Introduction

The shape and mass of the skeleton are dictated by the actions of osteoblasts and osteoclasts, which generation and function are regulated by osteocyte-produced signaling molecules [1, 2]. Dis-regulation on the pattern of gene expression and protein synthesis and secretion by these cells lead to abnormalities in bone acquisition during growth, in peak bone mass, and in the maintenance of bone mass with aging. Further, changes in bone mass and shape also alter bone strength.

Bone cell generation and function are regulated in a paracrine manner by cytokines produced locally by other cells in the bone marrow and by systemic factors such as hormones produced by endocrine glands [2]. Conversely, in addition to the bone intrinsic functions of osteoblasts, osteocytes, and osteoclasts, it is now recognized that bone cells are able to secrete molecules that have paracrine and endocrine functions, that is, can alter cellular responses of bone marrow cells and of cells in distant tissues and organs. Besides osteoclasts, osteoblasts, and osteocytes, other cells present in the bone microenvironment are known to exhibit paracrine roles and even to secrete molecules that can affect other organs. These cells include osteomacs, cells present in the blood vessels walls (for example, endothelial cells and smooth muscle cells), nerve cells, and cells present in the bone marrow, including bone cell progenitors, not described in this manuscript (for reviews on these factors, see [3,4,5,6]).

Bone-derived factors have been associated with several non-skeletal disorders, such as altered adipocytes and β-cell function, insulin metabolism regulation, fertility, food intake control, angiogenesis, immune system, and phosphate metabolism [7,8,9]. Interestingly, the RANKL/RANK system has been shown to participate in the immune response and in cancer initiation, progression, and metastasis (including bone metastasis), in particular, for breast and prostate cancer [10,11,12]. Bone cell-derived factors are also involved in other skeletal and non-skeletal pathologies, such as van Buchem disease, X-linked hypophosphatemia, autosomal hypophosphatemic rickets, McCune–Albright syndrome, and tumor-induced osteomalacia [13,14,15].

In this review, we will describe the molecules produced by osteoblasts, osteocytes, and osteoclasts that have been identified as endocrine factors, focusing on a few examples for each cell type.

Osteoblasts: Professional Secretory Cells

Osteoblasts, originated from progenitor cells present in the bone marrow, are the cells responsible for bone formation [2]. These cells are cuboidal with large nuclei located close to the basal membrane, enlarged Golgi apparatus, and extensive endoplasmic reticulum, consistent with their function as producers of extracellular matrix proteins and paracrine/endocrine factors.

The major function of osteoblasts is to produce the organic components to the bone extracellular matrix [2]. About 90% of the organic compartment is type I collagen, with smaller amounts of type III and V collagen. The remaining 10% are non-collagenous proteins (NCPs) with a vital role in regulating collagen formation and matrix mineralization, such as matrix extracellular phosphoglycoprotein (MEPE), pyrophosphate, matrix Gla protein (MGP), bone δ-carboxyglutamic acid-containing protein (osteocalcin, BGLAP), proteolipids, and bone alkaline phosphatase. Also, osteoblasts produce cell attachment proteins and small integrin-binding ligand with N‐linked glycoproteins (SIBLING proteins), including fibronectin, osteopontin, sialoprotein 1 (SSP1), osteonectin, and several other bone sialoproteins.

