Skeletal inflammation and attenuation of Wnt signaling, Wnt ligand expression, and bone formation in atherosclerotic ApoE-null mice

Abstract

ApoE-null (ApoE-KO) mice fed a high-fat diet (HFD) develop atherosclerosis, due in part to activation of vascular inflammation by oxidized low-density lipoprotein. Since bone loss also occurs in these mice, we used them to investigate the impact of oxidized lipids on bone homeostasis and to search for underlying pathogenic pathways. Four-month-old female ApoE-KO mice fed a HFD for three months exhibited increased levels of oxidized lipids in bone, as well as decreased femoral and vertebral trabecular and cortical bone mass, compared with ApoE-KO mice on normal diet. Despite HFD-induced increase in expression of Alox15, a lipoxygenase that oxidizes LDL and promotes atherogenesis, global deletion of this gene failed to ameliorate the skeletal impact of HFD. Osteoblast number and function were dramatically reduced in trabecular and cortical bone of HFD-fed mice, whereas osteoclast number was modestly reduced only in trabecular bone, indicating that an imbalance in favor of osteoclasts was responsible for HFD-induced bone loss. These changes were associated with decreased osteoblast progenitors and increased monocyte/macrophages in the bone marrow as well as increased expression of IL-1β, IL-6, and TNF. HFD also attenuated Wnt signaling as evidenced by reduced expression of Wnt target genes, and it decreased expression of pro-osteoblastogenic Wnt ligands. These results suggest that oxidized lipids decrease bone mass by increasing anti-osteoblastogenic inflammatory cytokines and decreasing pro-osteoblastogenic Wnt ligands.

epidemiological evidence points to a linkage between atherosclerosis and osteoporosis in elderly humans (17, 29, 51), as well as hip fractures, regardless of age (52); thus, similar pathogenic factors may contribute to both conditions. Crucial mechanisms underlying atherogenesis have been elucidated using mice in which clearance of lipoproteins is disrupted by deletion of the low-density lipoprotein (LDL) receptor or deletion of apolipoprotein E (ApoE), a component of LDL that mediates binding to the LDL receptor (25, 46, 54). Circulating levels of LDL therefore increase dramatically, especially when the mice are fed a high-fat diet (HFD). As a result, LDL accumulates in the subendothelial extracellular matrix and undergoes oxidative modification (ox) by reactive oxygen species (ROS), peroxidases, and lipoxygenases. OxLDL binds to scavenger receptors and Toll-like receptors on endothelial cells and resident macrophages. This stimulates the production of cytokines that recruit and activate additional macrophages and lymphocytes from the circulation. Activated macrophages then accumulate oxLDL to become lipid-containing “foam” cells that, together with smooth muscle cells, initiate the formation of calcified atherosclerotic plaques. Activated macrophages also accelerate atherogenesis via 12,15-lipoxygenase (Alox15)-mediated oxidation of LDL. The contribution of this mechanism has been shown by the attenuation of HFD-induced atherogenesis in mice with systemic or macrophage-specific deletion of Alox15 (11, 20, 26).

Consistent with the evidence for a linkage between atherosclerosis and osteoporosis in humans, ApoE-null (ApoE-KO) mice or LDL receptor-null mice fed a HFD exhibit reduced bone mass (24, 48, 49, 53). The mass and structural integrity of the skeleton are maintained by teams of osteoclasts and osteoblasts that, respectively, resorb old bone and replace it with new (38). Loss of bone with advancing age, estrogen deficiency, or glucocorticoid excess is due to disruption in the focal balance between resorption and formation. Whether HFD-induced bone loss in atherosclerotic mice is due to increased osteoclast number and excessive resorption, decreased osteoblast number and compromised bone formation, or both, remains unknown.

We have therefore performed a comprehensive investigation of the pathogenesis of bone loss in HFD-fed ApoE-KO mice. We show that HFD causes a reduction in osteoblast number and a focal imbalance between resorption and formation, leading to loss of both trabecular and cortical bone. HFD also induces a bone inflammatory response and suppressed pro-osteoblastogenic Wnt signaling, suggesting a mechanistic explanation for the reduction in osteoblast number and bone formation.

MATERIALS AND METHODS

Animal experiments.

All animal procedures were approved by the Institutional Animal Care and Use Committees of the University of Arkansas for Medical Sciences and the Central Arkansas Veterans Healthcare System. Breeders from Jackson Labs (Bar Harbor, ME) were used to establish colonies of ApoE-KO mice (no. 002052) and Alox15-KO mice (no. 00277) on C57BL/6J background. Appropriate crosses were made to generate ApoE-KO;Alox15-KO mice and ApoE-KO control littermates. Mice were genotyped using DNA obtained by clipping the tail at the time of weaning. For ApoE gene determination, we used the following primers: ApoE forward (F) 5′-GCCTAGCCGAGGGAGAGCCG-3′, ApoE reverse (R)1, 5′-TGTGACTTGGGAGCTCTGCAGC-3′, ApoE R2, 5′-GCCGCCCCGACTGCATCT-3′. The WT gene was detected with ApoE F and ApoE R1, WT product size of 155 bp. The deleted gene was detected with ApoE F and ApoE R2, KO product size of 254 bp. For Alox15 gene analysis, we used the following primers: Alox15 F 5′-GGCTGCCTGAAGAGGTACAG-3′, Alox15 R1 5′-CCATAGACGAGACCAGCACA-3′, Alox15 R2 5′-GGGAGGATTGGGAAGACAAT-3′. The WT gene was detected with Alox15 for and Alox15 R1, WT product size of 417 bp. The deleted gene was detected with Alox15 F and Alox15 R2, KO product size of 200 bp.

