Rosiglitazone Causes Bone Loss in Mice by Suppressing Osteoblast Differentiation and Bone Formation
http://www.100md.com
内分泌学杂志 2005年第3期
Division of Endocrinology & Metabolism, Center for Osteoporosis and Metabolic Bone Diseases, Central Arkansas Veterans Healthcare System, University of Arkansas for Medical Sciences, Little Rock, Arkansas 72205
Address all correspondence and requests for reprints to: Robert L. Jilka, Ph.D., Division of Endocrinology and Metabolism, Slot 587, University of Arkansas for Medical Sciences, 4301 West Markham, Little Rock, Arkansas 72205. E-mail: jilkarobertl@uams.edu.
Abstract
Because osteoblasts and marrow adipocytes are derived from a common mesenchymal progenitor, increased adipogenesis may occur at the expense of osteoblasts, leading to bone loss. Our previous in vitro studies indicated that activation of the proadipogenic transcription factor peroxisome proliferator-activated receptor isoform 2 with rosiglitazone suppressed osteoblast differentiation. Here, we show that 5-month-old Swiss-Webster mice receiving rosiglitazone for 28 d exhibited bone loss associated with an increase in marrow adipocytes, a decrease in the ratio of osteoblasts to osteoclasts, a reduction in bone formation rate, and a reduction in wall width—an index of the amount of bone formed by each team of osteoblasts. Rosiglitazone had no effect on the number of early osteoblast or osteoclast progenitors, or on osteoblast life span, but decreased the expression of the key osteoblastogenic transcription factors Runx2 and Osterix in cultures of marrow-derived mesenchymal progenitors. These effects were associated with diversion of bipotential progenitors from the osteoblast to the adipocyte lineage, and suppression of the differentiation of monopotential osteoblast progenitors. However, rosiglitazone had no effect on osteoblastic cells at later stages of differentiation. Hence, rosiglitazone attenuates osteoblast differentiation and thereby reduces bone formation rate in vivo, leading to bone loss. These findings provide a mechanistic explanation for the recent evidence that peroxisome proliferator-activated receptor isoform activation is a negative regulator of bone mass and suggest that the increased production of oxidized fatty acids with age may indeed be an important mechanism for age-related osteoporosis in humans.
Introduction
OSTEOBLASTS AND MARROW adipocytes are derived from a common multipotential mesenchymal stem cell (MSC) progenitor (1). It has been hypothesized that the close relationship between these lineages underlies the reciprocal relationship between increased adipocytes and decreased bone formation that occurs during aging (2, 3, 4). Indeed, age-related bone loss is associated with a deficit in the number of osteoblasts needed to replace the bone removed by osteoclasts during remodeling as evidenced by a decrease in mean wall thickness (5).
Lineage commitment of MSCs is determined by expression and/or activation of specific transcription factors: Runx2 and Osterix in the case of osteoblasts (6, 7), and CCAAT enhancer binding protein ? and peroxisome proliferator-activated receptor isoform (PPAR) in the case of adipocytes (8). PPAR is broadly expressed and plays an important role not only in adipogenesis but also in glucose homeostasis, atherosclerosis, and cancer (9). It is a member of the nuclear hormone receptor family of transcription factors, and exists in two isoforms, PPAR1 and PPAR2, resulting from alternative promoter usage and alternative splicing. The transcriptional activity of PPAR is induced by a variety of ligands including 15-deoxy-12,14 prostaglandin J2, oxidized fatty acid metabolites, as well as synthetic thiazolidinediones like troglitazone and rosiglitazone (10, 11, 12, 13, 14, 15).
Mounting evidence indicates an important role of PPAR in skeletal metabolism. Specifically, PPAR haploinsufficient mice exhibit increased bone mass associated with increased osteoblastogenesis and decreased adipogenesis (16). Using murine marrow-derived MSCs, as well as a model marrow-derived stromal cell line overexpressing PPAR2, we have shown that activation of PPAR2 with rosiglitazone stimulates adipogenesis and inhibits osteoblastogenesis, at least in part by down-regulating Runx2 (17); and that fatty acid peroxidation products such as 9,10-dihydroxyoctadecenoic acid and 9,10-epoxyoctadecenoic acid can attenuate osteoblast differentiation in a PPAR2-dependent fashion (13). We have therefore hypothesized that increased PPAR activation may contribute to the reduced osteoblast production and bone loss that occurs with aging.
Rosiglitazone causes bone loss in rats and mice (18, 19), but the cellular basis underlying this skeletal response has not been determined. Here, we report that adult mice receiving rosiglitazone exhibit bone loss characterized by an increase in marrow adipocytes, and a decrease in the osteoblast/osteoclast ratio, wall width, and bone formation rate. In addition, we show that rosiglitazone diverted bipotential mesenchymal progenitors from the osteoblast to the adipocyte lineage and also suppressed the differentiation of monopotential osteoblast progenitors. These inhibitory effects were strongly associated with a reduction in the expression of Runx2 and Osterix.
Materials and Methods
Animals
Swiss-Webster mice were obtained from Harlan (Indianapolis, IN). The University of Arkansas for Medical Sciences Division of Laboratory Animal Medicine approved the animal use protocol. Food was blended with rosiglitazone (Tularik, Inc., South San Francisco, CA), which was dissolved at 4.25 mg/ml of a 1:3 mixture of ethanol and glyceroloctaside (Stepan Chemical, Maywood, NJ), 0.1% Tenox 20 (Eastman Chemical, Kingsport, TN) before mixing with a blender to give a concentration of 0.03 mg or 0.15 mg of rosiglitazone/g normal rodent chow (Agway RMH 3000; Arlington Heights, IL). Mice were pair fed to avoid potential artifacts secondary to the hyperphagia that has been reported to occur when rats or mice are given a thiazolidinedione (20, 21). Five-month-old mice were housed singly, and average daily consumption of vehicle chow was determined to be 5–6 g/d over a 2-wk period. Animals were fed 5 g chow/d containing vehicle (four males and two females), 5 μg/g·d rosiglitazone (two males and three females), or 25 μg/g rosiglitazone (three males and three females) for 28 d. These doses have previously been shown to give moderate to strong antihyperglycemic activity in diabetic ob/ob mice (22). By weighing the food dish before placement in the cage and after removal, it was determined that each mouse consumed more than 90% of the chow throughout the course of the experiment. Tetracycline (30 mg/kg) was administered ip at 8 and 3 d before the end of the experiment.
Bone densitometry
Dual-energy x-ray absorptiometry (DEXA) was used to determine bone mineral density (BMD) as previously described using a Hologic 2000 Plus (19, 23). Over the past 3 yr, the coefficient of variation of the measurement done on a plastic-embedded whole mouse skeleton was 1.8% (n = 202). BMD determinations were performed before initiation of the study and used to allocate the animals into groups with equivalent spinal BMD values before entry into the treatment protocol. Measurements were made again after 2 wk, and at termination of the experiment after 4 wk of treatment using the Compare program of Hologic software to ensure accurate repositioning for serial measurements. Global BMD refers to the entire skeleton excluding the skull. Spinal and hindlimb subregions were measured as previously described (23).
Bone histomorphometry
The left femur and the lumbar vertebrae (L1–L4) were fixed, embedded undecalcified in methyl methacrylate, and stained using modified Massons’s trichrome (23, 24, 25). Histomorphometric measurements were performed using a computer and digitizer tablet (Osteomeasure version 3.00, Osteometrics Inc., Atlanta, GA) interfaced to a Zeiss Axioscope (Carl Zeiss Inc., Thornwood, NY) with a drawing tube attachment. The identity of the sample was unknown to the reader. Measurements were made with a Zeiss Plan-Neofluar 40 x (0.75 numerical aperture) objective on the entire secondary spongiosa, which comprised approximately 60 fields in at least three of the vertebrae from L1–L4, and approximately 30 fields in the distal femur. Variables were measured and reported as previously described (23, 25, 26) using the terminology recommended by the Histomorphometry Nomenclature Committee of American Society for Bone and Mineral Research (27). The number of fat cells within the tissue area measured, as well as individual fat cell area and diameter, were also determined. They are expressed as total fat cell number per tissue area, and the diameter of fat cells. Osteoclast and osteoblast number per tissue area are reported to facilitate direct comparison to fat cell number.
Apoptotic osteoblasts were detected by in situ nick end labeling using Klenow terminal deoxynucleotidyl transferase (Oncogene Research Products, Cambridge, MA) as previously described (28). Vertebrae from orchidectomized adult mice were used as a positive control. Slides were counterstained with 2% methyl green. Apoptotic osteoblasts were identified as in situ nick end labeling-positive cells lining the osteoid covered cancellous bone surface.
Quantification of osteoclast and osteoblast progenitors
Marrow cells were obtained from the right femur of each animal and replicate cultures established for the determination of osteoclast progenitors and colony-forming osteoblast progenitors (CFU-OB) as previously described (29, 30). Briefly, osteoclast progenitors were determined by coculturing marrow cells with UAMS-32 stromal/osteoblastic cells for 8 d in the presence 10 nM 1,25(OH)2D3. Osteoclastic cells were enumerated after staining for tartrate-resistant acid phosphatase. CFU-OB number was determined by culturing cells for 25 d with irradiated guinea pig feeder cells in MEM supplemented with 15% preselected fetal bovine serum and 1 mM ascorbate-2-phosphate to promote osteoblastogenesis. Colonies containing osteoblasts were visualized by Von Kossa staining.
CFU-OB replication in vitro was determined as described previously (30). One aliquot of cells was used to determine CFU-OB number in the initial isolate, as described above. A second aliquot was used to establish replicate cultures of cells in type I collagen gels, which were then maintained in the absence or presence of 100 nM rosiglitazone for 6 d. The cells were dispersed using bacterial collagenase, and the number of CFU-OB within each gel was determined using the colony-forming assay described above. To calculate the fold increase in CFU-OB during culture in the collagen gels, the number of CFU-OB obtained per gel (after 6 d of culture) was divided by the number of CFU-OB initially added to the collagen gel.
The effect of rosiglitazone on adipocyte and osteoblast differentiation in vitro was determined by culturing bone marrow cells at 2.5 x 106 per 10 cm2 well with irradiated guinea pig cells and maintained for 25 d in medium supplemented with 1 mM ascorbate-2-phosphate, either in the absence or presence of 100 nM rosiglitazone, added at various times after establishment of the culture. On d 25, cells were stained either with Oil Red O to visualize colonies containing adipocytes or Von Kossa stain to visualize colonies containing osteoblasts elaborating a mineralized matrix.
Quantification of gene expression in cultured bone marrow cells and osteoblastic cells
Murine bone marrow cells were cultured as described above. Osteoblastic cells were obtained from neonatal murine calvarial by collagenase digestion as previously described (31), and cultured in MEM supplemented with 15% preselected fetal bovine serum and 1 mM ascorbate-2-phosphate. Total RNA was extracted using Ultraspec reagent (Biotecx Laboratories, Inc., Houston, TX). RNA (2 μg) was reverse-transcribed using a High Capacity cDNA Archive Kit (Applied Biosystems, Foster City, CA). The transcripts of interest, and that of the housekeeping gene ribosomal protein S2 (ChoB), were amplified from first-strand cDNA by real-time PCR using TaqMan Universal PCR Master Mix and Assay by Demand or Assay by Design primer and probe sets (Applied Biosystems). Amplification and detection were carried out with an ABI Prism 7300 Sequence Detection System (Applied Biosystems) as follows: 5-min denaturation at 95 C for 10 min, 40 cycles of amplification including denaturation at 94 C for 15 sec and annealing/extension at 60 C for 1 min. Gene expression was quantified by subtracting the ChoB threshold cycle (Ct) value from the Ct value of the gene of interest, and expressed as 2-Ct, as described by the protocol of the manufacturer.
