Different skeletal effects of the peroxisome proliferator activated receptor (PPAR)α agonist fenofibrate and the PPARγ agonist pioglitazone

The Animal Welfare Committee at Trondheim University Hospital approved this study. Thirty-five two-month-old female Fischer rats (203 ± 11.5 g) were housed solely in wire-top cages with aspen woodchip bedding from B&K Universal Ltd. Room temperature was 24 ± 1.0°C with a relative humidity of 40%–50% and a 12-h light/dark cycle. The Rat and Mouse Diet of B&K, and tap water were provided ad libitum. One group received methylcellulose (control, n = 12), and the other two groups methylcellulose with either fenofibrate (n = 11) or pioglitazone (n = 12), 35 mg/kg body weight daily given for 4 months by gavage. The dose of pioglitazone was chosen on the background of previous studies on the long-term skeletal effects of TZDs in rodents [14, 16, 28]. The dose of fenofibrate was chosen according to other studies with long-term fenofibrate treatment in rats [29, 30].

The rats were weighed at the start and weekly throughout the study. Before all procedures and sacrifice, the animals were anesthetized with 2.0 ml/kg body weight of a combination of fluanison (2.5 mg/ml), fentanyl (0.05 mg/ml), and midazolam (1.25 mg/ml). Blood was collected by heart puncture during the finale anesthezia, and stored at -80°C until assaying. The liver weights were also registered. Both femurs were collected and stored at -80°C for further analyses.

Bone mineral density (BMD) and bone mineral content (BMC) in femur and whole body, fat mass and lean mass were measured by DXA in intact animals, using a Hologic QDR 4500A, and small animal software. Measurements were performed in duplicate at the start and the end of the study in intact animals. Coefficients of variation were 0.6% for whole body BMD, 0.7% for femur BMD, 0.5% for whole body BMC, 3.0% for femur BMC, 2.2% for total fat content, and 0.3% for lean body mass.

Histomorphometry of the left femur was performed as described previously and was evaluated by two persons (27). Transverse sections were cut close to the patellar ridge and also 5.0 mm proximal from it. They were grounded to a thickness of 50 um. The sections close to the patellar ridge were used for calculation of trabecular bone volume (TBV%), and proximal sections were used for calculation of cortical thickness. The sections for calculation of TBV% were stained with Goldener. For calculation of TBV% a Merz grid at a magnification of 10 × 10 was used. As many squares as possible, usually about 9–11, were calculated. For calculation of cortical thickness the central point of the medullary canal was identified. By means of an eyepiece the diameters were calculated in four directions with a 45 degrees angle between each. Medullary, cortical and total cross-sectional areas were calculated from the mean values. All histomorphometric measurements were performed blindly.

The right femurs were thawed in Ringers^® solution before mechanical testing of the femoral shaft and the femoral neck was performed. The diaphyses were fractured 18.7 mm from the femoral condyles in three point cantilever bending as previously described [31]. The proximal femur was fixed in a clamp, the cam of the rotating wheel engaged the femoral condyles and a fulcrum positioned anteriorly 18.7 mm from the condyles was the third point of force application. All tests were done at a loading rate of 0.095 radians/second (5.43°/second). The load in the test apparatus, an MTS 858 Mini Bionix^® Axial/Torsional Test System (MTS Systems Corporation, Minnesota, USA), was measured with a MTS Test Star TM Sensor Cartridge Force 250 N load cell and registered in MTS Test Star II software. Ultimate moment, ultimate energy absorption, stiffness and deflection were read directly or calculated from the computer recordings [32].

After biomechanical analyses, the right femurs were homogenized by crushing in liquid nitrogen, and 200 mg tissue of each sample was lysed in Trizol (Invitrogen, CA, USA) followed by protein isolation according to the manufacturer's instructions.

Using radioimmunoassay (RIA) (Linco Research, St. Charles, Missouri 63304, USA), leptin and adiponectin levels were quantified in plasma and in the isolated femoral protein fraction, and insulin and OPG levels were measured in plasma. Detection limits were 3.0 pg/ml for leptin, 0.78 ng/ml for adiponectin, 18.6 pg/ml for insulin, and 2.3 pg/ml for OPG. Intra- and interassay variations for all RIAs were < 4.5% and < 9.0%, respectively.

Osteocalcin in plasma was determined by a Rat-MID Osteocalcin enzyme-linked immunosorbent assay kit (Nordic Bioscience Diagnostics A/S, Denmark), according to the manufacturer's protocol. The detection limit was 50 ng/ml, and intra- and interassay variations were 5.0% and 5.5%, respectively. Bone resorption markers in plasma (fragments of type 1 collagen) were analyzed by a RatLaps ELISA kit (Nordic Bioscience Diagnostics A/S) according to the instructions from the manufacturer. The detection limit was 3.0 ng/ml, and intra- and interassay variations were 5.6% and 10.5%, respectively.

