Skeletal unloading induced by hindlimb suspension in rats reduces bone formation and induces osteopenia, but its effect on adipogenesis is unknown. We assessed the effects of unloading and transforming growth factor (TGF) β2 on bone marrow stromal cell adipocyte differentiation in relation with osteoblast differentiation. Skeletal unloading rapidly (4-7 days) decreased osteoblast transcription factor Runx2, osteocalcin (OC), and type I collagen messenger RNA (mRNA) levels and reduced bone formation in the long bone metaphysis. Conversely, unloading increased expression of the adipocyte transcription factor peroxisome proliferator-activated receptor γ2 (PPARγ2) at 4 days and increased expression of the adipocyte differentiation genes lipoprotein lipase (LPL) and aP2 in the bone marrow stroma at 7 days. Consistently, unloading increased the number and volume of adipocytes in the bone marrow stroma. Continuous (0-7 days) and late (4-7 days) treatments with TGF-β2 corrected the abnormal expression of Cbfa1/Runx2, OC, and type I collagen mRNAs and normalized bone formation in unloaded metaphyseal bone. Moreover, both TGF-β2 treatments decreased PPARγ2 and C/EBPα mRNA levels at 4 days and normalized aP2 and LPL expression and adipocyte number and volume at 7 days. These results show that skeletal unloading increases adipocyte differentiation concomitantly with inhibition of osteoblast differentiation. These abnormalities are prevented and reversed by TGF-β2, suggesting a role for TGF-β in the control of adipogenic differentiation in the bone marrow stroma.
SKELETAL IMMOBILIZATION is known to result in bone loss in humans and animal models.(1,2) Hindlimb suspension in the rat, a model of skeletal unloading,(3) induces osteopenia by inhibiting bone formation in long bones.(4) These effects result from impaired recruitment of osteoblasts and decreased expression of bone matrix proteins.(5–8) In addition, to alter osteoblastogenesis, skeletal immobilization increases adipogenesis in human bone marrow.(1) Osteoblasts and adipocytes are believed to derive from common mesenchymal precursor cells in the bone marrow stroma,(9–11) and these cells are known to differentiate toward osteoblasts or adipocytes under the control of local and hormonal factors.(12–15) This raises the possibility that in addition to affecting osteoblastogenesis, skeletal unloading may affect adipocyte differentiation in the bone marrow stroma. However, the effects of unloading on bone marrow stromal cell adipogenic differentiation are not known.
Transforming growth factor (TGF) β is a multifunctional factor that regulates a variety of cell types.(16) In bone, TGF-β is an important regulator of endosteal and periosteal bone formation.(17) TGF-β controls bone formation by modulating osteoblastic cell proliferation and differentiation in vitro.(18–20) Moreover, TGF-β increases osteoblast number in vivo.(21–23) However, the effects of TGF-β are complex because it depends on the differentiation stage (osteoprogenitors vs. mature osteoblasts) and the expression of its receptors.(17–19) Recent data indicate that TGF-β signaling plays a role in the control of bone formation induced by unloading.(24) Skeletal unloading in rats is associated with rapid and transient reduction in TGF-β messenger RNA (mRNA) levels,(7) associated with decreased expression of TGF-β receptor II (TGF-βRII).(8) The implication of TGF-β in the decreased bone formation induced by unloading is supported by our finding that exogenous TGF-β2 promotes osteoprogenitor cell proliferation and osteoblast function and prevents the trabecular bone loss in unloaded rats.(25) In addition, to regulate skeletal cells, TGF-β is an important modulator of adipocyte differentiation in vitro. TGF-β inhibits the differentiation of preadipocyte cells in vitro.(26–28) TGF-β also is a potent adipogenic antagonist in bone marrow stromal cells.(29) In adipocytic cells, TGF-β acts by reducing the expression of CCAAT-enhancer binding protein α (C/EBPα) and peroxisome proliferator-activated receptor γ2 (PPARγ2),(27,28) two important transcription factors involved in adipocyte differentiation.(30,31) Moreover, TGF-β reduces adipocyte gene expression independently of cell growth.(32,33) A role for TGF-β in the control of adipogenesis is supported further by the finding that overexpression of TGF-β1 in transgenic mice impairs brown and white adipocyte differentiation.(34) However, the implication of TGF-β in the control of bone marrow stromal cell differentiation into adipocytes in vivo is not known.
