The development of the skeletal system, as well as its repair, are dependent on the differentiation of both chondrocytes and osteoblasts from their common progenitor, the mesenchymal stem cell (MSC). There are two processes from which bone is formed. During intramembranous osteogenesis, bone is formed directly from MSC differentiation to osteoblasts. This process gives rise to flat bones, such as those of the skull and clavicle. In contrast, endochondral osteogensis requires MSC differentiation to chondrocytes and the formation of a cartilage template, which is followed by ossification by osteoblasts. This process is responsible for the formation of the long bones of the skeleton, and remains active in the growth plates of growing bones (reviewed in Refs Reference Olsen, Reginato and Wang1, Reference Hartmann2). Bone fracture repair recapitulates the events of skeletogenesis and is, therefore, used as an experimental model of bone formation (Ref. Reference Schindeler3). These processes are regulated by paracrine actions between osteoblasts and chondrocytes, and thus, proper skeletogenesis requires precise control over the differential differentiation of the MSC to these lineages. Failure of chondrocyte or osteoblast precursors to proliferate and differentiate leads to various types of skeletal dysplasias, depending on the pathway involved (reviewed in Ref. Reference Baldridge4).
Skeletogenesis: an overview
The osteochondroprogenitor
The majority of what is known about MSC differentiation along the skeletal lineage comes from embryology. Endochondral and intramembranous skeletogenesis begins with the proliferation and migration of mesenchymal cells to form condensations. These condensations are characterised as tightly packed cells that express specific condensation markers, and will determine the position, shape and size of the skeletal elements (Ref. Reference DeLise, Fischer and Tuan5). The transforming growth factor-β and Hox family of molecules, derived from the epithelium, have been implicated in directing the formation of condensations. These molecules, along with the cell–cell and cell–matrix interactions that result from the compact nature of these cells, are thought to trigger MSC differentiation (Refs Reference DeLise, Fischer and Tuan5, Reference Hall and Miyake6).
As the osteochondroprogenitor is multipotent, lineage fate decisions must be made and maintained. The transcription factor runt-related transcription factor 2 (Runx2) is believed to determine osteoblast versus chondrocyte differentiation. Various studies have shown that upregulation of Runx2 induces osteoblastic differentiation, whereas its persistent expression in chondrocytes causes premature maturation and mineralisation (Ref. Reference Enomoto7). Runx2 is expressed in prechondrogenic and preosteogenic condensations, but during early embryogenesis, outside signals regulate its expression (Ref. Reference DeLise, Fischer and Tuan5). Homeobox protein A2 (Hoxa-2) and Wnt (wingless-type mouse mammary tumor virus (MMTV) integration site) signalling have both been implicated in this process. Through specific inactivation in the developing limbs of mice, β-catenin, a downstream effector protein of Wnt signalling, was found to be essential to bone development as these embryos had impaired osteoblast differentiation (Ref. Reference Hill8). Also, it was found that if β-catenin is activated, it upregulates Runx2 expression and MSCs differentiate to osteoblasts, rather than chondrocytes (Refs Reference Hill8, Reference Day9, Reference Hu10). These results implicate β-catenin in the determination of osteo- versus chondrogenesis.
Chondrogenesis
Chondrocytes are the first skeletal cells to arise, and their differentiation depends on the expression of the transcription factor sry (sex determining region Y)-box 9 (Sox9). The importance of Sox9 was discovered when a genetic mutation in humans was found to be the cause of campomelic dysplasia, characterised by severe cartilage abnormalities (Refs Reference Foster11, Reference Wagner12). Genetic deletion of Sox9 in mice blocks chondrocyte differentiation at the point of mesenchymal condensation, indicating that it is necessary for the induction of chondrocyte differentiation (Ref. Reference Augello and De Bari13). In the prechondrogenic mesenchyme, Runx2 is expressed along with Sox9 (Ref. Reference Smith14). However, studies show Sox9 to be dominant over Runx2 (Ref. Reference Zhou15) by promoting Runx2 protein degradation as well as inhibiting its transcriptional activity (Ref. Reference Cheng and Genever16). In addition, Nkx3.2 inhibits Runx2 transcription, enforcing the differentiation to chondrocytes (Ref. Reference Lengner17). Sox9, along with co-activators Sox5 and Sox6, binds and activates promoters of chondrocyte-specific genes. These include collagen 2α1 (Col2α1) and aggrecan (Ref. Reference DeLise, Fischer and Tuan5).
