Osteoclast-Derived Complement Component 3a Stimulates Osteoblast Differentiation


  • For a Commentary on this article, please see Martin (J Bone Miner Res. 2014;29:1519–1521. DOI: 10.1002/jbmr.2276)


Bone remodeling is regulated by a coupling of resorption to subsequent formation; however, the “coupling factor” and underlying mechanism are not fully understood. Here, we found that the condition medium (CM) of mature osteoclasts contains a humoral factor that stimulates the differentiation of primary osteoblasts, as determined by alkaline phosphatase (ALP) activity. We purified osteoblastogenesis-stimulating activity from 3 L of osteoclast CM through successive ion exchange chromatographies by monitoring the ALP activity of osteoblasts and identified complement component 3 (C3). Expression of the C3 gene increased during osteoclastogenesis, and the cleavage product C3a was detected by ELISA in the CM of osteoclasts but not in that of bone marrow macrophages. The osteoblastogenesis-stimulating activity present in osteoclast CM was inhibited by a specific antagonist of the C3a receptor (C3aR), SB290157. Conversely, the retroviral expression of C3a as well as treatment with the C3aR agonist, benzeneacetamide, stimulated osteoblast differentiation. C3 gene expression in bone was increased in the high bone turnover states of ovariectomy (OVX) or a receptor activator of NF-κB ligand (RANKL) injection, and blocking the action of C3a with the daily administration of SB290157 resulted in the attenuation of bone formation elevated by OVX and the exacerbation of bone loss. These results suggest that osteoclast-derived C3a functions in the relay from bone resorption to formation and may be a candidate for a coupling factor. © 2014 American Society for Bone and Mineral Research.


Bone remodeling is a dynamic process regulated by two cellular mechanisms, osteoblastic bone formation and osteoclastic bone resorption. These functions are tightly regulated through reciprocal crosstalk to maintain bone mass. An imbalance between these activities results in metabolic bone diseases such as osteoporosis and osteopetrosis.[1] A previous study showed that preceding bone resorption is a prerequisite for subsequent bone formation,[2] and several candidates for the coupling factor, which bridges the two processes, have been proposed.[3]

TGF-β and IGF-1, which are abundantly stored in bone matrix and released and activated during osteoclastic bone resorption, have been reported to function in the coupling of bone resorption to formation.[4, 5] Using the gene expression profiles of osteoclasts by microarray analysis, the candidate molecules underlying coupling, which are produced by osteoclasts and stimulate osteoblast differentiation, have recently been reported and include ephrinB2,[6] collagen triple helix repeat containing 1 (Cthrc1),[7] and sphingosin-1-phosphate (S1P).[8, 9] Bidirectional signaling between ephrinB2 on osteoclasts and the receptor EphB4 on osteoblasts was shown to link bone resorption to formation through direct cell-cell contact.[6] In view of the expression of gp130 and LIF receptor on osteoblasts, one of their ligands, cardiotrophin-1 (CT-1), has been proposed as an osteoclast-derived coupling factor.[10] Cthrc1, a secreted protein of osteoclasts, whose gene expression is induced only when osteoclasts are activated and engage in bone resorption, is capable of stimulating the recruitment and differentiation of osteoblasts, and was previously shown to be involved in coupling in vivo.[7] The expression of sphingosine kinase 1 (Sphk1) was reported to be upregulated during osteoclastogenesis, and S1P produced by Sphk1 stimulated osteoblast recruitment, which implied that S1P acts as a coupling factor.[8, 9, 11] However, no study has attempted to biochemically characterize and purify osteoclast-derived bioactivity that is capable of stimulating osteoblastic bone formation.

In the present study, we established a coculture system of primary osteoblasts with mature osteoclasts generated ex vivo to test whether the presence of osteoclasts promoted osteoblastogenesis. Having detected such bioactivity, we then attempted to purify it from the conditioned medium of mature osteoclasts using biochemical techniques. Mass spectrometric analysis of an enriched fraction yielded peptide sequences matching complement component 3 (C3), which prompted us to study the function of its bioactive fragment C3a in the communication between osteoclasts and osteoblasts both in vitro and in vivo.

Materials and Methods

Reagents and recombinant proteins

SB290157, a C3a receptor antagonist, was purchased from Calbiochem (cat#559410, La Jolla, CA, USA), and benzeneacetamide (α-cyclohexyl-N-[1-[1-oxo-3-(3-pyridinyl)propyl]-4-piperidinyl]-benzeneacetamide), a C3a receptor agonist, from Santa Cruz Biotechnology, Inc. (sc-214644, Santa Cruz, CA, USA). These reagents were dissolved in DMSO. Recombinant murine macrophage colony-stimulating factor (M-CSF) and GST-receptor activator of NF-κB ligand (RANKL) were prepared as described previously.[12] Except where stated, all miscellaneous reagents were obtained from Sigma Chemical Co. (St. Louis, MO, USA).

