Lysophosphatidic Acid Receptor Type 1 (LPA1) Plays a Functional Role in Osteoclast Differentiation and Bone Resorption Activity*

Background: Lysophosphatidic acid (LPA) is a bioactive lipid with pleiotropic activities due to activation of six receptors (LPA1–6). Results: Genetic deletion and pharmacological blockade of LPA1 inhibit differentiation and bone degradation activity of osteoclasts. Conclusion: LPA controls bone homeostasis through the activation of LPA1 expressed by osteoclasts. Significance: LPA1 is a new therapeutic target of diseases with excess bone degradation. Lysophosphatidic acid (LPA) is a natural bioactive lipid that acts through six different G protein-coupled receptors (LPA1–6) with pleiotropic activities on multiple cell types. We have previously demonstrated that LPA is necessary for successful in vitro osteoclastogenesis of bone marrow cells. Bone cells controlling bone remodeling (i.e. osteoblasts, osteoclasts, and osteocytes) express LPA1, but delineating the role of this receptor in bone remodeling is still pending. Despite Lpar1−/− mice displaying a low bone mass phenotype, we demonstrated that bone marrow cell-induced osteoclastogenesis was reduced in Lpar1−/− mice but not in Lpar2−/− and Lpar3−/− animals. Expression of LPA1 was up-regulated during osteoclastogenesis, and LPA1 antagonists (Ki16425, Debio0719, and VPC12249) inhibited osteoclast differentiation. Blocking LPA1 activity with Ki16425 inhibited expression of nuclear factor of activated T-cell cytoplasmic 1 (NFATc1) and dendritic cell-specific transmembrane protein and interfered with the fusion but not the proliferation of osteoclast precursors. Similar to wild type osteoclasts treated with Ki16425, mature Lpar1−/− osteoclasts had reduced podosome belt and sealing zone resulting in reduced mineralized matrix resorption. Additionally, LPA1 expression markedly increased in the bone of ovariectomized mice, which was blocked by bisphosphonate treatment. Conversely, systemic treatment with Debio0719 prevented ovariectomy-induced cancellous bone loss. Moreover, intravital multiphoton microscopy revealed that Debio0719 reduced the retention of CX3CR1-EGFP+ osteoclast precursors in bone by increasing their mobility in the bone marrow cavity. Overall, our results demonstrate that LPA1 is essential for in vitro and in vivo osteoclast activities. Therefore, LPA1 emerges as a new target for the treatment of diseases associated with excess bone loss.

ferent cell types, including bone cells (6 -9). These activities involve the induction of cytoskeletal rearrangement, cell motility, survival, proliferation, and differentiation (10). Blood platelets are the major source of LPA (11), but we have previously shown that LPA is produced at bone sites in the context of bone metastasis, acting on cancer cells and osteoclasts to promote progression of osteolytic lesions and bone loss (12). More recently, osteoblasts were shown to produce LPA through the involvement of purinergic P2X7 receptor, suggesting that these cells might be an additional source of LPA in bone (13). However, the role of LPA and its cognate receptors in vivo on bone cells and bone turnover is not well understood.
LPA plays a key role during bone development as revealed by a low bone mass phenotype of Lpar1 Ϫ/Ϫ mice associated with a low osteoblastic activity and growth retardation due to inhibition of chondrocyte proliferation, defects in endochondral ossification (28), and increased bone formation in Lpar4 Ϫ/Ϫ animals (33). These reports suggest that LPA 1 and LPA 4 might have opposite functions in osteoblasts during bone development. In addition to adipogenesis and glucose tolerance defects (27), Lpar1 Ϫ/Ϫ mice present developmental abnormalities in the olfactory bulb that markedly impaired pup suckling behavior (19). Deficiencies of the cerebral cortex, food intake, and metabolic regulations are prone to compromise bone homeostasis indirectly. In addition, LPA 1 is the most ubiquitous LPA receptor in mammalian tissues (35), expressed in both osteoblasts and osteoclasts (8,28). Therefore, the bone phenotype of Lpar1 Ϫ/Ϫ mice is likely to be a consequence of multiple con-straints of bone remodeling involving disruption of both osteoblast and osteoclast activity. Here, we provide the first demonstration that LPA 1 controls in vitro osteoclast differentiation regulating nuclear factor of activated T-cells cytoplasmic 1 (NFATc1) and dendritic cell-specific transmembrane protein (DC-STAMP) expressions, bone resorption establishing functionally active sealing zone, and in vivo controlling motility of osteoclast progenitors in the bone marrow cavity and osteoclast-mediated bone loss induced by ovariectomy.

