Wnt-Lrp5 Signaling Regulates Fatty Acid Metabolism in the Osteoblast

Abstract

The Wnt coreceptors Lrp5 and Lrp6 are essential for normal postnatal bone accrual and osteoblast function. In this study, we identify a previously unrecognized skeletal function unique to Lrp5 that enables osteoblasts to oxidize fatty acids. Mice lacking the Lrp5 coreceptor specifically in osteoblasts and osteocytes exhibit the expected reductions in postnatal bone mass but also exhibit an increase in body fat with corresponding reductions in energy expenditure. Conversely, mice expressing a high bone mass mutant Lrp5 allele are leaner with reduced plasma triglyceride and free fatty acid levels. In this context, Wnt-initiated signals downstream of Lrp5, but not the closely related Lrp6 coreceptor, regulate the activation of β-catenin and thereby induce the expression of key enzymes required for fatty acid β-oxidation. These results suggest that Wnt-Lrp5 signaling regulates basic cellular activities beyond those associated with fate specification and differentiation in bone and that the skeleton influences global energy homeostasis via mechanisms independent of osteocalcin and glucose metabolism.

INTRODUCTION

Wnt signaling regulates nearly all aspects of osteoblast function, from initial fate specification (1) to the control of osteoclast differentiation (2). In this pathway, low-density-lipoprotein (LDL)-related receptor 5 (Lrp5) and the closely related Lrp6 participate in the stabilization and activation of the transcription factor β-catenin by facilitating the interaction of Wnt ligands with frizzled receptors (3, 4). Osteoblasts express all the components of the Wnt/β-catenin pathway, and most have now been linked with bone development and maintenance in humans and mouse models (5–7). Mutations in the LRP5 gene, in particular, can result in premature and generalized osteoporosis as in the rare condition osteoporosis pseudoglioma (8) or a high-bone-mass phenotype (9, 10) likely due to an increase in the number of mineralizing osteoblasts (11).

Like other metabolically active cells, osteoblasts require a supply of energy-rich molecules to fuel the synthesis, deposition, and mineralization of bone matrix (12). When energy input fails to meet demand, normal bone accrual ceases, a phenomenon that is evident clinically by the arrest of longitudinal bone growth and osteopenia observed in undernourished children and adults (13, 14). Therefore, osteoblasts must possess mechanisms to acquire and regulate the utilization of fuel macromolecules, as well as the ability to communicate energy needs with other tissues.

Recent studies have delineated a role for the osteoblast in a bone-pancreas endocrine loop that contributes to the regulation of glucose metabolism, as well as bone acquisition. Insulin receptor signaling in the osteoblast regulates the activity of the osteogenic transcription factor Runx2 and is required for the attainment of a mature phenotype as well as normal postnatal bone acquisition (15). In addition, insulin actions regulate the production and bioavailability of osteocalcin (15, 16), a bone-derived hormone that in its undercarboxylated form favors pancreatic insulin production and insulin sensitivity in peripheral tissues (17). Osteocalcin administration partially corrects the impairment in glucose metabolism evident in mice lacking the insulin receptor specifically in osteoblasts (15) and abrogates the metabolic disturbances associated with feeding wild-type mice a high-fat diet (18, 19). Genetic studies suggest osteoblasts also contribute to the regulation of whole-body metabolism via osteocalcin-independent mechanisms (20).

Osteoblasts express the enzymatic requirements for glycolysis, metabolize glucose to lactate (21), and store glycogen granules (22, 23), which suggests that glucose metabolism may be required for osteoblastic maturation and matrix production. However, osteoblasts also oxidize fatty acids, and this process may account for 40 to 80% of their energy requirements (24). Bone takes up a significant fraction of postprandial lipoproteins (25), and osteoblasts produce apolipoprotein E (26). Moreover, osteoblasts grown in the absence of lipoproteins exhibit severe defects in proliferation that cannot be rescued by growth factor supplementation (27). Such observations accord with the perceived need for the osteoblast to maintain a high level of ATP production to support matrix production and mineralization, but how the bioenergetics of bone remodeling contribute to whole-body fuel distribution remains poorly understood.

Disruptions in Wnt signaling have also been implicated in metabolic disease, which suggests that in addition to osteoblast differentiation, this pathway may regulate osteoblast metabolism. Wnt signaling inhibits adipogenesis (28, 29), and polymorphisms in pathway components are associated with an increased risk for the development of diabetes (30, 31), obesity (32), and hyperlipidemia (33, 34). In addition, the Lrp5 coreceptor was originally identified as a candidate gene in the IDDM4 locus with linkage to insulin-dependent diabetes mellitus (35, 36), and mice globally deficient for Lrp5 are glucose intolerant and exhibit increased plasma cholesterol levels when fed a high-fat diet (37, 38).

We report here on the regulation of osteoblast metabolism by the Wnt signaling coreceptor Lrp5. Mice lacking Lrp5 specifically in osteoblasts and osteocytes exhibited marked increases in body fat with corresponding reductions in whole-body energy expenditure. These phenotypes were not observed in mice lacking the closely related Lrp6 coreceptor in osteoblasts. The increased adiposity was associated with increased levels of circulating lipids and reductions in the oxidative potential of Lrp5-deficient osteoblasts. Mechanistically, Wnt signaling through Lrp5 and the stabilization of β-catenin appear to regulate the expression of key enzymes in fatty acid metabolism. Our results suggest that Lrp5 regulates fatty acid oxidation by osteoblasts and osteocytes and that reduced lipid utilization by bone results in the reallocation of fuel sources and alterations in body composition.

MATERIALS AND METHODS

Animal models.