Preparation of bone marrow macrophages and osteoclasts

Bone marrow macrophages (BMMs) were prepared from the whole bone marrow of 8- to 10-week-old C57BL/6 mice (Clea Japan Inc., Shizuoka, Japan) and cultured as described previously.[13] Briefly, BMMs were cultured in α-MEM (Gibco, Life Technologies Corp., Carlsbad, CA, USA) containing 10% fetal bovine serum (FBS) and penicillin/streptomycin (Gibco; complete medium) in the presence of M-CSF (1/10 vol. CMG14-12 culture supernatant) and GST-RANKL (100 ng/mL) for 2 days and used as TRAP-positive mononuclear preosteoclasts (pOC). These pOCs were further cultured for 2 days in the presence of M-CSF and GST-RANKL to generate mature osteoclasts (mOC).

Preparation of conditioned medium

BMMs or mOC were cultured in the presence of M-CSF or M-CSF and GST-RANKL, respectively, for 2 days, and the conditioned medium (CM) was harvested. Each CM was concentrated 20-fold using the Amicon Ultra Filter Unit (10K, Millipore Corporation, Billerica, MA, USA) and was then used as BMM CM and OC CM, respectively. Complete medium containing M-CSF and RANKL were also concentrated and used as the control.

Osteoblast culture, alkaline phosphatase assay, TRAP solution assay, and TRAP staining

Primary osteoblastic cells were obtained from the calvaria of 1-day-old C57BL/6 mice, as previously described.[14] Briefly, calvariae were resected and placed into ice-cold PBS to remove the surrounding tissue and pooled. Sequential digestion of the calvariae with 0.1% collagenase (Wako Pure Chemical Industries, Ltd., Osaka, Japan) and 0.2% dispase (Gibco) in PBS for 10 minutes was repeated four times. The cells obtained from each digestion except the first fraction were pooled and plated in complete medium (α-MEM + 10% FBS) and grown at 37°C with 5% CO2. Adherent osteoblasts were harvested after 4 to 5 days with trypsin/EDTA (Gibco), and stored in liquid nitrogen until use.

Calvarial osteoblasts were seeded at 1 × 104 cells/well in a 96-well-plate or 2.3 × 104 cells/well in 48-well plate for the alkaline phosphatase (ALP) assay. Cells were cultured in the presence of the C3a receptor agonist or OC CM (1/10 vol.) with different doses of the C3a receptor antagonist, SB290157. DMSO (0.1%) was used as vehicle control for the antagonist and agonist. Briefly, after removing the culture medium, cells were washed with PBS and were then lysed in 0.15 M NaCl, 3 mM NaHCO3, 0.1% Triton X-100, pH 9.3. The cell lysate was incubated at 37°C in buffer containing 1 M diethanolamine and 0.5 mM MgCl2, pH 9.8, for 30 minutes using p-nitrophenylphosphate as a substrate. ALP activity was measured, as described previously.[15] Enzyme activity was expressed as micromoles of p-nitrophenol produced per minute per mg protein. Protein content was determined using the BCA protein assay kit (Pierce Chemical Co., Rockford, IL, USA). ALP staining was performed using an alkaline phosphatase staining kit (Sigma).

To assess the mineralization of the osteogenic culture condition with ascorbic acid (50 μg/mL) and β-glycerophosphate (10 mM), cells were washed with PBS twice and with water. After fixation with ethanol for 15 minutes, calcified nodules were stained with 1% alizarin red S solution for 15 minutes and were then washed with water and air-dried. The areas of mineralization were determined with the NIH Image program (http://rsb.info.nih.gov/nih-image/).

The TRAP solution assay was performed to assess osteoclast differentiation in the coculture with osteoblasts. TRAP enzyme activity per well was measured by the conversion of p-nitrophenylphosphate (20 nM) to p-nitrophenol in the presence of 80 mM sodium tartrate and was expressed as optical density at 405 nm. For TRAP staining, cultured cells were fixed with 10% formalin for 5 minutes and then with ethanol:aceton (50:50 v/v) for 1 minute at room temperature, and incubated in acetate buffer (pH 4.8) containing naphtol AS-MX phosphate (Sigma), fast red violet LB salt (Sigma), and 50 mmol/L sodium tartrate. After washing with distilled water and air-drying, the number of TRAP-positive mature osteoclasts was counted.