Animals were group housed under specific pathogen-free conditions, maintained with a constant temperature of 23°C, a 12:12-h light-dark cycle, and had access to food and water ad libitum. Mice were assigned to normal diet (ND) or HFD treatment groups at 4 mo of age, based on bone mineral density (BMD) of the lumbar spine. Specifically, mice were rank-ordered by BMD and then assigned the number 1 or 2, successively. Animals with the same number were assigned to the same treatment group to give identical group means. The ND (TD22/5 Rodent Diet from Harlan Laboratories) contains 5.5% fat, 0.003% cholesterol, 0.9% phosphorus (0.6% nonphytate), and 1.1% Ca and has 17% of kcal from fat. The HFD diet (TD90221 from Harlan Laboratories) contains 15.8% fat (∼50% from cocoa butter, which comprises >95% triglycerides), 1.25% cholesterol and 0.5% sodium cholate, 1.0% phosphorus (0.8% nonphytate), and 0.9% Ca and has 37.1% of kcal from fat. To allow dynamic bone histomorphometry, tetracycline (30 mg/kg ip) was administered at 6 and 2 days before animals were euthanized. Animals were euthanized by CO₂ inhalation. Genotype was rechecked after euthanasia.

To obtain bone marrow cells and plasma appendicular bones, the proximal and distal ends were removed and the resulting cortices then centrifuged in a 1.5-ml Eppendorf tube at 2,500 rpm for 5 min. The supernatant was removed, and the cells were rinsed with 50 μl of PBS. Supernatants were combined and frozen; marrow cells were resuspended for further analysis. Osteoblast progenitors in bone marrow cell isolates were quantified by their ability to form a colony of osteoblastic cells, as previously described (15), in triplicate cultures of cells pooled from four mice of each treatment group.

Flow cytometry and immunomagnetic separation of marrow cells.

Marrow cells were harvested from one femur and two humeri. For flow cytometry, 0.5 × 10⁶ cells were incubated for 30 min on ice with anti-CD3-FITC (BD Pharmingen, Cat. no. 555274), anti-CD11b-PE (e-Bioscience, Cat. no. 12-0112-81), and anti-CD19-APC (BD Pharmingen, Cat. no. 550992). After rinsing, cells were fixed with PBS + 2% paraformaldehyde and then analyzed by flow cytometry using the BD LSR Fortessa running BD FACSDiva software. Compensation settings were adjusted during data acquisition, using compensation controls for each fluorochrome. Further analysis was performed using FlowJo v.10. For immunomagnetic separation, 10 × 10⁶ cells were incubated for 10 min at 4°C in staining buffer with anti-CD3e or anti-CD11b kit or anti-CD19 Ab (Miltenyi Biotech Cat. nos. 130-094-973, 130-097-142, and 130-097-144, respectively), according to manufacturer's instructions. Cell separation was performed using MS columns (Miltenyi Biotech, Cat. no. 130-042-201). The purity of enriched cells was checked by FACS.

Serum and plasma measurements.

Total serum Ca (no. 0150), inorganic phosphate (no. 0830), and creatinine (no. 0430) were quantified using kits from Stanbio Laboratory. Parathyroid hormone [PTH(1–84)] was measured by ELISA (Immutopics, no. 60-2305). Osteocalcin was measured by ELISA (Biomedical Technologies, #BT-470). Collagen type I cross-linked C-telopeptides (CTx) was determined using a kit from Immunodiagnostic Systems Ltd (no. AC-02F1). Thiobarbituric acid reactive substances (TBARS) were determined using a kit from Cayman Chemical (no. 10009055), and marrow plasma murine TNF was determined with a kit from R&D Systems (no. MTA00B).

Bone imaging.

BMD was determined by dual-energy X-ray absorptiometry (DEXA) with a PIXImus densitometer (GE Lunar) as described (44). The mean coefficient of variation of BMD of a proprietary phantom (performed prior to each use) during the conduct of these studies was 0.40%. Architectural features of the femur and the 2nd lumbar vertebra (L2) were quantified by microcomputed tomography (μCT) with a μCT40 (Scanco Medical, Basserdorf, Switzerland) using Eval Program, v.6.0 as described (28). The mean coefficient of variation of the μCT phantom (performed weekly) during the conduct of these studies was 1.23%. For cortical bone porosity measurements, femurs were scanned from a point immediately distal to the third trochanter to the beginning of the distal growth plate at medium resolution (nominal isotropic voxel size = 12 μm). After defining endosteal and periosteal boundaries, an additional image processing script (“peel-iter = 2”) was used to eliminate false voids caused by imperfect wrap of the contours to the bone surface. Images were binarized with a threshold of 365 mg/cm³, and overall porosity was determined with the “cl_image” script to obtain bone volume and void volume.

Histology.

For preparation of nondecalcified sections, bones were fixed in Millonig's fixative (Leica Microsystems), then embedded in methyl methacrylate. For decalcified sections, bones were fixed in B-Plus Fix (Biochemical), decalcified in formic acid, and then embedded in paraffin. For frozen sections, bones were fixed in Millonig's, decalcified in 14% EDTA, and stored in 30% sucrose for at least 2 days. Bones were then embedded in Cryo-Gel (Electron Microscopy Sciences) for preparation of cryosections.

Histomorphometric analysis was done using 5-μm-thick midsagittal sections of femurs and ventral planar sections of vertebrae (L1 and L2). Measurements were made in a blinded fashion on nondecalcified sections using Osteomeasure v. XP 3.1 (Osteometrics). Sections were either unstained for determination of fluorescent tetracycline labeling to measure dynamic indexes of bone formation or stained for tartrate-resistant acid phosphatase (TRAP) to visualize osteoclasts, with toluidine blue counterstaining for measurement of static indexes of osteoclast and osteoblast number and function, as previously described (2). Osteoblasts were identified as teams of cells (≥2) overlying osteoid. Histomorphometric data are reported using the nomenclature recommended by the American Society for Bone and Mineral Research (14). Lipids were visualized in frozen sections following fixation with 4% paraformaldehyde and staining with Oil Red O. Alox15 immunostaining was performed on paraffin sections using 15-lipoxygenase-1 rabbit polyclonal antiserum, 1:1,000 (Cayman Chemical, no. 160704), followed by DAB staining (DAKO) and methyl green counterstain.

