Cadherin 11 is a homophilic cell-to-cell adhesion molecule expressed on joint synovial fibroblasts. Absence of cadherin 11 in a mouse rheumatoid arthritis (RA) model led to striking reductions in cartilage erosion. Matrix metalloproteinases (MMPs) are enzymes expressed by synovial fibroblasts important for cartilage erosion. The objective of this study was to determine if synovial fibroblast MMP production is regulated by cadherin 11.
To mimic cadherin 11 engagement, human RA synovial fibroblasts were stimulated with a chimeric construct consisting of the cadherin 11 extracellular domain linked to the human IgG1 Fc domain (Cad-11-Fc). Effects on MMP production were measured by enzyme-linked immunosorbent assay, quantitative reverse transcription–polymerase chain reaction analysis, and immunoblotting.
Human Cad-11-Fc up-regulated MMP-1 and MMP-3 protein production by RA synovial fibroblasts, both alone and in synergy with tumor necrosis factor α. This up-regulation required cell cadherin 11 engagement, since a mutant Cad-11-Fc with reduced binding affinity stimulated significantly less MMP production. Also, short hairpin RNA (shRNA) cadherin 11 silencing almost completely inhibited Cad-11-Fc–induced MMP expression. Cad-11-Fc stimulation increased RA synovial fibroblast MMP messenger RNA levels. It also increased the phosphorylation of the MAPKs JNK, ERK, and p38 kinase, the phosphorylation of NF-κB p65, and the nuclear translocation of activator protein 1 transcription factor. MAPK and NF-κB inhibitors partially blocked RA synovial fibroblast MMP expression.
Cadherin 11 engagement stimulates increased synthesis of several MMPs by RA synovial fibroblasts in a MAPK- and NF-κB–dependent manner. These results underscore the existence of a pathway by which cadherin 11 regulates MMP production and has important implications for joint destruction in RA.
The cadherin family consists of at least 6 subfamilies of cell adhesion molecules characterized by the presence of 110–amino acid immunoglobulin-like extracellular cadherin domains (1). The classic cadherins are comprised of an extracellular N-terminal region consisting of 5 cadherin domains, a single-pass transmembrane domain, and a cytoplasmic tail (2). Cell-to-cell adhesion is mediated by homophilic binding of the extracellular domains, namely, a cadherin on one cell binds a cadherin of the same type on an adjacent cell. On each cell, the cadherins organize into adherens junctions to link to and regulate the actin cytoskeleton through interactions at their cytoplasmic tails with the catenin molecules: β-catenin, α-catenin, and p120 catenin (3). Besides connecting with actin, cadherins and their associated catenins interact with a variety of signaling molecules, including growth factor receptors, soluble kinases/phosphatases, and Rho family small GTPases (4, 5).
Cadherins play a critical role in embryogenesis, where temporospatial expression of specific cadherins results in cell aggregation, directing migration of cell sheets and tissue morphogenesis (2, 6). Cadherin expression continues to show tissue specificity important for the maintenance of adult tissue architecture. For example, expression of E-cadherin is a hallmark of epithelial layers, and loss of E-cadherin results in epithelial lining layer disruption (7).
Cadherin expression is also important in the synovium, the lining tissue of diarthrodial joints that provides lubrication for movement and nourishment for articular cartilage (8). The normal synovium consists of a cellular lining layer overlying a mainly connective tissue sublining layer. The synovial lining is generally 1–3 cell layers thick and consists of both fibroblast and macrophage cell types lying in close proximity. The sublining layer is mainly extracellular matrix, with scattered blood vessels, fibroblasts, and innate immune cells. A cloning technique identified cadherin 11 expression in synovial tissues, and further studies confirmed strong expression of this molecule by lining layer synovial fibroblasts (9). Furthermore, the presence of cadherin 11 appears essential for the development of normal synovial architecture, since mice deficient in cadherin 11 show a hypoplastic synovial lining characterized by loss of both lining cells and extracellular matrix (10).
