Role of the P2Y13 Receptor in the Differentiation of Bone Marrow Stromal Cells into Osteoblasts and Adipocytes

Authors

  • Galadrielle Biver,

    1. Institute of Interdisciplinary Research, Institut de Recherche Interdisciplinaire en Biologie Humaine et Moléculaire, Université Libre de Bruxelles, Gosselies, Belgium
    Search for more papers by this author
  • Ning Wang,

    1. The Mellanby Centre for Bone Research Department of Human Metabolism, The University of Sheffield, Sheffield, United Kingdom
    Search for more papers by this author
  • Alison Gartland,

    1. The Mellanby Centre for Bone Research Department of Human Metabolism, The University of Sheffield, Sheffield, United Kingdom
    Search for more papers by this author
  • Isabel Orriss,

    1. Department of Cell and Developmental Biology, University College London, London, United Kingdom
    Search for more papers by this author
  • Timothy R. Arnett,

    1. Department of Cell and Developmental Biology, University College London, London, United Kingdom
    Search for more papers by this author
  • Jean-Marie Boeynaems,

    1. Institute of Interdisciplinary Research, Institut de Recherche Interdisciplinaire en Biologie Humaine et Moléculaire, Université Libre de Bruxelles, Gosselies, Belgium
    2. Department of Medical Chemistry, Erasme Hospital, Brussels, Belgium
    Search for more papers by this author
  • Bernard Robaye

    Corresponding author
    1. Institute of Interdisciplinary Research, Institut de Recherche Interdisciplinaire en Biologie Humaine et Moléculaire, Université Libre de Bruxelles, Gosselies, Belgium
    • Ph.D., 10 Rue A. Bolland, Université Libre de, Bruxelles, Gosselies 6041, Belgium. E-mail: brobaye@ulb.ac.be Telephone: +3226509839; Fax: +3226509820

    Search for more papers by this author

  • Authors contributions: G.B.: conception and design, collection and assembly data, data analysis and interpretation, and manuscript writing; N.W.: collection and assembly data; A.G. and T.R.A.: data analysis and interpretation; I.O.: collection and assembly data and data analysis and interpretation; J.-M.B.: conception and design, data analysis and interpretation, manuscript writing, and final approval manuscript; B.R.: conception and design, collection and assembly data, data analysis and interpretation, manuscript writing, and final approval manuscript.

Abstract

Accumulating evidence indicates that extracellular nucleotides, signaling through purinergic receptors, play a significant role in bone remodeling. Mesenchymal stem cells (MSCs) express functional P2Y receptors whose expression level is regulated during osteoblast or adipocyte differentiation. P2Y13-deficient mice were previously shown to exhibit a decreased bone turnover associated with a reduction in the number of both osteoblasts and osteoclasts on the bone surfaces. We therefore examined whether P2Y13R activation was involved in the osteogenic differentiation of MSC. Our study demonstrated that ADP stimulation of P2Y13R+/+ (but not P2Y13R-/-) adherent bone marrow stromal cells (BMSCs) increased significantly the formation of alkaline phosphatase-colony-forming units (CFU-ALP) as well as the expression of osteoblastic markers (osterix, alkaline phosphatase, and collagen I) involved in the maturation of preosteoblasts into osteoblasts. The number of CFU-ALP obtained from P2Y13R-/- BMSC and the level of osteoblastic gene expression after osteogenic stimulation were strongly reduced compared to those obtained in wild-type cell cultures. In contrast, when P2Y13R-/- BMSCs were incubated in an adipogenic medium, the number of adipocytes generated and the level of adipogenic gene expression (PPARγ2 and Adipsin) were higher than those obtained in P2Y13R+/+ MSC. Interestingly, we observed a significant increase of the number of bone marrow adipocytes in tibia of P2Y13R-/- mice. In conclusion, our findings indicate that the P2Y13R plays an important role in the balance of osteoblast and adipocyte terminal differentiation of bone marrow progenitors. Therefore, the P2Y13 receptor can be considered as a new pharmacological target for the treatment of bone diseases like osteoporosis. STEM Cells 2013;31:2747–2758

Introduction

Bone marrow is a heterogeneous tissue composed of differentiated cells, lineage restricted progenitors, and stem cells. Among these are the multipotent mesenchymal stem cells (MSCs) that are able to differentiate into a number of distinct cell lineages including osteoblasts, adipocytes, chondrocytes, and myocytes [[1, 2]]. Osteoblasts and marrow adipocytes share a common mesenchymal precursor and their ability to differentiate along one of these two lineages is determined by growth factors, cytokines, and various signaling molecules. Factors promoting osteoblast differentiation are generally considered as inhibitory for adipocyte maturation and vice versa [[3-9]]. In age-related osteoporosis or osteopenia, the decreased bone volume is at least partly due to the fact that bone marrow stromal cells (BMSCs) have a reduced capacity to differentiate into osteoblasts but an increased capacity for adipocyte differentiation. This shift favoring the adipocyte over osteoblast lineage leads to progressive accumulation of fat cells in bones [[4, 10]]. Therefore, the correction of such disrupted balance is a potential therapeutic target for osteoporosis treatment. Over the last decade, a new strategy involving MSCs transplantation has been developed to cure bone formation deficiency. Because osteoblasts and adipocytes share a common precursor among MSC population, the main issue of this strategy is to avoid the differentiation of these MSCs into adipocyte cell lineage. Therefore, many studies focused on indentifying factors promoting osteogenesis and inhibiting adipogenesis in order to direct the differentiation toward osteogenic differentiation in vitro before transplantation [[11-13]].

The differentiation into both osteoblasts and adipocytes is typically described as a two-step process: determination and maturation. First, MSCs differentiate into preosteoblasts or preadipocytes and this requires the expression of the early transcription factors Runx2 in preosteoblasts [[14-16]] and CEBP/β and PPARγ1 in preadipocytes [[3, 17]]. An essential feature of these predifferentiated progenitors is that they exhibit a plasticity which allows a process of trans-differentiation into the other (reciprocal) lineage [18].

