The G-Protein-Coupled Receptor GPR103 Regulates Bone Formation

Abstract

GPR103 is a G-protein-coupled receptor with reported expression in brain, heart, kidney, adrenal gland, retina, and testis. It encodes a 455-amino-acid protein homologous to neuropeptide FF2, neuropeptide Y2, and galanin GalR1 receptors. Its natural ligand was recently identified as 26RFa, a novel human RF-amide-related peptide with orexigenic activity. To identify the function of GPR103, we generated GPR103-deficient mice. Homozygous mutant mice were viable and fertile. Their body weight was undistinguishable from that of their wild-type littermates. Histological analysis revealed that GPR103^−/− mice exhibited a thinned osteochondral growth plate, a thickening of trabecular branches, and a reduction in osteoclast number, suggestive of an early arrest of osteochondral bone formation. Microcomputed tomography confirmed the reduction in trabecular bone and connective tissue densities in GPR103 knockout animals. Whole-body radiography followed by morphometric analysis revealed a kyphosis in mutant animals. Reverse transcription-PCR analysis showed that GPR103 was expressed in human skull, mouse spine, and several osteoblast cell lines. Dexamethasone, a known inhibitor of osteoblast growth and inducer of osteoblast differentiation, inhibited GPR103 expression in human osteoblast primary cultures. Altogether, these results suggest that osteopenia in GPR103^−/− mice may be mediated directly by the loss of GPR103 expression in bone.

G-protein-coupled receptors (GPCRs, or GPRs) contain seven transmembrane domains and transduce extracellular signals through heterotrimeric G proteins (21, 26, 31). The family of GPCRs includes receptors for signals and small molecules such as light, peptidic hormones, neurotransmitters, amino acids, lipids, prostanoids, and odorants. There are ∼1,000 genes encoding such receptors in the human genome, and these receptors regulate almost all physiological processes, such as blood pressure regulation, inflammatory response, and feeding behavior, to name a few. Importantly, these molecules have been successful targets for the development of new therapeutic agents for a variety of therapeutic indications: nearly 350 new approved drugs over the past 15 years targeted GPCRs.

Bone is constantly regenerated through continuous formation and resorption in a process called bone remodeling. This physiological process occurs throughout adult life to maintain a constant bone mass. Bone resorption is primarily regulated by the number and activity of osteoclasts, a cell type derived from the monocyte/macrophage lineage, while the osteochondral lineage gives rise to osteoblasts and chondroblasts and drives bone and cartilage formation. Osteoporosis is a common health problem in postmenopausal women and patients taking long-term glucocorticoid treatments (15), characterized by a proportional increase of bone resorption relative to bone formation and leading to an increased incidence of fracture, particularly in hips and spines.

The molecular mechanisms underlying osteoclastogenesis have been largely elucidated (for a review, see reference 2). They involved growth and differentiation factors such as macrophage colony-stimulating factor; adhesion molecules such as osteopontin and its receptor, α_vβ₃ integrin; the activation of osteoclasts by a tumor necrosis factor-like molecule secreted by osteoblasts, RANKL (12), its cell surface receptor, RANK, and osteoprotegerin, which is a competitive decoy receptor for RANKL; and finally, the regulation of osteoclast apoptosis by transforming growth factor β.

In contrast, our knowledge of the mechanisms underlying bone formation is limited to a few genes. The transcription factor Cbfa1 is a master regulator of the osteoblast lineage (10, 11). Transforming growth factor β is also known to prevent differentiation of osteoblasts via the SMAD3 signaling pathway (4, 8).

More recently, it has become clear that central mechanisms also play a role in mediating bone formation and remodeling. Insight into this aspect of bone biology was first suggested by the observation that obesity protects from osteoporosis (15). Several molecules involved in orexigenic behavior, such as leptin (9, 33, 34), CART (13), and neuropeptide Y (NPY) (1, 9) and melanin-concentrating hormone (MCH) (3) receptors have been shown to regulate bone formation, most likely through a hypothalamic relay. The nature of the sympathetic signal from the hypothalamus to the bone is at least in part mediated by β2-adrenergic receptors (13, 35).