Statistics
Data were analyzed using SigmaStat (SPSS Science, Chicago, IL) or SAS software (SAS Institute Inc., Cary, NC). All values are reported as the mean ± SD. Differences between group means in histomorphometry studies were evaluated with Student’s t test, or by ANOVA after P values were adjusted with Bonferroni’s correction. Serial changes in body weight and BMD were analyzed using a model of repeated measures (32). Pearson correlation analysis of fat cell number with osteoblast number and mineralizing surface was done after adding three to total fat cell number (to avoid large negative values for mice with no adipocytes), and then taking the natural logarithm to obtain a normal distribution. Differences in group means in studies of gene expression and colony number were evaluated with Student’s t test.
Results
Rosiglitazone causes bone loss and decreases osteoblast number and bone formation rate
Rosiglitazone was fed to adult (5 months old) male and female Swiss-Webster mice at 5 or 25 μg/g body weight·d for 28 d. Vehicle was added to the food of control mice. There was no change in body weight in mice receiving the vehicle, but both doses of rosiglitazone caused an approximately 2% increase in body weight (Fig. 1A). As reported previously (33), administration of 25 μg/g·d of rosiglitazone increased interscapular fat (Fig. 1B), indicating that drug administration was efficacious. Liver weight was unaffected (Fig. 1C).
FIG. 1. Effect of rosiglitazone on body weight, interscapular brown adipose tissue (BAT) weight, and liver weight. Values represent percentage of change in body weight from baseline determined at the beginning of the experiment (A) or tissue weight at the end of the experiment (B and C). *, P < 0.01 vs. vehicle as determined by repeated measures analysis (body weight) or ANOVA (BAT and liver weight).
Serial bone densitometry using DEXA revealed that after 28 d of administration, mice given rosiglitazone at 25 μg/g·d exhibited a decrease in vertebral BMD, compared with mice fed the vehicle control (Fig. 2). The same trend was noted in mice fed this dose of rosiglitazone when BMD was measured over the entire skeleton (global) and in hindlimbs, but the changes were not statistically significant. The 5 μg/g·d dose of rosiglitazone had no discernable effect on BMD. Thus, histologic and histomorphometric studies were performed only on mice fed the 25-μg dose.
FIG. 2. Rosiglitazone causes loss of BMD. Values represent percentage of change in BMD from the baseline value determined at the beginning of the experiment. Repeated measures analysis indicated a significant group x time interaction for spinal BMD (P = 0.007) and indicated a stronger negative effect for the 25 μg/g dose of rosiglitazone than the 5 μg/g dose.
Mice receiving rosiglitazone exhibited increased marrow fat in both vertebral and femoral bone compared with mice receiving the vehicle control (Fig. 3). Vertebral bone from control animals was practically devoid of adipocytes, but rosiglitazone increased their number by approximately 20-fold (Table 1). Adipocyte number in femoral metaphyseal bone of control mice was considerably higher than that of vertebral bone, and rosiglitazone caused a further 7-fold increase (Table 2). Fat cell diameter was unaffected, indicating that rosiglitazone did not enhance lipogenesis in existing adipocytes. Consistent with the densitometric findings, rosiglitazone caused loss of cancellous vertebral bone that was characterized by a decrease in trabecular width, an increase in trabecular spacing, and a decrease in trabecular number (Table 1). Similar microarchitectural changes were noted in femoral cancellous bone (Table 2).
FIG. 3. Increased marrow fat and decreased cancellous bone in mice receiving rosiglitazone. Nondecalcified bone sections were stained with Masson trichrome stain. Calcified bone is stained blue; adipocytes were present in the white globular areas. The lower panels of each set of photomicrographs represent high-power views of the field denoted by the yellow square in the upper panel.
TABLE 1. Effect of rosiglitazone on vertebral cancellous bone
TABLE 2. Effect of rosiglitazone on distal femoral cancellous bone
Rosiglitazone caused a reduction in the number of osteoblasts per square millimeter of vertebral and femoral cancellous bone (Tables 1 and 2). However, when expressed on a bone perimeter rather than bone area referent, osteoblast number did not exhibit a statistically significant decline, nor was osteoid surface affected in mice receiving rosiglitazone. The drug had no effect on the prevalence of apoptotic osteoblasts. Osteoclast number was also unaffected, whether expressed per millimeter of bone perimeter or square millimeter of bone area. However, osteoblast number in both vertebral and cancellous bone declined by 2-fold when expressed as the ratio of osteoblasts to osteoclasts, calculated from the number of each cell type per millimeter of bone perimeter. Moreover, there was an inverse correlation between adipocyte number and osteoblast number whether expressed per square millimeter of tissue area or as the ratio of osteoblasts to osteoclasts, in both vertebral and femoral cancellous bone (Table 3). The biological impact of the reduction in the ratio of osteoblasts to osteoclasts in mice receiving rosiglitazone was revealed by a decrease in vertebral bone wall width (Table 1).
TABLE 3. Inverse correlation between marrow adipocytes and osteoblast number, as well as mineralizing surface
These histological findings establish that rosiglitazone caused a deficit in the number of osteoblasts when compared with the number needed to refill the resorption cavities created by osteoclasts. Consistent with this, rosiglitazone caused a 2-fold decrease in bone formation rate in vertebral bone (Table 1). This was due to a reduction in mineralizing surface, rather than mineral apposition rate. Fat cell number inversely correlated with mineralizing surface (Table 3). Rosiglitazone had no effect on width of the osteoid layer between osteoblasts and mineralized bone, or on mineralization lag time, the interval between deposition and mineralization of osteoid. Therefore, the drug did not cause defective mineralization. The small amount of cancellous bone that remained in the femurs of mice fed rosiglitazone (2–6% of tissue area) precluded ascertainment of potential changes in wall width or indices of bone formation.
Rosiglitazone inhibits the differentiation of osteoblasts from CFU-OB
Analysis of ex vivo cultures of mice from the experiment described above established that rosiglitazone had no effect on the number of MSCs, as measured by ex vivo assay of CFU-OB, or on the number of osteoclast progenitors (Fig. 4). Consistent with the lack of effect on CFU-OB number, rosiglitazone also had no effect on the replication of CFU-OB in vitro. Specifically, progenitors increased by 4.3 ± 0.7-fold during 6 d of culture with vehicle vs. 3.5 ± 0.8-fold in the presence of 100 nM rosiglitazone (n = 3 per group). Thus, the reduction in osteoblast number and bone formation rate seen in mice receiving rosiglitazone must be due to actions of the drug on progeny of MSC, as measured by CFU-OB (30).
FIG. 4. Rosiglitazone has no effect on the number of osteoclast and osteoblast progenitors in the bone marrow. Femoral marrow cells were obtained from the animals receiving vehicle (veh) or rosiglitazone (rosi) and ex vivo femoral marrow cell cultures were established for the quantification of osteoblast progenitors (CFU-OB) (A) or osteoclast progenitors (B).
The production of osteoblasts from MSC in vitro proceeds via sequential stages of replication and lineage commitment during the first 2 wk of culture, after which development of the mature matrix-synthesizing osteoblast occurs (30, 34). To determine the stage at which rosiglitazone influences osteoblast differentiation, the drug was added on d 3, 7, 11, or 15 after establishment of ex vivo cultures of femoral marrow cell cultures maintained in ascorbate-2-phosphate to stimulate osteoblast differentiation. Calcium deposition and osteocalcin secretion, markers of differentiated osteoblasts, were quantified after 25 d of culture. When rosiglitazone was added on d 3 or d 7 of culture, calcium deposition and osteocalcin secretion were reduced by 90% or more, whereas only a modest inhibitory effect was observed when the drug was added on d 11 or 15 (Fig. 5A).
FIG. 5. Rosiglitazone inhibits osteoblastogenesis only when added in the early phase of differentiation. Marrow cell cultures from femurs of 5-month-old Swiss-Webster mice were established as described in Materials and Methods. Vehicle (dimethylsulfoxide) or 100 nM rosiglitazone was added to two sets of parallel cultures on d 3, 7, 11, or 15. On d 25, one set of cultures (A) was analyzed for the content of osteocalcin in the medium, and for the content of calcium in the matrix. The second set of cultures (B–F) was stained with Von Kossa to visualize colonies containing mineralizing osteoblasts and with Oil Red O to visualize adipocytes. Data shown represent the mean (±SD) of triplicate cultures. *, P < 0.05 vs. vehicle control at the same time point, by ANOVA.
The influence of rosiglitazone on the differentiation of colony-forming cells was examined in more detail in a parallel set of cultures using Von Kossa and Oil Red O staining to detect the presence of osteoblasts and adipocytes, respectively. Rosiglitazone reduced the total number of colonies by approximately 50% when added on d 3 or 7 (Fig. 5B), as we had previously observed when rosiglitazone was added on the first day of culture (13); however, the drug had no effect on colony number when added after d 7. Colonies containing only adipocytes, only osteoblasts, both cell types, or neither cell type were quantified (Fig. 5, C–F). Most of the colony-forming cells gave rise to adipocytes when rosiglitazone was added on d 3, and osteoblastogenesis was practically abolished (Fig. 5, C–E). The percentage of colonies containing only adipocytes declined progressively with later addition of rosiglitazone, suggesting loss of the ability to differentiate into adipocytes with longer culture time in pro-osteogenic medium (Fig. 5E). When rosiglitazone was added on d 7, the number of colonies containing only osteoblasts was still greatly reduced, but this inhibitory effect was much less evident when the ligand was added on d 11 or 15 (Fig. 5C).
When rosiglitazone was added on d 7 or later, approximately one third of the colonies contained both osteoblasts and adipocytes (Fig. 5D). Such colonies must have arisen from a progenitor that gives rise to two kinds of daughter cells: one that can be driven into the adipocyte lineage with rosiglitazone, and one that differentiates into osteoblasts in the presence ascorbate-2-phosphate. We have designated this colony-forming cell as a bipotential progenitor but still recognize that it could differentiate into other cell types under the proper conditions. When rosiglitazone is added on d 3, few colonies containing both osteoblasts and adipocytes were observed. This indicates that rosiglitazone diverted bipotential progenitors into the adipocyte lineage. However, the ligand could also have suppressed the differentiation of osteoblasts from a monopotential osteoblastogenic daughter of the bipotential cell. Finally, about 20% of colonies contained neither osteoblasts nor adipocytes, but their number was also reduced by rosiglitazone at d 3 and 7, perhaps accounting for part of the decrease in overall colony number at these times (Fig. 5F).
To gain additional insight on the suppressive effect of rosiglitazone on osteoblastogenesis, we examined the expression of osteoblast- and adipocyte-specific transcription factors and proteins using real-time PCR. In this experiment, the ligand was added to cultured bone marrow cells on d 7, i.e. before expression of the osteoblast phenotype as indicated by the lack of secretion of osteocalcin (Fig. 5A). Transcript levels were determined just before addition of the ligand, and at d 10, 15, and 25 of culture. On d 7, PPAR2 transcripts were approximately 10-fold less than PPAR1 transcripts, but rosiglitazone rapidly increased expression of PPAR2 by approximately 30-fold, consistent with previous evidence that PPAR stimulates its own expression (35). PPAR1 transcripts were either not affected or modestly suppressed at d 15 of culture (Fig. 6). More importantly, rosiglitazone fully prevented the increase in expression of Runx2 and Osterix, which are required for osteoblast differentiation. This effect coincided with a failure to induce expression of the bone matrix proteins Col1A1, osteocalcin and bone sialoprotein, consistent with the findings of Fig. 5 showing almost complete suppression of osteoblast differentiation when rosiglitazone was added on d 7 of culture.
FIG. 6. Rosiglitazone reduces expression of osteoblast-specific transcription factors and matrix proteins, but up-regulates PPAR2 and adipogenesis in cultured bone marrow cells. Marrow cell cultures were established as in Fig. 5, and vehicle (dimethylsulfoxide) or 100 nM rosiglitazone was added on d 7. RNA was isolated immediately before addition of these agents, and on d 10, 15, and 25 of culture. The indicated transcripts, and the housekeeping gene ChoB, were quantified by TaqMan PCR. Data shown represent the mean (±SD) level of expression for each transcript relative to ChoB, determined in triplicate cultures. *, P < 0.05 vs. contemporaneous vehicle control by t test. BSP, Bone sialoprotein.