We hypothesized that the altered bone formation induced by skeletal unloading is associated with abnormal adipocyte differentiation of marrow stromal cells and that exogenous TGF-β may prevent or correct the alterations of osteoblast and adipocyte differentiation of unloaded bone marrow stroma in vivo. Therefore, we determined the alterations of osteoblastogenesis and adipogenesis induced by skeletal unloading and the effects and mechanisms of action of exogenous TGF-β on these cellular abnormalities.
MATERIALS AND METHODS
Animals and treatment
Forty adult 4-week-old mature Wistar male rats weighing ∼130 g (Janviez, Le Genest, France) were assigned randomly to seven groups (3-5 animals per group). Animals were not suspended (loaded) or suspended by the tail (unloaded) for 4 days or 7 days as previously described(5,8,25) after approval of the study by our local Animal Care Committee. The base of the tail was attached via a clip to the top of a specially designed Plexiglas cage (CERMA-Biomeca, Bretigny, France) to have hindlimbs nonweight bearing. In this model, hindlimb elevation causes minimal transient stress, is well tolerated, and allows normal physical activity by the animals, which had free movement in the cage by using their forelimbs.(5) The rats were maintained on a 12-h light/12-h dark cycle and body weight (BW) was recorded every 2 days. The animals were fed a standard chow containing 1% calcium and 0.8% phosphorus (UAR, Vilemoisson, France).
Figure 1 depicts the experimental protocol used in this study. Three groups of suspended animals were treated with recombinant human TGF-β2 (rhTGF-β2; kindly provided by Novartis Pharma, Basel, Switzerland). The stock solution of TGF-β2 was diluted in saline with 1 mg/ml of bovine serum albumin (BSA) and administered using osmotic minipumps (Alza Corp., Palo Alto, CA, USA) delivering 2 μg/kg BW per day, a dose that we found to be effective in improving bone formation in unloaded rats.(25) Previous studies showed that osmotic minipumps deliver appropriate concentrations of growth factors in unloaded rats and that TGF-β2 infusion at this dose does not induce side effects.(25) To determine the responsiveness of meta-physeal bone cells and bone marrow stroma to TGF-β2 treatment, one group received TGF-β2 from 0 to 4 days and one group was treated from 0 to 7 days of suspension (continuous treatment). Another group was treated with 4-7 days of suspension (late treatment; Fig. 1). Unloaded and loaded rats were sham-operated and were not treated. All animals were given two doses of calcein (10 mg/kg BW) 4 days and 2 days before death, to label the sites of mineralization. At the baseline time point (day 0) and after 4 days and 7 days of suspension, loaded and unloaded animals were anesthetized and killed. The content of the minipumps was measured to ensure appropriate delivery of TGF-β2. The right tibia metaphysis was processed for histomorphometric analysis and the left tibia and femur were used for extraction of total RNA.
The proximal halves of tibias were fixed in 10% phosphate-buffered formaldehyde, dehydrated in methanol, and embedded in methylmethacrylate resin without decalcification.(5) Undecalcified 5-μm-thick longitudinal sections were prepared using a Leica 2055 microtome (Leica, Rueil-Malmaison, France) equipped with a tungsten carbide blade. Sections were stained with Goldner trichrome, and 10- to 15-μm-thick sections were unstained for visualization of calcein labels under fluorescent microscopy. Histomorphometric indices were measured under blind conditions as previously described(5,8,25) using a Leitz Aristoplan microscope (Leica) connected to a Sony DXC-930P color video camera (Sony, Paris, France). An automatic image analyzer (Nachet NS1500; Microcontrole, Evry, France) was used for bone volume measurements. A semiautomatic image analyzer coupled to a digitizer table (Newtec, Nimes, France) was used for the evaluation of bone cells. Cancellous bone parameters were measured in the secondary metaphyseal area of the proximal ends of the tibia in a standardized zone located 400-600 μm from the growth cartilage and 200-300 μm from the cortices. The following indices were measured: the trabecular bone volume (BV/TV; percent of bone tissue composed of calcified and uncalcified matrix), the osteoblast surface (Ob.S/BS; percent of bone surface covered with osteoblasts), and the osteoid surface (OS/BS; percent of bone surface covered with osteoid). The mineral apposition rate (MAR; mean distance between the double fluorescent labels divided by the interval labeling time) and the double-labeled mineralizing surface (MS/BS; percent of the bone surface with double fluorescent markers) also were measured. The bone formation rate (BFR) at the tissue level was calculated by multiplying the MAR by the double fluorescent-labeled surface.(35) In addition, the volume and number of adipocytes were measured in the secondary metaphyseal and diaphyseal areas. The adipocyte volume was expressed as a percentage of the marrow volume and the number of adipocytes was expressed per unit area (mm2) of marrow.