Recently, Lee et al. (25) identified a novel G-protein-coupled receptor, GPR103, from a hypothalamus cDNA library encoding a deduced 455-amino-acid protein that shares between 35% and 38% sequence identity in the transmembrane regions with various peptide receptors, including neuropeptide FF2, neuropeptide Y2, and galanin GalR1 receptors. Northern blot analysis of human brain regions revealed widespread expression in the thalamus, hypothalamus, and pituitary gland. GPR103, also referred to as SP9155, was also found to be expressed in heart, kidney, retina, and testis (20). The ligand for GPR103 was recently identified as a 26-amino-acid RF-amide peptide, also called 26RFa or QRFP (20). This peptide was reported to induce the concentration of plasma aldosterone(17) and to have orexigenic activity (36).

We generated GPR103-deficient mice to identify the physiological functions of GPR103 and surprisingly found that mice deficient in GPR103 suffered from osteopenia. Microcomputed tomography (microCT) bone imaging revealed a reduction in trabecular bone density in both males and females. The decreased bone density was more prominent in females, and in addition, female mice exhibited the characteristic kyphotic “hump” of osteoporotic patients. Histological analysis revealed that GPR103^−/− female mice suffered from a thinning of the osteochondral growth plate, a thickening of trabecular branches, and a reduction in osteoclast number, which is suggestive of an early arrest of osteochondral bone formation. Ovariectomized GPR103^−/− female mice had an increased level of bone loss compared to that of wild-type females.

We also report that GPR103 is expressed in bone and in various osteoblast cell lines and that its expression in human osteoblast primary cultures is modulated by glucocorticoids. Altogether, these results show that GPR103 regulates bone formation in mammals and suggest that this effect might at least in part be mediated directly by GPR103 activity in the osteochondral growth plate.

MATERIALS AND METHODS

Animals.

GPR103 knockout mice were produced by Lexicon Genetics Incorporated (The Woodlands, TX). The entire exon 1 from GPR103 was replaced by a 5.3-kb LacZ/Neo cassette in the GPR103 targeting vector, along with a total of approximately 7 kb of homology: a ∼5-kb 5′ arm and a ∼2-kb 3′ arm. The targeting vector was introduced into embryonic stem cells derived from the 129SvEv^Brd mouse strain. Targeting events were confirmed by Southern blot analysis of embryonic stem (ES) cell genomic DNA after digestion with BglII by using a probe located outside of the arms of homology. As expected, the targeted clone gave two bands of 7.7 and 12.9 kb, respectively, while the untransfected clone showed a band at 7.7 kb only. Targeted ES cell clones were injected into host blastocysts, and resulting chimeric mice were bred to C57BL/6J albino mice to generate F₁ heterozygotes. F₁ heterozygotes were backcrossed to C57BL/6J mice for one generation. The resulting N₁F₁ heterozygous males and females were intercrossed to generate homozygous mice. GPR103 homozygous mutant mice were obtained according to Mendelian transmission ratios. They appeared normal and reproduced normally.

Genotyping was performed by PCR using three oligonucleotides, Neo 3a (5′-GCA GCG CAT CGC CTT CTA TC-3′), TL12-22 (5′-GCT ACT AAC ATA CTA GAT TAC AG-3′), and TL12-56 (5′-CGT TCT AAC GCA TGA GTG ATG-3′) in the following sequence of temperatures: 2 min at 94°C followed by 35 cycles of 1 min at 94°C, 1 min at 60°C, 1 min at 72°C, and a final 10-min extension period at 72°C. The reaction tube was then stored at 4°C until separated on an agarose gel. The wild-type band amplified between TL12-22 and TL12-56 had an expected size of 760 bp. The mutant band amplified between TL12-22 and Neo3a had an expected size of 870 bp.

Whole-body microradiographs.

Mice were euthanized by CO₂ inhalation. An incision was made in their abdomen, and the carcasses were fixed in formalin (10% formaldehyde solution). Contact radiographs were taken from animal carcasses (Skeletech Inc.). The entire animal body was placed on a piece of film and exposed in a Faxitron machine at 30 kV for 20 s. The angle of curvature of the spine and other morphometric values were determined from the contact radiographs. Four lines were drawn on the contact microradiograph: the first one was drawn from the beginning of the thoracic vertebra to the height of vertebral column, the second one was drawn from the height of the vertebral column to the end of the lumbar vertebra, and the third one was made by connecting the bases of the two lines; the fourth line was drawn at the maximum height of the spine curvature. The primary angle was determined at the intersection of the first two lines. The height and length of the base were measured, and its ratio was calculated.

