Impaired Osteoblast Function in GPRC6A Null Mice

GPRC6A is a widely expressed orphan G protein–coupled receptor that senses extracellular amino acids, osteocalcin, and divalent cations in vitro. GPRC6A null (GPRC6A−/−) mice exhibit multiple metabolic abnormalities including osteopenia. To investigate whether the osseous abnormalities are a direct function of GPRC6A in osteoblasts, we examined the function of primary osteoblasts and bone marrow stromal cell cultures (BMSCs) in GPRC6A−/− mice. We confirmed that GPRC6A−/− mice exhibited a decrease in bone mineral density (BMD) associated with reduced expression of osteocalcin, ALP, osteoprotegerin, and Runx2-II transcripts in bone. Osteoblasts and BMSCs derived from GPRC6A−/− mice exhibited an attenuated response to extracellular calcium-stimulated extracellular signal-related kinase (ERK) activation, diminished alkaline phosphatase (ALP) expression, and impaired mineralization ex vivo. In addition, siRNA-mediated knockdown of GPRC6A in MC3T3 osteoblasts also resulted in a reduction in extracellular calcium-stimulated ERK activity. To explore the potential relevance of GPRC6A function in humans, we looked for an association between GPRC6A gene polymorphisms and BMD in a sample of 1000 unrelated American Caucasians. We found that GPRC6A gene polymorphisms were significantly associated with human spine BMD. These data indicate that GRPC6A directly participates in the regulation of osteoblast-mediated bone mineralization and may mediate the anabolic effects of extracellular amino acids, osteocalcin, and divalent cations in bone. © 2010 American Society for Bone and Mineral Research.

GPRC6A is expressed in osteoblasts, (4,16) but its function in bone is not clear. Preliminary characterization of the skeleton of GPRC6A À/À mice indicates that loss of this receptor is associated with decreased bone mineral density (BMD) and impaired mineralization of bone, (6) but the presence of other abnormalities confounded the ability to ascertain the direct and indirect effects of GPRC6A on skeletal function. To determine if GPRC6A is a potential candidate for the purported extracellular calciumsensing receptor in osteoblasts, we performed a more detailed assessment of the skeletal phenotype of GPRC6A À/À mice, examined the function of primary osteoblasts and bone marrow stromal cells derived from these mutant mice ex vivo, and evaluated whether polymorphisms in GPRC6A are associated with skeletal abnormalities in humans by a gene association analysis.

GPRC6A knockout mice
The GPRC6A-deficient mouse model was created by replacing exon 2 of the GPRC6A gene with the hygromycin resistance gene, as described previously. (6) Mice were maintained and used in accordance with recommendations of theNational Research Council's (1985) Guide for the Care and Use of Laboratory Animals (DHHS Publication NIH 86-23, Institute on Laboratory Animal Resources, Rockville, MD) and following guidelines established by the University of Kansas Medical Center Institutional Animal Care and Use Committee.
For quantitative real-time RT-PCR assessment of bone marker expression, we isolated and reverse transcribed 2.0 mg of total RNA from the long bones of 8-week-old mice as described previously. (17) PCR reactions contained 100 ng of template (cDNA or RNA), 300 nM each of forward and reverse primer, and 1Â iQ SYBR Green Supermix (Bio-Rad, Hercules, CA, USA) in 50 mL. Samples were amplified for 40 cycles in an iCycler iQ Real-Time PCR Detection System (Bio-Rad) with an initial melt at 958C for 10 minutes, followed by 40 cycles of 958C for 15 seconds and 608C for 1 minute. PCR product accumulation was monitored at multiple points during each cycle by measuring the increase in fluorescence caused by the binding of SybrGreen I to dsDNA. The threshold cycle (C t ) of tested-gene product from the indicated genotype was normalized to the C t for cyclophilin A. Dissociation analysis was used to confirm the presence of a single transcript and lack of primer-dimer amplification in all PCR reactions.
