Neuronal Control of Bone Remodeling

Abstract

Although the brain is well established as a master regulator of homeostasis in peripheral tissues, central regulation of bone mass represents a novel and rapidly expanding field of study. This review examines the current understanding of central regulation of the skeleton, exploring several of the key pathways connecting brain to bone and their implications both in mice and the clinical setting. Our understanding of central bone regulation has largely progressed through examination of skeletal responses downstream of nutrient regulatory pathways in the hypothalamus. Mutations and modulation of these pathways, in cases such as leptin deficiency, induce marked bone phenotypes, which have provided vital insights into central bone regulation. These studies have identified several central neuropeptide pathways that stimulate well-defined changes in bone cell activity in response to changes in energy homeostasis. In addition, this work has highlighted the endocrine nature of the skeleton, revealing a complex cross talk that directly regulates other organ systems. Our laboratory has studied bone-active neuropeptide pathways and defined osteoblast-based actions that recapitulate central pathways linking bone, fat, and glucose homeostasis. Studies of neural control of bone have produced paradigm-shifting changes in our understanding of the skeleton and its relationship with the wider array of organ systems.

Despite its fixed and stable appearance, the skeleton is a dynamic tissue, with the necessity to adapt to changing mechanical strain, ensuring sufficient strength is maintained to withstand the forces generated during movement, while still providing calcemic demands. This adaption is primarily achieved through the basic multicellular units, osteoblasts and osteoclasts, which carry out bone formation and resorption, respectively. Although endocrine control of bone remodeling has been well established, central regulation of bone mass represents an emerging area of study that is identifying novel regulatory axes between the nervous system and bone homeostasis.

The brain is a powerful regulator of peripheral tissues, involved in a multitude of homeostatic processes across many organ systems in the body. Recent studies have outlined regulatory axes between the brain and bone, which originate from one of the most powerful regulatory regions in the body: the hypothalamus, via efferent neuronal discharges which are processed through the brainstem. The semipermeable nature of the blood-brain barrier in the hypothalamus allows detection and integration of signals from peripheral tissue through both the circulation and neural return. A direct regulatory link between the nervous system and the skeleton was strongly supported after the presence of innervation and neuropeptide receptors were identified in bone cells using immunocytochemistry. This connection was further supported following the identification of neural tracts between femoral bone marrow and the central nervous system using retrograde transsynaptic signaling (Denes et al. 2005). These direct central pathways represent a novel method of bone cell regulation, indicating the presence of several regulatory axes between the brain and bone. The study of these processes has revealed unexpected insights into not only control of bone mass but also previously unknown regulatory connections involving skeletal tissue, indicating a level of complexity and interconnectedness that has been previously unappreciated. This review examines the development of understanding within the field of central and neural control of skeletal metabolism and remodeling.

Leptin

Leptin is a hormone that plays a pivotal role in whole body homeostasis. The small polypeptide is secreted into the peripheral circulation by adipocytes and can be used as a measure of total body adiposity. Leptin is detected by its cognate receptor in the hypothalamus, which in turn powerfully regulates energy balance and suppression of food intake, most potently when leptin signaling is reduced, as during starvation. The study of a mutant mouse with nonfunctional leptin provided the compelling evidence of a direct connection between the brain and bone, representing a vital step in the understanding of central control of bone homeostasis. Mutations associated with the leptin gene (ob/ob) or its receptor (db/db) block leptin signaling and result in significant changes to the regulation of food intake and fat storage as well as endocrine function (Friedman and Halaas 1998; Halaas et al. 1995; Tartaglia et al. 1995) but also lead to marked bone phenotypes. These mice with leptin signaling disruption exhibited increased cancellous bone volume despite displaying hypogonadism and hypercorticism, which usually result in decreased cancellous bone due to elevated bone loss (Ducy et al. 2000). The increase in bone mass was caused by an imbalance in bone remodeling, where the elevated resorption was surpassed by a significant increase in formation. In further experiments, hypothalamic leptin was implicated in bone mass regulation when cerebral infusion of leptin decreased cancellous bone volume in both ob/ob and wild-type mice, with serum leptin levels remaining constant (Ducy et al. 2000). In a similar experiment, leptin was centrally infused into only one individual within a pair of parabiosed ob/ob mice (Takeda et al. 2002). The leptin-deficient phenotype was only rescued in the recipient mouse, despite successful cross-circulation between the two, indicating that the regulation of bone by central leptin was not via the circulation. These studies established that inhibition of bone formation in cancellous bone is mediated through leptin in the hypothalamus and a nonendocrine intermediate in a similar manner to the central regulation of obesity.