Rosiglitazone also down-regulated Wnt-10b expression, a response that is required for expression of the adipocyte phenotype (36) (Fig. 6). In addition, expression of the adipocyte markers fatty acid binding protein 4 (aP2), adiponectin, and lipoprotein lipase (LPL) were dramatically increased, consistent with the up-regulation of PPAR2 and the important role of this form of the transcription factor in ligand-stimulated adipogenesis (35). These findings confirm the data of Fig. 5 showing a dramatic increase in Oil Red O-positive adipocytes when rosiglitazone was added on d 7 of culture.
In view of the evidence of Fig. 5 that rosiglitazone had no effect on osteocalcin secretion or mineralization when added to bone marrow cultures at late stages of culture, we examined the effect of rosiglitazone on gene expression in preparations of osteoblastic cells obtained from neonatal murine calvaria. In contrast to the response of bone marrow progenitors, addition of rosiglitazone did not increase PPAR2 or decrease Wnt-10b transcripts in calvarial osteoblastic cells (Fig. 7). Moreover, transcript levels of osteoblast-specific transcription factors and matrix proteins were unaffected by the ligand. The adipocyte markers aP2, adiponectin ,and LPL were increased, but the absolute level of expression achieved by addition of rosiglitazone was approximately 20-fold lower than in bone marrow cells. This probably reflects the formation of only a few adipocytes, rather than expression of adipocyte markers in osteoblastic cells because a small number of Oil red O-positive cells were present in rosiglitazone-treated calvaria cells (not shown). These findings demonstrate that, unlike early progenitors, osteoblastic cells in the later stages of differentiation are resistant to the suppressive effects of rosiglitazone.
FIG. 7. Rosiglitazone has no effect on expression of osteoblast-specific transcription factors and matrix proteins in cultures of calvaria-derived osteoblastic cells. Osteoblastic cells from neonatal murine calvaria were grown to confluence, at which time vehicle (dimethylsulfoxide) or 100 nM rosiglitazone was added. RNA was isolated immediately before addition of these agents, and at 2, 4, and 8 d thereafter, and quantified by TaqMan PCR. Data shown represent the mean (±SD) level of expression for each transcript relative to ChoB, determined in triplicate cultures. *, P < 0.05 vs. contemporaneous vehicle control by t test. BSP, Bone sialoprotein.
Discussion
We demonstrate herein that the bone loss caused by administration of rosiglitazone to mice is due at least in part to a reduction in the differentiation of osteoblasts from early progenitors. Specifically, rosiglitazone decreased the ratio of osteoblasts to osteoclasts, reduced the bone formation rate, and reduced wall width. These effects of rosiglitazone were associated with an increase in marrow adipocytes.
The cellular target of the adverse effect of rosiglitazone on the skeleton appears to be downstream of the MSC as evidenced by the lack of effect on the marrow content of CFU-OB or their replication. Moreover, rosiglitazone did not affect osteoblast life span. Instead, in vitro studies indicate that rosiglitazone diverts bipotential mesenchymal progenitors from the osteoblast to the adipocyte lineage, and also suppresses the differentiation of monopotential osteoblast progenitors. Consistent with our previous findings using marrow stromal cell lines (13, 17), these inhibitory effects were strongly associated with increased expression of PPAR2 and a reduction in the expression of Runx2 and Osterix. Decreased expression of these transcription factors, which are required for osteoblastogenesis (6, 7), may explain how rosiglitazone blocked osteoblastogenesis in these cultures as evidenced by the failure to increase expression of matrix proteins like collagen1A1, osteocalcin, and bone siaoloprotein, and to deposit mineral into the extracellular matrix.
In contrast with early marrow-derived osteoblast progenitors, osteoblastic cells in the later stages of differentiation, either formed from marrow progenitors or isolated from neonatal murine calvaria, were resistant to the negative effect of rosiglitazone on the expression of the osteoblast phenotype. The reason for this difference is unclear but may be related to the inability of the ligand to increase PPAR2 expression and/or to decrease Wnt-10b in the more differentiated osteoblastic cells.
PPAR haploinsufficiency in mice causes an increase in bone mass associated with increased osteoblastogenesis and decreased adipogenesis from marrow-derived MSC (16), indicating that PPAR restrains osteoblast differentiation under normal physiological circumstances. We have previously shown that certain oxidized fatty acid ligands of PPAR suppress osteoblast differentiation in vitro (13). Oxidized fatty acids produced by 12/15-lipoxygenase are evidently active in the skeleton because mice lacking this enzyme exhibited increased bone mass, and administration of lipoxygenase inhibitors increased bone mass in wild-type mice (37). Thus, the adverse skeletal effects of rosiglitazone reported herein may reflect increased PPAR-mediated restraint of bone formation beyond that already activated by endogenous oxidized fatty acids, albeit putative PPAR-independent mechanisms of rosiglitazone action have not been excluded. The production of oxidized fatty acids increases with age (38), raising the possibility that age-related bone loss is due at least in part to an increase in the inhibitory effects of PPAR on osteoblast differentiation. Indeed, rosiglitazone reproduced several of the histologic indices of age-related bone loss including increased marrow fat and decreased wall width (2, 3, 4, 5). These findings highlight the need for future studies to establish whether endogenous PPAR2 ligands are involved in age-related bone loss.
The common origin of adipocytes and osteoblasts contributes to the reciprocal changes in the number of these cells caused by rosiglitazone, but suppression of the differentiation of monopotential osteoblast progenitors may also be involved. This situation is reminiscent of the cell autonomous effects of FosB overexpression that attenuates adipocyte differentiation via actions in early mesenchymal progenitors and stimulates bone formation via actions in osteoblasts (39). Moreover, reciprocal changes in marrow adipocyte and osteoblast number may not always accompany PPAR activation. Indeed, we have shown that the ligand 9-hydroxyoctadecadienoic acid stimulates adipogenesis but does not attenuate osteoblast differentiation, whereas 9,10-epoxyoctadecenoic acid suppresses osteoblast differentiation but does not stimulate adipogenesis (13).
The reduction in the ratio of osteoblasts to osteoclasts shown here largely accounts for the reduced trabecular width, wall width, and the extent of bone surface undergoing mineralization in mice receiving rosiglitazone. However, we also noted a decrease in trabecular number and an increase in the spacing between those remaining, indicating that some trabeculae have been removed (40). Osteoclastic perforation of a trabecular plate that was not repaired because of an osteoblast defect could be responsible, but trabeculae could have been lost due to increased bone resorption in the early phase of the experiment, leaving no trace by 28 d of how it had occurred (41). Increased resorption is inconsistent with evidence that activation of PPAR with 15-deoxy-12,14 prostaglandin J2 or thiazolidinediones inhibits rather than stimulates osteoclast differentiation and bone resorption in vitro (42, 43), but is consistent with evidence that, in cultured human aortic smooth muscle cells, thiazolidinediones strongly suppress the synthesis of osteoprotegerin, which antagonizes the pro-osteoclastogenic effects of receptor activator of nuclear factor B ligand (44).
In contrast with our findings, Rzonca et al. (19) reported that rosiglitazone reduced the mineral apposition rate in mice by 3.8-fold. Procedural differences may account for this discrepancy because we measured the entire secondary spongiosa of distal femoral bone and three lumbar vertebrae, whereas Rzonca et al. selected six fields in the proximal tibia—a comparatively small bone that may respond to rosiglitazone in a different fashion than other skeletal sites. Mineral apposition rate mainly reflects the vigor of individual osteoblasts, whereas mineralizing surface is a measure of the number of osteoblasts. The in vitro studies of the present report show that rosiglitazone inhibited the formation of osteoblasts from early progenitors, but had little if any effect on the activity of osteoblasts in the late stages of differentiation. This evidence is consistent with the reduction in mineralizing surface and lack of effect on mineral apposition rate we observed in vivo.
The adverse effect of thiazolidinediones on the rodent skeleton raises the possibility that this class of antidiabetic drugs could cause bone loss. Okazaki et al. (45) reported that, after taking troglitazone for 1 month, type II diabetics exhibited decreased bone remodeling as indicated by a reduction in the bone resorption markers deoxypyridinoline collagen type I C-terminal peptide, as well as the bone formation marker bone specific alkaline phosphatase. However, these markers returned toward baseline after 6 and 12 months of treatment (46). Most studies report normal to increased BMD and reduced fracture incidence in type II diabetics (47, 48, 49), perhaps due to low bone turnover (50). Further studies are needed to investigate whether thiazolidinediones adversely impact the seemingly favorable effects of type II diabetes on the skeleton.
In summary, we have elucidated that, as suggested by earlier in vitro work, rosiglitazone does indeed attenuate osteoblast differentiation and thereby reduces bone formation rate in vivo, leading to bone loss. Strong evidence indicates that activation of PPAR by natural ligands, such as oxidized fatty acids, plays an important role in aging in general (38), and the development of atherosclerosis in particular (51). Furthermore, PPAR activation is now an established negative regulator of bone mass (16, 37), and atherosclerosis and osteoporosis coexist (52). Collectively, these lines of evidence fully justify the need for further work to establish whether increased activation of PPAR secondary to increased level of oxidized fatty acid ligands with age represents a major mechanism for age-related osteoporosis in humans.
Note Added in Proof
While this manuscript was under review, Soroceanu et al. (53) reported that, similar to our findings, administration of rosiglitazone to growing mice for 90 d caused bone loss. Albeit, in contrast to our report, Soroceanu et al. found that the bone loss in their study was associated with a 4-fold increase in the combined prevalence of apoptotic osteoblasts and osteocytes. In our studies, we failed to detect an increase in osteoblast apoptosis (Tables 1 and 2). In addition, when we retrospectively examined vertebral sections for osteocyte apoptosis, we found no change with rosiglitazone administration (vehicle, 3.6 ± 1.3%; rosiglitazone, 4.3% ± 0.6%). The difference between the two sets of observations could be due to differences in the age of the animals or duration of treatment.
Acknowledgments
The authors thank S. Berryhill, T. Chambers, J. Crawford, B. Lecka-Czernik, X. Liu, C. O’Brien, R. Shelton, W. Webb, C. Wiggins III, and R. Wynne for their contributions to this work.
