Introduction

L-Glutamate (Glu) is accepted as an excitatory amino acid neurotransmitter in the mammalian central nervous system (CNS). Extracellular levels of glutamate are determined by the glutamate transporter1, 2. The diverse actions of L-glutamate in the CNS result from the existence of multiple glutamate receptors (GluR). These have been divided into two classes, metabotropic (mGluR) and ionotropic (iGluR), according to their molecular heterologies and differing intracellular signal transduction mechanisms. The iGluRs are further classified into N-methyl-D-aspartate (NMDA), DL-α-amino-3-hydroxy-5-methylisoxasole-4-propionate (AMPA), and kainite(KA) receptors. NMDA receptors are glutamate-gated ion channels characterized by a very high Ca2+ conductance3.

In mammalian bone, NMDA receptors are expressed in osteoblasts and osteoclasts as revealed by RT-PCR, in situ hybridization, immunohistochemistry, and electrophy-siology4, 5. Bone cells and neurons possess similar, and in some cases identical glutamate signaling machinery and receptors5.

Bone loss is associated with a reduction in nerve endings that immunostain for glutamate. Glutamate-containing neuronal endings have been described in a dense and intimate network in bone tissue6. Chenu et al reported that bone loss was induced in ovarectomized (OVX) rats with a reduction in neuronal density and concluded there was a functional link between the nervous system and bone loss after ovariectomy7. Hinoi et al reported that administration of glutamate significantly prevented the decreased bone mineral density in both the femur and the tibia in ovariectomized mice in vivo8. All these findings suggest that the neurotransmitter glutamate may play a role in bone remodeling.

NMDA has been shown to promote the proliferation and differentiation of hippocampal neural progenitor cells (NPCs) in vitro by activating NMDA receptors9. NMDA is mitogenic for MC3T3-E1 osteoblastic cells and glutamate has been reported to promote the viability of primary human osteoblasts in vitro10. Blockade of NMDA receptors in rat primary osteoblasts inhibits expression of markers of bone formation in vitro4, 11. Previously, we demonstrated that activation of NMDA receptors promoted rat primary osteoblast differentiation and that one of the possible mechanism was ERK1/2 activation12.

In the central nervous system (CNS), activation of the Ca2+-permeable NMDA receptors results in an increase in Ca2+ influx. The Ca2+ signals then activate several Ca2+-dependent kinases13. Kinases which have been suggested to be important include protein kinases C and A (PKC and PKA), and PI3 kinase (PI3K). PKC is localized at excitatory synapses containing NMDA receptors and might be involved in NMDA-evoked ERK1/2 phosphorylation14. In hippocampal neurons NMDA played a neuroprotective role through the PKA signaling pathway15, and in striatal and cortical neurons, PI3K is critically involved in NMDA receptor activation16, 17, 18. Thus, the above signaling pathways may positively respond to signals from ionotropic types of glutamate receptors in the CNS.

Locally elevated extracellular Ca2+ levels have been suggested to play a role in regulating bone remodeling. Signaling pathways such as phospholipase C (PLC), PKC, mitogen-activated protein kinase (MAPK), and PI3K have been implicated in the modification of cellular function by Ca2+19, 20. Since these signaling pathways are involved in NMDA receptor regulation in the CNS and were also suggested to be important by our initial microarray analysis, we hypothesized that they are also involved in NMDA-induced bone remodeling. The aim of this study was to investigate the mechanism of the effects of NMDA and MK801 (the noncompetitive antagonist of NMDA receptors) on differentiation of rat primary osteoblasts.

Materials and methods

Primary cultures of osteoblasts

Osteoblasts were prepared from calvaria of 1-day-old Sprague-Dawley rats by a sequential enzymatic digestion method as described previously21. The bones were cut into chips and washed with calcium- and magnesium-free phosphate-buffered saline (PBS). Calvaria were incubated at 37 °C for 20 min with 0.25% trypsin and 1 h with 0.1% type II collagenase in PBS with gentle mixing. Incubation with type II collagenase was then repeated. Cells released from the bone chips were collected in α-modified minimum essential medium (α-MEM) containing 10% fetal bovine serum (FBS), 100 units/mL penicillin (Gibco) and 100 μg/mL streptomycin (Gibco), followed by centrifugation at 1000 revolutions per minute for 10 min. The resulting pellets were re-suspended in α-MEM containing 10% FBS. Cells were plated at appropriate density, and cultured at 37 °C under 5% CO2. Culture medium was changed every 2 d.