Osteoblasts secrete a wide range of different molecules with paracrine actions including receptor activator of nuclear factor kappa B ligand (RANKL/TNFSF11), monocyte colony-stimulating factor (M‐CSF‐1), osteoprotegerin (OPG/TNFRSF11B), Wnt gene family 5A (WNT5A), and WNT16, which regulate osteoclastogenesis [2, 16, 17]. RANKL and M‐CSF are molecules that bind cognate receptors (RANK and CSFR) on osteoclast precursors and facilitate their differentiation, whereas OPG functions as a RANKL decoy receptor, controlling RANK/RANKL interaction and osteoclast differentiation. Other direct interactions between osteoblasts and osteoclasts involve cell surface proteins, such as Semaphorin 3A (SEMA3A), Ephrin B4 (EPHB4), and FAS Ligand (FASL) [17]. Osteoblasts also secrete vascular endothelial growth factor (VEGF) that regulates skeletal development, osteoblast and osteoclast differentiation, bone repair, and angiogenesis [18, 19]. More recent studies described the axon guidance cue SLIT3 as a bone cell-derived angiogenic factor [20]. SLIT3 was described in 2018 as both an osteoblast-derived molecule required for fracture repair and bone healing [20] and as a osteoclast-derived factor coupling bone resorption and bone formation [21, 22]. However, in a side-by-side study it was shown that SLIT3 expression was negligible in osteoclasts compared to its expression in osteoblasts [23]. Further, deletion of Slit3 in cathepsin K-expressing osteoclasts did not alter the mouse skeletal or muscular phenotype, whereas SLIT3 global knockouts or mice with deletion of the gene in osterix-expressing cells (osteoblast precursors) exhibit vascular alterations [20, 23]. In addition to auto- and paracrine regulation of bone cell function, osteoblasts secrete molecules with endocrine action such as osteocalcin and lipocalin-2 (LCN2), which are the focus of this review, and fibroblast growth factor 23 (FGF23), reviewed elsewhere [7,8,9, 24].

Osteocalcin

Osteocalcin is the most abundant non-collagenous protein of the bone extracellular matrix produced by osteoblasts and its deletion in osteocalcin-deficient mice (Ocn−/− mice) results in a high bone mass phenotype due to an increase in bone formation without impairing bone resorption [25]. In addition, osteocalcin acts as an endocrine hormone on multiple organs, including adipose tissue, liver, muscle, pancreas, testis, and brain (Fig. 1) [7,8,9]. Before secretion by osteoblasts, osteocalcin is posttranslationally modified to γ-carboxylated osteocalcin (GlaOCN). The acidic environment generated by osteoclasts is responsible to decarboxylation of GlaOCN into undercarboxylated osteocalcin (GluOCN), decreasing its affinity for hydroxyapatite and helping osteocalcin reach the circulation [8]. GluOCN is the bioactive osteocalcin form with action as an endocrine hormone.

Fig. 1
figure 1

Communication of bone cells with other tissues. Schematic representation of the effects of osteoblasts, osteocytes, and osteoclasts in other cells and tissues, highlighting the molecules described in the text. Parts of the figure were drawn using pictures from Servier Medical Art. Servier Medical Art by Servier is licensed under a Creative Commons Attribution 3.0 Unported License (https://creativecommons.org/licenses/by/3.0/). RANKL, receptor activator of nuclear factor kappa B ligand; PGE2, prostaglandin E2, FGF23, fibroblast growth factor 23; OCN, osteocalcin; LCN2, lipocalin-2; PDGF-BB, platelet-derived growth factor BB; PTH, parathyroid hormone; NT, neurotransmitters

Full size image

GluOCN functions as a hormone, regulating β-cell function and adipocyte gene expression [7, 26]. Thus, Ocn−/− mice are abnormally fat due to increased fat mass, adipocyte number, and serum triglyceride levels. In addition, Ocn−/− exhibit higher blood glucose and lower insulin serum levels compared to wild-type mice. Further, mice lacking osteocalcin specifically in osteoblasts show decreased β-cell proliferation, glucose intolerance, and insulin secretion and resistance, all actions that result from lack of activation of the GPRC6A receptor [27, 28]. Additional studies showed that the effect of administration of undercarboxylated osteocalcin on serum insulin levels are mediated by secretion of the glucagon-like peptide-1 (GLP-1) from the gut [26].