The pathogenesis of rheumatoid arthritis (RA) involves infiltration of the synovial sublining by leukocytes and transformation of the normal synovial lining into a hyperplastic, invasive tissue pannus capable of eroding bone and cartilage (8). Cadherin 11 on synovial fibroblasts appears to be critical for the full development of this pathologic process. When inflammatory arthritis was induced in cadherin 11–null mice via the K/BxN serum–transfer model, cadherin 11–null mice developed ∼50% less joint inflammation as compared to wild-type mice (10). Strikingly, cadherin 11–null mice were almost completely protected from cartilage erosions, despite ongoing bone erosion. These studies support a central role for synovial fibroblasts in cartilage invasion and damage in RA and reveal cadherin 11 as an important regulator of this synovial fibroblast behavior.
Key to cartilage erosion in RA are the matrix metalloproteinases (MMPs), a diverse family of zinc-dependent endopeptidases with broad specificity against extracellular matrix components (11). In particular, the collagenases (MMPs 1, 8, 13, and 14), belong to a very restricted subset of enzymes that are capable of cleaving intact fibrillar collagens, like the type II collagen matrix that provides tensile strength to cartilage (12). Several MMPs are up-regulated in RA synovial fluid and tissues (13, 14), and MMP expression has been positively correlated with increased synovial fibroblast invasion (15–18). Furthermore, increased MMP-3 expression is a specific biomarker of joint damage in RA patients (19).
In this study, we found that cadherin 11 engagement on RA synovial fibroblasts by a fusion protein linking the cadherin 11 extracellular domain to the human IgG1 Fc domain (Cad-11-Fc) induced the expression of several MMPs, both alone and, strikingly, in synergy with inflammatory cytokines, such as tumor necrosis factor α (TNFα). Increased MMP production was dependent on the activation of RA synovial fibroblast MAPK and NF-κB pathways. These results point to an unexpected pathway by which cadherin 11 regulates RA synovial fibroblast degradative and invasive behavior, suggesting that the formation and turnover of cadherin 11 contacts in pannus tissue may facilitate cartilage erosion through up-regulation of MMP expression.
MATERIALS AND METHODS
Cell culture and media.
Human synovial tissues from patients with RA and osteoarthritis (OA) were obtained after synovectomy or joint replacement surgery performed as part of indicated clinical care at Brigham and Women's Hospital. Synovial fibroblasts were released from synovial tissue by mincing followed by collagenase digestion, were purified by serial passage as previously described (20), and were used in experiments between passages 5 and 10. Purified normal human lung and skin fibroblast lines were obtained commercially (Lonza). Fibroblast cell lines were cultured at 37°C under an atmosphere containing 10% CO2 in 10% serum-containing medium: Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS; Gemini Bio-Products), 2 mML-glutamine, 100 units/ml of penicillin, 50 μM 2-mercaptoethanol, and essential and nonessential amino acids (Gibco BRL). RA synovial fibroblast cell lines were serum-starved prior to the assays by incubating them for 24–72 hours in medium identical to that used for routine culture, except that the FBS content was reduced to 1%. All enzyme-linked immunosorbent assays (ELISAs) were performed on culture supernatants obtained from 15,000 synovial fibroblasts/well stimulated in 96-well plates.
Cadherin 11 fusion proteins, antibodies, and other reagents.
Cad-11-Fc fusion proteins were stably expressed in HEK 293 cells after cloning the cadherin 11 extracellular domain (residues 1–1,827) into the pFuse-human IgG1-Fc1 vector (InvivoGen) (21). Cad-11-Fc was purified from culture supernatants with a protein A column (Bio X Cell). The following antibodies were used: human IgG1 isotype control (Sigma), mouse IgG1 isotype control (clone MOPC-21; Bio X Cell), anti–cadherin 11 monoclonal antibody 23C6 (10), donkey anti-mouse IgG phycoerythrin conjugate (Jackson ImmunoResearch), and anti–phosphorylated JNK (catalog no. 4668), anti–total JNK (catalog no. 9252), anti–phosphorylated ERK-1/2 (catalog no. 4370), anti–total ERK-1/2 (catalog no. 4695), anti–phosphorylated p38 kinase (catalog no. 4115), anti–phosphorylated p65 (catalog no. 3033), and anti–total p65 (catalog no. 4764) (Cell Signaling Technology).
