Energy Metabolism of Bone

Abstract

Biological processes utilize energy and therefore must be prioritized based on fuel availability. Bone is no exception to this, and the benefit of remodeling when necessary outweighs the energy costs. Bone remodeling is important for maintaining blood calcium homeostasis, repairing micro cracks and fractures, and modifying bone structure so that it is better suited to withstand loading demands. Osteoclasts, osteoblasts, and osteocytes are the primary cells responsible for bone remodeling, although bone marrow adipocytes and other cells may also play an indirect role. There is a renewed interest in bone cell energetics because of the potential for these processes to be targeted for osteoporosis therapies. In contrast, due to the intimate link between bone and energy homeostasis, pharmaceuticals that treat metabolic disease or have metabolic side effects often have deleterious bone consequences. In this brief review, we will introduce osteoporosis, discuss how bone cells utilize energy to function, evidence for bone regulating whole body energy homeostasis, and some of the unanswered questions and opportunities for further research in the field.

Osteoporosis

Osteoporosis is a major public health condition commonly related to aging. It is a skeletal disorder characterized by reduced bone mineral density (BMD) and microarchitectural deterioration of bone tissue (Kanis et al. 2013). The reduction in BMD is caused by an imbalance between osteoclast-mediated bone resorption and osteoblast-mediated bone formation (i.e., uncoupled remodeling). Osteoporosis compromises bone strength, leading to an increased susceptibility to fractures. In the United States, it is expected that the number and cost of fractures will increase almost 50% by the year 2025 (Burge et al. 2007). In addition to occurring with aging and after menopause, osteoporosis can be attributed to many clinical conditions that impact whole-body energy metabolism, such as diabetes mellitus (DM) and anorexia nervosa (AN).

Postmenopausal Osteoporosis

Postmenopausal osteoporosis is associated with low BMD and changes in body composition. Postmenopausal women are predisposed to fracture, in particular at the hip and wrist regions (Burge et al. 2007). During menopausal transition, women usually experience accelerated bone loss, mostly due to changes in sex hormone levels. Estrogen deficiency is an important factor that contributes to postmenopausal osteoporosis development. The drop in estrogen levels induces osteoclast recruitment and activation by increasing receptor activator of nuclear factor κ β ligand production by osteoblastic cells, which favors bone resorption activity and consequently causes rapid bone loss (Clarke and Khosla 2010). The onset of postmenopausal osteoporosis is also associated with age-related changes in body composition. Lean mass, which exerts a positive influence on the skeleton, decreases with age and becomes more evident in women over 65 years (Zhang et al. 2013; Chen et al. 2015). On the other hand, there is an increase in body fat with advancing age (Ho et al. 2010; Sornay-Rendu et al. 2012). However, the increase in adipose tissue may have different effects on bone mass (Compston 2015). Fat mass can have a positive association with BMD in postmenopausal women (Chain et al. 2017; Ho-Pham et al. 2010), suggesting that a slight accumulation of body fat may be advantageous for these women and it would not induce bone loss (Chain et al. 2017). However, low trauma fracture is observed in obese postmenopausal women with normal BMD (Premaor et al. 2010; Compston 2015). Obesity is associated with the production of a specific set of cytokines and adipokines by visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) that may have a negative or positive effect on bone mass (Russell et al. 2010). Thus, it needs to be taken into consideration that not only the amount of fat mass but also VAT/SAT proportion is important to postmenopausal women bone health, as it will determine the type of cytokines and adipokines secreted having a diverse impact on bone mass.

AN

AN is a psychiatric disease characterized by a voluntary food restriction leading to severe weight loss and, consequently, bone loss with increased fracture risk. Bone loss is caused in part by decrease in mechanical loading, as these patients exhibit a loss of lean and fat mass compartments. Concomitantly, marrow adipose tissue (MAT) expands in AN and has a negative association with bone mass. This increase in MAT is an evident feature of AN, and it might represent a compensatory mechanism related to the lack of peripheral fat or a shift toward fat lineage instead of bone cell as a result of impaired osteoblastogenesis (Bredella et al. 2009). Curiously, MAT content is decreased in women who had begun to recover their nutritional status. Therefore, this may indicate that MAT is more susceptible to nutritional changes than the other fat depots, and its reduction can induce a bone mass improvement in these patients (Fazeli et al. 2012).

