Bone is the primary tissue in the skeleton system, providing the function of structure support, hematopoiesis location, and mineral storage. To perform these functions, bone has a complex structure that contains compact bone and cancellous bone, with the composition of organic and inorganic composites containing living cells [5]. The bone remodeling process is a cascade of regular activities in which three cell types, including osteoblasts, osteoclasts, and osteocytes, play crucial roles (shown in Figure 1). Bone undergoes continuous resorption carried out by osteoclasts and formation by osteoblasts, which are linked cell activities that form a dynamic equilibrium in bone tissue [6]. To briefly summarize this complex process, osteoclast precursors are activated to start resorption stages by cytokines such as receptor activator of nuclear factor-kB (RANK) and macrophage colony-stimulating factor (M-CSF) released by bone lining cells upon receiving a biochemical signal such as hormones (e.g., parathyroid) or mechanical signal (due to bone damage). Osteoclasts, specialized cells developed from the monocyte/macrophage lineage, can contact the bone matrix to form a sealed bone resorption area, playing a role in degrading bone tissue [7]. The activity of the osteoclast is terminated by the formation of resorption pits on the bone surface. During the formation process, osteoblasts deposit the osteoid into the resorbed site to form a mineralized matrix. Osteoblasts are then either apoptotic or buried as osteocytes in new bone layers. It is crucial to maintain a balance between osteoclasts and osteoblasts. Bone remodeling is a dynamic and lifelong process. An imbalance in this process leads to bone disorders. It can cause several diseases, such as overactive osteoclasts and bone turnover, which would lead to osteopenia and osteoporosis, while abnormal osteoblasts elevate the risks for hyperosteogeny and tumor [3,8].

Cell differentiation is an integral part of bone homeostasis, as the osteoclast is derived from hematopoietic cells, while osteoblasts are produced from the differentiation of bone marrow stem cells (MSCs). The differentiation from MSC to osteoblast and from monocyte-macrophage lineage to osteoclast involves metabolic changes and adaptions. Metabolomics is one of the key methodologies to study these processes and deepen the understanding of osteogenesis and osteoclastogenesis and their relationship with bone diseases. MSCs, the precursor of osteoblasts, predominantly utilize glycolysis as the energy source within the bone marrow; however, the energy metabolism shifts to oxidative phosphorylation during the differentiation [9,10]. Hypoxia-inducible factor 1 α (HIF-1α) works as a regulator of MSC energy metabolism during osteogenic and chondrogenic differentiation [11]. HIF-1α can promote glycolysis and inhibit oxidative phosphorylation (OxPhos) and is most active in undifferentiated MSCs [12]. MSCs have been shown to grow and proliferate in the glycolysis pathway in hypoxic condition. At the same time, they undergo a metabolic shift to the TCA pathway to supply the required energy to differentiate into the bone cells [9]. As a result, a higher OxPhos activity with an unaltered or decrease in glycolysis happens in MSCs differentiation [9,12,13]. In mature osteoblasts, glycolysis produces most of the needed energy for osteoblasts to maintain the functionality [14]. The well-known osteogenic factor, bone morphogenetic protein, can boost glucose metabolism and therefore enhance the development of bone and cartilage [15].

During osteoclastogenesis, both glycolysis and OxPhos were found increased, as indicated with a higher glucose consumption and lactate production [16]. This means that glycolysis and mitochondrial processing both play a role in the differentiation and absorption. Increased expression of glycolytic genes, including hexokinase, phosphofructokinase, and pyruvate kinase, was found during osteoclastogenesis [17]. The final osteoclasts may become apoptosed, which makes them difficult to study [18]. However, there is some evidence which indicates an alternative fate for RANKL-stimulated osteoclasts-osteomorphs production, [19]. The lifespan of osteoclasts is generally considered to be as long as 2 weeks [20]. Nevertheless, Jacome-Galarza et al. described a mechanism upon which osteoclasts could be maintained in bone, leading to compensate for osteoclast deficiency in vivo [21].

Lipid metabolism can also alter the differentiation and function of bone cells. Long-chain polyunsaturated fatty acids, such as docosahexaenoic acid and arachidonic acid, can directly suppress the differentiation and activity of osteoclasts [22]. As part of lipid rafts to control the signal transduction during osteoclastogenesis, cholesterol can alter the RANK-RANKL signal transduction, resulting in increased bone resorption under a high content of cholesterol [23]. Similarly, phosphoinositide, a membrane lipid, regulates calcium signaling and influences osteoclast differentiation [23]. The changes of 18 metabolites, most of which were lysophosphatidylcholines, were detected in the study of differentiation inhibition with estradiol [24]. The enzymatic expression changes explained this inhibition during osteoclast differentiation and oxidative/anti-oxidative imbalance during osteoclast proliferation. It also should be pointed out that lipids are used as an energy source which in turn can influence cell functions. Low lipid levels in the skeletal stem cells environment leads to chondrogenesis rather than osteogenesis due to the osteoblasts dependency on the fatty acid oxidation (FAO) [25].