Upregulation of Runx2 expression is not sufficient for triggering osteoblast differentiation. Its transcription factor activity is regulated through the mitogen-activated protein kinase (MAP kinases) signaling pathway in cooperation with the small G protein Rho [[14, 16, 19, 20]]. Conversely, agents elevating intracellular cyclic AMP upregulate the expression and activity of the transcription factor, PPARγ [[17, 21]]. In mouse as well as in human, this nuclear receptor is present in two isoforms, PPARγ1 and PPARγ2, generated by alternative splicing. It is well-established that they are both involved in the differentiation of MSCs into adipocytes, but several studies support that PPARγ2 is expressed later and more selectively in fat cell lineage. Moreover, PPARγ2 is known to have a greater specificity to induce adipogenesis [[22-24]].

Growing evidence suggests that extracellular nucleotides play important roles in the regulation of bone metabolism [[25-31]]. Nucleotides are mainly intracellular but can be released through various mechanisms (i.e., cell lysis, exocytosis, and membrane channels) [32]. In the extracellular environment, they activate ionotropic P2X (PX1-7) and metabotropic G-protein-coupled P2Y receptors (P2Y1, 2, 4, 6, 11, 12, 13, 14) [33]. The P2Y13R is activated by ADP and is coupled to Gαi and inter alia adenylyl cyclase inhibition. Expression of P2Y13R mRNA has been detected in brain, liver, spleen, and bone marrow [[34-36]]. Several P2Y receptor subtypes are activated by ADP (P2Y1, P2Y12, and P2Y13). Since the availability of specific agonists and antagonists for these receptors is limited, the P2Y-deficient mice strains constitute a powerful tool for the identification of their physiological roles.

Knockout mice have been generated for the P2Y1, 2, 6 and 13 receptors and each strain displays a unique bone phenotype [[37-40]]. In particular, P2Y13-deficient (P2Y13R-/-) mice exhibit a decreased bone turnover associated with a reduction in the number of both osteoblasts and osteoclasts [39]. We have used P2Y13R-/- mice to investigate the functional expression of the P2Y13R in BMSCs and its role in the differentiation of these cells into osteoblasts versus adipocytes. This study provides evidence that the P2Y13R physiologically stimulates the differentiation of osteoblasts and inhibits that of adipocytes.

Materials and Methods

Animals

P2Y13R-/- mice that we previously generated were crossed with C57BL/6 mice for up to 10 generations on a C57BL/6 genetic background [34]. All the mice used in this study were generated using heterozygous parents and housed in a conventional animal facility. All the animal studies were authorized by the Animal Care Use and Review Committee of the Université Libre de Bruxelles (CEBEA-IBMM protocol 44).

Preparation and Culture of Mouse Bone Marrow Cells

Mice were killed by cervical dislocation. Mouse BMSCs were obtained from femur and tibia of 8-month-old female P2Y13 wild-type (WT) and knockout C57BL/6 mice. The two femurs and tibias were collected and bone marrow cavities were flushed with phosphate buffered saline (PBS) or bones were centrifuged as previously described [41]. Cells were cultured in αMEM medium (Invitrogen, Carlsbad, NM) supplemented with 10% fetal bovine serum (FBS), penicillin (50 units/ml), and streptomycin (50 μg/ml) at 37°C in a humidified atmosphere containing 5% CO2. Nonadherent cells were removed after 24 hours and 5 days by washing with PBS.

Flow Cytometry Analysis of Cell Surface Markers

Cells collected by trypsin treatment (2.5 mM in PBS/EDTA, 10 minutes, 37°C) were washed in fluorescence-activated cell sorting (FACS) medium (0.5% bovine serum albumin (BSA)/0, 1% NaN3 [wt/vol] in PBS). Cells (106) were incubated with antibody (2 μg/ml) in 100 μl of FACS medium. After washing, the cell labeling was analyzed using a Cytomics FC500 cytometer (Beckman Coulter). For each antibody, gating was determined based on appropriately stained controls. Anti-CD45, -Sca1, -PDGFRα, -CD90, and -CD44 antibodies were purchased from eBiosciences.

Osteogenic and Adipogenic Differentiation

For osteogenic differentiation, cells were cultured with osteogenic differentiation media (10 nM dexamethasone, 50 μg/ml ascorbic acid, and 5 mM β-glycerophosphate [Sigma-Aldrich, St. Louis, MO, USA]) for 7 days after the primary culture. For adipogenic differentiation, cells were cultured in the adipogenic differentiation media (0.5 mM 3-isobutyl-methylxanthine [IBMX], 10 μg/ml insulin, and 10 nM dexamethasone [Sigma-Aldrich, St- Louis, MO, USA]) for 14 days after the primary culture. The media were replaced every 3 or 4 days.

Alkaline Phosphatase Staining and Measurement of Alkaline Phosphatase-Colony-Forming Units and Colony Forming Unit-Fibroblast Colonies

Osteoblasts were detected using Naphthol Fast Red (Sigma-Aldrich, St-Louis, MO, USA) staining as described previously [42]. After 7 days in osteogenic differentiation media, cells were washed with PBS and incubated for 20 minutes in Naphthol AS-MX/Fast Red solution (0.5 mg/ml Naphthol, 1 mg/ml Fast Red in Tris 0.1 M pH 7.2) in order to detect the alkaline phosphatase (ALP)-positive colonies (colony-forming unit [CFU]-ALP). Cells were then counterstained with Mayer's hematoxylin in order to detect the total colonies formed (CFU-fibroblast [CFU-F]). CFU were detected using binocular microscope.

Alizarin Red Staining

Alizarin red staining was performed after 14 days culture in osteogenic medium. Cells were washed with PBS and fixed with 4% formaldehyde for 30 minutes at room temperature. The cell layers were washed with deionized water and allowed to air dry. The fixed cells were stained with 2% Alizarin red S pH 4.2 in deionized water (Sigma-Aldrich, St-Louis, MO, USA). After 20 minutes, cells were washed with deionized water and observed under light microscope in order to detect mineralized matrix nodules.

