeral epiphyseal ossification center occurs around a gestational age of 38 weeks which coincides with attainment of peak levels of TH (36–42 weeks). Thus, the time of appearance of secondary ossification in several species including humans, rats, and mice coincides with the time when peak levels of TH are attained (Xing, Cheng, Wergedal, & Mohan, 2014).

The epiphyseal region in long bones contains many regions of interest, including the chondrocyte growth plate, the SOC, and the articular cartilage. All of these regions become specified during different periods, although the entire region is populated by immature chondrocytes during embryonic and very early postnatal development. Chondrocytes in the growth plate then form a hierarchical organization involving slowly dividing stem-like cells, rapidly proliferating cells, and finally prehypertrophic and hypertrophic cells. A similar region surrounds the SOC, albeit in a circular, not linear, formation. Unlike primary ossification, which occurs during embryonic development, secondary ossification occurs during early postnatal development; however, both primary and secondary ossification take place via endochondral bone formation, and the method by which endochondral bone formation initiates during both stages of development has recently been brought into question. Of note, given recent studies of chondrogenic to osteogenic transdifferentiation in regions of primary ossification (Ono, Ono, Nagasawa, & Kronenberg, 2014; Yang, Tsang, Tang, Chan, & Cheah, 2014; Zhou et al., 2014), it may be likely that early secondary ossification involves a similar mechanism, given that ossification begins prior to vascular invasion.

Many studies have suggested that TH plays a key role in the development of these regions and is required for their ossification. Furthermore, our studies have suggested that the increase in TH levels during the prepubertal growth period is essential for initiation and progression of SOC formation (Xing et al., 2016, 2014). TH is a known factor for promoting the terminal differentiation of growth plate chondrocytes (Ishikawa, Genge, Wuthier, & Wu, 1998; Robson, Siebler, Stevens, Shalet, & Williams, 2000) and osteoblasts. Indeed, hypothyroid mice exhibit delayed or incomplete ossification of the SOC as well as a disrupted growth plate such that it inhibits differentiation (Fig. 3; Bassett et al., 2008). Conversely, hyperthyroidism can lead to adverse phenotypes such as premature epiphyseal fusion caused by increased early ossification prior to growth plate proliferation (Segni & Gorman, 2001). While TRα is ubiquitously expressed in chondrocytes and osteoblasts, TRβ is expressed only in prehypertrophic cells post-TH treatment, which necessitates mediation through TRα (Stevens et al., 2000). Moreover, recent findings suggest that TRβ, not TRα, plays a key role in thyroid signaling to the SOC and that Ihh plays an important role as a downstream effector of this process (Xing et al., 2016, 2014). Unlike sonic hedgehog (Shh), Ihh is typically involved in promoting differentiation. In a recent study of zebrafish jawbone regeneration, Ihh regulated the differentiation of chondrocytes to osteoblasts (Paul et al., 2016). Notably, TH also increases the expression of several osteogenic factors such as osteocalcin and alkaline phosphatase in chondrocytes and in euthyroid mice in part via promotion of osterix expression in hypertrophic chondrocytes (Xing et al., 2014).

Fig. 3.

TH deficiency in Tshr^hyt/hyt mice impairs formation of the epiphysis. Micro-computed tomography images of the distal femurs and proximal tibiae of 14-day-old euthyroid Tshr^hyt/+ and hypothyroid Tshr^hyt/hyt mice show epiphyseal defects in the femur and tibia of Tshr^hyt/hyt mice, indicated by arrowheads.

4. TH IN THE ADULT SKELETON

Although this review primarily focuses on studies involving the developing skeleton, we will include a brief overview of the effects of TH in the adult skeleton for the purpose of contrast. In the developing skeleton, the effects of TH are broadly anabolic; juvenile hypothyroidism leads to short stature due to a lack of skeletal growth, and juvenile thyrotoxicosis leads to short stature due to accelerated skeletal growth and premature closure of the growth plates. By contrast, TH affects adult bone maintenance by increasing bone turnover leading to net bone resorption. Adult hyperthyroidism is associated with increased markers of both bone formation and resorption (Garnero, Vassy, Bertholin, Riou, & Delmas, 1994; Guo, Weetman, & Eastell, 1997; Harvey et al., 1991; Toivonen, Tähtelä, Laitinen, Risteli, & Välimäki, 1998), and initiation of therapy for thyrotoxicosis in premenopausal women led to a 4% increase in BMD within the first year of treatment (Udayakumar, Chandrasekaran, Rasheed, Suresh, & Sivaprakash, 2006). Moreover, thyrotoxicosis leads to increased risk of hip fracture (Vestergaard & Mosekilde, 2003). Furthermore, adult hypothyroidism is associated with decreased osteoblast and osteoclast activity (Eriksen, Mosekilde, & Melsen, 1985, 1986; Mosekilde & Melsen, 1978), and, while hypothyroidism is not directly related to fracture risk, overly aggressive TH supplementation in hypothyroidism may lead to an increased risk of fracture (Ko et al., 2014). For a more comprehensive discussion of the TH effects on the adult skeleton, see the recent review by Bassett and Williams (2016).