RNA extraction and reverse-transcription-polymerase chain reaction analysis
The left femur and tibia were cleaned of soft tissues and aseptically dissected in sterile phosphate-buffered saline (PBS). After removal of the periosteum and epiphyseal area, the bones were sectioned longitudinally to expose the metaphysis and diaphysis. In each rat, the metaphyseal bone and marrow stroma were collected separately. The meta-physeal areas from one tibia and its associated femur were carefully removed and pooled. The marrow stroma from the tibia and associated femur was collected also by scraping and then pooled, therefore providing one sample of metaphysis and one sample of marrow stroma per rat. The samples were suspended in a monophasic solution of phenol and guanidine isothiocyanate lysis buffer (Extract-All; Eurobio, Les Ulis, France) and frozen at −80°C until RNA isolation. The steady-state expression of marker genes for the osteoblast or adipocyte phenotypes was determined by reverse-transcription-polymerase chain reaction (RT-PCR) analysis and confirmed in all cases by Southern blotting. We opted to use RT-PCR/Southern blotting rather than Northern analysis to avoid the possibility of false negatives, which may arise because of the relative insensitivity of the latter for detecting messages constitutively expressed at very low levels. This assay was validated previously in rat bone.(8) In the metaphysis, we determined the expression of transcripts for the osteoblast markers osteoblast-specific factor 2/core-binding factor a1 (Cbfa1/Runx2), pro-α1 type I collagen (ColIA1), and osteocalcin (OC). In the marrow stroma, we determined the expression of the adipocyte-specific transcription factors PPARγ1, PPARγ2, and C/EBPα; adipocytic differentiation-related genes adipocyte binding protein (aP2), and lipoprotein lipase (LPL).(30) To determine the changes in osteoblast differentiation induced by TGF-β2 of marrow stromal osteoprogenitors, we also determined the expression of alkaline phosphatase (ALP) and Cbfa1/Runx2 mRNA levels in the marrow stroma. Total RNA was extracted by a modified method of Chomczynsky and Sacchi(36) using Extract-All according to the manufacturer's protocol (Eurobio). RNA concentration was determined spectrophotometrically and quality was checked on a 1% agarose gel containing ethidium bromide. In preliminary experiments, optimization of RT-PCR was performed, the optimal cycle number within the linear range of the amplification reactions (23-30 cycles) was determined, and the experiments were conducted in these appropriate conditions.(8) The synthesis of complementary DNA (cDNA) from 10 μg of the isolated RNA was carried out for 40 minutes at 37°C with 200 U of Moloney murine leukemia virus (MMLV) reverse transcriptase (Gibco BRL Life Technologies, Eragny, France) in the presence of 100 ng of oligo(dT)15 primer (Promega, Charbonnière, France), 10 mM of dithiothreitol (DTT), 1 mM of deoxynucleoside triphosphates (dNTPs), and 40 U of RNAse inhibitor (Promega) in a final volume of 20 μl. Five microliters of this reverse transcribed cDNA was submitted to PCR in 50 μl of reaction mixture containing 1.25 U of Taq DNA polymerase (Extrapol; Eurobio), 2 mM of MgCl2, and 20 pmol of sense and antisense primers specific for the genes of interest and the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Primer sequences for the osteoblast markers Cbfa1/Runx2, ColIA1, OC, ALP, and GAPDH were reported previously.(8
(antisense); for LPL, 5′-ACTGCCACTTCAACCACAGC-3′ (sense) and 5′-AATACTTCGACCAGGCGACC-3′ (antisense); and for aP2, 5′-CGTCTCCAGTGAGAACTTCG-3′ (sense) and 5′-TCATGACACATTCCACCACC-3′ (antisense); PCR was performed after a denaturing step of 5 minutes at 95°C. Each cycle was composed of a melting step at 94°C for 1 minute, annealing at 60°C for 1 minute, and elongation at 72°C for 1 minute in a Perkin Elmer Cetus Thermal Cycler (Perkin Elmer, Cetus, Les Ulis, France). The final step was extended to 10 minutes at 72°C. For each transcript, an amplification curve was constructed based on ascending cycle number, and an optimal cycle number of 30 cycles, representing the ascending portion of the amplification curve, was chosen. Negative control reactions for RT-PCR were performed in each assay using all reagents except RNA. Southern blots were performed by running aliquots of amplified cDNAs on 1.5% agarose gel, visualized by ethidium bromide and transferred onto nylon membranes (Appligene-Oncor, Illkirch, France) by the alkaline transfer method. The membranes were hybridized overnight with antisense oligonucleotide probe that was end-labeled with (γ32P)ATP. Then, membranes were washed twice in 2× SSC/0.1% sodium dodecyl sulfate (SDS) at room temperature for 10 minutes and then in 0.1× SSC/0.1% SDS at 50°C for 10 minutes. Filters were exposed to X-ray films with intensifying screen at −80°C. Autoradiographic bands were quantified by densitometric analysis using a scanning densitometer.