Adenosine receptor subtype expression and activation influence the differentiation of mesenchymal stem cells to osteoblasts and adipocytes


  • Borzo Gharibi,

    1. Center for Endocrine and Diabetes Sciences, School of Medicine, Cardiff University, Cardiff, United Kingdom
    2. Department of Child Health, School of Medicine, Cardiff University, Cardiff, United Kingdom
    Search for more papers by this author
  • Anju A Abraham,

    1. Department of Child Health, School of Medicine, Cardiff University, Cardiff, United Kingdom
    Search for more papers by this author
  • Jack Ham,

    1. Center for Endocrine and Diabetes Sciences, School of Medicine, Cardiff University, Cardiff, United Kingdom
    Search for more papers by this author
  • Bronwen AJ Evans

    Corresponding author
    1. Department of Child Health, School of Medicine, Cardiff University, Cardiff, United Kingdom
    • Department of Child Health, School of Medicine, Cardiff University, Heath Park, Cardiff CF14 4XN, United Kingdom.
    Search for more papers by this author


Osteoblasts and adipocytes differentiate from a common precursor cell, the mesenchymal stem cell (MSC). Adenosine is known to signal via four adenosine receptor subtypes, and significantly, recent findings indicate that these may play a role in MSC differentiation. We therefore investigated adenosine receptor expression and activation during the differentiation of MSCs to osteoblasts and adipocytes. The A2BR was dominant in MSCs, and its expression and activity were transiently upregulated at early stages of osteoblastic differentiation. Both activation and overexpression of A2BR induced the expression of osteoblast-related genes [Runx2 and alkaline phosphatase (ALP)], as well as ALP activity, and stimulation increased osteoblast mineralization. The expression of A2AR was upregulated during later stages of osteoblastic differentiation, when its activation stimulated ALP activity. Differentiation of MSCs to adipocytes was accompanied by significant increases in A1R and A2AR expression, and their activation was associated with increased adipogenesis. Enhanced A2AR expression was sufficient to promote expression of adipocyte-related genes (PPARγ and C/EBPα), and its activation resulted in increased adipocytic differentiation and lipid accumulation. In contrast, the A1R was involved mainly in lipogenic activity of adipocytes rather than in their differentiation. These results show that adenosine receptors are differentially expressed and involved in lineage-specific differentiation of MSCs. We conclude, therefore, that fruitful strategies for treating diseases associated with an imbalance in the differentiation and function of these lineages should include targeting adenosine receptor signal pathways. Specifically, these research avenues will be useful in preventing or treating conditions with insufficient bone or excessive adipocyte formation. © 2011 American Society for Bone and Mineral Research


Osteoporosis and obesity are currently major health problems worldwide and, furthermore, are growing in prevalence. They are both disorders of body composition, and while they were once thought to be mutually exclusive, they are now thought to be closely related and to share several features.1, 2 One of these shared features is that osteoblasts and adipocytes differentiate from a common precursor cell in the bone marrow, the mesenchymal stem cell (MSC). The pluripotency of MSCs is well known, and their ability to differentiate into osteoblasts, adipocytes, chondrocytes, and myoblasts has been described extensively.3 Although numerous studies4–7 have shown that many substances, as well as mechanical agents, contribute to these differentiation processes, the mechanisms involved are yet to be completely defined. There is, however, a large body of evidence to support the notion that there is an inverse relationship between the differentiation of MSCs to osteoblasts and adipocytes. This seems to be true during attainment of peak bone mass8, 9 and in an aging population.10 Furthermore, such a relationship has been shown in individuals with high bone mass11 and following mechanical loading,12 whereas secondary causes of osteoporosis (eg, diabetes, glucocorticoids, and immobility) are associated with bone marrow adiposity.1

Adenosine receptors exist as four subtypes, A1R, A2AR, A2BR, and A3R, that signal through primarily the stimulation (A2AR and A2BR) and inhibition (A1R and A3R) of cyclic adenosine monophosphate (cAMP) and are possible drug targets for many diseases, including osteoporosis.13 We recently described the expression and function of adenosine receptors in a human osteoprogenitor cell line and in primary bone marrow MSCs and showed that A2BR is functionally dominant in these cells.14 We also showed that adenosine has a potent stimulatory action on interleukin 6 (IL-6) release and an inhibitory action on osteoprotegerin expression, suggesting that adenosine receptors are regulators of MSC differentiation, as well as bone formation and resorption.14

Adenosine receptors, in particular, A2AR, have been shown to mediate anti-inflammatory actions and have been implicated in tissue repair and damage.15–18 A2AR has been shown recently to be involved in the proliferation and differentiation of murine bone marrow–derived MSCs.19 A2Rs are also expressed on preadipocytes, whereas A1R is expressed predominantly on mature adipocytes.20, 21 Furthermore, it is known that A1R activation mediates leptin secretion22 and protects against obesity-related insulin resistance.23

In this study we investigated the expression of the adenosine receptor subtypes, as well as adenosine metabolic and catabolic enzymes, in MSCs as they differentiated to osteoblasts and adipocytes. Furthermore, we used receptor agonists and antagonists, as well as receptor overexpression, to dissect out the role of the receptor subtypes during these differentiation processes.

