Calcified Tissue International

, Volume 82, Issue 5, pp 361–372

Soluble RANKL Induces High Bone Turnover and Decreases Bone Volume, Density, and Strength in Mice

Authors

  • S. A. J. Lloyd
    • Department of BioengineeringClemson University
  • Y. Y. Yuan
    • Department of BioengineeringClemson University
  • P. J. Kostenuik
    • Metabolic DisordersAmgen, Inc.
  • M. S. Ominsky
    • Metabolic DisordersAmgen, Inc.
  • A. G. Lau
    • Department of BioengineeringClemson University
  • S. Morony
    • Metabolic DisordersAmgen, Inc.
  • M. Stolina
    • Metabolic DisordersAmgen, Inc.
  • F. J. Asuncion
    • Metabolic DisordersAmgen, Inc.
    • Department of BioengineeringClemson University
Article

DOI: 10.1007/s00223-008-9133-6

Cite this article as:
Lloyd, S.A.J., Yuan, Y.Y., Kostenuik, P.J. et al. Calcif Tissue Int (2008) 82: 361. doi:10.1007/s00223-008-9133-6

Abstract

Receptor activator for nuclear factor-κ B ligand (RANKL) is an essential mediator of osteoclastogenesis. We hypothesized that administration of soluble RANKL to mice would result in high turnover and deleterious effects on both cortical and trabecular bone. For 10 days, 10-week-old C57BL/6J female mice (n = 12/group) were given twice-daily subcutaneous injections of human recombinant RANKL (0.4 or 2 mg/kg/day) or inert vehicle (VEH). Bone turnover was greatly accelerated by RANKL, as evidenced by the 49–84% greater levels of serum TRAP-5b (bone resorption marker) and 300–400% greater levels of serum alkaline phosphatase (bone formation marker). RANKL resulted in significantly greater endocortical bone erosion surface (79–83%) and periosteal bone formation rate (64–87%) vs. VEH. Microcomputed tomographic (microCT) analysis of the proximal tibia indicated a reduction in trabecular volume fraction (–84%) for both doses of RANKL. Cortical bone geometry and strength were also negatively influenced by RANKL. MicroCT analysis of the femoral diaphysis indicated significantly lower cortical bone volume (−10% to –13%) and greater cortical porosity (8–9%) relative to VEH. Biomechanical testing of the femur diaphysis revealed significantly lower maximum bending load (−19% to –25%) vs. VEH. Bone strength remained correlated with bone mass, independent of RANKL stimulation of bone turnover. These findings are consistent with the hypothesis that soluble RANKL could be an important etiologic factor in pathologic bone loss. RANKL also has potential utility as a model for studying the consequences of high bone turnover on bone quality and strength in animals.

Keywords

RANKLBone strengthTurnoverMicrocomputed tomographyOsteoporosis

Bone strength is determined by bone mass and bone quality [1, 2]. Bone mass reflects the balance between bone formation and bone resorption, which involves cellular regulation of osteoblast/osteoclast number and activity. Moreover, the initiation of bone formation and resorption is coupled: cells of the osteoblast lineage regulate the recruitment and activity of osteoclasts through expression of receptor activator for nuclear factor-κ B ligand (RANKL) and osteoprotegerin (OPG) [3, 4]. Balanced RANKL/OPG levels may be critical for maintaining precise, homeostatic control over bone remodeling [36]. Bone quality is determined by a number of variables, including microarchitecture, microdamage, degree of mineralization, and bone turnover. Abnormally high turnover reduces bone mass and may also reduce bone quality by accelerating the normal resorption and formation phases of the remodeling cycle, leading to reduced matrix mineralization [2, 7].

RANKL is an essential mediator of osteoclast formation, activation, and survival [3]. RANKL binds to its receptor, RANK, on the surface of osteoclasts, leading to their activation and differentiation and the subsequent induction of bone resorption [3, 810]. OPG, a member of the tumor necrosis factor (TNF) receptor superfamily, is a soluble decoy receptor that inhibits the interaction of RANKL with its receptor, RANK [4, 11]. Transgenic mice overexpressing RANKL and OPG knockout mice (OPG–/–) both develop severe osteoporosis, accompanied by high bone turnover, low bone mineral density (BMD), and increased cortical porosity [3, 12].

RANKL exists in both soluble and membrane-bound forms. Membrane-bound RANKL, a member of the membrane-associated TNF family, is expressed on the surface of osteoblasts and stromal cells [3, 6]. Both forms of RANKL mediate bone resorption by binding to RANK on osteoclasts. In osteoblast–osteoclast coculture systems, direct cell-to-cell contact facilitates osteoclast formation via RANKL–RANK interaction [3, 13]. RANKL can also be directly secreted in a soluble form [12, 14, 15]. Previous in vitro studies have suggested that membrane-bound RANKL might be more potent than soluble RANKL in its induction of osteoclastogenesis [13, 14]. A role for soluble RANKL in bone disease is suggested by the associations between increased soluble RANKL levels and bone loss in nonhuman animals [15] and in certain human disease states [16, 17]. However, the precise contribution of soluble vs. membrane RANKL to the overall regulation of bone remodeling has not been properly elucidated.