(8) The signal for each gene was related to that for GAPDH. All RT-PCR analyses were carried out using cDNA obtained in individual rats and in 3-5 rats per group. Each analysis was repeated three to five times to account for the possible variations that may occur from rat to rat or from different assays.
All data are expressed as the mean ± SE. The data were analyzed by two-factor analysis of variance (ANOVA) using the statistical package super-ANOVA (Macintosh, Abacus Concepts, Inc., Berkeley, CA, USA). A minimal level of p < 0.05 was considered significant.
Exogenous TGF-β2 corrects the defective osteoblast gene expression and bone formation in unloaded rats
We first determined the biological activity of TGF-β2 on osteoblastogenesis in unloaded rats. The histomorphometric analysis showed that the Ob.S and OS in the tibia metaphy-sis were decreased in unloaded rats at 4 days and 7 days compared with loaded rats (Figs. 2A and 2B). Continuous treatment with TGF-β2 for 0-7 days restored normal OS and Ob.S at 7 days of unloading. The late treatment with TGF-β2 for 4-7 days also corrected these abnormalities (Figs. 2A and 2B). As expected, skeletal unloading decreased BFR in the tibia metaphysis at 4 days and 7 days of unloading. Continuous treatment with TGF-β2 corrected BFR after 7 days of treatment, and the same effect was obtained with the late treatment given for 4-7 days (Fig. 2C). In unloaded rats, the decreased bone formation at 4 days resulted in decreased BV/TV at 7 days. TGF-β2 had beneficial effects on BV/TV in the tibia metaphysis. Both the continuous and late treatments with TGF-β2 decreased the bone loss induced by unloading at 7 days (Fig. 2D). This shows that exogenous TGF-β2 rapidly corrected the defective osteoblastogenesis and bone formation and reduced the metaphyseal bone loss induced by unloading.
To determine the kinetics of action of TGF-β2 on bone formation, we analyzed the effect of exogenous TGF-β2 on osteoblast gene expression in tibia and femur metaphyses. The expression of osteoblast genes was determined by RT-PCR analysis in untreated and treated rats (Fig. 3A). In untreated unloaded bone, the expression of Cbfa1/Runx2 mRNA levels was decreased markedly at 7 days of suspension compared with control rats (Figs. 3A and 3B). Continuous treatment with TGF-β2 for 7 days prevented the decrease in Cbfa1/Runx2 mRNA levels induced by unloading. The late treatment with TGF-β2 for 4-7 days increased Cbfa1/Runx2 mRNA levels above values in the control rats (Figs. 3A and 3B). Consistent with the alteration of Cbfa1/Runx2 mRNA levels in long bone metaphyseal osteoblasts, OC mRNA levels were decreased at 7 days of suspension compared with control rats (Figs. 3A and 3C). The continuous treatment with TGF-β2 for 7 days or the late treatment for 4-7 days completely corrected the abnormal OC expression in unloaded metaphyseal bone (Figs. 3A and 3C). ColIA1 mRNA level was reduced at 7 days in unloaded metaphyseal bone compared with control rats (Figs. 3A and 3D). Continuous treatment with TGF-β2 corrected ColIA1 expression at 7 days of treatment (Figs. 3A and 3D). The late treatment with TGF-β2 for the last 4 days of suspension increased ColIA1 mRNA levels above values in control rats (Figs. 3A and 3D). These results show that exogenous TGF-β2 restored the expression of osteoblast marker genes in unloaded metaphyseal bone, leading to correct bone formation despite unloading. In an attempt to determine whether TGF-β may act directly on the common precursor cell to promote its commitment in the osteoblastic lineage, we also looked for the expression of ALP and Cbfa1/Runx2 expression in the bone marrow under treatment with TGF-β2. However, Cbfa1/Runx2 expression and ALP expression could not be detected consistently in the bone marrow stroma of unloaded rats (data not shown).