Also, important to chondrogenesis is chondrocyte maturation. This involves the transition of chondrocytes from proliferating, to nonproliferating and hypertrophic, and eventually apoptotic (Ref. Reference Hartmann2). The expression of Sox9, along with the activation of cAMP response element binding protein (CREB) and c-Fos, maintains the chondrocytes in a proliferative state (Ref. Reference Hartmann2). In order for the chondrocytes to exit the cell cycle and become hypertrophic, Runx2 must be upregulated and Sox9 suppressed. The exit from the cell cycle and apoptosis is necessary for the eventual invasion by osteoblasts and ossification of the bone matrix (Ref. Reference Hartmann2).
Osteoblastogenesis
The transcription factor Runx2 is absolutely necessary for osteoblast differentiation. Runx2 knockout (KO) mice show a total absence of differentiated osteoblasts and, therefore, lack any bone (Refs Reference Otto18, Reference Komori19). As Sox9 suppresses Runx2 activity, osteoblast differentiation occurs when Runx2 is upregulated and stabilised. Twist proteins, transcription factors with important roles in embryogenesis, also negatively regulate Runx2, by blocking its DNA binding domain and inhibiting its ability to upregulate osteoblast-specific genes (Ref. Reference Bialek20). Therefore, the inhibition of Twist proteins, specifically Twist-1 and/or Twist-2, is thought to initiate osteoblastogenesis (Ref. Reference Bialek20). Members of the distal-less (Dlx) family of homeobox proteins (Dlx5 and Dlx6) are important for endochondral ossification in the developing appendicular and axial skeleton (Ref. Reference Kraus and Lufkin21). Dlx5 has been shown to act directly upstream of Runx2, by binding its gene promoter, (Ref. Reference Lee22) and Dlx5/Dlx6 double KO mice exhibit delayed ossification (Ref. Reference Robledo23). Other homeobox transcription factors are msh homeobox 1 and 2 (Msx1/Msx2), although whether their regulation of osteoblast differentiation is positive or negative is still controversial (Ref. Reference Komori24). Msx1/Msx2 may be necessary for the expression of Runx2, as the Msx1/Msx2 double KO fails to express Runx2 (Ref. Reference Han25), although more direct evidence is needed (Fig. 1 summarises major transcriptional regulators of osteoblastogenesis). Wnt signalling has also been implicated in triggering osteoblast differentiation (reviewed in Refs Reference Lin and Hankenson26, Reference Milat and Ng27). There is a β-catenin responsive TCF/Lef binding site on the Runx2 gene promoter, and therefore, may upregulate Runx2 expression (Ref. Reference Dong28). Once upregulated, Runx2 triggers osteoblast differentiation and bone development by binding promoters of osteoblast-specific genes, including Osterix, alkaline phosphatase and Osteocalcin. Once its expression is stabilised, Runx2 inhibits Sox9 transcriptional activity (Ref. Reference Topol29), further pushing the MSC to the osteoblast lineage. In addition to regulating osteoblast differentiation, Wnt signalling also regulates osteoblast number and function. Post-natal deletion of β-catenin in Osterix-expressing cells causes osteopenia, however the number of osteoblasts is increased (Ref. Reference Chen and Long30). Therefore a continued investigation of the role of Wnt signalling in osteoblast function, particularly osteoclast regulation, is needed.
Transcriptional regulators of mesenchymal stem cell (MSC) differentiation to skeletal lineages. Osteochondroprogenitors arise from MSCs and express both Runx2 and Sox9. Multiple transcriptional regulators (including Dlx5, Nkx3.2 and Msx1/Msx2) have been found to modify the expression of Runx2 and Sox9 and, therefore, drive differentiation to chondrocytes or osteoblasts. Supportive references are included in the text.
In summary, it appears that the differentiation to chondrocytes may occur by default, when Runx2 and osteoblast differentiation are suppressed. It is also possible that signals that upregulate Sox9 are responsible for the suppression of the osteoblast differentiation programme. Figure 2 illustrates these two paths of differentiation.
Osteoblast versus chondrocyte differentiation. Differentiation to the chondrocyte lineage may occur, by default, with suppression of Runx2 and osteoblastogenesis. Conversely, osteoblast differentiation may be actively suppressed by Sox9 and/or its regulators.
