Article Text
Abstract
Tetraspanins function as membrane adaptors altering cell-cell fusion, antigen presentation, receptor-mediated signal transduction and cell motility via interaction with membrane proteins including other tetraspanins and adhesion molecules such as integrins. CD82 is expressed in several malignant cells and well described as tumour metastasis suppressor. Rheumatoid arthritis (RA) is based on persistent synovial inflammation and joint destruction driven to a large extent by transformed-appearing activated synovial fibroblasts (SF) with an increased migratory potential.
Objective CD82 is upregulated in RA synovial fibroblasts (RASF) compared with osteoarthritis (OA) SF as well as within RA compared with OA synovial lining layer (LL) and the role of CD82 in RASF was evaluated.
Methods CD82 and integrin immunofluorescence was performed. Lentiviral CD82 overexpression and siRNA-mediated knockdown was confirmed (realtime-PCR, Western blot, immunocytochemistry). RASF migration (Boyden chamber, scrape assay), attachment towards plastic/Matrigel, RASF-binding to endothelial cells (EC) and CD82 expression during long-term invasion in the SCID-mouse-model were evaluated.
Results CD82 was induced by proinflammatory stimuli in SF. In RA-synovium, CD82 was expressed in RASF close to blood vessels, LL, sites of cartilage invasion and colocalised with distinct integrins involved in tumour metastasis suppression but also in RA-synovium by RASF. CD82 overexpression led to reduced RASF migration, cell-matrix and RASF-EC adhesion. Reduced CD82 expression (observed in the sublining) increased RASF migration and matrix adhesion whereas RASF-EC-interaction was reduced. In SCID mice, the presence of CD82 on cartilage-invading RASF was confirmed.
Conclusion CD82 could contribute to RASF migration to sites of inflammation and tissue damage, where CD82 keeps aggressive RASF on site.
- rheumatoid arthritis
- fibroblasts
- inflammation
Statistics from Altmetric.com
Introduction
The proteins of the tetraspanin superfamily function as membrane adaptors facilitating protein interactions.1 2 There are 34 known tetraspanins in mammals and of those 33 are expressed in humans. Tetraspanins alter processes including antigen presentation, receptor-mediated signalling and cell motility. The effects are mediated by forming multimeric complexes consisting of tetraspanins and other proteins including kinases, adhesion molecules and adaptor proteins.2 Tetraspanins form interlinked networks of functional hubs, called tetraspanin-enriched microdomains regulating the clustering of associated receptors on the cell.1
Structure and splicing variants of the tetraspanin CD82 are well described3 4 with varying glycosylation.5 6 Although CD82 is expressed in different tissues including bone marrow as well as hematopoietic cells,5 its function in inflammatory diseases is unclear. CD82 is a well characterised metastasis suppressor of different solid malignant tumours without affecting primary tumour growth4 7 and a recognised biomarker to predict metastatic potential.5 CD82 inhibits multiple steps of the metastatic cascade including cell motility, invasion, cell-cell-interaction, proliferation, apoptosis and senescence.2 Regarding cell-cell-adhesion, CD82 overexpression enhances homophilic aggregation of different cells including tumour cells and leucocytes2 4 and mediates hematopoietic progenitor cell/osteoblast and tumour cell/endothelial cell (EC) interactions.8 9 In cell-matrix-adhesion, CD82 is negatively correlated with solid tumour cell adhesiveness onto matrices,3 whereas in leukaemia cells CD82 had opposite effects.7