EXPERIMENTAL PROCEDURES
Drugs and Reagents-Three competitive inhibitors of LPA signaling pathways dependent on LPA 1 and LPA 3 receptors VPC12249 (Coger, Morillon, France), Ki16425 (Interchim, Montluçon, France), and Debio0719 (generous gift of Dr. Murone, Debiopharm Group, Lausanne, Switzerland) were used (36 -38). No antagonist activity of Ki16425 was detected on LPA 4 , LPA 5 , and LPA 6 (37,39,40) . Animal experiments carried out at the University of Osaka were performed according to National Institutes of Health institutional guidelines and Osaka University animal experimental guidelines under approved protocols. Studies were routinely inspected by the attending veterinarian to ensure continued compliance with the proposed protocols. Mice were maintained on a 12:12-h light/dark cycle, and autoclaved water and mouse chow were provided ad libitum. Lpar1 Ϫ/Ϫ -and Lpar3 Ϫ/Ϫ -deficient mice originally generated in a C57Bl/6 background were backcrossed more than 12 times with BALB/c females (19,30). Lpar2 Ϫ/Ϫ knock-out mice generated in a C57BL/6 -129/SvJ background were described previously (29). Animals were genotyped by PCR following the procedures described previously (19,29,30). CX 3 CR1-GFP knock-in mice (41) were obtained from The Jackson Laboratory. Mutant mice were genotyped by PCR. BALB/c mice were obtained from Janvier SA (Le Genest St. Isle, France). Female BALB/c nude mice were obtained from Charles River (Kent, UK).
(saline). Mice were euthanized 30 days after OVX. To confirm the efficiency of OVX, uteri were removed and weighed at the end of the experiments. Bones from all euthanized mice were dissected out and used either for CT imaging or histological analysis. Total RNA extraction from bone was carried out on 12-week-old female BALB/c nude mice (Charles River, Kent, UK). On day 0, 18 mice per group were injected with 100 g/kg zoledronic acid or 0.1 ml of saline intraperitoneally. Three days after injection, nine mice from each group were either ovariectomized or sham-operated before sacrifice on day 10. Tibiae and femurs from three mice per group were pooled and flashfrozen in liquid nitrogen.
Two-photon Intravital Bone Tissue Imaging-CX 3 CR1-EGFP mice were subcutaneously administered either Debio0719 (20 mg/kg body weight) or vehicle (DMSO in PBS). Four hours after administration, bone tissues of these mice were visualized to measure the mobility of EGFP ϩ cells by intravital microscopy. Experiments were carried out on mouse calvaria bone tissues using a protocol modified from a previous study (42). Mice were anesthetized with isoflurane (Escain; 2.0% (v/v) vaporized in 100% (v/v) oxygen), and hair at the neck and scalp was removed with hair removal lotion (Epilat). The frontoparietal skull was exposed, and the mouse head was immobilized in a custommade stereotactic holder. The imaging system was composed of a multiphoton microscope (A1-MP; Nikon) driven by a laser (Chameleon Vision II Ti: Sapphire; Coherent) tuned to 880 nm and an upright microscope (APO, N.A. 1.1; Nikon) equipped with a 25ϫ water immersion objective (HCX APO, N.A. 1.0; Leica). The microscope was enclosed in an environmental chamber in which anesthetized mice were warmed by heated air. Fluorescent cells were detected through a bandpass emission filter at 525/50 nm (for EGFP). Vessels were visualized by injecting 70 kDa of Texas red-conjugated dextran (detected using a 629/56 nm filter) i.v. immediately before imaging. Image stacks were collected at a 5-m vertical step size at a depth of 100 -150 m below the skull bone surface. The time resolution was 1 min. Raw imaging data were processed with Imaris (Bitplane) with a Gaussian filter for noise reduction. Automatic three-dimensional object tracking with Imaris Spots was aided with manual corrections to retrieve cell spatial coordinates over time.
Three-dimensional CT-Microcomputed tomography analyses were carried out using a CT scanner Skyscan 1174 (Skyscan; Kontich, Belgium). The x-ray tube was set to a voltage of 50 kV and a current of 800 mA. A 0.5-mm aluminum filter was used to reduce beam-hardening artifacts. Samples were scanned in 70% ethanol with a voxel size of 10.3 m. For each sample, 265 section images were reconstructed with NRecon software (version 1.6.1.8, Skyscan). Three-dimensional modeling and analysis of bone volume/trabecular volume, were obtained with the CTAn (version 1.9, Skyscan) and CTVol (version 2.0, Skyscan) software. The dissected bones were then processed for histological analysis.