Isolation of RNA, Reverse Transcription, and Quantitative PCR

Total RNA was extracted using TriPure reagent (Roche, Basel, Switzerland) and quantified on a NanoDrop 1000 (Thermo Fisher Scientific, Waltham, MA, USA). After DNase I treatment (Fermentas, Hanover, Germany), RNA was reverse-transcribed using the RevertAid H Minus M-MuLV reverse transcriptase kit following manufacture instructions (Fermentas). PCR was performed using the QiagenTaq Core Kit (Qiagen, Venlo, The Netherlands). Quantitative RT-PCR was performed with the quantitative RT-PCR (qPCR) MasterMix for SYBR Green I (Biorad) on CXF86 Real-time system. For qPCR, housekeeping genes were hypoxanthine-guanine phosphoribosyltransferase, RPL32, and RPL12. Primers were purchased from EUROGENTEC and their sequences are shown in Table 1.

Table 1. List of specific primers used in this study
 GeneForward (5′–3′)Reverse (5′–3′)
  1. Abbreviations: ALP, alkaline phosphatase; COLI, collagen type I; HPRT, hypoxanthine-guanine phosphoribosyltransferase; OSX, osterix.
qPCRP2Y13CATATCCAGTGCCTGTGTGGTGTTGCATGTGCCTCTCTTC
 RUNX2CCCAGCCACCTTTACCTACATATGGAGTGCTGCTGGTCTG
 ALPCCAGCAGGTTTCTCTCTTGGCTGGGAGTCTCATCCTGAGC
 COLIGAGCGGAGAGTACTGGATCGGCTTCTTTTCCTTGGGGTTC
 OSXACTCATCCCTATGGCTCGTGGGTAGGGAGCTGGGTTAAGG
 PPARγ1TTTTCAAGGGTGCCAGTTTCAATCCTTGGCCCTCTGAGAT
 PPARγ2CTTCGCTGATGCACTGCCTACATCACGGAGAGGTCCACAG
 ADIPSINTGCACAGCTCCGTGTACTTCCACCTGCACAGAGTCGTCAT
 CD39L1AACCAGTCCATCTGAAGATCCAGATAATAGTAGAAAGCAGAAAAGGCTATGAAGTT
 RPL12ATGACATTGCCAAGGCTACCCAAGACCGGTGTCTCATCTGC
 RPL32GGCACCAGTCAGACCGATATCAGGATCTGGCCCTTGAAC
 HPRTAAGCTTGCTGGTGAAAAGGATTGCGCTCATCTTAGGCTTT
RT-PCRP2Y13ATTCGTGGGTTGAGCTAGTAAATCAGGGACCAGACGGAAAT
 HPRTCCTGCTGGATTACATTAAAGCACTGGTCAAGGGCATATCCAACAACAAAC

Oil Red O Staining

Lipid droplets accumulation was assessed by an oil red O staining. After 14 days in adipogenic medium, culture dishes were washed in PBS and cells were fixed in 4% formaldehyde for 1 hour. Cells were then washed with deionized water followed by 60% isopropanol. The cells were stained with 0.6% oil red O solution in isopropanol. After four washes with deionized water, oil red O dye was eluted with 100% isopropanol and quantified by measuring the optical absorbance at 490 nm.

Nile Red Staining

Adipocytes were detected using Nile Red staining (Sigma-Aldrich, St-Louis, MO, USA) as described previously [6]. Briefly, after 14 days in adipogenic differentiation media, cells were washed in PBS and fixed in 4% formaldehyde for 30 minutes. The cells were stained with 1 μg/ml Nile Red for 20 minutes at 4°C. Nile Red staining was detected using an inverted fluorescence microscope Zeiss Axio Observer Z1 with excitation at 485 nm and emission at 572 nm. The percentage of differentiated adipocytes was determined by counting cells stained by the lipid-specific dye Nile Red and normalizing them to the total cell number determined via nuclear 4,6-diamidino-2 phenylindole staining. An average value was determined by counting cells in at least eight microscopic fields (50–70 cells/field).

ATP and ADP Assay

Five days after primary culture, cells were incubated in Hanks’ balanced saline solution supplemented or not with differentiation medium for 15 minutes. The amount of ATP or ADP in cell cultures supernatants was determined using the ATP bioluminescence assay kit or the ADP colorimetric/fluorometric assay kit (Sigma-Aldrich, St-Louis, MO, USA). Data are expressed as nanomoles of ATP or ADP produced per milligram of cell protein (determined by Lowry assay). The pannexin-1 mimetic blocking peptide 10panx1 (WRQAAFVDSY) was supplied by R&D Systems (Minneapolis, MN, USA).

Histological Analysis

Female WT and P2Y13R-/- mice were euthanized at age of 20 weeks. Left tibiae of mice were dissected free of soft tissue, fixed in 10% buffered formalin, and embedded in paraffin wax. Three-micrometer sections were cut using a Leica Microsystems Microtome and followed by Hematoxylin & Eosin (H&E) staining [43]. The number of adipocytes (N.Adi) was counted in a 0.75 mm2 field, starting 0.25 mm from the growth plate and viewed on a DMRB microscope (Leica Microsystems), and normalized to the bone marrow area (Ma.Ar) measured.

Dual Energy X-Ray Absorptiometry Analysis of P2Y13R Knockout Mice

Percentage fat content of P2Y13R knockout mice and corresponding WT animals was measured by dual energy x-ray absorptiometry (DEXA) using a small animal scanner (Lunar PIXImus, GE Healthcare, Chalfont St. Giles, U.K.). Data were analyzed using the manufacturer's software (version 1.8).

Statistical Analysis

Nonparametric Mann-Whitney U tests were performed using Prism, Version 5.00, for Windows (Graph Pad, San Diego, CA, USA).