Protein purification

Mature osteoclasts were cultured in complete medium with M-CSF and GST-RANKL for 2 days and CM was harvested for protein purification. CM was concentrated with buffer exchanged into 10 mM HEPES buffer (pH 6.0, buffer 1) using Pericon XL Filter 10K (Millipore). Concentrated CM was applied to a Mono UNOsphere Q column (Bio-Rad, Hercules, CA, USA) equilibrated with buffer 1. After washing with 50-column volumes of buffer 1, bound proteins were eluted by a gradient of 0% to 100% 1 M NaCl in buffer 1. The eluate containing ALP-stimulating activity was collected and applied to a ConA Sepharose 4B column (GE Healthcare, Willys, Uppsala, Sweden) equilibrated with 1 mM CaCl2, 1 mM MgCl2, and 10 mM HEPES pH 7.3 (buffer 2). After washing with 50-column volumes of buffer 2, bound proteins were eluted by a gradient of 0% to 100% 0.5 M methyl α-D-mannopyranoside, 0.1 M NaCl in buffer 2. The eluate containing ALP-stimulating activity was collected and applied to a UNOsphere S column (Bio-Rad) equilibrated with 10 mM HEPES (pH 5.5, buffer 3). After washing with 50-column volumes of buffer 3, bound proteins were eluted by a gradient of 0% to 100% 1 M NaCl in buffer 3. The eluate with highly concentrated ALP-stimulating activity was dissolved in SDS-PAGE sample buffer and was resolved in SDS-PAGE. Protein bands were visualized by Coomassie brilliant blue staining, and all protein bands were excised with a scalpel. Samples were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) by Shimadzu Corporation (Kyoto, Japan). The data were submitted to the MASCOT program for identification.

Retroviral expression of C3a and knockdown of C3 by shRNA

To construct a mouse C3a retrovirus expression vector, signal sequences encoding the 26 hydrophobic amino acids from the 1Met to 26Pro of C3 were fused to the C3a portion encoding the 78 amino acids from 671Ser to 748Arg, and the DNA fragment was inserted into the pMX-puro retrovirus expression plasmid. cDNA fragments of signal sequences and the C3a portion of C3 were amplified from the cDNA of osteoclasts using the primers of 5'-CCCAATTGGGCCATATAAAGAGCCAGCGG-3' and 5'-AACGTGCACTGAGGGGATCCCCAGAGCTAATG-3' for the signal sequence, and the C3a portion primers of 5'-GTCTGCAGTCAGTACAGTTGATGGAAAG-3' and 5'-TTTGCGGCCGCTCACCTGGCCAGGCCCAGCACGTG-3'. These two DNA fragments were double-digested with MfeI and ApaL1, and SfcI and NotI, respectively, and the cDNAs were ligated into the EcoRI/NotI fragment of the pMX-puro vector (a gift from Dr Toshio Kitamura, University of Tokyo). The resulting C3a expression vector was cotransfected with the pVSV-G vector into GP2-293 cells (Clontech Laboratories Inc., Mountain View, CA, USA). Calvarial osteoblasts were infected with the retrovirus containing supernatants from GP2-293 cultures, and infected cells were selected with 2 μg/mL of puromycin for 2 days. After selection, cells were cultured in osteogenic media containing ascorbic acid and β-glycerophosphate, and RNAs were extracted for quantitative RT-PCR.

Lentiviral shRNA expression vectors for the C3 gene (C3 shRNA 1: MISSION shRNA TRCN0000066882, C3 shRNA 2: MISSION shRNA TRCN0000334477) and non-target shRNA control (SHC002) were obtained from Sigma. shRNA-mediated knockdown was performed according to the manufacturer's instruction, as previously described.[16] Briefly, lentivirus expression vectors were cotransfected with MISSION Lentiviral Packaging Mix (Sigma) into 293T cells. BMMs were then infected with the lentivirus containing supernatants from 293T cultures, and infected cells were selected with 2 μg/mL of puromycin for 2 days and used for the preparation of OC CM.

RNA isolation and RT-PCR

Total RNAs were extracted from cells using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and used for RT-PCR analyses. The primers used for RT-PCR are summarized in Tables 1 and 2. Quantitative PCR amplification of cDNA was performed using the power SYBER green PCR master mix on the 7300 fast real-time PCR system (Applied Biosystems, Carlsbad, CA, USA).