To quantify aortic lesions, the aorta was removed by gripping a piece of diaphragm attached to the end of the aorta and dissecting away the connective tissue attaching the aorta to the muscle wall of the thoracic and abdominal cavity. After flushing with PBS using a 23-gauge needle, adventitial fat was removed. The aorta was then cut along the longitudinal axis, followed by fixation in 4% paraformaldehyde in PBS for 24 h and pinning to a black wax surface. Lesions were visualized with Oil Red O in 70% 2-propanol and quantified as the ratio of lesion area to aortic area as measured from the aorta root to the kidney, using Osteomeasure.

Western blotting.

Protein extracts of bones were resolved by SDS-PAGE. Alox15 was detected with 15-lipoxygenase-1 rabbit polyclonal antiserum, 1:1,000 dilution (Cayman Chemical, no. 160704). 4-Hydroxynonenal (4-HNE) was detected with goat polyclonal antisera, 1:1,000 dilution (Novus Biologicals, no. NB100-63093). Proteins were expressed relative to α-tubulin, detected with monoclonal mouse anti-α-tubulin 1:1,000 dilution (Sigma Aldrich, no. T 6199). Intensity of the bands was quantified with a VersaDoc imaging system (Bio-Rad).

Quantitative PCR.

Freshly dissected bones were cleaned of adherent tissue and frozen immediately in liquid N₂. Total RNA was extracted with Ultraspec (Biotecx Laboratories, Houston, TX) and reverse-transcribed using the High-Capacity cDNA Archive Kit (Applied Biosystems). Taqman PCR was performed using primers and probes manufactured by TaqMan Gene Expression Assays (Applied Biosystems); they are listed in Table 1. Transcript levels were calculated by normalizing to the housekeeping gene mitochondrial ribosomal protein S2 using the ΔC_T method (34).

Table 1.

TaqMan reagents used for quantification of mRNA by qPCR

NCBI Gene Name	Gene Name	TaqMan #
Alox12	Arachidonate 12-lipoxygenase	Mm00545833_m1
Alox15	Arachidonate 15-lipoxygenase	Mm01250458_m1
Axin2	Axin2	Mm00443610_m1
Mrps2	Mitochondrial ribosomal protein S2 (chob)	Mm01962382_s91
Cx43	Gap junction membrane channel protein alpha 1	Mm01962382_s53
Dkk1	Dickkopf homolog 1	Mm00438422_m1
IL-1ß	Interleukin-1β	Mm00434228_m1
IL-6	Interleukin-6	Mm00446190_m1
Nkd2	Naked cuticle 2 homolog	Mm00472240_m1
Sost	Sclerostin	Mm00470479_m1
Sfrp4	Secreted frizzled-related protein 4	Mm00840104_m1
TNF	Tumor necrosis factor-α	Mm00443258_m1
Tnfrsf11b	Tumor necrosis factor receptor superfamily, member 11b (osteoprotogerin, OPG)	Mm00435452_m1
Wisp1	WNT1 inducible signaling pathway protein 1	Mm00457574_m1
Wnt4	Wingless-related MMTV integration site 4	Mm00437341_m1
Wnt5a	Wingless-related MMTV integration site 5A	Mm00437347_m1
Wnt6	Wingless-related MMTV integration site 6	Mm00437353_m1
Wnt10b	Wingless-related MMTV integration site 10B	Mm00442104_m1
Wnt16	Wingless-related MMTV integration site 16	Mm00446420_m1

Statistics.

Data are shown as mean ± SD or as box plots that represent the 25th and 75th percentile of data points. Whiskers represent the 10th and 90th percentile of data points, dots represent data points outside the 10th or 90th percentiles, and the line denotes the median value. Group mean values were compared, as indicated in the figure and table legends, by Student's unpaired two-tailed t-test or by two-way ANOVA with a Holm-Sidak multiple comparison test. Longitudinal data were analyzed by paired t-test or by repeated-measures ANOVA with Hom-Sidak P value adjustment for multiple comparisons. When variance was not equivalent, data were log transformed, or the nonparametric Mann-Whitney rank sum test was used instead of Student's t-test. P values < 0.05 were considered significant.

RESULTS

Increased lipid oxidation and Alox15 expression in HFD-fed ApoE-KO mice.

Administration of a HFD to 4-mo-old female ApoE-KO mice for 3 mo caused loss of BMD in both the femur and lumbar spine, as determined by DEXA (Table 2). Body weight and fat mass were not affected. HFD-fed mice exhibited extensive extracellular accumulation of lipid within the bone marrow, as determined by Oil Red O staining of frozen vertebral sections (Fig. 1A). This diet also increased lipid oxidation as measured by TBARS in the circulation and in plasma obtained from femoral bone marrow (Fig. 1B), and by proteins containing 4-HNE adducts extracted from bone (Fig. 1C).

Table 2.

HFD-induced loss of bone mass in female ApoE-KO mice

	ND (5)			HFD (6)
	Initial	Final	Change	Initial	Final	Change
BMD (femur), mg/cm2	75.4 ± 8.0	80.1 ± 0.6	4.7 ± 8.4	79.9 ± 6.1	59.9 ± 4.6	−20.0 ± 7.7*
BMD (spine), mg/cm2	55.7 ± 3.0	54.5 ± 6.2	1.2 ± 4.6	57.3 ± 2.6	46.5 ± 3.2	−10.8 ± 3.9*
Body weight, g	29.8 ± 4.6	34.7 ± 2.6	4.9 ± 3.0	30.1 ± 1.8	36.7 ± 2.3	6.6 ± 2.6
Body fat, %	20.1 ± 4.3	24.2 ± 3.2	4.1 ± 1.8	19.1 ± 3.4	21.1 ± 1.6	2.0 ± 3.0

Values are means ± SD; Number of animals examined is indicated in parentheses. ApoE-KO, apolipoprotein E knockout; ND, normal diet; HFD, high-fat diet; BMD, bone mineral density.

^*P < 0.01 by paired t-test.

Fig. 1.