The role of osteocalcin on the regulation of glucose homeostasis was also put in evidence in in vitro studies showing that low concentrations of osteocalcin (0.03 to 0.3 ng/ml) regulate cell proliferation and insulin expression in β-cells and adipocytes. In addition, higher concentrations of osteocalcin (10 to 30 ng/ml) increased adiponectin, Pgc1, and Ucp1 expression in adipocytes. These in vitro observations were confirmed in 8-week-old wild-type mice, which received osteocalcin by delivering pumps. These studies showed that osteocalcin is able to regulate insulin secretion, insulin sensitivity, and fat mass in vivo [29].

In liver, osteocalcin was shown to prevent lipid accumulation by partially restoring insulin sensitivity and glucose tolerance in mice fed with a high-fat diet and receiving daily osteocalcin injections [29]. In addition, high-fat diet-fed mice showed increased liver expression of the gene encoding for TNFα, which was normalized after osteocalcin injections.

Osteocalcin also increased energy expenditure and enhanced uptake and utilization of glucose in skeletal muscle, protecting from diet-induced obesity [30]. Moreover, osteocalcin increased the expression of the Nrf1 and Mcad genes implicated in energy consumption and enhanced mitochondrial activity in muscle [27].

A recent study in animal models proposed that OCN attenuates the endothelial dysfunction in atherogenic abdominal aorta seen in controls, by a mechanism independent of the GPRC6A receptor [31]. But no positive or negative impact of osteocalcin was observed in healthy aortic rings. Further, the association between osteocalcin levels and vascular function in humans is inconsistent, as recently reviewed by Tacey et al. [32].

GPRC6A has been proposed as an osteocalcin receptor in Leydig cells, in which osteocalcin promotes testosterone synthesis and is required for full fertility in male mice [33]. This conclusion is based on studies showing that testis size and weight and the size of epididymis and seminal vesicles, as well as sperm count are significantly decreased in Ocn−/− mice. However, no changes in testosterone levels were observed in men after 2 years of zoledronic acid therapy, which results in substantial reduction in osteocalcin levels, arguing against a biologically significant role for osteocalcin in the regulation of testosterone in adult men [34].

Another potential target for osteocalcin actions has been proposed to be the brain [35]. Thus, the passive behavior observed in Ocn−/− mice suggested a role of osteocalcin in central nervous system. Consistent with this hypothesis, osteocalcin, which is able to cross the blood–brain barrier, inhibits the synthesis of monoamine neurotransmitters by reaching neurons of the brainstem, midbrain, and hippocampus. Further, osteocalcin intracerebroventricular infusions corrected the neurotransmitter deficiency, normalized tryptophan hydroxylase 2 (Tph2), tyrosine hydroxylase (Th), glutamate decarboxylase 1 and 2 (Gad1/Gad2) expression, corrected anxiety and depression, and partially improved the spatial learning in Ocn−/− mice.

Lipocalin-2

LCN2 is a ubiquitously expressed protein that has been associated with processes, such as cellular differentiation, inflammation, and cancer, and it is known to induce survival/apoptotic signals [36]. In bone, at the local level, osteoblast-derived LCN2 negatively alters bone development and reduces trabecular number and bone mass by affecting growth plate development and osteoblast differentiation [37]. Further, it increases bone resorption by enhancing osteoclast activity, as demonstrated in transgenic mice overexpressing LCN2 in bone. In addition, an endocrine function of LCN2 has been proposed, by which the molecule regulates energy homeostasis, increasing insulin secretion and sensitivity, and glucose tolerance [38]. LCN2 deletion in osteoblastic cells (Lcn2osb−/− mice), but not in adipocytes, results in decreased glucose tolerance, insulin sensitivity and secretion, and increase in food intake. On the other hand, Lcn2fat−/− mice lacking LCN2 in adipocytes did not exhibit differences in these parameters. LCN2 decreases food intake acting on hypothalamus (Fig. 1) by activating MC4R-dependent anorexigenic signaling function, as demonstrated following LCN2 administration to Mc4r−/− mice [39]. Osteocalcin expression and activity were not affected suggesting a direct effect of LCN2 on pancreatic islets. However, the role of the bone on body weight remains unclear. In addition, previous studies reported that LCN2 deficiency in global Lcn2−/− mice improved systemic insulin sensitivity but does not affect food intake [40, 41].