PIXImus bone densitometer analysis, X-ray, and bone histology PIXImus bone densitometer analysis, X-ray, and bone histology Bone mineral density (BMD) values of whole skeletons and femurs were assessed at 6, 8, 12, and 16 weeks of age using a PIXImus bone densitometer (Lunar Corp., Madison, WI, USA) as described previously. (18) Femurs were dissected free of muscle and X-rayed with a Faxitron model MX-20 specimen radiography system (Faxitron X-Ray Corp., Lincolnshire, IL, USA). In addition, length and width (measured at 50% of the femur length) were assessed in isolated femora. Skeletons of mice were prelabeled twice with calcein (Sigma C-0875, 30 mg/g of body weight; Sigmal Chemical Company, St. Louis, MO, USA) by intraperitoneal injection at 8 and 3 days prior to sacrifice. Tibiae and femora were removed from 8-and 16-week-old mice, fixed in 70% ethanol, prestained in Villanueva stain, and processed for methyl methacrylate embedding. Then 10-mm Villanueva-prestained sections were evaluated under fluorescent light, as reported previously by our laboratory. (18) Micro-computed tomographic (mCT) analysis The distal femoral metaphyses were scanned using a mCT 40 device (Scanco Medical AG, Wayne, PA, USA); 167 slices of the metaphysis under the growth plate, constituting 1.0 mm in length, were selected. The 3D images were generated using the following values for a gauss filter (s ¼ 0.8, support 1) and a threshold of 275. A 3D image analysis was performed to determine bone volume (BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), and trabecular separation (Tb.Sp). Cortical bone was measured on the midshaft region of cortical bone in 50 slices of the diaphysis, constituting 0.3 mm in length. The mean cortical thickness (Ct.Th) was determined at eight different points on the cortical slice.

Primary bone marrow stromal cells and osteoblastic cells culture
The femora and tibiae from 8-week-old wild-type and GPRC6A À/À mice were dissected, the ends of the bones were cut, and marrow was flushed out with 2 mL of ice-cold a modified essential medium (a-MEM) containing 10% fetal bovine serum (FBS) by using a needle and syringe. A suspension of bone marrow cells was obtained by repeated aspiration of the cell preparation through a 22G needle, and nucleated cells were counted with a hemocytometer. Cells were seeded into 6-well plates at a density of 3 Â 10 7 cells/mL and cultured for 3 days in a-MEM supplemented with 10% FBS, 100 kU/L of sodium penicillin G, and 100 mg/L of streptomycin sulfate in a humidified incubator with 5% CO 2 and 95% air at a temperature of 378C. On day 3, all nonadherent cells were removed with the first medium change, and then the adherent cells [representing bone marrow-derived mesenchymal stem cells (BMSCs)] were grown for additional periods of up 3 days in the same medium. After overnight quiescence, the cells were stimulated for 5 minutes with calcium, NPS-R568, and arginine at the concentrations indicated.
We used modifications of a nonenzymatic method for obtaining the osteoblastic cell lines. (19) A fragment of the frontal and/or parietal bone from the single calvarium was aseptically removed from a 3-to 7-day-old mouse. Suture lines and endosteum were dissected away, and the bone fragment was placed in a culture dish. One or two metal strips were positioned on the endocranial surface and incubated for 3 to 4 days in DMEM nutrient mixture F-12 (Invitrogen, Carlsbad, CA, USA) containing 10% (v/v) FBS, 100 kU/L of sodium penicillin G, and 100 mg/L of streptomycin until the outgrowth of osteoblasts. The metal strips were removed, and the cells were allowed to grow until approximately 60% confluent. The cells were subcultured and propagated by incubation in a-MEM (Invitrogen) containing 10% FBS, 100 kU/L of sodium penicillin G, 100 mg/L of streptomycin, and 50 mg/mL ascorbic acid in a humidified atmosphere of 5% CO 2 and 95% air at 378C.

Agonist stimulation and Western blotting
Agonist stimulation was performed in quiescent cells. Quiescence was achieved in subconfluent cultures by removing the medium and washing with Hank's balanced salt solution (Invitrogen) to remove residual serum, followed by incubation for an additional 24 hours in serum-free medium. After agonist treatment at the specified concentrations and durations, cells were washed twice with ice-cold PBS and scraped into 250 mL of lysis buffer (25 mM HEPES, pH 7.2, 5 mM MgCl 2 , 5 mM EDTA, 1% Triton X-100, and 0.02 tablet/mL of protease inhibitor mixture). Equal amounts of lysates were subjected to 10% SDS-PAGE and transferred onto Immun-Blot PVDF (Bio-Rad). For phospho-ERK, the phospho-ERK1/ 2 levels were determined by immunoblotting using anti-phospho-ERK1/2 mitogen-activated protein kinase antibody (Cell Signaling Technology Inc., Beverly, MA, USA). To confirm that variations in the amount of ERK did not contribute to stimulated ERK activity, we used an anti-ERK1/2 mitogen-activated protein kinase antibody (Cell Signaling Technology, Inc.) to measure ERK levels for control protein loading. The signals were quantified by densitometric scanning (FotoDyne, Hartland, WI, USA).