Rat osteoblasts (1.5×105/cm2) were plated in 6-well plates in α-MEM containing 10% FBS, 100 units/mL penicillin, 100 μg/mL streptomycin, 50 μg/mL ascorbic acid (Sigma) and 5 mmol/L sodium β-glycerophosphate (Sigma) for measurement of differentiation markers. After 4 d of culture, cells were starved with serum-free α-MEM and 0.2% bovine serum albumin (BSA) for 12 h. Cultures were also exposed to fresh serum-free medium with 0.2% BSA with or without NMDA, MK801 (both from Tocris Cookson Ltd UK), or other inhibitors.

RNA, cDNA preparation, and quantitative real-time PCR

To ensure statistical significance of microarray analyses, cultures of osteoblasts were incubated in osteogenic medium with 0.2% BSA containing NMDA or MK801 for 48 h. Total RNA was isolated from primary rat calvarial cells with TRIzol reagent (Invitrogen). RNA (1 μg) was reverse transcribed to cDNA with the Invitrogen Superscript Kit. Corresponding probes were hybridized to Illumina GeneChip arrays and subjected to bioinformatics analyses. cDNA was amplified with the Takara SYBR Green RT-PCR kit using gene-specific primers (Table 1) in the Real-Time PCR Detection System (Roche). Quantification and normalization to GAPDH amplicons was performed. Statistical analyses were performed with Prism 5.0 (GraphPad Software, San Diego, CA).

Table 1 Primer sequences used for PCRs.
Full size table

As might be expected, many genes that have previously been demonstrated to be NMDA or MK801-regulated can be found in our lists of genes as commonly regulated in the treatment regimens (Table 2 or 3). This implies that the microarray results and analyses are reliable. To validate this further, we used quantitative real time RT-PCR to examine the effect of NMDA or MK801 on individual genes. We examined the following ten genes (Nov, Stc1, Anxa1, Tspan8, Dab2, Nfkb2, Mmp12, Fmo1, Colec12, and Fap) which were commonly up-regulated by NMDA and down-regulated by MK801 as shown in Table 2. We also examined thirteen genes (Cib1, Ddit3, Gaa, Herpud1, Ninj1, Trappc2, Rpo1-2, Slc20a1, Slc3a2, Trib3, Yars, Serpinb2, and Thop1) which were commonly down-regulated by NMDA and up-regulated by MK801 as shown in Table 3.

Table 2 The common genes up-regulated by NMDA and down-regulated by MK801 based on GO analysis.
Full size table
Table 3 The common genes down-regulated by NMDA and up-regulated by MK801 based on GO analysis.
Full size table

Gene ontology analysis

http://omicslab.genetics.ac.cn/GOEAST/php/illumina.php was applied to GO analysis. We chose only GO categories that had a P-value of <0.001 and a log odds-ratio of >1.5-fold.

Alkaline phosphatase synthesis

Rat calvarial osteoblasts were plated at a density of 1.5×105/well into 6-well culture plates. After treatments, cells were washed three times with ice-cold PBS, scraped into 0.5 mL of ice-cold 0.5% Triton X-100 solution and lysed by ultrasonication in an ice bath for 2 min. The supernatant was centrifuged at 14 000×g and 4 °C for 5 min, then stored frozen at -20 °C until measurement of alkaline phosphatase levels. For the determination of these levels, cell lysates were incubated in a 96-well plate with 100 μL of 0.1 mol/L NaHCO3-Na2CO3 buffer (pH 10.0) containing 0.5% Triton X-100, with p-nitrophenylphosphate as substrate for 30 min at 37 °C. The absorbance of p-nitrophenol liberated in the reactive solution was read at 405 nm. Diluted cell lysates were measured at 740 nm for total protein content using the BCA method (Bio-Rad Protein Assay kit, Bio-Rad Laboratories, Richmond, CA, USA). ALP activity in the cells was normalized for total protein content of the cell lysate.

Western blotting

Cells treated with 0.5 mmol/L NMDA and wortmannin (100 nmol/L) (Cell Signaling, USA) for the indicated times were lysed and the protein concentrations were determined by Bio-Rad Protein Assay (Bio-Rad Laboratories, Richmond, CA). For Western blotting, 50 mg of total cell lysates was subjected to SDS-polyacrylamide gel electrophoresis. The protein was transferred to polyvinylidene difluoride membranes using transfer buffer (50 mmol/L Tris, 190 mmol/L glycin, and 10% methanol) at 120 V for 2 h. The membranes were incubated with blocking buffer (50 mmol/L Tris, 200 mmol/L NaCl, 0.2% Tween 20, and 3% bovine serum albumin) overnight at 4 °C. After washing three times with washing buffer (blocking buffer without 3% bovine serum albumin) for 10 min each, the blot was incubated with primary antibody (PI3K, phosphorylated PI3K) for 12 h, followed by horseradish peroxidaselabeled secondary antibody for 1 h. The membranes were washed again, and detection was performed using an ECL kit (ECL Plus Western Blotting Detection System, GE Healthcare UK Limited, Little Chalfont, UK) and exposed to film.