Results

Adenosine-5′-O-(2-thiodiphosphate) Stimulates MSC Differentiation into Osteoblasts Through the P2Y13R

To study the role of the P2Y13R in osteoblast differentiation, we used the classic model of adherent bone marrow cells as a source of stem cells [2]. We first observed that the P2Y13R mRNA is present in bone marrow cells as well as in 5 days cultured adherent cells (Fig. 1A). In order to determine whether ADP through P2Y13R activation could affect differentiation of MSC cells into osteoblasts, we stimulated both WT and P2Y13R-/- bone marrow adherent cells with adenosine-5′-O-(2-thiodiphosphate) (ADPβS) (a relatively stable analog of ADP) for 7 days and identified the presence of CFU-ALP. As shown in Figure 1B, we observed that after ADPβS treatment the number of WT CFU-ALP was more than doubled compared to untreated control condition. This ADPβS-induced increase of CFU-ALP was absent in P2Y13R-/- cultures. By quantitative RT-PCR, we observed that ADPβS stimulated osterix (OSX), ALP, and collagen type I (COLI) mRNA expression in WT-derived cells; this effect was significantly reduced in P2Y13R-/- cell cultures (Fig. 1C). In contrast, ADPβS had no effect on Runx2 mRNA expression. The Rho signaling pathway has been shown to be important in the process of MSC differentiation toward the osteoblastic lineage [[20, 44]]. As P2Y13R have been shown to activate this pathway [45], we tested the effect of the Rho inhibitor Y27632. In the presence of this Rho inhibitor, the stimulatory action of ADPβS on the expression of OSX, ALP, and COLI mRNA was reduced by 49%, 35%, and 36%, respectively (Fig. 1D). These data suggest that the activation of the P2Y13R supports the differentiation of MSCs into osteoblasts in a Rho-dependent way and that this action occurs downstream of Runx2 upregulation.

Figure 1.

Osteogenic differentiation of bone marrow stromal cells (BMSCs): effect of ADPβS. (A): Detection of P2Y13 messenger: RT-PCR (29 cycles) was performed using cDNA synthesized from RNAs extracted from freshly isolated BMSCs or cells cultured for 5 days in minimal essential media (MEM). HPRT gene was used to control cDNA quality. (B): CFU-ALP formation: 5 × 106 marrow cells collected by flushing method were seeded on six-well plates and cultured for 5 days in MEM medium followed by 7 days in MEM medium supplemented or not with 50 μM ADPβS (replaced every 2 days). To identify CFU-ALP, cells were stained with Naphthol-FastRed as described in Materials and Methods. Data are expressed as mean ± SD of six cell preparations; each point is the mean of three wells from each of these preparations. *, p < .05. (C): Runx2, OSX, ALP, and COL I expression: 7.5 × 106 cells collected by centrifugation were seeded on Petri dishes and cultured for 5 days in MEM medium followed by 1, 3, or 5 days in MEM supplemented or not with ADPβS (50 μM). RNAs are extracted 1, 3, or 5 days after ADPβS additions. Total RNAs were extracted as described in Materials and Methods and mRNA expression was determined by RT-qPCR. Quantitative mRNA values were normalized by the amount of HPRT, RPL13, and RPL32, and results represent the ratio between ADPβS-stimulated and control cultures. Data are means ± SD of six cell preparations. Data were normalized to those of the housekeeping gene HPRT, RPL32, and RPL12. *, p < .05; **, p < .01. (D): Effect of Rho inhibitor Y27632: 7.5 × 106 cells collected by centrifugation were seeded on Petri dishes and were cultured for 5 days in MEM medium followed by 1 or 3 days in MEM supplemented with ADPβS (50 μM) with or without Y7632 (10 μM). RNA extraction and mRNA quantification were performed as described above. Data are means ± SD of five (OSX and RUNX2), six (ALP), or four (COLI) cell preparations. Data were normalized to those of the housekeeping gene HPRT, RPL32, and RPL12. *, p < .05; **, p < .01. Abbreviations: CFU-ALP, alkaline phosphatase-colony-forming units; COLI, collagen type I; NS, nonstimulated; HPRT, hypoxanthine-guanine phosphoribosyltransferase; OSX, osterix; WT, wild type.

P2Y13R Gene Deletion Inhibits Osteoblastic Differentiation in an Osteogenic Medium

To determine whether these effects of P2Y13R activation have a physiological relevance, we compared the generation of CFU-ALP induced by an osteogenic medium containing dexamethasone-βglycerophosphate-ascorbic acid (DGAA) in P2Y13R-/- and WT MSCs cultured in vitro. We first observed that P2Y13R mRNA expression was stable during osteoblast differentiation (Fig. 2A). As expected, the average number of CFU-ALP detected was higher when WT cells were treated with osteogenic inducers (mean of 9.5 CFU-ALP per dish) than with ADPβS treatment (mean of 1.8 CFU-ALP per dish) (Figs. 1A, 2B). The number of CFU-ALP formed in culture dishes of P2Y13R-/- bone marrow cells was significantly lower than that counted in WT cultures. To confirm that the targeting of the P2Y13 gene affected the differentiation of MSCs into osteoblasts, we studied another specific characteristic of these bone forming cells, that is, the formation of mineralized matrix nodules. As expected, the number of Alizarin red-S stained nodules formed in culture dishes of P2Y13R-/- culture dishes was also significantly lower than that counted in WT cultures (Fig. 2C). Furthermore, the DGAA-induced upregulation of OSX, ALP, and COLI mRNA was reduced in cell cultures from P2Y13R-/- mice, while Runx2 expression was unchanged (Fig. 2D).

Figure 2.