MicroCT analysis of LV4.

MicroCT analysis of lumbar vertebra 4 (LV4) was performed at Skeletech Inc., Seattle, WA. Four wild-type and four knockout mice of each gender were used. A Scanco μCT-20 computed tomography scanner (Scanco Medical, Basserdorf, Switzerland) was used for nondestructive high-resolution computed tomography imaging of murine vertebrae. Each vertebra was imaged with identity blinded, at an 18-μm voxel resolution. Using Scanco software (μCT-20, version 3.1), trabecular bone structure measurements were made for a core cylinder centered through the vertebral body. Using standard filtering and thresholding algorithms (22), the following parameters were assessed for each cylindrical volume: total volume, bone volume, fraction of bone volume, trabecular number, trabecular thickness, trabecular separation, connectivity density, and structure model index.

Histology.

The spinal column and distal half of femurs of three wild-type and five knockout GPR103 females were collected for histological analysis. In the case of the spinal column, the thoracic and first few lumbar vertebrae were removed from the animal carcass and trimmed of remaining tissue. The excised tissues were decalcified using 10% EDTA, pH 7.0, until they were radiotranslucent based on radiography. Decalcified tissues were dehydrated in graded alcohol (70%, 80%, 90%, and 100%), cleared through xylene, and embedded in paraffin. Paraffin blocks were oriented longitudinally, and a sagittal cut was used to bisect the vertebral column and femur. Sections (4 μm) were stained histochemically for tartrate-resistant acid phosphatase plus methyl green:thionin counterstaining.

Expression analysis: in situ hybridization, reverse transcription (RT)-PCR, and quantitative PCR.

In situ hybridization was performed as described previously by Chuang et al. (7). Briefly, ³³P-labeled mouse GPR103 sense and antisense riboprobes spanning the entire open reading frame were hybridized to paraformaldehyde-fixed, paraffin-embedded mouse brain sections. In situ signals were visualized by autoradiography and light microscopy.

Quantitative RT-PCR was performed on human total RNA from various tissues (Biochain or Clontech) using the Taqman Gold kit (Applied Biosystems) and primer and probe set 5′-GGGCTTTCACAATGCTAG-3′, 5′-GGAAGTCATATTTGATCTC-3′, and 5′-6-carboxyfluorescein-TGGCAGTCATCGTAGGAT-6-carboxytetramethylrhodamine-3′. The GPR103 expression level was quantitatively analyzed using the ABI Prism 7700 sequence detector (Applied Biosystems).

For mouse tissues, quantitative RT-PCR was performed on RNA either isolated from GPR103^+/+ and GPR103^−/− mice or obtained from Clontech with the following primer and probe set: 5′-AAGGCAACTCAAGCGACA-3′, 5′-CAAATGATATTAGCTATGA-3′, and 5′-6-carboxyfluorescein-CTGAAAACTCTACTTTCGG-6-carboxytetramethylrhodamine-3′.

For RT-PCR, total RNA from human osteoblasts and other human and mouse cell lines was isolated using TRIzol reagent (Gibco BRL). Primary human osteoblasts (Clonetics) from a 39-year-old donor were treated with different concentrations of dexamethasone for 24 h followed by total RNA extraction. RNA samples were reverse transcribed into cDNAs using a SuperScript First-Strand Synthesis kit (Invitrogen) with random primers. The human and mouse GPR103 fragment (318 bp) was amplified using primers 5′-TAGGATCACCCATGTGGCACGT-3′ and 5′-AAGAGAGCCACCACTGTCACCATC-3′ by performing 35 cycles of PCR (94°C for 30 s, 60°C for 30 s, and 72°C for 60 s). The PCR products were analyzed on a 2% agarose gel stained with ethidium bromide.

DEXA scanning and ovariectomy.

Female mice were anesthetized by intraperitoneal injection of tribromoethanol (250 mg/kg) in phosphate-buffered saline and then placed in a prone position on the platform of the PIXImus Densitometer (Lunar Inc.) for a dual-energy X-ray absorptiometry (DEXA) scan. Using Lunar PIXImus software, the bone mineral density (BMD), fat composition (percent fat), and total tissue mass were determined in the regions of interest (i.e., whole body, vertebrae, and both femurs).

