Biochemical Interaction Between Muscle and Bone: A Physiological Reality?
In elderly with a sedentary lifestyle, often suffering from sarcopenia to osteopenia, a training intervention could be an effective countermeasure for bone as well as muscle. Both bone and muscle adapt their mass and strength in response to mechanical loading in part via similar signaling pathways. Bone as well as muscle produces a wide variety of growth factors and cytokines in response to mechanical loading, which are important for their adaptations. It has been hypothesized that in addition to mechanical stimuli, muscle and bone communicate by these factors. Whether such biochemical interaction between both tissues is physiological is a still subject of debate. Here, we provide an overview of a range of biological factors possibly involved in the biochemical cross talk between bone and muscle. In addition, we discuss the plausibility that such interactions are involved in non-pathological adaptation of both tissues, either in paracrine or in endocrine fashion. As yet, convincing experimental evidence for biochemical cross talk between muscle and bone is very limited. Several studies have shown that muscle-derived factors are involved in bone fracture healing as well as in bone adaptation in case of muscle pathology. For involvement of cross talk between muscle and bone in physiological adaptation, there is no definite proof yet. Detailed knowledge of the biochemical interactions between muscle and bone is of clinical importance. It can help to discover pharmacological treatment to be used alone or in parallel with exercise training, thereby reducing the need for high-impact exercise.
KeywordsSkeletal muscleBoneGrowth factorCytokineHypertrophyBone remodelingCross talk
Degenerative musculoskeletal disorders are very common in elderly and patients suffering from chronic diseases such as heart failure, cancer and arthritis [1, 2]. A consequence of these disorders is a reduced physical function resulting in a more sedentary lifestyle. This reduction or lack of mechanical loading of muscle and bone which further impairs muscle function results in a vicious cycle and ultimately causes severe muscle atrophy and reduced bone mineral density. Muscle strength training and exercises with high-impact loading of bone are generally acknowledged to be effective as countermeasures. However, in case of frailty, the load-bearing capacity is rather low and high impact exercise is dissuaded. Therefore, development of effective treatment requires strategies which are physically less demanding (i.e., low-intensity training and/or pharmacological treatment). To this end, a profound insight is required in the mechanisms of how muscle and bone regulate their mass and functional properties as well as how these tissues interact.
Muscle and bone are mechanically connected and are both highly adaptive to mechanical loading. Muscles consist of a muscle belly that inserts directly onto bone or indirectly via aponeuroses and tendons. The muscle belly consists of muscle fibers (i.e., large multinucleated cells, densely packed with contractile proteins actin and myosin), being embedded in a honeycomb-like network of connective tissue sheets. The tendon consists of collagen fibrils in which fibroblasts resided for the collagen production. The potential of muscles to generate force is proportional to the muscle physiological area [3, 4]. The latter is determined by the number of muscle fibers and their size and connective tissue content. The number of muscle fibers is established shortly after birth  and only decreases with age. Therefore, increasing muscle mass can only occur by hypertrophy of muscle fibers. Hypertrophy occurs in case of positive net difference in the rate of muscle protein synthesis and degradation. The rate of synthesis is determined by the number of myonuclei, the mRNA content and rate of translation, while muscle protein degradation is largely determined by the activity of the proteasome system . The muscle stem cells [satellite cells (SCs)] lying between the basal lamina and the plasma membrane are activated in case of injury and in response to mechanical loading to add nuclei to the pool of myonuclei within the mature muscle fiber .
The regulation of bone mass is distinct from that of muscle. Whereas in muscle, all systems of protein synthesis and degradation reside within one cell (i.e., de muscle fiber), in bone, different cell types orchestrate bone formation and resorption. Bone remodeling occurs through the concerted action of a functional cohort of cells termed the basic multicellular unit (BMU). The BMU consists of the osteoclasts resorbing bone, the osteoblasts replacing bone, the osteocytes within the bone matrix, the bone lining cells covering the bone surface and the capillary blood supply. As bone remodeling is a multicellular event, signaling and cross talk between the cells involved is important in controlling this process. Coordination of the temporal activation and function of the different types of cells is regulated by “coupling factors” within the BMU.
Despite the differences in bone and muscle regarding their systems for protein synthesis and degradation, bone mass and muscle size are proportionally related [8, 9], suggesting that triggers for adaptation are interrelated. Mechanical stress is a prime cue for both muscle and bone cells to change the net rate of protein turnover. Muscle has the ability to contract and actively exert force onto bone, while the other way around only holds during periods of extensive growth (i.e., childhood and puberty). Long bones can induce stretching of the muscles and hence exert tensile stress onto the muscle fibers. The strong correlation between bone structure/strength and muscle mass was explained first by the mechanostat model postulated by Frost . This model predicts that bone adapts its mechanical properties to the required functional demands in daily life. According to this theoretical concept, there is mutual mechanical interaction between muscle and bone. This adaptation of bone in response to mechanical loading is regulated by loading thresholds (i.e., “set points”) above or below which tissue remodeling is changed. Circulating agents or pathological conditions can modulate the set point for this remodeling. In addition, in response to mechanical loading both muscle and bone cells produce a multitude of growth factors and cytokines which play an important role in muscle adaptation and bone remodeling via autocrine–paracrine signaling within the respective tissues and/or possibly by changing the sensitivity to mechanical loading.
