|Home | About | Journals | Submit | Contact Us | Français|
Smad3 deficiency accelerates chondrocyte maturation and leads to osteoarthritis. Primary chondrocytes without Smad3 lack compensatory increases of TGF-β signaling factors, but BMP-related gene expression is increased. Smad2 or Smad3 overexpression and BMP blockade abrogate accelerated maturation in Smad3−/− chondrocytes. BMP signaling is increased in TGF-β deficiency and is required for accelerated chondrocyte maturation.
Disruption of TGF-β signaling results in accelerated chondrocyte maturation and leads to postnatal dwarfism and premature osteoarthritis. The mechanisms involved in this process were studied using in vitro murine chondrocyte cultures.
Primary chondrocytes were isolated from the sterna of neonatal wildtype and Smad3−/− mice. Expressions of maturational markers, as well as genes involved in TGF-β and BMP signaling were examined. Chondrocytes were treated with TGF-β and BMP-2, and effects on maturation-related genes and BMP/TGF-β responsive reporters were examined. Recombinant noggin or retroviral vectors expressing Smad2 or Smad3 were added to the cultures.
Expression of colX and other maturational markers was markedly increased in Smad3−/− chondrocytes. Smad3−/− chondrocytes lacked compensatory increases in Smad2, Smad4, TGFRII, Sno, or Smurf2 and had reduced expression of TGF-β 1 and TGFRI. In contrast, Smad1, Smad5, BMP2, and BMP6 expression was increased, suggesting a shift from TGF-β toward BMP signaling. In Smad3−/− chondrocytes, alternative TGF-β signaling pathways remained responsive, as shown by luciferase assays. These non–Smad3-dependent TGF-β pathways reduced colX expression and alkaline phosphatase activity in TGF-β–treated Smad3−/− cultures, but only partially. In contrast, Smad3−/− chondrocytes were more responsive to BMP-2 treatment and had increased colX expression, phosphoSmads 1, 5, and 8 levels, and luciferase reporter activity. Overexpression of both Smad2 and Smad3 blocked spontaneous maturation in Smad3-deficient chondrocytes. Maturation was also abrogated by the addition of noggin, an extracellular BMP inhibitor.
These findings show a key role for BMP signaling during the chondrocyte maturation, occurring with loss of TGF-β signaling with important implications for osteoarthritis and cartilage diseases.
During embryonic limb development, mesenchymal stem cells undergo chondrogenesis, and a cartilaginous template preforms the skeletal elements. In the center of the cartilage template, chondrocytes hypertrophy, matrix calcifies, and vascular invasion occurs.(1) Terminally differentiated chondrocytes undergo apoptosis, and the calcified cartilage matrix acts as a scaffold for primary bone formation. As skeletal growth progresses, two growth plates migrate to opposite ends of the long bones and remain active during postnatal and adolescent growth. The growth plate is surrounded by a perichondrium that contains mesenchymal precursor cells. The perichondrium is an important signaling center during embryonic development but is likely less important during adolescent growth.
The growth plate is a highly organized structure; chondrocytes form columns of proliferating cells that progress through maturation, terminal differentiation, and apoptosis. Chondrocyte maturation is characterized by a 5- to 10-fold increase in cell volume and expression of specific genes including colX, MMP13, and alkaline phosphatase.(2–4) Chondrocytes undergo terminal differentiation, and the matrix calcifies in the lower hypertrophic zone. Chondrocytes within the lower hypertrophic region express osteocalcin and VEGF before undergoing apoptosis.(5–7) Vascular ingrowth and primary bone formation occurs in the vacant columns previously occupied by hypertrophic chondrocytes. Important interactions occur with the perichondrium where there is a transition from endochondral to intramembranous ossification and the generation of signals that influence endochondral ossification.(8,9) The events of chondrocyte proliferation, maturation and hypertrophy, and terminal differentiation are recapitulated during adult endochondral fracture healing.(10)
This process of endochondral ossification is highly regulated; chondrocytes in the developing limb bud and adolescent growth plate are highly responsive to growth factors and signaling molecules.(11) A particularly important regulatory event in endochondral ossification involves the transition from proliferation to hypertrophy. Models of embryonic limb development have shown a critical role for a signaling pathway involving PTH-related peptide (PTHrP) and Indian hedgehog (Ihh), which act in a negative manner, reducing the rate of chondrocyte maturation.(11–13) PTHrP completely arrests chondrocyte maturation and prevents expression of Ihh, which is present in a narrow zone of prehypertrophic chondrocytes.(12,13) PTHrP knockout mice and mice defective in PTHrP signaling show accelerated chondrocyte maturation and disorganization of the growth plate and develop severe skeletal malformations.(12,14) In contrast, chondrocytes in mice with overexpression of PTHrP have delayed chondrocyte maturation and also exhibit widespread skeletal malformations.(15)
Similar to PTHrP, TGF-β inhibits expression of colX, alkaline phosphatase, MMP13, and other genes associated with chondrocyte maturation in numerous cell culture models, including embryonic chick sternal chondrocytes, rat neonatal limb chondrocytes, rat periosteal cells, and rabbit growth plate chondrocytes.(16–20) However, whereas PTHrP null mice have developmental abnormalities at birth, mice deficient in TGF-β signaling develop normal skeletons and have normal growth until 4 weeks of age.(21,22) At that point, both Smad3−/− mice and mice over-expressing dominant negative transforming growth factor type II receptor (TGFRII) develop disorganized growth plates, with premature chondrocyte maturation and subsequent runting.(21,22) Furthermore, both Smad3−/− mice and mice overexpressing dominant negative TGF-β express markers of maturation in articular chondrocytes and develop premature osteoarthritis.(21,22) Thus, in contrast to Ihh/PTHrP, TGF-β is primarily involved in postnatal regulation of chondrocyte maturation and has critical involvement in the maintenance of articular cartilage and in the pathogenesis of osteoarthritis.