Ovariectomy was performed using standard procedures (37) on six wild-type and four knockout 24-week-old female mice. Six weeks after ovariectomy, a second DEXA scan was performed to access subsequent BMD loss.

RESULTS

GPR103 knockout female mice suffer from kyphosis.

Homozygous GPR103 knockout males and females are viable and fertile and appear normal based on direct home cage observation. After anesthesia or euthanasia of GPR103 mice for routine tissue collection and analysis, most GPR103 females appeared to have a pronounced arched back (Fig. 1A), approximately 80% from more than 15 females observed. Occasional observations of arched back in GPR103 males have been made but were infrequent (1 in more than 10 animal observed). Whole-body microradiography confirmed that the dorsal “hump” of GPR103^−/− animals resulted from an increased backward curvature of the spine, called kyphosis. (Fig. 1B and C). To quantitate this morphological observation, four lines were drawn on the contact microradiograph taken from a lateral view as shown in Fig. 1D and E and as described in Materials and Methods. The height and length of the base of the resulting triangle were measured, and the ratio between the two was calculated. Wild-type animals displayed a primary angle (the angle formed between the lumbar and thoracic vertebral lines) of approximately 90°, whereas the primary angle formed in GPR103^−/− mice was reduced by greater than 20%. The reduction in the primary angle in GPR103^−/− mice was also reflected by a reduction in the size of the base in these animals, although the height did not obviously differ between wild-type and GPR103^−/− mice. As would be expected from the aforementioned data, the height/base ratio was greater in knockout animals (Table 1). Hence, we concluded that knockout animals had a more severe kyphosis than wild-type animals.

FIG. 1.

Whole-body microradiographs of GPR103 wild-type and knockout mice. (A) Direct photography of a lateral view of wild-type (left) and knockout (right) mice after euthanasia. (B to E) X-ray microradiography of a lateral view of wild-type (B and D) and knockout (C and E) mice. In panels D and E, four lines were drawn to illustrate how the spine curvature measurements were collected: the first one (dotted line) was drawn from the beginning of the thoracic vertebra to the height of vertebral column, the second one (dotted line) was drawn from the height of the vertebral column to the end of the lumbar vertebra, and the third line was made by connecting the bases of the two lines (dashed line); the fourth line (solid line) was drawn at the maximum height of the spine curvature. The primary angle was determined at the intersection of the first two lines. The height and length of the base were measured. Lines are representative indicators for reference. Actual measurements were performed on copies of the original microradiographs.

TABLE 1.

Spine curvature measurements^a

Sample	Transgenic genotype	Primary angle (°)	Ht (mm)	Base (mm)	Ht/base ratio
1	WT	99	14	42	0.33
2	WT	94	14	43	0.33
3	WT	87	11	34	0.32
Mean		93	13	40	0.33
SEM		3.5	1.0	2.8	0.0
4	KO	70	15	31	0.48
5	KO	65	15	32	0.47
6	KO	75	14	32	0.44
Mean		70	15	32	0.46
SEM		2.9	0.3	0.3	0.0

^aThe primary angle and the height and length of the base measured on contact microradiograph, as shown in Fig. 1, of three wild-type (WT) and three knockout (KO) animals were used to calculate the height/base ratio.

The dorsal view of microradiographs showed that GPR103^−/−mice did not suffer from lordosis and that the defect was along one axis only (data not shown). No other skeletal abnormalities were observed.

GPR103 knockout mice have reduced trabecular bone density.

Histological analysis of thoracic vertebrae with a few lumbar vertebral segments on the caudal end was performed to determine whether morphological abnormalities were reflected in the cellular organization of the skeleton. All histological features of a vertebral segment were present in both groups; however, numerous differences were observed between knockout and wild-type groups. The vertebral column from wild-type animals had a flatter curvature than those from knockout animals (Fig. 2A and B), although the general length and height of the vertebral segments were not obviously different between these two groups. The nucleus pulposus from the knockout group was wider, causing the disks to occupy larger spaces. The end plates of the wild-type group (Fig. 2C) displayed columns of chondrocytes typical of osteochondral bone formation, while in the knockout group (Fig. 2D), a wider hypertrophic zone and mildly disrupted calcified zone were observed. Trabeculae present within primary spongiosa of knockout animals exhibited little branching compared to the trabecular structure observed in wild-type animals, and in addition, the trabeculae in GPR103^−/− mice appeared slightly thinned as confirmed by microCT (see below). The number of tartrate-resistant acid phosphatase (TRAP)-positive-staining cells in the knockout group was lower than that in the wild-type group, particularly in the secondary spongiosa.