Increased RANKL has been demonstrated in patients with postmenopausal osteoporosis [18] and in ovariectomized rodents [19]. It is possible that RANKL mediates some of the skeletal changes associated with estrogen deficiency as the inhibition of RANKL was shown to increase BMD in cases of osteoporosis [20] and ovariectomy [19, 21]. A recent study from our laboratory investigated the effects of continuous RANKL administration to rats via a subcutaneous osmotic pump [22]. Over a 4-week period we observed significant declines in cortical bone volume and strength. In the present study, it was hypothesized that administration of soluble recombinant RANKL would reproduce many of the skeletal changes that were described in this rat study and in other investigations of OPG knockout mice and estrogen-deficient osteopenia in humans, including deleterious effects on bone volume, geometry, density, and strength [23, 24]. Confirmation of this hypothesis would further reinforce the potential of RANKL to serve as an accelerated, surgery-free model for bone pathologies characterized by high turnover.

Materials and Methods

Study Design

Thirty-six female C57BL/6J mice (Jackson Laboratory, Bar Harbor, ME), aged 10 weeks, were assigned to one of three groups (n = 12/group): VEH (vehicle, phosphate-buffered saline [PBS]), LOW (0.4 mg/kg/day RANKL), or HI (2 mg/kg/day RANKL). The RANKL used in this study was a recombinant human form produced by Amgen (Thousand Oaks, CA). The HI dose was selected as half the maximum dose administered in a previous study by Lacey and colleagues [18], while the LOW dose was calculated as 20% of HI. For 10 days, all mice received twice-daily subcutaneous injections of RANKL or VEH (approximately 0.2 mL/injection). Body weight was measured every 2 days, and drug volume was adjusted accordingly. In order to monitor new bone growth, the fluorescent bone label calcein (20 mg/kg) was subcutaneously injected on day 2. On day 10, animals were anesthetized with isoflurane and killed by cardiac puncture exsanguination, followed by cervical dislocation. Both hindlimbs were removed and cleaned of all nonosseous tissue. All animal procedures were approved by the Institutional Animal Care and Use Committee at Clemson University (Clemson, SC).

Serum Bone-Turnover Markers

Markers for bone formation and resorption were measured from serum collected when animals were killed. Serum alkaline phosphatase (ALP) and total calcium and phosphorus levels were measured with an automated chemistry analyzer (Hitachi, Palo Alto, CA). Serum tartrate-resistant acid phosphatase-5b (TRAP-5b), a marker of bone resorption, was measured by enzyme-linked immunosorbent assay (SBA Science/IDS, Turku, Finland).

Microcomputed Tomography

Microcomputed tomographic (microCT) images were generated from air-dried right femurs (9 μm isotropic voxel size). Cortical and trabecular morphometric parameters were determined using a μCT20 scanner (Scanco, Bassersdorf, Switzerland) [25, 26]. Cortical bone parameters were obtained by scanning an 8-mm longitudinal section of the femoral diaphysis (the same span length examined in later mechanical testing). A total of 81 slices were analyzed, with 100-μm increments between each slice. To determine bone volume and polar moment of inertia (pMOI), contours were traced at the periosteal surface and calculated by Scanco IPL-Moment software. To quantify the porosity of femoral cortical bone, two contours were traced on the periosteal and endocortical surfaces of each slice and the enclosed bone was evaluated by Scanco software. During evaluation, the femur diaphysis was separated into three sections: proximal diaphysis (2.5 mm long, slices taken at the level of the third trochanter), mid-diaphysis (3 mm long), and distal diaphysis (2.5 mm long). Porosity data were also obtained for each of these sections.

Right tibiae were fixed in 10% neutral buffered formalin for 2 days, rinsed with distilled water, and stored in 70% ethanol. Trabecular bone parameters were obtained with axial microCT scans (100 slices, 9 μm each) of trabecular bone at the proximal end of the tibia, immediately distal to the growth plate (i.e., the metaphysis). Parameters measured included trabecular bone volume (BV), total volume (TV), and volume fraction (BV/TV).

Bone Mineral Content Analysis

Mineral content analysis was performed on the fractured femur. Prior to analysis, the metaphyses at both the proximal and distal ends were separated. Mineral content data were obtained separately from the metaphyses and the diaphysis. Dry mass (Dry-M) was measured after heating the bones for 24 hours at 105°C. Mineral mass (Min-M) was measured after the bones had been heated for an additional 24 hours at 800°C. Percent mineralization (%Min) was calculated as %Min = (Min-M/Dry-M) * 100.

