Objectives Chondrocytes, the only cells in the articular cartilage, play a pivotal role in osteoarthritis (OA) because they are responsible for maintenance of the extracellular matrix (ECM). Follistatin-like protein 1 (FSTL1) is a secreted protein found in mesenchymal stem cells (MSCs) and cartilage but whose function is unclear. FSTL1 has been shown to modify cell growth and survival. In this work, we sought to determine whether FSTL1 could regulate chondrogenesis and chondrogenic differentiation of MSCs.
Methods To study the role of FSTL1 in chondrogenesis, we used FSTL1 knockout (KO) mice generated in our laboratory. Proliferative capacity of MSCs, obtained from skulls of E18.5 embryos, was analysed by flow cytometry. Chondrogenic differentiation of MSCs was carried out in a pellet culture system. Gene expression differences were assessed by microarray analysis and real-time PCR. Phosphorylation of Smad3, p38 MAPK and Akt was analysed by western blotting.
Results The homozygous FSTL1 KO embryos showed extensive skeletal defects and decreased cellularity in the vertebral cartilage. Cell proliferation of FSTL1-deficient MSCs was reduced. Gene expression analysis in FSTL1 KO MSCs revealed dysregulation of multiple genes important for chondrogenesis. Production of ECM proteoglycans and collagen II expression were decreased in FSTL1-deficient MSCs differentiated into chondrocytes. Transforming growth factor β signalling in FSTL1 KO cells was significantly suppressed.
Conclusions FSTL1 is a potent regulator of chondrocyte proliferation, differentiation and expression of ECM molecules. Our findings may lead to the development of novel strategies for cartilage repair and provide new disease-modifying treatments for OA.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Osteoarthritis (OA) is the most common degenerative joint disease. It is characterised by loss of articular cartilage, changes in the subchondral bone, and osteophytosis. The exact cause of the disease remains unknown, and there is no effective therapy for OA, except joint replacement surgery. Typically, OA begins with aging-related disruption of the articular cartilage surface, although specific risk factors, such as joint trauma or metabolic disorders increase the risk of developing OA. Chondrocytes, the only cells in the articular cartilage, play a pivotal role in OA because they are responsible for maintenance of the extracellular matrix (ECM), the primary target of osteoarthritic cartilage degradation. OA cartilage cells display dysregulation of anabolic and catabolic processes as well as profound reduction in cell number due to a decrease in cell proliferation and an increase in apoptotic cell death.
Since articular cartilage has little or no regenerative capacity, the use of mesenchymal stem cells (MSCs) has become an attractive approach for cartilage repair.1 MSCs are capable of proliferating and differentiating into chondrocytes in culture; however, the process of in vitro chondrogenesis is not completely understood. For the future success of cartilage-regenerative medicine, it is critical to elucidate the regulatory mechanisms underlying chondrogenesis.
A number of factors, such as insulin-like growth factor (IGF), transforming growth factor β (TGFβ), bone morphogenetic proteins (BMPs), fibroblast growth factors (FGFs), the sex determining region Y-type high-mobility group box (SOX) family of transcription factors, and other cell differentiation-related signals, including Indian Hedgehog (IHH) and Wnt proteins have been implicated in the process of chondrocyte proliferation and differentiation.2–5 Proper cartilage development is linked to a delicate balance in the activity of the above-mentioned factors. Their activities are controlled at multiple levels, both extracellularly and intracellularly. Despite the progress made in recent years in understanding signalling mechanisms, the role of different factors involved in the fine regulation of chondrogenic pathways remains to be elucidated.