Recent evidence suggests that muscle-derived factors are involved in bone fracture healing and that muscle–bone interactions are not only mechanical . Bone tissue in rat regenerates faster after experimentally induced fracture when muscle-derived soluble factors are allowed to diffuse to the fracture site . It has been hypothesized that muscle and bone communicate by active biological agents such as growth factors, steroids and cytokines (soluble factors) [11–13]. Evidence for such communication is very scarce because of several reasons: (1) biochemical signaling in these tissues is related to their mechanical loading, and it is impracticable to study them separately, (2) it is very difficult to distinguish whether factors produced in one tissue are present and active in the other and vice versa and (3) changes in expression levels of biological agents in one tissue will likely also have their local effects on tissue adaptation with concomitant changes in the mechanical loading of the other tissue (e.g., muscle hypertrophy results in higher muscle force applied to the bones  ). If a two-way biochemical communication between muscle and bone exists, its impact may be substantial as their respective body fractions are 30–40 %  and 5 % . Therefore, any such communication will likely have major implications for development of training and medication to maintain muscle mass and bone mineral density in case of muscle wasting and osteoporosis. However, whether biochemical communication between muscle and bone is physiological is a still subject of investigation and direct evidence is very limited.
Here, we provide an overview of a range of biological factors potentially involved in direct cross talk between muscle and bone via biochemical factors and we discuss the plausibility that such interaction between both tissues occur in a physiological context.
Structures and Systems Involved in the Interaction Between Muscle and Bone
In addition to these more direct routes for biochemical cross talk, an alternative route may be that soluble factors derived from both bone cells and/or muscle fibers as well as from their SCs reach other tissue(s) via the circulation. In response to exercise training, expression levels of soluble factors in muscle and bone change substantially (e.g., [21, 22]), which is reflected in similar changes in serum levels (e.g., [21, 23, 24]).
From the above, it is suggested that several routes exist via which muscle and bone cells could affect each other with soluble factors in a paracrine and endocrine manner. This raises the question which biochemical agents could be involved in such signaling. Naturally, agents expressed and functioning in both muscle and bone (i.e., common factors) can affect the other tissue whenever it reaches the surrounding of this tissue. Biological agents expressed specifically in one tissue can only exert their effect in the other when the corresponding receptor is expressed. In the following sections, an overview is given of biological agents that are expressed in both bone and muscle tissue and that may have an effect on the other. First, we discuss biochemical factors commonly expressed in muscle and bone, thereafter factors which are known for their function in either muscle or bone but of which their receptors are expressed in both tissues.
Common Growth Factors and Cytokines in Muscle and Bone and Their Involvement in Regulation of Muscle Fiber Size and BMD
Muscle and bone share many similarities regarding the regulation of adaptation in response to mechanical loading. The regulation of mechanical loading-induced adaptation of muscle size and BMD involves soluble anabolic and catabolic factors. However, such mechanical loading-induced adaptations also implicate an increased metabolic rate and a larger demand for energy-rich phosphates, which requires mitochondrial biosynthesis and increased oxygen supply, particularly in muscle as the mechanical loading in this tissue is actively generated by contractile filaments. Therefore, soluble factors involved in energy metabolism also may be involved in the biochemical cross talk between muscle and bone.
Insulin-Like Growth Factor-1 (IGF-1)
In both muscle and bone, IGF-1 is generally acknowledged as an important anabolic factor [25–27]. Several splice variants have been identified from the IGF-1 gene, of which IGF-1Ea and IGF-1Eb/c [also referred to as “mechano growth factor” (MGF)]  are induced by mechanical loading in muscle fibers and myoblasts [29, 30], in fibroblasts in tendons  as well as in osteocytes [13, 18] and osteoblasts . The splice variants differ with respect to their nucleotide sequence in exon 5 which causes an insert of about 23 amino acids (i.e., E-peptide).
Within skeletal muscle, IGF-1Ea stimulates muscle fiber hypertrophy by activating the phosphoinositide 3-kinase/Akt/mammalian target of rapamycin (PI3K/Akt/mTOR) pathway which increases the rate of mRNA translation and inhibits the expression of E3 ligase muscle RING finger protein 1 (Murf1) and muscle atrophy F-box (MAFfbx) . In addition, IGF-1 Ea also acts on SCs by stimulation of the differentiation into myotubes [33, 34]. In contrast, MGF E peptide has been shown to activate proliferation of SCs which results in an increase in the myonuclear number within the muscle fibers [33, 35].
In bone, the two IGF-1 splice variants show more or less similar functions as reported for skeletal muscle. MGF E peptide has been shown to promote MC3T3-E1 osteoblast proliferation , whereas IGF-1 Ea has been shown to stimulate osteocyte and osteoblast survival [37, 38], but also differentiation of osteoblasts and matrix production . In addition to its osteoanabolic effects, IGF-1 Ea has a role in bone resorption. While in muscle, IGF-1 Ea inhibits degradation, in bone, IGF-1 Ea promotes also osteoclastogenesis [40, 41].
Since IGF-1 splice variants are expressed locally in the muscle belly and tendon as well as in bone, interactions may occur by diffusion through tendons and/or the fascia separating muscle and bone. However, the abundant changes in IGF-1 Ea and MGF expression in muscle and bone are also reflected in changes in serum levels of IGF-1 (no distinction between IGF-1 Ea and MGF). IGF-1 serum concentrations have been shown to increase after strength training [42–45], suggesting that local availability of IGF-1 in muscle and bone could also be affected via transport through the circulatory system. Note that mechanical loading of muscle and bone may also change the expression of IGF-binding proteins , which could enhance but also inhibit the effects of IGF-1 .
