|Home | About | Journals | Submit | Contact Us | Français|
In the synovial membrane of patients with rheumatoid arthritis (RA), a strong expression of laminins and matrix degrading proteases was reported.
To investigate the regulation of matrix metalloproteinases (MMPs) in synovial fibroblasts (SFs) of patients with osteoarthritis (OA) and RA by attachment to laminin‐1 (LM‐111) and in the presence or absence of costimulatory signals provided by transforming growth factor β (TGFβ).
SFs were seeded in laminin‐coated flasks and activated by addition of TGFβ. The expression of genes was investigated by quantitative reverse transcriptase‐polymerase chain reaction (qRT‐PCR), immunocytochemistry and ELISA, and intracellular signalling pathways by immunoblotting, and by poisoning p38MAPK by SB203580, MEK‐ERK by PD98059 and SMAD2 by A‐83‐01.
Attachment of SF to LM‐111 did not activate the expression of MMPs, but addition of TGFβ induced a fivefold higher expression of MMP‐3. Incubation of SF on LM‐111 in the presence of TGFβ induced a significant 12‐fold higher expression of MMP‐3 mRNA, and secretion of MMP‐3 was elevated 20‐fold above controls. Functional blocking of LM‐111–integrin interaction reduced the laminin‐activated MMP‐3 expression significantly. Stimulation of SF by LM‐111 and TGFβ activated the p38MAPK, ERK and SMAD2 pathways, and inhibition of these pathways by using SB203580, PD98059 or A‐83‐01 confirmed the involvement of these pathways in the regulation of MMP‐3.
Attachment of SF to LM‐111 by itself has only minor effects on the expression of MMP‐1 or MMP‐3, but it facilitates the TGFβ‐induced expression of MMP‐3 significantly. This mode of MMP‐3 induction may therefore contribute to inflammatory joint destruction in RA independent of the proinflammatory cytokines interleukin (IL)1β or tumour necrosis factor (TNF)α.
In the course of rheumatoid arthritis (RA), fibroblast‐like cells of the synovial tissue become activated and express elevated levels of matrix‐degrading proteases and cytokines.1,2 Among the cytokines found to be elevated in RA synovial tissue or fluid, proinflammatory factors such as interleukin (IL)1β and tumour necrosis factor (TNF)α contribute to synovial pain, swelling and cartilage degradation.3,4 Mitogenic factors such as TGFβ1 or bFGF may promote synovial hyperplasia.5 However, the synovial fibroblasts (SFs) do not only respond to cytokines. Interactions of these cells with extracellular matrix proteins such as fibronectin, collagens and laminins (LMs) may activate SFs and thus contribute to articular pathologies in RA, synovial inflammation and hyperplasia.6,7 It has been shown that fibroblast‐like synoviocytes express increased levels of integrins, which correlate with their enhanced binding to matrix proteins.8,9 Integrin engagement is known to induce proliferation of SFs.10 Binding of intact fibronectin to cells via integrins suppressed the induction of matrix metalloproteinases (MMPs), but fibronectin fragments induced elevated MMP expression.11 Further, attachment to fibronectin suppressed p53‐mediated apoptosis of fibroblasts.7,12 Thus interaction of extracellular matrix compounds with SFs can modulate the joint pathology in RA or other disorders.
LMs are heterotrimeric, extracellular matrix proteins. LMs are composed of α, β and γ chains. In the human genome, five α, three β and three γ chains encoding genes have been identified and characterised. To date, 16 LM isoforms are known, designated LM‐111–LM‐523.13 Elevated expression of different isoforms of LMs was reported in RA specimen when compared with osteoarthritis (OA) samples.8 Binding to LM fragments or LM‐derived peptides induced elevated expression of different MMPs in macrophages.14,15 In mice, LM‐111, previously termed LN‐1, is typically expressed early in embryogenesis.16 In the adult organism, it shows a rather limited expression pattern, but LM‐111 was found to be expressed in tumour cells, in normal adult brain, kidney, testis, placenta, adrenal gland, liver and in cultured cells,16,17 and it was upregulated in the airway in a murine asthma model.18 In humans, elevated LM‐111 expression was described in endometriotic lesions and anti‐LM‐111 autoantibodies were found in infertile patients.19 Enhanced expression of LM‐111 was also reported in astrocytes of diseased areas in the frontal cortex of Alzheimer's brain tissues,20 and reduced expression was observed in acinar basement membranes of patients with Sjögren's syndrome,21 indicating that LM expression patterns change during pathological processes.