Cyclic-AMP as a regulator of osteoblastogenesis and chondrogenesis
Although there are master regulators that direct MSC differentiation, various other molecules have been demonstrated to modulate the process. Here, we will focus on those that signal through cyclic-AMP (cAMP). cAMP is a ubiquitous second messenger that is synthesised from ATP by adenylyl cyclases. cAMP levels are regulated by different stimuli, one major effector being G-protein-coupled receptors. These receptors are classified either as stimulatory (Gαs) or inhibitory (Gαi) of adenylyl cyclase. Changing levels of cAMP is translated to the cell through cAMP's action on cyclic nucleotide-gated ion channels, on exchange proteins known as Epacs, and on protein kinase A (PKA) (reviewed in Ref. Reference Sands and Palmer31). PKA activation further perpetuates the signal by phosphorylating different target proteins. Ultimately, cellular transcription can be modified through the cAMP-dependent transcription factors CREB, cAMP response element modulator (CREM) and ATF-1 (reviewed in Refs Reference Sands and Palmer31, Reference Mayr and Montminy32, Reference Lee and Masson33). cAMP is degraded by phosphodiesterases, which remove a phosphodiester bond and produce AMP. These enzymes play a major role, not only in terminating the signal, but also in regulating the amplitude and duration of the signal (Ref. Reference Beavo34).
The role of Gαs signalling in osteoblast differentiation was spurred by the finding that individuals with mutations in the Gαs gene have a bone phenotype. Inactivating mutations in the Gαs gene (GNAS) cause Albright hereditary osteodystrophy (Ref. Reference Wilson and Hall35), whereas activating mutations cause fibrous dysplasia of bone (Ref. Reference Riminucci36). To explore its role in bone development, Hsaio et al. created a mouse expressing an engineered Gαs receptor in osteoblasts and found it to have drastically increased bone mass (Ref. Reference Hsiao37). Conversely, complete KO of Gαs receptors in osteoblasts reduced the number of osteoblasts and impaired bone formation (Refs Reference Sakamoto38, Reference Wu39). Although interpretation of these types of experiments is limited, they support a potential role for cAMP signalling in MSC differentiation. In accordance, parathyroid hormone (PTH), a hormone critical for bone development and homoeostasis, signals through a G-protein-coupled receptor (reviewed in Ref. Reference Datta and Abou-Samra40), reinforcing the importance of cAMP signalling in skeletogenesis. The PTH receptor binds PTH, as well as PTH-related protein (PTHrP), and is coupled to Gαs as well as Gαq (Ref. Reference Abou-Samra41). There is a long history of PTH's effect on osteoblast differentiation. Studies in PTH receptor KO mice show that the signalling is not required for osteoblast differentiation (Ref. Reference Lanske42), however PTH has been demonstrated to enhance osteoblast differentiation (reviewed in Ref. Reference Jilka43). Recently, it was found that PTH interacts with canonical Wnt signalling and FGF-2 signalling and that this interaction enhances osteoblast differentiation (reviewed in Ref. Reference Fei and Hurley44). Also, PTH signalling was found to interact with bone morphogenetic protein signalling, which additionally enhances osteoblast differentiation (Refs Reference Yu45, Reference Takase46, Reference Nakao47, Reference Chan48).
The direct effect of cAMP on chondrogenesis has not been fully investigated. In one study, treatment of rabbit chondrocyte cultures with cAMP analogues was found to suppress terminal differentiation and hypertrophy of chondrocytes. Similarly to osteoblasts, PTH is not required for chondrocyte differentiation (Ref. Reference Lanske42). However, PTH treatment of C3H10T1/2 cells can enhance the early stages of chondrocyte differentiation while suppressing chondrocyte maturation (Ref. Reference Hollnagel, Ahrens and Gross49). In addition, limb explants from PTHrP KO mice displayed accelerated chondrocyte maturation (Ref. Reference Vortkamp50).
cAMP effect on transcriptional regulators of osteoblastogenesis
Studies have shown that increasing intracellular levels of cAMP in cell lines or in primary mouse and human MSCs, through cAMP analogues or forskolin (a direct adenylyl cyclase activator), has a positive effect on Runx2 expression (Refs Reference Ghayor51, Reference Siddappa52, Reference Lo53, Reference Carroll54). In addition, activation of receptors that signal through cAMP increases Runx2 expression in MSC cell lines and in primary rodent MSCs. These include the PTH receptor (Ref. Reference Krishnan55) and the A2B adenosine receptor (Refs Reference Carroll54, Reference Gharibi56). This increase in expression may be owing to cAMP activation of CREB, as the Runx2 promoter contains a functional CREB domain (Ref. Reference Xiao57). Another likely effect of cAMP is its modulation of Runx2 activity. Runx2 has putative PKA phosphorylation sites in its activation domain, and treatment with a PKA-specific cAMP analogue increases its phosphorylation (Refs Reference Selvamurugan58, Reference Selvamurugan59) and activation of the collagenase-3 promoter (Ref. Reference Selvamurugan58) and the matrix metalloproteinase-13 promoter (Ref. Reference Selvamurugan59). The role that PKA phosphorylation of Runx2 has in vivo during osteoblast differentiation remains to be determined.