This study examined the contribution to articular cartilage homoeostasis of follistatin-like protein 1 (FSTL1), an extracellular protein whose functional role in physiological and pathological processes is still unclear. FSTL1 was cloned as a TGFβ-inducible protein from a mouse osteoblastic cell line.6 Our previous studies showed that FSTL1 is an important mediator in the pathogenesis of rheumatoid arthritis and other systemic autoimmune diseases.7–9 We also found that FSTL1 enhances the ability of T cells and monocytes/macrophages to respond to inflammatory signals.10–12 But, unlike most proinflammatory proteins, haematopoietic cells do not produce FSTL1.7 Rather, it is expressed in cells of the mesenchymal lineage, endothelial cells and neurones.13–16 It is noteworthy that FSTL1 is strongly expressed in the mesenchyme of multiple organs during embryogenesis.17–20 Loss-of-function studies in mice demonstrated that homozygous FSTL1 knockout (KO) animals die at birth due to abnormalities of the respiratory system, particularly, tracheomalacia.17 ,18 FSTL1 KO mice also display extensive skeletal defects, further supporting the role of FSTL1 in chondrogenesis.17 ,18 FSTL1 has been found to affect cell growth and survival. For example, FSTL1 inhibits apoptosis in cardiomyocytes and endothelial cells and promotes endothelial cell function and blood vessel growth.14 ,16 Recent studies17 ,18 ,21 also showed that FSTL1 may interact with key mediators of MSC chondrogenesis, proteins of the TGFβ superfamily or their receptors, and modulate their activities. Thus, we hypothesise that FSTL1 may play an important role in maintaining cartilage homeostasis by enhancing cell proliferation, survival and anabolic activity in articular chondrocytes. In the present work, we sought to determine whether FSTL1 could regulate chondrogenesis as well as to identify the molecular mechanisms for FSTL1-mediated regulation of differentiation of MSCs into chondrocytes.
Generation of FSTL1-deficient mice
All animal experiments were approved by the University of Iowa Animal Care and Use Committee. The targeting vector and the FSTL1WT/flox mice were generated by Ozgene (Australia) (see online supplementary methods for details of the breeding strategies).
Skeletal preparations and histological examination
Alcian blue/alizarin red staining of cartilage and bone in E18.5 embryos was performed following the standard protocol. For histological examination of lumbar vertebrae, FSTL1 KO and wild-type (WT) embryos at day E18.5 were fixed in 5% formaldehyde and then transferred to 70% ethanol the next day and subjected to paraffin embedding, sectioning and H&E staining using standard protocols.
Mouse MSC cultures and chondrogenic differentiation of MSCs
MSCs were isolated from skulls of E18.5 embryos by collagenase digestion and differentiated into chondrocytes in a pellet culture system (see online supplementary methods for details).
Quantitative reverse transcriptase (RT)-PCR
Total cDNA was produced from MSCs or chondrocyte pellets using the RNeasy Mini Kit (Qiagen) and the SuperScript II Reverse Transcriptase Kit (Invitrogen). PCR was performed in a LightCycler (ABI7000) using the Brilliant SYBR Green QPCR Master Mix (Agilent) as described.11 The copy number (number of transcripts) of amplified product was calculated from a standard curve obtained by plotting known input concentrations of plasmid DNA (see online supplementary table S1 for primer sequences).
Analysis of proliferative capacity of MSCs
MSCs were loaded with carboxyfluorescein succinimidyl ester (CFSE) and analysed after 1 week by flow cytometry. Data were analysed with FlowJo software (Tree Star).
Adenoviral vector and overexpression of human FSTL1
A replication-defective adenovirus vector encoding the human FSTL1 gene (Ad-FSTL1) was generated as described.22 To overexpress FSTL1, A549 cells were infected with Ad-FSTL1 adenovirus (1000 viral particles/cell). The cells were cultured in serum-free medium for 4 days; the supernatant was collected and analysed by ELISA for human FSTL1.12 The supernatant was used as the starting material for further FSTL1 purification (see online supplementary methods).
RNA processing, microarray and data analysis
Total RNA was extracted from MSCs isolated from WT and KO embryos using the RNeasy kit (Qiagen). Microarray analysis was performed with the MouseWG-6 V2.0 Expression BeadChip (Illumina) containing more than 45 200 transcripts by the Genomics and Proteomics Core Laboratories, University of Pittsburgh. Each sample was analysed using two separate arrays, and values for replicate arrays were averaged. For a comparison of FSTL1 KO MSC versus WT control, a threshold (cut-off point) >2 was selected.
For the signalling study, MSCs were incubated with 5 ng/mL TGFβ for 1 h, followed by protein extraction for western blot analysis. In some experiments, the KO cells were incubated in the presence of FSTL1 (1 or 5 μg/mL) overnight before TGFβ stimulation.
MSCs were lysed with Cell Lysis Buffer (Sigma) supplemented with proteinase inhibitor cocktail (Sigma) and phenylmethanesulfonyl fluoride following the manufacturer's protocol. The amount of protein was measured by BCA assay (Sigma). Sodium dodecyl sulfate/polyacrylamide gel electrophoresis and immunoblotting were performed as described.11 Phospho-p38 MAPK, p38 MAPK, phospho-Smad3, Smad3, phospho-Akt, Akt and actin were detected with rabbit polyclonal antibodies (Cell Signalling). Blots were developed using horseradish peroxidase-conjugated secondary antibodies (Thermo) with a chemiluminescent substrate (Thermo). Membranes were scanned by the LAS-4000 imaging system (Fuji Film) and images were analysed by Multi Gauge software.