The Institutional Animal Care and Use Committee of the Johns Hopkins University approved all procedures involving mice. The generation of mice lacking Lrp5 (ΔLrp5) and Lrp6 (ΔLrp6) in osteoblasts was described previously (6). In brief, osteocalcin-Cre (Oc-Cre^TG/+) mice (39) were crossed with mice in which the second exon of Lrp5 or Lrp6 is floxed. Lrp5KI mice contain a floxed Lrp5 allele that contains the high bone mass causing G171V mutation (40). When crossed with Oc-Cre mice (Oc-Cre^TG/+; Lrp5^KI/+), Lrp5^G171V expression is induced in osteoblasts. PCR analysis of ear or tail biopsy specimens was used to confirm genotypes; strategies are available upon request. All mice were maintained on a C57BL/6 background. Male littermates were selected for detailed metabolic analyses.

Culture of primary osteoblasts.

Mouse osteoblasts were isolated from the calvaria of 1- to 3-day-old neonates by serial digestion in 1.8 mg of collagenase/ml. For in vitro deletion of Lrp5 or Lrp6 or expression of LRP5^G171V, osteoblasts containing floxed alleles were infected with adenovirus encoding Cre recombinase or green fluorescent protein (Vector Biolabs). Infection with adenovirus expressing cytomegalovirus (CMV)/β-catenin was used to overexpress β-catenin (Vector Biolabs). A multiplicity of infection of 100 was used in all experiments, and gene deletion/overexpression was confirmed by quantitative PCR (qPCR). Osteoblast differentiation was induced by supplementing alpha minimal essential medium (αMEM) containing 10% serum with 10 mM β-glycerol phosphate and 50 μg of ascorbic acid/ml. Differentiation was confirmed by Alizarin red S staining for mineralization according to standard techniques. Lithium chloride (20 mM, Sigma-Aldrich) and Wnt10b (100 ng/ml, R&D Systems) were added 18 h prior to experimental assays. Human low-density lipoprotein (LDL) labeled with DiI (1,19-dioctadecyl-3,3,39,39-tetramethylindocarbocyanine perchlorate) was purchased from Life Technologies and added to cell cultures (10 μg/ml) for 2 h prior to visualization by microscopy.

Gene expression studies.

Total RNA was extracted from osteoblast cultures or mouse tissues using TRIzol (Life Technologies). For skeletal tissue, the femur was cleaned of soft tissue, the growth plate was removed, and the marrow cavity was flushed with phosphate-buffered saline. Reverse transcriptase reactions were carried out using 1 μg of RNA and the iScript cDNA synthesis system (Bio-Rad). Real-time qPCR was carried out using iQ Sybr green Supermix (Bio-Rad) using primer sequences obtained from PrimerBank (http://pga.mgh.harvard.edu/primerbank/index.html). Reactions were normalized to endogenous β-actin reference transcripts. Antibodies for FLAG and Lrp5 were obtained from Cell Signaling.

Microarray studies.

Microarray expression profiles were generated using Illumina MouseRef-8 v2.0 BeadChips according to standard protocols. Hybridization was performed in the Genome Sciences Laboratory in the Center for Public Health Genomics at the University of Virginia. Raw expression values were transformed using variance stabilizing transformation (41) and normalized with the robust spline normalization algorithm using the LumiR R package (42). After the preprocessing step and before the differential expression analysis, genes with a detection P value of >0.05 in more than 3/4 of the samples were filtered out. For the differential analysis step the Limma R package (43) was used to fit a linear model for each gene with the experimental conditions as factors and the expression levels (in log₂ space) as the response variable, and then contrasts were performed to assess the log fold changes between the experimental conditions. Limma increases the statistical power to detect differential expression by moderating the standard errors of the estimated fold changes using an empirical Bayes method. Genes with an adjusted P < 0.05 were deemed differentially expressed.

In vitro metabolic studies.

Fatty acid, pyruvate, and glucose oxidation were measured in flasks with stoppers equipped with center wells as previously described (44). Cultures were differentiated for 7 days prior to analysis and then incubated at 37°C in media containing 0.5 mM l-carnitine, 0.2% bovine serum albumin, and either [¹⁴C]oleate (Perkin-Elmer), [¹⁴C]glucose, or [¹⁴C]pyruvate. ¹⁴CO₂ was captured and counted by the addition of 1 N perchloric acid to the reaction mixture and 1 M NaOH to the center well containing Whatman filter paper. The acidified reaction mixture was incubated overnight at 4°C and centrifuged at 4,000 rpm for 30 min before aliquots of the supernatant were counted for ¹⁴C-labeled acid soluble metabolites. Lactate levels in cells and conditioned medium were assessed using a lactate assay kit (Sigma-Aldrich).

Imaging.

High-resolution images of the mouse femur and L5 vertebra were acquired using a desktop microtomographic imaging system (Skyscan 1172; Bruker) in accordance with the recommendations of the American Society for Bone and Mineral Research (45). Bones were scanned at 50 keV and 200 μA using a 0.5-mm aluminum filter with an isotropic voxel size of 10 μm. In the femur, trabecular bone parameters were assessed in the 500 μm proximal to the growth plate and extending for 2 mm (200 computed tomographic [CT] slices). In the spine, trabecular bone parameters were assessed between the cranial and caudal growth plates. The resulting two-dimensional cross-sectional images are shown in gray scale. Body composition and bone mineral density were measured by quantitative nuclear magnetic resonance (qNMR; Echo MRI) and dual-energy X-ray absorptiometry (DXA) (Lunar PIXImus II; GE Healthcare), respectively. For histological examination of bone marrow adiposity, femurs were decalcified in 10% EDTA before sectioning and staining with hematoxylin and eosin according to standard techniques.