Bone Histology-Hind limbs were fixed, decalcified with 16% EDTA, and embedded in paraffin. Five-m tissue sections were stained with Goldner's Trichrome and processed for histomorphometric analysis. In situ detection of osteoclasts was carried out on tissue sections using the tartrate-resistant acid phospha-tase (TRAP) activity kit assay (Sigma). The resorption surface (Oc.S/BS) was calculated as the ratio of TRAP-positive trabecular bone surface (Oc.S) to the total trabecular bone surface (BS) using the computerized image analysis system Morpho-Expert (Exploranova).
Osteoclastogenesis Assay-Bone marrow cells from hind limbs of 6-week-old male mice were collected and seeded in 12-well tissue culture plates at a density of 2 ϫ 10 5 cells per well in ␣-minimal essential medium (Invitrogen) supplemented with 10% v/v FBS (Life Technology, Saint Aubain, France), 1% penicillin/streptomycin (Life Technology), 1% L-glutamine (Sigma), 25 ng/ml macrophage-CSF (R&D Systems, Lille, France), 100 ng/ml receptor-activated nuclear receptor factor B ligand (RANK-L). Culture media were then supplemented with or without Ki16425 (10 M), Debio0719 (1 M), or VPC12249 (10 M). After 6 days, mature osteoclasts were enumerated under a microscope on the basis of the number of nuclei (n Ն3) and the TRAP activity (Sigma). Results were expressed as the number of osteoclasts per well.
Podosome Belt and Sealing Zone Quantification-Osteoclasts in culture on glass coverslips were fixed in 4% paraformaldehyde, pH 7.2, for 10 min, permeabilized with 0.2% Triton X-100 in PBS for 5 min, then saturated with 2% BSA in PBS, followed by incubation with rhodamine-conjugated phalloidin to reveal F-actin. Osteoclasts in culture on bovine cortical bone slices were fixed with 4% paraformaldehyde in PBS for 10 min at 37°C and permeabilized using PBS ϩ 0.2% Triton X-100 in PBS. Then cells were incubated for 20 min with Alexa Fluor 488 phalloidin (Invitrogen) diluted 1:100. Image acquisition was performed with an inverted confocal microscope Zeiss LSM780 using GaAsP PMT and an EC-Plan-Neofluar 20ϫ/0.5 objective. The surface area was measured manually with ImageJ software.
Resorption Assay-Osteoclast efficacy to resorb mineralized matrix was carried out using apatite collagen complex (Corning Glass). Two hundred osteoclasts of each genotype were plated in each well and left to resorb for 48 h. Cells were removed by a gentle shaking in 0.001% Triton buffer lysis for 1 h; the matrix was stained with silver nitrate, and the resorption pits were detected under a light microscope. To measure the resorbed surface area, a series of 99 micrographs were imaged using a DMI system, and the mosaic of each well was reconstituted using a plugin of ImageJ software. The pit surface per single osteoclast was quantified with ImageJ, and data were reported as resorbed mineralized surfaces in m 2 /Oc. Alternatively, an equivalent number of pre-osteoclasts isolated from WT BALB/c mice were cultured for 4 days. Then the culture media was supplemented with Ki16425 (10 M) for an additional 3 days. The culture was then stopped, and pit surfaces were determined as described above.
Reverse Transcription and RT-PCR-Total RNAs from osteoclast cultures were extracted using Nucleospin RNAII kit (Macherey-Nagel, Hoerdt, France) and from powdered whole bone with TRIzol (Invitrogen). Complementary DNA from osteoclasts and bones were synthesized by reverse transcription using iScript cDNA synthesis kit (Bio-Rad) and Superscript III (Invitrogen), respectively. Expression of target genes was quantified by real time quantitative RT-PCR on an Eppendorf Mas-tercycler RealPlex (Eppendorf, Sartrouville, France) using the SYBR Green PCR kit (VWR International, Fontenay-Sous-Bois, France) and sets of specific primers (Table 1). Quantifications were normalized to corresponding RNA L32 values and expressed as relative expression using the 2(Ϫ⌬⌬C T ) method (43).
Statistical Analysis-Differences between groups were determined by one-or two-way ANOVA followed by Bonferroni post-test using GraphPad Prism Version 5.0c software. Single comparisons were carried out using two-sided unpaired t test. p Ͻ 0.05 was considered significant.