Hepatocyte Growth Factor (HGF)
HGF is a growth factor which is expressed in various tissues . HGF mRNA expression within, muscle in vivo, has been shown to be stimulated by mechanical overloading . This increase in HGF may derive from its expression by SCs since subjecting these cells to cyclic stretch showed increased HGF mRNA expression levels . In addition, stretching of ex vivo cultured mature muscle fiber from mouse liberates HGF from the ECM that activates SCs resident within these muscle fibers . Such activation occurs via the HGF receptor c-Met . As such, in muscle, HGF is considered to be involved in regeneration from injury as well as hypertrophy by stimulating SCs and myonuclear accretion.
Within bone tissue, HGF seems to have a dual role as it is involved in both bone formation and bone resorption. HGF has been shown to stimulate osteoblast differentiation  and inhibit bone mineralization . In addition, HGF is also involved in bone resorption, as it stimulates osteoclast formation [55, 56]. Little is known regarding the type of bone cells that produce HGF. Mechanical loading of MLOY4 osteocyte by pulsating fluid flow decreased HGF mRNA expression levels, while HGF protein concentrations within the culture medium were increased . The observation that HGF is expressed in both muscle and bone cells and that cells in both tissues express c-Met receptors suggests that cross talk between bone and muscle by HGF could be a possibility. As fibroblasts seem to express HGF at very low levels and HGF has an inhibitory effect on collagen synthesis , signaling from muscle to bone by HGF implicates its diffusion from muscle fibers through tendons and/or fascia surrounding the muscle, rather than from fibroblasts within the tendons and connective tissue. Alternatively, HGF may also enter the circulation and act as an endocrine factor. However, to the best of our knowledge, changes in HGF serum levels in response to exercise or mechanical loading have not been reported yet.
Fibroblast Growth Factor (FGF)
Fibroblast growth factors (FGFs) regulate a broad spectrum of biological functions, including cellular proliferation, survival, migration and differentiation . FGF was originally identified as a protein capable of promoting fibroblast proliferation and is now known to comprise 23 members. FGFs exert multiple functions through the binding into and activation of fibroblast growth factor receptors (FGFRs). The main signaling occurring through the stimulation of FGFRs is the RAS/MAP kinase pathway . However, FGFs also activate the PI3-kinase/Akt signaling pathway in mouse myoblasts and fibroblasts [60, 61]. For muscle, both these signaling pathways are implicated in the regulation of hypertrophy . Skeletal muscle is a source of several FGF members [62–64]. In bone, FGF-FGFR signaling plays an essential role in both endochondral and intramembranous bone development . FGF signaling pathways are important for the earliest stages of limb development and throughout skeletal development . In adult bone, data on FGF expression is mostly limited to fracture healing. FGF 1, 2, 5, 6, 9, 16 and 18 and the four FGFRs have been shown to be temporarily expressed in the callus after fracture. Mice lacking FGFR1, 2 or 3 show deficits in osteoblast differentiation or have decreased bone mineral density and osteopenia .
FGF23 is an exceptional FGF variant since it is exclusively produced in bone and functions as a hormone regulating calcium phosphate homeostasis and vitamin D metabolism . This FGF is discussed below in more detail.
Since FGFs are expressed in bone cells, muscle cells and fibroblasts , paracrine signaling between muscle and bone by FGF may occur at the tendinous insertions of the muscle at the bone as well as via intra- and extramuscular connective and pereosteum. Although, free FGFs are readily degradable in vivo , several FGFs (FGF2, FGF21 and FGF23) have been detected in serum or plasma [69–71]. This suggests that these factors may also be involved in the communication between bone and muscle in an endocrine fashion.
Bone Morphogenetic Proteins (BMP)
BMPs were originally identified because of their unique activity, inducing heterotrophic bone formation in skeletal muscle . At this moment, however, BMPs are considered multi-functional cytokines, which belong to the TGF-beta family, as their activity is transduced by the specific transmembrane kinase receptors type I and II, downstream resulting in Smad 1/5/8 phosphorylation . BMP2, BMP6 and BMP7 are basally expressed in bone cells, and their expression levels increase in response to mechanical loading [74, 75]. In bone, BMPs promote osteoblast differentiation .
In myotubes, basal expression levels of BMP2 and BMP4 have been reported  as well as their downstream signaling proteins . In vitro studies using C2C12 myoblasts demonstrate that BMP2 changes C2C12 myogenic cell lineage into an osteoblastic cell lineage . Furthermore, activation of the TGF-β/BMP signaling pathway inhibits differentiation of myoblasts into myotubes [80, 81]. Clinically, the importance of BMP signaling in muscle has been demonstrated by the rare disastrous disease fybrodisplasia ossificans progressive, in which BMP2 receptor is constantly activated due to a mutation in the activin receptor Type IA, or ACVR1, resulting in ectopic bone formation in skeletal muscles . Also muscles and myoblasts from dystrophic mice show elevated expression levels of BMPs  which contribute to the impaired regenerative capacity of dystrophic muscle .
BMP expression has also been shown by fibroblasts within healthy and injured patellar tendon in particular at the tendon–bone interface [84, 85]. Since tendons connect muscle and bone, it is conceivable that BMPs expressed in tendons diffuse into bone and muscle belly and have their effects via paracrine signaling . As the effects of BMPs on mesenchymal stem cells and myoblasts mainly involve regulation of differentiation pathways, BMPs may therefore be important in the communication toward SCs in muscle rather than toward their host muscle fibers. Whether BMPs are active from one tissue to another via the circulation remains undetermined. To the best of our knowledge reports on BMP serum level measurements are scarce and associated with diseases . As yet, it seems most likely that if communication between muscle and bone by BMPs exists, this occurs via tendons and fascia.