The classic TGF-β–mediated signaling pathway involves Smad activation. Smads are a family of intracellular proteins that comprise three classes of signaling molecules: receptor-associated Smads (2 and 3 for TGF-β; 1, 5, and 8 for BMP signaling), the co-factor Smad4, and the inhibitory Smads (6 and 7).(1,23–25) The receptor-associated Smads bind to the type I receptor, and on ligand binding and activation, are phosphorylated and released into the cytoplasm. The activated receptor-associated Smads form a trimeric heterodimer with the co-factor Smad4, translocate to the nucleus, and influence gene transcription.(26) Prior work by our laboratory and others has established that the TGF-β Smads inhibit chondrocyte maturation, whereas the BMP-related Smads accelerate maturation.(27–29) Thus, basal BMP signaling is likely one of the pathways that commits chondrocytes to complete maturation.
An alternative and parallel TGF-β signaling pathway is mediated by a protein called TGF-β–activating kinase (TAK), which targets the MAP kinase, p38, and leads to phosphorylation and activation of the transcription factor activating transcription factor-2 (ATF-2).(30–34) Whereas ATF-2 has been shown to associate with c-jun and bind to the cAMP response element (CRE), it has also been found to interact with Smad3/4 hetero-oligomers through the MH1 region and the basic leucine zipper region of ATF-2.(35,36) Thus, ATF-2 plays a central role in TGF-β signaling by acting as a common nuclear target of both Smad and TAK1 pathways. In addition, in some cells and under certain conditions, TGF-β has been shown to stimulate protein kinase C (PKC), as well as c-Jun N-terminal kinase (JNK) and extracellular signal–related protein kinase (ERK) kinase signaling.(23)
In this study, we examined the intracellular signaling mechanisms underlying the accelerated chondrocyte maturation that occurs in Smad3−/− mice. For these experiments, we developed a novel model of chondrocyte maturation in primary mouse sternal chondrocytes and defined the patterns of differentiation in wildtype and Smad3−/− chondrocytes. Our findings show that loss of Smad3 results in an upregulation in BMP signaling molecules and a corresponding decrease in TGF-β signaling that results in accelerated chondrocyte differentiation. This acceleration is inhibited by the addition of the extracellular BMP antagonist noggin, suggesting a key role for BMP signaling. Thus, whereas Smad3−/− chondrocytes remain responsive to the inhibitory effects of TGF-β, the net effect of this loss is an enhanced rate of chondrocyte maturation because of upregulated BMP signaling.
All studies were performed under the auspices of the University Committee on Animal Resources. Smad3−/−mice derived from a C57/B6 lineage, in which exon 8 of the Smad3 gene is deleted (a kind gift from Dr CX Deng, NIH, Bethesda, MD, USA), were bred using heterozygote pairs.(22,37) Three-day-old neonatal mice were killed, and genotyping was performed on tail snips obtained at the time of death. The anterior rib cage and sternum were harvested en bloc, washed with sterile PBS, and digested with pronase (Roche Laboratory, Nutley, NJ, USA) dissolved in PBS (2 mg/ml) in a 37°C water bath with continuous shaking for 60 minutes. This was followed by incubation in collagenase D (Roche) solution (3 mg/ml dissolved in serum-free DMEM) for 90 minutes at 37°C. The soft tissue debris was carefully removed. The remaining sterna and costosternal junctions were further digested in fresh collagenase D solution in petri dishes in a 37°C cell incubator for 4–6 h with intermittent shaking. This step allows remnant fibroblasts to attach to the petri dish while the chondrocytes remain afloat in the medium. The digestion solution was filtered through swinex to remove residual tissue fragments. The solution was centrifuged, and the cells were resuspended in complete medium (DMEM with 10% FBS, 1% penicillin/streptomycin, 100 mM L-glutamine, and 50 μg/ml ascorbic acid, pH 7.1). Chondrocytes were counted and plated at appropriate concentrations. To remove any remaining fibroblasts, cells were treated with 0.05% trypsin for 1 minute after plating for 24 h, which lifts the chondrocytes from the culture dish while allowing the fibroblasts to remain attached. Where indicated, recombinant murine noggin (R&D Systems, Minneapolis, MN, USA) was added to the cultures at a concentration of 500 ng/ml.