Metabolic studies and bioassays.

Indirect calorimetry was conducted in an open-flow indirect calorimeter (comprehensive lab animal monitoring system; Columbus Instruments). Calorimetry, daily body weight, and daily food intake data were acquired during a 4-day experimental period. The data from the first 3 days were used to confirm acclimation to the calorimetry chamber, and the fourth day was used for analyses. The rates of oxygen consumption (VO₂, ml/kg/h) and carbon dioxide production (VCO₂) were measured for each chamber every 20 min throughout the study. Respiratory exchange ratio (RER; RER = VCO₂/VO₂) was calculated by Oxymax software (v4.90) to estimate relative oxidation of carbohydrate (RER = 1.0) versus fat (RER approaching 0.7), not accounting for protein oxidation. Energy expenditure (EE) was calculated as EE = VO₂ × [3.815 + (1.232 × RER)] (46) and normalized for subject body mass (kcal/kg/h). Glucose levels were measured using a OneTouch Ultra hand-held glucose monitor. For glucose tolerance testing, glucose (2 g/kg of body weight [BW]) was injected intraperitoneally after an overnight fast. For insulin tolerance testing, mice were fasted for 4 h and then injected intraperitoneally with insulin (0.2 U/kg BW). Plasma triglycerides (Sigma), β-hydroxybutyrate (Sigma), cholesterol (BioAssay Systems), free fatty acids (Sigma), and glycerol (Sigma) were measured colorimetrically in plasma collected from random fed mice or after an overnight fast. Plasma insulin (Alpco) and undercarboxylated osteocalcin (TaKaRa) were assessed by using an enzyme-linked immunosorbent assay. Insulin signaling in white adipose tissue (WAT) and liver was assessed by injection of insulin (2 U/kg) into the portal vein before excising tissue and snap-freezing for immunoblot analysis (47, 48). Pancreata were fixed and stained, and islet morphometry was assessed as previously described (49). Frozen sections of gonadal white adipose tissue were stained with hematoxylin and eosin.

Statistics.

All results are presented as means ± the standard errors of the mean (SEM). Statistical analyses were performed using unpaired, two-tailed Student t tests or analysis of variance, followed by post hoc tests. A P value less than 0.05 was considered significant.

Microarray data accession number.

Microarray data have been deposited in NCBI's Gene Expression Omnibus database under accession number GSE55900.

RESULTS

Peripheral adiposity is increased in mice deficient for Lrp5 in osteoblasts.

Recent studies in our laboratory have used mice deficient in either Lrp5 or Lrp6 in osteoblasts to examine the contribution of each Wnt coreceptor to postnatal bone acquisition (6). Although both mutants exhibit the expected reduction in bone mineral density (Fig. 1A) and vertebral trabecular bone volume (Fig. 1B and C), we noted striking increases in visceral fat in the Lrp5 mutants (Lrp5^flox/flox; Oc-Cre^TG/+ [referred to here as ΔLrp5]) that were not observed in Lrp6 mutants (Lrp6^flox/flox; Oc-Cre^TG/+ [ΔLrp6]). To formally characterize this phenotype, we generated cohorts of control (Lrp5^flox/flox or Lrp6^flox/flox), ΔLrp5, and ΔLrp6 mice and assessed body weights and body composition by qNMR as they aged. In agreement with the anecdotal observations and despite normal longitudinal growth (Fig. 1D) and organ and muscle weights (Fig. 1J and K), ΔLrp5 mice, but not ΔLrp6 mice, exhibited progressive increases in fat mass (Fig. 1F) and gonadal fat pad weight (Fig. 1H and I) and an age-related reduction in body mass (Fig. 1E) that may be attributed to a decrease in lean mass relative to control littermates (Fig. 1G) and the progressive loss of bone mass (6). Since there was no difference in the relative abundance of marrow adipocytes in control and ΔLrp5 mice (Fig. 1L), the increased adiposity evident in the mutant animals is unlikely to result from a shift in lineage allocation.

FIG 1.

ΔLrp5 mice, but not ΔLrp6 mice, have increased body fat. (A) Whole-body bone mineral density assessed by DXA at 12 weeks of age (n = 11 to 21 mice). There were no differences in the bone mineral densities of Lrp5^flox and Lrp6^flox mice, so data were combined as a control. (B) Representative micro-CT images of the L5 vertebrae at 24 weeks of age. (C) Quantitation of trabecular bone volume per tissue volume (BV/TV, %) in the L5 vertebrae at 24 weeks of age (n = 4 to 9 mice). (D) Representative images of control and ΔLrp5 mice at 8 weeks of age showing normal longitudinal growth. (E) Body weights (n = 7 to 20 mice). (F) Fat mass assessed by qNMR at 8 and 24 weeks of age (n = 6 to 11 mice). (G) Lean mass assessed by qNMR. (H) Representative images of the gonadal fat pad isolated from control and ΔLrp5 mice at 24 weeks of age. (I) Gonadal fat pad mass assessed at 8 and 24 weeks of age (n = 6 to 11 mice). (J and K) Wet tissue weights of major organs (J) and muscle groups (K) of 8-week-old control and ΔLrp5 mice (n = 5 to 6 mice). (L) Representative histological sections showing the distal femur of 24-week-old control and ΔLrp5 mice. No difference in marrow adiposity is apparent. (M) Daily food intake at 4 and 8 weeks of age (n = 4 to 6 mice). (N) Ambulatory activity measured via beam breaks in Oxymax system in 24 h (n = 4 to 6 mice). (O to Q) Indirect calorimetry at 8 weeks of age (n = 4 to 6 mice). (O) VO₂ (ml/kg/h). (P) Respiratory exchange ratio (RER). (Q) Energy expenditure (kcal/kg/h). The data are represented as means ± the SEM. *, P < 0.05.