FIG. 2.

Histological analysis of thoracic vertebrae (A to D) and of femurs (E and F) from wild-type (A, C, and E) and knockout (B, D, and F) mice. Sagittal sections of vertebrae (A to D) and of distal halves of femurs (E to F) were stained for TRAP and counterstained with methyl green-thionin. Cells positive for TRAP staining and the osteochondral growth plate (Gp) are indicated by arrows. Trabecular branches (Tb) and bone marrow (bm) are shown by their respective labels. Pictures were taken at ×20 (A and B), ×100 (C and D), and ×40 (E and F) magnifications.

The proportional differences in disk space, end plate morphology, trabecular pattern, and distribution of TRAP-positive cells appear to arise from a combination of early arrest of osteochondral bone formation and reduced bone remodeling in the secondary spongiosa. In addition, the larger disk space in GPR103 knockout animals could provide the laxity causing the vertebral column to become kyphotic.

The structure and spatial arrangement of the femoral trochlear groove, epiphysis, growth plate, primary and secondary spongiosa, trabeculae, and cortical bone in the mid-shaft of the femur were similar in wild-type and knockout mice (Fig. 2E and F). The similarity in structure and spatial arrangement of femurs from these animals indicated that there is no obvious difference between wild-type and knockout mice in long bones.

MicroCT was used to quantify the trabecular bone loss observed by histopathology in the vertebrae. The lumbar vertebrae (LV4) of four males and females of each wild-type and knockout genotype were examined (Fig. 3 and Table 2). This experiment confirmed the decrease in trabecular volume and connective density. The differences were more prominent in females (30% reduction) than in males (25% reduction). These results, combined with the aforementioned histological observations, demonstrate that GPR103 deficiency in mice leads to a reduction of trabecular bone volume.

FIG. 3.

MicroCT analysis of the lumbar vertebra LV4 in GPR103 mice. MicroCT images of the trabecular bone in LV4 of a wild-type mouse (A) and a knockout mouse (B) are shown.

TABLE 2.

MicroCT analysis of LV4^a

Genotype	Gender	Mean BV/TV (%) (SEM)	Mean Conn.D (SEM)	Mean SMI (SEM)	Mean Tb.N. (no./mm) (SEM)	Mean Tb.Th. (mm) (SEM)	Mean Tb.Sp. (mm) (SEM)
+/+	Female	18.8 (1.5)	65.361 (1.630)	1.348 (0.168)	3.527 (0.068)	0.063 (0.001)	0.293 (0.007)
−/−	Female	13.4 (1.8)	56.951 (7.994)	1.942 (0.220)	3.178 (0.179)	0.055 (0.001)	0.334 (0.025)
+/+	Male	24.1 (4.1)	115.499 (3.624)	1.176 (0.309)	4.584 (0.355)	0.060 (0.006)	0.219 (0.021)
−/−	Male	20.0 (2.3)	94.184 (15.026)	1.725 (0.155)	4.671 (0.303)	0.058 (0.003)	0.216 (0.016)

^aThe trabecular bone structure measurements were made for a core cylinder centered through the vertebral body. The following parameters were assessed for each cylindrical volume: total volume (TV), bone volume (BV), fraction of bone volume (BC/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp.), connectivity density (Conn.D), and structure model index (SMI).

Ovariectomy in GPR103 knockout mice.