Materials and Methods


General and cell culture materials were from Sigma-Aldrich (Poole, Dorset, UK), Fisher Scientific (Loughborough, UK), Lonza (Slough, UK), Biosera (Ringmer, UK), and Invitrogen (Paisley, UK). Molecular biology reagents were from Promega UK Ltd. (Southampton, UK) or as indicated. Antisera were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA), Abcam (A1R, A2aR, and A3R; Cambridge UK), and Alpha Diagnostic International (A2bR; San Antonio, TX, USA).

Adenosine receptor agonists and antagonists (Tocris Bioscience, Avonmouth, Bristol, UK) were universal agonists, adenosine, and 5'-N-ethylcarboxamidoadenosine (NECA); A1R agonist and antagonist, 2-chloro-N6-cyclpentyladenosine (CCPA) and 1-butyl-8-(hexahydro-2,5-methanopentalen-3a(1H)-yl)-3,7-dihydro-3-(3-hydroxypropyl)-1H-purine-2,6-dione (PSB36); A2AR agonist and antagonist, 3-[4-[2-[6-amino-9-[(2R,3R,4S,5S)-5-(ethylcarbamoyl)-3,4-dihydroxy-oxolan-2-yl]purin-2-yl]amino]ethyl]phenyl]propanoic acid (CGS21680) and 2-(2-furanyl)-7-[3-(4-methoxyphenyl)propyl]-7H-pyrazolo[4,3-e][1,2,4]triazolo[1,5-c]pyrimidin-5-amine (SCH442416); and A2BR antagonist, N-(4-acetylphenyl)-2-[4-(2,3,6,7-tetrahydro-2,6-dioxo-1,3-dipropyl-1H-purin-8-yl)penoxy]-acetamide (MRS1706).

The natural adenosine receptor agonist adenosine is rapidly metabolized in cell culture environments; thus this work relies on the use of the more stable adenosine receptor agonist NECA, although results with adenosine are also discussed where possible. Furthermore, since a subtype-specific A2B agonist is not currently available commercially for use, we have used throughout this study a universal adenosine receptor agonist (NECA), a specific A2A receptor agonist (CG62180), and specific A2A and A2B receptor antagonists (SCH442416 and MRS1706, respectively) to dissect out the roles of the A2A and A2B receptors in MSC differentiation. Thus we have characterized A2B receptors by relying on the lack of effectiveness of compounds that are potent and selective agonists of other receptor types.

Cell culture

Rat bone marrow MSCs resuspended in normal growth medium—α-minimal essential medium (α-MEM) L-glutamine (2 mM), penicillin (50 U/mL), streptomycin (50 µg/mL), and 10% fetal bovine serum (FBS)—were plated (2 × 106 cells/cm2) in 75-cm2 flasks and incubated at 37°C in 5% CO2 and 95% air for 3 days. The medium then was changed to remove nonadherent cells and again every 3 to 4 days thereafter, and the cells were used at passages 0 to 3.

For osteoblast differentiation, cells were seeded (5 × 103 cells/cm2) in normal growth medium, and 24 hours later, differentiation was induced with L-ascorbic acid 2-phosphate (50 µg/mL) and dexamethasone (10−8 M). For mineralization experiments, β-glycerol phosphate (2 mM) also was added. For adipocyte differentiation, MSCs were cultured in normal growth medium until 50% confluent, and adipogenic differentiation was induced with dexamethasone (10−8 M), indomethacin (60 µM), and insulin (10 µg/mL). For most experiments, the medium was replenished every 2 to 3 days. For mineralization experiments involving adenosine, the medium was changed every weekday.

Quantitative real-time polymerase chain reaction (qPCR)

Total RNA was prepared using TRIzol (Invitrogen), and contaminating DNA was removed using the TURBO DNA Free Kit (Applied Biosystems, Warrington, UK). cDNA was prepared from 1 µg of RNA by incubation for 1 hour at 37°C with Moloney Murine Leukemia Virus reverse transcriptase, RNasin, deoxyribonucleotide triphosphate mix, and oligodeoxythymidilic acid [oligo(dT)15]. qPCR was performed on an MX3000P thermal cycler (Stratagene, La Jolla, CA, USA). Each reaction contained SYBR Green Master Mix (12.5 µL; Stratagene), primer mix (1 µL; 10 pmol of each primer, Table 1), cDNA template (1 µL), and nuclease-free water (10.5 µL). The program (40 cycles) consisted of 90°C (30 seconds), 60°C (1 minute), and 72°C (30 seconds). β-Actin was used as an invariant housekeeping gene. The relative quantitative expression of the gene of interest (GOI) relative to β-actin was calculated, and this ratio for basal or untreated cells was assigned a value of 1.

Table 1. Primer Sequences
Primers (all 5'→3')ForwardReverseProduct size (bp)

Western blotting

Cells were lysed in radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 0.1% sodium dodecyl sulfate, and 0.5% sodium deoxycholate) containing protease inhibitors cocktail solution (chymostatin, leupeptin, antipain, and pepstatin all at 10 µg/mL), 1 mM sodium orthovanadate, and 0.1 mg/mL of phenylmethylsulfonyl fluoride (PMSF). Protein concentrations were quantified with the Bio-Rad protein assay (Bio-Rad Laboratories, Ltd., Hertfordshire, UK). Lysates (15 to 30 µg of protein) and prestained protein markers (Cell Signaling Technology, Beverly, MA, USA) were mixed with an equal volume of loading buffer (4% sodium dodecyl sulfate, 2% β-mercaptoethanol, 20% glycerol, 100 mM Tris-HCl, pH 6.8, and 0.004% pyronin Y) and subjected to SDS-PAGE. Proteins were transferred onto polyvinylidene fluoride (PVDF) membranes and incubated overnight at 4°C with specific antisera (used at the recommended concentrations, usually 1:1000 or 1:2000). Secondary antibodies (1:5000) conjugated with horseradish peroxidase were applied for 1 hour at room temperature, and proteins were visualized with ECL Plus reagent. Blots were stripped and reprobed with anti-β-actin (1:1000) for 1 hour at room temperature. Band densities were calculated using an AlphaImager documentation and analysis system (Alpha Innotech Corporation, Santa Clara, CA, USA).