Osteogenic differentiation of bone marrow stromal cells (BMSCs): comparison between P2Y13R-/- and WT mice. (A): P2Y13R expression: quantitative RT-PCR was performed using cDNA synthesized from RNAs extracted from cells cultured for 5 days in minimal essential media (MEM) and cells cultured additionally for 1, 3, 5, and 7 days in MEM supplemented with 50 μg/ml ascorbic acid, 5 mM β-glycerophosphate, and 10 nM dexamethasone. Data are means ± SD of four cell preparations. Data were normalized to those of the housekeeping gene hypoxanthine-guanine phosphoribosyltransferase (HPRT), RPL32, and RPL12. (B): CFU-ALP formation: 5 × 106 marrow cells collected by flushing method were seeded on six-well plates and cultured for 5 days in MEM followed by 7 days in MEM supplemented with 50 μg/ml ascorbic acid, 5 mM β-glycerophosphate, and 10 nM dexamethasone. Top panel: data are expressed as mean of CFU-ALP ± SD; nWT = 14, nP2Y13R-/- = 16, ***, p < .001. Bottom panel: picture of CFU-ALP in culture dishes. (C): Mineralized nodule formation: 5 × 106 marrow cells collected by flushing method were seeded on six-well plates and cultured for 5 days in MEM followed by 14 days in MEM supplemented with 50 μg/ml ascorbic acid, 5 mM β-glycerophosphate, and 10 nM dexamethasone. Top panel: data are expressed as mean ± SD; nWT = 11, nP2Y13R-/- = 10, ***, p < .001. Bottom panel: picture of mineralized nodules in culture dishes. (D): RUNX2, OSX, ALP, and COLI expression: quantitative RT-PCR for RUNX2, OSX, ALP, and OSX was performed as in Figure 1C, but cells were treated for 5 days in MEM, followed by 1 (RUNX2 and OSX), 3 (ALP) or 5 (COLI) days in MEM supplemented with 50 μg/ml ascorbic acid, 5 mM β glycerophosphate, and 10 nM dexamethasone. Data are means ± SD of six cell preparations. Data were normalized to those of the housekeeping gene HPRT, RPL32, and RPL12. *, p < .05; **, p < .01. (E): CFU formation: BMSCs were seeded and treated as described in Figure 2A. Total CFU (left) were detected using hematoxylin staining and the percentage of CFU-ALP among total CFU (right) was quantified. Data are expressed as mean of CFU ± SD, nWT = 14, nP2Y13R-/- = 16, ***, p < .001. Abbreviations: CFU-ALP, alkaline phosphatase-colony-forming units; COLI, collagen type I; DGAA, dexamethasone-βglycerophosphate-ascorbic acid; NS, nonstimulated; OSX, osterix; WT, wild type.

This altered differentiation of MSCs toward the osteoblastic lineage could be due to an alteration in the differentiation capability of progenitor cells or to a decreased number of these progenitors. Both mechanisms seem to be involved since we observed that the total number of CFU (detected by a hematoxylin coloration) counted in P2Y13R-/- cultures was reduced by 44%, and the percentage of CFU-ALP among total CFU was also diminished by 54% in cultures of P2Y13R-/- MSCs (Fig. 2E).

Among bone marrow cells, MSCs are very rare [44]. The experimental protocol we followed allows isolating them by adherence. However, the population we got remained heterogeneous as shown by flow cytometry. On a Forward and Side light scatter (FFC/SSC) dot plot (data not shown), we identified a low size/low granularity population (population A) and a medium size/medium granularity one (population B); the former is mainly CD45 (95%) while the latter is for the most part (75% of cells) CD45+. These hematopoietic cells are predominantly of the myeloid lineage as they were mainly CD11b+ (data not shown). To identify MSCs, we studied the expression of Sca-1 (stem cell antigen 1), CD90, CD44, and PDGFRα (platelet-derived growth factor receptor α) among CD45 cells [45]. As expected, we detected Sca-1+/CD90+, Sca-1+/CD44+, and Sca-1+/ PDGFRα+ cells but in a very low proportion (Table 2). Among CD45/Sca-1+ cells of both A and B subpopulations, we could not observe significant differences between WT and P2Y13R-/- cell cultures in terms of CD90 or CD44 expression. Interestingly, among Sca-1+ cells, we detected a significantly reduced proportion of PDGFRα+ cells in P2Y13R-/- cell cultures in the A population compared to WT cells.

Table 2. Immunophenotyping of adherent bone marrow cells
 WTP2Y13R-/- 
  1. Abbreviations: N.S., not significant; PDGFRα, platelet-derived growth factor receptor α; Sca-1, stem cell antigen 1; WT, wild type.
Population A
CD45+4.76 ± 1.724.00 ± 2.66N.S.
CD45-95.57 ± 1.4693.45 ± 4.61N.S.
CD45-Sca-1+CD90+0.08 ± 0.050.07 ± 0.07N.S.
CD45-Sca-1+CD44+0.14 ± 0.050.16 ± 0.07N.S.
CD45-Sca-1+PDGFRα+0.18 ± 0.050.10 ± 0.05p < .05
Population B   
CD45+74.8 ± 7.882.16 ± 10.43N.S.
CD45-24.5 ± 2.8119.23 ± 3.66N.S.
CD45-Sca-1+CD90+0.7 ± 0.160.63 ± 0.27N.S.
CD45-Sca-1+CD44+3.97 ± 1.384.22 ± 1.68N.S.
CD45-Sca-1+PDGFRα+3.07 ± 2.411.77 ± 1.11N.S.

ATP Release and Metabolism

These data suggested that maximal osteoblast differentiation requires P2Y13R activation and implied that ADP may be released in the extracellular space constitutively or in a regulated way. Therefore, we measured the concentration of ATP, the precursor of ADP, in the culture medium and observed that it was increased following addition of the osteogenic inducers and that this stimulatory effect was affected by a 10Panx inhibitor of Pannexin channel (Fig. 3A). Interestingly, we also noticed that the DGAA treatment induced upregulation of NTPDase2 (also known as CD39L1) mRNA (Fig. 3B). This ectonucleotidase preferentially hydrolyzes ATP into ADP, thereby leading to the accumulation of extracellular ADP [46]. Indeed, osteogenic inducers stimulated the accumulation of ADP in the culture supernatant (Fig. 3C).

Figure 3.