Mineralization assay
The formation of in vitro mineralization nodules was determined by alizarin red S histochemical staining. (15) The 14-day cultured cells in the a-MEM containing 10% FBS, 100 kU/L of sodium penicillin G, 100 mg/L of streptomycin sulfate, 50 mg/mL of ascorbic acid, and 4 mM of arginine were fixed for 24 hours in a solution of 10% Formalin, methanol, and water (1:1:1.5); the fixative was removed; and the fixed cells and matrices were stained for 15 minutes with a 2% (w/v) solution of alizarin red S at pH 4.0. The stained samples were washed three times with water and then air-dried.
Immunohistochemistry for osteocalcin (OC) and measurement of aklaline phosphatase (ALP) activity Anti-osteocalcin antibody was obtained from Dr Larry Fisher (NIDCR, National Institutes of Health). After deparaffinization and rehydration, the bone sections were immersed in 3% hydrogen peroxide to quench endogenous peroxidase and further digested with 1 mg/mL trypsin for 30 minutes at 378C. Sections then were blocked with 1% bovine serum albumin at room temperature for 2 hours. The primary antibodies were added to the sections and incubated overnight at 48C. After washing, the sections were coated with biotinylated second antibody (Vector Laboratories Burlingame, CA, USA) at a dilution of 1:200 and then incubated at room temperature for 60 minutes. The sections were washed again and incubated with the ABC reagent (Vector Laboratories, Burlingame, CA, USA) at room temperature for 60 minutes. The 3,3'-diaminobenzidine substrate was used to visualize immunoreaction sites. Sections were counterstained with hematoxylin and mounted on glass slides. Negative controls were obtained by substituting the primary antibody with normal IgG.
Alkaline phosphatase (ALP) activity was detected by an enzyme-substrate assay in frozen sections of bone. Briefly, fresh nondecalcified spines from 8-week-old wild-type and GPRC6A knockout mice were dissected and embedded with Optimum Cutting Temperature (OCT, Sakura Finetek, USA, Inc., Torrance, CA, USA) medium. ALP activity was detected in 12-mm-thick cryostat sections at alkaline pH by adding ALP substrate 4-nitro blue tetrazolium (NBT) and 5-bromo-4-chloro-3-indolyl phosphate (BCIP) directly on the slides according to the manufacturer's instructions (Roche Applied Science, Indianapolis, IN, USA). Endogenous ALP activity was visualized as a purple-blue color reaction with methyl green (nuclei) counterstain.
In situ hybridization for osteocalcin mRNA expression A mouse osteocalcin cDNA plasmid, EcoRI/Pst I subclone inserted into pSP65, was obtained from Dr Stephen Harris (University of Texas Health Science Center, San Antonio, TX, USA). A digoxigenin (DIG)-labeled cRNA probe was prepared by using the RNA Labeling Kit (Roche Applied Science). To make the antisense probe, a partial mouse osteocalcin cDNA (0.5 kb) was cleaved by Hind III and labeled with Sp6 polymerase. Sections of the tibia from a 8-week-old mouse were dewaxed, rehydrated, fixed with 4% paraformaldehyde, treated with 2% glycine and proteinase K, acetylated using an acetic anhydride-TEA solution, and then hybridized with a DIG-labeled probe. After washing the sections twice with 50% formamide, 5% saline sodium citrate, and 5% SDS for 30 minutes at 708C and once with 50% formamide and 2% saline sodium citrate for 30 minutes at 658C, we incubated them with antibody to digoxigenin conjugated with alkaline phosphatase and then with nitro blue etrazolium/ 4-bromo-5-chloro indolylphosphate, which yields a purple-blue color. We also counterstained the sections with methyl green (nuclei). The hybridization signals were photographed with a Nikon E800 microscope (Nikon, Melville, NY, USA) and a MagnaFire camera (Optronics, Muskogee, OK, USA).

Serum and urine biochemical measurements
Serum was collected using a retro-orbital bleeding technique. For urine samples collection, mice were placed in metabolic cages (Hatteras Instrument Cary, NC, USA), and urine was collected for 24 hours. The urine volume was measured before storage at À708C. Serum and urinary calcium was measured by the colorimetric cresolphthalein-binding method, and phosphorus was measured by the phosphomolybdate-ascorbic acid method. (18) Serum tartrate-resistant acid phosphatase (TRACP) was assayed with the ELISA-based SBA Sciences mouseTRAP assay (Immutopics, Inc., San Clemente, CA, USA). Serum parathyroid hormone (PTH) and 1,25dihydroxyvitamin D were measured with an Immunodiagnostic Systems device (Immunodiagnostic Systems, Ltd., Scottsdale, AZ, USA). Serum Fgf23 levels were measured using an FGF-23 ELISA kit (Kainos Laboratories, Inc., Tokyo, Japan) following the manufacturer's protocol. Creatinine was measured by the colorimetric alkaline picrate method (Sigma Kit 555). Urinary protein and deoxypridinolme (Dpd) were measured by Bio-Rad and Metra Biosystems (Hercules, San Diego, CA, USA) devices, respectively. siRNA transfection Two siRNAs (siRNA.m1638: CCAACACAGCTGTTGCTAT and siR-NA.m2553: GCAGAAGACTAACACCAAA) specific for marine GPRC6A (GenBank Accession Number NM_153071), identified using the ''siRNA Target Finder'' at the Web site (www.ambion.com/techlib/misc/siRNA_finder.html), were used as a template for synthesizing siRNA. Double-stranded siRNA was synthesized from DNA oligonucleotides using the Silencer siRNA construction kit (Ambion, Austin, TX, USA) according to manufacturer's instructions.