Hih-fat diet (HFD) increases lipid oxidation and expression of Alox15 (12,15-lipoxygenase) in apolipoprotein E knockout (ApoE-KO) mice. A: Oil Red O staining of frozen sections of vertebral bone (L5). Bars = 20 μm. B: TBARS (thiobarbituric acid reactive substances) concentration in plasma and in femoral marrow plasma [normal diet (ND), n = 4; HFD, n = 5]. C: Western blotting and quantification of proteins with 4-hydroxynonenal (4-NHE) adducts in the bracketed area of the gel (ND, n = 6, HFD, n = 5). D: quantification of Alox15 and Alox12a transcripts by qPCR of vertebral bone (ND, n = 5, HFD, n = 6). E: quantification of Alox15 protein by Western blotting of vertebral bone extracts. F: Alox15 immunostaining (arrows) of sections of femoral bone. Bars = 20 μm. G: Alox15 transcripts in monocyte/macrophages (CD11b+), T lymphocytes (CD3+) and B lymphocytes (CD19+) isolated from pooled femoral marrow cells by immunomagnetic separation. *P < 0.05 vs. ND by t-test.

The HFD also increased the expression of Alox15 transcripts in vertebral bone (Fig. 1D), which was confirmed by Western blotting (Fig. 1E). There was no effect on expression of Alox12, and expression of other lipoxygenases was too low to be reliably measured by qPCR. The number of Alox15-expressing cells in the femoral bone marrow of HFD-fed mice increased as measured by immunostaining (Fig. 1F). Alox15 transcripts were high in monocytes/macrophages isolated from the bone marrow (Fig. 1G) compared with T cells and B cells. However, Alox15 expression per cell was not affected by the HFD, indicating that the rise in Alox15 measured at the organ level mainly reflects an increase in the number of monocyte/macrophages expressing this enzyme.

HFD induces bone loss independently of Alox15.

Based on the evidence that the HFD increases expression of Alox15, and that Alox15 is involved in both atherogenesis (11, 20, 26) and skeletal homeostasis (31), we used ApoE-KO;Alox15-KO mice to determine whether this lipoxygenase plays a functional role in HFD-induced bone loss. BMD values were indistinguishable in 4-mo-old female ApoE-KO and ApoE-KO;Alox15-KO mice and remained stable when maintained on the ND for the following 3 mo (Fig. 2A). Both strains exhibited progressive loss of femoral and vertebral bone mass when fed the HFD, and the time course and magnitude of this loss were not affected by Alox15 deficiency.

Fig. 2.

HFD reduces both trabecular and cortical bone mass independently of Alox15. A: sequential determination of bone mineral density (BMD) in femora and lumbar vertebrae determined by DEXA in ApoE-KO mice fed a ND (n = 13) or HFD (n = 14), and in ApoE-KO;Alox15-KO mice fed a ND (n = 14) or a HFD (n = 12), beginning at 4 mo of age. At the beginning of the experiment, BMDs of the 2 strains were indistinguishable by t-test (femur: ApoE-KO, 0.072 ± 0.003 mg/cm²; ApoE-KO;Alox15-KO, 0.070 ± 0.004 mg/cm²; lumbar vertebrae: ApoE-KO, 0.059 ± 0.003 mg/cm²; ApoE-KO;Alox15-KO, 0.060 ± 0.004 mg/cm²). * P < 0.05 vs. ND of the same genotype by repeated-measures 2-way ANOVA. B and C: μCT determination of trabecular and cortical bone architecture of the femur (B) and the 2nd lumbar vertebra (C) at 7 mo of age in the animals shown in A. Data were analyzed by 2-way ANOVA. *P < 0.05 vs. ND of the same genotype. Abbreviations: trabecular bone volume per tissue volume (BV/TV), trabecular thickness (Tb.Th) trabecular number, (Tb.N), trabecular separation (Tb.Sp), connectivity density (Conn.D), cortical thickness (Ct.Th), endosteal perimeter (En.Pm), periosteal perimeter (P.Pm), cortical porosity (Ct.Por).

Femoral trabecular bone volume as well as connectivity were lower in HFD-fed ApoE-KO mice compared with ND controls as determined by μCT, and the magnitude of these differences was unaffected by absence of Alox15 (Fig. 2B). The decrease in trabecular bone was due to reduced trabecular thickness in both strains, whereas trabecular number and trabecular separation were unaffected. In addition, the thickness of cortical bone, determined at the diaphysis, was lower in HFD-fed mice than in ND controls of both strains. The reduced cortical thickness was due to removal of bone from the endosteal surface, as indicated by an increase in endosteal perimeter but unchanged periosteal perimeter. Femoral cortical porosity was unaffected by the HFD in both strains. HFD-fed ApoE-KO and ApoE-KO;Alox15-KO mice did not exhibit a reduction in vertebral trabecular bone volume or connectivity, but trabecular thickness was reduced in both strains (Fig. 2C). Vertebral cortical thickness was lower in HFD-fed mice of both strains. In contrast to the skeletal response to HFD, Alox15 deletion attenuated HFD-induced atherogenesis as measured by aortic calcification, although circulating TBARS was unaffected, as previously reported (12, 26) (Fig. 3).

Fig. 3.

Deletion of Alox15 attenuates HFD-induced vascular calcification. A: representative images of Oil Red O-stained aorta. B: histomorphometric quantification of Oil Red O staining (n = 6/group). C: quantification of circulating TBARS (n = 4–6/group). Data were analyzed by 2-way ANOVA. *P < 0.05 vs. HFD of the same genotype; †P < 0.05 vs. ApoE-KO response to HFD.

Reduced osteoblast number and bone formation in HFD-fed ApoE-KO mice.

We next investigated the cellular basis of HFD-induced bone loss in ApoE-KO mice. Histomorphometric measurements of bone sections revealed a two- to threefold reduction in osteoclast number in femoral and vertebral trabecular bone of HFD-fed mice compared with ND controls (Table 3). In contrast, femoral endocortical osteoclast number was not affected.

Table 3.