ATP release and metabolism. (A): ATP release: cells were cultured for 5 days in MEM medium and 15 minutes in minimal essential media (MEM) supplemented with 50 μg/ml ascorbic acid, 5 mM β glycerophosphate, and 10 nM dexamethasone in the presence or not of Pannexin 1 inhibitor ,10PANX (100 μM). The concentration of ATP in the supernatant was determined by luminescence using an ATP assay kit and normalized to protein content as described in Materials and Methods. Data are means ± SD of four cell preparations. *, p < .05. (B): CD39L1 expression: quantitative RT-PCR for CD39L1 was performed as in Figure 2D. Data are means ± SD of four cell preparations. Data were normalized to those of the housekeeping genes hypoxanthine-guanine phosphoribosyltransferase, RPL32, and RPL12. *, p < .05. (C): ADP release: cells were cultured for 5 days in MEM medium and 15 minutes in Hanks’ balanced saline solution supplemented with 50 μg/ml ascorbic acid, 5 mM β glycerophosphate, and 10 nM dexamethasone. The concentration of ADP in the supernatant was determined using the ADP Colorimetric/Fluorometric Assay Kit and normalized to protein content as described in Materials and Methods. Data are means ± SD of five cell preparations. *, p < .05. Abbreviations: DGAA, dexamethasone-βglycerophosphate-ascorbic acid; NS, nonstimulated.

P2Y13 Gene Targeting Increases Adipogenic Differentiation

As mentioned above, the osteoblast and adipocyte differentiations are linked and mutually exclusive. Therefore, we wanted to ascertain the effect of P2Y13R gene deletion on adipocyte differentiation of bone marrow adherent cells. The expression of the P2Y13 messenger was stable during adipocyte differentiation induced by the dexamethasone-insulin-IBMX (DII) cocktail (Fig. 4A). Culture of P2Y13R+/+ MSCs in a medium containing the adipogenic inducers induced the accumulation of cells with lipid droplets that can be evaluated by oil red O and Nile Red staining. As shown in Figure 4B, the amount of oil red O dye eluted from P2Y13R-/--treated cells was higher than that eluted from WT-treated cells. Moreover, the number of Nile Red-positive cells was significantly increased in P2Y13R-/- MSC cultures (Fig. 4C) while we did not find any difference in cell or droplet sizes (data not shown). Furthermore, the percentage of MSCs that spontaneously differentiated into adipocytes was significantly increased in P2Y13R-/- cell cultures (Fig. 4D). These data show that the deletion of the P2Y13R favors the formation of new adipocytes from precursor cells but do not affect the fat cell size.

Figure 4.

Adipogenic differentiation of bone marrow stromal cells: comparison between P2Y13R-/- and WT mice. (A): P2Y13R expression: quantitative RT-PCR was performed using cDNA synthesized from RNAs extracted from cells cultured for 5 days in minimal essential media (MEM) and cells cultured additionally for 1, 3, 5, and 7 days in MEM supplemented with 10 nM dexamethasone, 0.5 mM IBMX, and 10 μg/ml insulin. Data are means ± SD of four cell preparations. Data were normalized to those of the housekeeping gene hypoxanthine-guanine phosphoribosyltransferase (HPRT), RPL32, and RPL12. (B): Lipid droplet accumulation: 2.5 × 106 marrow cells collected by the flushing method were seeded on six-well plates and cultured for 5 days in MEM medium followed by 14 days incubation in MEM medium supplemented with 10 nM dexamethasone, 0.5 mM IBMX, and 10 μg/ml insulin. The lipid droplets accumulation was assessed by oil red O assay as described in Materials and Methods. Data are means ± SD (nWT = 12 and nKO = 9). *, p < .05. (C): Adipocyte formation: 5 × 105 marrow cells collected by the flushing method were seeded on Cellview dishes and cultured for 5 days in MEM medium followed by 14 days incubation in MEM medium supplemented with 10 nM dexamethasone, 0.5 mM IBMX, and 10 μg/ml insulin. The lipid droplets of adipocytes were identified by Nile Red staining as described in Materials and Methods. Left panel: quantification of Nile Red positive cells; data are expressed as mean ± SD, (nWT = 6 and nP2Y13R-/- = 6), **, p < .01. Right panel: fluorescence microphotographs taken at ×20 of Nile Red staining. (D): Adipocyte formation: 5 × 105 marrow cells collected by the flushing method were seeded on Cellview dishes and cultured for 21 days in MEM medium. The lipid droplets of adipocytes were identified by Nile Red staining as described in Materials and Methods. Data are expressed as mean ± SD (nWT = 5 and nP2Y13R-/- = 5), **, p < .01. (E): CEBP/β, PPARγ1, PPARγ2, and adipsin expression: 7.5 × 106 cells, collected by centrifugation, were seeded on Petri dishes and cultured for 5 days in MEM medium followed by 1 (CEBPβ), 3 (PPARγ1 and PPARγ2), 7 (PPARγ2), or 10 (ADIPSIN) days in MEM supplemented with 10 nM dexamethasone, 0.5 mM IBMX, and 10 μg/ml insulin. Total RNA was extracted as described in Materials and Methods and mRNA expression was determined by RT-qPCR. Quantitative mRNA values were normalized to the amount of HPRT, RPL13, and 32, and results are given as the ratio between adipogenic and control medium. Data are means ± SD of six cell preparations. Data were normalized to those of the housekeeping gene HPRT, RPL32, and RPL12. *, p < .05; **, p < .01. (F): ATP release: cells were cultured for 5 days in MEM medium and 15 minutes in MEM supplemented with 10 nM dexamethasone, 0.5 mM IBMX, and 10 μg/ml insulin. The concentration of ATP in the supernatant was determined by luminescence using ATP assay kit and normalized to protein content as described in Materials and Methods. Data are means ± SD of four cell preparations. *, p < .05. (G): ADP release: cells were cultured for 5 days in MEM medium and 15 minutes in Hanks’ balanced saline solution supplemented with 10 nM dexamethasone, 0.5 mM IBMX, and 10 μg/ml insulin. The concentration of ADP in the supernatant was determined by ADP Colorimetric/Fluorometric Assay Kit and normalized to protein content as described in Materials and Methods. Data are means ± SD of four cell preparations. *, p < .05. Abbreviations: NS, nonstimulated; WT, wild type.