CD82 regulates the signalling of membrane molecules by altering their distribution, thus inhibiting migration and invasiveness of solid tumour cells.2 4 CD82 interacts with cell surface molecules such as integrins via its extracellular loops and mediates rearrangement and clustering of these molecules on the cell surface. Trafficking of cell adhesion proteins, for example, integrins and CD44, is modified by CD82 thus altering their initiated outside-in signalling and their cell surface presence.3 10 11 CD82 is partially localised in lipid rafts.12 DARC, a CD82 binding partner, is a cell surface protein expressed in endothelium.8 DARC/CD82 engagement inhibits tumour cell proliferation and induces senescence. However, experimental CD82 overexpression did not inhibit proliferation of cancer cells.3
CD82 expression in melanoma cells reduced IL-8 secretion facilitating adherens junction disassembly and tumour cell extravasation.13 14 In turn, silencing CD82 in leukaemia stem cells reduced IL-10 and MMP915 while overexpression induced IL-10 in melanoma.16 CD82 serves as negative regulator of epidermal growth factor receptor (EGFR) signalling via association with the EGFR17 18 and suppresses the TGF-β1 pathway, which in turn affects CD82-mediated migration inhibition.19 Interestingly, CD82 expression is induced by proinflammatory cytokines such as IL-6, TNFα, IL-1β, growth factors and hypoxia.2 4
Rheumatoid arthritis (RA) is a chronic autoinflammatory joint disease and with key therapeutic targets being TNFα, IL-6, IL-1β.20 CD82 was increased in RA synovial fibroblasts (RASF) compared with primarily non-inflammatory osteoarthritis (OA)SF21 as well as in the hyperplastic synovial lining layer (LL) of patients with RA.22 RASF are permanently activated cells showing increased migration towards and invasiveness into cartilage of affected joints, thereby being a key cell promoting joint inflammation and destruction but also carrying the disease to non-affected areas.23 24 Due to the increased presence of CD82 in the RA synovial LL and close to the microvasculature, the effects of CD82 on RASF were evaluated with focus on RASF adhesion, migration and proliferation.
Methods
For detailed methods, see online supplementary file 1.
Supplemental material
Tissues and cells
Synovium and cartilage were obtained during knee replacement surgeries (Agaplesion Markus-Hospital Frankfurt/Main). Patients fulfilled the ACR classification criteria.25–27 Non-arthritic synovial fibroblasts (NSF) served as control. The study was approved by the local ethics committee. All patients gave written informed consent. Synovium was snap frozen or digested for cell isolation. SF were cultured in DMEM (GE Healthcare, Germany) containing 10% fetal calf serum (FCS) (Sigma-Aldrich, Germany), 100 U/mL penicillin/streptomycin (Applichem, Germany), 10 mM HEPES (GE Healthcare).28 Proteins for 16 hours stimulation: 50 ng/mL TNFα, 10 ng/mL IL-1β, 2 ng/mL TGF-β (R&D-Systems, Germany), 25 µg/mL adiponectin, 100 ng/mL visfatin (Biovendor, Heidelberg, Germany), 10 ng/mL lipopolysaccheride (LPS) (Sigma-Aldrich).
CD82 overexpression/knockdown
For lentiviral Cd82 transduction, Precision LentiORF individual clone and Precision LentiORF red fluorescent protein *(RFP) were used (MOI=5, GE Healthcare). After 48 hours, blasticidin-S-hydrochloride was added resulting in stable CD82 overexpression (>90%). Mock treated cells served as control.
The Amaxa Nucleofector II (program U23) and Basic Nucleofector Kit (primary mammalian fibroblasts, Lonza, Germany) were used. siRNA: ON-TARGETplus SMART pool CD82 siRNA; non-targeting-pool (NTP) siRNA (PerbioScience, Germany). At day 5–6 after nucleofection, maximum CD82 protein reduction was confirmed by Western blot.
Western blot
Cells were lysed (1% Nonidet P-40, Sigma-Aldrich) with 0.05M Tris-HCl pH6.7 containing complete protease inhibitor mix (Roche). Antibodies: anti-human CD82 (ab59509, Abcam, Cambridge, UK); anti-mouse horseradish peroxidase(HRP)-conjugated secondary (Agilent Technologies). Loading control: Cyclophilin B (Abcam). Band intensity was quantified using the ImageJ software.