Effect of HFD on static and dynamic indexes of bone remodeling

	Vertebral Trabecular Bone		Femoral Trabecular Bone		Femoral Endocortical Bone
	ND (8)	HFD (9)	ND (8)	HFD (9)	ND (8)	HFD (9)
Static indexes
N.Oc/B.Pm, /mm	3.4 ± 0.7	1.7 ± 0.6	3.5 ± 1.0	1.2 ± 0.7	3.2 ± 1.3	4.3 ± 1.4
N.Ob/B.Pm, /mm	17.9 ± 3.3	3.1 ± 1.6	20.2 ± 1.3	1.5 ± 1.5	17.8 ± 5.0	3.5 ± 1.2
O.Pm/B.Pm, %	3.1 ± 0.9	0.3 ± 0.7	2.7 ± 2.1	0.1 ± 0.3	6.2 ± 4.8	0.6 ± 1.0
O.Wi, μm	2.2 ± 0.4	0.6 ± 0.9	2.0 ± 0.9	0.4 ± 0.7	2.3 ± 0.5	0.9 ± 0.8
W.Wi, μm	9.0 ± 1.6	7.5 ± 0.9	9.3 ± 1.6	7.8 ± 1.2	19.8 ± 3.2	12.1 ± 3.7
N.Ob/N.Oc	5.5 ± 1.5	1.9 ± 0.9	6.0 ± 1.7	1.3 ± 1.2	7.0 ± 4.9	0.4 ± 0.2
N.Ad/T.Ar, /mm2	n.d	n.d.	31.9 ± 19.3	5.4 ± 4.3	n.d.	n.d.
Dynamic indexces
sL.Pm/B.Pm, %	15.8 ± 3.4	8.8 ± 5.4	6.3 ± 3.8	2.3 ± 2.4	16.8 ± 6.1	2.7 ± 1.7
dL,Pm/B.Pm, %	3.6 ± 1.5	0.3 ± 0.4	1.1 ± 0.5	0.1 ± 0.1	3.1 ± 2.3	0.1 ± 0.1
MS/B.Pm, %	11.6 ± 2.7	4.7 ± 3.0	4.3 ± 2.2	1.2 ± 1.2	11.5 ± 4.2	1.4 ± 0.9
MAR, μm/d	1.3 ± 0.2	0.6 ± 0.3	1.2 ± 0.2	0.4 ± 0.2	1.5 ± 0.5	0.5 ± 0.4
BFR/BS, μm3/μm2/d	0.152 ± 0.051	0.031 ± 0.011	0.077 ± 0.065	0.004 ± 0.004	0.171 ± 0.069	0.007 ± 0.010

Values are means ± SD; Number of animals examined is indicated in parentheses Bone sections were obtained from the animals shown in Fig. 3. Abbreviations: number of osteoblasts per mm bone perimeter (N.Ob/B.Pm), osteoid perimeter (O.Pm.), osteoid width (O.Wi), wall width (W.Wi), number of adipocytes per tissue area (N.Ad/T.Ar), number of osteoclasts (N.Oc), single labeled perimeter (sL.Pm), double labeled perimeter (dL.Pm), mineralizing surface (MS), mineral apposition rate (MAR), bone formation rate per bone surface (BFR/BS); n.d. not determined. Bones from all mice exhibited tetracycline labeling, but double labeling was not observed in trabecular vertebral bone from 4 of the mice fed the HFD, and in femoral trabecular and endocortical bone from 7 of the mice fed the HFD. In these cases, MAR and BFR/BS were calculated based on the recommended minimum detectable value of 0.3 μm/day for MAR. Boldface numbers denote a statistically significant effect, P < 0.05, of the HFD by t-test or by nonparametric Mann-Whitney rank sum test.

The number of osteoblasts in femoral and vertebral trabecular bone, as well as in femoral endocortical bone, of HFD-fed mice was 5- to 10-fold lower than in mice maintained on the ND (Table 3). Similarly, wall width (a measure of the amount of bone made by a team of osteoblasts), osteoid surface, and osteoid width were lower in HFD-fed mice. Unlike the typical plump morphology of osteoblasts of mice maintained on a ND, osteoblasts of mice on the HFD were thinner (Fig. 4A). The number of marrow-derived osteoblast progenitors, detected by their ability to form a colony of osteoblastic cells in ex vivo cultures, was twofold lower in HFD-fed mice (ND: 6.0 ± 1.3 colonies per 10⁶ marrow cells; HFD: 2.8 ± 1.1 colonies per 10⁶ marrow cells; P < 0.05). Importantly, the HFD reduced the ratio of osteoblasts to osteoclasts by threefold or greater at all sites examined (Table 3). Although hyperlipidemia might be expected to increase marrow adipocytes, their number decreased in the distal femoral metaphysis of HFD-fed mice (Table 3; Fig. 4B). Few adipocytes were evident elsewhere in femoral marrow or in vertebral marrow of mice fed either diet (not shown). Bone formation rate was greatly attenuated by the HFD in both trabecular and cortical bone (Table 3). This was due to a reduction in both mineralizing surface and mineral apposition rate. Bone sections from some HFD-fed mice exhibited no discernible double-labeled surface, suggesting a decrease in osteoblast function.

Fig. 4.

Histological features of bone of ApoE-KO mice fed ND or HFD. Photomicrographs of nondecalcified vertebral bone sections stained with toluidine blue depicting osteoblasts (red arrowhead) overlying osteoid (black arrows) (A) and adipocytes (yellow arrowheads) in bone marrow (B). Bars = 20 μm.

The effects of HFD on femoral osteoclasts and osteoblasts were examined in a separate experiment comprising three female and three male ApoE-KO mice in each treatment group. Two-way ANOVA disclosed an overall effect of diet, but not sex, on osteoclasts, osteoblasts, and bone formation rate that was similar to the data shown in Table 3 except that endosteal osteoclast number was increased by the HFD in this second study (not shown). However, there was insufficient power to detect statistically significant changes in some of these measurements in males and females when analyzed separately.

HFD-fed ApoE-KO mice exhibited increased circulating levels of CTx, a marker of bone resorption, and decreased levels of osteocalcin, a marker of bone formation (Table 4). Serum levels of Ca, inorganic phosphate, parathyroid hormone (PTH), and creatinine were also increased, consistent with an earlier report of renal damage in HFD-fed ApoE-KO mice (50). Reduced clearance of serum markers of bone resorption and formation has been reported in patients with renal failure (55). Thus, HFD-induced changes in circulating levels of CTx and osteocalcin might not accurately portray the magnitude of the changes in osteoclasts and osteoblasts measured at the tissue level (Table 3). Be that as it may, the HFD evidently overwhelmed any skeletal impact of hyperparathyroidism, as there was no effect on osteoid width, which is diagnostic of the defective mineralization that characterizes renal osteodystrophy (22) (Table 3). Moreover, the HFD did not cause marrow fibrosis (Fig. 4), nor did it increase either osteoclast or osteoblast number (Table 3), which are inexorable features of hyperparathyroidism (36).