Wnts are a large family of 19 secreted glycoproteins that trigger multiple signaling cascades essential for embryonic development, tissue regeneration and protein sysnthesis. Wnt signaling in particular the canonical Wnt/β-catenin signaling pathway in osteoblasts plays an important role in regulating bone formation and mass . Several proteins are involved in the modulation of Wnt signals. Wnt can be activated by R-spondins (RSPOs), which in bone promote osteogenesis and in muscle enhances satellite cell differentiation . On the contrary, secreted frizzled related proteins (SFRPs), Wnt inhibitory factors (WIFs), Dickkopf proteins (DKKs), sclerostin (SOST) and SOSTDC1 (also referred to as Wise) are inhibitors of the Wnt signaling. Overexpression of these proteins leads to abnormal cell cycle control and/or altered cell fate decisions are often implicated into abnormal cell cycle control and/or altered cell fate decisions . SFRPs and WIFs directly bind Wnts and prevent their interactions with receptors, while other secreted proteins including Dkks, SOST and SOSTDC1 bind to Lrp5/6 membrane receptors. Some Wnts (e.g., Wnt5a) can regulate Wnt signaling by competing with other Wnts (e.g., Wnt3a) that induce β-catenin stabilization .
Wnts are also expressed during myogenesis where they are involved in skeletal muscle embryonic development , as well as in muscle hypertrophy and muscle stem cell renewal [92, 93]. The Wnt/β-catenin signaling is activated during muscle overload ; however, whether this is due to enhanced expression of Wnts or due to changes in factors interacting with Wnt signaling is unknown.
Due to post-translational modification by glycosylation and palmification, Wnt proteins are extremely insoluble and whether these proteins diffuse from bone to muscle and vice versa is unknown. As yet, elevated serum levels for Wnts have only been reported for Wnt1 under pathological conditions , suggesting that endocrine effects of Wnts between muscle and bone are not impossible. If cross talk between muscle and bone occurs via Wnts, it seems more likely that this occurs by Wnt inhibitors. Several of the Wnt-inhibitors are specifically expressed in bone and are discussed below with respect to their ability to communicate between muscle and bone in paracrine or endocrine fashion.
Although the name suggests otherwise, OPN is a phosphoglycoprotein, which is secreted by a variety of cell types including, inflammatory cells, osteoblasts and skeletal muscle myoblasts [96, 97]. OPN, also known as secreted phosphoprotein one (SPP1), shows multiple proteolytic cleavage sites giving rise to different isoforms, which undergo variable posttranslational modifications [98, 99]. In bone, OPN is an extracellular structural protein and therefore an organic component of bone; however, OPN has also been recognized as a key inflammatory cytokine . It has a role in anchoring osteoclasts to the bone matrix during bone resorption . Recently, OPN was defined as a constraining factor on hematopoietic stem cells within the bone marrow microenvironment , implicating a directing role in the hematopoietic stem cell niche, possibly affecting the differentiation routes of these cells toward either osteoclast or lymphocytes .
OPN also seems to play a role in the regulation of muscle size. In mice, OPN mRNA levels increased in response to experimentally induced muscle damage . Gene promoter polymorphisms are associated with severity of muscle dysfunction in Duchenne muscular dystrophy , but also with muscle size in healthy humans and mice . Additionally, OPN plays a non-redundant role in muscle remodeling following injury . Involvement of OPN in biochemical communication between muscle and bone will depend on its diffusion characteristics from one tissue to the other or whether it is transported via the circulation. Recently, increased OPG plasma levels have been reported in different disease models [106, 107], suggesting a potential endocrine communication between muscle and bone.
OGN is one of the small leucine-rich proteoglycans (SLRPs), which play a key role in tissue development and repair by regulating the production of components of the extracellular matrix (ECM) as well as cell proliferation, differentiation, adhesion and migration within the ECM . In vitro studies revealed that OGN produced by C2C12 myoblasts affects MC3T3 osteoblast in the production of alkaline phosphatase, collagen type I and β-catenin . OGN is also demonstrated to be expressed in vascular smooth muscle cells . However, in bone, OGN expression has been demonstrated at both mRNA and protein albeit at low levels . Further research is necessary for the candidacy of osteoglycin as a factor in the communication between muscle and bone.
PGs are signaling molecules enzymatically derived from polyunsaturated fatty acids of cyclooxygenases (COX), or prostaglandin synthases. Receptors for PGs have been found on almost all cell types. PGs have a variety of actions throughout the body, such as vasodilatation or constriction and (dis)aggregation of platelets . In bone, the most abundantly expressed PG and most important one is that of the type 2-serie (PGE2), which can have both stimulatory and inhibitory effects on bone formation. PGs generally stimulate osteoblastic proliferation and differentiation in vitro ; however, at high concentrations, PG inhibits collagen synthesis . PGs are produced in response to mechanical loading and by exposure to serum factors among which cytokines and steroids . Because of their very short half-life , PGs can only function in an autocrine–paracrine manner.
In mature skeletal muscle, elevated PGE2 levels have been related to an increase in the rate of protein degradation  and possibly the induction of atrophy. However, in case of muscle injury, PGE2 may enhance muscle regeneration as lack of PGE2 inhibits differentiation of C2C12 myoblast into myotubes . This latter observation is in line with the report that COX expression in human skeletal muscle is increased after resistance exercise  and that blocking of COX induced production of PGs after resistance exercise is paralleled by a reduced rate of protein synthesis . Like in bone, it seems that in muscle, PGs have an important, but also dual role in the regulation of the rate of protein turnover.
With regard to biochemical communication between bone and muscle, PGE2 could be involved in this, as cross talk between osteocytes and myoblasts was demonstrated in vitro . Via what routes this cross talk also exists in vivo remains undefined. As PGs enter the circulation, such cross talk via PGs seems conceivable.