For transfection, MC3T3 cells were plated at a density of 1.0 Â 10 4 cells per well in a 6-well dish and grown overnight. Cells were transiently transfected with 25 nM of siRNA using 4 mL of Lipofectamine 2000 (Invitrogen) for 4 hours, after which growth medium was replaced. Then, 48 hours after transfection, cells were quiescence overnight and then stimulated by 10 mM of calcium for 5 minutes and lysed for Western blot analysis. Scrambled siRNA (Select Negative Control 1 siRNA, Ambion) was used as negative control.

Study populations and genotyping
The 1000 unrelated American Caucasians of European origin studied were identified from our established and expanding genetic repertoire at UMKC. Signed informed-consent documents were obtained from all study participants before they entered the study. The population genetic research on GPRC6A was approved by the Institutional Review Board. The inclusion and exclusion criteria are the same as those described elsewhere. (20) The basic characteristics of the studied subjects are listed in Supplemental Table 1. Briefly, the sample includes 251 elderly men (age range at recruitment 52.9 to 87.9 years), 250 younger men (age range at recruitment 19.1 to 49.9 years), 250 postmenopausal women (age range at recruitment 54.3 to 80.6 years), and 249 premenopausal women (age range at recruitment 19.2 to 53.2 years). Area BMD (g/cm 2 ) at the spine and hip region was measured by dual-energy X-ray absorptiometry (DXA) with Hologic QDR 4500W densitometers (Hologic, Inc., Bedford, MA, USA). The scanners were calibrated daily. The coefficient of variation (CV) values of the DXA measurements were about 1.98% and 1.87% for spine and hip BMD, respectively.
The nine single-nucleotide polymorphisms (SNPs) of interest in vicinity of or inside the GPRC6A gene were genotyped successfully with the Affymetrix GeneChip human mapping 500K array set (Affymetrix, Inc., Santa Clara, CA, USA) as part of the large-scale genotyping project. The genotyping was performed at the Vanderbilt Microarray Shared Resource at Vanderbilt University Medical Center (Nashville, TN, USA) using the standard protocol recommended by the manufacturer. The genotyping procedure is detailed elsewhere. (20) The average per-SNP call rate of these nine SNPs across the 1000 subjects is 99.4% (97.2% to 99.9%).

Statistical analysis
We evaluated differences between groups by one-way analysis of variance. All values are expressed as means AE SEM. All computations were performed using the Statgraphic Statistical Graphics System (STSC, Inc., Rockville, MD, USA).
The Hardy-Weinberg equilibrium (HWE) test and linkage disequilibrium (LD) plotting of the interested SNPs were performed using the Haploview 4.1 (Broad Institute, Cambridge, MA) . (21) The default option ''confidence intervals'' in Haploview 4.1 was used to define LD blocks. (22) Before performing the marker-trait association analysis, parameters such as age, age 2 , sex, age/age 2 -by-sex interaction, height, and weight were tested for their associations with BMD at the spine and hip; then the significant ( p .05) terms were included as covariates to adjust the raw BMD values for subsequent analyses. All marker-trait association analyses were performed using association tests implemented in HelixTree 5.3.1 (Golden Helix, Bozeman, MT, USA). To circumvent the problem of false-positive results incurred by multiple testing, the SNPSpD program developed by Nyholt was used to establish the experiment-wise significance threshold needed to keep the type I error rate less than 5%, considering the redundancy of information provided by SNPs that were in LD. (23)

Skeletal imaging
By X-ray analysis, we observed no significantly gross developmental abnormalities in the skeleton (Fig. 1A). Consistent with prior reports, (6) BMD as assessed by dual-energy X-ray absorptrometry (DEXA) was significantly lower in 16-week-old GPRC6A À/À mice compared with wild-type mice (Fig. 1B). We observed no difference in femur length (Fig. 1C, D), but there was a significant reduction in femur width between GPRC6A À/À and wild-type mice (Fig. 1C, E). In addition, we found that BMD was reduced by mCT in both the metaphyseal area, which predominately consists of trabecular bone, and the midshaft region, which is composed of cortical bone (Fig. 1F). These differences were not due to effects of a mixed genetic background because these animals were crossed onto C57BL/ 6 mice for more than six generations. There were no demonstrable changes, however, in bone structural parameters, including bone volume (BV/TV) and cortical thickness (Ct.Th), between GPRC6A À/À and wild-type mice, as assessed by mCT, suggesting that the decreased BMD was not due to structural changes in bone architecture.