In the RA synovial tissue, elevated levels of integrin receptors and LMs colocalise in the synovial lining layer, with elevated expression of MMPs.8 Hence, it was hypothesised that stimuli contributing to enhanced integrin or LM expression might also induce MMPs in synovial cells. However, the metabolic changes in SFs on attachment to LMs have not been studied so far. We reasoned that attachment of RA‐SF to LM via integrin could also modulate the expression of MMPs and, therefore, set out to investigate whether attachment to LMs influences the expression of MMP‐1 and MMP‐3, two important proteases contributing to cartilage degradation in joints of patients with RA.
We report that TGFβ activates SFs to produce, on attachment to LM‐111, 20‐fold more MMP‐3 when compared with mock‐treated controls. SFs bind to LM‐111 through integrin receptors, and blocking of β1‐integrins reduced this effect of coactivation. TGFβ‐induced expression of MMP‐3 is mediated through the ERK and SMAD2 pathways and enhanced by phosphorylation of p38MAPK and extended or intensified phosphorylation of ERK on LM‐111 binding. The SAPK/JNK‐1 seems to regulate the expression of MMP‐3 as well.
SFs were isolated from surgical samples of 16 patients with RA diagnosed according to the revised American College of Rheumatology (ACR) criteria for RA and of 12 patients with OA after obtaining written consent22 (table 11 and supplementary files available on http://ard.bmj.com/supplemental). The cells were expanded in vitro as described.23 In brief, fibroblasts were isolated from synovial membranes by cutting the tissue, followed by enzymatic degradation, and expanded for 3 to 4 passages in 10% fetal calf serum‐Dulbecco's Modified Eagle's medium. All studies were approved by the local ethics committee.
For induction experiments, cells were stimulated for 24 h with 10 ng/ml recombinant human TGF‐β1, or were incubated on LM‐111‐coated or LM‐511/521‐coated flasks. Cells incubated in uncoated cell culture flasks w/o TGFβ served as controls.
To modulate the effects of LM, SFs were preincubated with 10 μg/ml anti‐CD29 monoclonal antibodies (anti‐β1 integrin, clone 4B4, Beckman Coulter, Krefeld, Germany) for 60 min on ice to block integrin receptors. Then the cells were incubated for 24 h in the presence or absence of TGFβ and LM‐111 and changes in the expression of MMPs were enumerated by quantitative reverse transcriptase‐polymerase chain reaction (qRT‐PCR).
To investigate signal transduction pathways, cells were incubated on LM‐111, in the presence of 10 ng/ml recombinant human TGFβ, or on LM‐111 and TGFβ. Untreated cells served as controls. 20 μM SB203580 was used to block the p38MAPK pathway,24 20 μM PD98059 to block the MEK‐ERK pathway,25 and ALK‐5 blocker A‐83‐01 to block TGFβ dependent SMAD‐2 signalling (100 nM, TGFβR‐I serine/threonine kinase blocker,26 all from Calbiochem, Schwalbach, Germany) were used. After overnight incubation, cells were harvested to measure changes in the expression of genes by qRT‐PCR.
The RNA was extracted from cells using a spin column technique (RNeasy, Qiagen, Hilden, FRG) to generate cDNA. Gene‐specific cDNA was enumerated by qRT‐PCR (qRT‐PCR, LightCycler, Roche, Mannheim, Germany) using commercially available primer pairs (SearchLC, Heidelberg, FRG). The qRT‐PCR was performed as touch‐down PCR in 35 PCR cycles (denaturing 95°C, 10 s; annealing 5 cycles 68°C, 10 s; then touch‐down to 58°C, 10 s; extension 72°C, 16 s). Quantification of GAPDH mRNA and a recombinant standard DNA (Roche) served as controls in each PCR.27 To detect LM‐111 mRNA, end point RT‐PCRs were performed (30 cycles; denaturing 94°C, 60 s; annealing 56°C, 60 s; extension 72°C, 120 s, respectively), and amplification products were separated and visualised on agarose gels.
The secretion of MMPs was measured in SF supernatants by ELISA. The cells were seeded in flasks, activated for 24 h by addition of 10 ng/ml rhTGFβ1, by attachment to LM‐111, or with both stimuli. Controls were incubated in uncoated cell culture flasks w/o TGFβ or LM‐111. Supernatants were harvested after 24 h of incubation, precleared by centrifugation (20000×g, 4°C, 5 min), and MMPs were measured by ELISA (R&D Systems, Minneapolis, Minnesota, USA).