Table 4.

Effect of HFD on serum markers of bone remodeling, calcium homeostasis, and kidney function

	ND	HFD
CTx, ng/ml	10.6 ± 3.7 (10)	27.2 ± 9.6* (9)
osteocalcin, ng/ml	42.9 ± 3.7 (10)	19.3 ± 15.2 (9)
Ca, mg/dl	9.1 ± 0.7 (10)	13.0 ± 2.0* (9)
Pi, mg/dl	8.3 ± 1.5 (10)	28.7 ± 6.7* (9)
PTH, pg/ml	158 ± 63 (9)	282 ± 111* (9)
creatinine, mg/dl	0.44 ± 0.16 (9)	2.62 ± 1.44* (8)

Values are means ± SD; number of samples examined is shown in parentheses. Samples were obtained from the animals shown in Table 3.

^*P < 0.01 by t-test.

HFD induces an inflammatory response in the bone marrow.

We next investigated whether, as in the vasculature, the HFD elicits a bone inflammatory response. Transcript levels of IL-1, IL-6, and TNF were increased in bone of HFD-fed mice, as measured by qPCR of RNA extracted from tibiae or vertebrae (Fig. 5, A and B). Moreover, TNF protein levels in femoral marrow plasma increased fivefold (Fig. 5C). FACS analysis of bone marrow cells revealed that the HFD increased the number of monocytes/macrophages (CD11b+) about twofold (Fig. 6, A and B). Although the number of T lymphocytes (CD3+) was unaffected by the HFD, B lymphocytes (CD19+) were markedly reduced. TNF transcripts were highest in macrophages, but expression per monocyte/macrophage was unaffected by the HFD (Fig. 6C). Histological examination of the femoral bone marrow failed to disclose the presence of lipid-filled macrophages (Fig. 4).

Fig. 5.

HFD increases inflammatory cytokines in ApoE-KO mice. Cytokine mRNA levels measured in whole bone extracts of tibiae (A) or vertebrae (B) (ND, n = 8; HFD, n = 9). C: TNF protein levels in femoral marrow plasma (ND, n = 5, HFD n = 9). *P < 0.05 vs. ND by t-test.

Fig. 6.

HFD increases monocytes/macrophages in bone marrow of ApoE-KO mice. A: representative flow cytometry dot plots depicting fluorescence, side scattering (SSC-A), and gates used to quantify percentage of monocyte/macrophages (CD11b+), T-lymphocytes (CD3+), and B-lymphocytes (CD19+) in pooled marrow cells from femora and humeri; 10,000 cells were evaluated. Total cell yield: ND, 41.0 ± 3.1; HFD, 48.8 ± 8.7; n = 4 animals per group. The mean percentage (±SD) of each cell type in the marrow isolate is shown in each panel. B: number of each cell type in total marrow isolate. C: TNF mRNA levels in monocytes/macrophages, T cells, and B cells isolated with immunomagnetic beads. *P < 0.05 vs. ND by t-test.

HFD attenuates expression of Wnt target genes, Wnt ligands, and Wnt antagonists.

The sharp decline in the number of osteoblast progenitors and osteoblasts in HFD-fed mice suggests that the diet interfered with the genesis of these cells. In view of extensive evidence that Wnt signaling plays an essential role in these processes (6), we next investigated whether Wnt signaling was affected by the HFD.

Wnt ligands comprise a family of locally secreted lipidated glycoproteins. Some Wnt ligands activate so-called canonical Wnt signaling by increasing the level of β-catenin, which subsequently binds to and the promotes T cell/lymphoid enhanced binding factor (TCF/LEF)-mediated transcription required for osteoblast differentiation. As shown in Table 5, the HFD caused a reduction in transcript levels of several canonical Wnt target genes, including Cx43, Wisp1, and Tnfrsf11b (aka osteoprotogerin, OPG) in tibiae and vertebrae and in femoral cortical shafts devoid of bone marrow cells. However, Axin2 was unaffected, and expression of Nkd2 was reduced only in femoral shafts. HFD-fed mice also exhibited reduced expression of several Wnt ligands (Table 5), including Wnt6 and Wnt10b, that stimulate osteoblastogenesis by canonical Wnt signaling (7, 9). Expression of Wnt5a, which can affect osteoblastogenesis via a poorly understood noncanonical pathway, (37, 45) was also reduced. Nevertheless, high variance precluded detection of a HFD-induced change in Wnt6 and Wnt5a in vertebrae. Interestingly, expression of sclerostin, Dkk1, and sFRP4, which are secreted antagonists of canonical Wnt signaling, was decreased by the HFD in all bones examined. The HFD also suppressed expression of Wnt4 and Wnt16, but neither is essential for osteoblastogenesis (42, 56).

Table 5.

Effect of HFD on expression of Wnt target genes, Wnt ligands, and Wnt antagonists