In addition to the effects described for osteoblastic LCN2 and consistent with an outside-in effect of LCN2, kidney-produced LCN2 increases FGF23 synthesis by osteocytes in response to inflammation and in chronic kidney disease (CKD) [42, 43].

Osteocytes: Sending Signals from within the Bone

Osteocytes originate from osteoblasts once they become completely surrounded by mineralized bone matrix [2, 44]. Because of their localization within the bone matrix and the extensive network of projections that allow the communication among osteocytes and with cells on the bone surface, it has been long thought that osteocytes are responsible of sensing mechanical signals, triggering bone formation and/or resorption, depending on the magnitude of the mechanical signals or whether bones are loaded or not [45]. It is now accepted that osteocytes are also the target of hormonal and pharmacological stimuli and that in turn they release factors that alter the formation and activity of osteoblasts and osteoclasts [44]. The main products of osteocytes affecting bone remodeling are Sost/sclerostin, a potent inhibitor of bone formation, and RANKL/OPG, pro- and anti-osteoclastogenic cytokines indispensable for normal bone resorption [45]. In this section, we will describe the components of the osteocyte secretome that have effects on distal tissues and organs.

Mechanical stimulation of osteocytic cells leads to the release of small molecules, namely prostaglandin E2 (PGE2), adenosine triphosphate (ATP), nicotinamide adenine dinucleotide (NAD+), and nitric oxide (NO) [45, 46]. In vivo studies showed that these molecules are involved in the anabolic response to mechanical loading. Interestingly, studies have shown that skeletal muscle from aged humans (79 ± 2-year-old) exhibit reduced expression of prostaglandin receptors, compared to younger individuals (25 ± 1-year-old), raising the possibility that reduced response to osteocyte-derived prostaglandins is involved in the ameliorated response to exercise in the elderly [47]. However, the effect of small molecules released by osteocytes as a consequence of mechanical stimulation in cells other than osteoblasts and osteoclasts has not been tested.

Osteocytes are the main source of the receptor activator of Nfκb ligand (RANKL) for osteoclast differentiation and bone resorption [48, 49]. In addition to its osteoclastogenic actions, RANKL has been shown to reduce skeletal muscle mass and strength in mice overexpressing human RANKL [50]. Thus, mature skeletal muscle cells express RANK, the RANKL receptor, and its activation regulates Ca2 + storage, function, and phenotype in denervated skeletal muscles [51]. In particular, skeletal muscle deletion of RANK using the muscle creatine kinase-Cre increases muscle fatigability, results in altered fiber-type composition in both control and denervated muscles and reduces the sarcoendoplasmic reticulum Ca2 + -ATPase (SERCA) activity in the fast twitched extensor digitorum longus, EDL, muscle. Further, muscle cells secrete the RANKL decoy receptor osteoprotegerin (OPG) in vitro, and administration of OPG-Fc, a long-lasting OPG derivate, improves skeletal muscle integrity and function and reduces skeletal muscle inflammatory cell infiltration in the mdx mouse model of muscular dystrophy [52]. In addition, elevated RANKL levels are associated with skeletal muscle atrophy in a model of non-metastatic ovarian cancer, whereas anti-RANKL treatment reduced myotube atrophy and the decrease in skeletal muscle mass and strength in this animal model [12].

More recently, a novel RANKL receptor, leucine-rich repeat-containing G-protein-coupled receptor 4 (LGR4, also known as GPR48) has been described [53, 54]. LGR4 competes with RANK for RANKL binding, suppressing downstream signaling [53]. As expected, deletion of LGR4 results in increased osteoclasts and bone resorption. Other studies showed that absence of LGR4 increases lipid oxidation-related gene expression and reduces glucose transporter type 4 (Glut4) levels thereby regulating energy expenditure in skeletal muscle [55]. Yet, whether LGR4 mediates, at least in part, the effects of RANKL in skeletal muscle is not known. It should be noted, however, that the source of RANKL was not determined in these studies and, even though it has been proposed the osteocytic RANKL is the main regulator of osteoclastogenesis [48, 49], the possibility that other cells are the source of skeletal muscle-affecting RANKL, such as T cells [56, 57], cannot be ruled out.