After 24 h of culture, the cells were lysed in Golden lysis buffer supplemented with protease inhibitor (Boehringer Mannheim, Indianapolis, IN, USA), 1 mM sodium orthovanadate, 1 mM ethyleneglycol-bis-(β-amino ethyl ether), 1 mM sodium fluoride, and 1 μM microcysteine (Sigma Chemical, St Louis, MO, USA). The protein concentration was determined using the Coomassie Plus Protein Assay kit (Pierce, Rockford, IL, USA). SDS-PAGE was used to separate the protein extract (30 μg). After transfer to a polyvinylidene fluoride (PVDF) membrane (NEN Life Science Product, Boston, MA, USA), and blocking with 5% milk, the blots were probed with goat anti-Smad1 (1 μg/ml; Santa Cruz Biotechnology, Santa Cruz, CA, USA) and rabbit anti-Smad2 (1 μg/ml; Zymed, San Francisco, CA, USA), anti-Smad3 (1 μg/ml; Zymed), anti-Smad5 (1 μg/ml; Zymed), anti-Smad6 (3 μg/ml; Zymed), and anti–phosphoSmad 1/5/8 (1 μg/ml; pSmad1[Ser 463/465]/pSmad5[Ser 463/465]/pSmad8[Ser 426/428]; Cell Signaling, Beverly, MA, USA) antibodies. A monoclonal anti-β-actin antibody (Sigma) at a dilution of 1:8000 was used to confirm equal protein loading. After washing, the membrane was incubated with appropriate alkaline phosphatase–conjugated secondary antibody (BioRrad Laboratories, Hercules, CA, USA) for 2 h at room temperature. The immune complexes were detected using one-step NBT/BCIP substrate (Pierce, Rockford, IL, USA). The blot was repeated using two independent cell preparations.
Twelve hours after plating in 12-well plates, the following promoter-reporter constructs (500 ng/well) were transiently transfected using Superfect reagent (Qiagen, Valencia, CA, USA) per the manufacturer’s protocol: P3TP-Luciferase (Luc; TGF-β–responsive element), 4× SBE-Luc (Smad3-specific reporter), and 9× GCG-Luc (BMP-specific reporter). SV-40 renilla luciferase construct (10 ng/well) was co-transfected with the above firefly reporters to standardize results for transfection efficiency. Luciferase activity in cell lysate was determined in a luminometer (Opticom 1; MGM Instruments, Hamden, CT, USA). Each cell preparation was tested in quadruplicate; the experiment was repeated using three independent cell preparations.
After 48-h culture, chondrocytes were rinsed with 150 mM NaCl and lysed with either passive lysis buffer (Promega, Madison, WI, USA) or M-PER (Pierce) at room temperature for 15–30 minutes. The lysates were centrifuged at full speed at 4°C for 20 minutes, and the supernatant was collected without disturbing the pellet. Supernatant (100 μl) was added into 12-well plates containing 1 ml of reaction buffer (0.25 M 2 methyl-2-amino propanol, 1 mM MgCl2, and 2.5 mg/ml of p-nitrophenyl phosphate; pH 10.3) and incubated at 37°C until color change occurred; the time required to elicit this change was noted. The reaction was stopped by the addition of 0.5 ml of 0.3 M Na3PO4 (pH =12.3). Alkaline phosphatase activity was determined spectrophotometrically at 410 nm by comparison with standard solutions of p-nitrophenol and an appropriate blank. Alkaline phosphatase activity was normalized to protein concentrations and incubation time. Protein concentrations were determined by adding 20 μl of cell lysate to 1 ml dilute BioRad protein assay reagent (1:5; BioRad) and measuring absorbance at 595 nm, which was compared with a standard solution of protein and an appropriate blank. Each cell preparation was tested in triplicate; experiments were repeated on three independent cell preparations.
Total RNA was extracted from cultures using either RNAeasy kit (Qiagen) or TRIzol (Immunogen) per the manufacturer’s protocol. RNA (1 μg) was reversed transcribed using Advantage RT-for-PCR kit (BD Bioscience Clontech, Palo Alto, CA, USA) or iScript (BioRad) according to the manufacturer’s protocol. RT-PCR was performed using the RotorGene real-time DNA amplification system (Corbett Research) and the fluorescent dye SYBR Green I to monitor DNA synthesis (SYBR Green PCR Master Mix; Applied Biosystems). The primers used in this study are shown in Table 1. The PCR protocol included a 95°C denaturation step for 10 minutes followed by 40 cycles of 95°C denaturation (20 s), annealing (20 s), and 72°C extension (30 s). Detection of the fluorescent product was carried out at the end of the 72°C extension period. PCR products were subjected to a melting curve analysis, and the data were analyzed and quantified with the RotorGene analysis software. Dynamic tube normalization and noise slope correction was used to remove background fluorescence. Each sample was tested at least in triplicate and repeated for three independent cell preparations.