Other reagents that were used are as follows: human MMP-1 and MMP-3 ELISAs (R&D Systems), human TNFα and interleukin-1β (IL-1β; R&D Systems), PathScan phospho–NF-κB p65 (Ser536) ELISA (Cell Signaling Technology), lipopolysaccharide (LPS) (from Salmonella; Sigma), JNK inhibitors JNK I (HIV-TAT-JNK–binding domain peptide construct) and SP600125 (EMD Chemicals), ERK inhibitors U0126 and FR180204 (EMD Chemicals), p38 kinase inhibitor SB203580 (EMD Chemicals), and NF-κB (IKK-2) inhibitors IKK-2 inhibitor IV and SC514 (EMD Chemicals). Inhibitors SP600125, U0126, FR180204, SB203580, IKK-2 inhibitor IV, and SC514 were first dissolved in DMSO before diluting in DMEM-containing media, while the JNK I inhibitor was directly soluble in media.
Cell adhesion assay.
RA synovial fibroblasts were released from culture flasks using 0.02% trypsin (Worthington) in HEPES buffered saline containing calcium (HBSCa) (20 mM HEPES, 137 mM NaCl, 3 mM KCl, 1 mM CaCl2, pH 7.4) for 5 minutes at 37°C to minimize cadherin 11 proteolysis. After adding 1 volume of 0.04% trypsin soybean inhibitor (Sigma) in HBSCa, cells were washed and fluorescence labeled for 1 hour with 15 μg/ml of calcium AM (Invitrogen) in 10% serum-containing medium. Labeled RA synovial fibroblasts (30,000 cells/well) were resuspended in HBSCa containing 0.1% bovine serum albumin (BSA; Sigma) and added to Cad-11-Fc–coated 96-well microplates that had been blocked with 1% BSA. Cells were incubated for 30–90 minutes at 37°C to allow adhesion. The percentage adhesion was calculated by dividing the fluorescence remaining after sequential washing by the starting fluorescence and multiplying by 100, as previously described (22).
Synovial fibroblast lentiviral transfection.
Lentivirus containing short hairpin RNA (shRNA) against cadherin 11 (TRCN0000054334 and TRCN0000054335; Thermo Scientific Open Biosystems) or a control sequence (Mission Control; Sigma) were generated by transfecting 293T cells with a combination of packaging plasmid pCMV-dR8.91 (Broad Institute), envelope plasmid VSV-G/pMD2.G (Broad Institute), and hairpin-pLKO.1 vector. RA synovial fibroblasts plated at confluency in 6-well plates were spun at 700g for 30 minutes in medium containing equivalent virus concentrations and 6 μg/ml of Polybrene (Sigma). After 2 days, 3.5 μg/ml of puromycin (Sigma) was added for 2–3 days to select for infected cells. Cells were then trypsinized (0.02% trypsin, 5 mM EDTA) and replated for 1 day in 10% serum-containing medium before use in experiments.
Cadherin 11 surface expression after infection was analyzed by flow cytometry (FACSCanto; BD Biosciences) as previously described (9). Anti–cadherin 11 23C6 or mouse IgG1 isotype control antibodies were used for primary staining and donkey anti-mouse IgG phycoerythrin conjugate for secondary staining.