Table 1. Oligonucleotide Primers for General PCR
GeneForward primer (5'-3')Reverse primer (5'-3')
Table 2. Oligonucleotide Primers for Quantitative PCR
GeneForward primer (5'-3')Reverse primer (5'-3')

Ovariectomy and SB290257 administration to mice

Ten-week-old female ddy mice were ovariectomized (OVX) or sham-operated and were treated ip with SB290157 (30 mg per kg body weight) or vehicle (5% ethanol/PBS) daily for 6 days. Nine days after the last injection, mice were euthanized and subjected to bone analysis by micro-computed tomography (micro-CT) and bone histomorphometry.

Bone analysis

Bone histomorphometry was performed on undecalcified sections, with tetracycline and calcein double labeling. Histomorphometric parameters were measured at the Niigata Bone Science Institute (Niigata, Japan).

Micro-CT scanning was performed on proximal tibias using a μCT-40 scanner (SCANCO Medical AG, Bassersdorf, Switzerland) with a resolution of 12 μm, and microstructure parameters were calculated three-dimensionally, as described previously.[17] The proximal tibia was positioned so as to be scanned cranio-caudally using 320 slices with 12-micron increments at 45 kVp and 177 μA. On the original 3D image, morphometric indices, including bone volume (BV), tissue volume (TV), and trabecular number (Tb.N), were directly determined from the binarized volume of interest (VOI). Nonmetric parameter, such as connectivity density (Conn. Dens), was also described previously,[17] and Conn. Dens was determined as the number of trabecular connections per cubic millimeter. The nomenclature for micro-CT and histomorphometry followed the recommendations of the published guidelines.[18, 19]

Statistical analysis

Data were expressed as the mean ± SD. We carried out statistical analysis using Student's t test. Values were considered statistically significant at *p < 0.05, **p < 0.01, and ***p < 0.001.

Study approval

All experiments were performed in accordance with the National Center for Geriatrics and Gerontology's ethical guidelines for animal care, and the experimental protocols were approved by the Animal Care Committee.


The presence of osteoclasts promotes osteoblast differentiation

Osteoclasts are known to be generated in cocultures of bone marrow macrophages (BMMs) with calvarial osteoblasts stimulated with osteotrophic hormones such as 1α,25-dihydroxyvitamin D3 [1α,25(OH)2D3] or PTH.[14] To address whether mature osteoclasts influence osteoblastic differentiation, we set up a coculture system of calvaria-derived primary osteoblasts with BMM-derived pOCs in the presence of increasing doses of RANKL, and ALP activity was measured after 4 days, as a differentiation marker for osteoblast. Evidently, RANKL induced osteoclast differentiation in a dose-dependent manner, as determined by TRAP activity of their cell lysates (Fig. 1A), and the ALP activity of osteoblasts concomitantly increased in parallel with TRAP activity (Fig. 1B). To confirm that this effect was mediated humorally, we prepared CM from mature osteoclasts and added it to the osteoblast culture. CM from mature osteoclasts markedly induced ALP activity in cell lysates (Fig. 1D) as well as the staining of ALP (Fig. 1C), whereas neither CM from BMMs nor medium alone stimulated ALP activity (Fig. 1D and data not shown). The CM of mature osteoclasts also markedly increased mineralization, as shown in alizarin red staining, under osteogenic culture conditions with ascorbic acid and β-glycerophosphate (Fig. 1E, F). The osteoblastogenesis-stimulating activity of osteoclast CM was labile to heating and low/high pH, implying that it is a protein factor. The molecular size estimation by ultrafiltration also suggested that it is distinct from S1P (data not shown). Collectively, these results demonstrate that mature osteoclasts secrete a protein factor(s) that promotes osteoblastogenesis.

Figure 1.

The conditioned medium of osteoclasts contains osteoblastogeneis-stimulating activity. (A, B) Coculture of osteoclasts with osteoblasts stimulated osteoblastogenesis. Calvaria-derived primary osteoblasts were cultured with preosteoclasts (pOC) in the presence of RANKL at the indicated doses for 4 days. TRAP activity (A) by solution assay and the ALP activity (B) of cell lysates from those cultures were measured. Data are expressed as the mean ± SD, n = 3. *p < 0.05, **p < 0.01. (C) The conditioned medium (CM) of mature osteoclasts stimulated ALP activity in osteoblasts. The osteoclast (OC) CM was collected and concentrated 20-fold by ultrafiltration, and concentrated CM was added to the osteoblast culture at 1/10 vol. and cultured for 2 days (C), 1 day and 2 days (D), or 3 weeks (E, F) for the ALP staining (C), ALP assay (D), or Alizarin red staining (E, F), respectively. Mineralized area was determined with the NIH Image program (F). Scale bars = 200 μm. Data are expressed as the mean ± SD, n = 3. *p < 0.05, **p < 0.01, ***p < 0.001.