To investigate a possible cause for the alteration in body composition in Lrp5 mutants, we next examined global energy balance in age-matched cohorts of ΔLrp5 mice and control (Lrp5^flox/flox) littermates. Although food intake (Fig. 1M) and ambulatory activity (Fig. 1N) in the mutant mice were similar to those in control mice, indirect calorimetry demonstrated that ΔLrp5 mice had a decreased oxygen consumption rate (VO₂, Fig. 1O), without a corresponding change in the respiratory exchange ratio (RER, Fig. 1P). Thus, the increased fat mass evident in Lrp5 mutants was likely secondary to a decreased rate of oxidative metabolism that resulted in an overall decrease in the rate of energy expenditure relative to control littermates (Fig. 1Q).

ΔLrp5 mice have elevated triglyceride and free fatty acid levels.

Based upon our previous studies (15) and the potential for cross talk between Lrp5 and insulin signaling (50), we initially suspected the alterations in body composition and energy expenditure might be the result of impaired glucose homeostasis. However, fasting and random fed glucose levels in 8-week-old ΔLrp5 mice were comparable to controls (Fig. 2A), and the mutant mice exhibited normal fasting insulin levels (Fig. 2B) and responses during glucose tolerance and insulin tolerance testing (Fig. 2C to E). Moreover, the mutant did not exhibit a defect in pancreatic islet morphometry (data not shown); the ability of white adipose, liver, or cultured osteoblasts to respond to insulin (data not shown); or the levels of serum undercarboxylated osteocalcin (Fig. 2F). Rather, ΔLrp5 mice exhibited elevated levels of plasma triglycerides (Fig. 2G) and free fatty acids (Fig. 2H) relative to controls, with normal plasma glycerol (Fig. 2I) and cholesterol (Fig. 2J) levels. This phenotype was specific to the genetic ablation of Lrp5 as ΔLrp6 mice exhibited normal triglyceride and free fatty acid levels even at 24 weeks of age (Fig. 2K and L).

FIG 2.

Plasma lipids are increased in ΔLrp5 mice. (A) Blood glucose levels in random fed control and ΔLrp5 mice and after an overnight fast at 8 weeks of age (n = 5 to 12 mice). (B) Insulin levels in mice after an overnight fast at 8 weeks of age (n = 8 mice). (C) Glucose tolerance testing at 8 weeks of age (n = 7 to 12 mice). (D) Insulin tolerance testing at 8 weeks of age (n = 7 to 8 mice). (E) Area under the curve (AUC) analysis for GTT (C) and ITT (D). (F) Undercarboxylated osteocalcin (Glu) levels in 8-week-old mice (n = 10 mice). (G to J) Plasma analysis of random fed or fasted 8-week-old control and ΔLrp5 mice (n = 8 to 10 mice). (G) Plasma triglyceride levels (mg/dl). (H) Free fatty acid levels (mmol/liter). (I) Glycerol levels (mg/dl). (J) Cholesterol levels (mg/dl). (K and L) Plasma triglyceride levels (K) and free fatty acid levels (L) in random fed 24-week-old control and ΔLrp6 mice (n = 5 to 7 mice). (M) β-Hydroxybutyrate levels (mmol/liter) in randomly fed or fasted 8-week-old control and ΔLrp5 mice (n = 8 to 10 mice). (N) Representative images of the liver isolated at 8 weeks of age. (O to S) qPCR analysis of tissues isolated from 8-week-old control and ΔLrp5 mice (n = 6 mice). (O) Lrp5, Ldlr, and Vldlr expression in liver. (P) Expression of genes involved in lipid synthesis in the liver. (Q) Expression of genes involved in lipid oxidation in the liver. (R) Lrp5, Ldlr, and Vldlr expression in the gonadal fat pad. (S) Expression of genes involved in lipid storage in the gonadal fat pad. (T) Representative histological images of adipocyte morphology in the gonadal fat pad. The data are represented as means ± the SEM. *, P < 0.05.

To understand the physiologic basis for this increase in circulating lipids, we examined gene expression profiles in the liver, white adipose, muscle, and brown adipose. Consistent with normal ketone levels (Fig. 2M), the liver appeared normal in ΔLrp5 mice, without signs of hepatic steatosis (Fig. 2N). Likewise, the expression of genes involved in lipid synthesis (Fig. 2P) and lipid oxidation (Fig. 2Q) was normal or modestly increased in the liver of ΔLrp5 mice relative to controls. The increased expression of Lrp5 in the liver of the mutant mice (Fig. 2O) may represent a compensatory mechanism to increase lipid uptake from the circulation (37, 38). In white adipose tissue isolated from the gonadal fat pad, the expression of Ldlr and Vldlr was increased (Fig. 2R) in ΔLrp5 mice relative to controls, as was the expression of genes involved in lipid storage (Fig. 2S). Consistent with this change in gene expression, adipocyte size was noticeably larger in the gonadal fat pad of ΔLrp5 mice compared to controls (Fig. 2T). The expression of genes involved in lipid oxidation in muscle and brown adipose in ΔLrp5 mice was comparable to that in controls, as was the weight of the interscapular brown adipose depot (data not shown). Taken together, these data suggest that the increased adiposity evident in ΔLrp5 mice is associated with increases in circulating lipids and fatty acid storage by adipose without marked changes in the metabolism of the liver, muscle, or brown adipose.