5. SKELETAL EFFECTS OF TR SIGNALING

Before ligation by T3, TRs are located in the nucleus where they can form homodimers or heterodimers with retinoid X receptors (RXR). These heterodimers can bind to specific genomic TH-response element (TRE) sequences which exist in the promoter regions of T3 target genes (Fig. 4). In the absence of T3, the TR–RXR complex associates with corepressor proteins (e.g., nuclear receptor corepressor and the silencing mediator for retinoid and TR) which can recruit histone deacetylases and inhibit gene transcription (Astapova et al., 2008, 2011; Chassande, 2003; Hashimoto et al., 2001; Venero et al., 2005; Wallis et al., 2008). When T3 enters the cell, either directly or via conversion from T4 by D1/D2, the corepressors are released, and coactivator proteins (e.g., CBP/p300, pCAF, and steroid receptor coactivator 1) are recruited to activate transcription of T3 target genes. Indeed, the fact that TRs both inhibit and activate transcription of target genes depending on whether they are T3 bound is responsible for the large effect size observed in T3-mediated transcriptional modulation (Bassett, Harvey, & Williams, 2003; Brent, 2012; Cheng, Leonard, & Davis, 2010; Harvey & Williams, 2002; Kim et al., 2005; Vella et al., 2014).

Fig. 4.

T3 and T4 enter the cell via a combination of direct diffusion and membrane transport proteins. The levels of active T3 in the cell are regulated by deiodinase conversion of T4 to T3 or inactivation to rT3 and T2. Before ligation with T3, TRα in a heterodimer with RXR binds to specific TH-response element (TRE) sequences, recruiting corepressors to inhibit transcription of target genes. When T3 binds TRα, the corepressors are replaced by coactivators, allowing target gene transcription.

5.1. TRα Signaling in Postnatal Growth

Genetic mouse models have been used to study the role of TRα in skeletal development (Table 1). As per the discussion in Section 2.3, the THRA gene encodes several TRα splice variants and truncated isoforms. Mice lacking all TRα isoforms (TRα^0/0 mice) are systemically euthyroid; however, these mice exhibit transient developmental growth retardation and delayed endochondral bone formation including impaired chondrocyte differentiation and reduced mineral deposition (Bassett et al., 2008; Gauthier et al., 2001). Moreover, mice with dominant-negative TRα1 mutations do experience some amount of systemic hypothyroidism and display more severe skeletal development phenotypes than TRα^0/0 mice. For example, TRα1^PV/+ mice have severe and unremitting postnatal growth retardation with delayed endochondral bone formation and impaired bone mineralization (Bassett & Williams, 2009; Kaneshige et al., 2001). The fact that the presence of a dominant-negative TRα causes a more severe skeletal phenotype than the simple absence of the TRα demonstrates the powerful antagonistic effect of the unbound TRα1 receptor which, when unable to bind T3, recruits repressor proteins and represses expression of T3 target genes such as fibroblast growth factor receptors 1 and 3 (Barnard et al., 2005; Bassett, Nordström, et al., 2007; Chassande, 2003). In clinical cases of a heterozygous dominant-negative TRα receptor, patients present with a type of resistance to thyroid hormone (RTH) which is accompanied by short stature and delayed bone development, among other symptoms (Bochukova et al., 2012; van Mullem et al., 2012).

Mechanistically, TRα1 is required for longitudinal growth as it promotes expression of Shh, an important regulatory factor which is known to favor chondrocyte proliferation and survival, which lead to longitudinal growth, over differentiation, and hypertrophy. Shh increases expression of master chondrocyte regulator Sox9, and overexpression of Shh in chondrocytes led to impaired endochondral bone formation (Desouza et al., 2011; Gil-Ibáñez, Bernal, & Morte, 2014; Kim & Mohan, 2013; Tavella et al., 2004).

5.2. TRβ Signaling in Postnatal Growth

As with TRα, mutant mouse models (Table 1) have been studied to investigate the effect of TRβ signaling in skeletal development (Fig. 5). Since TRβ is responsible for sending negative feedback to the hypothalamus to regulate the HPT axis set point, TRβ-null (TRβ^−/−) mice lack that negative feedback and exhibit RTH. These mice display the effects of skeletal hyperthyroidism including persistent short stature due to premature closure of the growth plates, advanced endochondral and intramembranous bone formation, accelerated chondrocyte differentiation, and increased mineral deposition (Abel et al., 1999; Bassett et al., 2008; Forrest et al., 1996). Furthermore, mice with a dominant-negative TRβ mutation (TRβ^PV/PV mice) have a more severe RTH and skeletal hyperthyroid phenotype characterized by accelerated prenatal growth and advanced endochondral and intramembranous bone formation (Kaneshige et al., 2000; O’Shea et al., 2005, 2003). Since these mice have high levels of T3 which results in increased TRα signaling in bone, skeletal expression of T3 target genes FGFR1 and 3 is increased (Barnard et al., 2005; Bassett, Nordström, et al., 2007; Bassett et al., 2008). Clinical reports of patients with RTH due to dominant-negative TRβ mutations include developmental and growth delay (Ferrara et al., 2012; Refetoff, DeWind, & DeGroot, 1967; Refetoff & Dumitrescu, 2007; Weiss, 2008).