Quantitative Histomorphometry

Left femora were placed in 10% neutral-buffered formalin for 48 hours, rinsed with distilled water, and stored in 70% ethanol. All bones were allowed to air-dry and then embedded with noninfiltrating Epo-Kwick epoxy (Buehler, Lake Bluff, IL). The epoxy disks were sectioned with a low-speed saw (Buehler, 12.7 cm × 0.5 mm diamond blade) at the mid-diaphysis of the femur. The sections were wheel-polished to a flat, smooth surface with 600-, 800-, and 1,200-grit carbide paper, followed by a cloth impregnated with 6 μm diamond paste. This allowed micrographs at x50 magnification to be taken of bone cross sections under a blue light (400 nm). Green calcein labels were visualized, indicating the presence of bone formation sites during the period of study. Quantitative histomorphometric analysis was performed using these photographs and SigmaScan Pro software (SPSS, Inc., Chicago, IL).

Measurements of cortical bone morphology [27] included tissue volume, medullary volume (Me.V), and BV (calculated as tissue volume – Me.V). Endocortical and periosteal bone surfaces were also measured (Ec.BS and Ps.BS, respectively). The green calcein label injected at day 2 of the study allowed measurement of bone formed between the label and bone surface. Bone formation rate (BFR) was calculated by dividing the volume of new bone formed in 8 days and is reported as surface referent (BFR/BS). The length of the labeled surface was determined to be the mineralizing surface (Ps.MS and Ec.MS) and referenced by the corresponding bone surface (MS/BS). Mineral apposition rate (MAR) was calculated as BFR/MS for the periosteal (Ps.MAR) and endocortical (Ec.MAR) surfaces. Endocortical eroded surface (Ec.ES) was estimated by quantifying the portion of the nonlabeled eroded surface and is referenced to bone surface (Ec.ES/BS).

Biomechanical Testing

Mechanical properties of right femora were tested following microCT analysis. In order to simulate in vivo properties, all bones were rehydrated in PBS for 90 minutes prior to mechanical testing [28]. Three-point bending tests were performed using an Instron 5582 and Merlin, series IX, software (Instron, Norwood, MA). Femora were tested to failure with an 8 mm span length and deflection rate of 5 mm/minute. Elastic force (Pe, N) and deflection (δe, mm) were measured at the elastic limit. Maximum and structural failure (fracture) loads were also measured. Stiffness (S) was calculated from Pee.

Statistics

Unless otherwise indicated, all statistical comparisons were performed using SigmaStat software (Systat Software, San Jose, CA) and a one-way analysis of variance (ANOVA) with a Tukey test for follow-up comparisons. Differences in starting and final animal mass were made using a paired t-test. A 95% level of significance (type I error) was used for all tests. Linear regressions were performed and coefficients of determination (r2) obtained from a Pearson product moment correlation test. Data are presented as mean ± standard error (SE), unless otherwise indicated.

Results

Animal Mass and Serum Chemistry

Mice treated with vehicle or low-dose RANKL showed no significant change in body mass or serum calcium level during the 10-day study (> 0.05). In the HI group, 11% weight loss was observed at death (< 0.001 vs. VEH, Table 1); this was accompanied by symptoms of lethargy, suggestive of distress from hypercalcemia. Total serum calcium levels were 16% greater in the HI group vs. VEH (< 0.001). Serum phosphorus levels in the HI group were 20% greater than VEH (< 0.001), while those in the LOW group remained unchanged (> 0.05) (Table 1). Serum calcium, serum phosphorus, and loss of body mass were significantly greater in mice in the HI group compared to the LOW group (< 0.001).
Table 1

Animal body mass and serum chemistry values

 

Vehicle

0.4 mg/kg RANKL (LOW)

2 mg/kg RANKL (HI)

Starting body mass (g)

18.6 ± 0.4

18.1 ± 0.2

18.5 ± 0.3

Final body mass (g)

18.8 ± 0.3

18.5 ± 0.2

16.5 ± 0.4*

Change in body mass (%)

1.4 ± 1.4

2.3 ± 0.9

–10.9 ± 1.6*

Serum calcium (mg/dL)

8.9 ± 0.1

8.4 ± 0.1

10.3 ± 0.1*

Serum phosphorus (mg/dL)

6.9 ± 0.3

6.8 ± 0.2

8.3 ± 0.4*

Data are presented as means ± SE

* P < 0.001 vs. VEH and 0.4 mg/kg RANKL (LOW). Differences in starting and final animal mass were made using a paired t-test

Serum Bone-Turnover Markers

Biochemical markers of bone turnover were increased following 10-day treatment with RANKL. The serum bone formation marker ALP was found to be 224% and 321% greater than VEH for LOW and HI dose RANKL, respectively (< 0.001 for both) (Fig. 1A). The bone resorption marker serum TRAP-5b was 94% and 49% greater than VEH in the LOW and HI groups, respectively (< 0.05 for both) (Fig. 1B).
https://static-content.springer.com/image/art%3A10.1007%2Fs00223-008-9133-6/MediaObjects/223_2008_9133_Fig1_HTML.gif
Fig. 1