Total RNA (10 μg/sample) was separated on denaturing formaldehyde/agarose gels and transferred to GENE Screen nylon membranes (DuPont, Wilmington, DE, USA) by capillary transfer. The blotted RNA was cross-linked by UV and hybridized with [32P]labeled probes specific for murine type II collagen and type X collagen (generated by PCR reaction using the primers listed in Table 1). The probe was hybridized at 68°C for 1 h in QuickHyb hybridization solution (Stratagene, La Jolla, CA, USA). After washes, Kodak BioMax film (Eastman Kodak, Rochester, NY, USA) was exposed to the radiolabeled membranes at −80° for 24–48 h. This was repeated in duplicate using independent cell preparations.
Retroviral plasmids LPCX-F-Smad2 and LPCX-F-Smad3 (a kind gift from Dr Rik Derynck) were transfected into 293P cells using Superfect transfection reagent (Qiagen). A retrovirus expressing green fluorescent protein (GFP) was used as a control. The supernatant of the cell culture was collected every 24 h for 5 days, beginning 24 h after transfection. Viral particles were obtained by centrifuging the supernatant at 20,000 rpm for 2 h. The pelleted viral particles were resuspended in 1 ml of DMEM. Concentrated virus was stored at −80°C.
Sternal chondrocytes from either wildtype (WT) or Smad3−/− (KO) mice were plated in 60-mm dishes for 24 h. The medium was removed and replaced with 1 ml of concentrated virus solution from above, expressing either Smad2, Smad3, or GFP (MOI =5). After 48 h, 1 ml of complete media was added to each dish together with hexadimethrine bromide (polybrene, 4 mg/ml; Sigma) and incubated for another 48 h. The infection rate was examined using a fluorescence microscope; GFP expression was found in >50% of the chondrocytes. ColX mRNA expression, Western blot, and alkaline phosphatase activity were determined as previously described.
Statistical comparisons were made between the groups using either ANOVA or Student’s t-test as appropriate. p values of <0.05 were considered significant and are denoted in each of the figures.
Sternal chondrocytes were separately harvested from the sterna of 3-day-old mouse littermates with either a Smad3+/+ (wildtype) or Smad3−/− (knockout) genotype. Total RNA was obtained from monolayer chondrocyte cultures after 2, 4, and 8 days, and the expression of the matrix genes col2 and colX was examined by both RT-PCR and Northern blot. Both wildtype and knockout cells had abundant expression of col2, and in both cultures, there was a gradual decrease in col2 expression over time (Figs. 1A and 1B). However, the level of col2 expression was significantly higher in Smad3−/− cells compared with wildtype cells (4- to 6-fold between 2 and 8 days in culture). ColX expression was minimal in wildtype cells at 2 days, and there was no evidence of maturation of these cells over time (Figs. 1C and 1D). In contrast, Smad3−/− chondrocytes had abundant colX expression by 2 days, indicating spontaneous maturation in cells lacking Smad3 expression. Collectively, the data establish the involvement of Smad3 in the regulation of chondrocyte-specific matrix genes and suggest a suppressive role on chondrocyte maturation.
Additional experiments were performed to assess the expression of other maturation-associated genes, including alkaline phosphatase, MMP9, MMP13, VEGF, and osteocalcin (Fig. 1E). Alkaline phosphatase, MMP13, and MMP9 have been associated with chondrocyte hypertrophy, whereas VEGF and osteocalcin are associated with chondrocyte terminal differentiation.(2–7) Consistent with a role for Smad3 as a regulator of chondrocyte maturation, both alkaline phosphatase, an early maturation marker, and the later maturational markers, VEGF and osteocalcin, were increased in 48-h cultures of Smad3−/− chondrocytes compared with wildtype chondrocytes. In contrast, neither MMP9 nor MMP13 expression was altered in Smad3−/−chondrocytes, suggesting that these genes are regulated independent of Smad3 signaling. Altogether, the data establish that Smad3 is an inhibitor of the rate of chondrocyte maturation.
The TGF-β signaling pathway is complex and involves a number of regulatory steps, including growth factor production, secretion, activation, receptor binding, activation of downstream signals, and modulation of signaling by associated regulatory factors.(38) Experiments were performed to determine if the absence of Smad3 results in compensatory changes in other components of the TGF-β signaling pathway (Fig. 2). The expression levels of Smads 1 through 8 were examined by RT-PCR in cultures harvested at 48 h (Fig. 2A). As anticipated, Smad3 expression was absent in knockout cells, and there was no change in the expression levels of other components of TGF-β–mediated Smad signaling, Smad2 and Smad4. In contrast, there was a slight upregulation of the BMP receptor associated Smads, Smad1 and Smad5, whereas Smad8 was unchanged. The BMP inhibitory Smad, Smad6, was slightly increased in the knockout cells, whereas there was no change in Smad7. A Western blot examining the protein levels of Smads 1, 2, 3, 5, and 6 confirmed the absence of a compensatory increase in TGF-β–associated Smad expression in Smad3−/− chondrocytes (Fig. 2B).