	Tibia		Vertebra (L6)		Femoral Shaft
	ND (8)	HFD (9)	ND (8)	HFD (9)	ND (5)	HFD (9)
Wnt target genes
Axin2	1.00 ± 0.40	0.92 ± 0.72	1.00 ± 0.24	1.10 ± 0.23	1.00 ± 0.58	1.92 ± 0.99
Cx43	1.00 ± 0.28	0.38 ± 0.10	1.00 ± 0.18	0.49 ± 0.12	1.00 ± 0.18	0.40 ± 0.14
Nkd2	1.00 ± 0.43	0.69 ± 0.31	1.00 ± 0.19	0.89 ± 0.30	1.00 ± 0.23	0.72 ± 0.21
Tnfrsf11b	1.00 ± 0.37	0.55 ± 0.30	1.00 ± 0.21	0.95 ± 0.24	1.00 ± 0.13	0.45 ± 0.25
Wisp1	1.00 ± 0.41	0.39 ± 0.17	1.00 ± 0.17	0.35 ± 0.14	1.00 ± 0.44	0.17 ± 0.11
Wnt ligands
Wnt4	1.00 ± 0.27	0.57 ± 0.28	1.00 ± 0.54	0.72 ± 0.41	1.00 ± 0.13	0.64 ± 0.19
Wnt5a	1.00 ± 0.25	0.43 ± 0.36	1.00 ± 0.38	0.72 ± 0.58	1.00 ± 0.32	0.23 ± 0.10
Wnt6	1.00 ± 0.50	0.42 ± 0.38	1.00 ± 0.43	1.26 ± 1.07	1.00 ± 0.07	0.61 ± 0.19
Wnt10b	1.00 ± 0.39	0.49 ± 0.37	1.00 ± 0.63	0.43 ± 0.34	1.00 ± 0.39	0.40 ± 0.19
Wnt16	1.00 ± 0.67	0.34 ± 0.24	1.00 ± 0.42	0.44 ± 0.15	1.00 ± 0.59	0.19 ± 0.11
Wnt antagonists
Dkk1	1.00 ± 0.85	0.35 ± 0.33	1.00 ± 0.38	0.28 ± 0.12	1.00 ± 0.43	0.40 ± 0.27
sFRP4	1.00 ± 0.37	0.36 ± 0.15	1.00 ± 0.24	0.42 ± 0.10	1.00 ± 0.28	0.43 ± 0.23
Sost	1.00 ± 0.45	0.34 ± 0.15	1.00 ± 0.32	0.26 ± 0.11	1.00 ± 0.37	0.17 ± 0.11

Values are means ± SD; number of samples examined is shown in parentheses. Transcript levels are expressed relative to samples from animals fed a ND. Boldfaced numbers denote a statistically significant effect of the HFD, as determined by t-test or by nonparametric Mann-Whitney rank sum test.

DISCUSSION

The results presented herein demonstrate that HFD causes a dramatic reduction in osteoblast number and activity in trabecular and endocortical bone of ApoE-KO mice. The number of osteoclasts was also decreased in femoral and vertebral trabecular bone, but the magnitude of this decrease was less than the profound reduction in osteoblasts and bone formation. Osteoclast number in endocortical femoral bone was unaffected by the HFD. Overall, HFD reduced the ratio of osteoblasts to osteoclasts threefold or greater, depending on the site examined. These studies were performed with skeletally mature mice in which balanced remodeling maintains skeletal homeostasis. We therefore conclude that the HFD caused a focal imbalance between osteoclast and osteoblast number, resulting in loss of trabecular and endocortical bone of the femur and loss of trabecular thickness in the spine.

HFD-induced reduction in trabecular and cortical bone, as well as increased CTx and decreased osteocalcin in adult ApoE-KO mice reported here, are consistent with earlier reports in both ApoE-KO and LDL receptor-null mice (24, 48, 49, 53). However, in the earlier studies, the HFD was begun at ∼1 to ∼3 mo of age, when growth and bone formation are high (18). The HFD thus restrained bone acquisition in the earlier work. Consistent with this, bone formation rate is suppressed in growing HFD-fed atherosclerotic LDL receptor-null mice (24, 48). Nevertheless, osteoblast number was unchanged, suggesting that the reduced bone mass is due to attenuated osteoblast function in juvenile mice. These findings stand in contrast to the HFD-induced reduction in both osteoblast number and function in remodeling bone of adult mice reported here. Therefore, the mechanisms responsible for HFD-induced changes in bone mass of growing vs. adult atherosclerotic mice are probably not the same.

In agreement with previous studies (49), we found increased lipid oxidation products in HFD-fed mice, including 4-HNE, which induces oxidative stress. Like oxLDL, 4-HNE inhibits the differentiation and survival of cultured osteoblastic cells (8, 27). Thus, the HFD-induced reduction in osteoblasts and bone formation may be due in part to direct actions of lipid oxidation products on cells of the osteoblast lineage. Perhaps more important, we also found that, as in the vasculature, the proatherogenic diet provokes an inflammatory response in the bone marrow characterized by increased numbers of monocytes/macrophages, expression of inflammatory cytokines, and levels of TNF protein. The latter effect is consistent with a report of increased TNF in serum of HFD-fed LDL receptor-null mice (48). Mice lacking TNF, or functional TNF receptors, exhibit increased bone mass because of increased bone formation (33), indicating that even under normal physiological conditions endogenous TNF restrains bone formation. It is therefore likely that HFD-induced increase in TNF contributes to suppression of bone formation in ApoE-KO mice.

Our studies also identified reduced Wnt signaling as a potential culprit in HFD-induced reduction in bone formation. Increased TNF may be involved in this effect as this cytokine inhibits pro-osteoblastogenic Wnt signaling by increasing oxidative stress, which diverts β-catenin from TCF/LEF- to FoxO-mediated transcription (5). Our findings also support the notion that diminished Wnt signaling in HFD-fed mice results from attenuated expression of the pro-osteoblastogenic ligands Wnt6 and Wnt10b (7, 9). Increased TNF may contribute to decreased synthesis of Wnt ligands in view of evidence that bone marrow cultures established from transgenic mice over-expressing TNF exhibit reduced levels of Wnt10b transcripts (21). Surprisingly however, the HFD also decreased expression of the Wnt antagonists sclerostin, Dkk1 and sFRP4. Thus, this pro-atherogenic diet disrupts both the positive and negative regulators of Wnt signaling that are largely responsible for the strict temporal and spatial control of osteoblast differentiation and bone formation (6), although the underlying mechanisms are unknown. That bone formation is reduced despite loss of the restraining effects of Wnt antagonists on bone formation attests to the potential importance of increased oxidized lipids and TNF, and lower Wnt ligand levels, to the HFD-induced reduction in osteoblast number.