DM

DM is a metabolic disease characterized by a hyperglycemic state. Bone loss is one of the classical complications of type 1 DM (McCabe 2007; Starup-Linde et al. 2016; Vestergaard 2007). Conversely, type 2 DM patients usually have an adequate or increased BMD (Vestergaard 2007; Starup-Linde et al. 2016; de Araujo et al. 2017). However, fracture risk is increased in both types of diabetes even though there are discrepancies in bone phenotype (Starup-Linde et al. 2016). Nutritional status is one factor that differs between types of diabetes: unlike type 1 DM, type 2 DM patients are overweight or obese with visceral fat accumulation (de Araujo et al. 2017). Higher body weight exerts a greater mechanical loading on the skeleton (Reid 2010), and it can explain the absence of bone loss in these individuals. On the other hand, adipose tissue produces certain cytokines and adipokines, which have a deleterious effect on bone (Russell et al. 2010) and predispose obese patients to fracture. In type 1 DM animal models, as in caloric restriction models, trabecular bone mass reduction is associated with an increase in MAT (Botolin and McCabe 2007). However, MAT expansion is not observed in type 1 DM condition in humans (Slade et al. 2012). Likewise, some have reported that increased MAT is not a feature in obese type 2 DM subject (de Araujo et al. 2017). Yet, others have described a greater MAT content in lumbar spine and femoral metaphysis of morbid obese subjects with type 2 DM using insulin therapy or antidiabetic oral agents (Yu et al. 2017). This highlights that MAT function and its determinants in diabetes condition need to be better understood. In addition to aging, MAT was related to serum lipid levels in type 1 diabetes subjects (Slade et al. 2012), and hyperglycemia has been shown to be a determinant factor of MAT in type 2 diabetes condition (Yu et al. 2017). Moreover, available data in the literature did not observe any association between MAT and homeostatic model assessment of insulin resistance (HOMA-IR; Yu et al. 2017; de Paula et al. 2015; de Araujo et al. 2017). Another important aspect to be considered is the relationship between saturated and unsaturated lipids of the MAT and bone mass. Saturated lipids are higher in type 2 DM subjects with previous fractures (Patsch et al. 2013), suggesting that the lipid profile of MAT might be a predictive marker of bone health (Patsch et al. 2013; Yeung et al. 2005).

Bone Cell Energetics

Due to the impact of metabolic diseases on bone health, it is not surprising that changes in bone cell energy metabolism modify differentiation and function. Osteoblasts, osteocytes, and marrow adipocytes all originate from common progenitor mesenchymal stem cells. Osteoclasts, which are bone-resorbing cells, have a monocytic origin (Bianco and Robey 2004; Xiao et al. 2015). The fuel sources and pathways that are utilized by these cells to generate adenosine 5′-triphosphate (ATP) are currently under intense scrutiny. ATP is the most widely used nucleoside triphosphate to provide chemical energy for biochemical reactions. It is generated in the cytoplasm through glycolysis and in the mitochondria through oxidative phosphorylation. The specialized cells that make up bone need to generate substantial amounts of ATP to maintain a normal healthy skeleton. During development, osteoblasts produce and secrete α-1 type 1 collagen and mineralize bone (Rodan 1992). Osteocytes, which are embedded in mineralized bone, secrete a number of osteokines like sclerostin and function in a highly hypoxic environment as mechanical sensors (Weivoda, Youssef, and Oursler 2017). Osteoclasts, on the other hand, need to generate H⁺ to acidify and help resorb bone, hydrolyzing ATP to generate resorption pits. In this section, we will review the current understanding of the exogenous fuel sources and bioenergetics pathways that are utilized by cells in the skeleton to meet this ATP need.

Osteoblasts

Recent studies have shown that bone marrow stromal cells utilize oxidative phosphorylation to generate ATP preferably over glycolysis, while concomitantly increasing antioxidant enzymes generated to combat the detrimental effects of reactive oxygen species (Chen et al. 2008; Shum et al. 2016). In contrast to some of these data, Wnt3a, a known osteogenic factor specifically induces glycolysis through an mTORC2-dependent pathway (Esen et al. 2013). There is an increase in Glut1, the insulin-independent glucose transporter, along with several glycolytic pathway enzymes, in a number of precalvarial osteoblasts and bone marrow stem cell lines at the later stages of osteoblast differentiation. The preferential use of glycolysis to generate ATP is akin to the Warburg effect that has been described in a variety of cancer cells (Diaz-Ruiz, Rigoulet, and Devin 2011). The reasons for this seem to be multifold, namely, need for (1) generating nicotinamide dinucleotide (NAD+) when lactate is generated from pyruvate, (2) generating ribose precursors and nicotinamide adenine dinucleotide phosphate (NADPH) through pentose phosphate pathway, and (3) generating ATP at a faster rate. However, which of these are relevant for osteoblasts is currently not clear.