Alkaline phosphatase (ALP) activity assay

Alkaline phosphatase (ALP) activity was determined by an assay based on the hydrolysis of p-nitrophenylphosphate to p-nitrophenol (Sigma-Aldrich). Cells were washed with PBS, and 50 µL of phosphatase substrate/alkaline buffer (pH 10.3) was added to each well. After 20 minutes in the dark, absorbance (405 nm) was read on a SpectraCount microplate photometer (Canberra Packard, Ltd., Oxfordshire, UK). ALP activity was normalized to protein concentration in parallel experimental plates.


Mineralized matrix was stained with alizarin red S (1:100 in distilled water, adjusted to pH 4.2, and filtered). Cells were washed with PBS, fixed (4% formaldehyde in PBS) for 15 minutes, washed with distilled water, stained for 10 minutes, washed (five times) in 50% ethanol, and air dried. For quantification, cells were destained overnight in 10% (w/v) cetylpyridinium chloride at room temperature with continuous agitation. The absorbance (562 nm) was read using a SpectraCount microplate photometer.

Oil red O and Nile red staining of adipocytes

Accumulation of intracellular lipid was assessed by staining with oil red O — a 5% stock solution in isopropanol was used to prepare the working solution of 3% in water.24 Cells were fixed (15 minutes with 4% formaldehyde in PBS), stained for 15 minutes, and then washed with 60% isopropanol and PBS. Lipid accumulation was viewed microscopically, and images were captured with a Penguin 150 CL camera (Pixera Corporation, Santa Clara, CA, USA) through Viewfinder 3.0.1 (Pixera Corporation). For quantification, the dye was extracted with 0.2 mL of isopropanol (1 minute), and the absorbance (490 nm) was read using a SpectraCount microplate photometer.

Nile red (1 mg/mL in dimethyl sulfoxide stock24) was diluted 1:100 in PBS immediately before use. Cells were trypsinized, fixed for 30 minutes (4% formaldehyde in PBS) at 4°C, centrifuged, resuspended in Nile red (1 mL) for 30 minutes at 4°C, and analyzed with a FACSCalibur flow cytometer (Becton Dickinson Biosciences, Oxford, UK). For each sample, 15,000 events were collected, and data were analyzed using CellQuest 3.1 software (Becton Dickinson Biosciences).

Transient transfection of A1R, A2AR, and A2BR into MSCs

Gene constructs for human A1R, A2AR, and A2BR (kind gift from KN Klotz, Institute for Pharmacology und Toxicology, University of Würzburg, Würzburg, Germany) were cloned or subcloned into the plasmid pcDNA3.1. MSCs were transfected using the Human MSC Nucleofector Kit with the Nucleofector machine (Lonza). Cells were trypsinized, and 5 × 105 cells were resuspended in 100 µL of nucleofector solution and treated with 2 µg of pcDNA3.1/A1, pcDNA3.1/A2A, pcDNA3.1/A2B, pcDNA3.1 empty vector (vector control), or positive control vector pmaxGFP to monitor transfection efficiency. The cells were plated in 6-well plates, and after overnight incubation, the medium was changed and efficiency of transfection was determined by light and fluorescence microscopy. RNA was prepared 48 hours after nucleofection and subjected to qPCR analysis.

Statistical analysis

Statistical comparisons were determined using one-way ANOVA (SPSS 16, SPSS, Inc., Chicago, IL, USA) and the Tukey multiple-comparison test. Results were considered significant if p < .05.


Undifferentiated MSCs

Initial experiments demonstrated that the A1R, A2AR, A2BR, and possibly A3R (observed only at a Ct value of 35 or above) mRNA, as well as adenosine metabolic and catabolic enzyme (ie, adenosine deaminase, adenosine kinase, and CD73) mRNA, were expressed in undifferentiated MSCs. Furthermore, Western blotting showed that three receptor protein subtypes (A1R, A2AR, and A2BR) were present, with the A3R not being apparent (Fig. 1). In addition, a cAMP assay indicated that the A2B subtype was predominant, although these assays did not preclude the presence of other functional adenosine receptor subtypes (results not shown). Furthermore, the A3R protein was not detected, whereas MSCs underwent osteoblastic or adipogenic differentiation, and at the same time there was no convincing evidence of the presence of A3R mRNA during these differentiation processes (Figs. 2D and 4D), Thus we did not continue further with studies of the A3R in MSCs.

Figure 1.

Western blot analysis of adenosine receptor protein expression. (A) Expression of A1R, A2AR, and A2BR proteins is indicated by dashed oval shapes. Positive control (PC; BON-1) for the A3R protein is also shown. Left margin indicates the sizes (kDa) of protein markers. (B) Western blot analysis of BON-1 (lane 1) and KRJ-1 (lane 2) cell extracts (positive controls) with A3R antibody. BON-1 and KRJ-1 are human neuroendocrine tumors derived from pancreas and intestine, respectively.