Immunohistochemistry/immunofluorescence
Acetone-fixed sections or chamber slides were blocked in 2% BSA. Antibodies: anti-human CD82 (ab109529), isotype control (ab125938, Abcam). Secondary antibodies: biotinylated goat anti-rabbit (550338, BD Biosciences) and HRP-conjugated streptavidin (016-030-084, Dianova, Germany). Peroxidase substrate kit (AEC, Vector-Laboratories, Germany) detection was used.
Fluorescent double-staining: Blocking: FCS, chicken serum and BSA (10%). First antibodies: anti-human Integrin α6 (MAB1350, R&D), αV (MAB-1980, Merck Millipore), β1 (CP26, Merck Millipore), CD82 (PA5-20356, Thermo Fisher). Secondary antibodies: anti-mouse (Alexa-Fluor 488, A-11001), anti-rabbit (Alexa Fluor 546, A11071, Thermo Fisher). Nuclei were DAPI stained (Sigma Aldrich). Staining was evaluated as indicated in the online supplementary method.
SCID-mouse-model
Female immunodeficient Crl-scidBR mice (Charles River, Germany) were kept pathogen-free with water and food ad libitum. Animals underwent inverse-wrap implantation29 with subcutaneous implantation of human RASF together with human cartilage (intact areas from OA cartilage) in a Gelfoam matrix (Pfizer, USA) at the ipsilateral side. Contralaterally, cartilage without RASF was implanted.23 30 Implants removed after 60 days were snap frozen, 5 µm sections H/E-stained and scored.23 29 30
Laser-mediated microdissection (LMM)
Eight µm cryosections on PEN-membrane coated slides (P.A.L.M., Germany) were fixed, nuclei haematoxylin-stained and dehydrated. LMM was performed using a Robot-MicroBeam laser microscope (P.A.L.M. Microlaser Technologies). Dissected areas (3000 cells) from different tissues/implants were collected in RLT-lysis buffer (RNA from different tissues was pooled before isolation).
Real-time PCR
For RNA isolation of LMM samples, the RNeasy Micro Kit, otherwise Mini Kits (Qiagen) were used. RNA was transcribed using AMV reverse transcriptase (Promega, Madison, USA) and random hexamer primers (Roche Diagnostics). Real-time PCR followed by melting curve analysis was performed using a LightCycler (Roche).5 6 Primers: human CD82 forward 5′-ggtgtggatcctggccgacaagagc-3′/reverse 5′-atgcagcccaggaagcccatgagc-3′. Results were normalised to 18S rRNA and analysed using the LightCycler software.31
Migration assay
The lower compartment of a Boyden chamber containing a 8 µm pore membrane (GE Water and Process Technologies, Germany) coated with fibronectin was filled with DMEM containing 10% FCS (as chemoattractant) and RASF (harvested with Accutase) in DMEM with 2% FCS placed on top (three replicates each). After 16 hours, migrated cells on membranes were fixed, haematoxylin-stained and nuclei counted. 6.5 mm polycarbonate filters were used in the transwell assay (Corning, USA) under the same conditions. Migrated cells at the lower filter sides were detached and counted. For statistics, means per biological sample were used.
Cell motility assay
Fibroblasts (100% confluency) on uncoated plates were wounded with micropipette tips (scratch). Supernatants were replaced, cells incubated for 17 hours, then photographed. Gap closure was evaluated in five experimental replicates. The number of nuclei moved into the scratch was counted (mean per biological sample used for statistics).
Adhesion assay
SF detached with Accutase were added to 24-well plates coated with/without Matrigel (BD Biosciences) and incubated for 1 hour. The plate was shaken 5 min full speed thrice. Cells were stained with 0.1% crystal-violet dye in methanol and counted (three wells per population, means used for statistics).