Fig. 5.

TRβ as a homodimer exerts genomic actions via a similar mechanism to TRα, binding to different TRE sequences to regulate transcription of different target genes. In addition, TRs can also exert nongenomic actions. For example, TRβ can interact directly with the p85 subunit of phosphoinositide 3-kinase (PI3K). In the presence of T3, TRβ releases PI3K to activate downstream signaling cascades involving AKT and β-catenin.

In contrast to TRα regulation of longitudinal growth via upregulation of Shh and the resulting increase in chondrocyte proliferation and survival, TRβ promotes chondrocyte differentiation by stimulating Ihh signaling. Increased Ihh induces a number of growth and transcription factor signaling pathways including osterix, Runx2, β-catenin, IGF-I, and RANKL, leading to chondrocyte differentiation, hypertrophy, and even transdifferentiation into osteoblasts to carry out endochondral bone formation (Amano, Densmore, Nishimura, & Lanske, 2014; Paul et al., 2016; Xing et al., 2016, 2014).

5.3. Nongenomic TH Signaling

Nongenomic TH actions have been observed in the region of the plasma membrane, within the cytoplasm, and at the mitochondria. For example, T4 has been shown to act via the MAPK pathway to stimulate proliferation and angiogenesis, an effect mediated by the αVβ3 integrin. Furthermore, palmitoylated TRα activates the nitric oxide/protein kinase G2/Src pathway to increase MAPK and PI3K/AKT signaling which mediate rapid T3 actions in osteoblasts, and T3 acts via TRβ to activate the PI3K/AKT pathway (Fig. 5; Bassett et al., 2003; Bergh et al., 2005; Cao, Kambe, Moeller, Refetoff, & Seo, 2005; Cheng et al., 2010; Davis, Goglia, & Leonard, 2016; Davis et al., 2006; Furuya, Hanover, & Cheng, 2006; Gauthier & Flamant, 2014; Kalyanaraman et al., 2014; Martin et al., 2014; Moeller & Broecker-Preuss, 2011; Moeller, Dumitrescu, & Refetoff, 2005; Storey et al., 2006). AKT-mediated phosphorylation of β-catenin leads to nuclear accumulation of β-catenin to promote its transcriptional activity (Fang et al., 2007). Thus, nongenomic signaling by TH adds to the complexity needed for the multitude of TH effects in various cell types.

6. SUMMARY AND CONCLUSIONS

In summary, TH is a critical regulator of skeletal development and postnatal growth (Fig. 6). During the very early stages of development (the first few postnatal days in mice), low levels of T3 stimulate constitutive TRα in bone to upregulate expression of Shh and Sox9. Under these conditions, chondrocytes remain immature and are stimulated to proliferate, leading to longitudinal growth. However, when serum T3 levels rise during postnatal days 5–14, TRβ expression is induced. T3 then binds to TRβ, shifting the balance from predominant Shh expression in chondrocytes to Ihh expression (Aghajanian, Xing, Cheng, Pourteymoor, & Mohan, 2016). Ihh induces Osx, Runx2, β-catenin, IGF-I, and RANKL, causing chondrocytes to differentiate, hypertrophy, and transdifferentiate into endochondral bone-forming osteoblasts. While much progress has been made in recent years in our understanding of the roles played by TH in skeletal development, much still remains to be discovered. In order to determine the specific roles of each TR during endochondral bone formation, conditional knockout of the various TRs in specific cell types during specific stages of development will be required. Furthermore, the implications of chondrocyte to osteoblast transdifferentiation for the traditional model of endochondral bone formation are still in the early stages of being elucidated, and whether TH regulation of postnatal development in other tissues involves transdifferentiation is an open question. Additionally, the role of nongenomic TH signaling in bone development is still largely unexplored. Continuing to develop our understanding of these mechanisms of TH regulation of bone will open the door to development of novel therapies for musculoskeletal conditions which plague so many worldwide.

Fig. 6.

Model of TH regulation of the secondary ossification center in the epiphysis. The increase in serum levels of THs during the early postnatal growth period leads to an increase in TRβ1 expression in epiphyseal chondrocytes. Activation of TH signaling in epiphyseal chondrocytes via TRβ1 leads to an increase in Ihh and other growth factor pathways (IGF-I, Wnt, BMPs, and RANKL) resulting in chondrocyte to osteoblast transdifferentiation and bone formation.