Constitutive Activation of Integrin α9 Augments Self-Directed Hyperplastic and Proinflammatory Properties of Fibroblast-like Synoviocytes of Rheumatoid Arthritis

Abstract

Despite advances in the treatment of rheumatoid arthritis (RA), currently approved medications can have significant side effects due to their direct immunosuppressive activities. Additionally, current therapies do not address residual synovial inflammation. In this study, we evaluated the role of integrin α9 and its ligand, tenascin-C (Tn-C), on the proliferative and inflammatory response of fibroblast-like synoviocytes (FLSs) from RA patients grown in three-dimensional (3D)-micromass culture. FLSs from osteoarthritis patients, when grown in the 3D-culture system, formed self-directed lining-like structures, whereas FLSs from RA tissues (RA-FLSs) developed an abnormal structure of condensed cellular accumulation reflective of the pathogenic features of RA synovial tissues. Additionally, RA-FLSs grown in 3D culture showed autonomous production of proinflammatory mediators. Predominant expression of α9 and Tn-C was observed in the condensed lining, and knockdown of these molecules abrogated the abnormal lining-like structure formation and suppressed the spontaneous expression of matrix metalloproteinases, IL-6, TNFSF11/RANKL, and cadherin-11. Disruption of α9 also inhibited expression of Tn-C, suggesting existence of a positive feedback loop in which the engagement of α9 with Tn-C self-amplifies its own signaling and promotes progression of synovial hyperplasia. Depletion of α9 also suppressed the platelet-derived growth factor-induced hyperplastic response of RA-FLSs and blunted the TNF-α-induced expression of matrix metalloproteinases and IL-6. Finally, α9-blocking Ab also suppressed the formation of the condensed cellular lining by RA-FLSs in 3D cultures in a concentration-related manner. This study demonstrates the central role of α9 in pathogenic behaviors of RA-FLSs and highlights the potential of α9-blocking agents as a nonimmunosuppressive treatment for RA-associated synovitis.

FIGURE 1.

Aggressive behavior of RA-FLSs reproduced in the 3D-micromass system. (A) Typical microscopic images of arthritic synovial lining layer and in vitro established lining-like structure. Frozen sections of synovial tissues from OA and RA (upper), or 3D-cultured micromass architecture formed by OA- and RA-FLSs (lower), were stained with DAPI. Scale bars, 100 μm. (B) Section from 3D-culturted micromass architecture formed by RA-FLSs costained with Alexa Fluor 488–phalloidin (left) and DAPI (right). Scale bars, 200 μm. (C) Comparison of thickness of lining-like structure established by OA- and RA-FLSs. Frozen sections of micromass architectures formed by OA- (n = 5) and RA-FLSs (n = 5) were stained by Alexa Fluor 488–phalloidin. Thickness of the structure was calculated by image analysis and expressed as mean ± SE. (D) Analysis of proinflammatory gene expression in OA- (n = 5) and RA-FLSs (n = 5) cultured in either monolayer (2D) or 3D-culture system (3D). Relative expression level of indicated genes was determined by real-time PCR and expressed as mean ± SE. Statistical analyses were performed by a Student t test (C and D). *p < 0.05, **p < 0.01. ns, not significant.

FIGURE 2.

Involvement of FAK activation in the autonomous expression of proinflammatory genes in 3D-cultured RA-FLSs. (A) Fluorescent microscopic analyses of FAK phosphorylation status in RA and OA synovial tissues. Frozen sections from RA (upper) and OA synovial tissues (lower) were stained with isotype-matched control Ab (iso), anti-phosphorylated FAK Ab with Alexa Fluor 488–labeled secondary Ab (pFAK), and DAPI. The slides were photographed under fluorescent microscopy with the same exposure time. Representative images from three independent patients’ tissues are shown. Scale bars, 200 μm. (B) Western blot analysis of phosphorylated status of FAK in 3D-cultured RA-FLSs. Cell lysates prepared from plate-cultured (2D) or 3D-cultured (3D) RA-FLSs were analyzed in Western blotting with anti-pFAK (upper) or anti-total FAK (lower) Abs. (C) Real-time PCR analysis of proinflammatory gene expression in 3D-cultured RA-FLSs treated with FAKi (10 μM PF-573228, n = 5) or the control (0.01% DMSO, n = 5). The data were expressed as mean ± SE. Statistical analyses were performed by a paired t test. ns, not significant.

FIGURE 3.