The concentration of OPG in the mouse cell line MC3T3-E1 culture media was determined by ELISA as described by the manufacturer (RnD Systems, USA) using anti-mouse OPG-antibody and biotinylated anti-mouse OPG. Detection was performed by labeling with streptavidin-horseradish peroxidase (R&D Systems) and adding 1, 2 phenylenediamine dihydrochloride (OPD) (Dako, Glostrup, Denmark) as substrate. The enzymatic reaction was stopped after 20 min by adding 100 μl of 0.5 M H₂SO₄ to each well. The optical density (OD) was measured as absorbance at 490 nm. Recombinant murine OPG was used as standard. The detection limit was 10 pg/ml, and intra-assay and inter-assay variabilities were less than 15% and 9.0%, respectively. The amount of OPG was related to the amount of total protein in each sample. RANKL concentration in culture media was determined by an immunoassay kit for quantitative determination of free sRANKL from mouse and rat (Biomedica, Vienna, Austria) according to the manufacturer's protocol. The amount of RANKL was related to the amount of total protein in each sample. The amount of total protein in media was determined using Sigma Microprotein PR assay kit with Protein Standard Solution Calibrator (Sigma Diagnostics, Dorset, UK) according to the manufacturer's protocol. Analyses were performed using the Cobas Mira chemistry analyzer (Roche Diagnostics, Germany). Intra-assay and inter-assay variabilities were less than 2.4% and 3.2% respectively. Detection range for the assay was 10-2.0 × 10³ mg/l.

MC3T3-E1 (mouse preosteoblasts, ATCC) cells were maintained in α-MEM (Invitrogen) supplemented with 10% fetal calf serum (FCS) (EuroClone, Great Britain), 1 mM Na- pyruvate (Gibco BRL, Life Technologies Ltd, Scotland), 0.1 mg/ml L-glutamine (Gibco) and 10 U/ml penicillin/streptomycin (Gibco).

To study whether fenofibrate and pioglitazone affected osteoblast differentiation, MC3T3-E1 cells were seeded in 6-wells plates (3.0 × 10⁵ cells/well) and cultured until confluence (3 days). The cells were then cultured for up to 12 days in α-MEM/10% FCS, with or without fenofibrate or pioglitazone. Medium was changed every second day. The relative mRNA expression of the osteoblast differentiation markers including alkaline phosphatase (ALP), bone sialoprotein (BSP), CD44, collagen 1, osteocalcin and osteopontin was studied. The relative mRNA expression of PPARα, PPARγ and of the adipocyte differentiation gene lipoprotein lipase (LPL) was also determined.

Proliferation was studied using a kit from Roche Molecular Biochemicals, Mannheim, Germany. Briefly, MC3T3-E1 cells (4 × 10³ cells/well) were seeded in 96 well plates, and cultured for 24 h. Cells were washed once with 180 μl serum-free medium, before addition of new medium with or without test substances. After 4 h, 5-bromo-2'-deoxyuridine (BrdU)-labeling solution was added, and the cells were cultured for additional 18 h before incorporation of BrdU was measured as described by the manufacturer. After removing the labeling medium, cells were fixed and genomic DNA denaturized by adding 150 μl FixDenat per well for 30 min at room temperature. FixDenat-solution was removed and 100 μl of peroxidase-conjugated anti-BrdU antibody solution was added per well, followed by incubation at room temperature for 90 min. The cells were washed three times with 200 μl washing solution before 100 μl of substrate Luminol/4-idophenol was added. After 3 min, chemiluminescence was measured (RLU = relative luminescence units) in a micro plate luminometer (Fluoroscan Ascent FL, Labsystems).

Cells were homogenized and lysed in lysis/binding buffer and mRNA was isolated using magnetic beads (oligo (dT)₂₅) as described by the manufacturer (Dynal AS, Oslo, Norway). Beads containing mRNA were resuspended in 10 mM Tris-HCl, pH 8.0, and stored at -80°C until use. mRNA-containing solution was applied directly to obtain a first-strand complementary DNA (cDNA) using the iScript cDNA Synthesis Kit with oligo(dT) (Bio-Rad, CA, USA). cDNA samples were diluted 1:2 with nuclease free water.