Biomarkers/immunohistochemistry/gene expression
We found no evidence for GPRC6A effects on osteoclast function. In this regard, the osteoclastic marker tartrate-resistant acid phosphatase (TRACP) was not significantly different between wildtype and GPRC6A À/À mice (Table 1), consistent with our prior findings that neither urinary Dpd/creatinine ratio nor serum TRACP levels were different between GPR6CA À/À and GPR6CA þ/þ mice. (6) To better understand the effect of GPRC6A on osteoblasts, we examined gene expression profiles of whole-bone samples from GPRC6A À/À mice. Assessment of expression of osteoblast markers in bone from 16-week-old GPRC6A À/À mice revealed reductions in osteocalcin, ALP, osteoprotegerin, and Runx2-II message levels compared with wild-type mice by real-time RT-PCR (Table 1), consistent with a reduction in mineral apposition rates, as reported previously. (6) We also confirmed by in situ hybridization and immohistochemical staining that osteocalcin expression ( Fig. 2A-D) and ALP activity (Fig. 2E, F) were reduced in bone from 16-week-old GPRC6A À/À mice. The chondrocyte marker ColII and adipocyte markers aP2 and Lp1 were not significantly different between wild-type and GPRC6A À/À mice (Table 1). Impaired extracellular calcium response in BMSCs and calvarial-derived osteoblastic cells cultures from GPRC6A À/À mice To explore whether the observed skeletal abnormalities are due to a primary osteoblastic abnormality owing to loss of GPRC6A in Fig. 1. Characterization of the bone phenotype of GPRC6A À/À mice. (A) X-ray appearance for GPRC6A þ/þ and GPRC6A À/À mice at 16 weeks of age. (B) Comparison of the total BMD analysis by PIXI mus analysis in GPRC6A þ/þ and GPRC6A À/À mice at ages ranging from 6 to 16 weeks. (C) Abnormalities in femur bone of GPRC6A À/À mice by X-ray. (D, E) Comparison of femur length (D) and femur width (E) in GPRC6A þ/þ and GPRC6A À/À mice at 16 weeks of age. ( F) Comparison of the structure and BMD of the femur as assessed by mCT scanning in 16-week-old GPRC6A þ/þ and GPRC6A À/À mice. Data represent the mean AE SEM from 6 to 10 mice in each group. Ã Significant difference from GPRC6A þ/þ and GPRC6A À/À mice at p < .05. Denotes significant difference between GPRC6A þ/þ and GPRC6A À/À mice at p < .05. these cells, we evaluated the response to calcium, calcimimetic, and arginine of cultured of bone marrow stromal cell cultures (BMSCs) and calavarial osteoblasts obtained from wild-type and GPRC6A À/À mice (Fig. 3). In both BMSCs and primary osteoblasts, we observed a reduced ability of extracellular calcium to stimulate ERK activity in GPRC6A À/À mice compared with wild-type cells (approximately 50% reduction; Fig. 3A, B). In addition, a receptor allosteric modulator, the calcimimetic NPS-568, and the amino acid receptor ligand L-arginine both had an attenuated ERK activation in BMSCs and primary osteoblasts obtained from GPRC6A À/À mice compared with wild-type cells (Figs. 3C, D).
To further evaluate osteoblast dysfunction in GPRC6A À/À mice, we examined the impact of loss of GPRC6A on the capacity of cultured osteoblasts and BMSCs to undergo differentiation in vitro, as assessed by culture duration-dependent changes in alkaline phosphatase expression and activity. We found that wild-type osteoblasts and BMSC cultures increased alkaline phosphatase expression and activity during differentiation, but this increase was attenuated in GPRC6A À/À cells (Fig. 4A, B). These results are consistent with the gene expression data by real-time RT-PCR (Table 1) and immohistochemical staining (Fig. 2E, F) in intact bone. Moreover, the addition of arginine increased mineralization in wild-type BMSC cultures but not in GPRC6A À/À cells (Fig. 4C). These data indicate that the lack of GPRC6A impairs the ability of osteoblasts to sense calcium and amino acids as well as their ability to undergo differentiation and form a mineralized extracellular matrix.