Expression of LPA Receptors during Osteoclastogenesis-Os-
teoclastogenesis depends on a sequential gene activation program controlling the proliferation of osteoclast precursors that successively fuse to form large polykaryons that eventually differentiate into fully active multinucleated osteoclasts (44). Primary culture of bone marrow cells (BMCs) extracted from WT BALB/c mice in the presence of MCSF and RANK-L was performed. In agreement with previous reports showing high expression levels of LPA 6 (P2Y5) in hematopoietic cells and during monocyte-to-macrophage differentiation (45), we found by real time PCR (Fig. 1A) that LPA 6 had the highest expression levels in BMCs (data not shown) and at day 1 of osteoclastogenesis that decreased to high and stable values from days 2 to 5. LPA 3 was not detected in osteoclast progenitors and during osteoclastogenesis, whereas LPA 1 expression prevailed among all other LPA receptors at both early (days 1-3) and late (days 3-5) stages of osteoclast differentiation. Also, LPA 1 , LPA 2 , LPA 4 , and LPA 5 were up-regulated from days 3 to 4 concomitantly with the expression of osteoclast differentiation markers (ATP6v0d2, integrin ␤3 (Beta3), calcitonin receptor (CTR), and cathepsin K (CTSK)) (Fig. 1B). These data suggest that LPA 1 and LPA 6 potentially regulate the behavior of both osteoclast precursors and mature osteoclasts, whereas the activity of other LPA receptors might only play a role during the latter stages of osteoclast differentiation.
Inhibiting LPA 1 Activity Impairs Osteoclastogenesis-To determine the role of LPA receptors in osteoclastogenesis, primary cultures of BMCs collected from Lpar1 Ϫ/Ϫ , Lpar2 Ϫ/Ϫ , and Lpar3 Ϫ/Ϫ mice and WT littermates were performed in the presence of MCSF and RANK-L. After 6 days, we found that the number of tartrate-resistant acid phosphatase (TRAP)-positive multinucleated cells (mature osteoclasts) from Lpar1 Ϫ/Ϫ mice was significantly lower by 62% to that obtained with WT cells (316 Ϯ 82.2 and 831 Ϯ 78.6, p Ͻ 0.001) (Fig. 1, C and D). In marked contrast, the numbers of mature osteoclasts generated from Lpar2 Ϫ/Ϫ and Lpar3 Ϫ/Ϫ mice were identical to those obtained from WT cells with matched genetic background (Fig.  1, C and D). Inhibition of osteoclast differentiation in Lpar1 Ϫ/Ϫ cells was confirmed at the transcriptomic level, because the expression of osteoclast markers was markedly decreased compared with WT cells (Fig. 1E). Altogether these results suggest the implication of LPA 1 in osteoclast differentiation, although LPA 2 or LPA 3 appear not to be involved. Hematopoietic osteoclast precursors, monocytes, do not express LPA 3 (46,47). Because our previous observations showed no LPA 3 expression at all stages of osteoclastogenesis (Fig. 1A), we next considered cells from the osteoclast lineage as LPA 3 -null cells. Thus, to further investigate the role of LPA 1 in osteoclast differentiation, we decided to use a pharmacological approach to inhibit LPA 1 activity. We treated BMC precursors with nonspecific LPA 1 / Ϫ4.071 12.61 L32-R CTGCTCTTTCTACAATGGC LPA 3 antagonists (Ki16425, Debio0719, and VPC12249). Each LPA 1 inhibitor reduced MCSF/RANK-L-induced osteoclastogenesis of WT BMC precursors by 55% (VPC12249), 68% (Ki16425), or 83% (Debio0719) ( Fig. 2A). Interestingly, the extent of inhibition caused by LPA 1 antagonists was not statistically significantly different to that observed with Lpar1 Ϫ/Ϫ BMCs (62%) ( Fig. 2A). As presented above (Fig. 1A), LPA 2 was also up-regulated during osteoclastogenesis; therefore, we investigated whether blocking of both LPA 1 and LPA 2 activity would inhibit osteoclastogenesis to a further extent. We found that supplementing Lpar2 Ϫ/Ϫ BMC culture media with Ki16425 inhibited the differentiation of osteoclasts by 62% (Fig. 2B), which was not significantly different from the inhibition of osteoclast formation caused by Ki16425 treatment of WT BMCs (69%, Fig. 2A). This result indicates that blocking of both LPA 1 and LPA 2 activity does not inhibit osteoclastogenesis to a greater extent than that achieved by blocking LPA 1 alone. We next investigated whether LPA 1 plays a role in a particular phase of osteoclastogenesis (proliferation, fusion, and differentiation). We found that a short treatment period with Ki16425 during the proliferation phase of osteoclast precursors (days 1-3 and days 2 and 3), or during the late phase of mature osteoclast differentiation (days 5-7), did not significantly affect the final number of osteoclasts (Fig. 3, A and B). In contrast, when added during the fusion phase of osteoclast precursors (days 3-4, 3-5, and 4 -5), Ki16425 