RANKL administration resulted in greater levels of bone formation markers vs. vehicle control. (A) Compared to VEH, the bone formation marker serum ALP was threefold greater in 0.4 mg/kg RANKL (LOW) and fourfold greater in 2 mg/kg RANKL (HI) mice after 10 days of twice-daily injections. (B) The bone resorption marker serum TRAP-5b was 84% and 49% greater than VEH in low- and high-dose RANKL mice, respectively. Data are presented as means ± SE. * P < 0.05 vs. VEH, # < 0.05 vs. LOW

MicroCT

Trabecular volume fraction (BV/TV) at the proximal tibia (Fig. 2A) was decreased by 84% in both RANKL groups compared to VEH (< 0.001 for both, Fig. 2B). At the femur diaphysis, cortical BV in the LOW and HI groups was significantly lower than VEH (13% and 9%, respectively; < 0.001 for both) (Fig. 2C). In the central 3-mm region of the diaphysis, average cortical area was significantly lower than VEH in both the LOW and HI groups (9% and 6%, respectively; < 0.05 for both). Despite the decrease in cortical area, no significant difference was observed between VEH and either of the RANKL groups for pMOI at the femur midshaft (> 0.05). The resorption biomarker serum TRAP-5b was significantly negatively correlated with tibia metaphyseal BV/TV (r= 0.76, < 0.01) (Fig. 2D) and femur cortical area (r= 0.35, < 0.01) (Fig. 2E), demonstrating the impact of increased bone resorption on trabecular and cortical BV.
https://static-content.springer.com/image/art%3A10.1007%2Fs00223-008-9133-6/MediaObjects/223_2008_9133_Fig2_HTML.gif
Fig. 2

RANKL resulted in reduced trabecular and cortical volume vs. vehicle control. (A) Three-dimensional microCT images of 2-mm sections of trabecular bone at the proximal tibia illustrate severe bone loss after RANKL administration. (B) Trabecular bone parameters were measured from a 0.9-mm longitudinal section of trabecular bone at the proximal tibia metaphysis. RANKL reduced trabecular bone fraction (BV/TV) in both the LOW and HI groups by 84% (P < 0.001 for both vs. VEH). (C) Cortical bone volume measured in the central 3 mm of femur diaphysis was significantly reduced by 9% and 6% in the 0.4 mg/kg/day RANKL (LOW) and 2 mg/kg/day RANKL (HI) groups, respectively (< 0.05 vs. VEH for both). The bone resorption marker serum TRAP-5b was inversely correlated with (D) proximal tibia BV/TV and (E) femur cortical area (significant correlations, < 0.01). Data are presented as means ± SE. * Significant difference vs. VEH

After separating the femur diaphysis into proximal, mid-, and distal diaphysis components, RANKL was found to have site-specific effects on porosity. In the distal femur, porosity was 39% greater than VEH for the HI group (< 0.001) (Fig. 3A). In the central 3 mm of the femur diaphysis, there was a trend toward greater porosity for both RANKL groups (each +7% vs. VEH, ANOVA = 0.08) (Fig. 3B). The greatest effect was at the third trochanter (proximal), greater than VEH in both the LOW and HI groups, resulting in 55% and 82% greater porosity, respectively, vs. VEH (< 0.001 for both) (Fig. 3C). Cortical porosity of the total femur diaphysis was found to be 26% and 45% greater in the LOW and HI groups (< 0.001 vs. VEH for both) (Fig. 3D). Cortical porosity of the central and total femur diaphysis was significantly positively correlated with serum TRAP-5b (r= 0.19 and 0.29, respectively; < 0.02) (Fig. 3E, F), demonstrating the impact of greater bone resorption on cortical porosity.
https://static-content.springer.com/image/art%3A10.1007%2Fs00223-008-9133-6/MediaObjects/223_2008_9133_Fig3_HTML.gif
Fig. 3

RANKL resulted in increased cortical porosity. By dividing the femur diaphysis (8-mm-long scan area) to distal diaphysis (distal 2.5 mm), mid-diaphysis (middle 3 mm), and proximal diaphysis (proximal 2.5 mm), local changes in porosity were observed. (A) At the distal femur, 2 mg/kg/day RANKL (HI) resulted in 39% greater porosity (P < 0.001 vs. VEH); the change in the 0.4 mg/kg/day RANKL (LOW) group was not significant. (B) There was a trend toward greater porosity after RANKL administration at the mid-diaphysis (ANOVA P = 0.08). (C) The LOW and HI groups showed 55% and 82% greater porosity, respectively, at the proximal diaphysis (trochanter) (P < 0.001). (D) Total femur diaphyseal porosity was 26% and 45% greater in the LOW and HI groups, respectively (P < 0.001 vs. VEH). The bone resorption marker serum TRAP-5b was positively correlated with cortical porosity at the (E) central 3 mm of the femur diaphysis and (F) total femur (significant correlation, P < 0.01). Tibial two-dimensional microCT images from 4 mm distal to the tibial plateau (G) and 2 mm distal to the tibial plateau (H) at the proximal tibia were taken to demonstrate cortical porosity. Data are presented as means ± SE. * P < 0.001 vs. VEH, # P < 0.05 vs. LOW