Response of osteoblastic cell line MC3T3 to extracellular calcium through GPRC6A
To confirm the role of GPRC6A as a relevant calcium-and amino acid-sensing receptor, we examined the MC3T3 clonal osteoblastic cell line, which is known to express a putative calciumsensing receptor. (4,24) We examined the function of GPRC6A by siRNA-mediated knockdown in MC3T3 osteoblasts. We transfected MC3T3 cells with either GPRC6A siRNA.m2563 or siRNA.m1638. Mock-transfected cells and transfected cells with a random negative control siRNA plasmid were used as the control. Transfected MC3T3 cells GPRC6A siRNA.m2563 and siRNA.m1638 successfully downregulated the levels of mRNA expression of GPRC6A in MC3T3 cells compared with mocktransfected cells and negative-control siRNA-transfected cells by RT-PCR analysis (Fig. 5A). In addition, the activation of phospho-ERK stimulated by extracellular calcium in MC3T3 cells was significantly decreased by both transfected GPRC6A siR-NA.m2563 and siRNA.m1638 (reduced approximately 50% and 97%, respectively; Fig. 5B). These results suggest that endogenous GPRC6A accounts for the effects of extracellular calcium in stimulating ERK phosphorylation.
Association of GPRC6A polymorphisms with spine and hip BMD Next, we performed a gene association analysis to test whether GPRC6A gene polymorphisms are associated with the variation in human BMD. (25) Nine SNPs located from 21 kb downstream to and osteoblasts from GPRC6A À/À mice displayed impaired responses to extracellular calcium-mediated ERK activation. (C) ERK activation in response to the calcimimetic NPS-R568 also was impaired in BMSCs derived from GPRC6A À/À mice. (D) Osteoblasts derived from GPRC6A À/À mice also showed impaired ERK activation in response to the amino acid arginine. ERK phosphosphorylation was assessed by Western blot analysis using an anti-phospho-ERK antibody. Data represent three to four independent experiments. Fig. 4. Characterization of temporal maturational sequence in osteoblasts or BMSCs from GPRC6A þ/þ and GPRC6A À/À mice. (A) RT-PCR of alkaline ALP from 4-and 10-day cultured cells derived from 8-week-old GPRC6A þ/þ and GPRC6A À/À calvaria. (B) ALP activity. The GPRC6A À/À BMSCs had significantly lower ALP activity at days 10 and 14 of culture compared with age-matched wild-type mice. (C) Quantification of mineralization. Alizarin red S was extracted with 10% cetylpyridinium chloride and quantified as described under ''Materials and Methods.'' Data represent the mean AE SEM from three to four independent experiments. 42 kb upstream of the GPRC6A gene were tested for their association with hip and spine BMD variation in the human subjects studied. All nine SNPs are in HWE ( p > .01) with minor allele frequency (MAF) greater than 5%. The basic information on these SNPs is shown in Table 2.
For the SNP rs686708 (A/G), which showed the strongest evidence of association with spine BMD, its allele A was associated with lower spine BMD values in the studied subjects. The subjects with the AA or AG genotypes had significant lower spine BMD values than those with the GG genotype (raw spine BMD values were 1.016, 1.032, and 1.055 g/cm 2 for the AA, AG, and GG genotypes, p ¼ .029), showing an allele dose effect of 0.02 g/cm 2 lower spine BMD value per copy of the A allele. For the SNP rs571296 (A/G), its minor allele G was associated with lower spine BMD values. Subjects with the GG or AG genotypes had significant lower spine BMD values than those with the AA genotype (raw spine BMD values were 1.061, 1.037, and 1.019 g/cm 2 for AA, AG, and GG genotypes, p ¼ .032), showing an allele dose effect of 0.021 g/cm 2 lower spine BMD value per copy of the G allele. Linkage disequilibrium (LD) analysis based on the  For the SNP rs686708, the adjusted BMD values for the three genotypes AA, AG, and GG in the samples studied were as follows: À0.015, 0.000, and 0.022 ( p ¼ .023); For the SNP rs571296, the adjusted BMD values for the three genotypes AA, AG, and GG in the samples studied were as follows: 0.025, 0.005, and À0.012 ( p ¼ .048).
HapMap data set with Haploview software showed the existence of extended and strong intermarker LD in the genomic region between SNP rs571296 and SNP rs339321 (Supplemental Fig. S). SNP rs686708 was outside the LD block and linked very weakly with other SNPs presented (Supplemental Fig. 1).