Vascular Endothelial Growth Factor (VEGF)
Tissue adaptation and regeneration implicates an increase in protein turnover which requires de novo vascularization in order to meet the increase in demand for nutrients and oxygen supply during bone remodeling. Angiogenesis is mediated by a complex interplay of mechanical stimuli, hypoxia and biochemical agents such as VEGF, FGF, platelet-derived growth factor (PDGF), RANKL, HGF, BMPs and IGF-1 [122–124]. Since VEGF plays a central role in angiogenesis , while its direct effect on muscle hypertrophy, muscle stem cell activation or osteoblast and osteoclast differentiation remains elusive, VEGF is considered as a growth factor having a key indirect role in determining the oxidative metabolism of bone and muscle cells. VEGF expression increases within muscle in vivo, in response to mechanical muscle overload by reducing the function of synergistic muscles  as well as by physical exercise . Since muscle contractile activity is accompanied by a reduction in oxygen tension within the capillaries and muscle fibers [127, 128] as well as by changes in the expression of soluble angiogenic factors , these change in gene expression are likely an effect of the interplay between these stimuli. The increase in VEGF expression in muscle is not only derived from endothelial cells but also derived from muscle cells which is required for running exercise-induced angiogenesis in mouse m. gastrocnemius . Within in vivo bone, VEGF is likely expressed also by bone cells as osteocytes, osteoblasts and osteoclasts; all express VEGF in vitro [13, 130, 131]. Although VEGF expression levels within muscle increase by exercise, plasma levels remain unchanged  or decrease . The discrepancy between exercise-induced alterations in local muscle VEGF levels and those in plasma has been explained by an elevated peripheral uptake of plasma VEGF after exercise . Taken together, the abundant expression of VEGF within muscle and bone and its presence in the circulation make VEGF a candidate which is possibly involved in biochemical interaction between muscle and bone.
IL-6 is a cytokine which is generally recognized for its pro-inflammatory action; however, increasing evidence suggests that it has several other functions. IL-6 is produced by a variety of cell types such as T cells, macrophages, fibroblasts, smooth muscle cells as well as by skeletal muscle fibers and bone cells (osteoblasts and osteocytes) [133–136]. Expression of IL-6 in humans and rodent skeletal muscle is substantially increased by contractile activity [135, 137], which in human has been shown to be paralleled by elevated serum levels [138, 139]. Additionally, to its inflammatory effects, IL-6 appears to enhance insulin-stimulated glucose uptake by increasing GLUT4 translocation in muscle fibers . Moreover, adenosine monophosphate-activated protein kinase (AMPK), which stimulates mitochondrial biosynthesis in muscle when activated , is a downstream target of IL-6 [142, 143] and its activation by IL-6 stimulates an increase in fatty acid oxidation . As such, IL-6 seems to be a regulatory metabolic factor in skeletal muscle. However, recent studies indicate that IL-6 is also involved in the regulation of muscle of muscle fiber size. Chronic IL-6 administration directly to rat skeletal muscle induces muscle fiber atrophy . Recovery of gastrocnemius muscle from disuse atrophy is attenuated in IL-6 knockout mice . These reports suggest an important role of IL-6 in the regulation of muscle fiber size.
In bone, IL-6 is produced at high levels in osteoblast/stromal cells in response to stimulation by a variety of other cytokines and growth factors . IL-6 is also shown to be expressed by osteocytes when they undergo apoptosis in response to bone unloading . Moreover, IL-6 is involved in osteoblast differentiation as well as osteoclast activation in vitro and seems to play role in both bone ostoeogenesis and bone resorption [147, 148].
Given the abundant expression of IL-6 in both muscle and bone, diffusion of this cytokine between both tissues is likely to occur. Recent evidence suggests that IL-6 signaling from bone indeed exists in muscle of Duchenne muscular dystrophy patients. In these patients, also levels of circulating IL-6 were elevated which was associated with a reduction in bone mass . Moreover, when adding conditioned medium collected from unloaded C2C12 myotubes cultures was added to mouse osteoclast precursor cells, osteoclast formation was inhibited . This inhibition of osteoclast formation was partially abolished when cells were cultured in conditioned medium from mechanically loaded myotubes, containing elevated levels of IL-6. Taken together, these data suggest that IL-6 is likely involved in the biochemical communication between muscle and bone.
Myostatin, a Transforming Growth Factor Specifically Expressed in Muscle: Does Myostatin Affect Bone Cells?
Myostatin is a growth factor which belongs to the superfamily of transforming growth factors-β, which is exclusively expressed in muscle. Myostatin signals by first binding the Activin type II receptor, which then allows for interaction with type I receptors ALK4 or ALK5 . Myostatin is regarded as a negative regulator of muscle size as it inhibits satellite cell proliferation , attenuates the rate of protein synthesis  and stimulates the rate of protein degradation . Expression levels of myostatin in muscle decrease in response to increased contractile activity and in particular by eccentric contraction . Effects of mechanical loading on myostatin expression in human muscle are likely also reflected in their circulating levels as endurance and strength trainings show reduced plasma levels of myostatin [21, 154–156]. As myostatin expression is not detectable in bone cells  and bone marrow-derived mesenchymal stem cells as well as osteoblasts express activin type IIB receptors , the question has been addressed whether muscle-derived myostatin could also have osteoganabolic/osteocatabolic effects via paracrine or endocrine signaling .