Although leptin-dependent bone remodeling increased our understanding of neuronal control of skeletal physiology, many contradictions within the literature have indicated additional levels of complexity to this relationship. In numerous studies, leptin appears to have distinct effects on cortical bone that differ to the cancellous compartment. For example, ob/ob mice have been reported to display a decrease in both femoral length and total body bone mass (Baldock et al. 2006; Steppan et al. 2000). This phenotype was partly rescued in immature mice by peripheral administration of leptin (Steppan et al. 2000). In addition, the inactivating Gln269Pro mutation on the Ob-Rb leptin receptor gene is responsible for a decrease in femoral bone mineral density (BMD) and bone mineral content (BMC) in leptin resistant Zucker (fa/fa) rats (Mathey et al. 2002; Takaya et al. 1996). However, these reductions in bone length and BMC may relate to the growth plate rather than osteoblastic effects, as leptin is able stimulate growth plate chondrocyte proliferation and differentiation in vitro (Nakajima et al. 2003). The contrasts between axial and appendicular skeletons also provide further complexities to the study of ob/ob mice. The ob/ob phenotype displays increased vertebral length and lumbar BMC while femur midshaft cortical area and thickness and BMC were decreased (Hamrick et al. 2004). The extent to which reduced muscle mass, increased marrow adipogenesis, and disrupted growth plates are responsible for this observed variation in size and structure of cortical versus trabecular bone or appendicular versus axial skeleton is yet to be determined. However, neuropeptide Y (NPY) pathways have been implicated in leptin control of cortical bone but not cancellous as indicated in studies below.

Leptin and Bone in Humans

Although the association between leptin and skeletal homeostasis has been well established in mice, genetic-based human evidence has been limited and mixed. Researchers reported a loss in the markedly increased body fat and an increase in whole body BMD in a young girl in response to administration of leptin for 1 year (Farooqi et al. 1999). In contrast, a family study of 4 individuals carrying a missense mutation in the leptin gene found that, although all 4 were morbidly obese, only 1 exhibited low bone mass (Ozata, Ozdemir, and Licinio 1999). However, this observation was limited by the high degrees of consanguinity and multiple endocrine defects within the sample group. Another approach to gaining insights into the human association between leptin deficiency and bone mass is through the use of indirect clinical evidence. The severe calorie restriction that occurs in individuals with anorexia nervosa results in a reduction of serum leptin and reduced BMD and can lead to amenorrhea as well as extremely low insulin-like growth factor (IGF-1) level. These symptoms were consistent with the low bone mass, reduced fertility, and suppressed somatotropic activity in ob/ob mice (Miller et al. 2004; Soyka et al. 1999). Administration of leptin in patients with exercise or malnutrition-induced hypothalamic amenorrhea led to increased levels of both IGF-1 and bone formation markers (Welt et al. 2004), highlighting again the positive relationship between leptin and bone mass. However, there can be no guarantee that the changes observed in these studies are as a direct result of leptin rather than indirect effects through reproductive or other endocrine processes.

The discovery of leptin has led to a shift in research relating to the relationship between bone and fat, provided a novel mechanism for the associations between bone mass and overall body weight that had previously only been attributed to either mechanical load or fat-induced estrogen production (Reid 2002). Indeed, an association between serum leptin and BMD has been identified in pre- and postmenopausal women (Thomas et al. 2001) and with BMC in healthy nonobese women (Pasco et al. 2001). However, other studies have not replicated these findings (Rauch et al. 1998), and, in several cases, a correlation between leptin and BMC was identified in postmenopausal osteoporotic women but not in the controls (Odabasi et al. 2000). The relationship between leptin and skeletal regulation only increases in complexity when considering males. Similar studies on men have reported negative associations between leptin and BMD (Sato et al. 2001) or no relationship (Thomas et al. 2001), suggesting the effect might be gender specific in humans. In one of the few cases where a positive association between leptin and BMD was reported, markers of bone resorption and formation remained unchanged (Goulding and Taylor 1998), casting doubt over a direct regulatory role of leptin in bone remodeling; however, opposing evidence from in vitro experiments challenges this interpretation.