Reactions were performed and monitored using Stratagene's Mx3000P Real-time PCR system. The 2× iQ SYBR Green Supermix was based on iTaq DNA polymerase (Bio-Rad, Oslo, Norway). cDNA samples were analyzed both for the genes of interest and reference genes (β-actin) The amplification program consisted of a preincubation step for denaturation of the template cDNA (5 min, 95°C), followed by 40 cycles consisting of a denaturation step (30 s, 95°C), an annealing step (30 s, 60°C) and an extension step (30 s, 72°C). The Ct value, defined as the number of cycles required to produce a detectable product above background fluorescence, was measured for each sample, and arbitrary units were calculated using standard curves that consisted of serial dilutions of cDNA from a pool of samples or controls containing the highest amounts of the specific gene analyzed. Contamination by genomic DNA was ruled out by performing PCR analysis where RT-enzyme had been omitted in the RT reactions. β-actin RT-PCR was run in duplicates or triplets as control to monitor RNA integrity and to be used for normalization. Specificity of each primer pair was confirmed by melting curve analysis and agarose gel electrophoresis, and PCR products were sequenced for product confirmation. Table 1 show the primer sequences, the expected sizes of the PCR products and gene bank accession numbers used for design of the primer pairs. Intron-spanning primers were designed using the computer software Clone. Data is calculated from standard curves, related to housekeeping gene and presented normalized to untreated controls at each individual time point.

Primary human osteoclast precursor cells and differentiation media containing macrophage colony stimulating factor (M-CSF) and soluble RANKL were obtained from Lonza, and cultured according to the provided protocol. Ten thousand cells per well were seeded in a 96-well plate in differentiation medium. Fenofibrate or pioglitazone (10 nM, 0.1, 1.0 and 10 μM) were added in six parallel wells, and cultured for 12 days before cultures were stained for tartrate-resistant acid phosphatase (TRAP), activity using Naphtol AS-BI phosphate and Fast Garnet in the presence of sodium tartrate, as described by the manufacturer (Sigma-Aldrich, Norway). The number of TRAP positive, multinuclear (3 or more nuclei) cells was counted.

Human osteoclast activity was examined using the Osteolyse Assay kit (Lonza,). An osteolyse plate was seeded with 1.0 × 10⁴ human osteoclast precursor cells (Lonza) per well in medium containing M-CSF and soluble RANKL and cultured at 37°C for 7 days. Control cells were seeded with or without RANKL and M-CSF. Culture media were changed on day 7, and then cultured for another 1 to 3 days with or without fenofibrate (0.1, 1.0 and 10 μM) or pioglitazone (0.1, 1.0 and 10 μM), with six parallels for each condition. Cell medium supernatants were collected on day 8, 9 and 10, and the release of collagen type I was measured by EIA according to the manufacturer's protocol, with the exception that fluorescence was measured by excitation at 355 nm and emission of 650 nm, instead of 340/615 nm as recommended.

There were no differences in body weight, fat mass or lean mass at the beginning of the study (data not shown). After 4 months of drug administration, the body weight did not differ between the groups, and no difference in femoral length was observed, indicating absence of effects on longitudinal growth (Table 2). The rats given pioglitazone had significantly higher fat mass (P = 0.01 vs. controls, and P = 0.0001 vs. the group given fenofibrate). Lean mass was significantly lower in the pioglitazone group compared to controls (P = 0.005) and the fenofibrate group (P = 0.0001) (Table 2). In the fenofibrate group, fat mass did not differ from controls, while lean mass was significantly higher than in control rats (P = 0.006) (Table 2). The mean liver weight was significantly higher in the fenofibrate group compared to controls (9.64 ± 0.85 g and 6.98 ± 0.47 g, respectively, P = 0.0001). In the pioglitazone group, the liver weight was significantly lower (5.79 ± 0.39 g), both compared to controls (P = 0.0168) and to rats receiving fenofibrate (P = 0.0001). Since lean mass also include liver weight, the liver weight was subtracted to correct for this. After this correction, lean mass was still significantly higher in the fenofibrate group (185.3 ± 8.5 g), both compared to the control (176.4 ± 7.6 g, P = 0.016) and pioglitazone group (165.6 ± 8.6 g, P = 0.0001). Corrected lean mass was significantly lower in the pioglitazone group than in the controls (P = 0.008). Four animals died in the pioglitazone group, while none died in the other groups. The cause of death in the pioglitazone group was aspiration in connection with gastric gavage in 3 of the rats, in one of the rats no reason could be found. No differences in the well-being of the animals were, however, observed during the study.

There were no differences in femoral and whole body BMC or BMD at the beginning of the study (data not shown). After four months of treatment, femur BMC tended to be higher, albeit not significantly, in the fenofibrate group compared to controls, (+ 6.2%, P = 0.08), while the pioglitazone group had significantly lower femur BMC compared to the fenofibrate group (-10%, P = 0.0067), but not compared to controls (Table 2). Femoral BMD was significantly higher in fenofibrate rats (+ 5.8%, P = 0.004), while femoral BMD was similar in controls and pioglitazone treated rats (Table 2). The pioglitazone group had significantly lower femoral BMD than the fenofibrate group (-10%, P = 0.0003). Whole body BMC was significantly lower in the pioglitazone group after four months of treatment, relative to controls (-5.2%, P = 0.047), and relative to the fenofibrate group (-6.6%, P = 0.004). Whole body BMD was significantly lower in the pioglitazone group compared to controls (-9.8%, P = 0.0001), and also compared to the fenofibrate group (-8.4%, P = 0.0001), while there was no difference between the control group and the fenofibrate group (Table 2).