Figure 2.

Expression and function of adenosine receptors during the differentiation of MSCs into osteoblasts. mRNA expression of (A) A1R, (B) A2AR, (C) A2BR, (D) A3R, and (E) CD73 (mean ± SEM of three experiments in duplicate). (F) Representative Western blots probed for A1R, A2AR, and A2BR. (G–I) cAMP concentrations in MSCs differentiated to osteoblasts for 9 days and following incubation with (G) adenosine, (H) NECA, and (I) CGS21680 for 15 minutes in the presence of RO20-1724 (10−4 M) (mean ± SEM of two experiments in quadruplicate). *p < .05, **p < .01, and ***p < .001 when compared with untreated cells at day 0; ##p < .001 and ###p < .001 when compared with untreated cells at the same time point.

Adenosine receptor mRNA and protein expression during osteoblastogenesis

During osteoblastogenesis, the mRNA expression of A1R (Fig. 2A) was upregulated by up to eightfold between days 5 and 12, whereas the A2AR (Fig. 2B) and A2BR (Fig. 2C) mRNA showed small increases (twofold). Interestingly, A2BR mRNA expression also appeared to decrease beyond day 7. No changes were observed for A3R mRNA expression (Fig. 2D). CD73 mRNA decreased by 50% (p < .05) by day 2 of osteoblastic differentiation and then progressively returned to levels seen in undifferentiated cells (Fig. 2E). There were no changes in adenosine deaminase expression during osteoblast differentiation.

In contrast to changes in A1R mRNA expression during osteoblastogenesis, the A1R protein remained relatively unchanged. A representative blot is shown in Fig. 2F, and combining densitometric values from three separate experiments confirmed that A1R protein expression did not change significantly during differentiation to osteoblasts. This difference may be related to the already high expression of the protein in undifferentiated MSCs, which could limit further translation of mRNA. Changes in A2AR protein expression were similar to those seen with mRNA expression during differentiation. A relatively faint protein band was detected in undifferentiated MSCs that increased gradually in intensity during osteoblast differentiation and was maximal at days 7 to 12. The A2BR protein expression remained unchanged on day 1 but was strongly upregulated on days 2 and 7, but by day 12 its expression had fallen back to basal levels. A similar pattern of expression was seen for A2BR mRNA during osteoblastogenesis, but the increase in protein level persisted beyond that of the mRNA, probably because the protein signal is sustained after the mRNA had been degraded.

Effects of adenosine receptor agonists on cAMP accumulation

Treatment with adenosine (natural ligand) and NECA (universal agonist) increased cAMP accumulation in a dose-related manner at all time points, indicating the predominance of A2Rs at all stages of osteoblastogenesis (see Fig. 2G, H for day 9 time point). When compared with undifferentiated MSCs (day 0), the efficiency of adenosine and NECA in stimulating cAMP production increased during the time of osteoblastic induction. On day 0, adenosine (10−4 M) stimulated cAMP levels ninefold, and this was increased to 153-fold (p < .001) after 9 days of differentiation. The efficacy of NECA to increase cAMP also was increased significantly from 25-fold in MSCs to 275-fold at 9 days (p < .001) of osteoblastogenesis (Fig. 2H). Furthermore, MRS1706 (10−6 M) significantly reduced NECA (10−5 M) stimulated cAMP accumulation (19.4-fold reduced to 11.6-fold, respectively, p < .001) on day 9 of osteoblastogenesis. In addition, SCH442416 (10−6 M) significantly (p < .001) reduced this effect of NECA (21.2-fold reduced to 13.1-fold with SCH442416).

CGS21680 (A2AR agonist) had little effect on cAMP expression in undifferentiated MSCs (Fig. 2I), suggesting an absence of A2AR. On osteoblastic induction, however, CGS21680 significantly stimulated cAMP production, with maximum efficacy on day 9 (p < .001; Fig. 2I). During osteoblastogenesis, although we have shown an increase in A2AR expression, the observed maximal stimulation of cAMP accumulation by NECA was approximately 20-fold more than that seen with CGS21680 at all time points. Together, these data indicate that A2AR and A2BR expression increased during osteoblastogenesis of MSCs and concur with previous data showing increased mRNA and protein expression. Nevertheless the A2BR remained as the dominant receptor in MSC-derived osteoblasts. Furthermore, CCPA did not affect forskolin-stimulated cAMP production (results not shown).

Effects of adenosine receptor agonists and antagonists on osteoblast markers during the differentiation of MSCs to osteoblasts

All the osteoblast markers examined (ie, Runx2, ALP, collagen type 1, and osteocalcin) increased during induced osteoblastogenesis of MSCs, and Runx2 and ALP expression increased even further in the presence of NECA (10−6 M). Runx2 mRNA was increased fourfold on day 7 and was further enhanced by NECA to sevenfold (Fig. 3A; p < .05). Adenosine, however, had no effect. Both NECA (Fig. 3B) and adenosine (results not shown) stimulated ALP mRNA expression, but neither had a significant effect on the expression of collagen type 1 or osteocalcin mRNA. The late osteoblast marker, osteocalcin, however, showed a particularly large change during osteoblastogenesis, with a greater than 10,000-fold increase in mRNA after 7 days.

Figure 3.