Cell-to-cell binding assay
HUVEC or primary EC from human varicose veins (for details, see online supplementary file 1) were cultured on rat tail collagen-I-coated plates (Life Technologies) until confluency. RASF were Calcein-AM (Thermo Fisher) stained and detached using Accutase. RASF added to confluent EC-layers were incubated 30 min. Plates were shaken 5 min at full speed. Attached fibroblasts were counted (three different wells, mean per population used for statistics).
Proliferation assay
RASF were BrdU (proliferation 5-bromo-2′-deoxyuridine) labelled for 24 hours using a colorimetric BrdU-ELISA assay (Merck Millipore) according to the manufacturer and quantified using a TECAN reader.
Statistics
GraphPad Prism 6.0 was used. Data were analysed parametrically using RM one-way ANOVA with Geissner-Greenhouse correction followed by Tukey’s multiple comparisons test with individual variances computed for each comparison. Figure 2A shows the parametric paired two-tailed t-test. Figure 2B,C were calculated with Mann-Whitney test. Data in figures are shown as box-whisker plots with median, 25th/75th percentile (box) and lowest/highest value (whiskers) per data set. P values less than 0.05 were considered significant (*p<0.05/**p<0.005/***p<0.001/n.s.=not significant).
Results
Cellular and compartmental CD82 localisation
The increased CD82 expression in cultured RASF compared with OASF and in the RA synovial LL and close to blood vessels versus OA synovial tissue has previously been described21 22 and could be confirmed (figure 1) by immunofluorescence showing an increased presence of CD82 also in the invasion zone into cartilage. LMM out of the LL and sublining from n=6 different synovial tissues was performed, RNA from 12 tissues (different patients) was pooled (one pool of 6, two pools of 3 tissues) and subsequent real-time PCR performed. Comparison showed that tissue mRNA expression of CD82 was significantly stronger in the RA LL compared with sublining cells (figure 2A).
The percentage of positive tissues (12 in total) was calculated in LL, outer LL, vessel wall, EC layer, lymphocyte infiltrates and cartilage invasion zone of immunofluorescence stainings. One hundred per cent (12/12) RA tissues were CD82-positive in the outer LL and total LL. Lymphocyte infiltrates were CD82-negative. Several cells showed CD82 signals in sublining and vessel walls (100% of tissues). Cartilage invading cells were CD82-positive (100% of tissues).
αV-integrin was positive in 80% of outer LL, 100% were positive in the deeper LL. Several double stained cells were located in the outer LL (figure 1A). Single sublining cells were αV-positive (100%), but double-stained cells were detectable in only 50% of tissues. αV-positive vessel walls were detectable in 100% of tissues. CD82 and αV-double-positive cells were limited to the vessel wall (figure 1A). Cartilage invading cells showed weaker αV-integrin signals compared with CD82, not all cells being αV-positive.
β1-integrin was present in outer and deeper LL areas of all tissues (100%). In sublining, single cells were β1-positive. CD82/ß1-positive cells were mainly detectable in the outer LL (figure 1B). All tissues displayed β1-integrin positive vessel walls. Deeper areas but not EC showed a CD82/β1-overlay. CD82/β1-positive cartilage invading cells were detectable in all tissues (figure 1B).
α6-integrin was positive in the outer LL (100%) whereas deeper areas were negative or contained single positive cells (figure 1C). An overlay was only observed at the outer LL. In vessels, EC were α6-positive (100%) and double-positive for CD82. At the invasion zone, none of the tissues showed α6 signals.
CD82 induction in RASF by proinflammatory factors
To evaluate the factors responsible for the increased presence of CD82 in RA-synovium, proinflammatory factors increased in RA were used for SF stimulation and CD82 expression was measured. RASF/OASF were stimulated with proinflammatory factors (TNFα, IL-1β, LPS), TGF-β or adipokines (visfatin, adiponectin). All stimuli influenced CD82 expression on RASF (figure 2B). Although cultured SF already produce CD82 on RNA and protein level, CD82 was significantly induced by TNFα, IL-1β and LPS in RASF and OASF (figure 2B,C). Adipokines induced CD82 expression significantly but to a lower extent in both RASF and OASF (figure 2B, C).