As mentioned earlier, Osterix is another transcription factor essential for osteoblast differentiation and bone development. As Runx2 KO mice do not express Osterix (Refs Reference Otto18, Reference Komori19) but Osterix KO mice do express Runx2, it was deduced that Osterix is downstream of Runx2 (Ref. Reference Nakashima60). Analysis of the Osterix gene promoter identified a Runx2 binding site, and Runx2 was reported to upregulate Osterix expression (Ref. Reference Nishio61). A putative CREB site was also found in this gene promoter, suggesting cAMP may be able to regulate Osterix expression (Ref. Reference Nishio61). In support of this, treatment of cells lines or primary mouse MSCs with cAMP analogues increases Osterix expression (Refs Reference Ghayor51, Reference Carroll54, Reference Wang62). However, exposure to relatively high cAMP (1 mm) inhibits its expression in UM-106-01 cell line or in mouse primary osteoblasts (Ref. Reference Hong63). Similarly to Runx2 expression, Gαs receptor activation causes an increase in Osterix expression (Ref. Reference Carroll54), the mechanism of which remains to be demonstrated.
Dlx5, an upregulator of Runx2 (Ref. Reference Lee22), is phosphorylated by PKA, which increases Dlx5 protein levels by augmenting its stability. PKA signalling also increases Dlx5 transcriptional activity. Therefore, PKA signalling enhances Dlx5-induced osteoblast differentiation (Ref. Reference Han64).
Of note, some have reported a negative effect of cAMP on osteoblast differentiation. Yang et al. found that the PKA inhibitor, PKI, increased the expression of Runx2 in a human MSC-derived cell line whereas forskolin, 3-isobutyl-1-methylxantine and a cAMP analogue decreased the expression of osteopontin (Ref. Reference Yang65). Koh et al. found forskolin to decrease osteocalcin expression in MC3T3-E1 cells. Forskolin also decreased the number of mineralised nodules formed by rat primary calvarial cells (Ref. Reference Koh66). Tintut et al. reported that treatment of MC3T3-E1 cells with forskolin decreased the activity of alkaline phosphatase and inhibited mineralisation, as well as decreased the expression of alkaline phosphatase, bone sialoprotein, osteocalcin and osteopontin (Ref. Reference Tintut67). The discrepancy in reported effects of cAMP on osteoblastogenesis may be due, in part, to the relatively high concentration of forskolin or cAMP analogue used (10–100 µm). For instance, Turksen et al. found that treatment with 10 µm of forskolin inhibited osteoblast differentiation whereas 1 nm increased it (Ref. Reference Turksen68).
cAMP effect on Sox9 and chondrogenesis
Multiple studies have suggested synergism between cAMP signalling and Sox9 expression. The Sox9 promoter contains a CRE site (Ref. Reference Kanai and Koopman69) and, using a Sox9 gene promoter reporter construct in various MSC cell lines, it was reported that binding of this site by CREB increases Sox9 promoter activity (Ref. Reference Piera-Velazquez70).
Sox9 interacts with CREB binding protein (CBP) and p300 to increase its transcriptional activity. Using a Col2α1 gene promoter reporter construct in a chondrocyte cell line, as well as a Gal4-Sox9 fusion protein, it was reported that co-transfection with CBP and/or p300 increased Sox9 activity as a transcriptional activator (Ref. Reference Tsuda71).
A yeast two-hybrid screen of Sox9 binding partners identified the PKA catalytic subunit as interacting with Sox9 (Ref. Reference Huang72). This prompted the investigation of PKA-dependent Sox9 phosphorylation and two serine residues were found to be phosphorylated (Ref. Reference Huang72). Phosphorylation of these sites by PKA increased Sox9 activation of the Col2α1 gene promoter (Ref. Reference Huang72). Zhao et al. investigated this finding further by mutating the serine residues. The absence of PKA phosphorylation partially reversed PKA enhancement of Sox9 activity, suggesting that PKA augmented Sox9 activity through multiple mechanisms. The investigators found an interaction between Sox9 and CBP in chondrogenic differentiating C3H10T1/2 cells by using co-immunoprecipitation, and this interaction increased Sox9 transcriptional activity. The importance of this interaction was confirmed by mutating a CREB site in the region of the Sox9 interaction, which reversed the effect (Ref. Reference Zhao, Li and Zhou73). Therefore, cAMP signalling, by means of PKA, enhances Sox9 expression and transcriptional activity through direct phosphorylation and activation of CREB. In addition to findings in a cell line, differentiation of chick limb bud chondrocytes in the presence of the PKA inhibitor H89 blocked chondrogenesis. Yoon et al. found PKA to promote chondrogenesis by downregulating N-Cadherin through a PKC-dependent pathway (Ref. Reference Yoon74).