Total cellular RNA was extracted from equivalent numbers of cultured RA synovial fibroblasts using RNeasy Micro Columns (Qiagen) and reverse transcribed using a QuantiTect reverse transcription kit (Qiagen) according to the manufacturer's instructions. Equivalent volumes of complementary DNA were analyzed by qPCR using Brilliant SYBR Green QPCR Master Mix (Agilent Technologies) and the following primers (Integrated DNA Technologies): for MMP-1, 5′-TAAAGACAGATTCTACATGCGC-3′ (forward) and 5′-GTATCCGTGTAGCACATTCTG-3′ (reverse); for MMP-3, 5′-CTGCTGTTGAGAAAGCTCTG-3′ (forward) and 5′-AATTGGTCCCTGTTGTATCCT-3′ (reverse); for MMP-13, 5′-GACATTCTGGAAGGTTATCCC-3′ (forward) and 5′-AGTATCATCATATCTCCAGACC-3′ (reverse); for MMP-14, 5′-TCATGATCTTCTTTGCCGAG-3′ (forward) and 5′-GATGTCATTTCCATTCAGATCC-3′ (reverse); and for hypoxanthine guanine phosphoribosyltransferase, 5′-GGGCTATAAATTCTTTGCTGAC-3′ (forward) and 5′-CTGGTCATTACAATAGCTCTTCAG-3′ (reverse).
Cell lysates and Western blots.
RA synovial fibroblasts (150,000 cells/well, 6-well plates) were lysed using the following buffer: 50 mM Tris, 150 mM NaCl, 1% Triton X-100, 0.1% sodium dodecyl sulfate (SDS), 1 mM EDTA, 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, and a mixture of protease inhibitors (Complete; Roche Applied Science). Lysates were cleared of insoluble material by centrifugation, subjected to electrophoresis though SDS–polyacrylamide gels, and then transferred to Immobilon-P membranes (Bio-Rad). Membranes were blocked in TBS-T (25 mM Tris HCl [pH 7.2], 150 mM NaCl, 0.05% Tween) supplemented with 5% nonfat skim milk or 5% BSA (for detecting phosphorylated proteins) and incubated with antibodies against the target proteins. After washing, the membranes were incubated with horseradish peroxidase–conjugated anti-mouse IgG or anti-rabbit IgG before detection with a chemiluminescent substrate.
Activator protein 1 (AP-1) detection and NF-κB p65 ELISA.
To assay for AP-1, RA synovial fibroblast nuclear extracts were isolated by lysing cells in hypotonic buffer (10 mM HEPES, pH 7.9, 1 mM MgCl2, 10 mM KCl, 0.1% Triton X-100, 20% glycerol, 1 mM dithiothreitol, and protease inhibitors), resuspending the remaining pellet in hypertonic buffer (10 mM HEPES, pH 7.9, 0.1% Triton X-100, 400 mM NaCl, 1 mM EDTA, 20% glycerol, 1 mM dithiothreitol, and protease inhibitors), and then collecting the supernatant after centrifugation. AP-1 was detected in 5 μg of total protein/well using an anti–phosphorylated c-Jun antibody in the DNA-binding TransAM AP-1 ELISA (Active Motif) according to the manufacturer's instructions. For the ELISA for phosphorylated NF-κB p65, RA synovial lysates were sonicated using the buffer provided in the PathScan phospho–NF-κB p65 (Ser536) ELISA kit (Cell Signaling Technology). Phosphorylated p65 was detected in 5 μg of total protein/well according to the manufacturer's instructions.
Increased synovial fibroblast expression of MMP-1 and MMP-3 by Cad-11-Fc, both alone and synergistically with inflammatory cytokines.
Given the in vivo findings that cadherin 11 deficiency protects mice with inflammatory arthritis from cartilage degradation, we hypothesized that cadherin 11 engagement might regulate synovial fibroblast production of MMPs that are important for cartilage erosion in RA. Initially, we focused on MMP-1 and MMP-3, two MMPs that are up-regulated by RA synovial fibroblasts in response to inflammatory cytokines (23). MMP-1 is a collagenase that is capable of cleaving intact type II collagen fibrils found in cartilage, and retroviral MMP-1 silencing in RA synovial fibroblasts was shown to block cartilage invasion in an ex vivo assay (11, 16). MMP-3 is a stromelysin with activity against many noncollagen matrix proteins; it also cleaves other MMP zymogens, leading to their activation (24). Furthermore, increased serum levels of MMP-3 are a biomarker for joint damage in RA patient populations (19).