For OASF, the baseline expression of CD82 was lower, similar to previously published results.21 22 Therefore, the relative CD82 inducibility was higher in OASF due to the lower CD82 baseline levels. NSF showed comparable results to OASF except for TNF showing weaker CD82-induction (figure 2D). Although TGF-β significantly reduced the CD82 expression in RASF (figure 2A), the grade of regulation was low and not detectable for OASF (figure 2B).
CD82 overexpression in RASF
CD82 expression level in RASF from different patients varied. For lentiviral overexpression, RASF with moderate to intermediate CD82 expression were used. In comparison to mock and RFP-transduced cells, overexpression of CD82 in RASF resulted in an induction of CD82 on protein (figure 3A,B) and RNA level (n=5, figure 3C).
CD82 overexpression: fibroblast motility, migration, adhesion, proliferation
Lentiviral CD82 overexpression significantly reduced RASF migration (Boyden chamber) compared with RFP and to mock (figure 4A). In the cell motility assay, a significant reduction of RASF migration into the gap was observed for CD82-transduced RASF compared with mock (figure 4B). Apoptosis was not altered by CD82 (not shown) and RASF proliferation was non-significantly reduced after CD82 overexpression compared with controls (figure 4C). Cell-to-cell binding of RASF to confluent EC-layers showed reduced RASF-EC-interaction in CD82-overexpressing cells (figure 4D) and adhesion to culture plates was significantly reduced (figure 4E).
Confirmation of CD82-expression of RASF under in vivo conditions
To evaluate whether cultured RASF express similar CD82 levels compared with sites of cartilage invasion in vivo, the SCID-mouse-model was used. Human RASF (one population with intermediate CD82 expression) were implanted together with human cartilage subcutaneously at the ipsilateral site. Contralaterally, human cartilage was implanted without RASF. As shown previously, human RASF invade the coimplanted and contralateral cartilage.23 Using LMM, cells at the implant cartilage invasion zone were isolated (n=6 animals). Pooled RNA was evaluated for CD82 expression compared with cultured RASF used for implantation. The presence of human CD82 at sites of cartilage invasion (online supplementary file 2) was comparable to cultured RASF used for implantation with similar levels of human (not mouse) CD82 mRNA (online supplementary file 2). Of note, human CD82 expression was comparable in the invasion zone of both sites (data not shown).
Supplemental material
CD82 knockdown in RASF
To evaluate whether CD82 downregulation has opposite effects, siRNA-mediated CD82 knockdown was confirmed by Western blot (figure 5A). RASF adhesion was induced by CD82 knockdown in contrast to RASF cell-to-cell binding to EC, which was reduced versus NTP and mock (figure 5B). CD82 knockdown significantly increased RASF migration towards a higher FCS gradient as chemoattractant (figure 5C). Similarly, cell motility assays without chemoattractant showed increased RASF migration compared with NTP or mock (figure 5D).
Discussion
In RA-synovium, CD82 expression was mainly detectable in the hyperplastic LL and vessels.21 22 We observed a CD82 induction in RASF by proinflammatory factors, which could explain the increased CD82 expression in inflamed RA tissue similar to the situation in cancer.2 4 Interestingly, the proinflammatory stimuli were also able to induce CD82 in OASF. However, due to chronic inflammation in RA, the exposure of SF to these factors is increased in RA. Aside of the LL, we found increased CD82 levels in RASF at the cartilage invasion zone, which was also confirmed in the SCID-mouse-model. In contrast, the main localisation of CD82 in the sublining was around vessels. Therefore, CD82 expression in RASF appears dependent on the localisation within the tissue and on local factors.