Another function ascribed to osteocytes is the regulation of mineral metabolism (recently reviewed in [58]). As discussed elsewhere in the special issue [24] fibroblast growth factor 23 (FGF23), a phosphaturic factor, is secreted by osteocytes in response to increased blood phosphate and 1,25(OH)2 vitamin D [59, 60]. First identified as the cause of autosomal recessive hypophosphatemic rickets [61], FGF23 activates FGF receptor/Klotho1 complex in the kidney, resulting in increased urinary phosphate excretion (phosphaturia) (Fig. 1).

Osteocytes also secrete several molecules involved in the activation and inhibition of the Wnt signaling pathway, with potential impact outside of bone cells [1]. These include Dikkopf1 (Dkk1) and soluble frizzled-related protein 1 (Sfrp1), both with systemic effects. However, whether osteocytic Dkk1 and/or Sfrp1 have a systemic role is not clear. On the other hand, sclerostin, the product of the Sost gene and a very potent inhibitor of Wnt signaling and bone formation, produced mainly by osteocytes, has been shown to have systemic effects (Fig. 1). First identified in individuals with sclerosteosis and van Buchem disease, in which its reduction/absence results in high bone mass [62], was originally described as an osteocyte-produced BMP antagonist [63], later found to be a potent inhibitor of the Wnt signaling pathway [64, 65]. Due to its potent bone formation inhibitory effects, sclerostin was targeted for anabolic treatments and neutralizing antibodies were generated [66]. Studies in animals demonstrated the effectiveness of the neutralizing antibodies in increasing bone mass and strength and preventing deleterious effects of surgical and pharmacological approaches, as well as of traumatic events [67,68,69,70,71]. The anti-sclerostin antibody was tested in humans and demonstrated high effectiveness [72,73,74] and was approved by the FDA in 2019 for the treatment of patients at high risk of bone fractures [75]. However, there is evidence of increased cardiovascular events in individuals treated with anti-sclerostin, suggesting sclerostin might have an effect on the vasculature. Indeed, evidence of reduced tissue deterioration has been shown in the aorta of mice-treated recombinant sclerostin or transgenic for Sost, whereas human samples from aortic aneurysm showed reduced sclerostin levels [76]. However, it is not clear whether osteocyte-derived versus locally produced sclerostin are the cause of this cardiovascular pathology.

Osteoclasts: Secretion Versus Release of Matrix-Embedded Signaling Molecules

Osteoclasts are unique cells with the ability of bone resorption and therefore essential to remodeling and maintain bone integrity. They have a short lifespan but are highly active cells. Located on trabecular and endosteal bone surfaces, as well as on the periosteal surface during bone development and growth, these cells derive from hematopoietic stem cells differentiated by stimulation of growth factors and cytokines such as M-CSF and RANKL secreted by osteoblast and osteocytes, among other cells [77]. Differentiation gives rise to immature mononuclear cells that merge to form large, multinucleated osteoclasts with high movement and migration capacity [78].

Mature osteoclasts are polarized cells with a basolateral domain in contact to vascular stream and a resorptive domain facing bone surface, where the ruffled border is located [2]. The structure of these cells matches their physiological functions; they are responsible of bone resorption, protein transport, and waste products processing, but also secretion of molecules and proteins. For the resorption process, the osteoclasts attachment to bone matrix is facilitated by podosomes action, anchoring the cells in the sealing zone and therefore giving place to the resorption pit, a compartment where protons, chloride ions, and matrix-degrading enzymes like cathepsin K and tartrate-resistant acid phosphatase (TRAP) are secreted through the ruffled border. Due to low pH, enzyme activation occurs, enhancing mineral and proteins release from bone matrix [79].