Ovaries from GPR103^−/− females were removed to determine whether their response to estrogen depletion was different from that of wild-type mice (Fig. 4). BMD was measured immediately before and 6 weeks after ovariectomy using a DEXA scanner. As expected, BMDs of all mice in both long bones and spine were reduced after ovariectomy. The spine, which contains the highest proportion of trabecular bone, was more affected than the femurs in all animals. Importantly, GPR103 knockout mice exhibited a greater reduction in total BMD than wild-type mice. The greater decrease in BMD observed in GPR103^−/− mice was attributed almost entirely to a significant decrease in BMD in the spine of ovariectomized GRP103^−/− mice compared to spinal BMD of the wild-type mice. Body weight was not affected considerably by the surgery (data not shown), indicating that the observed effect did not result from a systemic cachexia.

FIG. 4.

Loss of bone mineral density in GPR103 wild-type (+/+) and knockout (−/−) females after ovariectomy. Total bone mineral density and bone mineral density of regions of interest, the spine and the left (L.) and right (R.) femurs, were measured in mice before and after ovariectomy. The percentage of loss of bone mineral density relative to the initial value is shown.

GPR103 is expressed in bone, kidney and brain of mouse and human tissues.

As a first step towards elucidating the mechanisms of action of GPR103 on bone formation, we performed expression analysis by RT-PCR (Fig. 5A to C) and in situ hybridization (Fig. 5D). Quantitative RT-PCR (Fig. 5A) showed that GPR103 was expressed in spine RNA but not in femur RNA of adult GPR103^+/+ male and female mice. GPR103 expression was not detected in spine or femur RNA from GPR103^−/− mice, thereby confirming that that the GPR103 gene was completely inactivated in GPR103 knockout mice. GPR103 expression was also detected in brain, spinal cord, and eye tissue but not in heart, kidney, liver, lung, lymph node, ovary, placenta, skeletal and smooth muscle, small intestine, colon, spleen, testis, thymus, and thyroid gland. In humans, quantitative RT-PCR showed that the highest levels of expression were found in fetal bone, testis, brain, kidney, and heart, while little or no expression was detected in lung, bone marrow, and skeletal muscle (Fig. 1B). In situ hybridization was performed on mouse brain tissue sections (Fig. 1D). The GPR103 signal was strongest in the hypothalamic ventromedial nucleus and was also observed in the hippocampus.

FIG. 5.

Expression analysis of GPR103 mRNA. (A) Quantitative RT-PCR of GPR103 mRNA was performed with RNA samples from various mouse organs as indicated. In some cases, RNA was isolated from GPR103 wild-type (+/+) or knockout (−/−) male (M) or female (F) mice. The relative mRNA expression levels are indicated. Sk. Muscle, skeletal muscle; Sm. Muscle, smooth muscle; Sm. Intestine, small intestine. (B) Quantitative RT-PCR of GPR103 mRNA was performed with RNA samples from various human organs as indicated. The relative mRNA expression levels are indicated. (C) RT-PCR performed on various cultured cells: osteoblast primary cultures derived from a 16-year-old individual (16Y); osteoblast primary cultures derived from a 39-year-old individual (39Y) cultured without (0) or with 10⁻⁹, 10⁻⁷, and 10⁻⁵ M concentrations of dexamethasone (Dex); primary human mesenchymal stem cells (huMSC); osteosarcoma cell line SaOS-2; MC3T3-E1; MC3T3-E1 subclone 4 (MC3T3.E1/c4); myoblast cell line C2C12; and bone marrow stromal cell line ST2. Brain RNA was included as a positive control. (D) In situ hybridization of GPR103 mRNA in a coronal section of mouse brain. The hypothalamic ventromedial nucleus is indicated by an arrow.

To determine whether GPR103 expression in bone was representative of its presence in the osteoblast lineage, we performed RT-PCR on various cultured cell lines and osteoblast primary cultures (Fig. 5C). The MC3T3-E1 cell line is a phenotypically heterogeneous cell line originating from mouse calveria that has been subcloned for selected levels of osteoblast differentiation. MC3T3-E1 subclone 4 (or MC3T3E1/C4) differentiates more readily than the parental clone. Interestingly, GPR103 mRNA is undetectable in the parental cell line but abundant in subclone 4. GPR103 was also expressed in the human osteosarcoma cell line SaOS-2 and in the bone marrow stromal cell line ST2 but was undetected in primary human mesenchymal stem cells. The presence of GPR103 mRNA in the mouse myoblast cell line C2C12 confirms that its expression is not restricted to cells of the osteoblast lineage. In addition, GPR103 mRNA was detected in osteoblast primary cultures derived from a 16-year-old individual and a 39-year-old individual, respectively. Dexamethasone, a glucocorticoid known to inhibit osteoblast growth, dramatically reduced the expression levels of GPR103 in a dose-dependent manner in primary human osteoblasts.