significantly inhibited the formation of osteoclasts by 48% at days 3-5 (p Ͻ 0.001) and days 4 and 5 (p Ͻ 0.001) and by 58% at days 3 and 4 (p Ͻ 0.001)) compared with vehicle-treated BMCs (Fig. 3, A and B). Additionally, the levels of inhibition were not statistically significantly different from the 69% inhibition of osteoclastogenesis resulting from addition of Ki16425 from days 1 to 7. When added between days 3 and 5, Ki16425 also significantly induced an accumulation of osteoclasts with a low number of nuclei (3 Յ n Ͻ5) compared with vehicle-treated cells (73 versus 40%; p Ͻ 0.05) and a decrease in the number of osteoclasts with a high number of nuclei; 5 Յ n Ͻ10 and n Ͼ10 (25% versus 50%; p Ͻ 0.001) and (2% versus 10%; p Ͻ 0.05), respectively. Also, a significant accumulation by 69% (p Ͻ 0.05) of osteoclasts with low number of nuclei (3 Յ n Ͻ5) was observed in Lpar1 Ϫ/Ϫ BMC cultures. Altogether, these results indicated that blocking LPA 1 activity inhibited the fusion of osteoclast precursors (Fig.  3C). In agreement with previous observations on Lpar1 Ϫ/Ϫ BMCs, inhibiting LPA 1 activity with Ki16425 significantly decreased the expression of markers of osteoclast activity (ATP6v0d2, CTR, Beta3, and CTSK) (Fig. 3D). In addition, compared with vehicle, Ki16425 had no effect on the expression of LPA 1 , the transcription factor c-Fos, and the MCSF receptor cFMS that are expressed at the early stage of osteoclast differentiation, whereas expressions of the pro-survival NFATc1 and the fusion protein DC-STAMP that are implicated in fusion and differentiation were significantly decreased by 50% (p ϭ 0.003) and 70% (p ϭ 0.021), respectively (Fig. 3D).
Inhibiting LPA 1 Activity Impairs the Cytoskeletal Organization in Mature Osteoclasts-Osteoclastic bone resorption activity is dependent on a highly specialized organization of the  actin cytoskeleton (48). Because induction of cytoskeletal rearrangements is a hallmark of LPA activity (49), we investigated the role of LPA 1 in osteoclast actin cytoskeleton organization. Mature osteoclasts from BMCs of Lpar1 Ϫ/Ϫ and WT animals were generated on glass coverslips and stained with rhodamine-phalloidin. The number of osteoclasts with a complete podosome belt was determined by epifluorescence microscopy. We found that the proportion of Lpar1 Ϫ/Ϫ osteoclasts with a complete podosome belt was significantly lower than that of WT osteoclasts (35% versus 58%; p Ͻ 0.001) (Fig. 4A). This experiment was repeated with WT cells seeded on bovine cortical bone slices. Under these conditions, osteoclasts form a complex podosome belt-like structure rich in F-actin named the sealing zone, which is required for successful osteoclastic bone resorption (48). We observed that within 30 min, Ki16425 (10 M) reduced the percentage of osteoclasts with detectable sealing zone by 48.6% as a consequence of a decrease in the percentage of osteoclasts with both complete (8% versus 25.6%; p Ͻ 0.001) and discontinued sealing zones (48% versus 74.4%; p Ͻ 0.05) (Fig. 4B). Moreover, Ki16425 decreased the area covered by the sealing zone in each osteoclast by 40.2% (Fig. 4B). These results suggest that inhibition of LPA 1 activity might reduce the capacity of osteoclasts to adhere to the bone matrix, impairing the formation of a fully efficient sealing zone and hence inhibit bone resorption.
Inhibiting LPA 1 Activity Impairs Osteoclast Bone Resorption Activity-To analyze the role of LPA 1 in osteoclast bone resorption activity, we first generated WT and Lpar1 Ϫ/Ϫ osteoclasts from BMCs on conventional culture plates, detached the multinucleated cells, and seeded the same number of osteoclasts on mineralized tissue culture plates (apatite collagen complex). We then imaged and quantified the resulting lacunae by phase contrast microscopy. The mean resorbed mineral surfaces generated by Lpar1 Ϫ/Ϫ osteoclasts was significantly lower compared with that of WT osteoclasts (1190 Ϯ 222 m 2 versus 329 Ϯ 81 m 2 ; p ϭ 0.0013) (Fig. 5A). Treatment of WT osteoclasts with Ki16425 (10 M) reduced the mean resorbed mineral surface area by 60% (2413 Ϯ 402 m 2 versus 946 Ϯ 118 m 2 ; p ϭ 0.005) (Fig. 5B). Additionally, 72 h of treatment of mature WT osteoclasts generated on dentin slices with Ki16425 caused a 26% decrease in the density of osteoclasts (542 Ϯ 23.1 versus 400 Ϯ 31.8 per cm 2 ; p ϭ 0.01) with noticeable overspreading of Ki16425-treated osteoclasts attached to the dentin (Fig. 5C). These results indicate that LPA 1 is essential for optimal bone resorption and that blocking LPA 1 activity might reduce the tight adhesion of osteoclasts to the mineralized matrix required for resorption activity.