Bone Mineral Content Analysis

Whole femur dry mass measured in the LOW and HI groups was significantly lower than VEH (−12% and −17%, respectively; < 0.001). The reductions in Dry-M and Min-M at the metaphyses were greater than in the diaphysis after RANKL administration. In the femoral metaphyses, Dry-M was reduced by 13% and 22% in the LOW and HI groups and Min-M was significantly reduced by 20% and 32% in the LOW and HI groups, respectively (all comparisons < 0.05 vs. VEH). In the femoral diaphyses, Dry-M was reduced by 12% and 9% in the LOW and HI groups and Min-M was significantly reduced by 16% and 13% in the LOW and HI groups, respectively (all comparisons < 0.05 vs. VEH). Whole femur percent tissue mineralization, the ratio of mineral to dry mass, was 6% and 7% lower than VEH for LOW and HI doses, respectively (< 0.05 for both) (Fig. 4A). These changes were attributed to significantly lower tissue mineralization in the femur metaphyses, −9% and −13% for the LOW and HI groups, respectively (< 0.001 vs. VEH for both) (Fig. 4B), and femur diaphysis (−4% and −3% for LOW and HI, respectively; < 0.05 vs. VEH for both) (Fig. 4C). Tissue mineralization was significantly negatively correlated with serum TRAP-5b (r= 0.45 for whole femur, r= 0.29 for central diaphysis; < 0.01 for both) (Fig. 4D), demonstrating the impact of greater bone resorption on cortical bone mineralization.
https://static-content.springer.com/image/art%3A10.1007%2Fs00223-008-9133-6/MediaObjects/223_2008_9133_Fig4_HTML.gif
Fig. 4

RANKL results in lower percent tissue mineralization in the femur. (A) Whole femur %Min was 6% and 7% less than VEH for LOW and HI RANKL, respectively. (B) At the femur metaphyses, %Min was 9% and 13% lower in the 0.4 mg/kg/day RANKL (LOW) and 2 mg/kg/day RANKL (HI) groups, respectively. (C) At the femur diaphysis, ash analysis revealed a 3–4% decrease in %Min after 10 days of RANKL administration. (D) The bone resorption marker serum TRAP-5b was inversely correlated with %Min at the diaphysis (significant correlation, < 0.01). Data are presented as means ± SE. * < 0.05 vs. VEH. # < 0.05 vs. LOW

Quantitative Histomorphometry

Quantitative histomorphometric analysis of the femur at mid-diaphysis (Table 2) revealed that endocortical bone resorption was greater than VEH for both RANKL-treated groups. Medullary volumes of the LOW and HI groups were found to be 9% and 8% greater than VEH, respectively (< 0.05 for both). Endocortical eroded surface was 75% and 73% greater than VEH for the LOW and HI groups, respectively (< 0.001). This greater erosion surface was associated with 6% and 8% lower cortical volume in the LOW and HI groups, respectively (< 0.05 for both vs. VEH).
Table 2

Quantitative histomorphometric results at femur mid-diaphysis

Measurement

Vehicle

RANKL-treated

0.4 mg/kg (LOW)

2 mg/kg (HI)

Volume (mm3)

    Tissue

1.55 ± 0.02

1.57 ± 0.02

1.56 ± 0.01

    Medullary

0.90 ± 0.01

0.98 ± 0.02**

0.97 ± 0.01**

    Bone

0.64 ± 0.02

0.60 ± 0.01**

0.59 ± 0.01**

Ct.Th (μm)

151 ± 4

136 ± 2**

135 ± 2**

BFR/BS (μm3/μm2/day)

    Endocortical

1.02 ± 0.09

0.76 ± 0.04**

0.80 ± 0.08

    Periosteal

0.34 ± 0.04

0.64 ± 0.07*

0.56 ± 0.04**

MS/BS (%)

    Endocortical

60.7 ± 2.2

36.8 ± 1.5*

40.3 ± 3.2*

    Periosteal

23.6 ± 1.2

28.6 ± 2.1

28.5 ± 1.1

MAR (μm/day)

    Endocortical

1.68 ± 0.10

2.06 ± 0.07**

1.96 ± 0.11***

    Periosteal

1.42 ± 0.11

2.15 ± 0.17*

1.96 ± 0.11**

Ec.ES/BS (%)

30.0 ± 2.3

52.4 ± 2.2*

51.9 ± 3.0*

Data are presented as means ± SE

< 0.001 vs. VEH, ** P < 0.05 vs. VEH, *** P < 0.08 vs. VEH

Based on histomorphometric analysis, cortical bone formation was greater than VEH control following RANKL administration. Significant elevations in periosteal bone formation (Ps.BFR/BS) with low-dose (+100%, < 0.001) and high-dose (+76%, < 0.05) were found in mice treated with RANKL compared to VEH. Endocortical bone formation was significantly lower (−26%, < 0.05) in the low-dose RANKL group vs. VEH, although no significant difference was observed for the high-dose group. Endocortical mineralizing surface was significantly lower than VEH for RANKL-treated mice at both doses (< 0.001 vs. VEH), while periosteal mineralizing surface was not significantly greater. Compared to VEH, RANKL administration was associated with greater mineral apposition rate at both the endocortical and periosteal surfaces, with values 23% (< 0.05) and 51% (< 0.001) greater, respectively, in the LOW group, and 17% (P = 0.08) and 38% (< 0.05) greater, respectively, in the HI group.