To model cadherin 11 engagement pharmacologically, we developed a soluble, recombinant Cad-11-Fc fusion protein that binds to cell cadherin 11 in a manner like that of cadherin 11 interactions between adhering cells (9, 25) (Figure 1A). These cadherin fusion proteins, when bound to latex beads, have been shown to interact with cell surface cadherins, recruiting catenins and linking to the actin cytoskeleton (21). Incubation of serum-starved human RA synovial fibroblasts with increasing concentrations of human Cad-11-Fc increased the production of MMP-1 and MMP-3 in a dose-dependent manner, as compared to no stimulation or stimulation with matched human isotype control (Figure 1B). The ability of cadherin 11 to regulate MMP production was not unique to RA synovial fibroblasts. Cultured OA synovial fibroblasts, normal skin fibroblasts, and normal lung fibroblasts also expressed cell surface cadherin 11 and up-regulated MMP-3 expression after incubation with Cad-11-Fc (data available upon request from the authors).
The inflammatory milieu in the RA joint contains cytokines such as TNFα and IL-1β that strongly induce the expression of MMP-1 and MMP-3 by RA synovial fibroblasts. To test whether cadherin 11 stimulation augments cytokine-driven MMP expression, RA synovial fibroblasts were incubated with Cad-11-Fc or low-dose TNFα, either alone or in combination, and the effect on MMP-3 expression was measured by ELISA (Figure 1C). The combination of Cad-11-Fc and low-dose TNFα induced more MMP-3 expression than the additive effects of each agent alone: for Cad-11-Fc alone, 1,164 pg/ml and for TNFα alone, 735 pg/ml; predicted stimulation based on additive effect 1,899 pg/ml; actual stimulation based on combination 4,720 pg/ml. On average, there was 2.5-fold increased MMP production by combined stimulation with Cad-11-Fc and TNFα as compared to the sum of each agent alone (n = 8 experiments). Similar results were seen with IL-1β stimulation (data not shown). These results suggest that cadherin engagement acts in synergy with inflammatory cytokines to stimulate MMP expression, complementing the effects of inflammatory cytokines on RA synovial fibroblast activation.
Dependence of Cad-11-Fc action on engagement of cell surface cadherin 11 molecules.
Two approaches were taken to determine if Cad-11-Fc acts specifically through engagement of cadherin 11, rather than nonspecific mechanisms. In the first approach, we created a mutant Cad-11-Fc fusion protein by exchanging tryptophans 2 and 4 in the first extracellular domain for alanines. Structural studies have shown that insertion of these tryptophans into a hydrophobic pocket in the first extracellular domain on the cadherin from another cell bound in trans helps stabilize cadherin-to-cadherin adhesion (26, 27). Consistent with the known important role of these amino acids in increasing cadherin 11 binding affinity, RA synovial fibroblasts adhered substantially less well in a static adhesion assay to plates coated with the mutant Cad-11-Fc (Cad-11-Fc W2,4A) than to plates coated with similar amounts of wild-type Cad-11-Fc (Figure 2A). Correspondingly, when RA synovial fibroblasts were stimulated with equivalent amounts of wild-type and mutant Cad-11-Fc (Figures 2B and C), the mutant Cad-11-Fc induced significantly less MMP-1 and MMP-3, consistent with its reduced binding affinity for cellular cadherin 11 (mean ± SD percentage decrease in MMP expression by mutant Cad-11-Fc stimulation compared to wild-type 59.8 ± 13.4% for MMP-1 [n = 3 cell lines] and 60.3 ± 25.3% for MMP-3 [n = 4 cells lines]).