Lipid oxidation is reduced in osteoblasts deficient for Lrp5.

Given the emerging biology of osteocalcin as a bone-derived hormone that influences whole-body metabolism (17, 51), it appeared reasonable to hypothesize that the increase in fat deposition in ΔLrp5 mice might be the result of alterations in another bone-derived regulator of metabolism. However, while treating primary preadipocytes with the fatty acid stearate recapitulated the gene signature of adipocytes isolated from ΔLrp5 mice (Fig. 3A), culturing cells in media conditioned by either Lrp5-deficient osteoblasts (Fig. 3B) or those expressing an allele associated with a high-bone-mass phenotype (Lrp5^G171V) (Fig. 3C and 4A) did not alter the expression of these same genes. Therefore, to investigate the mechanism responsible for the increased body fat and dyslipidemia in ΔLrp5 mice in an unbiased way, we profiled RNA isolated from ΔLrp5 osteoblasts differentiated for 7 days by microarray (Illumina MouseRef-8 v2.0). These analyses identified 1015 unique genes that were positively or negatively regulated in ΔLrp5 osteoblasts relative to controls (adjusted P ≤ 0.05) (see Table S1 in the supplemental material). In addition to the expected reductions in putative Wnt target genes (Gja1, Vegfa, Ptgs2, and Tgfb3) and osteoblastic markers (Col1a2, Ibsp, Enpp1, Atf4, and Mgp), the expression of a number of genes involved in lipid metabolism were inhibited in cultures of ΔLrp5 osteoblasts relative to those of controls. These included Acat2, Acsl1, Acaa1a, and Acadvl. qPCR analyses confirmed the downregulation of these genes and also revealed reductions in the mRNA levels of Acadl, Acads, Hadha, and Cpt1b, the rate-limiting enzyme in fatty acid oxidation (Fig. 3D). Similar and, in some cases, more severe reductions in the expression of these genes were also evident in RNA samples isolated from the femurs of ΔLrp5 mice (Fig. 3E). Several genes involved in oxidative phosphorylation and the TCA cycle, including Atp6vlb2, Ndufv1, Ndufs8, Cox5a, Cox5b, Cs, Idh2, and Idh3a, were upregulated by the loss of Lrp5 expression (see Table S1 in the supplemental material), which likely represents a compensatory response designed to maintain cellular energy levels.

FIG 3.

Lrp5 regulates the expression of genes involved in lipid oxidation in osteoblasts. (A to C) qPCR analysis of genes associated with lipid storage in primary preadipocytes after overnight treatment with vehicle or 100 μM stearate (A) or medium conditioned by control and ΔLrp5 osteoblasts (B) or control and Lrp5^G171V osteoblasts (C). (D and E) qPCR analysis of genes involved in lipid metabolism in control and ΔLrp5 osteoblasts after 7 days of differentiation (D) and in mRNA samples isolated from the femur of 8-week-old control and ΔLrp5 mice (E, n = 5 mice). (F) Oxidation of [¹⁴C]oleate to ¹⁴CO₂ by differentiating osteoblasts. (G) qPCR analysis of genes involved in lipid metabolism in differentiating osteoblasts. (H to J) Impact of etomoxir (ETO; 100 μM) on osteoblast function after 7 days of differentiation. (H) Oxidation of [¹⁴C]oleate to ¹⁴CO₂. (I) qPCR analysis of genes involved in osteoblast differentiation. (J) Alizarin red staining for calcium deposition. (K to N) Oxidation of ¹⁴C-labeled substrates by control and ΔLrp5 osteoblasts after 7 days of differentiation. The results are normalized to protein concentration and presented as relative to control. (K) [¹⁴C]oleate to ¹⁴CO₂. (L) [¹⁴C]oleate to ¹⁴C-labeled acid-soluble metabolites (ASM). (M) [¹⁴C]pyruvate to ¹⁴CO₂. (N) [¹⁴C]glucose to ¹⁴CO₂. (O) Visualization of DiI-labeled LDL uptake by control and ΔLrp5 osteoblasts after 7 days of differentiation. (P and Q) Cellular lactate levels in cultures of control and ΔLrp5 osteoblasts after 7 days of differentiation (P) and in undifferentiated calvarial cells (Q). (R) qPCR analysis of genes involved in glucose metabolism in cultures of undifferentiated control and ΔLrp5 osteoblasts. (S) qPCR analysis of genes involved in lipid metabolism in control and ΔLrp6 osteoblasts after 7 days of differentiation. The data are represented as means ± the SEM. *, P < 0.05.

FIG 4.