Biomechanical Testing

Both doses of RANKL resulted in significantly lower maximum strength and stiffness when compared to VEH (Table 3). Maximum bending loads of femoral diaphyses were 25% (LOW) and 19% (HI) lower than VEH, while structural stiffness was 39% (LOW) and 37% (HI) lower than VEH (< 0.001 for all). LOW RANKL administration resulted in significantly lower femoral fracture force (−30%) compared to VEH (< 0.001). We assessed the role of cortical bone volume, porosity, and tissue mineralization in determining bone strength in treated and untreated rat femurs by regression analysis. Positive correlations with femur maximum load were found for microCT-derived cortical area (Fig. 5A, r= 0.61) and ash-derived tissue mineralization (Fig. 5B, r= 0.34), while cortical porosity was negatively associated with load (Fig. 5C, r= 0.45) (< 0.001 for all correlations). The product of cortical area and percent mineralization provided an improved surrogate for femur maximum load (Fig. 5D, r= 0.75).
Table 3

Femur mechanical properties

Strength parameter

Vehicle (VEH)

0.4 mg/kg RANKL (LOW)

2 mg/kg RANKL (HI)

Stiffness (N/mm)

45.1 ± 2.0

27.8 ± 2.6*

28.2 ± 2.4*

Elastic force (N)

8.3 ± 0.4

7.0 ± 0.5

7.2 ± 0.5

Maximum force (N)

11.7 ± 0.3

8.8 ± 0.2*

9.5 ± 0.4*

Fracture force (N)

8.9 ± 0.8

6.1 ± 0.6**

6.9 ± 0.7

Data are presented as means ± SE

* < 0.001 vs. VEH, ** < 0.05 vs. VEH

https://static-content.springer.com/image/art%3A10.1007%2Fs00223-008-9133-6/MediaObjects/223_2008_9133_Fig5_HTML.gif
Fig. 5

Regression analyses of femur diaphyseal maximum load vs. cortical area and tissue mineralization in mice with and without 10-day RANKL administration. Cortical area (A, from microCT) and percent tissue mineralization (B, from ash analysis) were positively correlated with femur strength, while cortical porosity (C) was negatively associated with strength. (D) The product of cortical area and mineralization further improved the correlation with maximum load (all correlations < 0.001)

Discussion

Soluble RANKL is known to stimulate bone resorption and remodeling in mice. In the present study, we examined the consequences of RANKL-mediated bone turnover on the volume, geometry, density, and strength of mouse bone. A 10-day course of RANKL administration created a severely osteopenic phenotype that recapitulated many of the deleterious skeletal changes associated with postmenopausal osteoporosis.

High bone turnover and the attendant reduction in bone mass are thought to contribute to increased fracture incidence in women with postmenopausal osteoporosis [29]. One potential etiologic factor that may contribute to this high turnover state is RANKL [18], a TNF family member that is essential for osteoclast activation, differentiation, and survival [3, 8]. There are numerous animal models for studying the impact of high-turnover bone diseases on bone quality and bone strength, with ovariectomy (OVX) being the most commonly used. Despite their widespread use as an osteoporosis model, OVX rats do not typically show increases in cortical bone remodeling or reductions in cortical bone strength [30]. Mice have variable responses to OVX depending on their genetic background, but intracortical remodeling does not occur after OVX in most strains [31]. Cortical bone loss was modest in most strains of mice 4 weeks after OVX [32], as were reductions in cortical bone strength [31]. These findings stand in apparent contrast to the increased bone turnover and fracture incidence found at cortical sites in humans with postmenopausal osteoporosis. Compared to OVX in mice, the skeletal phenotype of OPG knockout mice tends to more faithfully model important postmenopausal changes in bone, including increased cortical porosity, reduced bone strength at cortical sites, and spontaneous fragility fractures [23, 24, 33]. We hypothesized that injections of soluble recombinant RANKL could create a skeletal phenotype similar to that of the OPG knockout mice, thereby creating a novel, practical, and nonsurgical animal model for osteoporosis research.