References
Aubin JE, Triffitt JT 2002 Mesenchymal stem cells and osteoblast differentiation. In: Bilezikian JP, Raisz LG, Rodan G, eds. Principles of bone biology. 2nd ed. San Diego: Academic Press; 59–81
Justesen J, Stenderup K, Ebbesen EN, Mosekilde L, Steiniche T, Kassem M 2001 Adipocyte tissue volume in bone marrow is increased with aging and in patients with osteoporosis. Biogerontology 2:165–171
Kajkenova O, Lecka-Czernik B, Gubrij I, Hauser SP, Takahashi K, Parfitt AM, Jilka RL, Manolagas SC, Lipschitz DA 1997 Increased adipogenesis and myelopoiesis in the bone marrow of SAMP6, a murine model of defective osteoblastogenesis and low turnover osteopenia. J Bone Miner Res 12:1772–1779
Meunier P, Aaron J, Edouard C, Vignon G 1971 Osteoporosis and the replacement of cell populations of the marrow by adipose tissue. A quantitative study of 84 iliac bone biopsies. Clin Orth Rel Res 80:147–154
Lips P, Courpron P, Meunier PJ 1978 Mean wall thickness of trabecular bone packets in the human iliac crest: changes with age. Calcif Tissue Res 26:13–17
Ducy P, Starbuck M, Priemel M, Shen J, Pinero G, Geoffroy V, Amling M, Karsenty G 1999 A Cbfa1-dependent genetic pathway controls bone formation beyond embryonic development. Genes Dev 13:1025–1036
Nakashima K, Zhou X, Kunkel G, Zhang Z, Deng JM, Behringer RR, De Crombrugghe B 2002 The novel zinc finger-containing transcription factor osterix is required for osteoblast differentiation and bone formation. Cell 108:17–29
Ren D, Collingwood TN, Rebar EJ, Wolffe AP, Camp HS 2002 PPAR knockdown by engineered transcription factors: exogenous PPAR2 but not PPAR1 reactivates adipogenesis. Genes Dev 16:27–32
Yki-Jarvinen H 2004 Thiazolidinediones. N Engl J Med 351:1106–1118
Forman BM, Tontonoz P, Chen J, Brun RP, Spiegelman BM, Evans RM 1995 15-Deoxy- 12, 14-prostaglandin J2 is a ligand for the adipocyte determination factor PPAR. Cell 83:803–812
Kliewer SA, Sundseth SS, Jones SA, Brown PJ, Wisely GB, Koble CS, Devchand P, Wahli W, Willson TM, Lenhard JM, Lehmann JM 1997 Fatty acids and eicosanoids regulate gene expression through direct interactions with peroxisome proliferator-activated receptors and . Proc Natl Acad Sci USA 94:4318–4323
Lehmann JM, Moore LB, Smith-Oliver TA, Wilkison WO, Willson TM, Kliewer SA 1995 An antidiabetic thiazolidinedione is a high affinity ligand for peroxisome proliferator-activated receptor (PPAR). J Biol Chem 270:12953–12956
Lecka-Czernik B, Moerman EJ, Grant DF, Lehmann JM, Manolagas SC, Jilka RL 2002 Divergent effects of selective peroxisome proliferator-activated receptor-2 ligands on adipocyte versus osteoblast differentiation. Endocrinology 143:2376–2384
Nosjean O, Boutin JA 2002 Natural ligands of PPAR: are prostaglandin J(2) derivatives really playing the part? Cell Signal 14:573–583
Huang JT, Welch JS, Ricote M, Binder CJ, Willson TM, Kelly C, Witztum JL, Funk CD, Conrad D, Glass CK 1999 Interleukin-4-dependent production of PPAR- ligands in macrophages by 12/15-lipoxygenase. Nature 400:378–382
Akune T, Ohba S, Kamekura S, Yamaguchi M, Chung Ui, Kubota N, Terauchi Y, Harada Y, Azuma Y, Nakamura K, Kadowaki T, Kawaguchi H 2004 PPAR insufficiency enhances osteogenesis through osteoblast formation from bone marrow progenitors. J Clin Invest 113:846–855
Lecka-Czernik B, Gubrij I, Moerman EJ, Kajkenova O, Lipschitz DA, Manolagas SC, Jilka RL 1999 Inhibition of Osf2/Cbfa1 expression and terminal osteoblast differentiation by PPAR2. J Cell Biochem 74:357–371
Jennermann C, Triantafillou J, Cowan D, Pennink BGA, Connolly KM, Morris DC 1995 Effects of thiazolidinediones on bone turnover in the rat. J Bone Miner Res 10:S241
Rzonca SO, Suva LJ, Gaddy D, Montague DC, Lecka-Czernik B 2004 Bone is a target for the antidiabetic compound rosiglitazone. Endocrinology 145:401–406
Pickavance LC, Buckingham RE, Wilding JP 2001 Insulin-sensitizing action of rosiglitazone is enhanced by preventing hyperphagia. Diabetes Obes Metab 3:171–180
De VP, Lefebvre AM, Miller SG, Guerre-Millo M, Wong K, Saladin R, Hamann LG, Staels B, Briggs MR, Auwerx J 1996 Thiazolidinediones repress ob gene expression in rodents via activation of peroxisome proliferator-activated receptor . J Clin Invest 98:1004–1009
Cantello BC, Cawthorne MA, Cottam GP, Duff PT, Haigh D, Hindley RM, Lister CA, Smith SA, Thurlby PL 1994 [[-(Heterocyclylamino)alkoxy]benzyl]-2,4-thiazolidinediones as potent antihyperglycemic agents. J Med Chem 37:3977–3985
Jilka RL, Weinstein RS, Takahashi K, Parfitt AM, Manolagas SC 1996 Linkage of decreased bone mass with impaired osteoblastogenesis in a murine model of accelerated senescence. J Clin Invest 97:1732–1740
Bellido T, Ali AA, Plotkin LI, Fu Q, Gubrij I, Roberson PK, Weinstein RS, O’Brien CA, Manolagas SC, Jilka RL 2003 Proteasomal degradation of Runx2 shortens parathyroid hormone-induced anti-apoptotic signaling in osteoblasts: a putative explanation for why intermittent administration is needed for bone anabolism. J Biol Chem 278:50259–50272
Weinstein RS, Jilka RL, Parfitt AM, Manolagas SC 1998 Inhibition of osteoblastogenesis and promotion of apoptosis of osteoblasts and osteocytes by glucocorticoids: potential mechanisms of their deleterious effects on bone. J Clin Invest 102:274–282
Weinstein RS, Jilka RL, Parfitt AM, Manolagas SC 1997 The effects of androgen deficiency on murine bone remodeling and bone mineral density are mediated via cells of the osteoblastic lineage. Endocrinology 138:4013–4021
Parfitt AM, Drezner MK, Glorieux FH, Kanis JA, Malluche H, Meunier PJ, Ott SM, Recker RR 1987 Bone histomorphometry: standardization of nomenclature, symbols, and units. Report of the ASBMR Histomorphometry Nomenclature Committee. J Bone Miner Res 2:595–610
Weinstein RS, Jia D, Powers CC, Stewart SA, Jilka RL, Parfitt AM, Manolagas SC 2004 The skeletal effects of glucocorticoid excess override those of orchidectomy in mice. Endocrinology 145:1980–1987
Fu Q, Jilka RL, Manolagas SC, O’Brien CA 2002 Parathyroid hormone stimulates receptor activator of NFB ligand and inhibits osteoprotegerin expression via protein kinase A activation of cAMP-response element-binding protein. J Biol Chem 277:48868–48875
Di Gregorio GB, Yamamoto M, Ali AA, Abe E, Roberson P, Manolagas SC, Jilka RL 2001 Attenuation of the self-renewal of transit-amplifying osteoblast progenitors in the murine bone marrow by 17?-estradiol. J Clin Invest 107:803–812
Jilka RL 1986 Parathyroid hormone-stimulated development of osteoclasts in cultures of cells from neonatal murine calvaria. Bone 7:29–40
Kuehl RO 2000 Design of experiments: statistical principles of research design and analysis. New York: Duxbury; 492–519
Tai TA, Jennermann C, Brown KK, Oliver BB, MacGinnitie MA, Wilkison WO, Brown HR, Lehmann JM, Kliewer SA, Morris DC, Graves RA 1996 Activation of the nuclear receptor peroxisome proliferator-activated receptor promotes brown adipocyte differentiation. J Biol Chem 271:29909–29914
Stein GS, Lian JB, Stein JL, Van Wijnen AJ, Montecino M 1996 Transcriptional control of osteoblast growth and differentiation. Physiol Rev 76:593–629
Mueller E, Drori S, Aiyer A, Yie J, Sarraf P, Chen H, Hauser S, Rosen ED, Ge K, Roeder RG, Spiegelman BM 2002 Genetic analysis of adipogenesis through peroxisome proliferator-activated receptor isoforms. J Biol Chem 277:41925–41930
Ross SE, Hemati N, Longo KA, Bennett CN, Lucas PC, Erickson RL, MacDougald OA 2000 Inhibition of adipogenesis by Wnt signaling. Science 289:950–953
Klein RF, Allard J, Avnur Z, Nikolcheva T, Rotstein D, Carlos AS, Shea M, Waters RV, Belknap JK, Peltz G, Orwoll ES 2004 Regulation of bone mass in mice by the lipoxygenase gene Alox15. Science 303:229–232
Spiteller G 2000 Peroxidation of linoleic acid and its relation to aging and age dependent diseases. Ageing Res Rev 122:617–657
Kveiborg M, Sabatakos G, Chiusaroli R, Wu M, Philbrick WM, Horne WC, Baron R 2004 FosB Induces osteosclerosis and decreases adipogenesis by two independent cell-autonomous mechanisms. Mol Cell Biol 24:2820–2830
Parfitt AM, Mathews CHE, Villanueva AR, Kleerekoper M, Frame B, Rao DS 1983 Relationships between surface, volume and thickness of iliac trabecular bone in aging and in osteoporosis. Implications for the microanatomic and cellular mechanism of bone loss. J Clin Invest 72:1396–1409
Parfitt AM 1988 Bone remodeling: relationship to the amount and structure of bone and the pathogenesis and prevention of fractures. In: Riggs BL, Melton LJ, eds. Osteoporosis—etiology, diagnosis and management. New York: Raven Press; 45–94
Mbalaviele G, Abu-Amer Y, Meng A, Jaiswal R, Beck S, Pittenger MF, Thiede MA, Marshak DR 2000 Activation of peroxisome proliferator-activated receptor- pathway inhibits osteoclast differentiation. J Biol Chem 275:14388–14393
Okazaki R, Toriumi M, Fukumoto S, Miyamoto M, Fujita T, Tanaka K, Takeuchi Y 1999 Thiazolidinediones inhibit osteoclast-like cell formation and bone resorption in vitro. Endocrinology 140:5060–5065
Fu M, Zhang J, Lin Y, Zhu X, Willson TM, Chen YE 2002 Activation of peroxisome proliferator-activated receptor inhibits osteoprotegerin gene expression in human aortic smooth muscle cells. Biochem Biophys Res Commun 294:597–601
Okazaki R, Miura M, Toriumi M, Taguchi M, Hirota Y, Fukumoto S, Fujita T, Tanaka K, Takeuchi A 1999 Short-term treatment with troglitazone decreases bone turnover in patients with type 2 diabetes mellitus. Endocr J 46:795–801
Watanabe S, Takeuchi Y, Fukumoto S, Fujita H, Nakano T, Fujita T 2003 Decrease in serum leptin by troglitazone is associated with preventing bone loss in type 2 diabetic patients. J Bone Miner Metab 21:166–171
Hanley DA, Brown JP, Tenenhouse A, Olszynski WP, Ioannidis G, Berger C, Prior JC, Pickard L, Murray TM, Anastassiades T, Kirkland S, Joyce C, Joseph L, Papaioannou A, Jackson SA, Poliquin S, Adachi JD 2003 Associations among disease conditions, bone mineral density, and prevalent vertebral deformities in men and women 50 years of age and older: cross-sectional results from the Canadian Multicentre Osteoporosis Study. J Bone Miner Res 18:784–790
Stolk RP, van Daele PL, Pols HA, Burger H, Hofman A, Birkenhager JC, Lamberts SW, Grobbee DE 1996 Hyperinsulinemia and bone mineral density in an elderly population: The Rotterdam Study. Bone [Erratum (1996) 19:566] 18:545–549
Lunt M, Masaryk P, Scheidt-Nave C, Nijs J, Poor G, Pols H, Falch JA, Hammermeister G, Reid DM, Benevolenskaya L, Weber K, Cannata J, O’Neill TW, Felsenberg D, Silman AJ, Reeve J 2001 The effects of lifestyle, dietary dairy intake and diabetes on bone density and vertebral deformity prevalence: the EVOS study. Osteoporos Int 12:688–698
Krakauer JC, McKenna MJ, Buderer NF, Rao DS, Whitehouse FW, Parfitt AM 1995 Bone loss and bone turnover in diabetes. Diabetes 44:775–782
Barish GD, Evans RM 2004 PPARs and LXRs: atherosclerosis goes nuclear. Trends Endocrinol Metab 15:158–165
Schulz E, Arfai K, Liu X, Sayre J, Gilsanz V 2004 Aortic calcification and the risk of osteoporosis and fractures. J Clin Endocrinol Metab 89:4246–4253
Soroceanu MA, Miao D, Bai XY, Su H, Goltzman D, Karaplis AC 2004 Rosiglitazone impacts negatively on bone by promoting osteoblast/osteocyte apoptosis. J Endocrinol 183:203–216(A. Afshan Ali, Robert S. )
Address all correspondence and requests for reprints to: Robert L. Jilka, Ph.D., Division of Endocrinology and Metabolism, Slot 587, University of Arkansas for Medical Sciences, 4301 West Markham, Little Rock, Arkansas 72205. E-mail: jilkarobertl@uams.edu.