Involvement of α9 and Tn-C in the self-directed abnormal behavior of RA-FLSs. (A) Gene expression levels of ITGA9 (left) and TNC (middle) in OA- (n = 5) and RA-FLSs (n = 5) were determined by real-time PCR and expressed as mean ± SE. Distributions of α9 and Tn-C in 3D-cultured RA-FLSs and RA synovial tissue was determined by costaining with Alexa Fluor 594–labeled Ab to α9 (red) and Ab to Tn-C with Alexa Fluor 488–labeled secondary Ab (green) (right). Scale bars, 100 μm. (B–E) RA-FLSs treated with shRNA either for ITGA9, TNC, or the control (Con) cultured in 3D micromass were analyzed the phenotypes. (B) Sections from the resultant architectures were costained with Alexa Fluor 488–phalloidin (green) and DAPI (blue). Representative data from three independent RA-FLSs are shown. Scale bars, 50 μm. (C) Cell lysates from 3D-cultured RA-FLSs were analyzed by Western blotting with Abs to pFAK (upper) and total FAK (lower). Representative data from five independent blots are shown. (D) Protein amounts of MMP-1, MMP-3, and IL-6 in the culture supernatants were determined by ELISA and indicated as mean ± SE (n = 5). (E) Western blotting of the cell lysates with Abs to MMP-14 (left), TNFSF11 (right) and GAPDH. A representative data from five independent blots was shown. Statistical analyses were performed by a Student t test (A) and a Dunnett test using within-subject error (D). *p < 0.05, **p < 0.01. ns, not significant.

FIGURE 4.

Expression levels of Tn-C and cadherin-11 in 3D-cultured RA-FLS treated with shRNA for ITGA9, TNC, or FAKi. (A) Gene and protein expression levels of Tn-C in 3D-cultured RA-FLSs treated with shRNA for ITGA9 were determined by real-time PCR (n = 5, left) and ELISA with the culture supernatants (n = 5, right), respectively. (B) Gene and protein expression levels of cadherin-11 in 3D-cultured RA-FLSs treated with the indicated shRNA were determined by real-time PCR (n = 5, left) and Western blotting with Abs to cadherin-11 and GAPDH (right), respectively. Representative data from five independent blots are shown. (C) Gene expression levels of TNC (left) and CDH11 (right) in 3D-cultured RA-FLSs treated with or without FAKi were determined by real-time PCR with total RNA prepared in Fig. 2C (n = 5) and expressed as mean ± SE (n = 5). Statistical analyses were performed by a paired t test (A and C) and a Dunnett test using within-subject error (B). *p < 0.05, **p < 0.01. ns, not significant.

FIGURE 5.

Phenotypes of RA-FLSs with depletion of α9 cultured in 3D-micromass system under stimulation with PDGF or TNF-α. RA-FLSs treated with shRNA either for ITGA9 or the control (Con) were cultured in a 3D-micromass system in the media supplemented with PDGF (PDGF-BB, 50 ng/ml) or TNF-α (20 ng/ml) and the phenotypes were analyzed. (A) Frozen sections from the resultant architectures were costained with Alexa Fluor 488–phalloidin (green) and DAPI (blue). Representative data from three independent experiments are shown. Scale bars, 50 μm. (B) Protein amounts of MMP-1, MMP-3, and IL-6 in the culture supernatants collected at the end of the study were determined by ELISA and expressed as mean ± SE (n = 5). Statistical analyses were performed by a paired t test. *^/#p < 0.05, **^/##p < 0.01. ns, not significant.

FIGURE 6.

Effects of pharmacological blockade of α9 in development of synovial hyperplasia in CAIA model. (A) Changes in the arthritis scores of each group. Data indicate mean ± SE. ●, control; □, MA9-413 (10 mg/kg); ○, MA9-413 (30 mg/kg). (B) Pathological analysis of the inflamed joints. Sections from the left hindlimbs were stained with H&E. The black boxes indicate the close-up images of synovial lining (red arrows). Scale bars, 500 μm (upper). Severity of synovial hyperplasia and bone resorption was scored (0–4) and expressed as dotted plot. ○, individual scores; bars, median. Statistical analyses were performed by Wilcoxon rank sum test (NT; normal versus control groups) and a Steel multiple comparison test (control versus MA9-413-treated groups). *p < 0.05, ^##/**p < 0.01. (C and E) Gene expression analysis of the inflamed joints. Total RNA from right hindlimbs of NT, control, and MA9-413–treated groups was analyzed by real-time PCR. The data indicate mean ± SE. Statistical analyses were performed by a Student t test (NT versus control groups) and a Dunnett test (control versus MA9-413–treated groups). ^##/**p < 0.01. ns, not significant. (D) Immunohistostaining of the CAIA joints. Sections from the left hindlimbs were stained with Ab to pFAK. Representative data from the indicated groups are shown. Scale bars, 500 μm.

FIGURE 7.

Pharmacological profile of ASP5094, a blocking Ab for α9. (A) Inhibition of adhesion of α9-expressing SW480 cells to plate coated with Tn-C peptide by ASP5094. A representative fitting curve from three independent experiments is shown. Each plot represents mean of quadruplicate assays with SD. (B–D) RA-FLSs were pretreated with ASP5094, etanercept, or natalizumab at indicated concentrations and 3D cultured in media containing the corresponding reagents. An equal volume of PBS was used as a control. (B) Typical images of the lining-like structure formed by RA-FLSs treated with PBS (left) or ASP5094 (10 μg/ml, right) stained with DAPI. Scale bar, 100 μm. (C and D) The thickness of the lining-like structure of the resultant architectures was calculated by image analyses and expressed as mean ± SE. (B), n = 9; (C), n = 5. Statistical analyses were performed by a Dunnett test using within-subject error. *p < 0.05, **p < 0.01. ns, not significant.