During bone resorption osteoclasts release from bone matrix different growth factors such as transforming growth factor β (TGF-β) and insulin-like growth type 1 (IGF-1), cytokines, and factors that stimulate bone formation [80, 81]. These released factors, such as bone morphogenetic protein 6 (BMP6), collagen triple helix repeat-containing 1 (CTHRC1), sphingosine-1-phosphate (S1P), Wnt family member 10B (WNT10B), semaphorin-4D (SEMA4D), and cardiotrophin-1 (CT-1) can carry out paracrine actions, while other factors can act on vascular tissue, including platelet-derived growth factor (PDGF).

Osteoclast-Released Factors with Paracrine Actions—Clastokines.

Studies demonstrated that, in addition to their traditional role in bone resorption and as cells able to release factors stored within the bone matrix [2], osteoclasts participate in the immune response. The localization and origin of the cells were a clue for this novel osteoclast role. Thus, on one hand osteoclasts are located at the same place where hematopoiesis occurs and on the other, similar to macrophages, they differentiate from cells of the myeloid lineage. These facts were key for investigators to find the role of osteoclasts on immunity: they suppress T cell response to the activators α-CD3/CD28. Osteoclast and T cell interaction could be cell contact independent, via the secretion of chemokines by osteoclasts. Osteoclasts were also shown to participate in the immunosuppressive microenvironment by releasing molecules, like death ligand 1 (PD-L1), CD200, and galectin-9 (Fig. 1) [82]. Another study has demonstrated that osteoclasts interact with T cells by mechanisms similar to those used by antigen-presenting cells [83]. Thus, osteoclast membranes express major histocompatibility complex (MHC) I and II molecules, which are needed for antigen presentation. Osteoclasts also express costimulatory molecules CD80, CD86, and CD40 and are able to release cytokines IL-10, TGF-β, IL-6, and TNF-α, required to activate T cells [84]. Recent studies and, in particular, genomic analyses have provided the molecular basis for the role of osteoclasts in the immune response, which is mediated by the secretion of cytokines, recently named “clastokines” [85]. Further, microarray transcriptomics revealed that mature human osteoclasts are transcriptionally closer to dendritic cells than to monocytes [82].

PDGF-BB

Platelet-derived growth factor (PDGF) is a factor secreted by preosteoclast involved in the maintenance of bone homeostasis and vessel formation during bone modeling and remodeling. Deletion of the Pdgfb gene in TRAP + preosteoclasts caused low cortical and trabecular bone mass and reduced angiogenesis in young healthy mice [86].

PDGFs are important serum factors that stimulate smooth muscle cell migration and proliferation and a role of PDGF-BB in the development of neointimal hyperplasia after joint injury and in atherosclerosis acting as vascular aging–inducing factor has been proposed [87]. In aged mice under metabolic stress, preosteoclasts exhibit increased PDGF-BB secretion and arterial stiffening and low bone mass compared to healthy mice, suggesting that high PDGF-BB negatively regulates osteoblasts differentiation and contributes to vascular aging in a paracrine manner [88].

Summary and Conclusion

Mounting evidence supports the role of bone cells not only in the regulation of bone mass and strength but also in controlling bone homeostasis. In this manuscript, we discussed how gene mutations in bone-released factors can affect other organs and cause a wide range of diseases. These interactions shed light to a side of the field that is not usually analyzed and highlights the need to pay attention to bone tissue physiology and functionalities affecting others biological process.

Although some studies are still controversial, osteocalcin, FGF23, and sclerostin stand out as bone-produced endocrine factors. Yet, further investigation is needed to tease out the contribution of bone release versus local production for some of those molecules. The challenge for the future will be to identify and find the specific functions of bone-released factors and understand the reciprocal interaction between bone and other organs in favor of a better understanding of skeletal and non-skeletal diseases.