DISCUSSION

GPR103 is a novel G-protein-coupled receptor of previously unknown function that was first identified on the basis of its homology to other known GPCRs and localization in the hypothalamus (25). To identify the function of GPR103 in mammals, we generated GPR103-deficient mice by homologous recombination into ES cells to demonstrate that GPR103 regulates bone formation.

Bone remodeling and osteoporosis.

Osteoporosis is a complex, multifactorial disease characterized by a reduction in bone mineral density and an increased risk of bone fracture. It may progress silently for decades without symptoms until fractures occur. Normally, bone is continuously being replaced from the osteochondral growth and differentiation of precursor cells into osteoblasts and chondrocytes, followed by osteoclast-driven resorption, thereby maintaining a balance between bone formation and resorption rates. Bone loss occurs when the balance between new bone growth and resorption is broken. There are two types of bones: trabecular bone, also called cancellous or spongy bone, and cortical bone, also called lamellar bone. Trabecular bone has a higher surface-to-volume ratio than cortical bone and is better suited for rapid turnover. Consequently, osteoporosis affects predominantly trabecular bones (15).

Bone loss can be reduced by treatment, but it is difficult to restore the microarchitecture of the skeleton once bone has been lost. Therefore, therapeutic treatment has focused on early intervention in order to slow the rate of bone loss in these patients. Some existing therapies have highly undesirable side effects. For example, estrogen replacement therapy inhibits osteoclastic bone resorption (for reviews, see references 6 and 32); however, its potential relationship with increased breast cancer and heart disease risks have raised concerns. Food supplements, such as calcium and vitamin D, have limited efficacy in preventing osteoporosis. Oral bisphosphonates, the current standard of care for osteoporotic patients, require that the patient remain upright for a period of time after dosing, and intravenous bisphosphonates have recently been implicated in osteonecrosis. Recently, teriparatide, a recombinant form of the parathyroid hormone, was approved by the FDA for the treatment of osteoporosis. This drug is important to the treatment of osteoporosis, since it is currently the only anabolic agent available with potential to restore some of the bone lost in osteoporosis; however, treatment requires a daily injection of the peptide, and teripeptide carries a black-box warning on the label indicating that osteosarcomas were observed in rats treated with this drug. Therefore, there is still an unmet need to increase our understanding of the mechanisms underlying bone formation and resorption in order to develop more efficacious and safer therapies with improved dosing regimens for bone diseases (for a recent review of therapeutic approaches to osteoporosis, see reference 28).

GPR103 is a novel mouse model for osteopenia.

Few mouse models to study osteoporosis exist. The smaller weight of mice and their proportionally high cortical bone content relative to that of humans renders them less susceptible to bone fractures. Still, the overall response of mouse bone to molecular signals such as estrogen depletion is similar to the response of human bones. In both cases, estrogen depletion causes a 10 to 20% loss in bone density. Surgical removal of ovaries (37) or chemical ablation of follicle functions by injection of 4-vinylcyclohexene diepoxide (29) in mice mimics the effect of estrogen depletion in postmenopausal women and induces osteopenia.

Knockout mutations of key regulators of bone formation induce often severe and pleiotropic abnormalities, rendering their use as mouse models for adult osteoporosis difficult. Mice lacking the calcium-regulating parathyroid hormone (PTH) were dysmorphic but viable and demonstrated diminished cartilage matrix mineralization, decreased neovascularization with reduced expression of angiopoietin-1, and reduced metaphyseal osteoblasts and trabecular bone (23, 24). The PTH/PTH-related protein receptor regulates fetal bone formation, although later analysis was not possible because PTH/PTH-related protein receptor null mutant mice died in midgestation.

Osteoporotic fractures are observed in osteoprotegerin (Opg) knockout mice (5). Opg is a secreted protein that inhibits osteoclast formation by acting as a decoy receptor for RANKL, an activator of osteoclasts. Osteopenia has also been observed in other transgenic models carrying null mutations in LRP5 (18), proteoglycan biglycan (Bgn) (30), and Smad3 (4), among others.