Abstract
Because osteoblasts and marrow adipocytes are derived from a common mesenchymal progenitor, increased adipogenesis may occur at the expense of osteoblasts, leading to bone loss. Our previous in vitro studies indicated that activation of the proadipogenic transcription factor peroxisome proliferator-activated receptor isoform 2 with rosiglitazone suppressed osteoblast differentiation. Here, we show that 5-month-old Swiss-Webster mice receiving rosiglitazone for 28 d exhibited bone loss associated with an increase in marrow adipocytes, a decrease in the ratio of osteoblasts to osteoclasts, a reduction in bone formation rate, and a reduction in wall width—an index of the amount of bone formed by each team of osteoblasts. Rosiglitazone had no effect on the number of early osteoblast or osteoclast progenitors, or on osteoblast life span, but decreased the expression of the key osteoblastogenic transcription factors Runx2 and Osterix in cultures of marrow-derived mesenchymal progenitors. These effects were associated with diversion of bipotential progenitors from the osteoblast to the adipocyte lineage, and suppression of the differentiation of monopotential osteoblast progenitors. However, rosiglitazone had no effect on osteoblastic cells at later stages of differentiation. Hence, rosiglitazone attenuates osteoblast differentiation and thereby reduces bone formation rate in vivo, leading to bone loss. These findings provide a mechanistic explanation for the recent evidence that peroxisome proliferator-activated receptor isoform activation is a negative regulator of bone mass and suggest that the increased production of oxidized fatty acids with age may indeed be an important mechanism for age-related osteoporosis in humans.
Introduction
OSTEOBLASTS AND MARROW adipocytes are derived from a common multipotential mesenchymal stem cell (MSC) progenitor (1). It has been hypothesized that the close relationship between these lineages underlies the reciprocal relationship between increased adipocytes and decreased bone formation that occurs during aging (2, 3, 4). Indeed, age-related bone loss is associated with a deficit in the number of osteoblasts needed to replace the bone removed by osteoclasts during remodeling as evidenced by a decrease in mean wall thickness (5).
Lineage commitment of MSCs is determined by expression and/or activation of specific transcription factors: Runx2 and Osterix in the case of osteoblasts (6, 7), and CCAAT enhancer binding protein ? and peroxisome proliferator-activated receptor isoform (PPAR) in the case of adipocytes (8). PPAR is broadly expressed and plays an important role not only in adipogenesis but also in glucose homeostasis, atherosclerosis, and cancer (9). It is a member of the nuclear hormone receptor family of transcription factors, and exists in two isoforms, PPAR1 and PPAR2, resulting from alternative promoter usage and alternative splicing. The transcriptional activity of PPAR is induced by a variety of ligands including 15-deoxy-12,14 prostaglandin J2, oxidized fatty acid metabolites, as well as synthetic thiazolidinediones like troglitazone and rosiglitazone (10, 11, 12, 13, 14, 15).
Mounting evidence indicates an important role of PPAR in skeletal metabolism. Specifically, PPAR haploinsufficient mice exhibit increased bone mass associated with increased osteoblastogenesis and decreased adipogenesis (16). Using murine marrow-derived MSCs, as well as a model marrow-derived stromal cell line overexpressing PPAR2, we have shown that activation of PPAR2 with rosiglitazone stimulates adipogenesis and inhibits osteoblastogenesis, at least in part by down-regulating Runx2 (17); and that fatty acid peroxidation products such as 9,10-dihydroxyoctadecenoic acid and 9,10-epoxyoctadecenoic acid can attenuate osteoblast differentiation in a PPAR2-dependent fashion (13). We have therefore hypothesized that increased PPAR activation may contribute to the reduced osteoblast production and bone loss that occurs with aging.
Rosiglitazone causes bone loss in rats and mice (18, 19), but the cellular basis underlying this skeletal response has not been determined. Here, we report that adult mice receiving rosiglitazone exhibit bone loss characterized by an increase in marrow adipocytes, and a decrease in the osteoblast/osteoclast ratio, wall width, and bone formation rate. In addition, we show that rosiglitazone diverted bipotential mesenchymal progenitors from the osteoblast to the adipocyte lineage and also suppressed the differentiation of monopotential osteoblast progenitors. These inhibitory effects were strongly associated with a reduction in the expression of Runx2 and Osterix.
Materials and Methods
Animals
Swiss-Webster mice were obtained from Harlan (Indianapolis, IN). The University of Arkansas for Medical Sciences Division of Laboratory Animal Medicine approved the animal use protocol. Food was blended with rosiglitazone (Tularik, Inc., South San Francisco, CA), which was dissolved at 4.25 mg/ml of a 1:3 mixture of ethanol and glyceroloctaside (Stepan Chemical, Maywood, NJ), 0.1% Tenox 20 (Eastman Chemical, Kingsport, TN) before mixing with a blender to give a concentration of 0.03 mg or 0.15 mg of rosiglitazone/g normal rodent chow (Agway RMH 3000; Arlington Heights, IL). Mice were pair fed to avoid potential artifacts secondary to the hyperphagia that has been reported to occur when rats or mice are given a thiazolidinedione (20, 21). Five-month-old mice were housed singly, and average daily consumption of vehicle chow was determined to be 5–6 g/d over a 2-wk period. Animals were fed 5 g chow/d containing vehicle (four males and two females), 5 μg/g·d rosiglitazone (two males and three females), or 25 μg/g rosiglitazone (three males and three females) for 28 d. These doses have previously been shown to give moderate to strong antihyperglycemic activity in diabetic ob/ob mice (22). By weighing the food dish before placement in the cage and after removal, it was determined that each mouse consumed more than 90% of the chow throughout the course of the experiment. Tetracycline (30 mg/kg) was administered ip at 8 and 3 d before the end of the experiment.
Bone densitometry
Dual-energy x-ray absorptiometry (DEXA) was used to determine bone mineral density (BMD) as previously described using a Hologic 2000 Plus (19, 23). Over the past 3 yr, the coefficient of variation of the measurement done on a plastic-embedded whole mouse skeleton was 1.8% (n = 202). BMD determinations were performed before initiation of the study and used to allocate the animals into groups with equivalent spinal BMD values before entry into the treatment protocol. Measurements were made again after 2 wk, and at termination of the experiment after 4 wk of treatment using the Compare program of Hologic software to ensure accurate repositioning for serial measurements. Global BMD refers to the entire skeleton excluding the skull. Spinal and hindlimb subregions were measured as previously described (23).
Bone histomorphometry
The left femur and the lumbar vertebrae (L1–L4) were fixed, embedded undecalcified in methyl methacrylate, and stained using modified Massons’s trichrome (23, 24, 25). Histomorphometric measurements were performed using a computer and digitizer tablet (Osteomeasure version 3.00, Osteometrics Inc., Atlanta, GA) interfaced to a Zeiss Axioscope (Carl Zeiss Inc., Thornwood, NY) with a drawing tube attachment. The identity of the sample was unknown to the reader. Measurements were made with a Zeiss Plan-Neofluar 40 x (0.75 numerical aperture) objective on the entire secondary spongiosa, which comprised approximately 60 fields in at least three of the vertebrae from L1–L4, and approximately 30 fields in the distal femur. Variables were measured and reported as previously described (23, 25, 26) using the terminology recommended by the Histomorphometry Nomenclature Committee of American Society for Bone and Mineral Research (27). The number of fat cells within the tissue area measured, as well as individual fat cell area and diameter, were also determined. They are expressed as total fat cell number per tissue area, and the diameter of fat cells. Osteoclast and osteoblast number per tissue area are reported to facilitate direct comparison to fat cell number.
Apoptotic osteoblasts were detected by in situ nick end labeling using Klenow terminal deoxynucleotidyl transferase (Oncogene Research Products, Cambridge, MA) as previously described (28). Vertebrae from orchidectomized adult mice were used as a positive control. Slides were counterstained with 2% methyl green. Apoptotic osteoblasts were identified as in situ nick end labeling-positive cells lining the osteoid covered cancellous bone surface.
Quantification of osteoclast and osteoblast progenitors
Marrow cells were obtained from the right femur of each animal and replicate cultures established for the determination of osteoclast progenitors and colony-forming osteoblast progenitors (CFU-OB) as previously described (29, 30). Briefly, osteoclast progenitors were determined by coculturing marrow cells with UAMS-32 stromal/osteoblastic cells for 8 d in the presence 10 nM 1,25(OH)2D3. Osteoclastic cells were enumerated after staining for tartrate-resistant acid phosphatase. CFU-OB number was determined by culturing cells for 25 d with irradiated guinea pig feeder cells in MEM supplemented with 15% preselected fetal bovine serum and 1 mM ascorbate-2-phosphate to promote osteoblastogenesis. Colonies containing osteoblasts were visualized by Von Kossa staining.
CFU-OB replication in vitro was determined as described previously (30). One aliquot of cells was used to determine CFU-OB number in the initial isolate, as described above. A second aliquot was used to establish replicate cultures of cells in type I collagen gels, which were then maintained in the absence or presence of 100 nM rosiglitazone for 6 d. The cells were dispersed using bacterial collagenase, and the number of CFU-OB within each gel was determined using the colony-forming assay described above. To calculate the fold increase in CFU-OB during culture in the collagen gels, the number of CFU-OB obtained per gel (after 6 d of culture) was divided by the number of CFU-OB initially added to the collagen gel.
The effect of rosiglitazone on adipocyte and osteoblast differentiation in vitro was determined by culturing bone marrow cells at 2.5 x 106 per 10 cm2 well with irradiated guinea pig cells and maintained for 25 d in medium supplemented with 1 mM ascorbate-2-phosphate, either in the absence or presence of 100 nM rosiglitazone, added at various times after establishment of the culture. On d 25, cells were stained either with Oil Red O to visualize colonies containing adipocytes or Von Kossa stain to visualize colonies containing osteoblasts elaborating a mineralized matrix.
Quantification of gene expression in cultured bone marrow cells and osteoblastic cells
Murine bone marrow cells were cultured as described above. Osteoblastic cells were obtained from neonatal murine calvarial by collagenase digestion as previously described (31), and cultured in MEM supplemented with 15% preselected fetal bovine serum and 1 mM ascorbate-2-phosphate. Total RNA was extracted using Ultraspec reagent (Biotecx Laboratories, Inc., Houston, TX). RNA (2 μg) was reverse-transcribed using a High Capacity cDNA Archive Kit (Applied Biosystems, Foster City, CA). The transcripts of interest, and that of the housekeeping gene ribosomal protein S2 (ChoB), were amplified from first-strand cDNA by real-time PCR using TaqMan Universal PCR Master Mix and Assay by Demand or Assay by Design primer and probe sets (Applied Biosystems). Amplification and detection were carried out with an ABI Prism 7300 Sequence Detection System (Applied Biosystems) as follows: 5-min denaturation at 95 C for 10 min, 40 cycles of amplification including denaturation at 94 C for 15 sec and annealing/extension at 60 C for 1 min. Gene expression was quantified by subtracting the ChoB threshold cycle (Ct) value from the Ct value of the gene of interest, and expressed as 2-Ct, as described by the protocol of the manufacturer.
Statistics
Data were analyzed using SigmaStat (SPSS Science, Chicago, IL) or SAS software (SAS Institute Inc., Cary, NC). All values are reported as the mean ± SD. Differences between group means in histomorphometry studies were evaluated with Student’s t test, or by ANOVA after P values were adjusted with Bonferroni’s correction. Serial changes in body weight and BMD were analyzed using a model of repeated measures (32). Pearson correlation analysis of fat cell number with osteoblast number and mineralizing surface was done after adding three to total fat cell number (to avoid large negative values for mice with no adipocytes), and then taking the natural logarithm to obtain a normal distribution. Differences in group means in studies of gene expression and colony number were evaluated with Student’s t test.