Expression of Lrp5^G171V increases lipid oxidation in vitro and reduces adiposity in vivo. (A and B) Detection of Lrp5^G171V-FLAG expression in calvarial osteoblasts by qPCR (A) and immunoblotting (B). (C) qPCR analysis of osteoblastic markers after 7 days of differentiation. (D) qPCR analysis of axin2 and Nkd2 mRNA levels in osteoblasts differentiated for 7 days. (E) Oxidation of [¹⁴C]oleate to ¹⁴CO₂. (F) Oxidation of [¹⁴C]oleate to ¹⁴C-labeled acid-soluble metabolites (ASM). (G) qPCR analysis of genes involved in lipid metabolism in control and Lrp5^G171V-expressing osteoblasts after 7 days of differentiation. (H) Oxidation of [¹⁴C]glucose to ¹⁴CO₂. (I) Cellular lactate levels. (J) Representative micro-CT images of the distal femur of 8-week-old controls and mice expressing LRP5^G171V in osteoblasts. (K) Representative micro-CT images of the L5 vertebrae from 8-week-old control and Lrp5^G171V mice. (L) Quantitation of trabecular bone volume in the distal femur and L5 vertebrae (n = 6 to 9 mice). (M) Body weights of 8-week-old control and Lrp5^G171V mice (n = 6 to 9 mice). (N and O) qNMR analysis of fat mass (N) and lean mass (O) in 8-week-old control and Lrp5^G171V mice (n = 6 to 9 mice). (P) Gonadal fat pad mass assessed at 8 weeks of age (n = 12 to 14 mice). (Q to W) Blood and plasma analysis of random fed mice at 8 weeks of age (n = 7 to 12 mice). (Q) Plasma triglyceride levels. (R) Free fatty acid levels. (S) Glycerol levels. (T) Cholesterol levels. (U) β-Hydroxybutyrate levels. (V) Blood glucose. (W) Insulin levels. (X) Undercarboxylated osteocalcin (Glu) levels. (Y) qPCR analysis of genes involved in lipid metabolism in mRNA samples isolated from the femurs of 8-week-old controls and mice expressing LRP5^G171V in osteoblasts (n = 5 mice). The data are represented as means ± the SEM. *, P < 0.05.

To directly examine the functional significance of these changes in gene expression, we examined fatty acid oxidation in cultures of differentiating primary calvarial osteoblasts. Osteoblasts were incubated with ¹⁴C-labeled oleate and the production of ¹⁴CO₂ was used to assess fatty acid metabolism. By this measure, fatty acid oxidation increased 3-fold during the differentiation of wild-type osteoblasts (Fig. 3F) and was accompanied by corresponding increases in the mRNA levels of genes involved in fatty acid metabolism (Fig. 3G). We next differentiated osteoblasts in the presence of etomoxir (ETO), an antagonist of carnitine palmitoyltransferase 1 (52), the rate-limiting enzyme in fatty acid oxidation. This manipulation, which resulted in a >90% decrease in oleate oxidation to ¹⁴CO₂ (Fig. 3H), impaired the expression of markers of osteoblast differentiation (Fig. 3I) and decreased matrix mineralization (Fig. 3J), which suggests that the utilization of fatty acids as a fuel source is critical for normal osteoblast function.

Consistent with the downregulation of enzymatic mediators of fatty acid oxidation in ΔLrp5 osteoblasts, oleate oxidation to CO₂ was reduced by 43% in cultures of ΔLrp5 osteoblasts differentiated for 7 days (Fig. 3K), while the production of acid-soluble metabolites, products of incomplete oxidation, was reduced by 13% relative to controls (Fig. 3L). Oxidation of pyruvate (Fig. 3M) and glucose (Fig. 3N) and the uptake of low-density lipoproteins (Fig. 3O) by ΔLrp5 osteoblasts were comparable to controls, indicating that Lrp5 deficiency did not result in a generalized defect in oxidative metabolism. In addition, we did not observe an impairment in the production of lactate in differentiated ΔLrp5 osteoblasts (Fig. 3P). Although in accordance with a recent analysis of osteoprogenitors (53), lactate production and genes involved in glycolysis were reduced in undifferentiated calvarial cells (Fig. 3Q and R). These data, together with the normal expression of genes involved in fatty acid metabolism in osteoblasts deficient for Lrp6 (Fig. 3S), suggest that the metabolic phenotype of Lrp5 mutant mice is the result of reduced fatty acid utilization by bone.

Expression of the high bone mass Lrp5^G171V allele increases fatty acid oxidation.

To further test the importance of Lrp5 signaling in the regulation of fatty acid metabolism, we examined the effect of expressing the Lrp5^G171V allele that results in a high-bone-mass phenotype in humans (9) on fatty acid metabolism and body composition. Calvarial osteoblasts were isolated from Lrp5KI mice, which contain a Cre-inducible Lrp5^G171V-FLAG cDNA inserted into the first intron of the endogenous Lrp5 gene (40), and infected with adenovirus expressing Cre recombinase to induce the expression of the mutant allele (here referred to as Lrp5^G171V) or green fluorescent protein as a control (Fig. 4A and B). In addition to the expected enhancements in osteoblast differentiation (Fig. 4C) and the expression of Wnt target genes (Fig. 4D), expression of Lrp5^G171V increased fatty acid oxidation in vitro. Oxidation of oleate to CO₂ and acid soluble metabolites was increased by 17 and 39%, respectively, in cultures of Lrp5^G171V-expressing osteoblasts differentiated for 7 days compared to controls (Fig. 4E and F) and was accompanied by an enhancement in the expression of genes involved in fatty acid metabolism (Fig. 4G). Glucose metabolism and lactate production were not affected by the expression of this mutant allele (Fig. 4H and I).

In vivo, the expression of Lrp5^G171V by osteoblasts and osteocytes (Lrp5^KI/+; Oc-Cre^TG/+) resulted in an increase in femoral trabecular bone volume and cortical tissue area, but consistent with a previous report (40), had no discernible impact on bone architecture in the spine (Fig. 4J to L) at 8 weeks of age. As expected, the metabolic phenotype of Lrp5^G171V mice was the opposite of that observed in ΔLrp5 mice. Fat mass determined by qNMR (Fig. 4N), gonadal fat pad weight (Fig. 4P), plasma triglycerides (Fig. 4Q), and free fatty acids (Fig. 4R) were all reduced relative to controls in Lrp5^G171V mice, while body weight (Fig. 4M) and lean body mass (Fig. 4O) were comparable to controls. Plasma glycerol, cholesterol, ketone, glucose, insulin and undercarboxylated osteocalcin levels were all normal in Lrp5^G171V mice (Fig. 4S to X). Moreover, the expression of genes involved in fatty acid metabolism was increased in RNA samples isolated from the femurs of Lrp5^G171V mice relative to controls (Fig. 4Y). Taken together, these data imply that fatty acid metabolism in bone cells and fat mass are correlated with the functional activity of Lrp5.