Inhibition of LPA 1 Activity Prevents Osteoclast-induced Cancellous Bone Loss-Decreased levels of estrogen are responsible for osteoporosis mediated by increased osteoclast activation and excessive bone loss in postmenopausal women. Osteoporosis can be mimicked in animals by OVX (50). Four weeks after surgical intervention, we collected distal femur and tibia metaphyses of ovariectomized and sham mice, extracted total bone RNAs, and measured expression of LPA receptors by RTquantitative PCR. Expression of LPA 6 was predominant in the bone of sham-operated mice, whereas LPA 1 , LPA 3 , PLA 4 , and LPA 5 were found at lower levels, and LPA 2 was barely detectable (Fig. 6). Following OVX, the expression of LPA 6 was not significantly modified, whereas LPA 1 was dramatically increased (ϫ11.3-fold) and LPA 4 to a lesser extent (ϫ1.7-fold), although other receptors were down-regulated compared with sham-operated animals: LPA 3 (ϫ7.1-fold) and LPA 5 (ϫ35.5-fold) (Fig. 6A). As expected, osteoclast activation markers (CTR and CTSK) were up-regulated in bone following OVX due in part to an increase in osteoclast differentiation (Fig. 6B). Remarkably, along with reducing the expression of osteoclast markers, the anti-resorptive agent zoledronic acid inhibited up-regulation of LPA 1 in the bone of ovariectomized animals (Fig. 6B). These results indicate that increased osteoclastic bone resorption in vivo is associated with up-regulation of LPA 1 .
We next investigated the contribution of LPA 1 during the process of hyper-resorptive bone loss in vivo. Nine-week-old female mice were ovariectomized and treated daily for 4 weeks with vehicle or the Ki16425 structurally related LPA 1 antagonist Debio0719 (20 mg/kg) (38) or the anti-resorptive agent risedronate (4 mg/kg). We used CT imaging analyses of the distal femurs to assess histomorphometric parameters of the trabecular bone. As expected, risedronate treatment inhibited trabecular bone loss, resulting in a bone volume/trabecular volume value not significantly different from that of sham-operated animals (23% Ϯ 2.0 versus 26% Ϯ 3.5) (Fig. 7A). Although the bone volume/trabecular volume value in animals treated with Debio0719 was lower than in sham mice, it was significantly higher than in OVX animals treated with vehicle (19% Ϯ 3.1 versus 14% Ϯ 4.1; p Ͻ 0.05) (Fig. 7, A and B). Overall, these data indicate that systemic treatment of animals with the LPA 1 antagonist Debio0719 prevented by 60% trabecular bone loss induced by OVX.
Serum bone resorption markers are known to gradually increase during the first 2 weeks following OVX and then rapidly return to base line the week after (51). Therefore, 4 weeks after OVX, osteoclast activity in the bone of animals treated with Debio0719 was analyzed by histological examinations of tissue sections of tibia epiphyses stained for TRAP activity. As expected, risedronate treatment completely blocked the OVXinduced increase in osteoclast number (Oc.S/BS) (Fig. 7B). There was no significant difference in Oc.S/BS ratio in animals treated with Debio0719 compared with the sham group (28.2 Ϯ 4.5% versus 28.7 Ϯ 4.8%). Altogether, these results indicate that systemic treatment with Debio0719 inhibited bone resorption through a mechanism involving impairment of the differentiation and/or the recruitment of osteoclast precursors at the bone surface.
Inhibiting LPA 1 Activity Impairs the Dynamic Behavior of Circulating Osteoclast Precursors-To investigate the role of LPA 1 on osteoclast precursor migration in vivo, we performed intravital two-photon imaging of calvaria from CX 3 CR1-EGFP ϩ mice, examining the migratory behavior of fluorescent monocytoid cells containing osteoclast precursors resident in the marrow spaces. First, we measured the expression of mRNAs encoding different LPA receptors in CX 3 CR1 ϩ osteoclast precursor monocytes. CX 3 CR1 ϩ cells from BMCs were collected using a cell sorter and analyzed by RT-PCR. In agreement with the results presented above, we found that LPA 1 , LPA 2 , LPA 4 , LPA 5 , and LPA 6 were expressed in CX 3 CR1 ϩ osteoclast precursors, whereas LPA 3 expression was not detectable ( Fig. 8A). Next we examined the effect of Debio0719 on the mobility of osteoclast precursors in live bone tissues by using intravital multiphoton microscopy (42). CX 3 CR1-EGFP mice were administered with either Debio0719 (20 mg/kg) or vehicle s.c., and bone tissues were visualized to measure the mobility of EGFP ϩ cells 4 h later (Fig. 8B and supplemental video 1 and  video 2). The mean tracking velocity of CX 3 CR1-EGFP ϩ circulating osteoclast precursors was significantly increased in mice treated with Debio0719 (Fig. 8C) suggesting decreased recruitment of osteoclasts at the bone surface because of increased osteoclast precursors mobility in the marrow. Therefore, LPA 1 may contribute to retaining circulating osteoclast precursors in bone marrow cavity.