Results
Rosiglitazone causes bone loss and decreases osteoblast number and bone formation rate
Rosiglitazone was fed to adult (5 months old) male and female Swiss-Webster mice at 5 or 25 μg/g body weight·d for 28 d. Vehicle was added to the food of control mice. There was no change in body weight in mice receiving the vehicle, but both doses of rosiglitazone caused an approximately 2% increase in body weight (Fig. 1A). As reported previously (33), administration of 25 μg/g·d of rosiglitazone increased interscapular fat (Fig. 1B), indicating that drug administration was efficacious. Liver weight was unaffected (Fig. 1C).
FIG. 1. Effect of rosiglitazone on body weight, interscapular brown adipose tissue (BAT) weight, and liver weight. Values represent percentage of change in body weight from baseline determined at the beginning of the experiment (A) or tissue weight at the end of the experiment (B and C). *, P < 0.01 vs. vehicle as determined by repeated measures analysis (body weight) or ANOVA (BAT and liver weight).
Serial bone densitometry using DEXA revealed that after 28 d of administration, mice given rosiglitazone at 25 μg/g·d exhibited a decrease in vertebral BMD, compared with mice fed the vehicle control (Fig. 2). The same trend was noted in mice fed this dose of rosiglitazone when BMD was measured over the entire skeleton (global) and in hindlimbs, but the changes were not statistically significant. The 5 μg/g·d dose of rosiglitazone had no discernable effect on BMD. Thus, histologic and histomorphometric studies were performed only on mice fed the 25-μg dose.
FIG. 2. Rosiglitazone causes loss of BMD. Values represent percentage of change in BMD from the baseline value determined at the beginning of the experiment. Repeated measures analysis indicated a significant group x time interaction for spinal BMD (P = 0.007) and indicated a stronger negative effect for the 25 μg/g dose of rosiglitazone than the 5 μg/g dose.
Mice receiving rosiglitazone exhibited increased marrow fat in both vertebral and femoral bone compared with mice receiving the vehicle control (Fig. 3). Vertebral bone from control animals was practically devoid of adipocytes, but rosiglitazone increased their number by approximately 20-fold (Table 1). Adipocyte number in femoral metaphyseal bone of control mice was considerably higher than that of vertebral bone, and rosiglitazone caused a further 7-fold increase (Table 2). Fat cell diameter was unaffected, indicating that rosiglitazone did not enhance lipogenesis in existing adipocytes. Consistent with the densitometric findings, rosiglitazone caused loss of cancellous vertebral bone that was characterized by a decrease in trabecular width, an increase in trabecular spacing, and a decrease in trabecular number (Table 1). Similar microarchitectural changes were noted in femoral cancellous bone (Table 2).
FIG. 3. Increased marrow fat and decreased cancellous bone in mice receiving rosiglitazone. Nondecalcified bone sections were stained with Masson trichrome stain. Calcified bone is stained blue; adipocytes were present in the white globular areas. The lower panels of each set of photomicrographs represent high-power views of the field denoted by the yellow square in the upper panel.
TABLE 1. Effect of rosiglitazone on vertebral cancellous bone
TABLE 2. Effect of rosiglitazone on distal femoral cancellous bone
Rosiglitazone caused a reduction in the number of osteoblasts per square millimeter of vertebral and femoral cancellous bone (Tables 1 and 2). However, when expressed on a bone perimeter rather than bone area referent, osteoblast number did not exhibit a statistically significant decline, nor was osteoid surface affected in mice receiving rosiglitazone. The drug had no effect on the prevalence of apoptotic osteoblasts. Osteoclast number was also unaffected, whether expressed per millimeter of bone perimeter or square millimeter of bone area. However, osteoblast number in both vertebral and cancellous bone declined by 2-fold when expressed as the ratio of osteoblasts to osteoclasts, calculated from the number of each cell type per millimeter of bone perimeter. Moreover, there was an inverse correlation between adipocyte number and osteoblast number whether expressed per square millimeter of tissue area or as the ratio of osteoblasts to osteoclasts, in both vertebral and femoral cancellous bone (Table 3). The biological impact of the reduction in the ratio of osteoblasts to osteoclasts in mice receiving rosiglitazone was revealed by a decrease in vertebral bone wall width (Table 1).
TABLE 3. Inverse correlation between marrow adipocytes and osteoblast number, as well as mineralizing surface
These histological findings establish that rosiglitazone caused a deficit in the number of osteoblasts when compared with the number needed to refill the resorption cavities created by osteoclasts. Consistent with this, rosiglitazone caused a 2-fold decrease in bone formation rate in vertebral bone (Table 1). This was due to a reduction in mineralizing surface, rather than mineral apposition rate. Fat cell number inversely correlated with mineralizing surface (Table 3). Rosiglitazone had no effect on width of the osteoid layer between osteoblasts and mineralized bone, or on mineralization lag time, the interval between deposition and mineralization of osteoid. Therefore, the drug did not cause defective mineralization. The small amount of cancellous bone that remained in the femurs of mice fed rosiglitazone (2–6% of tissue area) precluded ascertainment of potential changes in wall width or indices of bone formation.
Rosiglitazone inhibits the differentiation of osteoblasts from CFU-OB
Analysis of ex vivo cultures of mice from the experiment described above established that rosiglitazone had no effect on the number of MSCs, as measured by ex vivo assay of CFU-OB, or on the number of osteoclast progenitors (Fig. 4). Consistent with the lack of effect on CFU-OB number, rosiglitazone also had no effect on the replication of CFU-OB in vitro. Specifically, progenitors increased by 4.3 ± 0.7-fold during 6 d of culture with vehicle vs. 3.5 ± 0.8-fold in the presence of 100 nM rosiglitazone (n = 3 per group). Thus, the reduction in osteoblast number and bone formation rate seen in mice receiving rosiglitazone must be due to actions of the drug on progeny of MSC, as measured by CFU-OB (30).
FIG. 4. Rosiglitazone has no effect on the number of osteoclast and osteoblast progenitors in the bone marrow. Femoral marrow cells were obtained from the animals receiving vehicle (veh) or rosiglitazone (rosi) and ex vivo femoral marrow cell cultures were established for the quantification of osteoblast progenitors (CFU-OB) (A) or osteoclast progenitors (B).
The production of osteoblasts from MSC in vitro proceeds via sequential stages of replication and lineage commitment during the first 2 wk of culture, after which development of the mature matrix-synthesizing osteoblast occurs (30, 34). To determine the stage at which rosiglitazone influences osteoblast differentiation, the drug was added on d 3, 7, 11, or 15 after establishment of ex vivo cultures of femoral marrow cell cultures maintained in ascorbate-2-phosphate to stimulate osteoblast differentiation. Calcium deposition and osteocalcin secretion, markers of differentiated osteoblasts, were quantified after 25 d of culture. When rosiglitazone was added on d 3 or d 7 of culture, calcium deposition and osteocalcin secretion were reduced by 90% or more, whereas only a modest inhibitory effect was observed when the drug was added on d 11 or 15 (Fig. 5A).
FIG. 5. Rosiglitazone inhibits osteoblastogenesis only when added in the early phase of differentiation. Marrow cell cultures from femurs of 5-month-old Swiss-Webster mice were established as described in Materials and Methods. Vehicle (dimethylsulfoxide) or 100 nM rosiglitazone was added to two sets of parallel cultures on d 3, 7, 11, or 15. On d 25, one set of cultures (A) was analyzed for the content of osteocalcin in the medium, and for the content of calcium in the matrix. The second set of cultures (B–F) was stained with Von Kossa to visualize colonies containing mineralizing osteoblasts and with Oil Red O to visualize adipocytes. Data shown represent the mean (±SD) of triplicate cultures. *, P < 0.05 vs. vehicle control at the same time point, by ANOVA.
The influence of rosiglitazone on the differentiation of colony-forming cells was examined in more detail in a parallel set of cultures using Von Kossa and Oil Red O staining to detect the presence of osteoblasts and adipocytes, respectively. Rosiglitazone reduced the total number of colonies by approximately 50% when added on d 3 or 7 (Fig. 5B), as we had previously observed when rosiglitazone was added on the first day of culture (13); however, the drug had no effect on colony number when added after d 7. Colonies containing only adipocytes, only osteoblasts, both cell types, or neither cell type were quantified (Fig. 5, C–F). Most of the colony-forming cells gave rise to adipocytes when rosiglitazone was added on d 3, and osteoblastogenesis was practically abolished (Fig. 5, C–E). The percentage of colonies containing only adipocytes declined progressively with later addition of rosiglitazone, suggesting loss of the ability to differentiate into adipocytes with longer culture time in pro-osteogenic medium (Fig. 5E). When rosiglitazone was added on d 7, the number of colonies containing only osteoblasts was still greatly reduced, but this inhibitory effect was much less evident when the ligand was added on d 11 or 15 (Fig. 5C).
When rosiglitazone was added on d 7 or later, approximately one third of the colonies contained both osteoblasts and adipocytes (Fig. 5D). Such colonies must have arisen from a progenitor that gives rise to two kinds of daughter cells: one that can be driven into the adipocyte lineage with rosiglitazone, and one that differentiates into osteoblasts in the presence ascorbate-2-phosphate. We have designated this colony-forming cell as a bipotential progenitor but still recognize that it could differentiate into other cell types under the proper conditions. When rosiglitazone is added on d 3, few colonies containing both osteoblasts and adipocytes were observed. This indicates that rosiglitazone diverted bipotential progenitors into the adipocyte lineage. However, the ligand could also have suppressed the differentiation of osteoblasts from a monopotential osteoblastogenic daughter of the bipotential cell. Finally, about 20% of colonies contained neither osteoblasts nor adipocytes, but their number was also reduced by rosiglitazone at d 3 and 7, perhaps accounting for part of the decrease in overall colony number at these times (Fig. 5F).
To gain additional insight on the suppressive effect of rosiglitazone on osteoblastogenesis, we examined the expression of osteoblast- and adipocyte-specific transcription factors and proteins using real-time PCR. In this experiment, the ligand was added to cultured bone marrow cells on d 7, i.e. before expression of the osteoblast phenotype as indicated by the lack of secretion of osteocalcin (Fig. 5A). Transcript levels were determined just before addition of the ligand, and at d 10, 15, and 25 of culture. On d 7, PPAR2 transcripts were approximately 10-fold less than PPAR1 transcripts, but rosiglitazone rapidly increased expression of PPAR2 by approximately 30-fold, consistent with previous evidence that PPAR stimulates its own expression (35). PPAR1 transcripts were either not affected or modestly suppressed at d 15 of culture (Fig. 6). More importantly, rosiglitazone fully prevented the increase in expression of Runx2 and Osterix, which are required for osteoblast differentiation. This effect coincided with a failure to induce expression of the bone matrix proteins Col1A1, osteocalcin and bone sialoprotein, consistent with the findings of Fig. 5 showing almost complete suppression of osteoblast differentiation when rosiglitazone was added on d 7 of culture.
FIG. 6. Rosiglitazone reduces expression of osteoblast-specific transcription factors and matrix proteins, but up-regulates PPAR2 and adipogenesis in cultured bone marrow cells. Marrow cell cultures were established as in Fig. 5, and vehicle (dimethylsulfoxide) or 100 nM rosiglitazone was added on d 7. RNA was isolated immediately before addition of these agents, and on d 10, 15, and 25 of culture. The indicated transcripts, and the housekeeping gene ChoB, were quantified by TaqMan PCR. Data shown represent the mean (±SD) level of expression for each transcript relative to ChoB, determined in triplicate cultures. *, P < 0.05 vs. contemporaneous vehicle control by t test. BSP, Bone sialoprotein.