SF, especially activated fibroblasts in the LL, express increased amounts of different integrins, including the collagen receptor α1β1, the fibronectin receptor α5β1, the laminin receptors α3β1 and α6β1, the vitronectin receptor αvβ3.32 CD82 has been described to attenuate β1-integrin activation in prostate cancer cells33 but also to downregulate the level, activity and signalling of integrins,34 the latter correlating with decreases in focal adhesions, stress fibres34 35 and maturation of β1-integrin.34 In our study, β1-integrin was colocalised with CD82, especially in deeper areas of the LL but also in invading RASF. We observed a reduced migration after CD82-overexpression, likely contributing to RASF accumulation at sites of cartilage invasion, a process that could be increased by inflammation.24 36 CD82 colocalised with αVβ3-integrin on experimental overexpression in ovarian cancer cells.37 The expressions of CD82 and αV-integrin, however, was inversely correlated in cancer4 and EC.10 We observed the colocalisation of CD82 with αV-integrin mainly in single cells of the invasion zone and LL. Association of CD82 with α6-integrin decreased adhesion and migration of prostate cancer cells.38 In our study, α6-integrin was not detectable close to cartilage invasion and colocalisation with CD82 was limited to the LL. The strongest colocalisation with CD82 was present in the endothelium, supporting the role of endothelium in RASF recruitment and mobility. The RASF recruitment and the increased integrin expression at the site of cartilage invasion is well known.39 However, the CD82-induction at sites of cartilage erosion may contribute to hold activated RASF on-site.
CD82-overexpression significantly reduced RASF motility similar to the observations in tumour cells.4 7 Interestingly, cell-matrix adhesion and binding to EC were reduced by CD82 overexpression, suggesting a reduced potential for RASF-attachment to EC in local vessels. The increased expression of CD82 in the vascular wall and its close proximity22 suggests a role of CD82 in vascular activation including synovial neoangiogenesis.40
Due to lower CD82 expression in RASF in the sublining, a CD82 knockdown was performed showing that the migratory capability of RASF was significantly increased. However, cell-to-cell binding of RASF to EC was reduced in contrast to cell-matrix adhesion, which was increased in contrast to CD82 overexpression. Therefore, RASF in the sublining expressing less CD82 may be able to migrate but their capacity to attach to EC is not increased. Potentially, at sites of inflammation and matrix destruction CD82-induction takes place, contributing to RASF accumulation. A similar observation has been made for podoplanin (PDPN) and CD248 with a high expression of PDPN in the LL, whereas CD248 expression was restricted to sublining cells.41 In this study, TNF or IL-1β also increased PDPN expression whereas TGF-ß induced CD248. Local RASF-proliferation may also contribute to increased RASF numbers at the site of erosion. In our study, CD82-overexpression had only limited effects on RASF-proliferation.
CD82 appears to play a role in cell adhesion and motility in the inflamed rheumatoid synovium. Our data show that CD82 is induced by inflammation, both in RASF close to vessels and at sites of cartilage damage. On cellular level, this may be mediated by interaction of CD82 with surface integrins, expressed at the sites of cartilage invasion and LL. Reduced CD82 in the sublining may facilitate RASF migration to sites of inflammation and tissue damage.
Acknowledgments
We would like to thank Dirk Schröder for technical contributions as well as Simone Benninghoff, Carina Schreiyäck. We thank Mona Arnold for her help during the revision.
References
Footnotes
Handling editor Josef S Smolen
Contributors Project conception and design: EN, MSc, UM-L. Acquisition of data: MSc, RH, M-LH, MSa, SR. Analysis and interpretation of data: EN, MSc, M-LH, SC. Drafting and/or revision of article: EN, MSc, M-LH, UM-L.
Funding This project was funded by the Kerckhoff-Stiftung, the DFG excellence cluster for cardiopulmonary research (ECCPS) and the BMBF (IMPAM, project 01EC1008G).
Competing interests None declared.
Patient consent Not required.
Ethics approval Local ethics committee.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement There are no shared additional unpublished data from the study available.