Wnt stimulation enhances fatty acid oxidation in vitro.

Finally, to determine whether the impact of Lrp5 deficiency or Lrp5^G171V expression on fatty acid metabolism was solely due to changes in osteoblast differentiation and to discern the contribution of β-catenin in these phenotypes, we examined the effect of acute activation of Wnt signaling and β-catenin overexpression on fatty acid metabolism. First, wild-type osteoblasts were treated with lithium chloride (LiCl) to activate Wnt signaling (Fig. 5A) by inhibiting the activity of Gsk-3β (54). Consistent with a direct role for Wnt signaling in the regulation of fatty acid metabolism in osteoblasts, LiCl treatment resulted in a 20 to 25% increase in the oxidation of oleate relative to NaCl-treated controls (Fig. 5B and C) and enhanced metabolic gene expression (Fig. 5D). Second, osteoblasts expressing or rendered deficient for Lrp5 were treated with Wnt10b. Similar to LiCl treatment, Wnt-stimulation enhanced oleate oxidation (Fig. 5E) and the expression of genes involved in fatty acid metabolism (Fig. 5F). However, this effect was completely abolished in cells lacking Lrp5. Genetic ablation of the Lrp6 coreceptor had no effect on the ability of Wnt10b to increase oleate oxidation or alter gene expression (data not shown). Lastly, β-catenin overexpression (Fig. 5G and H), by infecting osteoblasts with adenoviral CMV–β-catenin, activated Wnt target gene expression, increased oleate oxidation, and augmented the expression of enzymatic mediators of lipid metabolism (Fig. 5H to J). Collectively, these data suggest a model wherein Wnt-initiated activation of β-catenin downstream of the Lrp5 coreceptor regulates the transcription of enzymes required for fatty acid metabolism. Reductions in fatty acid metabolism by bone cells in the case of Lrp5 deficiency result in the need to redistribute these fuel macromolecules to other tissues and the accumulation of peripheral adiposity.

FIG 5.

Acute Wnt stimulation increases lipid oxidation. (A to D) Primary osteoblasts were differentiated for 7 days and then treated for 18 h with 20 mM LiCl to stimulate Wnt/β-catenin signaling or NaCl (20 mM) as a control. (A) qPCR analysis of axin2 and Nkd2 mRNA levels. (B) Oxidation of [¹⁴C]oleate to ¹⁴CO₂. (C) Oxidation of [¹⁴C]oleate to ¹⁴C-labeled acid-soluble metabolites (ASM). (D) qPCR analysis of genes involved in lipid metabolism. (E and F) Primary osteoblasts expressing or deficient for Lrp5 were differentiated for 7 days and then treated with 100 ng of Wnt10b/ml for 18 h. (E) Oxidation of [¹⁴C]oleate to ¹⁴CO₂. (F) qPCR analysis of Acaa1a, Acadvl, and Cpt1b mRNA levels. (G to J) Primary osteoblasts differentiated for 4 days after infection with adenovirus expressing green fluorescent protein (ad-GFP) or β-catenin (ad-Ctnnb1). (G) Immunoblot analysis of β-catenin protein levels. (H) qPCR analysis of Ctnnb1, axin2, and Nkd2 mRNA levels. (I) Oxidation of [¹⁴C]oleate to ¹⁴CO₂. (J) qPCR analysis of genes involved in lipid metabolism. The data are represented as means ± the SEM. *, P < 0.05.

DISCUSSION

Mutations in LRP5 in humans have a profound impact on bone mass and skeletal integrity (8–10). In the present study, we uncovered a previously unrecognized role for Lrp5 in bone and provide evidence that Wnt signaling through Lrp5 but not Lrp6 regulates osteoblast metabolism. Genetic ablation of Lrp5 expression in osteoblasts and osteocytes resulted in the expected reduction in bone mass (6) and also led to an unexpected change in body composition and energy expenditure in mice. We attribute these changes in metabolism to a reduction in the utilization of fatty acids as a fuel source by osteoblasts.

A growing body of evidence indicates that the skeleton contributes to the regulation of whole-body metabolism (55, 56). On one hand, the bone-derived hormone osteocalcin allows the tissue to assist in the regulation of glucose metabolism (17). On the other, the sheer size of the skeleton and the near constant remodeling process that removes and replaces old and damaged bone tissue suggests that bone-forming osteoblasts require significant energy resources (57). Previous studies (24, 25, 27) and data presented here imply that the osteoblast obtains a significant fraction of its energetic requirements from the metabolism of fatty acids. It follows that cellular signaling pathways essential to osteogenesis and bone acquisition should also regulate bone cell metabolism.

In humans, polymorphisms in LRP5 exhibit genetic linkage with elevated LDL cholesterol levels, hypertension, increased body mass index, and obesity (32, 58–60). In our studies, mice lacking Lrp5 in osteoblasts and osteocytes exhibited an age-related increase in fat mass and reductions in both lean mass and bone mass. Osteoblasts deficient for Lrp5 exhibited a reduced capacity to fully oxidize oleate to CO₂ and impairments in the expression of genes involved in lipid β-oxidation. Wnt stimulation, β-catenin overexpression, or expression of Lrp5^G171V increased oxidation and gene expression. These data indicate that lipid metabolism is positively correlated with the level of signaling through Lrp5 and accord with the influence of Wnt-Lrp5 signaling on osteoblast differentiation and bone matrix mineralization (6, 61, 62). In addition, they suggest that the disturbances in bone architecture and whole-body metabolism arising in humans from mutations in LRP5 may be linked.