While the mRNA expressions of CEBP/β and PPARγ1 were not modified in P2Y13R-/- cells, that of PPARγ2 was increased at an early time (Fig. 4E). Indeed, in WT cultures the stimulation of PPARγ2 mRNA expression was detectable after 7 days of DII treatment, but in P2Y13R-/- cultures it was already observed at day 3. Moreover, the expression of adipsin (a marker of adipocyte differentiation [[10, 47]]) was higher in P2Y13R-/- cells compared to WT cells (Fig. 4E). It is interesting to notice that adipogenic differentiation stimulates accumulation of ATP and ADP (Fig. 4F, 4G). These data suggest that the deletion of the P2Y13R favors the differentiation of MSCs into adipocytes by accelerating the expression of PPARγ2.

P2Y13R Deletion Increases the Accumulation of Adipocytes In Vivo

To see whether the increased capability of P2Y13R-/- MSCs to differentiate into adipocytes observed in vitro translated to the in vivo situation, we performed histological analysis of the long bone cellular content. As illustrated in Figure 55A--5C, the number of adipocytes in tibial sections of 20-week-old female mice was more than doubled in P2Y13R-/- mice compared to WT. There was however no difference in the percentage of total body fat measured by DEXA between WT and P2Y13R-/- mice (Fig. 5D).

Figure 5.

Fat content in bone marrow and total body : comparison between WT and P2Y13R-/- mice. (A): Transparent round adipocytes can be identified in the tibial proximal end bone marrow, scale bar = 0.5 mm. (B): Adipocytes were marked as “a” in magnified view of bone marrow area, scale bar = 50 μm. (C): The number of adipocytes (N.Adi) normalized by area of bone marrow (Ma.Ar) was quantified for mice at 20 weeks of age, nWT = 5, nP2Y13R-/- = 6. Data are mean ± SEM, **, p < .01. (D): Dual energy x-ray absorptiometry analysis of P2Y13R-/- mice and their corresponding WT: results are expressed in % fat content (nWT = 7 and nP2Y13R-/- = 10). Abbreviations: NS, nonstimulated; WT, wild type.

Discussion

Growing evidence suggests a role for extracellular nucleotides and P2Y receptors in the differentiation of MSCs [[25, 30, 48]]. Zippel et al. have recently shown that the mRNAs of each P2 receptor subtype are detectable in human MSC derived from adipose tissue or dental follicles. Moreover, the expression of P2Y1, P2Y2, P2Y4, and P2Y14 mRNAs is reduced after osteogenic differentiation while the expression of the P2Y4 and P2Y14 receptors is downregulated after adipogenic differentiation and that of the P2Y11 receptor is upregulated [48]. It was previously shown that P2Y13R-/- mice exhibit a decreased bone turnover in vivo and that this phenotype is associated with a reduction in the number of osteoblasts at the surface of bone [39]. These data suggested that the P2Y13 receptor might be involved in the control of osteoblast and adipocyte differentiation from stem cells. Therefore, we have investigated the role of the P2Y13R in the generation of osteoblasts from adherent MSCs isolated from bone marrow. We show that ADP triggers the generation of CFU-ALP and stimulates the expression of ALP, OSX, and COLI mRNA, three osteoblast markers. These effects involve the activation of the P2Y13R since the stimulatory effect of ADP is reduced in P2Y13R-/- MSCs cell culture. To determine at which level of the differentiation process the stimulation of the P2Y13R favors the osteoblastic lineage, we tested the mRNA expression of the Runx2 transcription factor, which is one of the earliest determinants of osteoblast differentiation driving the differentiation of MSCs into preosteoblasts [14]. Its pro-osteogenic activity implies not only the upregulation of its expression but also post-translational modifications like phosphorylation, which depend on the Rho and MAP kinase signaling pathways [[16, 19]]. ADP does not modulate Runx2 expression in these cultures, but increases the mRNA level of OSX, another important transcription factor involved in osteoblast differentiation that is under the control of Runx2. ATP, the precursor of ADP, has been shown to stimulate Runx2 DNA-binding activity in osteoblast-like HOBIT cell line in an Erk1, two-dependent way [49]. Moreover, the RhoA/ROCK1 signaling pathway is linked to P2Y13R activation in murine osteoblasts isolated from calvaria and ADP stimulation of human hepatocytes leads to RhoA/ROCK1 activation [[39, 45]]. In agreement with these observations, we show here that an inhibitor of the Rho G protein (the compound Y27632) reduced ADP-induced upregulation of OSX, ALP, and COLI mRNA expression in MSCs. Therefore, we suggest that ADP activates the differentiation of preosteoblasts into osteoblasts by stimulating the DNA binding activity of Runx2 through RhoA/ROCK1 signaling pathways in a P2Y13R-dependent way.

The capability of the P2Y13R to modulate the differentiation of MSCs into the osteoblastic lineage is relevant at the physiological level, as the number of the CFU-ALP and mineralized nodules induced by the osteogenic cocktail DGAA is largely reduced in P2Y13R-/- cell cultures. This is due both to the reduction of total CFU obtained after culture of BMSCs and to their decreased ability to differentiate into osteoblasts. The reduction of total CFU could be correlated to the decreased number of CD45/Sca-1+/PDGFRα+ seen in a subpopulation of P2Y13R-/- adherent cells. Indeed, it has been recently shown that CD45/Sca-1+/PDGFRα+ cells isolated from the bone marrow have the highest capability to generate CFU-Fs after plating [45]. Human MSCs isolated from the bone marrow or adipose tissue express the P2Y13R [[25, 48]]. However, because the cell population we stimulated in this work was heterogeneous, we cannot exclude that the effect of the P2Y13R on osteoblast differentiation is partly indirect involving other cells that stem cells. Among cells we obtained after 5 days of preculture, we identified CD45+/CD11b+ myeloid cells. In this context, it is interesting to refer to the work of Pirraco et al. showing that the coculture of human macrophages with human bone marrow-derived MSCs leads to increase differentiation into osteoblasts compared to monoculture of MSCs [50].

We confirm that the influence of the P2Y13R on osteogenesis is downstream of the Runx2 upregulation, as its expression is not altered by the P2Y13R gene targeting. But the expression of OSX is significantly reduced in mutated MSCs. This suggests that the P2Y13R does not control the initiation of MSCs differentiation, but the maturation of preosteoblasts into osteoblasts.