The effects of recombinant RANKL on bone mineralization, formation, geometry, and strength in mice have not been previously described. An interesting observation from previous studies involving transgenic mice overexpressing RANKL was that the deleterious skeletal effects appeared to be less severe than the phenotype associated with OPG knockout mice. This could imply that the local and systemic absence of OPG, a RANKL inhibitor, resulted in skeletal changes that cannot be mimicked by systemic exposure to excess soluble RANKL. Alternatively, transgene expression may have been too modest in these animals to fully overcome the local inhibitory effects of OPG. The latter possibility is supported by data from the present study, whereby direct injection of high-dose soluble RANKL produced a skeletal phenotype that was similar to that associated with the total ablation of OPG [23, 24, 33, 34]. In the current mouse study, administration of soluble RANKL resulted in greater bone turnover, lower cortical and trabecular bone volume and mineralization, and lower cortical strength compared to control. Thus, while the osteoporotic phenotype in OPG knockout mice arises from unopposed activity of both membrane-bound and soluble RANKL, their phenotype can be faithfully mimicked by excess soluble RANKL in normal mice. These observations are consistent with the possibility that soluble RANKL is an important mediator of bone resorption in normal animals.

The increase in local and systemic bone turnover with RANKL was demonstrated by the greater serum bone turnover markers, endocortical eroded surface, and mineral apposition rate compared to vehicle control. Greater levels of the bone resorption marker TRAP-5b were expected due to the stimulation of osteoclasts by RANKL. The RANKL-mediated increase in the bone formation marker ALP can be attributed to the normal physiologic coupling between osteoblasts and osteoclasts, which has been demonstrated previously [23, 33]. The difference in serum markers between the LOW and HI groups is likely not practically relevant, although it is not without precedence for endogenous proteins to elicit variable responses depending on dose frequency and magnitude. Such is the case with parathyroid hormone, which is anabolic when given intermittently and catabolic when given continuously [35]. Similarly, macrophage-colony stimulating factor stimulates osteoclast activation at low doses but is inhibitory at higher doses [36].

The systemic increases in serum markers of bone resorption and formation were also reflected locally at the tissue level. Endocortical eroded surface was greater in RANKL-treated mice vs. control, coinciding with lower endocortical mineralizing surface, thus reflecting a shift toward a negative balance of bone turnover. The greater mineral apposition rate at both periosteal and endocortical surfaces in the RANKL groups could reflect a compensatory mechanism to counteract the deleterious effects of increased bone turnover on bone volume, mineralization, and strength. Increases in tissue-level stresses have also been reported to stimulate bone formation in mice [37]. Alternatively, increased periosteal bone formation could have resulted from coupling-related stimulation of endocortical bone resorption by RANKL.

Many of the cortical changes associated with RANKL were reminiscent of changes observed in humans during aging and after menopause. For example, periosteal expansion has been suggested to partially compensate for age- and menopause-related reductions in BMD and increased endocortical resorption [38]. More recent data from subjects with postmenopausal osteoporosis indicate that periosteal apposition failed to compensate for increased endocortical resorption, leading to reduced cortical thickness and reduced indices of bone strength [39]. This effect is in contrast to the significant bone formation that is possible with the relatively young, growing animals utilized in the present study. Indeed, we observed that RANKL treatment led to stimulation of periosteal bone formation, but the concomitant increase in endocortical resorption and reduction in BMD led to significantly lower bone area and bone strength. These results suggest that osteoporosis and RANKL are both associated with a negative bone turnover balance and that in both conditions periosteal apposition was an inadequate response for maintaining bone strength. These phenomena stand in contrast to the effects of OVX in rats, wherein periosteal apposition has been shown to effectively compensate for endocortical bone loss such that cortical strength does not decrease and might even increase [30].

Trabecular bone microstructure at the proximal tibia was also dramatically degraded after only 10 days of RANKL administration. Although trabecular strength was not assessed in this study, the deterioration of trabecular bone has been associated with increased fragility or reduced bone strength in animal and human studies. Consistent with the RANKL-mediated increase in endocortical eroded surface, medullary volume was significantly greater than control, while cortical bone volume was lower after 10 days of RANKL treatment.

Increased cortical porosity has been associated with a substantial age-related decline in human bone strength [40]. While intracortical remodeling is not a common finding in mice, excessive RANKL activity can lead to this pathologic change [33]. The RANKL-stimulated increase in cortical porosity was region-specific, with significantly greater increases in porosity observed in the proximal and distal metaphyses of the femoral diaphysis but not the mid-diaphysis. Cortical porosity in hip fracture patients has also been shown to have a distinct nonhomogenous distribution within the proximal femur [41]. Differences in the ability of RANKL to stimulate porosity at the proximal and distal femur could be the result of differences in the basal rate of bone turnover, cortical morphology, or bone marrow composition between these sites and the mid-diaphysis. The potential effect of the RANKL-mediated increase in distal and proximal femur porosity on bone strength was not determined and would be better suited to examination in a larger species, such as the rat. Although cortical porosity was not significantly lower in the RANKL groups at the femur midshaft, porosity measures were inversely correlated with femur maximum load.

