Inflammatory Fibroblast-Like Synoviocyte-Derived Exosomes Aggravate Osteoarthritis via Enhancing Macrophage Glycolysis

Abstract

The severity of osteoarthritis (OA) and cartilage degeneration is highly associated with synovial inflammation. Although recent investigations have revealed a dysregulated crosstalk between fibroblast-like synoviocytes (FLSs) and macrophages in the pathogenesis of synovitis, limited knowledge is available regarding the involvement of exosomes. Here, increased exosome secretion is observed in FLSs from OA patients. Notably, internalization of inflammatory FLS-derived exosomes (inf-exo) can enhance the M1 polarization of macrophages, which further induces an OA-like phenotype in co-cultured chondrocytes. Intra-articular injection of inf-exo induces synovitis and exacerbates OA progression in murine models. In addition, it is demonstrated that inf-exo stimulation triggers the activation of glycolysis. Inhibition of glycolysis using 2-DG successfully attenuates excessive M1 polarization triggered by inf-exo. Mechanistically, HIF1A is identified as the determinant transcription factor, inhibition of which, both pharmacologically or genetically, relieves macrophage inflammation triggered by inf-exo-induced hyperglycolysis. Furthermore, in vivo administration of an HIF1A inhibitor alleviates experimental OA. The results provide novel insights into the involvement of FLS-derived exosomes in OA pathogenesis, suggesting that inf-exo-induced macrophage dysfunction represents an attractive target for OA therapy.

Keywords: HIF1A; exosomes; fibroblast‐like synoviocytes; glycolysis; macrophages; osteoarthritis; synovitis.

Figure 1

Exosome secretion from fibroblast‐like synoviocytes (FLSs) is elevated under inflammation stimulation. a) Representative images of HE staining of synovial specimens (synovium marked as S on the images) from patients with and without OA (top), and mice subjected to sham or DMM operation (bottom) after 8 weeks. Scale bar, 100 µm. b) Western blot analysis of RAB27A in synovial tissues from patients with or without OA, using β‐actin as loading control. c) Quantification of (b) using ImageJ. n = 6 per group. d) Gene expression of Rab27a reanalyzed using publicly available RNA‐seq results of GSE89408 from the Gene Expression Omnibus (GEO) database, including healthy (n = 28) and OA (n = 22) synovium biopsies. e) Representative images of immunohistochemistry staining of RAB27A of synovial specimens from patients with and without OA (top), and mice subjected to sham or DMM operation (bottom) after 8 weeks. Boxed area is enlarged in the picture corner. Scale bar, 50 µm. f) Pearson's correlation analysis of Rab27a expression with OA‐related matrix degradation enzymes (Mmp13 and Adamts5) and inflammatory genes (Il1b, Tnf, Ptgs2) based on the RNA‐seq results of GSE89408 from the GEO database. g) Representative images of immunofluorescence co‐staining of FLS marker vimentin (green) and RAB27A (red) for synovium 8 weeks after sham or DMM operation. Scale bar, 10 µm. h) Immunofluorescence staining of RAB27A in mouse FLSs stimulated by interleukin 1‐beta (IL‐1β) for 48 h. Scale bar, 50 µm. i) and j) Western blot analysis and quantification of RAB27A protein in mouse FLSs stimulated by IL‐1β for 48 h. k) RT‐qPCR analysis of messenger RNA levels for Rab27a in mouse FLSs after being stimulated by IL‐1β for 24 h. l) Experiment design diagram of exosome isolation procedures by sequential centrifugation from cell culture supernatant of FLSs stimulated with or without IL‐1β, referred to as inf‐exo or ctr‐exo respectively later. m) Representative images of exosomes by transmission electron microscopy. Scale, 100 nm. n) Quantification of the exosomes derived from control or inflammatory FLS culture supernatant by nanoparticle tracking analysis (NTA). Data presented as mean ± SD. ** P < 0.01, * P < 0.05, versus the indicated groups, Student's t‐test.

Figure 2

Exosomes from inflammatory FLSs promote macrophage classical activation in vitro and in vivo. a) Experiment design diagram of co‐culturing FLSs with RAW264.7 macrophages free of direct contact using Transwell apparatus. FLSs were treated with or without IL‐1β (10 ng mL⁻¹) and GW4869 (20 µM) before co‐culture. b) and c) Western blot analysis and quantification of protein expression of iNOS, CD86, CD163, and CD206 in macrophages after co‐culturing with FLSs as indicated in (a). d) Schematic diagram for stimulating macrophages with ctr‐exo and inf‐exo in e) to f). e) Representative immunofluorescence images showing the uptake of DiI‐labelled exosomes (red) by DiO‐labelled macrophages (green) after incubation for 2 h. Scale bar, 20 µm. (f) Representative immunofluorescence staining of macrophage marker F4/80 (green) and M1 polarization marker iNOS (red) in macrophages stimulated by ctr‐exo and inf‐exo separately for 48 h. Scale bar, 50 µm. g) and h) Western blot analysis and quantification of protein expression of iNOS, CD86, NLRP3, CD163, and CD206 in RAW264.7 macrophages stimulated by ctr‐exo and inf‐exo separately. i) and j) Western blot analysis and quantification of protein expression of iNOS, CD86, and NLRP3 in THP‐1 macrophages stimulated by ctr‐exo and inf‐exo separately. k) Heatmap of selected chemokines and interleukins from the RNA‐seq results normalized using Z‐score. l) Expression of macrophage classical activation markers (Nos2, Cd86, Il1b, Il6, Ptgs2) in macrophages stimulated by ctr‐exo and inf‐exo from RNA‐seq results. m) RT‐qPCR for messenger RNA of inflammatory genes (Nos2, Cd86, Il1b, Il6, Ptgs2) in macrophages stimulated by ctr‐exo and inf‐exo. n) Diagram for intra‐articular injection of exosomes, the red rectangle indicating the synovial region for analysis. o) to r) Representative synovium images of hematoxylin and eosin (H.E.) staining, immunofluorescence staining for iNOS, and immunohistochemistry staining for IL‐1β and TNF‐α receiving injection of 20 µg exosomes suspended in 10 µL saline twice per week for 4 weeks. Scale bar, 50 µm. Data presented as mean ± SD. ** P < 0.01, * P < 0.05, ns, not significant, versus the indicated groups, Student's t‐test.