Adenosine receptor signal pathway stimulation of osteogenesis in MSCs. mRNA expression of (A) Runx2 and (B) ALP following treatment with NECA (10−6 M) for 2 and 7 days. (mean ± SEM of three experiments in duplicate). *p < .05 and ***p < .001 when compared with day 0. #p < .05 when compared with vehicle at the same time point. (C) ALP mRNA expression following treatment with 10−6 M SCH442416 and MRS1706 alone or in the presence of NECA (10−6 M) for 2 and 5 days (mean ± SEM of three experiments in duplicate). ***p < .001 when compared with untreated cells. ##p < .01 when compared with NECA-treated cells at the same time point. (D, E) ALP activity following treatment with NECA and CGS21680 for 2, 5, 7, 9, and 12 days (mean ± SEM of three experiments in quadruplicate). *p < .05, **p < .01, and ***p < .001 when compared with vehicle at the same time point. (F, G) ALP activity following treatment with 10−6 M SCH442416 or MRS1706 alone or in combination with NECA (10−6 M) (mean ± SEM of three experiments in quadruplicate). ***p < .001 when compared with untreated cells. ###p < .001 when compared with NECA alone–stimulated cells at the same time point. (H, I) Mineralization of MSCs following treatment with adenosine (10−4 M) alone or in combination with 10−6 M to 10−7 M MRS1706. (H) Representative photograph [control (c), adenosine (10−4 M, a), adenosine (10−4 M) + MRS1706 (10−7 M) (a*), and adenosine (10−4 M) + MRS 1706 (10−6 M) (a⁁)]. (I) Mineralization results following elution of alizarin red S and spectrophotometric measurement. (Mean ± SEM from three experiments in triplicate). ***p < .001 when compared with vehicle control. ##p < .001 and ###p < .001 when compared with adenosine-stimulated.

Initial cell number experiments with receptor antagonists (ie, A2AR, SCH442416, and A2BR, MRS1706) demonstrated that concentrations up to 10−6 M were without effect. These concentrations were used subsequently, and exposure to antagonists alone for 2 or 5 days did not induce significant changes in ALP expression (Fig. 3C). In the presence of either SCH442416 or MRS1706, however, the NECA-induced increase in ALP expression was significantly reduced (p < .01).

Effects of adenosine receptor agonists and antagonists on ALP enzyme activity

In addition to effects on ALP mRNA expression, NECA (Fig. 3D) and adenosine (results not shown) both induced a dose-dependent increase in ALP enzyme activity over 5 to 7 days of osteoblast differentiation. The effects of NECA occurred earlier (day 2) and appeared more potent than those of adenosine, probably a reflection of the stability of NECA in comparison with adenosine. Beyond day 7 of agonist exposure, NECA and adenosine had little effect on ALP enzyme activity. Similar stimulatory effects on ALP enzyme activity were seen with the CGS21680 (Fig. 3E); interestingly, this increase in enzyme activity seemed to persist for up to at least 12 days. CCPA did not modulate ALP enzyme activity.

When receptor antagonists (ie, A2AR, SCH442416, and A2BR, MRS1706) were used, exposure to antagonists alone for 2 or 5 days did not induce significant changes in ALP enzyme activity (Figs. 2G and 3F). In the presence of SCH442416, however, NECA-induced ALP enzyme activity was partially (∼75%) but significantly reversed (p < .001; Fig. 3F). In addition, the A2BR antagonist MRS1706 abolished the increase in ALP mRNA expression and ALP enzyme activity induced by NECA on days 2 and 5 of differentiation (Fig. 3G).

Effects of adenosine receptor agonists and antagonists on mineralization of MSC-derived osteoblasts

Since A2AR and A2BR seemed to play an important role in osteoblastogenesis, we also investigated whether activation of these receptors stimulated matrix mineralization. Moderate alizarin red S staining was observed in untreated MSCs after 10 to 12 days, and this was further enhanced (p < .001), respectively, by 109%, 50%, and 37% with adenosine (10−4 M), adenosine (10−5 M), or NECA (10−5 M) (Figs. 2I and 3H and results not shown). Mineralization, however, was unaffected by lower concentrations of NECA or adenosine. Surprisingly, CGS21680 (previously shown to induce ALP enzyme activity) had no effect on mineralization (Supplemental Fig. S1), and mineralization also was unaffected by the A1R agonist CCPA (Supplemental Fig. S1). The A2BR antagonist MRS1706 inhibited the increase in mineralization induced by adenosine by up to 95% (p < .001; Fig. 3I), whereas the A2AR antagonist SCH442416 had no effect (Supplemental Fig. S2). The lack of effect by CGS21680 and the lack of inhibition of adenosine-induced mineralization by SCH442416 suggest that the A2BR was responsible for the induction of mineralization.

Adenosine receptor mRNA and protein expression during adipogenesis

During adipogenesis, A1R mRNA expression was strongly upregulated with an 800-fold increase on day 12 (p < .001). Changes in A2AR mRNA paralleled those of A1R but only showed a 15- to 20-fold increase at 12 days (p < .001). There were no changes in the expression of A2BR, A3R (Fig. 4C, D), CD73, or adenosine deaminase (results not shown) during adipogenesis.

Figure 4.

Expression and function of adenosine receptors during the differentiation of MSCs into adipocytes. mRNA expression of (A) A1R, (B) A2AR, (C) A2BR, and (D) A3R (mean ± SEM of three experiments in duplicate). (E) Representative Western blots probed for A1R, A2AR, and A2BR. (F–H) cAMP concentrations following incubation with (F) adenosine, (G) NECA, (H) CGS21680, and (I) CCPA for 15 minutes in the presence of RO20-1724 (10−4 M) (mean ± SEM of three experiments in quadruplicate).*p < .05, **p < .01, and ***p < .001 when compared with untreated cells; ###p < .001 when compared with MSCs stimulated with the same concentration of agonist.