Bone Cell Responses to Leptin In Vitro

Evidence for presence of the signal-transducing Ob-Rb leptin receptor or leptin binding sites has been detected on ossifying fetal cartilage (Hoggard et al. 1997), immortalized marrow stromal cells (Thomas et al. 1999), chondrocytes (Cornish et al. 2002; Maor et al. 2002; Nakajima et al. 2003; Steppan et al. 2000), primary osteoblasts (Cornish et al. 2002; Enjuanes et al. 2002; Lee et al. 2002; Reseland et al. 2001; Steppan et al. 2000), and some but not all osteosarcoma cell lines (Reseland et al. 2001). Direct effects of leptin on a variety of bone cells have been observed during in vitro studies. Increases in the proliferation and differentiation of cultured growth plate chondrocytes have been detected in response to leptin (Nakajima et al. 2003). In addition, leptin has been observed to stimulate osteoblast differentiation while suppressing adipogenesis in human stromal cell lines (Thomas et al. 1999). Similar stimulatory effects on cell proliferation and differentiation, in addition to changes within cell function, have been directly linked to leptin in cultures of both human and rat osteoblasts (Cornish et al. 2002; Gordeladze et al. 2002). In contrast, an inhibition of osteoclastogenesis has also been reported in cultures in response to leptin (Burguera et al. 2001; Holloway et al. 2002). However, not all in vitro studies have reported peripheral actions of leptin in bone, indicating an additional layer of complexity to the previous findings. In one such study, investigators found no significant expression of leptin or the leptin receptor long form expression in primary mouse osteoblastic cultures and no change in the formation of mineralized nodules in response to leptin treatment (Ducy et al. 2000). Importantly, the disparity within the findings of these in vitro leptin studies can be resolved by the understanding that, while peripheral actions have a significant impact on bone cell response at high leptin concentrations, central anti-actions become the determining factor at lower levels, where leptin deficiency stimulated powerful centrally mediated energy homeostatic processes, such as NPY (see below).

Sympathetic Nervous System (SNS) Regulation of Bone Mass

The mechanism by which central leptin signaling altered bone remained elusive; however, humoral signaling was ruled out by parabiosis experiments on ob/ob mice (Takeda et al. 2002), indicating the likelihood that efferent neural signaling may be involved. Complex studies involving chemical lesioning and genetic models isolated the ventromedial hypothalamus (VMH) as the primary source of leptin signaling to cancellous bone (Takeda et al. 2002).

Subsequently, sympathetic activity was outlined as a potential key downstream mediator of central leptin signaling, as sympathetic tone is known to be altered in leptin deficiency. Studies led by Takeda and colleagues utilized dopamine β-hydroxylase deficient mice to investigate the relationship between the SNS and the central actions of leptin. These mice are incapable of producing epinephrine and norepinephrine and so are unable to activate adrenergic receptors. More importantly, dopamine β-hydroxylase deficient mice have high bone mass which is unaffected following intracerebroventricular (icv) infusion of leptin (Takeda et al. 2002). This not only indicates that the skeletal impacts of leptin require a functional autonomic nervous system but also identifies the sympathetic system as a key neuronal mediator (Takeda et al. 2002).

This complex relationship has also been explored through the use of agonists and antagonists of β-adrenergic receptor (β-AR). The disruption of adrenaline and noradrenaline synthesis (from dopamine β-hydroxylase) led to an increase in cancellous bone mass in wild-type mice and protected against cancellous bone loss following icv leptin. This restriction of leptin-induced cancellous bone loss was also demonstrated in ob/ob mice, following inhibition of adrenergic signaling using propranolol treatment (Takeda et al. 2002). Although equivalent effects in cortical bone were not fully explored, these studies established a vital role for β-adrenergic signaling via SNS in the central leptin signaling pathway to bone. The β-agonist isoproterenol not only restores sympathetic activity when administered to ob/ob mice but also reduces cancellous bone mass in both wild-type and ob/ob mice (Takeda et al. 2002). The bone change occurs without changing the overall body weight, consistent with β-adrenergic response acting downstream of leptin. Consistent with this, the bone mass of wild-type mice can be increased through the administration of the nonselective β-AR antagonist propranolol. Thus, changes in SNS activity were implicated in the modulation of bone remodeling and consequently the ob/ob phenotype (Takeda et al. 2002). In addition, deficiencies in β2-AR resulted in an increase in cancellous bone volume (Elefteriou et al. 2005), while investigators were able to limit age-related bone loss through the disruption of β2-AR signaling by deletion of adenylyl cyclase-5 (Yan et al. 2007).