Early studies using ex vivo techniques to measure bone cell energetics in the 1950s identified that parathyroid hormone (PTH) induces lactate via aerobic glycolysis. More recently, it was shown that this mechanism is through induction of insulin-like growth factor 1 (IGF-1) and glycolytic gene upregulation (Neuman, Neuman, and Brommage 1978; Esen et al. 2015). However, others have shown increases in both oxidative phosphorylation and glycolysis during the late stages of in vitro osteoblast differentiation in calvarial-derived osteoblast cells (Guntur et al. 2014).

The substrates used in these studies are exogenously added glucose, pyruvate, and glutamine (Guntur et al. 2014; Komarova, Ataullakhanov, and Globus 2000). All of these substrates are metabolized and processed through glycolysis and Krebs cycle. Glutamine, used by osteoblasts as an alternate fuel source, is converted to α-ketoglutarate to enter the Krebs cycle through the process of glutaminolysis. Karner et al. (2015) have shown that this process is also Wnt signaling mediated and is essential for osteoblast differentiation. However, the transporters and enzymes involved in this pathway in osteoblasts still need to be identified. Furthermore, more recent studies have also identified that early osteoblast differentiation in bone marrow stromal cells (BMSCs) requires intracellular fatty acids as energy sources. Other studies have shown that oxidation of fatty acids by osteoblasts for generating energy is controlled by Wnt-Lrp5 signaling, and a loss in this process will result in decreased bone mass along with increases in whole-body fat mass (Rendina-Ruedy, Guntur, and Rosen 2017; Frey et al. 2015).

Osteocytes

Osteocytes make up close to 95% of the cells in the adult skeleton (Bonewald 2011). There are currently no studies that have identified the bioenergetic pathways in these cells. Osteocytes are embedded in a hypoxic environment. This, combined with the fact that they are terminally differentiated osteoblasts leads to the hypothesis that they would be highly glycolytic in their energy production. These cells have been shown to generate protons and acidify their microenvironment (Jahn et al. 2017), though there are currently no published studies that have examined energy metabolism in osteocytes.

Osteoclasts

The origin of the osteoclast is the bone marrow macrophage. The fully differentiated osteoclast forms a sealing zone between its ruffled membrane and the area it needs to resorb. It then proceeds to generate an acidic environment within the sealing zone containing cathepsin K and matrix metalloproteases to initiate remodeling (Boyle, Simonet, and Lacey 2003). There are two studies that have identified the bioenergetic pathways in osteoclasts. The first study identified the need for glycolysis for normal osteoclast differentiation, which seems to be an underlying theme for all the different cells in the skeleton (Indo et al. 2013). The authors observed an increase in Glut1 and other glycolytic enzymes with osteoclast differentiation. The same study also identified an increase in glutaminolysis as seen with osteoblast differentiation in response to Wnt signaling. Modulating either pathway leads to a decrease in osteoclast differentiation and function. In a more recent study, Lemma et al. (2016) showed that there is increased oxidative phosphorylation with differentiation in osteoclasts with an increase in mitochondrial mass and biogenesis. This is followed by a need for glycolysis to properly function during resorption.

The sequence of events described in these studies suggests that the need to shift to glycolysis might be based on the function of these specialized cells. The generation of ATP for pumping protons into the sealing zone may need to occur at a rate that exceeds the ability of the cells to promote mitochondrial biogenesis. Thus, glycolysis in the cytoplasm might be a more readily available source of ATP than if it is generated through oxidative phosphorylation in the mitochondria and transported out through ADP/ATP transporters.