Identification of osteoclast-derived osteoblastogenesis-stimulating factor

To identify the osteoblastogenesis-stimulating factor, we attempted to purify ALP-stimulating bioactivity through successive, multistep ion exchange column chromatographies (Q, ConA, and S column) from 3 L of CM. The recovery of purified proteins through three steps of column chromatographies is shown in Table 3. As shown in Fig. 2A, B, fractions of the final S sepharose column chromatography were divided into five pools (E1 to E5), and the ALP activity of each fraction was determined. The highest ALP-stimulating activity was detected in E5, which was subjected to SDS-PAGE after concentration (Fig. 2C), and protein bands on SDS-PAGE were analyzed by LC-MS/MS. Several peptide sequences were obtained and identified as shown in Fig. 2C, and considering the humoral nature of osteoblastogenesis-stimulating activity, we focused on complement component 3 (C3) as a secreted protein. C3 peptide sequences detected by LC-MS/MS analysis were identical with murine C3 but not bovine C3 sequence (Fig. 2D).

Table 3. Purification Profiles of Three Steps of Column Chromatographies
ColumnProtein (mg/mL)Total protein (mg)Recovery (%)Relative ALP-increasing activity in each fraction
Figure 2.

Purification of osteoblastogenesis-stimulating activity and identification as C3a. (A, B) Osteoblastogenesis-stimulating activity was purified from osteoclast CM using an ion exchanged column (Q sepharose) and ConA column. ALP-stimulating activity was applied to an S sepharose column and eluted by an NaCl gradient (A). The protein content in each fraction and NaCl concentration are indicated by solid and dashed lines, respectively. Protein fractions were divided into 5 pools, E1 (fraction 2 to 11), E2 (12 to 22), E3 (23 to 30), E4 (31 to 38), and E5 (39 to 42). E1 through E5 fractions were added to osteoblast cultures, and ALP activity was determined by day 3 (B). Osteoclast (OC) CM (concentrated 20-fold) and PBS (Cont.) were added at 1/10 vol. and used as positive and negative controls, respectively. ALP activity was corrected by protein content and expressed as nmol/h/mg protein of the osteoblast lysate. Note that ALP activity was the highest in the E5 fraction, which was further examined for SDS-PAGE and LC-MS/MS analyses. (C) SDS-PAGE of fraction E5 with identified proteins by LC-MS/MS analysis. (D) Comparison of detected peptide sequences with murine and bovine C3. (E) Quantitative RT-PCR analysis for C3 expression during osteoclastogenesis. RNA was extracted from mouse bone marrow macrophages (BMM), preosteoclasts (pOC), and mature osteoclasts (mOC). Data are expressed as the mean ± SD, n = 3. *p < 0.05, **p < 0.01. (F) Mature osteoclasts, but not BMMs, secrete C3a. Culture supernatants from BMMs and mOC were assayed for C3a using the mouse C3a ELISA kit. Data are expressed as the mean ± SD; n = 3. *p < 0.05. ND = not detected.

Quantitative RT-PCR analysis demonstrated that the expression of C3 progressively increased during osteoclast differentiation (Fig. 2E). C3 was previously shown to be a component in the complement system that provides an important arm of innate immunity[20, 21] and is cleaved into C3a and C3b in the complement pathway. Because C3a is known to be a bioactive molecule that can induce local as well as systemic inflammatory responses, we chose to characterize C3a bioactivity in the CM of osteoclasts. ELISA specific for C3a demonstrated that C3a existed in the CM of mature osteoclasts but not in that of BMM (Fig. 2F), indicating that mature osteoclasts secrete C3a.

C3a stimulates osteoblast differentiation

We next examined whether C3a stimulates osteoblast differentiation and accounts for the ALP-stimulating activity detected in osteoclast CM. Because C3a is known to bind to and transmit signals through the C3a receptor (C3aR) on the target cell surface, the expression of C3aR was examined by RT-PCR. As shown in Fig. 3A, C3aR was expressed in stromal cell lines, such as C3H10T1/2 and ST2, as well as in primary calvarial osteoblasts. To assess the function of osteoclast-derived C3a on osteoblast differentiation, calvarial osteoblasts were cultured with osteoclast CM in the absence or presence of the C3aR antagonist, SB290157, which is known to competitively bind C3a receptor and inhibit C3a action.[22, 23] As shown in Fig. 3B, SB290157 dose-dependently inhibited the ALP activity induced by osteoclast CM, whereas SB290157 alone had no effect on ALP activity. These results suggest that C3a present in the CM of osteoclasts is responsible for the stimulation of ALP activity in osteoblasts.