Loss of myostatin function in mice has been shown to be associated with increased bone mineral density [158, 159] and cortical thickness , suggesting that in wild-type mice, the presence of myostatin could have had a detrimental effect on bone. Note however that the lack of myostatin is also accompanied by substantial larger muscles , likely causing higher forces to be exerted on the skeleton. In case of bone fracture in mice, myostatin deficiency is associated with increased fracture callus size . In line with these observations, exogenous myostatin release from a hydrogel injected within the fracture callus caused a reduction in endochondral ossification . These observation strongly suggests the existence of biochemical signaling (paracrine and endocrine) from muscle-derived myostatin onto bone cells, although effects of elevated muscle mass and muscle strength may contribute as well.
Bone-Specific Agents Possibly Affecting Muscle Fibers and Muscle Stem Cells
Bone cells produce many factors mainly involved in bone turnover and mineralization. Factors regulating the mineralization process are mainly produced by late osteoblasts and osteocytes and are thought to have local effects. Since bone turnover is a tightly regulated process in which bone resorption is coupled to bone formation, many factors, produced by bone cells, play an important role in the communication between the bone resorbing cell, the osteoclast, and the bone forming cell, the osteoblast. To the best of our knowledge, among these factors, the following ones seem to be specifically expressed in bone and not in muscle.
Osteocalcin is a small γ-carboxyglutamate protein preferentially expressed by osteoblasts, only in the late stage of their differentiation, after the arrest of proliferation . OC mRNA has also been detected in several non-osseous tissues, but the RNA splicing was found to be incomplete, and therefore, osteocalcin in these tissues remained non-functional . OC is therefore characterized as a bone-specific factor. The exact role of OC in bone is not yet completely understood, but its ability to regulate bone mineralization and bone turnover is evident .
To the best of our knowledge, any influence of OC on muscle tissue in vivo or myoblasts in vitro has not been described in the literature. For this reason, OC is yet not considered to be a candidate in bone–muscle communication. However, several epidemiologic, genetic and biochemical studies have suggested that there is a hormonal link between bone and adipose tissue, and energy metabolism. In vitro, OC stimulates insulin expression in β-cells and adiponectin expression, an insulin-sensitizing adipokine, in adipocytes . Adiponectin is also expressed in both bone and muscle, where it may improve insulin sensitivity and stimulates biosynthesis of mitochondria . Whether osteocalcin has any role in the adiponectin-related metabolic effects in these tissues remains to be determined.
Fibroblast Growth Factor 23 (FGF23)
FGF23 is exclusively produced in osteocytes and osteoblasts and functions as an endocrine hormone that regulates phosphorus homeostasis through binding to FGFR and klotho, its co-receptor in the kidney and parathyroid glands. The primary physiological actions of FGF23 are to augment phosphaturia by down-regulating the expression of sodium phosphate co-transporters in the renal proximal tubule and to decrease circulating concentrations of 1,25-dihydroxyvitamin D by inhibiting renal expression of the 1,25-dihydroxyvitamin D synthesizing CYP27B1 (1-α-hydroxylase) and stimulating expression of the catabolic CYP24 (24-hydroxylase) . Because of the endocrine character of FGF23 and FGFRs as well as klotho have been shown to be expressed in mature skeletal muscle (cf. [169–172]), FGF23 is one of de candidate factors in the biochemical communication from bone to muscle.
FGF23 is associated with greater cardiovascular risk and left ventricular mass as well as higher prevalence of left ventricular hypertrophy . FGF23 was demonstrated to induce hypertrophy  and activate prohypertrophic gene programs in isolated neonatal cardiomyocytes . FGF23 caused pathological hypertrophy of isolated rat cardiomyocytes via FGF receptor-dependent activation of the calcineurin-NFAT signaling pathway, but this effect was independent of klotho . It is likely that FGF23 also affects skeletal muscle; however, to the best of our knowledge, any effect of FGF23 on muscle fibers or myoblasts has not been reported yet. FGF23 is therefore a bone-specific agent, candidate for communication between bone and muscle.
The active metabolite of vitamin D is 1,25 dihydroxyvitamin D (1,25 (OH)2D), and the conversion of 25(OH)D into 1,25(OH)2D has been proven in osteoblastic cells . Whether the conversion of 25(OH)D also occurs in muscle remains unclear [176, 177]. As yet, vitamin D is considered a bone-derived factor that could influence muscle cells.
The role of 1,25(OH)2D in bone metabolism is to provide a proper balance of calcium and phosphorus to support mineralization. 1,25(OH)2D stimulates differentiation of osteoclasts as well as osteoblasts . In osteoblastic cells, 1,25(OH)2D induces mRNA expression of osteocalcin as well as osteopontin and suppresses synthesis of type I collagen in a maturation-dependent manner . Because of the different regulatory effects, vitamin D is considered both an osteoanabolic and osteocatabolic steroid.
Expression of vitamin D receptor (VDR) has been reported in many different tissues including muscle cells  and mature skeletal muscle . Note however that the presence of VDR in myoblasts is still a subject of debate .
The role of vitamin D in the skeletal muscle homeostasis is not unambiguous. Clinically, vitamin D deficiency is associated with muscle weakness, predominantly in the proximal muscle groups , and a reduction in movement speed . A meta-analysis of the effects of vitamin D supplementation on the reduction in fall risk showed that 25(OH)D serum concentrations higher than 24 ng/mL reduced substantially the risk of falls .