DISCUSSION
Bone remodeling controls bone mass through a complex regulation of the balance between bone formation mediated by osteoblasts and bone resorption through osteoclast activity (1). Here, we demonstrate that Lpar1 Ϫ/Ϫ animals also display altered osteoclast differentiation and activity, indicating that LPA/LPA 1 pathway plays a central role in bone remodeling and the interplay between osteoblasts and osteoclasts.
LPA is a serum-borne factor required for MCSF/RANK-Linduced osteoclastogenesis of BMCs (52). Compared with previous reports describing the role of LPA on osteoclast function, our in vitro experiments were carried out in the presence of serum, avoiding potential LPA-independent serum stress signals on osteoclast primary culture cells. Incubating mature osteoclasts for 12 h in serum-free medium induces apoptosis that is rescued by addition of LPA, demonstrating that LPA is an osteoclast survival factor (8). However, the role of LPA in bone resorption is quite controversial and not well characterized to date. In vitro, LPA is shown to increase the bone resorption activity of rabbit osteoclasts but to cause a significant reduction of pits formed of mouse osteoclasts (8,53). In our study, we found that supplementation of serum-containing medium with Ki16425 inhibited complete sealing zone formation and decreased resorption activity in mouse osteoclasts, indicating that through activation of LPA 1 , LPA promoted bone resorption. Sealing zones are dynamic structures delineating compartments in which protons and proteases are secreted by osteoclasts to dissolve and degrade the mineralized matrix (48). Sealing zones are found in myeloid cells and are composed of cytoskeletal structures, named podosomes, that relate to stress fibers and focal adhesions in mesenchymal cells sharing similar regulation via Rho GTPases (54). LPA 1 was primarily identified for its capacity to induce cytoskeleton rearrangement and cell contraction through a mechanism involving G␣ 12/13 and downstream activation of RhoA (20,55). Therefore, acting through RhoA, LPA 1 may likely control complete sealing zone formation, resulting in decreased resorption activity as observed on WT osteoclasts treated with Ki16425. Mature Lpar1 Ϫ/Ϫ osteoclasts showed a significant reduction of podosome belt formation and a decrease in the extent of mineral resorption, demonstrating that in the presence of serum LPA 1 controls osteoclastic bone resorption.
Increased cell survival, proliferation, and motility are hallmarks of LPA effects on a large range of cell types (56). Cells commonly co-express multiple LPA receptors, and defining the ones that are involved in specific functions remains challenging. Transcripts encoding LPA 1 , LPA 2 , LPA 4 , and LPA 5 are present in mature osteoclasts (8). We found expression of LPA 1 at all stages of osteoclast differentiation that was further up- regulated together with LPA 2 , LPA 4 , and LPA 5 concomitantly with the fusion and differentiation phases of osteoclast precursors. Lpar2 Ϫ/Ϫ BMCs had not altered osteoclastogenesis capacity, suggesting that LPA 2 may not contribute to bone remodeling via osteoclast activity. The role of LPA 4 , LPA 5 , and LPA 6 in osteoclast function remains to be established. LPA 1 transduces the mitogenic activity of LPA in fibroblasts and cancer cells (57)(58)(59)(60). Proliferation of osteoclast precursors is a prerequisite for successful osteoclastogenesis (44). Treatment of BMCs with Ki16425 during the proliferation phase of osteoclast progeni-  tors had no effect on the final number of mature osteoclasts, indicating that LPA 1 activity is not required for proliferation of osteoclast precursors. In addition, expression of c-Fos and c-Fms was not affected by Ki16425 treatment. RANK-L activates expression of c-Fos (61), and high concentrations of MCSF acting through c-Fms restores the expression of c-Fos in Beta3 Ϫ/Ϫ osteoclasts (62). Therefore, although LPA up-regulates c-Fos in various cells (63), its expression in osteoclasts was likely sustained by MCSF and RANK-L abundantly present in BMC culture media independently of LPA action. LPA activates NFATc1 in osteoclasts and promotes survival (8). Decreased expression of NFATc1 in BMCs treated with Ki16425 might decrease osteoclast precursor survival and differentiation and together with decreased expression of the fusion protein DC-STAMP could be responsible for the observed decrease in the fusion of osteoclast precursors. LPA 1 is the most ubiquitous LPA receptor (64). Expression of LPA 1 in bone cells is well documented (65). However, regulation of LPA 1 expression in bone is not known. OVX or continuous exposure to elevated levels of PTH results in an increase in