Taken together, it appears that cAMP signalling is capable of upregulating Runx2, a master regulator of osteoblasts, and Sox9, a master regulator of chondrocytes. Since Sox9 is an inhibitor of Runx2 expression, it is possible that the sum effect of cAMP on lineage determination depends on its concentration, cellular localisation and a possible regulatory loop.
Adenosine receptors
Adenosine is a regulatory metabolite and its receptors are comprised of a family of seven transmembrane domain G protein-coupled receptors. A1 and A3 adenosine receptors (A1AR and A3AR) are coupled to Gαi and are, therefore, adenylyl cyclase inhibiting, whereas A2A and A2B adenosine receptors (A2AAR and A2BAR) are coupled to Gαs, which stimulates adenylyl cyclase and produces a cAMP signal. The A2BAR may also be coupled to Gαq, which activates phospholipase C (Ref. Reference Fredholm75). A2BAR has a relatively low affinity for adenosine. High extracellular concentrations of adenosine can be achieved during cell injury or stress (Ref. Reference Fredholm76). A2AAR and A2BARs are widely expressed to varying degrees, with high expression in the vasculature (Ref. Reference St Hilaire77). Our laboratory and others have shown that A2BAR expression is inducible under stress conditions such as inflammation and hypoxia (reviewed in Ref. Reference Hasko78). A2AAR and A2BAR are expressed in MSCs (Refs Reference Gharibi56, Reference Costa79), and interestingly, high expression of the A2AAR and A2BAR receptors can be found in cartilage (Ref. Reference Koolpe, Pearson and Benton80).
Adenosine receptors and osteoblast and chondrocyte differentiation
Only recently has the role of adenosine receptors in osteoblast differentiation been examined. In a broad sense, purinergic signalling was looked at in the context of bone, and ATP receptors (P2X and P2Y) were found to promote differentiation and proliferation in an osteoblast cell line (MC3T3-E1) (Ref. Reference Shimegi81). It was found that after bone injury and when exposed to hypoxic conditions, rat osteoblasts secrete ATP in the high nm to μm range (Ref. Reference Orriss82). This extracellular ATP is available for catabolism to adenosine by ectonucleotidases, which are expressed on osteoblasts (Ref. Reference Costa79). In accordance, it was recently demonstrated that genetic ablation of the ectonucleotidase CD73 results in osteopenia and decreased osteoblast differentiation in mice (Ref. Reference Takedachi83).
In vitro studies found adenosine receptors to be expressed in both human (Ref. Reference Costa79) and rodent MSCs (Ref. Reference Gharibi56). Based on its expression and measurements of cAMP levels after agonist treatments of rat MSCs, Gharibi et al. concluded that the A2BAR is the dominant receptor, relative to other adenosine receptors, and that its expression increases during osteoblast differentiation (Ref. Reference Gharibi56). In human MSCs, A2BAR activation increased osteoblast differentiation, as determined by an increase in alkaline phosphatase activity (Ref. Reference Costa79). Similarly, in rat MSCs, activation of the A2BAR increased Runx2 and alkaline phosphatase expression, as well as the number of mineralised nodules (Ref. Reference Gharibi56). We found bone marrow-derived MSCs from A2BAR KO mice to have decreased osteoblast differentiating potential, with diminished expression of Runx2 and Osterix. Activation of the A2BAR with pharmacological agonists increased the expression of these transcription factors, as well as caused an increase in the number of mineralised nodules. Treatment with a cAMP analogue also increased the expression of Osterix, suggesting that cAMP may be the mechanism of action for A2BAR effect on differentiation. In addition, the A2BAR KO mouse had mild osteopenia, and a delayed or impaired bone fracture healing response (Ref. Reference Carroll54).