The metabolism studies on osteogenesis and osteoclastogenesis provide a guide to regulate the differentiation of precursor cells by altering the metabolic pathways. Monitoring the cell function upon exposure to a change in its surrounding environment allows for controlling cells’ behavior. Osteogenesis differentiation from MSCs can be guided by parathyroid hormone by enhancing glycolysis to increase bone formation [24]. Lee et al. attempted to regulate adipocyte and osteoblast differentiation from MSCs using several metabolites. Among them, ergosterol peroxide and 9,11-dehydroergosterol peroxide were found to inhibit the differentiation toward adipocytes. Diterpenes dehydroabietic acid, 7-oxocallitrisic acid, and pimaric acid, on the other hand, induced osteoblasts differentiation [24]. The effect of high and low amounts of energy source, oxygen, and ROS as environmental factors on a deviation of cellular behavior has been well reviewed recently [6].

The close connection between the immune system and bone tissue generated the conception of osteoimmunology [26]. Immune cells have a remarkable effect on bone physiology by acting on osteogenesis and osteoclastogenesis. The activity of bone resorption is also regulated by interleukins, cytokines secreted by immune cells. Among all immune cells, macrophages play an essential regulatory role in bone regeneration. Macrophages are divided into two polarized extremes, pro-inflammatory M1 and anti-inflammatory M2 [27]. Extracellular amino acid levels can alter the immune system functions through altering macrophage phenotypes. Arginine (Arg), ornithine, tryptophan, and glutamine can be considered amino acid-based immunomodulators [28]. Arginine metabolism plays a critical role in macrophage polarization and characterization. The prominent metabolic feature of M1 macrophages is Arg catabolism to nitric oxide (NO) and citrulline using nitric oxide synthase (NOS). While for M2 macrophages, Arg is catabolized to ornithine and urea through arginase [27]. Therefore, the catabolic and anabolic activity of NOS and arginase processes can regulate the polarization of macrophage and thus influence the inflammation and regeneration in hard tissue [29]. Indeed, the chronic inflammation of the target tissue can be pharmacologically controlled by regulating NO and Arg production [30].

In bone regeneration processes, activation of M1 macrophages in the early inflammatory phase leads to the secretion of inflammatory cytokines such as tumor necrosis factor α (TNF-α), Interleukin 6 (IL-6), and IL-1b, which results in activation of the osteoclastogenesis cascade with consequential bone resorption [31]. The M1 macrophage effect on mesenchymal stem cell (MSC) osteogenesis through the COX-2-prostaglandin E2 pathway has also been reported [32]. M2 macrophages are involved in the secondary stages of bone repair, which can lead to fibrous capsule or bone formation. Fibrous capsule formation prevents bone marrow stem cells being connected to the surface of a biomaterial for further progress in osteogenesis. Therefore, the domination of the inflammatory phase of macrophages guarantees the long-term failure or success of bone regeneration in the subsequent stages. The protracted stage of the M1 phenotype results in more fibrous inducible cytokines are amplified and fibrous capsules are formed, helping M2 phenotypes [31]. The immune response elicited by primary M1 macrophages in the early stages determines the decision of secondary M2 macrophages to secrete specific cytokines. In summary, macrophages positively contribute to bone healing by secretion of osteogenesis-related growth factors or negatively interfering through fibrous-induced inflammatory cytokines.

From an energy metabolism point of view, macrophages are constantly switching from glycolysis to oxidative phosphorylation (OxPhos). It was proved that M1 phenotypes use glycolysis and pentose phosphate pathways to uptake required ATP. M1 phenotypes lead to metabolic perturbations in several metabolites of the TCA cycle, including citrate, succinate, and itaconate. Prolonged M1 polarization can be selected as a metabolic marker for the diagnosis of inflammatory diseases such as rheumatoid arthritis and metabolic disorders such as diabetes and osteoporosis [28]. On the contrary, glycolysis has been shown to not be relevant for M2 polarization, while the consumption of glutamine Gln can be upregulated to fuel the TCA cycle [33]. Elevated FAO and OxPhos pathways indicate M2 macrophage activity and sustain the inflammatory response [27]. Gln metabolism has been reported as the synergistic support for macrophage activation through M2 phenotype differentiation [34]. Macrophages have the potential to uptake different types of lipids, including VLD, LDL, and oxidized lipoproteins. Lysosomes convert the consumed lipids to free fatty acids and cholesterol, which eventually are involved in OxPhos and ETC pathways. As lipid-based macrophage modulators, fatty acid synthesis (FAS) and FAO guide M1 and M2 macrophage polarization, respectively [28]. This suggests that the tracking of metabolites involved in macrophage energy homeostasis can serve as a marker to evaluate a transition from the M1 phenotype towards M2. In fact, it has been demonstrated that metabolomics, through an elucidation of changes in macrophage polarization, provides accurate information about an inflammatory disease [28,35].