Representative Western blots for A1R, A2AR, and A2BR protein expression during adipogenesis (Fig. 4E) showed that in parallel with changes in mRNA expression, A1R and A2AR protein expression also increased. There were no changes in A2BR protein expression (confirmed by combining densiometry values from three experiments) during adipogenesis, consistent with a lack of change in its mRNA.

Adenosine-mediated cAMP responses during adipogenesis

Figure 4FI shows the changes in cAMP when MSCs were grown with or without differentiation medium for 9 days. At 10−4 M, the fold increases in cAMP for MSCs and adipocytes were, respectively, 10- and 37-fold (adenosine; p < .001), 25- and 55-fold (NECA; p < .001), and 0- and 8-fold (CGS21680; p < .001). Although these data indicate an overall increase in A2R function in adipocytes compared with undifferentiated MSCs, the relative change in response to CGS21680 was particularly noticeable and is consistent with an increase in A2AR mRNA and A2AR protein on adipocyte differentiation. Furthermore, MRS1706 and SCH442416 significantly inhibited cAMP responses to NECA (19.4-fold with 10−5 M NECA reduced to 11.6-fold with 10−6 M MRS1706 and 18.1-fold with 10−5 M NECA reduced to 7.4-fold with 10−6 M SCH442416 (both p < .001).

Although there was a large increase in A1R mRNA during adipogenesis, this did not appear to be translated into protein, where only a small increase was observed. The A1R agonist CCPA failed to have an inhibitory action on forskolin-stimulated cAMP levels in either MSCs or adipocyte-differentiated MSCs; this suggests low expression of the A1R, particularly in the context of increased A2AR expression.

Effects of adenosine and NECA on lipid accumulation and adipocyte numbers

Treatment with NECA (10−4 M) increased, in a time-dependent manner, the size and number of cells that contained lipid (Fig. 5A). Following elution of the oil red O (a measure of total lipid content of the culture; Fig. 5B), the observed effects of NECA (10−4 M to 10−6 M) were dose-dependent, 10−6 M being the lowest concentration that caused a significant (p < .01) response. Adenosine (Supplemental Fig. S3) had less of an effect in comparison with NECA, and this probably reflects the short half-life, whereas NECA is regarded as being relatively stable. Treatment with adenosine (10−4 M) significantly (p < .01) increased adipogenesis from day 7, with the maximum effect on day 12. Nile red staining, however, showed that the number of adipocytes increased by up to 100% (p < .001) in the presence of NECA (10−4 M; results not shown).

Figure 5.

NECA stimulation of adipogenesis of MSCs. Lipid accumulation (A) visualized by staining with oil red O (×200) (NECA 10−4 M compared with control) and (B) quantified by measuring the absorbance of the dye retained by the cells at 490 nm (mean ± SEM of three experiments in triplicate). *p < .05, **p < .01, and ***p < .001 when compared with control at the same time point. (C) PPARγ, (D) C/EBPα, and (E) LPL mRNA expression following NECA treatment in adipogenic medium (mean ± SEM of three experiments in duplicate). *p < .05, **p < .01, and ***p < .001 when compared with day 0 (assigned a value of 1). #p < .05, ##p < .01, and ###p < .001 when compared with vehicle at the same time point.

Effect of NECA on the expression of adipocyte markers as MSCs differentiated to adipocytes

Adipogenic differentiation resulted in a gradual increase in the mRNA expression of PPARγ, C/EBPα, and LPL mRNA, which was further enhanced by the addition of NECA (Fig. 5CE). PPARγ mRNA expression was induced significantly from 3-, 7-, and 10- fold in untreated MSCs to 6- (p < .05), 21- (p < .001), and 31-fold (p < .01) in the presence of 10−4 M NECA for 2, 7, and 12 days, respectively. Treatment with 10−5 M NECA also tended to have a similar stimulatory effect, although statistical significance was not reached. NECA also induced similar but more pronounced changes in C/EBPα and LPL mRNA expression.

Effect of selective adenosine receptor agonists and antagonists on lipid accumulation

Since our experiments suggested that the differentiation of MSCs to adipocytes was associated with changes in expression and activation of A2AR and A1R, the effects of CCPA and CGS21680 on adipogenesis also were investigated. Both compounds stimulated adipogenesis in a time- and dose-related fashion (Fig. 6A, B). There were small but significant increases in adipogenesis in the presence of CCPA at 5, 7, and 9 days but only at the higher concentrations (10−5 and 10−6 M). Similar concentrations of CGS21680 had a much more pronounced effect on adipogenesis. The effects of CGS21680 were comparable to those observed when NECA was used (Fig. 5A, B). These results indicate an involvement of A1R and A2AR in the adipogenic differentiation of MSCs. To confirm the involvement of these two receptors, the selective antagonists PSB36 (A1R) and SCH442416 (A2AR) were incubated together with NECA (Fig. 5C). The concentrations of antagonist used were 10−6 M or less because higher concentrations caused a reduction in cell number. Addition of PSB36 or SCH442416 alone (10−6 and 10−7 M) had no effect on lipid accumulation. PSB36, however, had no effect on the stimulation of adipogenesis by NECA, suggesting that A1R may not have an important role in adenosine-mediated adipogenesis. On the other hand, SCH442416 (10−6 and 10−7 M) inhibited NECA stimulation of lipid accumulation by 80% (p < .001) and 75% (p < .01), respectively, after 12 days of treatment (Fig. 6C). This inhibitory action of SCH442416, however, was not seen at 7 days.