Additional experiments were performed to examine the expression levels of other genes involved in regulating TGF-β signaling (Fig. 2C). TGFβ1 and TGFBRI expression was reduced (40% and 45%, respectively) in Smad3−/−compared with wildtype chondrocytes. In contrast, the expression level of several other genes was unchanged, including TRIP1, TGFBRII, the nuclear cofactor c-Ski, and Smurf2, which is the E3 ubiquitin ligase involved in catabolism of TGF-β–associated Smads.(39) Among the factors examined, only the 47% percent reduction in the nuclear factor Sno would result in a compensatory increase in TGF-β/Smad signaling in knockout chondrocytes.
In contrast, BMP-2 and BMP-6 expression were increased ~2-fold in the knockout cells, whereas BMPs 4, 7, and 9 were unchanged (Fig. 2D). Coupled with increases in Smad1 and Smad5 (Fig. 2A), the findings suggest an increase in BMP signaling in Smad3−/− cells. Because BMP acts to stimulate maturation, the absence of Smad3 results in changes in gene expression that would tend to further accelerate chondrocyte differentiation.(28,40)
The effect of Smad3 deficiency on TGF-β and BMP signaling was directly measured in Smad3−/− chondrocytes in transient transfection experiments using reporters sensitive to TGF-β or BMP signaling. The P3TP-Luc reporter is derived from the plasminogen activator inhibitor-1 promoter and contains TGF-β–responsive elements including sequences responsive to Smad3 and ATF-2.(41,42) Smad3−/−chondrocytes have a 24% reduction in basal luciferase activity and a 30% reduction in TGF-β–stimulated activation of the P3TP-Luc reporter (Fig. 3A). However, in both wild-type and Smad3−/− chondrocytes, addition of exogenous TGF-β increased promoter activity compared with the respective controls, (4.7-fold in WT, 4.3-fold in KO), indicating that signaling components other than Smad3 remained operant in the knockout cells (Fig. 3A). In contrast, activation of the 4× SBE-Luc reporter, which contains only the Smad binding sequence, was markedly reduced in response to TGF-β treatment in Smad3−/− chondrocytes (Fig. 3B).(43,44) A 4.3-fold increase was present in wildtype chondrocytes compared with a 0.4-fold increase in Smad3−/−chondrocytes, confirming the specificity of Smad3 for SBE promoter activation (Fig. 3B).
BMP responsiveness was assessed using the 9× GCG-Luc promoter, which is responsive to BMP-associated Smad signaling (Fig. 3C).(45) Smad3−/− chondrocytes transiently transfected with the 9× GCG-Luc promoter had both higher basal luciferase activity and higher luciferase levels after treatment with 50 ng/ml of BMP-2 compared with wildtype chondrocytes (2-fold increase in basal activity; 3.9-fold increase in BMP-2–stimulated activity). To confirm increased BMP signaling, Western blot was performed using an antibody that detects phosphorylated Smad 1, 5, and 8, which is the activated form of these proteins involved in cell signaling (Fig. 3D). Consistent with the luciferase data, basal phosphoSmad 1/5/8 levels were substantially increased in Smad3−/− chondrocytes compared with wildtype cultures (Fig. 3D). Treatment with exogenous BMP (50 ng/ml) further enhanced phosphoSmads 1, 5, and 8 levels. The maximal induction was observed 15 minutes after BMP treatment, and similar to the basal levels, was elevated in the knockout chondrocytes. Altogether, these findings show that Smad3−/− chondrocytes have reduced TGF-β responsiveness, increased basal BMP signaling, and maintained a response to exogenous BMP-2.
Wildtype and Smad3−/− chondrocytes were treated with either control medium or medium containing either BMP-2 (50 ng/ml) or TGF-β (4 ng/ml) for up to 8 days in culture, and the effects on col2, colX, and alkaline phosphatase activity were examined. As previously observed (Fig. 1), basal levels of col2 were higher in Smad3−/− chondrocytes compared with wildtype chondrocytes (Fig. 4A). BMP-2 induced col2 expression in both the wildtype and knockout cells. The relative induction was slightly greater in wildtype chondrocytes (2.7-fold at 2 days, 5.4-fold at 4 days, and 9.7-fold at 8 days) compared with Smad3−/− chondrocytes (1.8-fold at 2 days, 4.1-fold at 4 days, and 3.4-fold at 8 days).