Discussion
GPRC6A is a widely distributed member of the amino acid-and calcium-sensing receptor family that appears to function as an nutrient-sensing anabolic receptor. (6) In this study we have demonstrated direct effects of GPRC6A on regulating osteoblast function. The predominant effect of GPRC6A deficiency in vivo was to impair bone mineralization, which was associated with a reduction in osteoblast gene expression markers without changing bone-resorptive markers in vivo. We show that these changes are due, at least in part, to direct effects of the loss of GPRC6A in osteoblasts. In this regard, a primary GPRC6Adependent effect on osteoblasts is suggested by the abnormal differentiation and attenuated response to calcium and arginine stimulation of osteoblasts derived from GPRC6A null mice in vitro. A direct role of GPRC6A in osteoblasts also was confirmed by siRNA-mediated knockdown of this receptor in MC3T3-E1 osteoblasts that was associated with reduction in calciumstimulated ERK activation. The human relevance of these observations is supported by the association between SNPs rs686708 and rs571296 and reductions in spinal BMD in Caucasians, raising the possibility that GPRC6A polymorphisms may contribute to human osteopenia.
Several limitations of this study prevent more definitive conclusions. First, since GPRC6A null mice represent a global knockout of this receptor, our in vivo findings are still confounded by the potential actions of concomitant alterations in the ratio of testosterone and estrogens known to be present in these mice or to other undefined secondary consequences of the generalized loss of GPRC6A. (6) The absence of increased bone resorption in GPRC6A null mice, however, along with the direct in vitro effects of GPRC6 on osteoblast function in two different models (GPRC6A null and GPRC6A knockdown), suggests that direct effects from receptor loss rather than alterations in sex hormones may account for the osteopenia in GPRC6A À/À mice. Indeed, testosterone deficiency typically leads to increased osteoclast-mediated bone resorption, (26) which was not observed in GPRC6A null mice. While the attenuation of calcium and amino acid response in isolated osteoblasts from GPRC6A null mice supports a direct role, further studies that compare the effects of testosterone replacement with administration of GPRC6A ligands (e.g., calcium, strontium, amino acids, and osteoclacin) to restore bone mass in GPRC6A À/À mice will be needed to determine the relative contribution of secondary alterations in sex hormones and primary loss of calcium-sensing receptor responses to the observed bone phenotype in GPRC6A null mice. Second, with regard to the role of GPRC6A in humans, SNPs rs686708 and rs571296 are not in the coding region of GPRC6A and therefore are not definitely the actual diseasecausing variants. Fine mapping of SNPs tightly linked to SNPs rs686708 and rs571296 and functional studies will be required to identify the causal variants in the GPRC6A region responsible for human osteopenia.
Our data also support the presence of more than one calciumsensing mechanism in osteoblasts. Indeed, the full complement of physiologically relevant receptors mediating the changes in amino acids, calcium, and osteocalcin released by osteoclasticmediated bone resorption continues to be explored. Recent data support a role for the amino acid-and calcium-sensing receptor CASR in regulating osteoblast function. (27,28) Conditional deletion of the 7-TM domain of CASR using Col1-2.3-Cre results in a severe skeletal phenotype. (28) In addition, some studies have implicated CASR in the differentiation of osteoblasts and growth plate chondrocytes. (28,29) On the other hand, the original CASR knockout mouse model (which lacks exon 5) fails to display a bone phenotype when hyperparathyroidism is corrected by either performing a ''molecular parathyroidectomy'' or ablating PTH receptor signaling. (30,31) Also, the anabolic bone effects of strontium ranelate on osteoblast replication and survival are independent of CASR. (32) The explanation for disparity in bone phenotype between the conditional and global CASR À/À lacking exon 5 knockout mouse models is not clear. One possibility is that global CASR À/À mice lacking exon 5 are hypomorphic owing to the persistent function of an alternatively spliced exon 5-deleted CASR in bone and cartilage. Although CASR null osteoblasts retain their calcium-sensing capabilities, so far no signal-transduction activity of the transfected alternatively spliced exon 5-deleted CASR has been identified. (33,34) Another theoretical possibility is that the actions of cre-recombinase in the conditional CASR model creates a secreted extracellular domain of CASR that acts as a dominant-negative ''decoy'' receptor to disrupt CASR function in nearby tissues. Alternatively, the expression of GPRC6A in bone and osteoblasts and the resulting bone phenotype raise the possibility that GPRC6A is a candidate for another osteoblastic calcium-sensing receptor (4,15) that is distinct from CASR (15) and which could account for some of these disparate observations. The residual calcium-sensing functions in GPRC6A null mice, after silencing of GPRC6A in MC3T3 cells, leaves open the possibility of more than one calcium-sensing receptor in osteoblasts.