Several pieces of experimental evidence indicate that the metabolic phenotypes evident in Lrp5 mutants are the direct result of alterations in osteoblast metabolism. First, gene expression patterns in tissues most likely to influence whole-body fatty acid metabolism, including the liver, muscle, and brown adipose, were nearly identical in control and Lrp5 mutant mice. In fact, the most obvious change in gene expression was compatible with an increase in lipid storage by white adipose. Second, medium conditioned by Lrp5-deficient osteoblasts did not alter the expression of genes in preadipocytes, and while we identified a number of genes involved in fatty acid β-oxidation in our microarray study, alterations in the expression of a factor that might fulfill a hormonal function was not immediately apparent. Serum undercarboxylated osteocalcin levels were not affected by Lrp5 deficiency, and it seems unlikely that alterations in the level of this hormone underlie the metabolic phenotype in Lrp5 mutants as glucose metabolism was normal. Third, studies in other tissues have shown a direct link between Wnt signaling and the regulation of lipid oxidation. In particular, the Wnt effector Tcf4 regulates the expression of Acadl, Acaa1a, and Cpt1b in the liver (63). Lrp5 signaling in osteoblasts regulates these same genes.

A recent study (53) reported that Wnt-Lrp5 signaling also regulates glucose metabolism in osteoblasts. In this context, Wnt3a signaling through Lrp5 in cell lines characteristic of osteoprogenitors activated mTORC2-Akt signaling to increase the conversion of glucose to lactate via the Warburg effect. We also found that lactate production and the expression of glycolytic genes are reduced in undifferentiated calvarial osteoblasts lacking Lrp5, but this effect was lost as osteoblasts matured in vitro. Taken together, these data suggest that the influence of Lrp5 on metabolic functions changes during the course of osteoblast maturation and, potentially, that the fuel requirements of osteoblasts are differentiation state-dependent. This idea may also explain the reduction in glycogen storage by differentiating osteoblasts (22, 23).

An intriguing finding in these studies relates to the apparent difference in the functions of Lrp5 and Lrp6 in osteoblasts and osteocytes. Although Lrp5 mutants accumulated body fat, Lrp6 mutants did not exhibit changes in body composition or circulating lipid levels. The two Wnt coreceptors are 71% homologous, have a similar structure, and form a distinct subfamily of the LDL receptor-related proteins (35, 64–66). However, previous studies have suggested that Lrp5 and Lrp6 exert distinct actions owing to differences in tissue distribution, affinity for individual Wnt ligands, and the ability to propagate Wnt-initiated signals (67–70). Indeed, our previous studies indicated that while both receptors are required for normal bone acquisition, they influence bone structure via different mechanisms (6). Whereas Lrp6 mutants have a reduced number of osteoblasts, Lrp5 primarily regulates the mineralization process. With regard to metabolism, both receptors retain the capacity to bind lipoproteins (71–73), and, much like LRP5, mutations in LRP6 are associated with metabolic dysfunction in humans (33, 74). One potential explanation for the functional difference in osteoblasts may relate to the expression pattern of the coreceptors during osteoblast differentiation. The expression of Lrp5 increases during differentiation, which is consistent with the 3-fold increase in oleate oxidation we observed in differentiating osteoblasts. The expression of Lrp6 remains stable during the differentiation process (6). In addition, the upregulation of Lrp5 mRNA levels may be sufficient to compensate for the loss of Lrp6 function in osteoblasts. Examining this possibility directly would be difficult since mice lacking both Lrp5 and Lrp6 either in early mesenchyme or mature osteoblasts die prematurely (6, 75).

The results presented here bring to light a number of important questions that will require additional studies. The first relates to the requirement for fatty acid metabolism in bone acquisition and the metabolic flexibility of bone cells. As indicated above, our studies and others imply this may be an important fuel source for osteoblasts and osteocytes. Testing this directly will require transgenic models, currently in development, that directly alter the metabolic potential of osteoblasts. A second concerns the lack of a change in glucose metabolism in the Lrp5 mutant mice. Increases in circulating lipids and ectopic lipid accumulation are often associated with the development of a state of insulin resistance (76). It is possible that the levels of circulating fatty acids in Lrp5 mutants did not reach levels sufficient to induce a change in insulin sensitivity. In this case, it would be interesting in future studies to determine if Lrp5 deficiency in osteoblasts renders mice more sensitive to disturbances in glucose metabolism when fed a high-fat diet. Mice globally deficient for Lrp5 are glucose intolerant, but this phenotype is likely due to deficits in insulin secretion by the β cell (37). Finally, our data suggest that Wnt targeted therapeutics might have a beneficial impact on metabolism. The increased levels of circulating sclerostin, a bone-derived inhibitor of Wnt signaling (77), in metabolic diseases (78–80) suggest that therapies designed to increase bone mass by targeting sclerostin (81, 82) might also impact circulating lipid levels. To date, body composition has not been a major endpoint in the studies designed to examine antisclerostin agents as a bone anabolic therapy. Such studies are under way in our laboratory.

In summary, our study suggests a direct link between bone cell metabolism and whole-body energy expenditure and supports the notion that fatty acid utilization by osteoblasts is regulated by the Wnt-Lrp5 pathway, which also regulates bone acquisition. These observations should impact our understanding of bone homeostasis as well as global energy balance.