TGF-β2 prevents the increased bone marrow adipogenesis induced by unloading
We then determined the effect of unloading on adipogenesis in the marrow stroma of long bones. Histological examination revealed that adipogenesis was increased in the bone marrow area of unloaded rats (Fig. 4B) compared with loaded rats (Fig. 4A) at 7 days of suspension. Treatment with TGF-β2 prevented the increased adipogenesis in the bone marrow stroma induced by unloading (Fig. 4C). To confirm this finding, the number and size of adipocytes were determined by histomorphometrical analysis. Adipocyte number and volume were not significantly altered at 4 days of suspension (data not shown). In contrast, unloading increased the number of adipocytes 2-fold compared with control rats at 7 days of suspension (Fig. 4D). The volume of adipocytes also was increased markedly in unloaded rats at 7 days compared with control rats, reflecting increased adipogenesis (Fig. 4E). Remarkably, these abnormalities were prevented by exogenous TGF-β2. Indeed, TGF-β2 administered for 7 days prevented both the increased adipocyte number and the volume induced by unloading (Figs. 4D and 4E). The late treatment with TGF-β2 administered from 4 to 7 days also markedly reduced adipocyte number (Fig. 4D) and volume (Fig. 4E). These data show that exogenous TGF-β can prevent and partially reverse the increased long bone marrow adipogenesis induced by unloading.
TGF-β2 reduces adipocyte master gene expression in the marrow stroma
We then determined the mechanisms of action of TGF-β2 on adipogenesis induced by unloading in the long bone marrow stroma. We first evaluated the changes in PPARγ2, a master gene that is known to play a critical role in adipogenic differentiation.(30,31) As shown in Fig. 5 A, PPARγ2 mRNA level in the marrow stroma was increased in unloaded rats compared with loaded rats. The densitometric analysis showed that PPARγ2 mRNA levels were increased at both 4 days and 7 days of suspension (Fig. 5B). The changes observed using primers for PPARγ1 and PPARγ2 did not differ significantly (data not shown). Treatment with TGF-β2 for 4 days reduced PPARγ2 mRNA levels in the marrow stroma (Fig. 5B). This early effect of TGF-β2 also was found at 7 days of treatment. In contrast, the late treatment with TGF-β2 from 4 to 7 days of suspension did not reduce PPARγ2 mRNA levels in unloaded rats (Fig. 5B), indicating that TGF-β2 acted early on PPARγ2 expression in the marrow stroma of unloaded long bones.
C/EBPα is another transcription factor that is involved in late adipogenic differentiation.(30,31) We found no difference in C/EBPα mRNA levels in the marrow stroma in unloaded and loaded rats at 4 days or 7 days of suspension (Figs. 5A and 5C). However, continuous treatment with TGF-β2 for 4 days significantly decreased C/EBPα mRNA levels in unloaded marrow stroma, and this inhibitory effect persisted at 7 days (Figs. 5A and 5C). In contrast, the late treatment for the last 4-7 days had no effect on C/EBPα mRNA levels in unloaded marrow stroma (Figs. 5A and 5C), indicating that TGF-β2 acted at an early stage to reduce the expression of this transcription factor in unloaded marrow stroma, as found for PPARγ2 expression.