There have been few reports on the effect of the cAMP inhibitory adenosine receptors, A1AR and A3AR on osteoblast differentiation. Overexpression of the A1AR in an osteoblast precursor cell line led to inhibition of osteoblast differentiation, and instead promoted differentiation to adipocytes (Ref. Reference Gharibi84). However, treatment of human MSCs with the A1AR agonist, N6-cyclopentyladenosine, caused an increase in osteoblast differentiation (Ref. Reference Costa79). In rat MSCs, very little A3AR was found (Ref. Reference Gharibi56) and in human MSCs, treatment with the A3AR agonist, 1-Deoxy-1-[6-[[(3-iodophenyl)methyll]amino]-9H-purin-9-yl]-N-methyl-β-d-ribofuranuronamide, had no effect on osteoblast differentiation (Ref. Reference Costa79).
A role for adenosine receptors in chondrocyte differentiation has not yet been examined and/or reported. However, as A1- and A2-type ARs are known to be expressed in MSCs (Refs Reference Gharibi56, Reference Costa79), and given the published influence of cAMP on Sox9 (as described above), it is likely that a role for these receptors on the differentiation of this lineage will be found.
Inflammation and osteoblast and chondrocyte differentiation: implication for adenosine receptors
Inflammation plays a significant role in bone development (reviewed in Ref. Reference De Benedetti85) and regeneration (reviewed in Ref. Reference Mountziaris and Mikos86). Although adenosine receptor signalling through cAMP has the potential to directly influence osteo- or chondrogenesis, cAMP signalling through these receptors can also affect the level of inflammatory cytokines systemically or at the cellular level (reviewed in Ref. Reference Blackburn87). Here, we will focus on a brief survey of the effects of inflammatory processes on bone cell differentiation, followed by a summary of A2-type adenosine receptors effects on inflammation and its potential influence on bone cell lineages.
The effect of inflammation on osteoblast differentiation
The effect of inflammation on the skeletal system, including osteoblast differentiation, has been well studied, particularly the effects of tumour necrosis factor-α (TNF-α) (reviewed in Ref. Reference Nanes88). Experiments by Gilbert et al. in both fetal calvarial cells and the osteoblast precursor cell line MC3T3-E1 show that TNF-α inhibits osteoblast differentiation, specifically at the early stage of lineage commitment (Ref. Reference Gilbert89). Further experimentation showed this inhibition to be associated with downregulation of Runx2 transcription and a subsequent decrease in nuclear Runx2 (Ref. Reference Gilbert90). Although these authors did not find NFκB signalling to be involved, Huang et al. found that inhibition of NFκB signalling by overexpression of IκB in ST2 cells, a MSC line, abolished the inhibitory effect of TNF-α on Runx2 gene expression (Ref. Reference Huang91). The inhibitory effect of TNF-α on mRNA expression and osteoblast differentiation was confirmed in primary mouse MSCs. Here, Lacey et al. also found interleukin-1β (IL-1β) to have a similar effect (Ref. Reference Lacey92).
These studies on mRNA expression were complimented by Kaneki et al. who found TNF-α to promote Runx2 protein degradation in C2C12 and 2T3 osteoblast precursor cells by upregulating the E3 ligases Smurf1 and Smurf2 (Ref. Reference Kaneki93). In order to examine this phenomenon in primary MSCs, Zhao et al. isolated an MSC-enriched fraction from the bone marrow of TNF-α overexpressing mice. In these cells they found upregulation of the E3 ligase Wwp1 relative to cells from wildtype (WT) mice. Further experimentation found Wwp1 to be upregulated by TNF-α and responsible for inhibiting osteoblast differentiation. However, in these primary cells, the inhibition was caused by the degradation of JunB, a promoter of osteoblast differentiation, rather than affecting Runx2 directly (Ref. Reference Zhao94).
In addition to their studies on Runx2, Lu et al. found TNF-α treatment to inhibit Osterix mRNA expression, and claim this to be a direct effect of TNF-α signalling through mitogen activated (MEK) and inhibition of the Osterix gene promoter (Ref. Reference Lu95). In addition, they found TNF-α-stimulated binding of paired mesoderm homeobox protein 1(Prx1), causing inhibition of Osterix gene promoter activity and transcription in MC3T3 and C3H10T1/2 cells (Ref. Reference Lu96). Interestingly, Prx1 is a developmental regulator of skeletogenesis that was previously thought to be silenced after embryogenesis (Ref. Reference Lu96).