In addition, to confirm that cadherin 11 expression at the cell surface is required for Cad-11-Fc activation, we tested the ability of Cad-11-Fc to induce MMPs after lentiviral shRNA silencing of cellular cadherin 11. Lentiviral infection with a cadherin 11–specific virus decreased cadherin 11 expression by >90% compared to control virus and compared to uninfected cells, as shown by surface staining and flow cytometry for cadherin 11 (Figure 3A) or by messenger RNA (mRNA) levels (data not shown). Stimulation of MMP-3 production by Cad-11-Fc was almost completely lost in cadherin 11–silenced cells compared to control infected cells (Figure 3B). In contrast, stimulation of MMP-3 production by either LPS or IL-1β was unaffected by cadherin 11 silencing, providing further evidence that Cad-11-Fc induces MMP expression specifically through binding to RA synovial fibroblast cadherin 11.
Increased mRNA expression of a subset of MMPs following Cad-11-Fc stimulation.
MMP activity is tightly controlled by several mechanisms, including transcription regulation, zymogen activation, and production of endogenous inhibitors (11, 28). Since MMP-1 and MMP-3 protein levels rose substantially after Cad-11-Fc stimulation, qRT-PCR was used to test whether Cad-11-Fc stimulated the mRNA expression of a panel of MMPs. Cad-11-Fc induced a rise in MMP-1, MMP-3, and MMP-13 mRNA levels compared to the appropriate isotype control (2 representative cell lines are shown in Figure 4). These 3 MMPs are known to be cytokine-inducible and dependent on activation of the transcription factor AP-1 (11). In contrast, the more constitutively expressed MMP-14 was not induced by Cad-11-Fc, suggesting that cadherin 11 engagement regulates a subset of, but not all, MMP transcription. As a control, Cad-11-Fc did not alter cadherin 11 mRNA levels (data not shown).
Increased MMP production by activating MAPKs and NF-κB following Cad-11-Fc stimulation.
MMPs 1, 3, and 13 belong to an inducible group of MMPs whose proximal promoter contains both a TATA box and AP-1 transcription factor consensus sequences (28). The constitutively expressed MMP-14, on the other hand, belongs to a subset of MMPs without either a TATA box or AP-1 consensus sequences in the proximal promoter. The ability of Cad-11-Fc to induce the transcription of MMPs 1, 3, and 13, but not MMP-14, suggests that it signals through AP-1 activation. This hypothesis is consistent with what has been described for MMP-1 and MMP-3 transcription by inflammatory cytokines in RA synovial fibroblasts (12, 29, 30). In RA synovial fibroblasts and many other cell types, MMP-1 and MMP-3 transcription by upstream stimuli is critically dependent on activation of the MAPK signaling cascade. Once phosphorylated, the MAPK family members ERK, JNK, and p38 kinase stimulate AP-1 activity, both directly through AP-1 phosphorylation and indirectly by increased synthesis of the Jun and Fos family members that comprise the AP-1 heterodimer (31).
To determine whether Cad-11-Fc also induces MMP expression by a MAPK and AP-1–dependent pathway, we first tested total cellular RA synovial fibroblast lysates made from unstimulated or Cad-11-Fc–stimulated cells for MAPK phosphorylation. Cad-11-Fc increased ERK-1/2, JNK, and p38 phosphorylation (Figure 5A), as compared to unstimulated controls. Next, we determined whether or not Cad-11-Fc increased AP-1 activation by isolating nuclear fractions from control or Cad-11-Fc–stimulated RA synovial fibroblasts. Cells incubated with Cad-11-Fc increased phosphorylated AP-1 in their nucleus, suggesting that Cad-11-Fc induces MMP transcription by accumulation of nuclear AP-1 upon MAPK activation (Figure 5B).