Nevertheless, the wide expression of GPRC6A also raises the possibility that GPRC6A could have a physiologic function in coordinating the responses of bone with other organ systems to changing nutritional cues. For example, there may be a physiologic need to coordinate renal calcium excretion and osteoblast-mediated bone formation. Theoretically, a primary decrease in bone formation and decreased buffering capacity for calcium, with consequent increased urinary expression of dietary calcium, could account for the relationship between osteopenia and hypercalciuria in GPRC6A À/À mice. Conversely, activation of GPRC6A would be predicted to stimulate osteoblast-mediated bone calcium accretion and renal calcium conservation to meet the need for bone mineralization. There are some enigmatic clinical disorders with features similar to those of GPRC6A À/À mice that support the possibility of coordinated effects between bone formation and renal conservation of calcium. In this regard, a subset of male patients with idiopathic osteoporosis described by Zerwekh and Pak (35) has the combined features of decreased osteoblast-mediated bone formation and hypercalciuria without evidence of hypogonadism, secondary hyperparathyroidism, or abnormal vitamin D levels. Some patients with what appears to be primary hypercalciuria and nephrolithiasis also have concomitant low bone density and low bone turnover (36) without increased PTH levels that would be expected from a sole defect in renal calcium handling. There are also examples of primary bone defects that are associated with hypercalciuria without increased PTH. These include osteogenesis imperfecta type 1 caused by Col1A1 and Col1A2 mutations, infantile hypophoshatasia caused by inactivating mutations of ALP, and McCune-Albright syndrome caused by activating mutations of GNAS1. (8) It will be interesting to determine if gene polymorphisms of GPRC6A are associated with osteopenic and hypercalciuric clinical disorders.
These findings also expand the repertoire of GPCRs that regulate the functional activity and provides greater insights into the function of Gai in bone. In this regard, GPRC6A is predominately a Gai-coupled receptor. (4) Compared with other well-characterized anabolic receptors, such as the PTH receptor, which is coupled to Gas and Gaq-dependent signaling pathways (37) that regulate proliferation and differentiation of osteoblasts, these studies suggest that Gai-dependent pathways play a distinct role in regulating osteoblast gene expression, mainly affecting the mineralization of bone. The classica effect of Gai activation in inhibiting adenylyl cyclase activity might be expected to oppose signaling of Gas-coupled GPCRs in osteoblasts. (38) There is little available information on the role of osteoblast signaling by Gai, but the pertussis toxindependent proliferative actions of fluoride and strontium on osteoblasts in vitro, (4,39) Gai-coupled apelin receptor stimulation of osteoblast proliferation in vitro. (40) development of osteoporosis in Gai-coupled CB2 cannabinoid receptor knockout mice, (41) as well as our findings in GPRC6A null mice, suggest that Gai signaling is important. Potential anabolic actions of Gaicoupled receptors, including GPRC6A, might be mediated through activation of mitogen-activated protein kinase (MAPK) pathways. (42) In contrast to our studies, Wellendroph and colleagues (16) generated a GPRC6A null mouse by targeting exon VI of GPRC6A, which encodes the seven transmembrane domain and C-terminal tail. Analysis of these mice failed to identify any skeletal abnormalities in 13-week-old mice. At present, we have no explanation for these disparate findings, but similar to the differences observed between existing CASR knockout models described earlier, (28,43) this may represent another example where disrupting the extracellular domain results in no expression of the GPCR, whereas disrupting the transmembrane domain may lead to partial translation of the extracellular domain and functional effects. Regardless, there are other instances of multiple laboratories independently targeting the same GPCR and failing to find similar phenotypes that have been attributed to a variety of different mechanisms. (44) In summary, we have shown that GPRC6A has a direct function in osteoblasts. The ligand profile of GPRC6A, which includes extracellular calcium, calcimimetics, amino acids, and osteocalcin, (1,2,4,45) along with the complex phenotype of GPRC6A null mice, suggests that GPRC6A may represent an anabolic receptor that responds to a variety of nutritional and hormonal signals and may serve to coordinate the functions of multiple organs, including bone, to changes in the local and systemic concentrations of these ligands.

Disclosures
LDQ serves as a consultant for Amgen, Shire, Cytochroma, Novartis, and Osteometrics and receives research support from Amgen, Genzyme, Servier, and VasoGenix. All the other authors state that they have no conflicts of interest.