Other regulators of osteoblast differentiation are affected by TNF-α. In culture, Msx2 has been shown to inhibit osteoblast differentiation (Ref. Reference Lee97). Treatment of C2C12 or fetal calvarial cells with TNF-α caused increased expression of Msx2 and reduced expression of alkaline phosphatase. This effect was independent of Runx2, as the phenomenon was maintained in Runx2 null cells, and overexpression of dominant negative IκB showed NFκB signalling to be involved. To model inflammation induced by wear of artificial limbs, macrophages were activated with titanium particles and their media, enriched in inflammatory cytokines, was used to treat MC3T3-E1 cells. Treatment with this conditioned media or TNF-α inhibited Runx2 expression and osteoblast differentiation and this was attributed to NFκB activation and increased expression of sclerostin, an inhibitor of the Wnt pathway (Ref. Reference Lee98).
The effect of inflammation on chondrocyte differentiation
Relative to osteoblastogenesis, little is known about the effect of inflammation on chondrocyte differentiation. IL-1 and TNF-α decreased Sox9 mRNA and protein expression in a chondrocytic cell line, MC615, and in primary mouse chondrocytes through the NFkB signalling pathway (Refs Reference Murakami, Lefebvre and de Crombrugghe99, Reference Sitcheran, Cogswell and Baldwin100). However, in these studies differentiation was not directly addressed. To study chondrocyte differentiation, Nakajima et al. differentiated a chondrocyte progenitor cell line, ATDC5, in the presence of IL-6. IL-6 decreased collagen 2α1 and collagen 10 mRNA expression in a dose-dependent manner and inhibited the formation of cartilage nodules (Ref. Reference Nakajima101). Wehling et al. differentiated human MSCs to chondrocytes in the presence of either IL-1 or TNF-α. They found that both cytokines decreased the size of the cartilage pellet and lowered the amount of glycosaminoglycan accumulation. IL-1 treatment decreased the expression of Col2α1 and aggrecan mRNA (Ref. Reference Wehling102).
Taken together, it is then quite possible that adenosine receptor modulation of inflammatory cytokines affects directly or indirectly the process of MSC differentiation into bone cell lineages. Figure 3 illustrates a proposed mechanism of A2BAR effects on osteoblast differentiation involving CREB and/or TNF-α.
Proposed mechanism for A2BAR action on osteoblast differentiation. Activation of the mesenchymal stem cell A2BAR triggers cAMP signalling, which may modulate the expression and/or activity of a key osteoblast transcription factor, Runx2, and promote osteoblast differentiation. A2BAR activation also decreases tumour necrosis factor-α level, which has an inhibitory effect on osteoblast differentiation.
Adenosine receptors and inflammation
One of the principal attributes of adenosine and its receptors is its pro- or anti-inflammatory effects, most of which are mediated via cAMP signalling. The concentration of our laboratory has been the A2BAR, and thus will be the focus of this section. Both pro- and anti-inflammatory effects of the A2BAR have been described, and these depend on the cell type and stimulus. In addition, these effects have been ascribed either protective or deleterious roles, depending on the context, e.g. chronic versus acute pathology (reviewed in Ref. Reference Johnston-Cox, Koupenova and Ravid103). Complete KO of the A2BAR gene results in a slight systemic inflammation, as KO animals have elevated plasma levels of TNF-α at baseline, and elevated levels of TNF-α and IL-6 expression in macrophages (Ref. Reference Yang104). These differences in cytokine levels are exacerbated upon stress or injury. Treatment of A2BAR KO mice with lipopolysaccharide causes an exceptional increase in TNF-α and IL-6 plasma levels and in macrophage expression, relative to WT mice (Ref. Reference Yang104).
The ability of the A2BAR to dampen inflammation in response to stimuli highlights its protective role during stress or injury. The importance of this role is confirmed by findings that the expression of the receptor is also induced by these stimuli. Treatment of vascular smooth muscle cells with TNF-α causes an upregulation of A2BAR expression. Further, this increase was shown to be mediated by NADPH oxidase 4 (Nox4) signalling (Ref. Reference St Hilaire105).
As NFκB has an important role in the regulation of cellular inflammation it has been a target of investigation in relation to A2BAR signalling. Recently, we found the A2BAR to directly bind p105, an inhibitor of NFκB, stabilising it and preventing its degradation (Ref. Reference Sun106). This mechanism helps explain the inflammatory phenotype in the A2BAR KO mice and the ability of the A2BAR to downregulate inflammation.
Considering the above described effects of TNF-α, IL-1 and IL-6 on bone cell differentiation, it is possible that some of the A2BAR's recently described protective effect in a mouse bone injury model (Ref. Reference Carroll54) is related to changes in cytokine levels.