Rosiglitazone also down-regulated Wnt-10b expression, a response that is required for expression of the adipocyte phenotype (36) (Fig. 6). In addition, expression of the adipocyte markers fatty acid binding protein 4 (aP2), adiponectin, and lipoprotein lipase (LPL) were dramatically increased, consistent with the up-regulation of PPAR2 and the important role of this form of the transcription factor in ligand-stimulated adipogenesis (35). These findings confirm the data of Fig. 5 showing a dramatic increase in Oil Red O-positive adipocytes when rosiglitazone was added on d 7 of culture.
In view of the evidence of Fig. 5 that rosiglitazone had no effect on osteocalcin secretion or mineralization when added to bone marrow cultures at late stages of culture, we examined the effect of rosiglitazone on gene expression in preparations of osteoblastic cells obtained from neonatal murine calvaria. In contrast to the response of bone marrow progenitors, addition of rosiglitazone did not increase PPAR2 or decrease Wnt-10b transcripts in calvarial osteoblastic cells (Fig. 7). Moreover, transcript levels of osteoblast-specific transcription factors and matrix proteins were unaffected by the ligand. The adipocyte markers aP2, adiponectin ,and LPL were increased, but the absolute level of expression achieved by addition of rosiglitazone was approximately 20-fold lower than in bone marrow cells. This probably reflects the formation of only a few adipocytes, rather than expression of adipocyte markers in osteoblastic cells because a small number of Oil red O-positive cells were present in rosiglitazone-treated calvaria cells (not shown). These findings demonstrate that, unlike early progenitors, osteoblastic cells in the later stages of differentiation are resistant to the suppressive effects of rosiglitazone.
FIG. 7. Rosiglitazone has no effect on expression of osteoblast-specific transcription factors and matrix proteins in cultures of calvaria-derived osteoblastic cells. Osteoblastic cells from neonatal murine calvaria were grown to confluence, at which time vehicle (dimethylsulfoxide) or 100 nM rosiglitazone was added. RNA was isolated immediately before addition of these agents, and at 2, 4, and 8 d thereafter, and quantified by TaqMan PCR. Data shown represent the mean (±SD) level of expression for each transcript relative to ChoB, determined in triplicate cultures. *, P < 0.05 vs. contemporaneous vehicle control by t test. BSP, Bone sialoprotein.
Discussion
We demonstrate herein that the bone loss caused by administration of rosiglitazone to mice is due at least in part to a reduction in the differentiation of osteoblasts from early progenitors. Specifically, rosiglitazone decreased the ratio of osteoblasts to osteoclasts, reduced the bone formation rate, and reduced wall width. These effects of rosiglitazone were associated with an increase in marrow adipocytes.
The cellular target of the adverse effect of rosiglitazone on the skeleton appears to be downstream of the MSC as evidenced by the lack of effect on the marrow content of CFU-OB or their replication. Moreover, rosiglitazone did not affect osteoblast life span. Instead, in vitro studies indicate that rosiglitazone diverts bipotential mesenchymal progenitors from the osteoblast to the adipocyte lineage, and also suppresses the differentiation of monopotential osteoblast progenitors. Consistent with our previous findings using marrow stromal cell lines (13, 17), these inhibitory effects were strongly associated with increased expression of PPAR2 and a reduction in the expression of Runx2 and Osterix. Decreased expression of these transcription factors, which are required for osteoblastogenesis (6, 7), may explain how rosiglitazone blocked osteoblastogenesis in these cultures as evidenced by the failure to increase expression of matrix proteins like collagen1A1, osteocalcin, and bone siaoloprotein, and to deposit mineral into the extracellular matrix.
In contrast with early marrow-derived osteoblast progenitors, osteoblastic cells in the later stages of differentiation, either formed from marrow progenitors or isolated from neonatal murine calvaria, were resistant to the negative effect of rosiglitazone on the expression of the osteoblast phenotype. The reason for this difference is unclear but may be related to the inability of the ligand to increase PPAR2 expression and/or to decrease Wnt-10b in the more differentiated osteoblastic cells.
PPAR haploinsufficiency in mice causes an increase in bone mass associated with increased osteoblastogenesis and decreased adipogenesis from marrow-derived MSC (16), indicating that PPAR restrains osteoblast differentiation under normal physiological circumstances. We have previously shown that certain oxidized fatty acid ligands of PPAR suppress osteoblast differentiation in vitro (13). Oxidized fatty acids produced by 12/15-lipoxygenase are evidently active in the skeleton because mice lacking this enzyme exhibited increased bone mass, and administration of lipoxygenase inhibitors increased bone mass in wild-type mice (37). Thus, the adverse skeletal effects of rosiglitazone reported herein may reflect increased PPAR-mediated restraint of bone formation beyond that already activated by endogenous oxidized fatty acids, albeit putative PPAR-independent mechanisms of rosiglitazone action have not been excluded. The production of oxidized fatty acids increases with age (38), raising the possibility that age-related bone loss is due at least in part to an increase in the inhibitory effects of PPAR on osteoblast differentiation. Indeed, rosiglitazone reproduced several of the histologic indices of age-related bone loss including increased marrow fat and decreased wall width (2, 3, 4, 5). These findings highlight the need for future studies to establish whether endogenous PPAR2 ligands are involved in age-related bone loss.
The common origin of adipocytes and osteoblasts contributes to the reciprocal changes in the number of these cells caused by rosiglitazone, but suppression of the differentiation of monopotential osteoblast progenitors may also be involved. This situation is reminiscent of the cell autonomous effects of FosB overexpression that attenuates adipocyte differentiation via actions in early mesenchymal progenitors and stimulates bone formation via actions in osteoblasts (39). Moreover, reciprocal changes in marrow adipocyte and osteoblast number may not always accompany PPAR activation. Indeed, we have shown that the ligand 9-hydroxyoctadecadienoic acid stimulates adipogenesis but does not attenuate osteoblast differentiation, whereas 9,10-epoxyoctadecenoic acid suppresses osteoblast differentiation but does not stimulate adipogenesis (13).
The reduction in the ratio of osteoblasts to osteoclasts shown here largely accounts for the reduced trabecular width, wall width, and the extent of bone surface undergoing mineralization in mice receiving rosiglitazone. However, we also noted a decrease in trabecular number and an increase in the spacing between those remaining, indicating that some trabeculae have been removed (40). Osteoclastic perforation of a trabecular plate that was not repaired because of an osteoblast defect could be responsible, but trabeculae could have been lost due to increased bone resorption in the early phase of the experiment, leaving no trace by 28 d of how it had occurred (41). Increased resorption is inconsistent with evidence that activation of PPAR with 15-deoxy-12,14 prostaglandin J2 or thiazolidinediones inhibits rather than stimulates osteoclast differentiation and bone resorption in vitro (42, 43), but is consistent with evidence that, in cultured human aortic smooth muscle cells, thiazolidinediones strongly suppress the synthesis of osteoprotegerin, which antagonizes the pro-osteoclastogenic effects of receptor activator of nuclear factor B ligand (44).
In contrast with our findings, Rzonca et al. (19) reported that rosiglitazone reduced the mineral apposition rate in mice by 3.8-fold. Procedural differences may account for this discrepancy because we measured the entire secondary spongiosa of distal femoral bone and three lumbar vertebrae, whereas Rzonca et al. selected six fields in the proximal tibia—a comparatively small bone that may respond to rosiglitazone in a different fashion than other skeletal sites. Mineral apposition rate mainly reflects the vigor of individual osteoblasts, whereas mineralizing surface is a measure of the number of osteoblasts. The in vitro studies of the present report show that rosiglitazone inhibited the formation of osteoblasts from early progenitors, but had little if any effect on the activity of osteoblasts in the late stages of differentiation. This evidence is consistent with the reduction in mineralizing surface and lack of effect on mineral apposition rate we observed in vivo.
The adverse effect of thiazolidinediones on the rodent skeleton raises the possibility that this class of antidiabetic drugs could cause bone loss. Okazaki et al. (45) reported that, after taking troglitazone for 1 month, type II diabetics exhibited decreased bone remodeling as indicated by a reduction in the bone resorption markers deoxypyridinoline collagen type I C-terminal peptide, as well as the bone formation marker bone specific alkaline phosphatase. However, these markers returned toward baseline after 6 and 12 months of treatment (46). Most studies report normal to increased BMD and reduced fracture incidence in type II diabetics (47, 48, 49), perhaps due to low bone turnover (50). Further studies are needed to investigate whether thiazolidinediones adversely impact the seemingly favorable effects of type II diabetes on the skeleton.
In summary, we have elucidated that, as suggested by earlier in vitro work, rosiglitazone does indeed attenuate osteoblast differentiation and thereby reduces bone formation rate in vivo, leading to bone loss. Strong evidence indicates that activation of PPAR by natural ligands, such as oxidized fatty acids, plays an important role in aging in general (38), and the development of atherosclerosis in particular (51). Furthermore, PPAR activation is now an established negative regulator of bone mass (16, 37), and atherosclerosis and osteoporosis coexist (52). Collectively, these lines of evidence fully justify the need for further work to establish whether increased activation of PPAR secondary to increased level of oxidized fatty acid ligands with age represents a major mechanism for age-related osteoporosis in humans.
Note Added in Proof
While this manuscript was under review, Soroceanu et al. (53) reported that, similar to our findings, administration of rosiglitazone to growing mice for 90 d caused bone loss. Albeit, in contrast to our report, Soroceanu et al. found that the bone loss in their study was associated with a 4-fold increase in the combined prevalence of apoptotic osteoblasts and osteocytes. In our studies, we failed to detect an increase in osteoblast apoptosis (Tables 1 and 2). In addition, when we retrospectively examined vertebral sections for osteocyte apoptosis, we found no change with rosiglitazone administration (vehicle, 3.6 ± 1.3%; rosiglitazone, 4.3% ± 0.6%). The difference between the two sets of observations could be due to differences in the age of the animals or duration of treatment.
Acknowledgments
The authors thank S. Berryhill, T. Chambers, J. Crawford, B. Lecka-Czernik, X. Liu, C. O’Brien, R. Shelton, W. Webb, C. Wiggins III, and R. Wynne for their contributions to this work.