The negative effects of vitamin D deficiency on muscle force generating capacity are presumably caused by muscle atrophy of mainly type II muscle fibers. However, muscle weakness in severe vitamin D deficiency could also be caused by secondary hyperparathyroidism and resultant hypophosphatemia . The reported reduction in falls was probably only partly due to an increase in muscle strength. Direct effects of vitamin D on myoblasts size are subject to controversy. 25(OH)2D was shown to stimulate myogenic differentiation of C2C12 myoblasts by inhibiting cell proliferation and modulating the expression of myogenic growth factors and myostatin . Furthermore, 25(OH)2 vitamin D activates Akt through PI3K as well as p38 mitogen AMPK to stimulate myogenesis in C2C12 cells .
In addition, 1,25(OH)2D confers a protective effect against NRTI-induced mitochondrial toxicity in skeletal muscle myoblasts and myotubes . This supports a protective role for vitamin D within muscle cells against free oxygen radicals. All together, it seems likely that the active metabolite of vitamin D, 1,25(OH)2D, can very well be formed in bone and act on skeletal muscles nearby.
OPG is the soluble factor in bone that decreases bone resorption by binding to RANK on the osteoclast. Whether OPG is involved in muscle homeostasis remains elusive. The solubility of OPG is one of the characteristics positive for its effect on skeletal muscle through paracrine and/or endocrine action. Elevated OPG serum levels have been associated with an increased rate of heart failure and are inversely related to BMD and serum myostatin . However, mechanistic studies refer to the stimulation of arterial calcification, resulting in atherosclerosis. Direct evidence for OPG on muscle cells has been described in vascular smooth muscle cells. OPG expression levels have been demonstrated to increase in pulmonary arterial hypertension and that it can regulate pulmonary arterial smooth muscle cell proliferation and migration . SiRNA-mediated knockdown of OPG in human vascular smooth muscle cells however did not alter the calcification process . Effects of OPG on skeletal muscle cells have not been described so far. Further research is needed to investigate the role of OPG on skeletal muscle cells.
Wnt Inhibitors Dickkopfs (DKKs) and Sclerostin (SOST)
Regulation of the Wnt signaling pathway is not bone specific. However, some of the inhibitors, SOST and DKK1 have been shown to induce extreme bone phenotypes in genetic diseases and are presumed to be bone specific (i.e., to our knowledge, these agents are not expressed in skeletal muscle). Whether these factors also affect Wnt signaling in myoblasts remains unclear. Down-regulation of the Wnt pathway may contribute to the inhibition of myogenic differentiation and resistance to apoptosis as has been shown in for instance embryonal rhabdomyosarcoma cases . Stimulation of Wnt signaling by RSPOs has been shown to activate myogenesis, while DKK1, but not sFRP1, effectively antagonized this induced activation of Wnt signaling. In addition, this antagonising effect of DKK1 effectively reduced MYF5 (a myogenic transcription factor) expression in both C2C12 myoblasts and primary SCs . Inhibition of Wnt signaling in muscle seems to have an effect on myogenesis. However, whether inhibition of Wnt signaling in muscle by bone-specific inhibitors of Wnt signaling occurs in vivo is unknown. The fact that both SOST and DKK1 can be detected in serum strongly indicates that these Wnt inhibitors are able to be transported from bone to muscle tissue in an endocrine manner, which makes it likely that they play a role in the biochemical communication between bone and muscle.
Biochemical Interaction Between Muscle and Bone: a Physiological Reality or Not?
The above inventory of biological agents expressed in muscle and/or bone reveals that theoretically, cross talk between muscle and bone via these agents should be feasible. Both tissues have similar signaling pathways, and many common biochemical factors may be involved in the communication between muscle and bone. However, most biochemical cross talk between muscle and bone described in the literature is based on in vitro studies, and hence, these provide only indications of the possibility that such communication exists. The observation that specific ligands (i.e., myostatin, OPG, FGF23, DKKs and SOST) are expressed in one of the two tissues, while their receptors or agonists are expressed in both, suggests, from a teleological perspective, that these receptors can only be activated by agents that originate from tissue at a distance and not by locally expressed biochemical agents. However, whether this communication exists under physiological conditions remains elusive and requires experimental testing. Investigating biochemical cross talk between bone and muscle in vivo encounters many difficulties and is a big challenge since the biochemical processes cannot be divorced from the mechanical components. Several pathologic conditions and innovative experimental approaches have been adopted, providing indications for a biochemical interaction between muscle and bone.
Overexpression of IGF-1 in mouse plantar flexor muscles by electroporating IGF-1 cDNA under control of a CMV promoter induced an increase not only in muscle mass but also in BMD and bone mineral content (BMC) . When the IGF-1 overexpression was paralleled by disuse due to hind limb suspension, the rate of muscle atrophy of the plantar flexor muscles as well as reductions in BMD and BMC were attenuated . These results indicate that IGF-1 expressed by the targeted muscles was likely released, moved toward bone and had osteoanabolic effects on tibial and fibular bones. Obviously, mechanical effects due to the elevated muscle mass cannot be excluded.
Patients with Duchenne muscular dystrophy show muscle weakness as well a severely reduced BMD which was shown to be related to down-regulation of osteogenic genes and up-regulation of osteoclastogenic genes . These changes were paralleled by high serum levels of IL-6 as well as high serum ratios of RANKL/OPG. Incubation of human primary osteoblast cultures with sera from DMD patients showed an impaired osteogenic response. These results support the notion that in patients and animals with muscular dystrophy, bone adapts to paracrine and endocrine signaling derived from muscle.