In addition to in vivo studies, β-adrenergic signaling has been further implicated in control of bone formation in several studies using osteoblast cell lines. β2-AR expression has been detected in both rat osteoblast-like cells and mouse primary osteoblast cultures (Moore et al. 1993; Takeda et al. 2002), and β1 and β2 ARs have been identified in human osteoblast-like cell lines (Kellenberger et al. 1998). This indicates a direct impact of sympathetic activity on osteoblast activity and provides a potential mechanism for sympathetic control over bone formation. Coupling of β2-AR to intracellular signaling pathways has been demonstrated in osteoblast-like cell lines. In one such study, expression of the c-fos gene was used to demonstrate increased downstream activity of β2-AR in response to the β2-AR-specific agonist, formoterol, in SaOS-2 human osteosarcoma cell line, which was reversed following administration of the corresponding antagonist (Kellenberger et al. 1998). β1AR is also expressed in bone cells and appears to act in an opposing manner to β2-AR, with reduced bone mass and mechanical responses in β1 and β1/β2 knockout (KO) mice compared to β2 KO alone (Pierroz et al. 2012). In addition to demonstrating adrenergic control of bone formation, several studies have indicated that adrenergic signaling is able to modulate the whole process of bone remodeling. Bone resorption has been stimulated in response to adrenaline in neonatal mouse calvariae in organ culture (Moore et al. 1993). However, the addition of adrenaline has been found to increase receptor activator of nuclear factor kappa-B ligand (RANKL) activity in MC3T3-E1 preosteoblastic mouse cell line (Takeuchi et al. 2000). This suggests that changes in bone resorption in response to adrenergic signaling could occur indirectly as a result of cross talk between osteoblasts and osteoclasts. Taken as a whole, the parallel responses of B2-AR and leptin-deficient models as well as the well-established sympathetic alterations in leptin deficiency identified the first pathway from brain to bone via neural pathways and linked the energy homeostatic regulation to control of bone mass.

Role of β-adrenergic Signaling in Regulation of Bone and Its Clinical Relevance

β-blockers are commonly used to treat a wide variety of cardiovascular diseases, though until recently there had been very little investigation into how they affect skeletal mass and strength. There is now some epidemiological evidence suggesting that β-adrenergic signaling regulates on bone homeostasis, following new studies into how bone turnover, BMD and fracture risk in human patients are influenced by β-adrenergic antagonists. In men and women between the ages of 30 and 79, β-blockers were linked with a decreased risk of fracture (Schlienger et al. 2004), and in women over the age of 50, there was again a link to a reduction in fracture risk, as well as increased BMD in the hips and forearm (Pasco et al. 2004), findings that reflect the data from mouse studies. However, these observations contrasted another study, which associated the use of β-blockers in premenopausal women with a 3-fold increase in fracture risk and reduced serum osteocalcin, a marker for bone formation via osteoblasts (Rejnmark et al. 2004), though one explanation for these conflicting observations may be the differences in experimental design. There are numerous conditions that may be treated by inducing bone formation (such as age-related osteoporosis), and hence, restriction of β-adrenergic access to bones could be of therapeutic benefit. To determine whether this is the case, placebo-controlled randomized clinical trials must be conducted.

NPY and the Y Receptors

NPY system

Following the studies into the role of leptin in the hypothalamus, several key neuronal pathways critical to bone activity were described, with the NPY system being one of the most significant. This system utilizes 3 ligands: NPY, peptide YY (PYY), and pancreatic polypeptide (PP). PP is involved in controlling pancreatic endocrine functions and central satiety following production in the pancreas via endocrine islet cells (Ueno et al. 1999). PYY is also involved in satiety control, as well as gastrointestinal regulation, and originates mainly from endocrine L cells in the gastrointestinal tract and pancreatic islets (Lundberg et al. 1984), though it has also been found to be expressed in neurons of the brainstem (the function of the ligand here remains unknown; Ekblad and Sundler 2002). The most well understood of these ligands, however, is NPY, a peptide 36 amino acids in length that is often coreleased with noradrenaline following nerve stimulation. It is abundantly expressed in both the central and peripheral nervous systems, in sympathetic and parasympathetic nerve fibers, as well as in the blood from both neuronal and adrenal origins. NPY’s role as a vasoconstrictor in these neurons has been well-documented (Morris 1994; Parker and Herzog 1999; Pernow et al. 1987), while central NPY activity has been linked to the control of pituitary hormone release, cardiac and respiratory action, and very powerful stimulation of food intake (Hokfelt et al. 1998; Wettstein, Earley, and Junien 1995). NPY-ergic neurons have been found to be highly concentrated in the hypothalamic arcuate nucleus (ARC) and the VMH, regions where the blood-brain barrier function is reduced and endocrine factors can signal directly to resident neurons (Hokfelt et al, 1998). These 3 ligands transmit their effects through G protein-coupled receptors, of which Y1, Y2, Y4, Y5, and Y6 have been identified (Blomqvist and Herzog 1997; Lin et al. 2005).