Cells were stimulated by LM‐111 and TGFβ, harvested, lysed and subjected to sodium dodecyl sulphate‐polyacrylamide agarose gel electrophoresis and immunoblot analysis as described.28 Aliquots of 100 μg cellular protein were mixed with Laemmli sample buffer,28 denatured and then separated by electrophoresis in 10% sodium dodecyl sulphate‐polyacrylamide agarose gel electrophoresis. Proteins were transferred onto nylon membranes, blocked and probed overnight at 4°C with monoclonal antibody (mAb) specific for phospho‐p38MAPK (Thr180/Tyr182), total‐p38MAPK, phospho‐ERK1/2 (Thr202/Tyr204), phospho‐JNK/SAPK (Thr183/Thr185), phospho‐SMAD2 (Ser465/467; all from Cell Signalling Technology, Beverly, Massachusetts, USA). After rinsing of the membrane, binding of the primary antibodies was detected by peroxidase‐labelled goat anti‐rabbit‐IgG antiserum (Dianova, Hamburg, Germany). The binding of antisera was visualised by enhanced chemoluminescence (ECL, Amersham Biosciences, Freiburg, FRG) and recorded by a luminescence‐sensitive CCD camera system (Diana, Raytest Straubenhardt, FRG).
Experimental data were analysed by a two‐sided modified Student's t test and p values of datasets 0.05, 0.01 or 0.001 were considered to be significant. Each data point given represents the mean value ±SD from individual experiments using SFs from n different patients (4n14). A detailed description of all specimens including materials and methods can be found in the supplementary files (available at http://ard.bmj.com/supplemental).
Elevated expression of LM8 and elevated concentrations of growth factors including TGFβ are reported in RA synovial fluid.29 Expression of LM‐111 by SFs in vitro was confirmed (see supplementary fig S1 and files available at http://ard.bmj.com/supplemental). Attachment of SFs to LM‐111‐coated flasks without additional stimulation induced only slight but significant increases of MMP‐1 (2.3‐fold ±1.61 induction, p0.015) or MMP‐3 mRNA (1.49‐fold ±0.66 induction, p0.037, fig 1A1A).). This moderate induction was also observed with other batches of LM‐111, and in the presence of anti‐TGFβ antiserum (not shown). In addition, incubation of SFs on LM‐511 was investigated as well, but an effect of LN‐511 to MMP expression was not observed (not shown). Activation of SFs by 10 ng/ml TGFβ1 resulted in a significant induction of MMP‐3 mRNA (5.3‐fold ± 3.13 induction, p0.018), but with no effects on expression of MMP‐1 mRNA (fig 1A1A).). Incubation of SFs in LM‐111‐coated flasks in the presence of 10 ng/ml TGFβ augmented the expression of MMP‐1 mRNA to some extent (2.73‐fold ±1 induction, p0.004), but a significantly stronger induction of MMP‐3‐encoding message (14.21‐fold ±7.27 induction, p<0.004, fig 1A1A)) was observed. This expression of MMP‐3 was significantly higher when compared with expression of MMP‐1 (5.2‐fold, p0.018). No difference in the induction patterns of MMP‐1 or MMP‐3 between RA‐SF and OA‐SF was observed. The induction of tissue inhibitors of metalloproteinases (TIMPs) and other MMPs was less pronounced (not shown).
The release of MMP‐1 and MMP‐3 was confirmed by ELISA (fig 1B1B).). SFs produced spontaneously some MMP‐1 (mean 49 pg/ml), but activation of MMP‐1 production was not induced significantly by either method (mean 70–123 ng/ml, fig 1B1B).). SF spontaneously expressed less MMP‐3 (mean 25 ng/ml), but substantially more on activation with TGFβ (sevenfold ±0.92 induction, p0.012, mean 168±16 ng/ml). The production of MMP‐3 was further enhanced by attachment to LM‐111 in the presence of TGFβ (23.19‐fold ±1.96 induction, p<0.004, mean (SD) 492±73 ng/ml, fig 1B1B).). This LM‐111 and TGFβ induced release of MMP‐3 was significantly higher when compared with the stimulation of SF by LM‐111 only (19‐fold, p0.004) or that by TGFβ only (3.3‐fold, p0.009).
Among the different integrins, α6β1 integrin represents a major receptor for LM‐111 binding on fibroblasts. The expression of α6β1 integrin was verified on SFs in vitro by immunocytochemistry (fig 2A,B2A,B).). Preincubation of SF with anti‐β1 integrin mAb before seeding the cell into LM‐111‐coated flasks reduced the number of cells attached in comparison with mock‐treated controls (fig 2C,D2C,D).). This confirmed that the mAb 4B4 interfered with the LM–integrin binding on the cell surface. In SFs coactivated with TGFβ and LM‐111, the elevated MMP‐3 mRNA expression was reduced significantly (p<0.05) by preincubation of SFs with anti‐β1 integrin mAb (fig 33).). The data confirm the expression of β1 integrin on SF in vitro and suggest a functional role of this integrin chain in the regulation of MMP‐3 expression.