MMP-1 and MMP-3 expression in synovial fibroblasts also depends on activation of the transcription factor NF-κB (32, 33). To determine whether cadherin 11 engagement increases NF-κB activity, total cell lysates isolated from unstimulated, Cad-11-Fc–stimulated, or TNFα-stimulated RA synovial fibroblasts were assayed for NF-κB p65 phosphorylation by either Western blotting or ELISA (Figure 5C). As was seen for MAPK/AP-1, Cad-11-Fc also induced NF-κB activation in RA synovial fibroblasts.
To determine whether JNK, ERK-1/2, p38 kinase, or NF-κB activation leads to MMP-3 expression, RA synovial fibroblasts were incubated with inhibitors specific for each of these pathways prior to stimulation with Cad-11-Fc. Inhibition of ERK-1/2, JNK, p38 kinase, or NF-κB activity all reduced MMP-3 expression by ∼25–50% in multiple RA synovial fibroblast cell lines (Figures 6A–D). These results confirm that both MAPKs and NF-κB play a role in translating the signal induced by Cad-11-Fc engagement at the cell surface to MMP expression by RA synovial fibroblasts.
There is a growing appreciation that synovial fibroblasts are key participants in cartilage erosion in RA. Early histologic studies showed close proximity of synovial fibroblasts to sites of cartilage erosion, indicating a role of these cells in mediating cartilage damage (34, 35). In vitro studies then demonstrated that synovial fibroblasts are invasive cells capable of secreting many MMPs that are important for cartilage degradation (13, 18, 29, 36), while ex vivo studies using cartilage implanted under the renal capsule in SCID mice showed that RA synovial fibroblasts, but not OA or normal synovial fibroblasts, were capable of mediating cartilage damage in the absence of an adaptive immune system (37). More recently, in vivo studies genetically depleting a molecule dominantly expressed by synovial fibroblasts, cadherin 11, confirmed that synovial fibroblasts are critical drivers of cartilage erosion. The absence of cadherin 11 resulted in almost complete protection of cartilage from erosions after induction of serum-transfer arthritis, even though the amount of inflammation was only partially reduced and bone erosions occurred (10).
Cartilage erosion by synovial fibroblasts likely involves several steps, including attachment to extracellular matrix, cell migration over and into cartilage, and digestion of the cartilage matrix. These steps are classically driven by integrin–matrix interactions. How cadherin cell-to-cell adhesion molecules influence cell migration, invasion, and matrix degradation is not fully understood, but there are growing examples of cross-talk between integrin and cadherin signaling pathways (4, 38). Furthermore, tumor models have shown that loss of epithelial E-cadherin with aberrant acquisition of mesenchymal cadherins, such as N-cadherin and cadherin 11, leads to an epithelial-to-mesenchymal transition, increasing tumor cell invasion and metastasis, thus providing evidence that cadherins contribute to invasive processes by several mechanisms (39, 40).
Previous studies in our laboratory showed that the absence or blockade of cadherin 11 on synovial fibroblasts reduced cell migration and invasion through basement membrane extracts (10, 41), while the expression of cadherin 11 in a cadherin-null fibroblast cell line promoted invasion (42). Here, we suggest an additional, unique pathway by which cadherin 11 may influence cartilage erosion. Incubation of RA synovial fibroblasts with Cad-11-Fc, a molecule that mimics cadherin 11 engagement, increased the synthesis of several inducible MMPs implicated in joint destruction in inflammatory arthritis. These MMPs include MMP-1, a collagenase shown to be a major mediator of fibroblast-driven cartilage erosion in an ex vivo model system (16), and MMP-3, a stromelysin that has proved valuable as a joint damage biomarker in RA trials (19). Interestingly, not all MMPs were regulated by cadherin 11. RA synovial fibroblast expression of MMP-14, a membrane-bound MMP that has been shown in 2 in vitro assay systems to be critical for synovial fibroblast cartilage invasion (15, 17), was not increased by Cad-11-Fc.