Figure 3.

Osteoclast-derived C3a stimulates osteoblast differentiation through the C3a receptor. (A) The C3a receptor (C3aR) was expressed in stromal cell lines and primary osteoblastic cells. RNA was extracted from the mouse mesenchymal cell line C3H10T1/2, mouse bone marrow stromal cell line ST2, and mouse calvaria-derived primary osteoblasts, and RT-PCR was performed for the expression of C3aR. (B) The C3aR antagonist inhibited the ALP-stimulating activity in OC CM. Primary osteoblasts were cultured with OC CM in the presence of the indicated doses of the C3aR antagonist, SB290157 (SB), for 3 days, and cell lysates were analyzed for ALP activity. Data are expressed as the mean ± SD, n = 3. *p < 0.05, **p < 0.01. (C–E) The knockdown of C3 expression by shRNAs inhibited the ALP-stimulating activity in OC CM. Lentiviruses for C3 shRNAs (1 and 2) were infected to BMMs, and transduced cells were differentiated into osteoclasts by M-CSF and RANKL. RNAs and conditioned medium were prepared and assayed for C3 expression (C) and C3a concentrations (D) by quantitative RT-PCR and C3a ELISA, respectively. Data are expressed as the mean ± SD, n = 3. **p < 0.01. Note that the CM of osteoclasts, in which C3 expression was inhibited by C3 shRNA, lost ALP-stimulating activity when added to osteoblast cultures for 4 days (E). Data are expressed as the mean ± SD, n = 3. **p < 0.01. (F) Stimulation of C3aR in osteoblasts increased ALP activity. Calvarial osteoblasts were cultured with increasing doses of the C3aR agonist, benzeneacetamide, for 3 days and ALP activity was examined. Data are expressed as the mean ± SD, n = 3. *p < 0.05. (G) Forced expression of C3a in osteoblasts stimulated ALP activity. C3a was retrovirally expressed in calvarial osteoblasts, and ALP activity was determined on the indicated days. Data are expressed as the mean ± SD, n = 3. *p < 0.05.

We then knocked down C3 gene expression in osteoclasts using lentiviral shRNA (Fig. 3C), and CM was assayed for ALP activity. shRNAs 1 and 2 against C3 led to a significant decrease in the concentration of C3a in the CM of osteoclasts (Fig. 3D), but did not affect osteoclastogenesis as evaluated by TRAP activity (data not shown). Both failed to stimulate the ALP activity of osteoblasts significantly compared with the control CM derived from osteoclasts transduced with scramble shRNA (Fig. 3E).

Conversely, benzeneacetamide, the C3aR agonist, which is known to bind C3aR and activate ERKs,[24] dose-dependently increased ALP activity when added to osteoblast cultures (Fig. 3F). Furthermore, when C3a was retrovirally expressed in calvarial osteoblasts, ALP activity was significantly higher than that in cells transduced with the vector alone (Fig. 3G). Thus, the results of pharmacological as well as genetic experiments indicate that osteoclast-derived C3a, but not C3a derived from the serum, stimulates osteoblast differentiation through C3aR expressed on osteoblasts.

C3a function in vivo

We used the OVX model, in which an elevation in bone resorption is associated with increased bone formation, to gain an insight into the function of C3a in bone coupling in vivo. As shown in Fig. 4A, C3 gene expression in bone was significantly higher in OVX-treated mice than in sham-operated mice. Osteoclast marker genes, such as TRAP, cathepsin K, and the calcitonin receptor, were also upregulated by OVX (data not shown). Accordingly, the concentration of C3a in the bone marrow was increased significantly after OVX, as measured by ELISA specific for C3a (Fig. 4B). The RANKL injection, which has been reported to cause transient elevations in osteoclastic bone resorption,[25] also increased C3 expression in bone along with the osteoclast marker genes (Fig. 4C). These results suggest that C3 expression in bone is increased in vivo under high bone turnover states with excessive osteoclast activation.

Figure 4.