Statistical analysis

Each experiment was performed with at least three independent samples. The results are expressed as the mean±SD, unless otherwise indicated. Statistical significance of the observed differences was analyzed by one-way ANOVA where appropriate. A P value <0.05 was considered statistically significant.

Results

cDNA microarray and gene ontology analysis

Genes whose expression was changed in opposite directions by NMDA and MK801 treatment were defined as being commonly up-regulated or down-regulated genes. According to the ratio values, 353 genes were up-regulated by NMDA, 106 genes down-regulated by MK801 and hence 83 genes were the commonly up-regulated genes. There were 50 genes down-regulated by NMDA, 297 genes up-regulated by MK801 and 27 were the commonly down-regulated genes (Figure 1). We then examined the significant GO categories and genes by http://omicslab.genetics.ac.cn/GOEAST/php/illumina.php. A P-value of <0.001 and a log odds-ratio >1.5-fold were selected as the significant criteria. This narrowed the numbers of genes to 10 in the group of common genes that were up-regulated by NMDA and down-regulated by MK801 (Table 2), and 13 in the group of common genes down-regulated by NMDA and up-regulated by MK801 (Figure 2, Table 3).

Figure 1
figure 1

Number of genes changed by NMDA or MK801 treatment regimens. Genes identified by Illumina GeneChip microarray were changed with treatment by at least 1.5-fold compared to vehicle. The number of genes (both up- and down-regulated) common to both treatments is noted in the overlap of the Venn diagram.

Full size image
Figure 2
figure 2

Verification of common genes according to GO analysis by real-time PCR. A: Genes up-regulated by NMDA and down-regulated by MK801; B: Genes down-regulated by NMDA and up-regulated by MK801. Primary calvarial osteoblasts were cultured in medium with 0.2% BSA containing 0.5 mmol/L NMDA or 100 μmol/L MK801 for 48 h. mRNAs were isolated after 48 h and subjected to quantitative real-time PCR with primers described in Table 1. Comparative threshold values represent the mean of three samples normalized to GAPDH levels. Values are relative to those obtained from the vehicle groups. bP<0.05, cP<0.001 by one-way ANOVA versus the vehicle groups.

Full size image

Effects of inhibiting PKC, PKA, and PI3K on indexes of osteoblast differentiation

From the results of the microarray analysis as well as some previously published work in other cell types, we noted several genes (Anxa1, Mmp12, Stc1, Trib3, NF-κB, Ddit3) involved in PKC, PKA, and PI3K signaling pathways (see discussion). To determine the involvement of PKC in NMDA-mediated osteoblast differentiation, serum-starved cells were pretreated with Calphostin C (0.05 μmol/L), a specific inhibitor of PKC, for 90 min followed by coincubation with 0.5 mmol/L NMDA for 48 h. Cell morphology was assessed using a commercial ALP staining kit. Basal levels of ALP activity were unaffected by treatment of osteoblastic cells with Calphostin C alone. However, this treatment did lead to an observable reduction in alkaline phosphatase levels (Figure 3).

Figure 3
figure 3

Effect of inhibitors of signal transduction on alkaline phosphatase levels. Cells were pretreated with vehicle or the PKA inhibitor H-89 (20 nmol/L), the PKC inhibitor Calphostin C (Cal C, 0.05 μmol/L), or the phosphatidylinositol 3-kinase (PI3K) inhibitor wortmannin (W, 100 nmol/L) for 90 min. Cells were then treated with 0.5 mmol/L NMDA for 48 h. A: ALP staining study; B: ALP activity study. Mean±SD (n=8). bP<0.05, cP<0.001 vs control; eP<0.05, fP<0.001 vs the group treated with NMDA only.

Full size image

To determine whether activation of PKA is involved in NMDA-induced osteoblast differentiation, cells were pretreated with the PKA inhibitor H-89 (20 nmol/L). After incubation in serum-free medium for 12 h, cells were pretreated with 20 nmol/L H-89 for 90 min, followed by coincubation with 0.5 mmol/L NMDA for 48 h. Inhibition of PKA by H-89 treatment led to a decrease in NMDA-stimulated ALP activity levels (Figure 3). The ability of this inhibitor to curtail the effects of NMDA on osteoblast differentiation suggests that PKA activation is involved in NMDA-induced osteoblast differentiation.