The Relationship between NPY and Leptin

As NPY has been shown to be a key regulator of energy homeostasis, it is unsurprising that it has a very close association in the hypothalamus with leptin; the majority of NPY-ergic neurons in the ARC also express leptin receptors (Mercer et al. 1996). Leptin and NPY share a reciprocal relationship, most powerful during periods of reduced leptin signaling. Specifically, leptin deficiency, such as during starvation and loss of fat mass, reduces leptin signaling to NPY-ergic the hypothalamus, increasing NPY production, which functions as a key downstream regulator, stimulating energy conservation and strong increases in appetite (Schwartz, Dallman, and Woods 1995; Spanswick et al. 1997; Spiegelman and Flier 1996). This upregulation of NPY is also seen in leptin KO ob/ob mice (Wilding et al. 1993). Leptin receptors have also been found in other regions of the hypothalamus such as the VMH, which was shown to be the source of the leptin pathway to cancellous bone (Takeda et al. 2002). Determining whether the ARC leptin receptors and with it NPY expression was capable of regulating bone was an important research question.

The Y2 receptor KO, Y2^−/− , mice (chosen due to the coexpression of Y2 and leptin receptors in ARC neurons) was the first model that investigated skeletal changes related to NPY (Baskin, Breininger, and Schwartz 1999; Broberger et al. 1997). These studies highlighted a role for the Y2 receptors in the central pathway and also provided the first specific gene deletion within the hypothalamus that affected bone remodeling. In both germ line Y2^−/− mice and those with Y2 receptors conditionally deleted in the hypothalamus, an increase in cortical and cancellous bone volume was observed as a result of a greater rate of bone formation (Baldock et al. 2002, 2006), while aside from a moderate increase in osteoclast number, bone resorption parameters were constant. The key finding from the investigation was that despite the skeletal changes observed in both models, there was no significant change to bone endocrine factors, indicating that the increased bone formation following Y2 receptor deletion was initiated by a neuronal pathway. Overexpression of NPY specifically in the ARC (using virally mediated methods) confirmed the importance of the ARC to NPY signaling in bone, showing a significant loss of bone (Baldock et al. 2005). Importantly, they also showed a simultaneous and very marked increase in body weight, just as was seen with triggering of starvation pathways and loss of leptin signaling in ob/ob and db/db mice. This observation indicated a potential role for NPY in regulation of the cortical bone changes in leptin-deficient models.

Subsequent studies assessed the skeletal effects of the leptin and NPY pathways. Y2 receptor and leptin double KO mice, Y2^−/− ; ob/ob, showed no additive effect on cancellous bone formation or volume (Baldock et al. 2006), suggesting overlap between the 2 pathways. However, other investigations highlighted differences between the 2 mechanisms. NPY⁻ mice exhibit a different bone phenotype to ob/ob mice, with both displaying increased cancellous formation (with no additive effects as with the previous studies) but inducing different changes to cortical bone. The leptin-deficient mice show a reduced whole body BMC (Steppan et al. 2000), arising from a significant reduction in cortical bone formation, leading to shorter, smaller bones, especially when the much greater body mass is considered (Baldock et al. 2006). The NPY^−/− mice, however, showed a consistent anabolic phenotype, displaying greater whole body BMC, greater bone formation, and larger cortical bones, in addition to gains in cancellous bone (Lee, Nguyen, et al. 2011). NPY^−/− ; ob/ob mice were produced to determine the interaction between leptin and NPY (Wong et al. 2013) and gave rise to very specific differences in the bones. In ob/ob mice, NPY is upregulated due to the lack of leptin receptor activation in NPY-ergic neurons (Erickson, Hollopeter, and Palmiter 1996), associated with a reduced cortical bone mass and formation. However, in the double mutant NPY^−/− ; ob/ob mice, the ob/ob cortical bone deficit is completely corrected. The findings indicate that the reduction in ob/ob bone mass results from ARC NPY signaling. NPY^−/− ; ob/ob also indicated that the VMH/SNS/β2 pathway is not influenced by NPY, as the cancellous bone was unchanged between ob/ob and NPY^−/− ; ob/ob mice. Taken together, this series of studies identified a signaling axis whereby 2 separate nuclei from the hypothalamus give rise to distinct pathways, both of which respond to an absence of the key indicator of energy balance in the body, leptin. The arcuate nucleus utilizes NPY to prevent bone formation as a means to conserve energy, while the VMH simultaneously sustains cancellous bone volume via the actions of SNS.