bone resorption activity, hypercalcemia, and bone loss (66,67), whereas subcutaneous intermittent PTH administration increases bone turnover and osteoblast activity resulting in net anabolic effects on bone and reduced risk of fractures (68). Lpar1 expression was not reported in the list of genes that were modulated in the bones of animals treated with intermittent administration of PTH (69), whereas Lpar1 was up-regulated by 1.5-fold in the bone of rats treated with continuous infusion of PTH(1-34) (70). We found Lpar1 was up-regulated by 4-fold in the bones of ovariectomized mice. These results strongly support the link between LPA 1 expression and osteoclast activity in vivo. Aminobisphosphonates (i.e. zoledronic acid and risedronate) are powerful inhibitors of osteoclast activity and thereby are the standard of care in patients with diseases associated with excessive bone resorption, such as osteoporosis, multiple myeloma, and bone metastases (71). Here, we showed that zoledronic acid inhibited up-regulation of LPA 1 in bones of ovariectomized mice, confirming that modulation of LPA 1 expression was linked to mature osteoclast activity. Blocking LPA 1 function in vivo with the LPA 1 antagonist Ki16425 inhibits the progression of osteolytic bone lesions in mouse models (72). The Ki16425 structurally related LPA 1 antagonist, Debio0719 (38), partially prevented bone loss induced by OVX compared with complete bone loss prevention with risedronate. Remarkably, Debio0719 was equally potent as risedronate at inhibiting the increase of osteoclast number at the bone surface of ovariectomized mice, but the underlying mechanisms may be different. Nitrogen-containing bisphosphonates are powerful inducers of apoptosis of osteoclasts due to inhibition of farnesyl pyrophosphate synthase activity (73). In vivo, resorbing osteoclasts undergo apoptosis due to uptake of bisphosphonate released from the bone surface. We demonstrate that antagonists of LPA 1 decreased in vitro osteoclast differentiation and in vivo the number of osteoclasts lining on the surface of trabecular bone of ovariectomized mice. Osteoclast precursors are nonadherent circulating hematopoietic cells (44). In vivo osteoclastogenesis depends on the recruitment and attachment of osteoclast progenitors to the bone surface initiating their fusion and subsequent differentiation. Our in vitro results demonstrate that blocking the LPA 1 pathway inhibits the expression of osteoclast markers, including the adhesion molecule ␤3 integrin, and the fusion of osteoclast precursors, and interferes with the cytoskeletal organization altering adhesion of mature osteoclasts to mineralized surfaces. Debio0719 reduces the number of osteoclasts attached to the bone surface in ovariectomized mice. The mode of action of Debio0719 in vivo was not completely determined but may include the inhibition of osteoclast differentiation and a decrease in the attachment of osteoclast precursors to the bone matrix. A reduced residence time of the nonadherent osteoclast progenitors at the bone surface combined with an increased motility of CX 3 CR1-GFP osteoclast precursors would decrease the number of fusion events and therefore the number of osteoclasts at the bone surface. Subsequently, the number of nonadherent osteoclast precursors floating in bone marrow would be increased. This hypothesis was supported by the increase in CX 3 CR1-GFP ϩ nonadherent osteoclast progenitor mean tracking velocity in the bone marrow of animals treated with Debio0719. However, the contribution of other LPA receptors (i.e. LPA 4 -6 ) to this phenomenon remains to be established. Nevertheless, bone mass recovery of ovariectomized mice treated with Debio0719 (50 mg/kg) was not complete. This may be a result of an inhibitory effect of the treatment on bone formation as the use of LPA 1 antagonists also inhibits osteoblast differentiation in vitro (7,74). Therefore, it is conceivable that systemic treatment with Debio0719 may inhibit overall bone remodeling by affecting both osteoclast and osteoblast function, thus decreasing bone resorption while limiting bone formation.
Overall, our results demonstrate that LPA 1 controls bone remodeling by regulating directly osteoclastogenesis and motility of osteoclast precursors in vivo and bone resorption. We have shown previously that LPA 1 is a potential target in the context of osteolytic bone metastases (72). A recent work showed that blocking LPA 1 inhibits the development of arthritis via cellular infiltration, Th17 differentiation, and osteoclastogenesis (75). Our study extends the field of applications of anti-LPA 1 therapy for patients suffering from bone fragility disorders due to excessive bone resorption.