We further explored signal transduction components related to NMDA-induced osteoblast differentiation by examining the involvement of PI3K using the PI3K inhibitor wortmannin. Cells were pretreated with wortmannin (100 nmol/L) for 90 min, followed by coincubation with 0.5 mmol/L NMDA for 48 h. This treatment protocol also led to a decrease in NMDA-stimulated ALP activity levels (Figure 3). The ability of these three inhibitors to curtail the effects of NMDA on markers of osteoblast differentiation suggests that activation of PKC, PKA, and PI3K is involved in the phenomenon of NMDA-induced osteoblast differentiation.

To assess whether PI3K were activated by NMDA, we assessed phosphorylation of PI3K using phospho-antibodies against phosphorylated peptides derived from PI3K. We found that 0.5 mmol/L NMDA induced a rapid increase in PI3K phosphorylation with maximal levels at 15 min. Prolonged NMDA stimulation up to 30 min, however, resulted in a decrease of phosphorylated PI3K levels toward baseline (Figure 4A).

Figure 4
figure 4

Effects of NMDA on PI3K activation in osteoblastic cells. Cell lysates were subjected to Western blot and incubated with PI3K or phosphorylated PI3K antibodies. (A) Cells are exposed to 0.5 mmol/L NMDA for 0, 10, 15, 20, and 30 min. PI3K was phosphorylated by NMDA and the peak reached at 15 min. (B) Cells incubated with wortmannin (100 nmol/L) for 90 min prior to treatment with 0.5 mmol/L NMDA, and then treated with or without NMDA for 15 min. The suppression of phosphorylated PI3K induced by NMDA was observed in the presence of wortmannin.

Full size image

To determine whether activation of PI3K is involved in NMDA-induced osteoblast differentiation, cells were pretreated with the PI3K inhibitor wortmannin. After incubation in serum-free medium for 12 h, cells were pretreated with 100 nmol/L wortmannin for 90 min, followed by coincubation with 0.5 mmol/L NMDA for 15 min. Inhibition of PI3K by wortmannin treatment led to a decrease in NMDA-stimulated phosphorylated PI3K levels (Figure 4B). The ability of this inhibitor to curtail the effects of NMDA on osteoblast differentiation suggests that PI3K activation is involved in NMDA-induced osteoblast differentiation (Figure 5).

Figure 5
figure 5

Schematic representations of signaling pathways from NMDA receptors to the multiple signaling system.

Full size image

Discussion

Our data suggest that NMDA promotes osteoblast differentiation via PKA, PKC and PI3K signaling pathways. These findings also demonstrate that NMDA directly acts on and affects osteoblasts and that the NMDA receptors expressed in primary rat calvaria osteoblasts are functional.

Within the large family of iGluRs, NMDARs constitute a subfamily identifiable by a specific molecular composition and unique pharmacological and functional properties22, 23. The most commonly used agonist at the glutamate recognition site of NMDA receptors is NMDA itself. However, it is not a substrate that promotes glutamate uptake24. Activation of NMDA receptors has been shown to be important in normal expression of bone matrix proteins4, 12. We have previously demonstrated that NMDA increased osteoblast ALP activity and osteocalcin (OC) in a dose-dependent manner, while the NMDA receptor antagonist MK801 reduced these effects12. We have also observed that NMDA increases ALP activity and OC expression in a time-dependent manner and that the peak of this increase reached 48 h after treatment (data not shown). Together, these observations suggest that NMDA regulates osteoblast differentiation via NMDA receptors.