Oxidized phospholipid components of oxLDL may also suppress Wnt signaling via activation of peroxisome proliferator-activated receptor-γ (PPARγ), which attenuates β-catenin/TCF/LEF transcription (1, 4, 47). Moreover, activated PPARγ promotes differentiation of adipocytes at the expense of osteoblasts from a common mesenchymal progenitor. Despite lipid accumulation within in the bone marrow, however, we found that adipocyte number was decreased in bone of HFD-fed ApoE-KO mice. TNF potently inhibits adipocyte differentiation and stimulates lipolysis (10), suggesting that HFD-induced inflammation prevents the increase in marrow adipocyte number that usually heralds activation of PPARγ.

The increased lipid oxidation in bone of HFD-fed ApoE-KO mice was associated with higher levels of Alox15, probably resulting from an increase in the number of macrophages expressing this enzyme. Nevertheless, deletion of Alox15 did not attenuate HFD-induced bone loss in ApoE-KO mice, but it did reduce vascular calcification, in agreement with previous reports (11, 20, 26). Thus, although osteoporosis and atherosclerosis may share some oxLDL-activated pathogenic factors such as increased inflammatory cytokines, the factors involved in the oxidation of LDL in the two conditions clearly differ. This may reflect a more diverse repertoire, if not higher levels, of lipoxidative enyzmes in bone and bone marrow compared with cells in the subendothelial space of the vasculature. Our observation that marrow of HFD-fed mice lacked the foam cells that characterize atherosclerotic plaques is another indication that the response of the two tissues to elevated LDL is quite different. Tissue-specific differences in the properties of macrophages could be involved, in view of recent evidence for the ontogenic and functional diversity of this cell type (32).

Similar to an earlier report in LDL receptor-null mice (48), we observed increased serum levels of creatinine, Ca, Pi, and PTH in HFD-fed ApoE-KO mice fed the same HFD, which contains cholate in order to increase fat absorption from the gut. This diet has been shown to cause renal damage in ApoE-KO mice, as evidenced by macrophage infiltration and increased inflammatory cytokine expression (50). In contrast, LDL receptor-null mice fed a cholate-free HFD do not exhibit renal damage or increased serum Ca and PTH; however, this HFD in combination with partial nephrectomy causes a rise in both (16). Thus, hypercalcemia and hyperparathyroidism occur in HFD-fed murine models of atherosclerosis regardless of whether reduced renal function results from surgical or chemical (cholate) intervention, albeit the hypercalcemia appears more extreme in the latter. Phosphate binders lower both Ca and PTH in nephrectomized LDL receptor-null mice fed a cholate-free HFD (13, 39), indicating that these mice experience secondary rather than primary hyperparathyroidism. By analogy, secondary hyperparathyroidism may have been responsible for the disturbed Ca homeostasis in ApoE-KO mice (reported herein) and in LDL receptor-null mice (48) fed cholate-containing HFD. Nevertheless, it is unclear why PTH remains high in the face of elevated serum Ca in these models. It is possible that the underlying molecular and cellular mechanisms leading to elevated PTH levels differ depending on the nature of the renal insult.

The deleterious effects of oxLDL on bone are not caused solely by kidney damage, since bone loss due to a decrease in osteoblast number also occurs in cholate-free HFD-fed LDL receptor-null mice with intact renal function (16). In these mice, partial nephrectomy not only increases Ca, Pi, and PTH but also further reduces osteoblast number, bone formation rate, and bone mass (13, 16). Interestingly, these changes are associated with a dramatic, but transient, increase in the level of circulating Dkk1, which is apparently produced by regenerating renal tissue. The deleterious skeletal effects of the combined cholate-free HFD and partial nephrectomy were attenuated by administration of a neutralizing anti-Dkk1 monoclonal antibody. Thus, the suppressive effect of cholate-containing HFD on osteoblast number and bone formation reported herein may have been amplified by increased circulating levels of this Wnt antagonist.

The HFD-induced reduction in osteoclasts in both femoral and vertebral trabecular bone noted in the present study is unexpected in the face of increased circulating PTH and inflammatory cytokines and decreased expression of osteoprotegerin, all of which stimulate osteoclast differentiation (23, 43). The HFD must therefore have negated the impact of these pro-osteoclastogenic signals specifically within the microenvironment of trabecular bone. In vitro studies have given conflicting results as oxLDL inhibits RANKL-induced differentiation of the RAW 264.6 pre-osteoclastic cell line (41), but increased basal levels of pro-osteoclastogenic RANKL in MG-63 human osteosarcoma cells (40). However, it is possible that the HFD-induced decrease in the expression of Wnt5a is responsible for the decrease in trabecular osteoclasts since this Wnt ligand promotes osteoclastogenesis in trabecular but not cortical bone (37).

In summary, we have provided evidence that HFD-induced bone loss in adult ApoE-KO mice is the result of compromised osteoblast supply and bone formation. This is probably caused by oxLDL-induced bone inflammation, increased TNF production, and reduced Wnt ligand expression. Taken together with evidence for reduced circulating levels of osteocalcin in patients with vascular calcification (30), we propose that similar changes underlie the pathogenesis of osteoporosis in atherosclerotic patients. Increased lipid oxidation and chronic low-level inflammation are prominent hallmarks of aging (19, 35). Therefore skeletal inflammation may also contribute to the reduced number and activity of osteoblasts that characterize involutional osteoporosis (3, 38).

GRANTS

This work was supported by the Biomedical Laboratory Research and Development Service of the Veterans Administration Office of Research and Development (I01 BX000514 to R. L. Jilka), the National Institutes of Health (P01 AG-13918 to S. C. Manolagas), and the University of Arkansas for Medical Sciences Tobacco Funds and Translational Research Institute (1UL1 RR-029884).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: Y.L. performed experiments; Y.L. and R.L.J. analyzed data; Y.L. and R.L.J. prepared figures; Y.L. and R.L.J. drafted manuscript; Y.L., M.A., R.S.W., C.A.O., S.C.M., and R.L.J. approved final version of manuscript; M.A., C.A.O., S.C.M., and R.L.J. conception and design of research; M.A., R.S.W., C.A.O., and R.L.J. interpreted results of experiments; M.A., R.S.W., C.A.O., S.C.M., and R.L.J. edited and revised manuscript.