The parametric Student's t test was used to assess the significance of differences between groups. Data are presented as mean±SEM unless otherwise specified. p<0.05 was considered significant.
Skeletal defects in FSTL1-deficient mice
To study the role of FSTL1 in chondrogenesis, we generated mice with a targeted loss-of-function mutation of the FSTL1 gene. PCR analysis confirmed the absence of the FSTL1 transcript in E18.5 KO embryos (figure 1A). The expected Mendelian ratios for genotypes were observed up to day E18.5; however, KO of both FSTL1 alleles was lethal at birth. The homozygous FSTL1 KO embryos were smaller in size than their WT littermates and showed several skeletal defects, including abnormal head position, spine curvature and anomalous long bones (figure 1B). The heterozygous FSTL1−/+ mice were viable and did not display any obvious defects (not shown). Skeletal abnormalities found in FSTL1 KO embryos imply that FSTL1 plays an important role in cartilage development.
Impaired proliferative capacity of FSTL1-deficient chondrocytes
To reveal the impact of FSTL1 deficiency on vertebral cartilage development, the spines from day E18.5 WT and homozygous KO mice were examined using H&E stain. Sections of vertebrae in FSTL1 KO mice displayed decreased cellularity in the embryonic cartilage (figure 2A). Quantitative analysis demonstrated a statistically significant reduction in the number of chondrocytes (figure 2B) and smaller lacunae (figure 2C) in the FSTL1 KO mice. These changes in cartilage probably reflect an intrinsic defect in chondrocyte proliferative capacity. To test this hypothesis, we made use of MSCs (the progenitors of chondrocytes) isolated from WT and KO embryos. Figure 2D shows that FSTL1 KO MSCs labelled with CFSE had a reduced rate of cell division.
Critical role of FSTL1 in chondrogenic differentiation of MSCs
Since FSTL1 expression is upregulated in response to TGFβ, one of the key regulators of chondrogenesis, we hypothesised that FSTL1-deficient MSCs might display defects in TGFβ-induced differentiation into chondrocytes. WT MSCs differentiated into chondrocytes normally, as demonstrated by Safranin O staining of ECM proteoglycans (figure 3A). Chondrocyte differentiation of FSTL1 KO MSCs was impaired, with decreased production of ECM proteoglycans (figure 3A). In addition, expression of the chondrocyte-characteristic marker gene, COL2A1, was significantly downregulated in TGFβ-differentiated FSTL1 KO MSCs (figure 3B). Together, these data indicate that poor MSC differentiation into chondrocytes is associated with functional impairments of FSTL1-deficient cells.
Analysis of expression of genes important for chondrogenesis
To further investigate the mechanistic basis of FSTL1 effects on MSC function, we performed global gene expression microarray analysis using RNA isolated from WT and FSTL1 KO MSCs. A number of genes known to be involved in chondrogenesis showed over- or under-expression in mutant cells (figure 4A). Among these genes, underexpressed SOX9 and COL2A1 are of particular interest because they play a crucial role in chondrogenic differentiation of MSCs. The alterations in gene transcription were also confirmed by real-time PCR (figure 4B).
Alteration in TGFβ-induced signalling in FSTL1 KO MSCs
To gain further insight into the molecular mechanisms by which FSTL1 affects chondrogenic differentiation of MSCs, we analysed the canonical TGFβ/Smad2/3 signalling pathway as well as alternative signalling pathways in MSCs that are stimulated with TGFβ. Protein extracts from WT and KO cells were analysed for phospho-Smad3, phospho-p38 MAPK and phospho-Akt by western blotting. TGFβ-induced phosphorylation of Smad3, p38 MAPK and Akt was markedly decreased in FSTL1 KO cells (figure 5). MSCs lacking FSTL1 had reduced levels of both phospho-Smad3 and total Smad3 (figure 5A), suggesting that an additional pathway (or pathways) regulates Smad3 signalling.