Increased C3 expression in bone in high bone turnover states. (A, B) OVX stimulated C3 expression in bone and increases the concentration of C3a in bone marrow. Ten-week-old female ddy mice were ovariectomized (OVX) or sham-operated. Four days after OVX, RNA was extracted from the femur and quantitative RT-PCR was performed for the expression of C3 (A). Data are normalized by Gapdh and expressed as the mean ± SD, n = 3. *p < 0.05. After bone marrow isolated from the femurs was diluted within 400 μL of PBS, the concentration of C3a was determined using the mouse C3a ELISA kit. Data are expressed as the mean ± SD, n = 3. *p < 0.05. (C) The RANKL injection increased C3 expression in bone. Ten-week-old male C57Bl/6 mice were injected sc with vehicle or GST-RANKL (1 mg per kg body weight) twice within a 24-h interval. Mice were euthanized 3 days after the first injection, and RNA was extracted from the femur. Quantitative RT-PCR analysis was performed for the expression of C3, tartrate-resistant acid phosphatase (Acp5), cathepsin K (Ctsk), and calcitonin receptor (Calcr). Data are normalized by Gapdh and expressed as the mean ± SD, n = 3. *p < 0.05.

To delineate the function of C3a in a high bone turnover state in vivo, especially in the context of the coupling of bone resorption to formation, we focused on the OVX mouse model, in which the coupling of bone resorption to formation is stimulated with a net decrease in bone mass. We speculated that if osteoclast-derived C3a was involved in the stimulation of bone formation after accelerated resorption, the secondary stimulation of bone formation should be inhibited when C3a action is blocked by the specific C3aR antagonist, SB290157, which would then exacerbate net bone loss. Ten-week-old female ddy mice were subjected to a sham operation or OVX, then treated daily with an intraperitoneal administration of SB290157 for 6 days, and the bone was analyzed at 14 days after OVX, as schematically shown in Fig. 5A.

Figure 5.

C3a in the coupling of bone formation to resorption in the OVX model. (A) Ten-week-old female ddy mice were sham-operated or ovariectomized (OVX) and injected ip daily with a vehicle (PBS) or the C3a receptor-specific antagonist SB290157 (SB; 30 mg per kg body weight). Mice were euthanized 14 days after OVX for bone analysis. (B) Histomorphometric analysis at the tibial metaphysis. The bone formation rate corrected by bone surface (BFR/BS) and mineral apposition rate (MAR) are shown. Data are shown as the mean ± SD, n = 8. *p < 0.05. (C) Micro-computed tomography analysis at the tibial metaphysis. Bone volume fraction (BV/TV), trabecular number (Tb.N), and connectivity density (Conn.Dens) are shown. Data are shown as the mean ± SD, n = 8. *p < 0.05. (D) Representative micro-CT images are shown. Scale bars = 500 μm.

Histomorphometric analysis at the tibial metaphysis revealed that, as expected, OVX caused a significant increase in bone formation, as assessed by the bone formation rate (BFR/BS) and mineral apposition rate (MAR) (Fig. 5B). Interestingly, the administration of SB290157 abolished the stimulation of bone formation induced by OVX (Fig. 5B). Furthermore, 3D bone analysis by micro-CT revealed that bone loss and the deterioration in trabecular architecture induced by OVX was exacerbated by the administration of SB290157 (Fig. 5C, D). These results are consistent with our concept that C3a, the expression of which is increased in bone after OVX, acts through C3aR to stimulate bone formation and maintain bone mass and structure.


To our knowledge, this is the first study to demonstrate that osteoclasts secrete C3a, especially in larger amounts in high bone turnover states, and that osteoclast-secreted C3a acts on osteoblastic cells through the C3a receptor to promote their differentiation. Earlier studies demonstrated that C3 expression was induced in bone marrow stromal cells when treated with 1α,25(OH)2D3 and that C3 stimulated osteoclast differentiation.[26, 27] Taken together with the present findings that osteoclasts secrete C3 and that the cleavage product C3a stimulates osteoblastogenesis, C3 is suggested to be involved in the bidirectional communication between osteoblasts and osteoclasts. Our results provide further in vivo evidence that C3a mediates the coupling of bone resorption to formation at least in a high turnover model caused by an estrogen deficiency, which indicates a new anabolic function of osteoclast-derived C3a in bone remodeling.