Osteoporosis (OP) is the most known and common bone disease. OP occurs worldwide in all populations, with a higher prevalence among postmenopausal women and aged people. A low bone mineral density, which increases fracture risk, is the hallmark of OP. OP is usually characterized by metabolic disorders of bone tissue, and usage of metabolomics to study pathophysiology has gained popularity [36]. Animal models are commonly used to study the molecular mechanisms of OP in vivo. Ovariectomized (OVX) mice is an animal model of OP that mimics postmenopausal women with low BMD. Using this model, ovarian extraction results in a decrease in estrogen and progesterone, with consequences elevating the rate of bone loss [36]. Lipid metabolisms, especially arachidonic acid metabolism, linoleic acid metabolism, and glycerophospholipid metabolism, are impacted by the decrease in estrogen [37,38]. Polyunsaturated fatty acids (PUFAs) are well known for their influence on BMD [39]. Two different families of PUFAs, n-3 PUFA, derived from α- linolenic acid, and n-6 PUFA, derived from linoleic acid, are frequently occurring in metabolomics studies with their opponent functions [22]. In fact, elevated levels of arachidonic acid (AA), an n-6 PUFA, have been observed in oophorectomized rats and postmenopausal women with OP [37,38,40]. Arachidonic acid can stimulate the expression of receptor activator of NF-kB ligand (RANKL), leading to a high plasma RANKL level [22]. As the cytokine essential for osteoclast differentiation, a high concentration of RANKL can remarkably promote osteoclastogenesis, resulting in the loss of bone tissue [41]. On the contrary, docosahexaenoic acid (DHA), an n-3 PUFA, suppresses osteoclast formation from human CD14-positive monocytes by the reduction of key signaling transduction pathways of kinases (JNK, ERK, and p38 MAPKs) [22,42]. Thus, DHA inhibits osteoclastogenesis by blocking RANKL-induced activation from primary macrophages. At the same time, prostaglandin and leukotrienes, the downstream metabolic products of arachidonic acid, were also proved to impact Wnt signaling (osteoblastogenesis) and osteoclastic resorption, respectively [43,44]. A high-level accumulation of other lipid metabolites has been found in postmenopausal women and oophorectomized rats as well [45,46,47]. Lipid metabolisms not only impact osteoclastogenesis but also influence osteoblastogenesis. The increased lipid oxidation, followed by high oxidative stress, can activate peroxisome proliferator-activated receptor γ, consequently inhibiting osteoblastogenesis and promoting adipogenesis [48]. PUFAs also play various roles in osteoblastogenesis. Fatty acids can activate peroxisome proliferator-activated receptor γ (PPARγ) [49]. PPARγ plays a pivotal part in cell-fate determination, guiding the MSCs to differentiate into adipocytes [50]. This could explain the bone loss of the patients with a high-fat diet. AA decreases the expression of osteogenic markers and the osteoprotegerin/RANKL ratio, causing the appearance of adipocytes in MSC differentiating during osteoblastogenesis [51]. However, n-3 PUFA such as DHA and eicosapentaenoic acid do not have such an impact on MSCs [51,52]. In an attempt to characterize the pathological mechanism of the postmenopausal OP, metabolomics was employed to analyze the OVX mice-related femur tissue. The obtained data indicated altered levels of 93 lipid metabolites such as fatty acyls, glycerolipids, glycerophospholipids, sphingolipids, and sterols, among which levels of many fatty acids were increased in the OVX model [53]. To sum up, PUFAs play a critical role in osteoporosis by (1) promoting osteoclastogenesis through the expression of RANKL and (2) altering the differentiation of MSCs by inducing adipogenesis.

Ageing is another leading cause of low BMD. The fractures due to senile OP are highly life-threatening for the aged population, especially those over the age of 70 [1]. It is known that the functionality of osteoblasts, adipocytes, and osteoclasts changes with ageing. A higher adipogenesis level was estimated with the accumulation of bone marrow fat, not found in postmenopausal OP, which may be a critical cause for lower osteoclastogenic activity [54,55]. An in vitro study verified that adipocytes, with their metabolites, inhibit osteoblast differentiation by downregulating histone acetylation [56]. Again, lipid metabolism appears to have an essential role in the age-associated reduction of BMD, as demonstrated by a metabolomics study on senescent osteoblasts [57]. The authors identified the downregulation of n-3 PUFAs in the fatty acid metabolism due to the severe oxidative stress damage. The decrease in n-3 PUFAs, as explained before, could result in an elevation of bone loss through osteoclasts.

At the clinical level, a plasma metabolites profiling conducted on 1552 individuals to identify BMD-associated metabolomic markers detected a higher level of creatine, dimethylglycine, and glycine [58]. The authors suggested that these metabolites were causally negatively associated with BMD by altering bioenergetic processes as well as glycine and threonine metabolism pathways. As a major component of the protein collagen, hydroxyproline is also considered an OP-specific marker [59]. Serum metabolite components of postmenopausal women with low BMD were investigated using CE-MS and showed that the level of hydroxyproline could be a marker of OP [60]. Increased concentration of hydroxyproline indicates the degradation of collagen type I from the bone matrix and it was reduced after treatments [61]. Similarly, other catabolites developed in bone from collagen, such as deoxypyridinoline (collagen stabilizer) and pyridinoline (cross-linking compound of collagen fibers), can also serve as the markers of OP [59]. However, changes in their levels are the results rather than the causes of OP; therefore, they are more suitable for the development of diagnosis but not treatment.