The fact that the differentiation of MSCs into osteoblasts requires P2Y13R expression implies that the osteogenic cocktail may favor accumulation of ADP in the extracellular space. This is supported by the observation that the incubation of MSCs with apyrase, that hydrolyzes extracellular nucleotides, inhibits osteoblast differentiation [48]. We observed that in osteogenic conditions the concentration of extracellular ATP, as well as of ADP, is enhanced compared to control conditions. In human BMSCs, it has been shown that ectonucleotidases are expressed and that the supplementation of culture medium with ATP leads to transient production of ADP [30]. We also detected the expression of the ectonucleotidases NTPDase2 (CD39L1) which preferentially hydrolyzes ATP into ADP [46]. Therefore, we conclude that the DGAA-induced maturation of preosteoblasts into osteoblasts involves partially the release of ATP and its conversion through CD39L1 into ADP that activates the P2Y13R. The release of nucleotides involves multiple mechanisms like cell lysis, secretory granules exocytosis, membrane transporter, and hemi-channels like connexin and pannexin [[32, 51]]. DAAG-induced ATP release seems to result from the opening of pannexin hemichannels as DGAA-induced ATP accumulation is reduced by 10Panx, an inhibitory peptide of Pannexin 1.

The impaired ability of P2Y13R-/- MSCs to differentiate into osteoblasts in vitro may explain the decreased number of osteoblasts present in P2Y13R-/-bones in vivo, leading to reduced trabecular bone volume [39]. It is important to mention that the ADP effects that we observed were not completely abolished in P2Y13R-/- cell cultures suggesting implication of the other purinergic receptors as P2Y1 or P2Y12 or adenosine receptors. Several studies support that the P2Y1 receptor promotes proliferation of undifferentiated cells and osteoblasts, but there are no indication to our knowledge regarding to a role of this receptor in MSCs differentiation [[31, 52]]. On the contrary, the expression of the P2Y1 mRNA is downregulated during osteogenic differentiation of human adipose-tissue-derived stem cells [48]. Recently, it has been shown that the expression of the P2Y12 receptor mRNA is upregulated in a model of rat osteoblast differentiation of precursors isolated from calvarial bones [53]. Finally, adenosine that can accumulate into extracellular space after ADP hydrolysis is well known to promote osteoblast differentiation [[54, 55]].

The differentiation of MSCs into osteoblastic or adipogenic lineages are not independent processes: molecular components promoting one cell fate inhibit the mechanisms governing the differentiation of the alternative lineage [[4, 7]]. This is also the case for the P2Y13R. Its expression is required for full osteoblastic differentiation as discussed above, while the P2Y13R gene targeting increases the potential of MSCs to generate adipocytes. Indeed, the number of adipocytes generated in P2Y13R-/- MSCs in response to the adipogenic cocktail and the expression of adipocyte markers like adipsin are enhanced compared to that obtained in WT culture. This suggests that the activation of the P2Y13R inhibits adipocyte differentiation and implies that the adipogenic stimulation stimulates the accumulation of ADP in the extracellular space, what we also observe. CEBP/β and PPARγ1 are transcription factors that regulate the first steps of adipocyte differentiation [[22, 56]]. The P2Y13R gene targeting does not modulate their mRNA expression. However, the earlier expression of PPARγ2 observed in P2Y13R-/- cultures correlates with increased adipocyte formation and adipsin expression, suggesting that the P2Y13R controls negatively the maturation of preadipocytes into adipocytes. It could be linked to the control of the transcriptional expression of PPARγ2 by cAMP-linked signaling pathway [21]. Indeed the P2Y13R belongs to the G protein-coupled receptor superfamily and activates Gαi/o proteins inhibiting cAMP production [36]. The gene deletion of Gαi/o-coupled type 1 cannabinoid receptor CB1 leads to an analogous phenotype in terms of MSCs differentiation: reduced production of osteoblasts but increased generation of adipocytes [3]. Moreover, it has been recently described that an impairment of Rho signaling upregulates PPARγ2 expression as well as adipocyte formation [[44, 57]]. Therefore, a cooperation between cAMP inhibition and Rho signaling pathways may be the effector mechanism downstream the P2Y13R stimulation to control negatively the differentiation of adipocytes. The fact that we observed a significant increase of adipocytes in noninduced P2Y13R-/- MSCs culture as well as in the bone marrow of P2Y13R-/- mice, but not in the total body fat, underscores the relevance of our observations in relation with bone diseases. Osteoporosis, for example, is a disease in which resorption by osteoclasts is not sufficiently compensated by bone formation [58] and is associated with increased bone marrow adipogenesis [[59, 60]]. The identification of molecular components able to favor osteogenesis and reduce adipogenesis will help to design new drugs for bone diseases cure.

Conclusion

In conclusion, this study demonstrates that the P2Y13R is a physiological determinant of MSCs differentiation. This receptor is one of the factors controlling the balance between osteogenesis and adipogenesis, at the level of the maturation of precommitted progenitor cells. Our in vitro observations are consistent with our in vivo data showing a high marrow fat content in the bone of the P2Y13 knockout mice and the previously reported decreased bone formation in these mice [40]. Agonists of the P2Y13R might be used to favor osteoblast differentiation and bone formation, in particular in case of MSC transplantation.

Acknowledgments

We thank Dominique Fokan for excellent help and technical assistance. This work was supported by Actions de Recherche Concertés de la Communauté Française de Belgique, by the Belgian Program on Interuniversity Poles of Attraction Initiated by the Belgian State, Prime Minister's Office, Federal Service for Science, Technology and Culture, by the Fonds National de la Recherche Scientifique Médicale, the Fonds Emile DEFAY, and the European Commission under the seventh Framework Programme (proposal #202231) performed as a collaborative project among the members of the ATPBone Consortium (Copenhagen University, University College London, University of Maastricht, University of Ferrara, University of Liverpool, University of Sheffield, and Université Libre de Bruxelles). G.B. was supported by the Fonds National de la Recherche Scientifique (FRIA, Belgium).

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.

Ancillary