Tu and colleagues reported that C3 or C3aR deficiency reduces osteoclastogenesis due to lower RANKL/OPG ratio than that from WT, whereas it has not been reported whether osteoclasts are reduced in number in the KO mice.[27] It is expected, therefore, that C3 KO mice may exhibit lower bone turnover than WT mice. In addition, it has also been reported that C3aR KO mice exhibit decreases in body weight and serum leptin, resulting in insulin resistance.[28] Because leptin is known to inhibit bone formation through the central nervous system,[29, 30] it is conceivable that C3aR KO mice may exhibit altered bone formation due to reduced serum leptin level. Thus, it may be difficult to evaluate a coupling function of C3a under the complex situations of the systemic KO mice. To obtain the definitive evidence that osteoclast-derived C3a acts directly on osteoblasts to regulate coupling in vivo, analysis of osteoclast-specific C3 and/or osteoblast-specific C3aR gene knockout models would be required.

In addition to TGF-β and IGF-1, which have been shown to be released from the bone matrix during osteoclastic bone resorption,[4, 5] recent studies suggest that osteoclasts produce various molecules themselves, both secreted and membrane proteins that stimulate osteoblast differentiation, such as S1P and ephrinB2.[6, 8, 9, 11] We recently identified Cthrc1 as a coupling factor produced by active, bone-resorbing osteoclasts,[7] whereas C3a is secreted by osteoclasts irrespective of their activation status (in this study). Thus, sequential interactions may occur between osteoclasts in the distinct phases of differentiation/function and osteoblast lineage cells. Identification of the osteoclast-derived factors, Cthrc1 and C3a, demonstrates that osteoclasts, both dependent and independent of their resorptive activity, secrete factors that stimulate osteoblastic bone formation.[31]

C3 plays a central role in activating the complement system, and C3a is produced by the proteolytic cleavage of C3 convertase in several processing pathways.[32, 33] Because C3 is known to be abundantly present in serum, it was previously thought that C3a was derived from C3 in the serum. However, the knockdown of C3 expression by shRNA in osteoclasts revealed that it decreased not only C3 mRNA but also C3a content in the conditioned medium, which demonstrated that C3a was derived from osteoclasts, not from the serum. We have examined the expression of complement factors involved in the generation of C3a from C3, including C1q, MASP-1, -2, MBL, Factor D, and Factor B; we could not detect the expression of these genes in osteoclasts (data not shown). On the other hand, we also confirmed that C3 expression was upregulated in osteoblasts treated with 1α,25(OH)2D3, which was consistent with the findings of a previous report;[34] however, the C3a protein could not be detected in their conditioned medium (data not shown). These findings suggest that there may be a particular processing system for C3a production in osteoclasts but not in osteoblasts.

Although C3 is produced in the liver and functions in innate immunity, locally produced C3a regulates physiologically important processes in various tissues. For example, C3a generated by immune cells acts on both antigen-presenting cells and T cells, and these paracrine and autocrine interactions are involved in T-cell proliferation and survival in adaptive immunity.[35] Previous studies showed that C3a promoted liver regeneration, which stimulated the proliferation of hepatocytes and also induced ischemia/reperfusion injury in the liver.[36-38] Furthermore, C3a present in injured tissues contributed to the recruitment of mesenchymal stem cells through the G-coupled receptor of C3aR by activating ERKs and Akt.[39] Thus, locally produced C3a acts not only as anaphylatoxin but also induces multiple cellular processes such as cell proliferation and chemotaxis in the adaptive immune response and tissue regeneration, respectively. Our results further provide a new function of C3a as a differentiation factor for osteoblast maturation. Therefore, the mechanisms by which C3a transmits osteogenic signals through the C3a receptor on osteoblasts remain to be determined.

In conclusion, we identified C3a with osteoblast-stimulating activity in the CM from mature osteoclasts and demonstrated that osteoclast-derived C3a stimulated osteoblastogenesis in vitro. In vivo evidence further showed that C3a was produced in bone marrow and acted to enhance bone formation in high bone turnover states by an estrogen deficiency. We propose that C3a is involved in the coupling of bone resorption to formation at the initial stage of osteoclast differentiation.


All authors state that they have no conflicts of interest.


We thank Mrs Mie Suzuki for her technical assistance and Dr Michio Kimura (Japan Pharmaceutical University) for technical advice on protein purification.

This study was supported in part by a grant from the program Grant-in-Aid for Science Research C (#22590330 to ST), for Young Scientists B (#24790371 to KM), and for Science Research on Innovative Areas (#22118007 to KI) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan, and a grant (ST) from Japan Foundation for Aging and Health.

Authors' roles: Study design: KM, KI, and ST. Data acquisition: KM, KP, MI, and ST. Data analysis and interpretation: KM, KP, MI, KI, and ST. Drafting manuscript: KM, KI, and ST. Approving final version of manuscript: KM, KP, MI, KI, and ST. ST takes responsibility for the integrity of the data analysis.