References
Aubin JE, Triffitt JT 2002 Mesenchymal stem cells and osteoblast differentiation. In: Bilezikian JP, Raisz LG, Rodan G, eds. Principles of bone biology. 2nd ed. San Diego: Academic Press; 59–81
Justesen J, Stenderup K, Ebbesen EN, Mosekilde L, Steiniche T, Kassem M 2001 Adipocyte tissue volume in bone marrow is increased with aging and in patients with osteoporosis. Biogerontology 2:165–171
Kajkenova O, Lecka-Czernik B, Gubrij I, Hauser SP, Takahashi K, Parfitt AM, Jilka RL, Manolagas SC, Lipschitz DA 1997 Increased adipogenesis and myelopoiesis in the bone marrow of SAMP6, a murine model of defective osteoblastogenesis and low turnover osteopenia. J Bone Miner Res 12:1772–1779
Meunier P, Aaron J, Edouard C, Vignon G 1971 Osteoporosis and the replacement of cell populations of the marrow by adipose tissue. A quantitative study of 84 iliac bone biopsies. Clin Orth Rel Res 80:147–154
Lips P, Courpron P, Meunier PJ 1978 Mean wall thickness of trabecular bone packets in the human iliac crest: changes with age. Calcif Tissue Res 26:13–17
Ducy P, Starbuck M, Priemel M, Shen J, Pinero G, Geoffroy V, Amling M, Karsenty G 1999 A Cbfa1-dependent genetic pathway controls bone formation beyond embryonic development. Genes Dev 13:1025–1036
Nakashima K, Zhou X, Kunkel G, Zhang Z, Deng JM, Behringer RR, De Crombrugghe B 2002 The novel zinc finger-containing transcription factor osterix is required for osteoblast differentiation and bone formation. Cell 108:17–29
Ren D, Collingwood TN, Rebar EJ, Wolffe AP, Camp HS 2002 PPAR knockdown by engineered transcription factors: exogenous PPAR2 but not PPAR1 reactivates adipogenesis. Genes Dev 16:27–32
Yki-Jarvinen H 2004 Thiazolidinediones. N Engl J Med 351:1106–1118
Forman BM, Tontonoz P, Chen J, Brun RP, Spiegelman BM, Evans RM 1995 15-Deoxy- 12, 14-prostaglandin J2 is a ligand for the adipocyte determination factor PPAR. Cell 83:803–812
Kliewer SA, Sundseth SS, Jones SA, Brown PJ, Wisely GB, Koble CS, Devchand P, Wahli W, Willson TM, Lenhard JM, Lehmann JM 1997 Fatty acids and eicosanoids regulate gene expression through direct interactions with peroxisome proliferator-activated receptors and . Proc Natl Acad Sci USA 94:4318–4323
Lehmann JM, Moore LB, Smith-Oliver TA, Wilkison WO, Willson TM, Kliewer SA 1995 An antidiabetic thiazolidinedione is a high affinity ligand for peroxisome proliferator-activated receptor (PPAR). J Biol Chem 270:12953–12956
Lecka-Czernik B, Moerman EJ, Grant DF, Lehmann JM, Manolagas SC, Jilka RL 2002 Divergent effects of selective peroxisome proliferator-activated receptor-2 ligands on adipocyte versus osteoblast differentiation. Endocrinology 143:2376–2384
Nosjean O, Boutin JA 2002 Natural ligands of PPAR: are prostaglandin J(2) derivatives really playing the part? Cell Signal 14:573–583
Huang JT, Welch JS, Ricote M, Binder CJ, Willson TM, Kelly C, Witztum JL, Funk CD, Conrad D, Glass CK 1999 Interleukin-4-dependent production of PPAR- ligands in macrophages by 12/15-lipoxygenase. Nature 400:378–382
Akune T, Ohba S, Kamekura S, Yamaguchi M, Chung Ui, Kubota N, Terauchi Y, Harada Y, Azuma Y, Nakamura K, Kadowaki T, Kawaguchi H 2004 PPAR insufficiency enhances osteogenesis through osteoblast formation from bone marrow progenitors. J Clin Invest 113:846–855
Lecka-Czernik B, Gubrij I, Moerman EJ, Kajkenova O, Lipschitz DA, Manolagas SC, Jilka RL 1999 Inhibition of Osf2/Cbfa1 expression and terminal osteoblast differentiation by PPAR2. J Cell Biochem 74:357–371
Jennermann C, Triantafillou J, Cowan D, Pennink BGA, Connolly KM, Morris DC 1995 Effects of thiazolidinediones on bone turnover in the rat. J Bone Miner Res 10:S241
Rzonca SO, Suva LJ, Gaddy D, Montague DC, Lecka-Czernik B 2004 Bone is a target for the antidiabetic compound rosiglitazone. Endocrinology 145:401–406
Pickavance LC, Buckingham RE, Wilding JP 2001 Insulin-sensitizing action of rosiglitazone is enhanced by preventing hyperphagia. Diabetes Obes Metab 3:171–180
De VP, Lefebvre AM, Miller SG, Guerre-Millo M, Wong K, Saladin R, Hamann LG, Staels B, Briggs MR, Auwerx J 1996 Thiazolidinediones repress ob gene expression in rodents via activation of peroxisome proliferator-activated receptor . J Clin Invest 98:1004–1009
Cantello BC, Cawthorne MA, Cottam GP, Duff PT, Haigh D, Hindley RM, Lister CA, Smith SA, Thurlby PL 1994 [[-(Heterocyclylamino)alkoxy]benzyl]-2,4-thiazolidinediones as potent antihyperglycemic agents. J Med Chem 37:3977–3985
Jilka RL, Weinstein RS, Takahashi K, Parfitt AM, Manolagas SC 1996 Linkage of decreased bone mass with impaired osteoblastogenesis in a murine model of accelerated senescence. J Clin Invest 97:1732–1740
Bellido T, Ali AA, Plotkin LI, Fu Q, Gubrij I, Roberson PK, Weinstein RS, O’Brien CA, Manolagas SC, Jilka RL 2003 Proteasomal degradation of Runx2 shortens parathyroid hormone-induced anti-apoptotic signaling in osteoblasts: a putative explanation for why intermittent administration is needed for bone anabolism. J Biol Chem 278:50259–50272
Weinstein RS, Jilka RL, Parfitt AM, Manolagas SC 1998 Inhibition of osteoblastogenesis and promotion of apoptosis of osteoblasts and osteocytes by glucocorticoids: potential mechanisms of their deleterious effects on bone. J Clin Invest 102:274–282
Weinstein RS, Jilka RL, Parfitt AM, Manolagas SC 1997 The effects of androgen deficiency on murine bone remodeling and bone mineral density are mediated via cells of the osteoblastic lineage. Endocrinology 138:4013–4021
Parfitt AM, Drezner MK, Glorieux FH, Kanis JA, Malluche H, Meunier PJ, Ott SM, Recker RR 1987 Bone histomorphometry: standardization of nomenclature, symbols, and units. Report of the ASBMR Histomorphometry Nomenclature Committee. J Bone Miner Res 2:595–610
Weinstein RS, Jia D, Powers CC, Stewart SA, Jilka RL, Parfitt AM, Manolagas SC 2004 The skeletal effects of glucocorticoid excess override those of orchidectomy in mice. Endocrinology 145:1980–1987
Fu Q, Jilka RL, Manolagas SC, O’Brien CA 2002 Parathyroid hormone stimulates receptor activator of NFB ligand and inhibits osteoprotegerin expression via protein kinase A activation of cAMP-response element-binding protein. J Biol Chem 277:48868–48875
Di Gregorio GB, Yamamoto M, Ali AA, Abe E, Roberson P, Manolagas SC, Jilka RL 2001 Attenuation of the self-renewal of transit-amplifying osteoblast progenitors in the murine bone marrow by 17?-estradiol. J Clin Invest 107:803–812
Jilka RL 1986 Parathyroid hormone-stimulated development of osteoclasts in cultures of cells from neonatal murine calvaria. Bone 7:29–40
Kuehl RO 2000 Design of experiments: statistical principles of research design and analysis. New York: Duxbury; 492–519
Tai TA, Jennermann C, Brown KK, Oliver BB, MacGinnitie MA, Wilkison WO, Brown HR, Lehmann JM, Kliewer SA, Morris DC, Graves RA 1996 Activation of the nuclear receptor peroxisome proliferator-activated receptor promotes brown adipocyte differentiation. J Biol Chem 271:29909–29914
Stein GS, Lian JB, Stein JL, Van Wijnen AJ, Montecino M 1996 Transcriptional control of osteoblast growth and differentiation. Physiol Rev 76:593–629
Mueller E, Drori S, Aiyer A, Yie J, Sarraf P, Chen H, Hauser S, Rosen ED, Ge K, Roeder RG, Spiegelman BM 2002 Genetic analysis of adipogenesis through peroxisome proliferator-activated receptor isoforms. J Biol Chem 277:41925–41930
Ross SE, Hemati N, Longo KA, Bennett CN, Lucas PC, Erickson RL, MacDougald OA 2000 Inhibition of adipogenesis by Wnt signaling. Science 289:950–953
Klein RF, Allard J, Avnur Z, Nikolcheva T, Rotstein D, Carlos AS, Shea M, Waters RV, Belknap JK, Peltz G, Orwoll ES 2004 Regulation of bone mass in mice by the lipoxygenase gene Alox15. Science 303:229–232
Spiteller G 2000 Peroxidation of linoleic acid and its relation to aging and age dependent diseases. Ageing Res Rev 122:617–657
Kveiborg M, Sabatakos G, Chiusaroli R, Wu M, Philbrick WM, Horne WC, Baron R 2004 FosB Induces osteosclerosis and decreases adipogenesis by two independent cell-autonomous mechanisms. Mol Cell Biol 24:2820–2830
Parfitt AM, Mathews CHE, Villanueva AR, Kleerekoper M, Frame B, Rao DS 1983 Relationships between surface, volume and thickness of iliac trabecular bone in aging and in osteoporosis. Implications for the microanatomic and cellular mechanism of bone loss. J Clin Invest 72:1396–1409
Parfitt AM 1988 Bone remodeling: relationship to the amount and structure of bone and the pathogenesis and prevention of fractures. In: Riggs BL, Melton LJ, eds. Osteoporosis—etiology, diagnosis and management. New York: Raven Press; 45–94
Mbalaviele G, Abu-Amer Y, Meng A, Jaiswal R, Beck S, Pittenger MF, Thiede MA, Marshak DR 2000 Activation of peroxisome proliferator-activated receptor- pathway inhibits osteoclast differentiation. J Biol Chem 275:14388–14393
Okazaki R, Toriumi M, Fukumoto S, Miyamoto M, Fujita T, Tanaka K, Takeuchi Y 1999 Thiazolidinediones inhibit osteoclast-like cell formation and bone resorption in vitro. Endocrinology 140:5060–5065
Fu M, Zhang J, Lin Y, Zhu X, Willson TM, Chen YE 2002 Activation of peroxisome proliferator-activated receptor inhibits osteoprotegerin gene expression in human aortic smooth muscle cells. Biochem Biophys Res Commun 294:597–601
Okazaki R, Miura M, Toriumi M, Taguchi M, Hirota Y, Fukumoto S, Fujita T, Tanaka K, Takeuchi A 1999 Short-term treatment with troglitazone decreases bone turnover in patients with type 2 diabetes mellitus. Endocr J 46:795–801
Watanabe S, Takeuchi Y, Fukumoto S, Fujita H, Nakano T, Fujita T 2003 Decrease in serum leptin by troglitazone is associated with preventing bone loss in type 2 diabetic patients. J Bone Miner Metab 21:166–171
Hanley DA, Brown JP, Tenenhouse A, Olszynski WP, Ioannidis G, Berger C, Prior JC, Pickard L, Murray TM, Anastassiades T, Kirkland S, Joyce C, Joseph L, Papaioannou A, Jackson SA, Poliquin S, Adachi JD 2003 Associations among disease conditions, bone mineral density, and prevalent vertebral deformities in men and women 50 years of age and older: cross-sectional results from the Canadian Multicentre Osteoporosis Study. J Bone Miner Res 18:784–790
Stolk RP, van Daele PL, Pols HA, Burger H, Hofman A, Birkenhager JC, Lamberts SW, Grobbee DE 1996 Hyperinsulinemia and bone mineral density in an elderly population: The Rotterdam Study. Bone [Erratum (1996) 19:566] 18:545–549
Lunt M, Masaryk P, Scheidt-Nave C, Nijs J, Poor G, Pols H, Falch JA, Hammermeister G, Reid DM, Benevolenskaya L, Weber K, Cannata J, O’Neill TW, Felsenberg D, Silman AJ, Reeve J 2001 The effects of lifestyle, dietary dairy intake and diabetes on bone density and vertebral deformity prevalence: the EVOS study. Osteoporos Int 12:688–698
Krakauer JC, McKenna MJ, Buderer NF, Rao DS, Whitehouse FW, Parfitt AM 1995 Bone loss and bone turnover in diabetes. Diabetes 44:775–782
Barish GD, Evans RM 2004 PPARs and LXRs: atherosclerosis goes nuclear. Trends Endocrinol Metab 15:158–165
Schulz E, Arfai K, Liu X, Sayre J, Gilsanz V 2004 Aortic calcification and the risk of osteoporosis and fractures. J Clin Endocrinol Metab 89:4246–4253
Soroceanu MA, Miao D, Bai XY, Su H, Goltzman D, Karaplis AC 2004 Rosiglitazone impacts negatively on bone by promoting osteoblast/osteocyte apoptosis. J Endocrinol 183:203–216(A. Afshan Ali, Robert S. )