Kyphosis has also been observed in knockout mice, including mice deficient in tetranectin (19). However, the tetranectin mutation affected only the mouse interdisc cartilage but not bone formation.

GPR103 knockout mice constitute a novel model of osteopenia in mice. Loss of bone mineral density seems to be more spatially restricted than in other models, i.e., specifically in the spine, where the ratio of trabecular cortical bone is the highest. Interestingly, GPR103 expression was detected in spine but not in femur RNA of adult mice. However, it is possible that the restricted pattern of expression reflects the low number of osteoblasts in the adult femur. Interestingly, human fetal bone, which undergoes active osteogenesis, expresses a high level of GPR103.

GPR103 mutant mice also had a reduced connectivity density, suggesting that the absence of GPR103 affects the ability of osteochondral progenitor cells to fully differentiate into chondrocytes as well as osteoblasts. Interestingly, preliminary analysis of tetranectin expression by microarray suggests that tetranectin mRNA levels are reduced in GPR103^−/− females (unpublished observations), whereas microarray analysis failed to show significant differences in Opg, Bgn, or Smad3 between knockout and wild-type mice.

Calcium and phosphate levels in the blood of GPR103^−/− mice were also unchanged relative to those of wild-type mice (unpublished observations). In addition, the expression of the 25-OHD₃ 1α-hydroxylase (Cyp27b1), an enzyme strongly induced by PTH and responsible for the metabolic conversion of vitamin D to its most active form, 1,25(OH)₂ vitamin D (calcitriol), is unchanged in knockout mice (unpublished observations). Those preliminary results are in line with the histological finding that GPR103 affects the early stage of osteochondral growth plate formation rather than the late differentiation process.

Is GPR103 acting through a hypothalamic relay or directly on osteoblasts?

GPR103 was originally cloned from a hypothalamus cDNA library (25). In situ hybridization performed here confirmed the presence of a strong GPR103 signal in the hypothalamic ventromedial nucleus. Recently, the hypothalamus has been shown to be an important center for the control of bone mass (9). Indeed, ob/ob mice, carrying a leptin null mutation, and db/db mice, carrying a leptin receptor null mutation, are obese and have an increased bone mass. Leptin is a small polypeptide hormone secreted by adipocytes regulating body weight via its binding to the leptin receptor in the hypothalamus. The increase in bone mass precedes the onset of obesity, suggesting that the increased bone mass is due to the signaling of the hypothalamic leptin receptor rather than a general response to increased body weight. In addition, leptin induces bone loss when injected intracerebroventricularly, and this effect was recently demonstrated to be mediated by β2-adrenergic receptors and the neuropeptide CART in the sympathetic nervous system (13).

NPY is a downstream modulator of leptin action on food intake (14). Intracerebroventricular injection of NPY also induces bone loss. NPY binds to a family of neuropeptide Y receptors found in the hypothalamus and elsewhere. Interestingly, Y2 knockout mice have a twofold increase in trabecular bone volume, indicating that NPY, like leptin, also mediates bone formation (1). Whether NPY2R-expressing neurons also contribute to the effect of leptin on bone is unclear. Recently, it was suggested that the increases in BMD seen in the NPY2R knockout mice may in part be attributed to another downstream orexigenic peptide, MCH, since MCH-R1 null mice have a reduction in cortical bone mass. However, it is worth noting that no difference was observed in the trabecular bone densities of MCH-R1 knockout mice (3). Whether MCH expression is increased in NPY2R knockout mice or whether β-adrenergic receptors also mediate the NPY2R knockout bone phenotype remains to be determined. The orexigenic peptide ghrelin has also been implicated in the regulation of bone formation (16, 27).

Given that GPR103 is a hypothalamic receptor for a peptide with orexigenic activity (36), it is possible that some of the effect of GPR103 genetic ablation on bone mass is mediated through this hypothalamic relay. However, the presence of GPR103 on osteoblastic cells and its regulation by glucocorticoids in osteoblast primary cultures suggest that GPR103 has the potential to promote bone formation directly by its receptor activity in osteoblasts. Future experiments using the RF26a peptide on primary cultures of osteoblasts collected from human donors and from wild-type and knockout mice will help elucidate those mechanisms.