Relevance of studying energy metabolism

Metabolic pathways can control gene expression through epigenetic modifications as most of the substrates and cofactors that are necessary for epigenetic modifications are generated through bioenergetic pathways. For example, NAD+ (oxidized form) is an essential electron acceptor for a number of dehydrogenases to generate nicotinamide adenine dinucleotide (NADH). In the mitochondria, NADH functions as the electron donor in the electron transport chain to generate a ΔpH and the mitochondrial membrane potential necessary for ATP generation. NAD+ is also absolutely essential for the enzymatic activity as a cofactor for sirutinin enzyme–mediated deacetylations (Imai et al. 2000). The effects of NAD+ on bone metabolism are currently not known though there is a keen interest in using nicotinamide mononucleotide (NAM) or nicotinamide riboside which are precursors that have been shown to affect the aging process. More importantly, long-term treatment of wild-type C57BL/6J mice with NAM showed significantly higher BMD compared to age-matched controls, suggesting that there could be a beneficial effect of these compounds on bone mass (Mills et al. 2016).

Acetyl CoA generated in the Krebs cycle in the mitochondria is also a crucial cofactor and substrate for acetylation reactions. It is used by histone acetyltransferases as an acetyl donor and known to exist in two separate pools: one in the mitochondria and the other in the cytoplasm. Karner et al. (2015) have recently shown that reduced nuclear acetyl CoA leads to suppression of osteoblast gene expression. This could be one potential explanation for the need for the cells to switch to glycolysis with osteoblast differentiation (Mills et al. 2016). Therefore, control of metabolic pathways and the flux through which these substrates can be increased or decreased has far-reaching consequences in controlling not only ATP generation but also gene expression. These are in turn controlled by substrate availability and cellular ATP demand, setting the tone for which pathway needs to be upregulated or downregulated.

Missing pieces

One of the crucial pieces of data that are lacking from most of these studies is the status of mitochondrial respiration and mitochondrial membrane potential, which at any point of time is a good indicator of cellular energetic pathways. The switch to glycolysis that is observed with osteoblasts and osteoclasts would indicate that there is a defect or insufficiency in mitochondrial metabolism, but whether this is because of some dysfunction that occurs with differentiation or due to specific programming is not clear. Studying ATP generation and mitochondrial dynamics with differentiation and identifying the amount of ATP that is generated from these pathways will be crucial for obtaining a complete bioenergetic picture.

Current studies have focused on identifying the pathways during normal osteoblast differentiation, so there needs to be a push to study the role of these energetic pathways in pathophysiological conditions. More careful study of bioenergetic pathways at the molecular level may identify novel mechanisms of pathological bone loss and subsequently novel pathways to target for therapies.

Regulation of energy metabolism by peripheral tissues

Energy balance is defined by the regulation of food intake, nutrient storage, and energy expenditure. These processes are all controlled by central mechanisms, and feedback from peripheral tissues is an essential part of these regulatory networks. Leptin, for example, is secreted from adipocytes that are storing lipids and acts in the hypothalamus to reduce food intake and increase energy expenditure (Roh, Song, and Kim 2016). On the other hand, during exercise, interleukin 6 (IL-6) is secreted from skeletal muscle and promotes sympathetic nervous system (SNS)-mediated mobilization of fat stores. Intracerebroventricular administration of IL-6 increases energy expenditure, and deletion of IL-6 promotes obesity (Febbraio and Pedersen 2002; Roh, Song, and Kim 2016). Alternately, low glucose levels in the blood (lowered by utilization or storage by tissues) stimulate the hypothalamic–pituitary–adrenal axis and the SNS to mobilize glucose and fatty acids (Routh et al. 2014). The latter example, however, is a more passive way in which peripheral tissues modify energy metabolism (i.e., simply by using energy). Bone will certainly use energy, causing a reduction in available fuel and further stimulation to eat or mobilize fuel stores, but active mechanisms through which bone regulates energy metabolism are currently being investigated. Although we have not currently identified ways in which the bone can feedback directly to the brain, there is evidence that bone can regulate insulin secretion and sensitivity, which we will summarize below.