Figure 6.

A1R and A2AR receptor activation promotes the adipogenesis of MSCs. MSCs incubated with adipogenic medium in the presence of (A) CCPA and (B) CGS21680 and lipid accumulation visualized by light microscopy of oil red O staining (×200) and quantified by measuring the absorbance at 490 nm (mean ± SEM of three experiments in triplicate). (C) Lipid accumulation in MSCs treated with SCH442416 at 10−6 M and 10−7 M in the presence or absence of NECA (10−5 M) in adipogenic medium (mean ± SEM of four experiments in triplicate). *p < .05, **p < .01, and ***p < .001 when compared with vehicle at the same time point; ##p < .01 and ###p < .001 when compared with NECA-stimulated cultures at the same time point.

Effects of adenosine receptor overexpression on osteogenic- and adipogenic-associated genes in MSCs

Transient overexpression of A1R, A2AR, and A2BR in MSCs revealed significant changes in expression of osteoblasts and adipocyte markers (Fig. 7AD). Specifically, in relation to osteoblast markers, A2BR overexpression increased the expression of the early marker Runx2 (p < .001; Fig. 7A), whereas A1R overexpression decreased the expression of the intermediate marker ALP (Fig. 7B). This latter observation could indicate that the A1R influences differentiation of MSCs to adipocytes and thus away from the osteoblast lineage and therefore ties in with our other data. Overexpression of A1R or A2BR did not change the expression of adipocyte markers (ie, PPARγ and C/EBPα; Fig. 7C, D). However, overexpression of the A2AR significantly increased the expression of PPARγ (p < .01; Fig. 7C) and C/EBPα (p < .001; Fig. 7D).

Figure 7.

Effects of adenosine receptor overexpression on osteogenic- and adipogenic-associated genes in MSCs. MSCs were transfected with positive control (pmaxGFP) and PCDNA3 plasmids containing adenosine receptor genes (A1R, A2AR, and A2BR). Relative levels of (A) Runx2, (B) ALP, (C) PPARγ, and (D) C/EBPα mRNA compared with plasmid only–containing cells (mean ± SEM of three experiments in duplicate). *p < .05, **p < .01,and ***p < .001 when compared with values for control at the same time point.


There has been increasing interest recently in determining the factors that regulate the differentiation of MSCs into osteoblasts and adipocytes. Such information would enable the development of strategies for treating diseases associated with an imbalance in the differentiation and function of these lineages. Adenosine is known to regulate the physiologic function of many cells by binding to cell membrane receptors. Under basal conditions, adenosine concentrations in cells and tissue fluids are low (30 to 300 nM). When cells become stressed (eg, during hypoxic events, inflammation, and tissue injury), however, there is an accumulation of adenosine in the extracellular space25, 26 resulting in concentrations of 30 µM.25, 27 Adenosine receptors are known to be expressed by MSCs and are important for the function of these cells.14, 19 In addition, MSCs are capable of producing adenosine because they all express CD73.28 Recently, adenosine also has been shown to help the localization of human MSCs and promote their differentiation into hepatocyte-like cells,29 suggesting that adenosine also has a role in the differentiation of MSCs into several possible lineages. Thus we have explored how adenosine receptors are modulated as MSCs are induced to differentiate into osteoblasts and adipocytes and how these processes may be influenced by adenosine.

In undifferentiated MSCs, the A2BR was dominantly expressed. Furthermore, its activation (unpublished observations) or overexpression in undifferentiated cells was sufficient to promote the expression of osteoblast lineage markers. Interestingly, A2BR gene and protein expression appeared to be transiently upregulated as cells differentiated into osteoblasts, suggesting that this receptor may have a role in early osteoblastic differentiation. On the other hand, A2AR gene and protein expression appeared to increase during later stages of osteoblastic differentiation. From these data, one can speculate that A2BRs are more important for commitment and differentiation to osteoblasts and that A2ARs are important for osteoblast maturation and maintenance of the osteoblast phenotype. There also was a progressive increase in A2AR-mediated cAMP expression as MSCs were induced to differentiate to osteoblasts, which also may support a role for this receptor in later differentiation or function of osteoblasts. Activation (unpublished observations) or overexpression of A2AR in fact did not alter the expression of Runx2 or ALP in undifferentiated MSCs, whereas it stimulated osteogenesis during the differentiation. In addition, A2AR has been implicated in many activities by mature osteoblasts, which might explain the upregulation of the receptors following osteogenesis.30, 31 Nevertheless, in our study, A2BR was the dominant receptor throughout the differentiation process, and its activation led to a stimulation of osteoblastogenesis and subsequent mineralization by osteoblasts. Our observations are supported by a recent publication by Larsen and colleagues,32 who show that the A2BR is upregulated by 4.2-fold in a human bone marrow stromal cell population capable of in vivo heterotropic bone formation when compared with a second population that does not have the capacity to form such bone.