TGF-β2 prevents and reverses the increased adipocyte gene expression in unloaded bone marrow
We then assessed the changes in the expression of aP2 and LPL that are involved in adipocyte differentiation.(30,31) Neither aP2 nor LPL mRNA levels in the marrow stroma were altered by unloading at 4 days of suspension (not shown), in keeping with the late expression of these genes during adipocyte differentiation.(30,31) In contrast, mRNA levels for aP2, a late marker of adipogenesis, were increased 2- to 3-fold in unloaded rat bone marrow stroma at 7 days of suspension compared with loaded rats (Figs. 6A and 6B), confirming the increased adipogenesis. Continuous treatment with TGF-β2 for 7 days markedly reduced aP2 mRNA expression in the marrow stroma. The late treatment with TGF-β2 for 4-7 days also reduced aP2 mRNA levels in the marrow stroma (Figs. 6A and 6B). We also found that LPL mRNA levels were increased markedly in unloaded marrow stroma compared with loaded rats at 7 days (Figs. 6A and 6C). Continuous treatment with TGF-β2 for 7 days decreased LPL mRNA expression to levels not different from those in loaded rats (Figs. 6A and 6C). The late treatment with TGF-β2 for 4-7 days induced a similar inhibitory effect, and this treatment corrected the abnormal expression of LPL mRNA levels induced by unloading (Figs. 6A and 6C). These results show that skeletal unloading drastically increases the expression of adipocyte differentiation genes in the bone marrow stroma and this effect can be prevented by exogenous TGF-β2 and corrected by a late treatment with TGF-β2.
This study shows that skeletal unloading induces increased adipocyte differentiation and adipogenesis in the bone marrow stroma concomitantly to the reduction in osteoblastogenesis. Our data also show that exogenous TGF-β2 prevents and reverses both the alterations of osteoblastogenesis and adipogenesis induced by skeletal unloading, which suggests a role for TGF-β in the control of adipogenic differentiation induced by unloading.
In line with the reduction in bone formation, skeletal unloading induced marked changes in osteoblast gene expression in the metaphysis. We showed that skeletal unloading reduced the expression of Cbfa1/Runx2, a transcriptional activator of osteoblast differentiation,(37,38) and this was associated with decreased expression of OC and ColIA1, genes known to be controlled in part by Cbfa1/Runx2 in postnatal organisms.(39) Thus, the loss of Cbfa1/Runx2 expression may contribute, with other transcriptional activator factors, to the decreased expression of OC and ColIA1 in unloaded osteoblasts. Concomitantly to the reduced bone formation in the metaphysis, unloading increased the expression of adipogenic differentiation genes in the bone marrow stroma. Notably, PPARγ2, a critical transcription factor involved in adipogenic differentiation,(40) was increased markedly in unloaded bone marrow stroma. Moreover, the rise in PPARγ2 preceded the increased expression of aP2 and LPL that are regulated by PPARγ2 in vitro.(40–42) Although C/EBPα, a factor that interacts with PPARγ2,(42–46) was expressed in the bone marrow stroma, its expression was not increased by skeletal unloading. This is consistent with the lack of change in C/EBPα during adipocyte differentiation of bone marrow stromal cells in vitro.(29,42) We were unable to detect mRNA for C/EBPβ, a transcription factor that is believed to play a role in adipogenesis.(47) However, we cannot rule that this gene may be expressed transiently during bone marrow adipocyte differentiation.(30) PPARδ is another transcription factor that may play a role in the differentiation program of adipocytic cells by promoting PPARγ expression.(48) Although we found significant amounts of PPARδ mRNA in marrow stroma, the levels were not changed by skeletal unloading (data not shown). Thus, the initial increase in adipocyte differentiation induced by unloading in the bone marrow stroma appears to involve mainly PPARγ2, an essential transcription factor involved in adipogenesis.(40,41) Because assessing mRNA expression levels to investigate adipogenesis in vivo presents some limitations, we confirmed the changes in adipocyte gene expression by histomorphometric analysis. Both the number and the volume of adipocytes were increased markedly in unloaded bone marrow stroma, confirming the increased adipocyte gene expression induced by unloading. Our finding that skeletal unloading induced concomitant and reciprocal alterations of the osteoblast and adipocyte differentiation programs supports the previous observations that osteoblasts and adipocytes show some plasticity in culture(10–15,49–56) as well as in vivo.(57–61)