Cartilage erosion in RA is a complex process. Current models suggest that degradative enzymes secreted by the synovium and synovial fluid inflammatory cells damage the cartilage surface, removing the protective proteoglycan layer (43, 44). This damage exposes the underlying cartilage matrix and allows deposition of immune complexes and fibrin (45, 46). The hyperplastic synovial lining then converts into pannus tissue, further perpetuating cartilage damage by attaching to, invading, and degrading the cartilage surface (47).
The overlapping expression and functions of the 23 human MMPs has made it challenging to define which MMPs play a role in the various steps that lead to cartilage damage. Experiments using intact collagen matrices have shown that, once activated, soluble MMPs are capable of digesting broad areas of matrix, while membrane MMPs mediate more focal, pericellular cell degradation and invasion (48). Therefore, cadherin 11, by inducing the expression of several soluble MMPs, may help promote early damage to the cartilage matrix that facilitates later synovial fibroblast attachment, allowing other MMP pathways to promote direct cell invasion into the cartilage matrix.
The formation of new cadherin contacts causes activation of cell signaling pathways, as measured by a transient increase in cell tyrosine phosphorylation (5). In vitro, we pharmacologically mimicked this process using Cad-11-Fc and showed both increased phosphorylation of the MAPKs ERK, JNK, p38, and NF-κB p65 and nuclear translocation of AP-1 (Figure 5). However, individually, NF-κB and MAPK activation contributed only partially to MMP expression (Figure 6), and combinations of MAPK inhibitors or MAPK/NF-κB inhibitors were unable to completely block MMP production (data not shown), suggesting that other signaling pathways are also important. In addition to AP-1 and NF-κB consensus sequences, various MMP promoters contain combinations of binding sites for many transcription factors, including Ets, STAT, and T cell factor/lymphoid enhancer factor family members (12, 28). The precise role other signaling pathways play in cadherin 11–stimulated MMP expression requires further study.
The upstream signals between Cad-11-Fc engagement at the cell surface and MAPK and NF-κB signaling are not known. Cadherins at the cell surface associate with growth factor receptor tyrosine kinases, soluble kinases and phosphatases, catenins, and Rho family small GTPases (4, 5), which alone or in combination might help transmit signals upon cadherin 11 binding. This study makes it clear that cadherins are not just passive mediators of cell-to-cell adhesion. Rather, engagement of cadherins provides a mechanism for communicating extracellular signals inside the cell.
Although cell-to-cell contacts are maintained in normal tissues, these contacts are continually remodeled, with individual cadherin molecules recycling in and out of the cell contact site, breaking and reforming cadherin binding interactions (49). This remodeling process may provide a homeostatic level of cell signaling appropriate to help maintain normal cell function and tissue architecture, not just for synovial fibroblasts but for fibroblasts from other tissues when cadherin 11 is expressed. In a noninflammatory state, any MMPs stimulated by this process would likely contribute to basal extracellular matrix remodeling.
In the RA synovium, however, histologic evaluation has shown that the number of cadherin 11 contacts is increased (50). We propose that either increased numbers and/or increased turnover of cadherin 11 contacts in the compacted, hyperplastic RA synovial lining stimulates cadherin 11–mediated cell signaling, activating RA synovial fibroblasts and increasing their sensitivity to inflammatory cytokines such as TNFα and IL-1β. This process results in increased MMP production (shown in this study) and inflammatory mediator secretion (21). Consistent with this model, this study showed that Cad-11-Fc acted synergistically with low-dose TNFα to promote marked MMP-3 production. These findings suggest a model in which the formation and turnover of cadherin 11 contacts in pannus tissue act together with the inflammatory cytokines to stimulate the production of enzymes that degrade the cartilage matrix, leading to the eventual destruction of joint structure in RA.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Brenner had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Noss, Chang, Brenner.
Acquisition of data. Noss, Chang, Watts.
Analysis and interpretation of data. Noss, Chang, Watts, Brenner.