Role of NPY in Bone Tissue

Initial studies that have reported NPY-positive autonomic nerves are expressed in bone tissue. These nerves are typically linked to blood vessels, indicating that their function may be to do with modifying vasculature as opposed to directly interacting with the bone cells (Ahmed et al. 1993; Bjurholm et al. 1988; Lindblad et al. 1994; Sisask et al. 1996). In addition, a reduction in NPY immunoreactive nerves can be seen in the bone periosteum following sympathetic denervation (Hill and Elde 1991). The presence of Y receptors has also been demonstrated, and a reduced response to parathyroid hormone and noradrenaline in osteoblastic cells following NPY treatment again indicating a regulatory function in bone for NPY (Bjurholm 1991; Bjurholm et al. 1992).

Y1 and Y2 are the 2 NPY receptors that have been shown to be linked to bone homeostasis and are also expressed in high levels in peripheral nerves and the hypothalamus (Kishi and Elmquist 2005; Kopp et al. 2002; Naveilhan et al. 1998). Mutant mouse models were again used to better define their roles in bone homeostasis, and Y1 KO mice showed similar anabolic effects to Y2 KOs, with greater cortical and cancellous bone, via increased bone formation, although with an additional increase in bone resorption (Baldock et al. 2007). However, activity of the Y1 receptors differ in that hypothalamic-specific deletion did not affect bone homeostasis, suggesting that Y1 does not act centrally, in contrast to the activity of arcuate Y2 receptors to inhibit bone formation.

Y1 receptors were identified in osteoblast cells in vivo (Baldock et al. 2007), confirming that bone anabolism occurs as a consequence of direct Y1 activity in the bone. Furthermore, calvarial osteoblast numbers were significantly reduced following NPY treatment, a change not seen in Y1 KO mice. This further indicated that Y1 receptors functioned on osteoblast cells, an idea enhanced by osteoblast-specific Y1 KO studies, which showed increased bone formation matching that of germ line Y1 KOs (Lee, Allison, et al. 2011). Y1 receptor deletions have also indicated a potential role for the receptor in controlling mesenchymal stem cell activity and mineralization of osteoblast cultures in vitro (Lee et al. 2010).

Further to this, osteoblasts have been shown to produce NPY and that increasing this production in osteoblast-specific transgenic mice reduces bone volume and bone formation, contrasting the Y1 receptor deletion (Matic et al. 2012). Differentiation was found to be critical in regulating NPY and Y1 receptor expression and that this expression responds to loading (one of the fundamental features of bone physiology), confirming the integration of NPY into many critical aspects of the osteoblast lineage (Igwe et al. 2009).

In addition to regulating bone cell activity, a novel study showed changes to energy usage and a decline in adiposity following osteoblast-specific deletion of p38a-Mapk14 (Rodriguez-Carballo et al. 2015). These alterations to the metabolism were linked to NPY production and secretion by the osteoblasts, with i.p. NPY administration preventing the metabolic changes. These findings are consistent with evidence form Y1 receptor KO mice, showing alterations in glucose balance and insulin production (Lee et al. 2015). These findings between them therefore show a functional NPY loop within the osteoblast lineage as well as an NPY pathway from the arcuate nucleus of the hypothalamus to osteoblasts. Additionally, they suggest that whole body energy homeostasis is regulated by both osteoblastic and central NPY from the hypothalamus, highlighting the recent notion that bone cells play an important functional role in interorgan communication.

Cannabinoid Receptors

Another central pathway to be identified in bone was endocannabinoid signaling, which has recently been suggested to influence bone homeostasis by regulating adrenergic signaling. Endocannabinoids are unusual in that whereas other neurotransmitters are released in preformed vesicles, endocannabinoids are produced as and when they are needed. The 2 most important ligands are 2-arachidonoylglycerol and N-arachidonoylethanolamine (anandamide or AEA; Devane et al. 1992; Mechoulam et al. 1995), which bind to the cannabinoid receptors CB1 and CB2 (Howlett et al. 2002). The CB2 receptor is expressed much more highly in peripheral tissues (Tam et al. 2008) and is particularly abundant in osteoblasts, osteocytes, and osteoclasts, whereas the CB1 receptor is responsible for most of the cannabinoid and endocannabinoid signaling in the CNS (Mackie 2008), where it is primarily found.