Signal transduction pathways contributing to the TGFβ and LM‐111 induced MMP‐3 expression were investigated by immunoblot analysis. A weak phospho‐ERK signal was seen in control cells before activation (fig 4A4A).). Addition of 10 ng/ml TGFβ to SFs induced a transient phosphorylation of ERK at 10–30 min after induction (fig 4A4A,, top). In contrast, phosphorylation of SMAD2 was not observed in control cells, but only on addition of TGFβ (fig 4A4A,, middle). The p38MAPK was detected in all samples (not shown), but phospho‐p38MAPK signals remained below detection levels (fig 4A4A,, bottom).
Coactivation of SFs by TGFβ in the presence of LM‐111 induced an enhanced and extended phosphorylation of ERK at 10–60 min after induction (fig 4B4B,, top). Again, phospho‐SMAD2 was not detected before stimulation, but a long‐lasting phosphorylation of SMAD2 was noted (fig 4B4B,, middle). Moreover, phospho‐p38MAPK signals were noted at 30 min (fig 4B4B,, bottom) and phosphorylation of SAPK/JNK‐1 was induced weakly as well (see supplementary fig S2 and files at http://ard.bmj.com/supplemental).
We further investigated whether the elevated expression of MMP‐3 could be blocked by addition of SB203580, a p38MAPK blocker, or by PD98059, a MEK/ERK blocker, or by A‐83‐01, interfering with the SMAD2 signalling pathway.26 Attachment of the cells to LM‐111 significantly enhanced the TGFβ‐induced MMP‐3 mRNA (3.1‐fold, p<0.003; fig 55).). Addition of SB203580, PD98059 or A‐83‐01 before stimulation of the SF reduced the elevated expression of MMP‐3 significantly (p0.005), confirming the importance of these pathways in TGFβ‐ and LM‐111‐dependent induction of MMP‐3 (fig 55).
Attachment of SFs to LM‐111 in the presence of TGFβ induced a significant expression of MMP‐3, a matrix protease that has been known for its involvement in tissue destruction in RA for a long time.30,31 MMP‐3 has a rather broad substrate specificity and degrades a variety of extracellular matrix proteins, including proteoglycans, collagens, fibronectin and LMs.32 Hence, MMP‐3 plays an important role in the pathomechanisms of RA. The induction of MMP‐3 is associated with inflammatory cytokines, especially IL1β or TNFα.
The additive effects of attachment of SFs to LMs together with TGFβ have not been investigated in detail so far, but some hints may have been derived from studies using Matrigel for investigation of the invasive properties of SF.33 In the present study, we show that attachment of SFs to LM‐111 sensitises these cells to TGFβ‐mediated signalling and significantly increases the expression of MMP‐3, a relevant protease in RA.
In recent studies, RA‐SF showed a higher invasive score on Matrigel when compared with control cells, and this correlated significantly with a higher expression of MMP‐1, MMP‐3 and MMP‐10.33 Matrigel contains LM‐111 as a major component. It therefore may activate SFs in a way comparable to our experiments. Using a ribozyme approach to limit the expression of MMP‐1 in RA‐SF confirmed that this increased invasiveness of SF on Matrigel was associated with the expression of MMP‐1.34 Our data, presented here, provide a molecular explanation for these findings, as we see an increased expression of MMP‐1 in SF on attachment to LM‐111. Addition of anti‐TGFβ antibodies to SF attached to LM‐111 did not change the LM‐induced effects in our experiments (data not shown). This finding confirms that our LM‐111 preparations used did not contain TGFβ. For elevated production of MMP‐1, cosignalling of LM‐111 and TGFβ is not sufficient, but LM‐111 together with TGFβ significantly elevated the expression of both MMP‐3 mRNA and protein.
As shown recently, most extracellular proteins derived from articular cartilage or connective tissue enhanced the attachment of SF mediated by integrins on the SF.8 Stronger adherence of RA‐SF to cartilage oligomeric protein was associated with their expression of αvβ1‐integrin.35 In other studies, elevated expression of α6‐ and β1‐integrins on RA‐SF correlated with their enhanced attachment to matrix proteins.9 In mice lacking β1‐integrin expression in chondrocytes, cell motility and proliferation were impaired,36 and the expression of MMP‐3, MMP‐9 and MMP‐13 was significantly reduced in these cells.37 We corroborated these results as treatment of SF with mAb 4B4 blocked the attachment of SF to LM‐111 and significantly reduced the induction of MMP‐3. Further, the induction of MMP‐3 described here was induced by TGFβ and LM‐111, without addition of IL1β or TNFα to the cells.