In contrast, col2 was disparately regulated by TGF-β in wildtype and Smad3−/− chondrocytes (Fig. 4A). After 2 days of treatment, TGF-β had minimal effects on col2 expression in either wildtype or knockout cells (Figs. 4A and and4B).4B). However, 4 and 8 days of continuous treatment resulted in a strong suppression of col2 in wildtype chondrocytes (87% and 99%, respectively), but a 62% and 80% induction in col2 was observed in Smad3−/− chondrocytes, respectively (Figs. 4C and and4D).4D). These findings suggest that Smad3 suppresses col2 expression in wildtype chondrocytes, whereas other TGF-β signals act to induce col2 in the absence of Smad3 signaling.
As shown in the initial experiments, wildtype chondrocytes have minimal expression of colX, whereas much higher basal levels are observed in Smad3−/− chondrocytes (Fig. 1). Addition of recombinant BMP-2 induced colX expression in wildtype and Smad3−/− chondrocytes (Figs. 5A–5C). The induction of colX by BMP-2 was sustained in the Smad3-deficient cells and increased over the 8 days of BMP-2 exposure. The increase, compared with control medium, in the knockout cells was 3-fold at 2 days, 27-fold at 4 days, and 106-fold at 8 days.
Wildtype chondrocytes do not express colX under basal conditions and thus do not provide a good model to study the effect of factors that slow the rate of chondrocyte maturation. However, Smad3−/− chondrocytes provide an excellent model to determine whether remaining signaling molecules downstream of TGF-β act to delay maturation. In Smad3−/− chondrocytes treated with TGF-β, colX expression was reduced ~50% compared with the untreated Smad3−/− control culture. Therefore, residual signaling molecules in the TGF-β pathway remain responsive and at least partly account for the inhibitory effects of TGF-β on maturation.
Alkaline phosphatase activity was measured in the cultures as a functional marker of chondrocyte maturation.(19,46) Alkaline phosphatase activity was increased in Smad3−/− chondrocytes compared with wildtype chondrocytes (4.2-fold higher), consistent with spontaneous maturation of these cells (Fig. 5D). BMP-2 treatment for 24 h increased alkaline phosphatase activity in both wildtype and knockout chondrocytes. The fold increase was slightly greater in the wildtype cells (4.4- versus 2.4-fold), although the overall magnitude of the effect was much greater in Smad3−/− chondrocytes because of the higher basal level of alkaline phosphatase activity. Unlike colX, which is only minimally expressed in wildtype cells, basal alkaline phosphatase activity was high enough to examine the suppressive effect of TGF-β on the activity of this enzyme both in wildtype and Smad3−/− chondrocytes. In wildtype chondrocytes, TGF-β essentially eliminated alkaline phosphatase activity, resulting in ~90% decrease. In contrast, effects on Smad3−/− chondrocytes were much smaller; only a 40% decrease was observed. These latter observations again show that Smad3-independent signals exist downstream of TGF-β that decrease the rate of chondrocyte maturation. Moreover, they confirm the role of Smad3 as a key regulator of the rate of chondrocyte maturation, because the TGF-β effect is much less in the absence of Smad3.
Smad3−/− chondrocytes were infected with retroviruses expressing GFP, Smad2, or Smad3. Western blot was performed to confirm viral induced protein production (Figs. 6A and 6B). Fluorescence microscopy showed a >50% rate of infection in the chondrocyte cultures (Figs. 6C and 6D).
After 48 h of infection, colX expression and alkaline phosphatase activity were examined (Figs. 6E and 6F). Smad3−/− chondrocytes infected with retrovirus-GFP had colX expression and alkaline phosphatase activity levels that were significantly greater than those observed in wild-type chondrocytes infected with retrovirus-GFP (Figs. 6E and 6F). Overexpression of Smad2 or Smad3 each suppressed colX expression by 70% in Smad3−/− chondrocyte cultures, confirming a key role for the TGF-β receptor-associated Smads as inhibitors of the rate of chondrocyte maturation. Similar to colX, overexpression of Smad2 and Smad3 reduced alkaline phosphatase activity in Smad3−/−chondrocytes, also with an ~70% suppression noted in cultures overexpressing Smad2 or Smad3.
To directly determine the role of BMP signaling in Smad3−/− chondrocyte maturation, recombinant murine noggin was added to Smad3−/− chondrocyte cultures, and colX mRNA expression and alkaline phosphatase activity were examined (Fig. 7). Noggin inhibited the increased expression of colX in Smad3−/− chondrocytes. In Smad3 −/−chondrocytes, which have a 12.5-fold greater basal expression of colX compared with WT, noggin reduced colX to a level lower than those observed in wildtype chondrocytes (Fig. 7A). Similar findings were observed on alkaline phosphatase activity (Fig. 7B). In Smad3−/− chondrocytes, addition of noggin to the cultures reduced alkaline phosphatase activity to a level similar to that observed in wildtype chondrocyte cultures. These experiments indicate that the loss of Smad3 signaling alone is not sufficient for the progression of maturation but requires the presence of BMP signaling. When BMP signaling is blocked in these cells, their accelerated maturation is abrogated.