When compared to control, RANKL administration resulted in significantly lower maximum load and tissue mineralization at the femur diaphysis. Cortical area and tissue mineralization were positively correlated with maximum load and accounted for 75% of the variation in load when combined as a surrogate for bone mineral content.

The present study has some important limitations, including the lack of an obvious dose response for many of the parameters investigated. It is clear that a lower dose range would have likely provided a more distinct dose response for the majority of parameters. Dose-dependent weight loss and hypercalcemia are consistent with the notion that the twice-daily injection of RANKL at 2 mg/kg (HI dose) was excessive and resulted in some degree of toxicity. It remains possible that a once-daily dose of RANKL at 0.4 mg/kg or lower could reproduce many of the deleterious skeletal effects described here, with fewer side effects. The lethargy that occurred as a result of the hypercalcemia could have also made a contribution to the observed decline in bone parameters through a reduction in applied muscle force. However, unlike disuse models such as hindlimb suspension [42] or sciatic nerve crush [43], the mice in the present study were not completely immobilized. A further limitation is that changes in trabecular bone strength after RANKL administration were not examined in this study due to the challenges in biomechanical testing of trabecular sites in the mouse. The rat would be the preferred model for characterizing the effects of RANKL on trabecular bone strength, and preliminary data have demonstrated loss of trabecular bone after soluble RANKL administration in this species [44]. Although the use of a single calcein bone label is a limitation to the interpretation of the histomorphometric results, the increased eroded surface found in RANKL-treated mice is consistent with the increased endocortical osteoclast surface reported in OPG knockout mice [34]. Despite the fact that there was not a baseline control group, the short duration of the study and the nonsignificant (1.4%) weight gain in the vehicle control group suggest that changes from baseline in bone morphology, mineralization, and strength in the untreated controls would be minimal. Furthermore, a study by Halloran and colleagues [45] demonstrated that proximal tibia BV/TV in C57BL/6 mice decreased approximately 16% from 3 to 5.5 months of age, far less than the 85% reduction observed with RANKL administration over 10 days in mice of a similar age and strain.

A recent study from our laboratory investigated the effects of continuous infusion of RANKL (35–175 μg/kg/day) to mature rats via subcutaneous osmotic pumps [22]. Although we were able to elicit similar responses in regard to increased bone turnover and decreased trabecular and cortical bone parameters, this previous study did require a considerably longer duration (4 weeks vs. 10 days). In addition, it is clear that, even considering differences in animal age, mice are far more sensitive to the deleterious effects of RANKL on the skeletal system. Although it must be weighed against the advantages of the rat skeletal system as an analogue of human physiology, intermittent RANKL administration to mice may represent a more efficient and reproducible model of pathologic bone loss.

Data from the present study establish the potential utility of a novel, nonsurgical model for rapid bone loss in mice. This model is characterized by cortical and trabecular changes that are more severe and more rapid than those associated with OVX, even in C57BL/6 mice that have been characterized as “slow losers” of bone mass after OVX [30]. OVX in C57BL/6 mice resulted in a nonsignificant 18% reduction in trabecular bone volume at the proximal tibia after 4 weeks [32], while cortical bone was not statistically significant until 16 weeks after OVX [30]. In the current study, 10 days of RANKL administration caused much greater osteoporotic changes in bone similar to that of OPG gene ablation. Further studies are needed to evaluate the relative role of soluble vs. membrane-bound RANKL in the regulation of bone resorption, but it is reasonable to conclude that soluble RANKL is strongly catabolic for bone. This model may be relevant for bone loss initiated by local and/or systemic changes associated with osteoporosis and joint destruction associated with inflammation as both conditions are associated with increased RANKL [18, 46].

In summary, the administration of soluble recombinant RANKL to mice significantly increased bone turnover, resulting in rapid decreases in bone volume, mineralization, and strength. The bone resorption marker TRAP-5b was correlated with cortical bone volume, porosity, and mineralization, thus demonstrating the link between the RANKL-mediated increase in bone resorption and its deleterious effects on bone. Increases in bone formation in RANKL-treated mice, demonstrated by increased ALP and femur diaphyseal mineral apposition rate, were insufficient to counter the increases in resorption, as evidenced by significant loss of cortical area. The loss of cortical bone strength after RANKL administration was positively correlated with the loss of cortical bone mass. Many of the skeletal changes associated with RANKL administration were qualitatively similar to those associated with postmenopausal osteoporosis, suggesting that this nonsurgical model might have utility for studying osteoporosis related to estrogen deficiency.

Acknowlegement

This work was supported by the National Space Biomedical Research Institute through NASA NCC 9–58, Amgen, Inc., and BioServe Space Technologies (through NASA NCC8–242). The editorial and formatting assistance of Jenny Bourne is greatly appreciated. Thanks also to Steven Adamu (Amgen, Inc.) for assistance in measuring biochemical markers and Michael Lemus for help with microCT analysis.

Copyright information

© Springer Science+Business Media, LLC 2008