Because TGF-β is an important regulator of both osteoblast and adipocyte differentiation in vitro, we attempted to prevent and rescue the alterations of the osteoblast and adipocyte differentiation programs induced by unloading using exogenous TGF-β2. As we previously reported,(25) exogenous TGF-β2 corrected the defective bone formation in unloaded rat bone. We show here that this beneficial effect of TGF-β2 was associated with increased expression of Cbfa1/Runx2, OC, and ColIA1 mRNA levels in unloaded rats, indicating that TGF-β2 acted by increasing the expression of these target genes in metaphyseal osteoblasts. However, correction of osteoblastic gene expression also may occur as the result of posttranscriptional effects of TGF-β2.(17) Concomitant with these effects of TGF-β2 in metaphyseal osteoblasts, we found that TGF-β2 corrected adipogenesis in the bone marrow stroma. This was evidenced by the reduced expression of PPARγ2 and C/EBPα in marrow stromal cells, which is consistent with the repressive effect of TGF-β on these genes in cultured adipocytes.(32,33) Because the activity of PPARγ may be regulated by post-translational modifications involving mitogen-activated protein kinase (MAPK)-mediated phosphorylation,(62,63) it is possible that, in addition to reducing PPARγ2 expression, TGF-β2 may modulate PPARγ activity in marrow stromal cells. Our finding that the continuous treatment with TGF-β2 rapidly reduced PPARγ2 expression whereas the late treatment had no effect suggests that TGF-β2 acted mainly on this transcription factor at early stages of adipocyte differentiation enhanced by unloading. Thus, the effects of TGF-β on adipogenesis appears to depend on the stage of adipogenic differentiation in the marrow stroma, as found in murine cell lines.(26–28,32,33)
The early decrease in PPARγ2 expression in response to TGF-β2 was followed by down-regulation of the adipocyte-specific genes aP2 and LPL, confirming that TGF-β2 inhibited the adipocyte differentiation program in the bone marrow stroma. Interestingly, treatment with TGF-β2 for 4 days was sufficient to correct PPARγ2, aP2, and LPL overexpression at 7 days, showing the rapid adipocyte suppressing activity of TGF-β2. This is consistent with the rapid antiadipogenic effects of TGF-β2 on bone marrow stromal cells in vitro.(29) This effect is not specific to TGF-β2 because both TGF-β1 and TGF-β2 are equally able to reduce adipogenic differentiation independently of cell growth.(32) Consistent with the reduction in mRNA levels for adipocyte genes, we found that TGF-β2 reduced the number and volume of adipocytes in unloaded bone, further showing that TGF-β2 abolished the adipocyte differentiation program induced by skeletal unloading in rats. Whether TGF-β also may have antiproliferative and apoptotic effects on marrow adipocytes remains to be determined. The present finding that TGF-β concomitantly corrected osteoblastogenesis and adipogenesis suggests that TGF-β may act directly on the common precursor cell to promote its commitment in the osteoblastic lineage at the expense of the adipocyte lineage. Further in vitro studies, now in progress, are required to clarify this point.
We hypothesize that the reciprocal alterations of the osteoblast and adipocyte differentiation programs induced by unloading may result from a local defect in TGF-β signaling. This hypothesis is supported by several points. First, TGF-β mRNA levels in the bone marrow are decreased rapidly during hindlimb suspension.(7) The source of TGF-β in long bone marrow is unknown but may include preadipocytes.(64) Second, the expression of TGF-βRII is decreased rapidly by unloading at 4 days in both the metaphysis(8) and the bone marrow stroma (data not shown) compared with normal loaded bones, which may contribute in impairing the antiadipogenic effect of endogenous TGF-β in the bone marrow stroma, as found in cultured preadipocytes.(64) Thus, we suggest that the reduced expression of endogenous TGF-β expression(7) and signaling(8) in unloaded bone results in up-regulation of PPARγ2 expression and adipocyte differentiation in the bone marrow stroma associated with down-regulation of the osteoblast differentiation program in the metaphysis. This provides a mechanism involving TGF-β in the altered program of differentiation toward adipocytes and osteoblasts in unloaded bone and suggests potential therapeutic value of exogenous TGF-β not only to improve bone formation but also to reverse the associated alteration of adipocyte differentiation induced by skeletal unloading.
The authors thank S. Renault and C. André (IMASSA-CERMA, Department de Physiologie Aérospatiale, Brétigny sur Orge) for their technical assistance, and Novartis Pharma (Basel, Switzerland) for the generous gift of rhTGF-β2. This work was supported in part by grants from the Centre National d'Etudes Spatiales (CNES, Paris, France).