Sympathetic neuron release of noradrenaline is prevented by CB1 receptor-mediated signaling on presynaptic nerve terminals in order to balance the suppression on bone formation via tonic sympathetic activity (Ishac et al. 1996; Niederhoffer, Schmid, and Szabo 2003). Interestingly, CB1 receptor inactivation caused more than just a rise in bone mineral. It also prevented osteoclastogenesis and bone resorption, which CB1 receptor–deficient mice were resistant to (indicating that the CB1 receptor controls osteoclasts via cannabinoid signaling), and also prevented bone loss induced by ovariectomy (Idris et al. 2005).

CB2 KO mice exhibit increased bone turnover, resulting in quickened age-related cancellous bone loss and cortical expansion (Ofek et al. 2006), consistent with in vitro pharmacological studies that showed osteoclasts to express activated CB2 receptors and human genetic association studies that showed a relationship between the CNR2 gene (which encodes CB2) and a decrease in female bone mass (Karsak et al. 2005; Yamada, Ando, and Shimokata 2007). The in vitro studies suggested that the CB2 receptor influences the regulation of bone mass by activating osteoblasts/stromal cells and inhibiting osteoclasts/monocytes. There is potential for therapeutic use of CB2 agonists in diseases associated with lone bone mass, as they showed an ability to reduce ovariectomy-induced bone loss (Ofek et al. 2006); however, the potential psychoactive properties of CB2 agonism would require consideration. Nonetheless, the cannabinoid system, together with the NPY system, also demonstrates that both central and tissue-specific expression of signaling receptors, working through different pathways, are fundamental to bone function.

Semaphorins

Semaphorins and their receptors (most commonly plexins and neuropilins) have been shown to play key roles in a variety of cell activities, such as differentiation, interaction, and morphology, with their most well understood function being as attractants or repellents for neurite extension (e.g., axons) and cellular migration (Derijck, Van Erp, and Pasterkamp 2010; Tran, Kolodkin, and Bharadwaj 2007). The semaphorin proteins can both be released from the cell or bound to the membrane and have been divided into 8 classes. In bone specifically, the semaphorin–plexin pathway has been shown to be key to interactions between osteoclasts and osteoblasts (Kang and Kumanogoh 2013; Negishi-Koga and Takayanagi 2012), with semaphorin 3A (SEMA3A) and semaphorin 4D (SEMA4D) particularly associated with bone homeostasis.

SEMA4D is exclusively expressed in bone by osteoclasts and prevents formation and differentiation of osteoblasts. In vitro studies have shown that this inhibition of osteoblast differentiation is a result of plexin-B1-mediated ERBB2-dependent ras homolog gene family, member A (RHOA) activation (plexin-B1 is the receptor for SEMA4D). An increase in bone mass is observed in SEMA4D null mice (or in mice lacking plexin B1; Dacquin et al. 2011; Negishi-Koga et al. 2011), and bone loss in ovariectomized mice can be reduced using an SEMA4D antibody (Negishi-Koga et al. 2011). In vivo studies have shown that these SEMA4D and plexin-B1 KO results can also be obtained by inducing osteoblast-specific expression of RHOA (Negishi-Koga et al. 2011).

SEMA3A shows the opposite function to SEMA4D in mice, being produced by osteoblasts to prevent osteoclast precursors from differentiation into osteoclasts (Hayashi et al. 2012). Unsurprisingly, therefore, a decrease in bone density is observed in mice lacking SEMA3A, while systemic administration of SEMA3A into a mouse model of menopause prevents bone loss, suggesting a potential for use as a treatment for diseases related to bone loss (Hayashi et al. 2012). SEMA3A has been shown to repel osteoclast precursors as well as inhibit their differentiation and also to increase osteoblast formation and differentiation (when it originates from paracrine or autocrine sources). Finally, more recent investigations have indicated that development of correct sensory innervation of bone tissue is dependent on SEMA3A and that neuronal SEMA3A (not osteoblastic SEMA3A) caused the bone loss in mice lacking SEMA3A (Fukuda et al. 2013). Therefore, SEMA3A controls bone metabolism via regulation of the development of sensory innervation.

Proopiomelanocortin (POMC) and Melanocortin System

Body weight homeostasis and coat color are among a wide variety of physiological roles that utilize the melanocortin system, in which peptides synthesized from POMC such as melanocortins, β-endorphin, and adrenocorticotropic hormone (ACTH) bind to melanocortin receptors (MCRs) and opioid receptors and transmit their pleiotropic effects. Melanocortin 4 receptor (MC4R), one of five MCRs found to be G-protein-coupled receptors (Beltramo et al. 2003; Nijenhuis, Oosterom, and Adan 2001), is found in particularly high levels in the hypothalamus. Patients with an absence of MC4R show decreased bone resorption (Farooqi et al. 2000), displaying a high BMD which is still observed even when accounting for obesity, a phenotypic marker for MC4R deficiency (Farooqi et al. 2000). A recent study has identified MC4R as a ligand for osteoblast-derived lipocalin-2, which acts to regulate food intake and glucose balance (Mosialou et al. 2017): Highlighting the existence of novel feedback loops from bone to brain.