Involvement of Bone in the Regulation of Energy Metabolism

Osteocalcin (OC) is the second most abundant protein in bone, after type 1 collagen, and its serum levels have long been used as a marker for bone turnover. However, a role of OC in regulating insulin sensitivity has been established and is well reviewed (Booth et al. 2013; Motyl, McCabe, and Schwartz 2010; Wei and Karsenty 2015). These findings were initiated by the observation OC knockout mice were overweight, with high blood glucose (Lee et al. 2007). However, these mice also had lower insulin levels, suggesting they were not simply prone to the classical obesity-induced insulin resistance, in which insulin levels are generally high (Lee et al. 2007). Instead, the authors did indeed find insulin resistance (marked by impaired insulin tolerance test and reduced glucose infusion rate in hyperinsulinemic euglycemic clamp studies) in the OC knockout mice, along with the intriguing finding that OC protein itself could stimulate insulin secretion from pancreatic β cells (Lee et al. 2007). More recently, Ferron et al. (2010) demonstrated that undercarboxylated OC release from bone during resorption stimulates insulin production in the pancreas. This combined with the known role of insulin in promoting bone remodeling suggests a bone–pancreas feed-forward loop regulating bone and insulin secretion. Although studies in humans point to the carboxylated form of OC as having a more prominent role in glucose metabolism, these studies nonetheless suggest that metabolic homeostasis can be modified by hormones released from bone (Booth et al. 2013).

More recently, non-OC-mediated mechanisms have been proposed for bone in the regulation of energy homeostasis. Ablation of osteoblasts with diphtheria toxin expression under control of the OC-Cre promoter resulted in hypoinsulinemia, hyperglycemia, and impaired insulin sensitivity (Yoshikawa et al. 2011). This finding is similar to the metabolic phenotype of the OC knockout mice; however, the authors also observed increased energy expenditure and reduced gonadal fat mass, which could not be rescued with OC administration. This suggests that osteoblasts or osteocytes produce another hormone(s) that contributes to the regulation of energy metabolism. Furthermore, increasing evidence suggests that neuropeptide Y expressed from osteoblasts is one of these essential molecules mediating changes in energy expenditure (reviewed in this issue; Rodriguez-Carballo et al. 2015; Baldock et al. 2007; Lundberg et al. 2007; Lee et al. 2015). The above examples modulated energy expenditure and glucose homeostasis, but it is becoming evident that bone can have impacts on brown/beige adipose tissue as well. In a recent study by Brun et al. (2017), deletion of peroxisome proliferator activated receptor gamma (PPARγ) in osteocytes (utilizing the Dmp1-Cre mouse) resulted in high bone mass, improved glucose tolerance and insulin sensitivity, and increased beiging of white adipose tissue. These mice were also resistant to high-fat diet–induced obesity and steatosis and could not be explained by changes in OC. The authors postulated that osteokines under the control of PPARγ were responsible for the effects, and although not tested directly, they identified bone morphogenetic protein 7 (BMP7) as a candidate molecule (Brun et al. 2017).

In addition to bone cells themselves regulating energy metabolism, bone marrow adipocytes within the bone marrow niche have been demonstrated to be important contributors to glucose control. In particular, marrow adipocytes expand in size and number during metabolic diseases, such as AN and DM. In part, this is thought to be due to altered mesenchymal lineage selection toward the adipogenic lineage at the expense of the osteogenic lineage. However, MAT is the major source of adiponectin in calorie restriction models, suggesting it too has a role in modifying insulin sensitivity (Cawthorn et al. 2014). Furthermore, stimulation of the SNS through cold exposure, or treatment with antipsychotic drugs, promotes marrow fat loss, suggesting fat is lipolyzed for fuel in these cases (Scheller et al. 2015; Motyl et al. 2015). How such acute changes in marrow fat globally modulate energy metabolism, however, is unknown.

Summary and Conclusions

Bone remodeling is a dynamic process that, when uncoupled, leads to osteoporosis. Although osteoporosis occurs with conditions such as aging and menopause, it can also occur secondary to metabolic diseases such as DM and secondary to treatment with pharmaceuticals that impact metabolism. Bone requires a significant portion of the available fuel of an organism; thus, it is not surprising that glucose homeostasis, adipose tissue metabolism, and bone remodeling appear to be tightly linked. There is increasing evidence that bone can regulate organismal energy metabolism through pathways other than just OC-mediated alone. However, we have not yet seen strong evidence that bone regulates central control of energy balance. The importance of studying these pathways is at least 2-fold. First, treatments that modulate energy metabolism may have unexpected impacts on bone and vice versa. For example, atypical antipsychotic drugs have known metabolic side effects, and these lead to trabecular bone loss in rodent models. Alternately, osteoporotic therapies may in turn have deleterious metabolic side effects. Second, studying the energetics of bone cells themselves will lead to a better understanding of the etiology of pathologic conditions like diabetes and the mechanism of action of bone therapies like PTH. In conclusion, understanding the energy metabolism of bone may lead to novel treatment regimens and/or therapeutics with improved efficacy and reduced side effects.