Trabecular bone volume in the distal femur was lower in the pioglitazone rats compared to control rats and fenofibrate rats (-27%, P = 0.006 and – 25%, P = 0.019, respectively), while no difference was found between the fenofibrate and the control group (Table 3). The medullary area in distal femur was significantly higher in the pioglitazone rats compared to the fenofibrate group, (+ 14.8%, P = 0.0001), and also significantly higher compared to controls (+ 8.0%, P = 0.05), (Table 3). In contrast medullary area was lower, in the fenofibrate group compared to controls (-7.2%, P = 0.05). There were no differences in cortical and total areas between the groups (Table 3). For the total material, there was a significant positive correlation between lean mass (corrected for liver weight), femoral BMD (Pearsons r = 0.513, P = 0.0001) and cortical area (Pearsons r = 0.673, P = 0.008).

The ultimate bending moment, deflection and energy absorption of femoral shafts were significantly lower in the pioglitazone rats compared to controls (P = 0.007, P = 0.039 and P = 0.014), and compared to the fenofibrate group (P = 0.019, P = 0.0025 and P = 0.014), while no difference was found between the fenofibrate group and the controls (Table 4). Biomechanical parameters in the femoral neck did not differ significantly between the groups (Table 4).

Plasma levels of leptin were significantly lower in both fenofibrate (P = 0.02) and pioglitazone groups (P = 0.03) compared to controls (Figure 1A). In femoral bone, the leptin concentration was lower in pioglitazone rats (P = 0.003 vs. control and P = 0.03 vs. fenofibrate group), while the fenofibrate group did not differ from controls (Figure 1B). Plasma levels of adiponectin were significantly lower in the fenofibrate group compared to controls (P < 0.0001) (Figure 1C), while the adiponectin levels were significantly higher in the pioglitazone group both compared to controls (P < 0.0001), and to the fenofibrate group (P < 0.0001) (Figure 1C). There were no differences in the adiponectin levels in femoral bone between the groups (Figure 1D).

There were no differences in plasma levels of insulin between the groups (controls: 1.42 ± 0.66 ng/ml, fenofibrate group: 1.35 ± 0.69 ng/ml and pioglitazone group: 1.32 ± 0.52 ng/ml). Plasma levels of osteocalcin were significantly higher in the fenofibrate group compared to controls (P = 0.012), while the pioglitazone group did not differ from controls (Figure 1E). Plasma OPG levels were significantly lower in the pioglitazone group compared to controls (P = 0.009), while the fenofibrate group did not differ from controls (Figure 1F). There were no differences in plasma levels of bone resorption marker (fragments of collagen type I) between the groups (data not shown).

After 12 days of stimulation, fenofibrate (0.1 μM) significantly increased the mRNA expression of ALP, BSP, CD44, collagen 1, and osteocalcin in MC3T3-E1 cells (Table 5). Pioglitazone (0.1 μM) increased the relative CD44 mRNA expression after 12 days of stimulation, but none of the other osteoblast differentiation genes examined (Table 5). Other concentrations of both fenofibrate and pioglitazone (1.0 μM and 10 μM) were examined, with similar patterns. The early adipocyte differentiation marker lipoprotein lipase (LPL) mRNA was not expressed at day 0, but was low and equally expressed in all cells at day 4, 8 and 12 of stimulation (Table 5). PPARα mRNA expression was significantly enhanced compared to control by both fenofibrate and pioglitazone after 4 days, and also after 8 and 12 days by fenofibrate. There was, however, no detectable expression of PPARα mRNA in pioglitazone stimulated cells after 12 days. PPARγ expression was equally expressed in all cells throughout the stimulation period (Table 5).

Fenofibrate (1.0 nM – 1.0 μM) increased proliferation of MC3T3-E1 preosteoblasts, whereas pioglitazone had no effect (Figure 2).

MC3T3-E1 cells were treated with fenofibrate and pioglitazone (1.0 nM to 10 uM) for 12, 24, 48 and 72 h to study the effect on release of OPG and RANKL. Fenofibrate increased OPG release in a dose-dependent manner after 12–48 h stimulation (P < 0.05) (Figure 3A). In contrast, pioglitazone had no effect on OPG release from MC3T3-E1 cells (Figure 3B). Neither fenofibrate nor pioglitazone induced significant changes in RANKL secretion from MC3T3-E1 cells compared to untreated cells (data not shown).

We found no effect of fenofibrate or pioglitazone on either differentiation or activity of primary human osteoclasts.