In vivo models have established that TGF-β signaling inhibits chondrocyte maturation and regulates postnatal skeletal growth and development.(21,22) Importantly, TGF-β is necessary for the maintenance of articular cartilage and likely has a critical role in the pathogenesis of osteoarthritis.(21,22) These in vitro experiments show that Smad3 has an essential role as a suppressor of chondrocyte maturation. Primary murine sternal chondrocytes lacking Smad3 had spontaneous differentiation with increased expression of colX, alkaline phosphatase, and other markers of chondrocyte maturation. TGF-β signaling effects were reduced in Smad3−/− chondrocytes, without a compensatory increase in the expression of related signaling molecules, such as Smad2, TGF-β 1, or the TGF-β receptors. In contrast, Smad3−/− chondrocytes had increased responsiveness to BMP signaling associated with increased levels of expression of Smad1, Smad5, BMP-2, and BMP-6, and increased levels of both basal and BMP-2 stimulated phospho-Smad1, 5, and 8 levels. Overexpression of Smad2 or Smad3 blocked spontaneous maturation in Smad3−/− chondrocytes, as did addition of the extracellular BMP antagonist noggin. Altogether the experiments show that loss of Smad3 results in accelerated chondrocyte maturation caused by a shift in signaling from TGF-β to BMP.
Analysis of downstream BMP signaling effectors was limited to the Smad signaling molecules, and increases in both Smad1 and Smad5 were observed. The importance of BMP receptor Smads for chondrocyte maturation and BMP responsiveness has been previously shown.(28,47,48) Overexpression of Smad1 and Smad5 induces maturation in the murine chondrocyte cell line MC615.(49) Work in our laboratory has shown that retinoic acid stimulation of chondrocyte maturation in chicken embryonic caudal sternal chondrocyte cultures is dependent on induction of Smad1 and Smad5 protein expression.(28) We have also shown that addition of 5-azacytidine induces primary chick articular chondrocytes, which typically do not undergo maturation, to express colX and alkaline phosphatase, and causes these cells to be responsive to the maturational effects of BMP-2. The 5-azacytidine–treated articular chondrocytes showed decreased Smad2 and 3 and increased Smad1 and 5 expression, similar to the finding in our murine model.(47) Activation of the type X collagen promoter has been shown to involve a cooperative interaction of Smad1 and Runx2.(29,48) Recent in vivo experiments showed that over-expression of either Smad6 or Smurf1, both of which target BMP/Smad signaling, reduce the rate of chondrocyte hypertrophy and maturation.(50) Similarly, overexpression of Smad6 in chondrocyte cultures inhibits both spontaneous and BMP-2–induced maturation, whereas loss of function through RNAi has the opposite effect.(51) Thus, multiple intracellular and extracellular targets that converge on BMP-related Smad signals regulate chondrocyte maturation and show the complexity of this process.
Smad3−/− chondrocytes had an increase in basal BMP signaling as measured by the BMP-responsive GCG-luciferase promoter.(45) Both BMP-2 and BMP-6 expressions were increased in Smad3−/− chondrocytes. Because both BMP-2 and -6 are more highly expressed by hypertrophic chondrocytes, their increases in Smad3−/− cultures are consistent with a more differentiated phenotype.(52) Whereas Smad 1 and 5 mRNA levels were slightly increased, and total protein levels were similar in the two cultures, chondrocytes lacking Smad3 had a marked increase in basal levels of phosphorylated Smad1, 5, and 8, showing increased receptor-mediated signaling events. These findings provide a mechanism for the spontaneous maturation of Smad3−/− chondrocytes in culture, which is supported by the ability of the BMP antagonist noggin to block the expression of maturational characteristics in these cells.
The inhibitory effect of TGF-β on maturation has similarly been related to signals mediated by the TGF-β receptor-associated Smads, Smad2 and Smad3, in various culture models. Overexpression of Smad2 and Smad3 inhibited differentiation in primary embryonic sternal chondrocytes and in murine MC615 chondrocytes.(19,42,49) In this study, over-expression of either Smad2 or Smad3 was compensatory and inhibited maturation in Smad3−/− chondrocyte cultures. In embryonic metatarsal organ cultures, Ihh stimulates TGF-β 2 expression in the perichondrium, resulting in the suppression of maturation through induction of PTHrP.(53,54) In this model, both Smad2 and Smad3 were shown to mediate the suppressive effect of TGF-β on chondrocyte maturation.(55)
Smad3 deficiency resulted in an increase in col2 expression. Because BMP-2 induces col2, the effect is likely caused in part by a relative increase in BMP signaling in Smad3−/− chondrocytes. BMP-2 induces col2 expression in human and equine articular chondrocytes(56,57) and in C3H10T1/2 cells, TMC23 cells, and primary murine chondrocytes.(49,58,59) However, in the absence of Smad3, TGF-β directly induced col2 expression. This finding suggests that Smad3 acts to inhibit expression of col2, whereas the remaining components of the TGF-β signaling pathway enhance col2 expression.