We next examined whether incubation with exogenous recombinant FSTL1 can rescue the Smad3 or p38 MAPK signalling phenotype in KO MCSs. As shown in figure 5D, phosphorylation of both Smad3 and p38 MAPK in KO cells was increased in the presence of FSTL1. Addition of FSTL1 to KO MSC culture medium at a concentration of 5 μg/mL returned phosphorylation of the signalling molecules to the levels seen in control WT cells. Exogenous FSTL1 also increased the amount of total Smad3 protein detected in FSTL1 KO cells.
In this study, we found that embryos without FSTL1 are neonatal lethal, although they survive up to the end of gestation at Mendelian ratios. The embryos showed multiple developmental abnormalities of the skeleton. This skeletal phenotype resembles the phenotype in FSTL1 KO mice reported by other groups.17 ,18
As was shown in earlier studies, knockdown of FSTL1 orthologues resulted in perturbation of early mesoderm development during zebrafish and chicken embryogenesis.19 ,20 Geng et al17 and Sylva et al18 also found that one cause of death of FSTL1 KO mice is apparent respiratory failure due to malformation of the cartilaginous tracheal rings. Based on these reports, we hypothesised that KO embryos display changes in the quality of cartilage, and we compared the cellularity in the fetal vertebral cartilage of FSTL1 KO embryos with that in WT. The KO cartilage had fewer cells, probably due to a difference in the rate of chondrocyte cell division, and MSCs from FSTL1 KO animals had reduced proliferation. In addition, gene expression analysis showed a profound decrease in expression of DIO3 (iodothyronine deiodinase type 3), a negative regulator of the hormone, 3,5,3′-l-tri-iodothyronine (T3), in FSTL1 KO MSCs. It has been shown that, in the absence of T3 hormone, cells maintain proliferation, whereas its binding to T3 receptors stimulates hypertrophic chondrocyte differentiation.23 Although these very large changes in cells grown several generations after isolation may or may not correlate with changes in the embryos as a whole, our results highlight a possible FSTL1-mediated mechanism of regulation of MSC proliferation through enhancement of DIO3 expression, and, as a consequence, preventing T3 access to the cells. Thus, we propose that FSTL1 is capable of regulating the rate of mesenchymal progenitor cell proliferation during chondrogenesis, thereby influencing the pace of their differentiation into chondrocytes.
Decreased cellularity found in the embryonic cartilage of FSTL1 KO mice might also reflect enhanced apoptosis of chondroprogenitor cells. We demonstrated downregulation of SOX9 expression in FSTL1 KO MSCs. Recently, it has been shown that the Bax/Bcl2 ratio was increased in SOX9-deficient MSCs, and, as a consequence, apoptosis was increased.24 Studies to reveal the role of FSTL1 in apoptosis of articular chondrocytes and to address the possibility that Bcl2 expression is downregulated in FSTL1 KO MSCs are currently underway.
MSCs are functionally defined by their self-renewal ability and their capacity to differentiate along the adipogenic, chondrogenic and osteogenic lineages.25 In our studies, loss of FSTL1 markedly inhibited TGFβ-induced differentiation of FSTL1 KO MSCs into chondrocytes, demonstrated by the absence of proteoglycan and the decreased expression of type II collagen, the prototypic gene of the chondrocyte lineage.26 It has previously been shown that COL2A and SOX9 expression is indispensible for the early stages of the differentiation process. SOX9 drives the expression of COL2A,27 and therefore decreased expression of COL2A might be secondary to underexpressed SOX9. SOX5 expression was inhibited in FSTL1 KO cells. Since it has been reported that the combination of SOX5, SOX6 and SOX9 (the SOX trio) provides signals sufficient for induction of permanent cartilage,28 it is most likely that, in the absence of FSTL1, the cells are not capable of maintaining a normal articular chondrocyte phenotype. Furthermore, FSTL1 KO cells express more COL1A, a distinctive feature of fibrocartilage, but not articular hyaline cartilage. Loss of FSTL1 led to alterations in several other pathways controlling chondrocyte proliferation and differentiation, including those regulated by BMP4, IGF and Wnt (table 1), suggesting that commitment of FSTL1 KO MSCs to become chondroprogenitor cells may be significantly distorted.