TGFβ is a multifunctional factor regulating cell division and differentiation. Depending on the individual experimental context, TGFβ may activate or reduce proliferation and expression of target genes in a dose‐ and time‐dependent manner.38 This factor has also been shown to influence the homeostasis of the extracellular matrix.39 TGFβ may repress the PMA‐induced expression of MMP‐1 by SMAD signalling.40 Without prestimulation by PMA, TGFβ activates the expression of MMPs including MMP‐3 using a p38MAPK signalling pathway.41 In our experiments, addition of TGFβ activated the ERK and SMAD2 but not the p38MAPK pathways. The importance of SMAD2 for regulation of MMPs is supported by the fact that SMAD2 transgenic mice show elevated expression of MMP‐1 and MMP‐2.42
Activation of ERK1/2 by expression of a recombinant constitutively active MEK1 was sufficient to induce MMP‐1 and MMP‐3, but strong expression of MMP‐1 and MMP‐3 was observed when ERK1/2 were activated in combination with either SAPK/JNK or p38MAP kinases. Further, ERK was associated with the induction of transcription factors c‐jun, junB and c‐fos, whereas p38α regulated MMP mRNA stability.24 In our experiments, SF from one patient failed to induce the expression of MMP‐3 when attached to LM‐111 in the presence of TGFβ. Investigation of these cells showed that phosphorylation of ERK had not occurred (data not shown). However, attachment to LM‐111 seems to address the ERK pathway in different mesenchymal cells.43
In our experiments, phosphorylation of p38MAPK was not observed by TGFβ alone but on coactivation by LM‐111 and TGFβ. We also quantified less MMP‐3 mRNA after blocking p38MAPK activity in activated SF by the kinase blocker SB203580, corroborating that p38MAPK is involved in LM‐111 and TGFβ induced expression of MMP‐3 expression. A very weak signal of phospho‐SAPK/JNK was noted in LM‐111 and TGFβ activated SF, but not in SF activated by TGFβ alone (see supplementary fig S2 and files at http://ard.bmj.com/supplemental), indicating that additional MAP kinases are addressed by these stimuli.44 However, a detailed investigation of the contribution of other MAP kinases to the TGFβ and LM‐111 induced expression of MMPs in SF must await additional experiments.45
In summary, we show that attachment of RA‐SF and OA‐SF to LM‐111 in the presence of TGFβ is sufficient to yield a significant MMP‐3 expression. Our data corroborate the importance of ERK, p38MAPK and SMAD2 signalling for efficient MMP‐3 induction by TGFβ and attachment to LM‐111 in SF. Other cytokines such as IL1β or TNFα were not necessary for the expression of MMP‐3 under these conditions. We conclude that pathological expression of LMs in the synovium contributes to the elevated expression of MMP‐3 in RA. This pathway may also explain in part the progression of cartilage destruction when patients with RA are treated with anti‐TNF (or anti‐IL1) therapies, as levels of TGFβ did not change during etarnecept treatment.46
Supplementary files are available on http://ard.bmj.com/supplemental
We thank Mrs T Abruzzese, A Hack and M Weis‐Klemm for excellent experimental support, Dr D Alexander for technical advice, and Prof Klaus von der Mark (University of Erlangen, FRG) for providing purified murine EHS LM‐111 for coating of culture flasks.
JNK - jun‐N kinase‐1
LM - laminin
mAb - monoclonal antibody
MAP - mitogen‐activated protein
MMP - matrix metalloproteinase
OA - osteoarthritis
qRT‐PCR - quantitative reverse transcriptase‐polymerase chain reaction
RA - rheumatoid arthritis
SAPK - stress‐activated protein kinase
SF - synovial fibroblast
SMAD - transcription factor (homologue to mothers against DPP and SMA genes)
TGF - transforming growth factor
TIMP - tissue inhibitor of metalloproteinase
Funding: This project was supported by DFG‐grants Ai16/10‐2 and Ai16/14‐1 to WKA and in part by institutional funding. SG was supported by the European Community's FP6 funding. This publication reflects only the authors' views. The European Community is not liable for any use that may be made of the information herein. There are no disclosures for this study by any of the authors.
Competing interests: None declared.
Supplementary files are available on http://ard.bmj.com/supplemental