Bone fracture appears to be a pathological condition in which biochemical interactions, apart from mechanical, are most evidently playing a role in the healing process. Several studies indicate that this healing is attenuated at sites where the muscle envelope is reduced . Muscle tissue surrounding bone has been shown to accelerate the rate of fracture healing by promoting revascularization as well as by recruitment of muscle stem cells and their accumulation in the fracture site [194, 195]. In addition, muscle-derived biochemical agents are also very likely involved in the healing of fractures. When experimental tibia fracture was followed by wrapping a nitrocellulose membrane around the tibia, the healing occurred more slowly and IGF-1 was shown to be present only after using membranes with particular pore size, suggesting that agents of particular size were required to induce IGF-1 expression within the fracture site . Interestingly, muscle atrophy after injection of quadriceps muscles with botulinum toxin was associated with a compromised femur fracture healing . As in experimentally denervated mouse gastrocnemius muscle  and in unloaded human vastus lateralis muscle , expression levels of myostatin were increased; effects of botulinum toxin on growth factor and cytokine expression within the targeted muscles could have contributed to the impaired bone healing.
Taken together, there are several indications that biochemical communication from muscle to bone is very likely to occur under pathological conditions or experimental conditions where muscle or bone function is compromised. Given the fact that bone also produces bone-specific factors which potentially could affect adaptation of muscle suggests that biochemical communication from bone to muscle in vivo should also be possible.
Whether biochemical communication in vivo between muscle and bone exists under physiological/non-pathological conditions is still an enigma? For the biochemical agents that are specifically expressed in either muscle or bone, it is apparent that if the corresponding signaling pathways are affected in the tissue where they are not expressed, this signaling must originate from the other organ.
Only recently, Hamrick et al. made a first attempt to study such interaction by testing whether inhibition of myostatin signaling in aged mice by injecting a myostatin inhibitor stimulated muscle hypertrophy and bone formation simultaneously . Indeed, this approach induced muscle hypertrophy; however, an increase in bone formation could not be shown . The lack of effect on bone formation suggests that within a physiological context, myostatin derived from muscle seems to be not involved in bone adaptation. However, it cannot be ruled out that myostatin from muscle was present in bone and its signaling inhibited, while bone adaptation did not have occurred yet. Clearly, biochemical cross talk between muscle and bone requires to be investigated more extensively at different levels of signaling pathways related to myostatin and other agents. In addition, it needs to be established whether under physiological/non-pathological conditions muscle- or bone-derived factors indeed diffuse from one tissue into the other. The use of tractable agents may be a requirement for this.
Apart from the factors specifically expressed in a tissue, for the common factors, it remains questionable whether presence of biochemical agents derived from another tissue will have an effect on top of the already locally expressed factors. Such effects will depend on the local concentrations of growth factors and cytokines in the vicinity of the membrane receptors. It is conceivable that the dose–response relation reaches a plateau at particular concentrations, implicating that paracrine and endocrine biochemical agents may not add to the signaling by locally expressed agents.
Clinical Implications of Biochemical Communication Between Muscle and Bone
This review mostly compiled the possible candidates yet known for their involvement in biochemical communication between muscle and bone, either by direct contact via the tendons and intermuscular connective tissue or through the circulation as endocrine factors. Knowledge on the biochemical interaction between bone and muscle is important in a clinical perspective, since sarcopenia and osteoporosis develop equally with aging. This implies that also prevention and treatment of these conditions should occur with a joint goal, namely to improve both bone as muscle mass and strength. The biomechanical cross talk between bone and muscle obviously refers to a joint approach. Insight into the biochemical interaction could fine-tune this approach, particularly toward an individualized therapy. So far, the design of training intervention programs to improve bone mass and bone quality mostly focused on high-impact loading on bone using gravitational forces to induce bone deformity, necessary to induce bone formation. However, biochemical cross talk between bone and muscle enables development of training programs that enhance muscle mass and muscle strength, which may indirectly also stimulate bone formation. This is a promising thought because vigorous training intervention in elderly is dissuaded. It should be realized that the growth factor expression levels in muscle are not exclusively increased by eccentric high-intensity contractions. Although eccentric contractions of rat gastrocnemius medialis muscle increase IGF-1 expression more than isometric or concentric contractions, also after these latter type of contraction, with lower contractile forces, IGF-1 expression levels within these were substantially increased . In addition, basal IGF-1 expression levels were shown to be higher in slow, high-oxidative rat soleus muscle than in fast, low-oxidative extensor digitorum longus muscle, while the opposite was shown for myostatin . These observations seem paradoxical as generally slow, high-oxidative muscle fibers remain relatively small. However, the rate of protein breakdown is also higher in high-oxidative muscles  which may explain the relatively high expression levels of anabolic factors in these fiber types.
Regarding the biochemical cross talk between muscle and bone, these observations suggest that also during low-intensity exercises, growth factors and cytokines are produced which have osteoanabolic effects outside the muscle. To make effective use of the biochemical communication between muscle and bone requires optimization of training strategies aiming to minimize the mechanical impact in case of frailty or whenever physical performance is impaired. In addition, pharmacological or nutritional interventions targeting at changing the biochemical factors and their signaling pathways involved in bone and muscle communication could be used to support training interventions.
Measurement of serum levels of biochemical factors involved in muscle–bone communication could also be used to predict the potential of training intervention treatment. This possibility mostly holds for the soluble bone and muscle-specific factors which are measurable in serum. At this moment, several assays for factors like FGF23, myostatin and OPG are commercially available. Use of these may give physicians also the opportunity to monitor early effects of training in patients. This may help to predict the efficacy of treatment and provide cues for continuation of the training intervention, which together enables individualized therapy .
We would like to thank Huib van Essen for technical assistance with the staining and Guus Baan for his help in the imaging and graphical design of the figures.
Conflict of interest
Richard Jaspers and Nathalie Bravenboer declare that they have no conflict of interest.
This article does not include any studies with human or animal subjects performed by the author.