The precise mechanism by which FSTL1 might control gene transcription is unclear, and we cannot rule out the possibility that the effect of FSTL1 on expression of various genes, such as SOX9, is indirect and related to the observed changes in expression of BMP4 and/or Wnt members. Reduced expression of SOX9 in FSTL1 KO MSCs might be due to the decrease in BMP4 expression and/or the increase in the expression of Wnt members, since it has been reported that SOX9 expression is activated by BMPs38 and repressed by excessive Wnt/β-catenin signalling.39
Recent studies have demonstrated that alterations in TGFβ signalling pathways could disrupt the anabolic–catabolic balance in articular cartilage and result in irreversible degradation of the ECM.40 TGFβ/Smad3 signals repress excessive/terminal differentiation of chondrocytes to the hypertrophic phenotype and are required for the formation of articular cartilage at the end of the long bones.41 Our data show that FSTL1 enhances the activity of various intracellular signalling molecules of the TGFβ pathway, which include canonic Smad3 as well as alternative p38 MAPK and Akt. It is noteworthy that we found less total Smad3 in TGFβ-induced FSTL1 KO MSCs. Our microarray data did not reveal a significant difference in Smad3 mRNA levels in control cells and cells lacking FSTL1, suggesting some post-transcriptional regulation of Smad3 expression. It has recently been shown that Smad3 is regulated at the protein level but not at the RNA level by microRNA (miRNA)-140.42 TGFβ suppressed the accumulation of miRNA-140, whereas miR-140 suppressed the TGFβ pathway through repression of Smad3, forming a double-negative feedback loop.42 We suggest that FSTL1, a TGFβ-inducible molecule, might negatively regulate the accumulation of miRNA-140, thereby promoting activation of Smad3 in response to TGFβ.
It has been demonstrated that SOX9 transcription can be regulated not only via the Smad3 pathway,43 but also via the p38 MAPK pathway,44 and inhibition of p38 MAPK activity leads to severe cartilage-degenerative changes.45 This is in agreement with our observation that FSTL1 KO MSCs have decreased SOX9 expression and less phosphorylated p38 MAPK. Activation of the phosphatidylinositol 3-kinase/Akt pathway in response to IGF-1 is reduced with aging and in OA.46 We suggest that decreased Akt activation in the absence of FSTL1 could play a role in the increased cell death and reduced response of chondrocytes to IGF-1 observed in OA cartilage.
Published information on FSTL1 expression in OA cartilage is limited and contradictory. In the study by Wang et al,47 little FSTL1 immunostaining was found in the chondrocytes of the articular cartilage from patients with OA; however, the results also showed weak FSTL1 staining in the articular cartilage from control trauma patients. In contrast, the proteomic analysis of chondrocyte secretomes from patients with OA and a low-Mankin-scored individual showed a significantly higher amount of FSTL1 in the macroscopically healthy cartilage.48 The study of conditional FSTL1 KO mice in various OA models would be useful to test if FSTL1 plays a role in the pathogenesis of OA. These mice have recently been generated in our laboratory, and in vivo studies are currently underway.
TGFβ, a known FSTL1 inducer,6 is also a potential tool for repair of OA cartilage or prevention of its degradation.49 However, there are side effects of TGFβ-based therapy, such as synovial fibrosis and osteophyte formation.50 In addition, chondrocytes in OA cartilage may not respond well to TGFβ because of the reported reduction of TGFβ receptor expression during OA.49 Our findings that FSTL1 may interact with the TGFβ signalling pathway and potentially correct its defects suggest that FSTL1 might be a potential therapeutic candidate for treating articular cartilage disorders.
In conclusion, in this study we show that FSTL1 is a crucial component of the regulatory mechanism controlling chondrocyte proliferation and differentiation and expression of ECM molecules. FSTL1 plays a critical role in the extensive crosstalk between the numerous signalling pathways that are activated in MSCs in response to TGFβ. Further research to determine how FSTL1 is integrated into the signal-transduction network maintaining cartilage function and integrity is important to be able to correct the imbalance in TGFβ signalling that occurs in OA. Our findings may lead to the development of novel strategies for cartilage repair based on targeted FSTL1 gene delivery and provide new disease-modifying treatments for OA.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Handling editor Tore K Kvien
Contributors All authors contributed to the conception and design of the study, acquisition and/or interpretation of data, drafting the article or revising it critically for important intellectual content, and approved the final version of the manuscript.
Funding This work was supported by the NIH (grants RO1AI073556 and RO1AR056959 to RH).
Competing interests The University of Pittsburgh has a patent ‘Immunomodulation of inflammatory conditions utilising Follistatin-like Protein-1 and agents that bind thereto’ (No 8334274) listing RH as an inventor.
Provenance and peer review Not commissioned; externally peer reviewed.