TGF-β signals are transmitted through a multitude of pathways other than Smad3, including Smad2 as well as Smad-independent signaling, including JNK, p38 MAPK, and Erk MAPK.(60) TGF-β activates TGF-β –activated kinase-1 (TAK1), a MAPKKK family member, possibly through XIAP (X-linked inhibitor of apoptosis).(61) In addition to activating p38 kinase and JNK, TAK1 has also been shown to activate IκB kinases to induce NFκB signaling, providing another Smad-independent pathway for intracellular signaling.(60,62) Prior work by our laboratory has shown that both Smad3 and ATF-2 are required for complete inhibition of chondrocyte maturation by TGF-β.(39) Altering the balance between these non-Smad pathways and Smad-mediated signaling changes the cellular response to TGF-β. Whereas we and others have shown induction of col2 by TGF-β in various culture conditions,(17,63–65) these findings show that the overall effect is determined by an integration of multiple signals and suggest that Smad3 acts to inhibit col2 expression.
The relative response to the various signaling pathways likely explains some of the other differences observed in gene expression between wildtype and Smad3−/− chondrocytes. This includes the increase in Smad6 and decrease in sno expressions observed in Smad3−/− chondrocytes. We and others have shown that Smad6 is increased by BMP signaling and acts in a compensatory manner to decrease BMP signaling.(66) Thus, the increase in Smad6 is part of the normal compensatory effect observed in response to increased BMP signaling in Smad3−/− chondrocytes. Sno is a TGF-β co-repressor that is induced by TGF-β and also acts in a negative regulatory loop to inhibit TGF-β signaling.(67) Similarly, reduced sno expression is consistent with the decrease in TGF-β signaling present in Smad3−/− chondrocytes. In contrast, Smad3 deficiency had no effect on either MMP9 or MMP13 expression. MMP-9 has been shown to be regulated by TGF-β through NF-κB.(68) Similarly, although TGF-β/Smad3 induce MMP13, the effect is dependent on p38 MAPK activity.(69) Thus, the similar expressions of MMP9 and 13 in Smad3−/− and wildtype chondrocytes is likely related to the importance or predominance of related signaling pathways.
Transfection experiments using several different reporters were used to distinguish the relative responsiveness of wildtype and Smad3−/− chondrocytes to TGF-β and BMP. The P3TP-luciferase reporter is derived from the plasminogen activator inhibitor-1 promoter and contains a Smad3-responsive binding element in addition to binding elements for other transcription factors, including ATF-2.(41,42) Because TGF-β stimulates MAP kinase signaling and activates p38 kinase and subsequently ATF-2, it is not surprising that Smad3−/− chondrocytes have reduced, but residual, P3TP luciferase responsiveness to TGF-β.(42) In contrast, the Smad binding element (SBE) reporter is specific to Smad3, and responsiveness is absent in Smad3−/− chondrocytes.(43,44) On the other hand, the BMP responsive reporter, GCG-luciferase, has increased basal activity and increased stimulation after BMP-2 treatment of Smad3−/−chondrocytes, consistent with the observed increase in BMP-associated Smad signaling molecules.(45) Western blot showed increased basal levels of phosphoSmads 1, 5, and 8 in Smad3−/− chondrocytes, confirming increased basal BMP signaling in these cells; BMP treatment further increased expression of the phosphoSmads.
The increased BMP signaling and accelerated maturation apparent in sternal chondrocyte cultures has important implications for articular cartilage diseases. Smad3−/− mice have premature arthritis(20) associated with inappropriate accelerated maturation of articular chondrocytes. The findings further support the notion that loss of Smad3 signaling allows articular chondrocytes, which normally remain quiescent and maintain a chondrocyte phenotype, to inappropriately proceed to terminal maturation.(20) Additionally, the experiments define BMP signaling as an essential component of the accelerated rate of maturation that occurs in Smad3 deficiency. Thus, the interplay of TGF-β and BMP signaling determine chondrocyte fate; alteration of the relative magnitude of these antagonistic signals can disturb homeostasis and induce differentiation. Maturation is prevented by restoring Smad3 to these cells, overexpressing alternative TGF-β signaling molecules such as Smad2, or by blocking BMP signaling, as shown by inhibition of colX expression and alkaline phosphatase activity after the addition of noggin to Smad3−/− chondrocyte cultures. These three methods show that manipulations that alter the balance of TGF-β/BMP signaling control chondrocyte fate and represent potential therapeutic targets for human osteoarthritis and other cartilage diseases. Further in vivo studies are necessary to test these hypotheses.
This work was supported by RO1 AR 38945 and AR48681 (RJO).
The authors have no conflict of interest.