In bone, few studies have focused on the expression of POMC and the peptides synthesized from it or on the effects of endogenous opioids, choosing instead to investigate the melanocortin peptides. However, recent studies have found the POMC-derived peptide ACTH to play an important role in bone, with glucocorticoid-induced osteonecrosis in the femoral head prevented by daily subcutaneous injection of 0.2 µg/kg ACTH (amino acids 1-24; Isales, Zaidi, and Blair 2010). In addition to this, it has been shown that ACTH modulates osteoblast maturation and survival by increasing vascular endothelial growth factor production through adrenocorticotropic hormone receptor (MC2R).

Collagen type I (the most important collagen in bone) is highly expressed during osteoblast differentiation. In human osteoblast-like SaOs2 cells from an osteosarcoma source, a biphasic effect on collagen transcripts was observed following ACTH administration, indicating that ACTH is not only involved in regulating maturation and survival of osteoblasts but also differentiation. At lower concentrations, ACTH inhibited osteoblast differentiation, but at 10 nM, it showed increased transcription of collagen I mRNA (and hence upregulated differentiation; Islaes et al. 2010). One study has also observed secretion and immunoreactivity of ACTH in osteoclasts in rats, though the functional importance of this remains to be determined (Sun et al. 2006). α-Melanocyte-stimulating hormone (α-MSH) appears to have contrasting effects to ACTH, with an in vitro 10⁻⁸ M administration into fetal rat osteoblasts causing no change to differentiation but enhancing proliferation. It has also been shown to induce osteoclast development from precursor cells in mouse bone marrow cultures, though no changes were observed in mature osteoclasts. The complexity of the relationship between the POMC network and bone confirms the importance of neural interactions to the control of bone homeostasis.

Bone Homeostasis and Cocaine and Amphetamine-regulated Transcript (CART)

CART is a neuropeptide precursor found primarily in the hypothalamus, adrenal glands, and pancreas that has been associated with energy use and food intake (Elefteriou et al. 2005), acting in an opposing manner to NPY. CART has been implicated in bone homeostasis and specifically with bone resorption and leptin signaling. Leptin KO mice showed simultaneously lower levels of expression of CART in the hypothalamus and a rise in bone resorption, an effect not seen in β2-AR null mice. This decrease in CART expression was overcome following intraperitoneal injection of leptin into the ob/ob mice (Kristensen et al. 1998). Similarly, an increase in bone resorption leads to an osteopenic phenotype in CART KO mice (Elefteriou et al. 2005). CART has also been associated with RANKL expression. RANKL expression is increased in CART-deficient mice compared with wild type. A local mechanism for altering central CART effects has been indicated by in vitro osteoclast differentiation studies (Elefteriou et al. 2005). These studies further identify the complexity of the leptin-deficient regulation of bone, which involves neural control of bone formation as well as resorption.

Conclusion

In recent years, substantial evidence has emerged that regulation of bone homeostasis is dependent not only on cells of the bone but also on neural inputs that link bone cells to the brain. These neuronal pathways, which originate from the hypothalamus, have not only been shown to play a critical role in regulating bone metabolism but are also vital in regulating energy availability across the entire body. These interactions enable the coordination of whole body energy balance and energy expenditure at the tissue level. It is understandable that bone is integral in this system, as it is not only a protein/nutrient and ion rich tissue (and hence a large energy/solute store for the body), but it is also a tissue that uses a significant amount of energy via the protein-synthetic actions of osteoblast cells that maintain bone structure. Interactions between energy availability and bone homeostasis are dynamic and context dependent, as is clearly evident in markers of a starvation-type situation including elevated NPY expression or reduced leptin, CART or POMC. These interactions will help in developing our understanding of how changes to energy and nutrient availability (e.g., obesity and anorexia) affect skeletal functioning and behavior. Many similarities have been observed between central pathways from the hypothalamus and local pathways within the bone tissue, indicating a potential for treatments that directly influence the bone and hence avoid the far more complicated option of targeting the hypothalamus. Due to the extent to which some of these pathways influence bone metabolism and homeostasis, it is possible that in the following years, such treatments will become powerful therapeutic agents.