Adenosine receptors as therapeutics for osteoporosis, skeletal injury and arthritis
A role for adenosine receptors in osteoporosis is gaining increasing momentum. In mouse bone marrow-derived cells, antagonism of the A1AR with 8-cyclopentyl-1,3-dipropylxanthine inhibited the differentiation of osteoclasts (bone reabsorbing cells) (Ref. Reference He and Cronstein107). Also in these cells, treatment with an A2AAR agonist, 4-[2-[[6-Amino-9-(N-ethyl-β-d-ribofuranuronamidosyl)-9H-purin-2-yl]amino]ethyl]benzene propanoic acid hydrochloride (CGS21680), inhibited the differentiation of osteoclasts, as well as inhibited their activity. In addition, the A2AAR KO mouse has an increased number of osteoclasts, as well as reduced bone volume (Ref. Reference Mediero108). Similarly, A2BAR KO mice display a mild reduction in bone density (Ref. Reference Carroll54). In our study, we also found that A2BAR KO mice have changes in bone fracture healing with decreased callus bone formation and an apparent delay in healing (Ref. Reference Carroll54). Although the anti-inflammatory effects of adenosine likely have an important role in improving skeletal injury repair, we contend that adenosine, through the A2BAR, may also improve healing by promoting osteoblast differentiation (Ref. Reference Carroll54). Therefore, agonism of the A2-type ARs may be useful as a therapeutic for osteoporosis and bone injury.
As caffeine is an antagonist of adenosine receptors, its effects on bone may be relevant here. In epidemiological studies of risk factors for osteoporosis in humans, caffeine was either found to be negatively associated with (Refs Reference Rapuri109, Reference Ilich110) or not associated with bone mineral density (Refs Reference Wetmore111, Reference Waugh112, Reference Hannan113, Reference Conlisk and Galuska114). It is possible that effects of caffeine are confounded by factors such as age, oestrogen levels and calcium intake. In a study of over 3000 individuals using the Framingham Cohort, it was found that caffeine intake was associated with a higher relative risk of hip fracture (Ref. Reference Kiel115). In experimental animals, caffeine has been shown to inhibit bone formation. When demineralised bone particles were implanted subcutaneously, rats that were treated with caffeine had decreased chondrogenesis and decreased mineralisation (Ref. Reference Barone116), suggesting that caffeine impairs new endochondral bone formation by inhibiting the proliferation and differentiation of chondroprogenitor cells. In addition, chick osteoblasts treated with caffeine had decreased collagen expression and alkaline phosphatase activity, resulting in reduced matrix formation (Ref. Reference Tassinari117). Finally, differentiation of osteoclasts from mouse bone marrow-derived cells was enhanced with caffeine treatment (Ref. Reference Liu118). Whether any or all of these negative effects of caffeine on bone formation and maintenance is solely because of antagonism of adenosine receptors has not been determined.
Adenosine receptors have been found to be protective against a variety of injuries, including but not limited to cardiovascular (reviewed in Ref. Reference Riksen and Rongen119), kidney (reviewed in Ref. Reference Bauerle120), lung (reviewed in Ref. Reference Eckle, Koeppen and Eltzschig121) and gastrointestinal (reviewed in Ref. Reference Colgan and Eltzschig122). As arthritis is an inflammatory disease of the joints, adenosine receptors have been investigated in the context of this disease. Direct infusion of adenosine into the joint in a rat arthritis model reduced the pathogenesis of the disease (Ref. Reference Green123). Methotrexate is a drug commonly used for the treatment of rheumatoid arthritis and ameliorates the condition by decreasing the inflammatory response in the joints. At least part of its inflammatory action is attributed to its ability to increase adenosine release in the joints (Ref. Reference Cronstein, Naime and Ostad124). Currently, the principal receptor to be implicated is the A3AR. An A3AR receptor agonist, CF-101, has been undergoing clinical trials for the treatment of rheumatoid arthritis; however, the improvement in rheumatoid arthritis has not yet reached statistical significance (Ref. Reference Silverman125).
Conclusions
There are a multitude of signalling molecules and pathways that converge during chondrocyte and osteoblast differentiation. The coordination of the spatial and temporal pattern of these molecules is necessary for correct bone formation. Through exploration of the downstream signalling of G-protein-coupled receptors, cAMP was found to be an important component of the signalling pathways needed for full differentiation of MSCs along the skeletal lineage. Continued investigation into the contribution of these receptors, adenosine receptors among them, to bone homoeostasis and regeneration could lead to important discoveries with clinically therapeutic implications.