The A1R, in contrast to the A2BR, was found to be involved in adipogenesis and adipocyte function. A1R expression is reported to be low in preadipocytes but increases on adipogenesis and remains at high levels in mature adipocytes.20, 21, 33 Our data concur with these findings. We showed that expression of the A1R in rat MSCs was low and that it increased markedly on differentiation to adipocytes. A1R mRNA expression increased by more than 800-fold as MSCs were induced to differentiate to adipocytes. This was accompanied by increases (but to a much lesser extent) in protein expression. On the other hand, when MSCs were differentiating to osteoblasts, there was a less than fivefold increase in A1R mRNA and no change in protein expression. Despite the remarkable increase in expression during adipocyte differentiation, A1R activation by the selective agonist CCPA had no effect on expression of markers of adipogenesis. A similar result was observed when MSCs were transfected with the A1R gene. However, CCPA was found to elevate lipid accumulation by MSCs owing to either stimulation of lipogenesis or inhibition of lipolysis. Adenosine, acting through the A1R, has been implicated in many of the physiologic functions of adipocytes and has been associated in particular with an inhibition of lipolysis.34 Many A1R agonists have been shown to induce antilipolytic effects.33, 35, 36 Furthermore, overexpression of A1Rs and subsequent inhibition of lipolysis in adipocytes have been suggested as a potential cause of a lower and slower rate of weight loss in obese African-American women.37 We also have evidence that the A1R stimulates lipogenesis during the differentiation of a preosteoblast cell line into adipocytes (unpublished observations).

Adipocyte differentiation of MSCs also was unexpectedly accompanied by increases in A2AR expression at both the mRNA and protein levels. This was surprising because others have reported a loss of A2R expression and function during adipogenesis of a preadipocyte cell line (Ob1771)21 or failed to detect A2R expression in primary mature adipocytes.20 Increases in A2AR mRNA and protein expression on induced adipogenesis also was supported by increases in cAMP stimulation in response to adenosine, NECA, and the A2AR agonist CGS21680. The importance of the cAMP/PKA/CREB pathway in the induction of adipocyte differentiation and maturation has been known for many years. Indeed, the cAMP pathway has been reported previously, in murine preadipocyte 3T3-L1 cells and in human MSCs, to stimulate adipogenesis and expression of adipocytic genes.38, 39 Other relevant observations are that adipocyte differentiation in mouse embryonic fibroblasts from CREB−/− mice is impaired,40 and siRNA-mediated depletion of CREB and the closely related activating transcription factor 1 (ATF1) in 3T3-L1 cells results in loss of adipogenic differentiation.41

Activation of A2AR by CGS21680 stimulated adipogenesis of MSCs with similar potencies to NECA. The stimulatory affect of NECA could be blocked by SCH442416 (A2AR-selective antagonist), suggesting an A2AR-dependent action. We also investigated whether enhanced expression of A2AR is able to stimulate adipogenesis or osteoblastogenesis in the absence of differentiation medium. Our data show that enhanced expression of A2AR is sufficient to induce the expression of key adipogenic master switches such as PPARγ and C/EBPα that are essential to adipogenic differentiation.42 Upregulation of these genes subsequently leads to an increase in differentiation, lipid accumulation, and activities of adipocytes. Enhanced A1R and A2BR expression, on the other hand, did not effect the expression of PPARγ and C/EBPα. These findings suggest that the A2AR is mainly associated with the differentiation of MSCs into adipocytes, whereas the A1R could be involved in lipogenic activity of adipocytes rather than their differentiation.

In summary, data presented here show that adenosine can be important for both osteoblast and adipocyte differentiation. The A2BR is the dominant subtype in undifferentiated MSCs and during osteoblastogenesis. Osteoblast differentiation is associated with increases in the A2AR and A2BR expression, and these receptors stimulate osteoblast differentiation. The A2BR in particular is the major subtype in the differentiation process because it enhances the three stages of initiation, maturation, and mineralization in osteoblastogenesis of MSCs. On the other hand, the adenosine receptors involved in adipogenesis of MSCs are A2AR and A1R. Despite most of the receptor subtypes being expressed in most, but not all, tissues, targeting adenosine receptors in pathophysiology with truly specific compounds is still a distinct possibility. Many serious attempts are currently being undertaken to develop new therapies in this field—for example, several companies are working in late clinical trial with A2AR antagonists for neurodegeneration.13 Our work presents a better understanding of adenosine receptor expression and function in MSCs, and the changes in expression and effects of activation of A1R, A2AR, and A2BR observed make them promising targets for developing drugs to prevent or treat conditions with insufficient bone or excessive adipocyte formation. Even if these drugs could not be used systemically, their use locally at fracture sites, in osteoarthritic joints, or to prime MSCs at the time of replacement in relation to tissue engineering, for example, is a distinct possibility. Furthermore, it may be possible to target adenosine receptors indirectly by modulating downstream effectors such as ion channels43 or putative interacting proteins such as glutamate receptors.


All the authors state that they have no conflicts of interest.


We acknowledge the School of Medicine, Cardiff University, for funding and technical assistance from Dr MD Lewis and Mrs C Elford. BG and BAJE also thank the William Morgan Thomas Fund, Cardiff University for financial support.

Author's roles: BG was involved with conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of the manuscript. AAA was involved with collection of data and data analysis for the revised manuscript. JH was involved with conception and design, financial support, provision of study material or patients, data analysis and interpretation, manuscript writing, and final approval of the manuscript. BAJE was involved with conception and design, financial support, provision of study material or patients, data analysis and interpretation, manuscript writing, and final approval of the manuscript.