In the present study, we employed cDNA microarray analysis, which was expected to be more sensitive than cDNA subtraction analysis, for the detection of specific genes related to the stimulation of NMDA receptors in rat osteoblasts. Annexin 1 (Anxa1) is one member of a family of phospholipid and calcium-binding proteins of which 20 are known at present25. It translocates from the cytoplasm to the outer cell surface via a Ca2+-dependent mechanism26, 27, 28. It is a substrate for protein kinase C and protein tyrosine kinases and has multiple phosphorylation sites as well as calcium and phospholipid binding properties29. Anxa1 was also shown to be activated as a consequence of sequential PI3K-PKC activation30, 31. Matrix metalloproteinases (MMPs) are a family of secreted or transmembrane zinc-dependent endopeptidases. The activation of MMP3, MMP12 and MMP13 was correlated with activation of the PI3K/Akt signaling cascade in microglial cells32. In epithelial cells and fibroblasts, MMP12 and MMP13 could be up-regulated in a PI3K, PKCδ and ERK1/2-dependent manner33. The stanniocalcins, comprising Stc1 and Stc2, are secreted homodimeric glycoprotein hormones with little homology to other proteins34. Stc1 was originally identified as a hormone secreted by the corpuscles of the stannius inteleost fish35 and in fish appears to function primarily in the prevention of hypercalcaemia mediated via the ability of Stc1 to reduce calcium uptake by gills and inhibit intestinal calcium transport in the gut36, 37, 38. In human endothelial cells Stc1 mRNA expression was up-regulated primarily through PKC, ERK and Ca2+ signaling pathways39. Trib3 (tribbles 3) is a mammalian homologue of Drosophila tribbles40, 41, 42. Expression of Trib3 was found to be PI3K-dependent in prostate cancer cells43. Furthermore, it has been suggested that the tribbles protein family is involved in regulation of the MAP kinase pathway44. Nuclear factor-κB (NF-κB) is a ubiquitous heterodimeric transcription factor that regulates inflammation and cell survival and differentiation45, 46. Many different kinases, including PKA, PKC, GSK-3β, PI3K, AKT, p38, NIK and even IKK, have been shown to induce phosphorylation of NF-κB directly or indirectly45, 46, 47. Regulation of NF-κB has also previously been suggested to depend on the PLCβ/PKC pathway48, 49. The PI3K/Akt signaling pathway seems able to positively regulate NF-κB activity50, 51. Studies have demonstrated that different stress response pathways mediating Ddit3 (DNA-damage inducible transcript 3) expression are regulated by protein kinases52, 53. Ddit3 is reported to be induced during serum starvation, glutamine deprivation, growth to confluency and upon exposure to tunicamycin, which promotes accumulation of proteins in the ER by preventing glycosylation54. Ddit3 protein was increased in pancreatic β-cells grown in high glucose that were exposed to U0126 to block ERK1/2 activity55. Our data showed Anxa1, MMP12, Stc1 and NF-κB were all up-regulated by NMDA in osteoblasts, while Trib3 and Ddit3 were down-regulated. In addition, it suggests that ERK, PKA, PKC and PI3K signaling pathways may be involved with changes of these genes' expression levels in NMDA-treated osteoblasts.

In the CNS, activation of NMDA receptors increases calcium influx, which is associated with spatial long term memory in the Morris water maze56 through the PKA signaling pathway56, 57, 58. Repetitive stimulation of glutamatergic NMDA receptors result in PKA and PKC activation in vivo59. The high Ca2+ permeability of NMDA receptor channels in bone is a characteristic similar to that of receptors expressed in the CNS5, 60. Ca2+ flux could trigger a number of second messenger responses that could ultimately be responsible for the anabolic effects of NMDA stimulation in bone. The MAPK, PKC, PKA and PI3K signaling pathways can all be directly activated downstream of calcium influx61. The activation of protein kinase C (PKC) and/or protein kinase A (PKA) plays an essential role in osteoblast differentiation62. In our previous study, we have observed activation of the ERK1/2 signaling pathway when osteoblasts were treated with NMDA12. With this background information, we further pursued the results of our microarray analysis information and specifically studied the PKC, PKA and PI3K signaling pathways to see if these signaling pathways are involved in NMDA-mediated osteoblast activation, independent of ERK1/2. We found that cells staining for ALP were significantly decreased by treatment with the PKA inhibitor H-89 or the PKC inhibitor Calphostin C. However, the ALP staining cells recovered after treatment with NMDA, which suggests a mechanism for NMDA-induced osteoblast differentiation that is, at least partly, dependent on PKA and PKC signaling. NMDA increased PI3K phosphorylation which is kinetically correlated to concomitant ERK1/2 phosphorylation in rat striatal neurons63. It was reported that intracellular Ca2+- regulated pathways signal through PI3K in osteoblasts subjected to mechanical strain64. The ability of the chemical inhibitor of PI3K, wortmannin, to significantly reduce NMDA-induced expression of the osteoblast differentiation marker ALP suggests that PI3K is involved in NMDA-mediated osteoblastic cell differentiation.

In summary, we have demonstrated that activation of NMDA receptors promotes osteoblast differentiation via PKA, PKC and PI3K signaling mechanisms which are independent of ERK1/2 signaling. The screening for specific genes carried out in this work allowed the discovery of signaling pathways which are associated with these genes. These findings yield new avenues by which to look for therapeutic targets in osteoporosis and provide useful direction for further investigation of the mechanisms involved.

Author contribution

Dr Jie-li LI designed the project, performed the research and wrote the paper; Dr Lin ZHAO worked on the data for revision; Dr Bin CUI helped on detailed research work; Dr Lian-fu DENG and Dr Guang NING contributed overview advice; Dr Jian-min LIU provided fundings on all the work and designed the project.