Figure 3

Inf‐exo‐primed macrophages promotes catabolism and apoptosis in chondrocytes. a) ELISA analysis of IL‐1β, IL‐6, and TNF‐α in macrophage culture supernatant after stimulated by inf‐exo or ctr‐exo. b) Experiment design diagram of co‐culturing mouse chondrocytes (ATDC5) with macrophages (RAW264.7) free of direct contact using the Transwell apparatus. Macrophages were stimulated by inf‐exo or ctr‐exo 24 h prior to co‐culture. c) Representative images of Alcian blue staining of chondrocytes after co‐culturing with exosome‐primed macrophages. The colored circles around the wells indicate the group for d) to f). d) Representative immunofluorescence images of COL2A1 and MMP13 in chondrocytes co‐cultured with exosome‐primed macrophages. Scale bar, 40 µm. e) and (f) Representative images of Live‐dead staining and TUNEL staining of chondrocytes after co‐culturing with exosome‐primed macrophages. Scale bar, 200 µm. g) Quantification of (d) to (f) using ImageJ software for three independent experiments. h) to k) Western blot analysis and quantification of ACAN, COL2A1, MMP3, MMP13, Cyclin D1, BAX, BCL2, and cleaved caspase‐3 in chondrocytes after co‐culturing with exosome‐primed macrophages. l) and m) RT‐qPCR for messenger RNA of Mmp3, Mmp13, Il1b, Tnf, Ptgs2, Bax, Bcl2, and Ccnd1 in chondrocytes after co‐culture. Data presented as mean ± SD. ** P < 0.01, * P < 0.05, ns, not significant, versus the indicated groups, Student's t‐test.

Figure 4

Intra‐articular injection of inf‐exo aggravates experimental osteoarthritis. a) Experiment diagram of intra‐articular injection of 20 µg inf‐exo or ctr‐exo resuspended in 10 µL PBS after DMM operation, twice a week for 4 and 8 weeks. b) and c) Quantitative analysis of Osteoarthritis Research Society International (OARSI) scale and synovitis score of mice described in (a). n = 6 per group. * P<0.05, student's t‐test. d) and e) Representative images of Safranin O‐fast green staining of joint cartilage area and H.E. staining of joint synovial membranes (n = 6). Scale bar, 100 µm. f) Representative images of immunohistochemistry staining of COL2A1 for joint cartilage (n = 6). Scale bar, 100 µm. g) Representative images of immunofluorescence staining of MMP13 in cartilage and synovium region (n = 6). Scale bar, 100 µm. h) Representative images and intensity quantification of immunofluorescence staining of F4/80 and iNOS in synovium region (n = 6). Scale bar, 100 µm. i) Quantification of intensity of IHC and IF staining in (f) and (g) using ImageJ. Data presented as mean ± SD. ** P < 0.01, * P < 0.05, ns, not significant, versus the indicated groups, Student's t‐test.

Figure 5

Inf‐exo‐triggered hyperglycolysis promotes macrophage inflammation. a) and b) Western blot analysis and quantification of GLUT1, HK2, PKM2, and LDHA in macrophages stimulated by ctr‐exo and inf‐exo, repeated in three independent experiments. c) and d) Representative images of immunohistochemistry staining for GLUT1 and HK2 in human (top panel) and mouse (bottom panel) synovium. Scale bar, 50 µm. e) and f) The extracellular acidification rate (ECAR) was measured using a XF96 Seahorse Analyzer, of which the glycolysis parameters were quantified. g) Heatmap of differentially expressed metabolites from the targeted metabolomics, fulfilling the criteria |log2 (fold change)| ≥ 1 and P < 0.05 (n = 3 each group). h) Relative concentration of glycolytic intermediates detected in targeted metabolomics using LC‐MS. Data shown in the histogram was normalized to those in macrophages stimulated by ctr‐exo. i) and j) Glucose consumption and lactate production in macrophage culture supernatants after exosome stimulation (n = 3 each group). k) and l) Western blot analysis and quantification of iNOS, CD86, and NLRP3 in macrophages stimulated by ctr‐exo and inf‐exo and glycolysis blockage using 2‐DG, repeated in three independent experiments. m) RT‐qPCR for messenger RNA of Ptgs2, Il1b, and Tnf in in macrophages stimulated by ctr‐exo and inf‐exo and blockage of glycolysis by 2‐DG (n = 3 each group). Data presented as mean ± SD. ** P < 0.01, * P < 0.05, versus the indicated groups, Student's t‐test.

Figure 6

HIF1A is the key transcription factor which mediates inf‐exo‐triggered hyperglycolysis. a) and b) Heatmap of glycolytic gene messenger RNA levels (Slc2a1, Hk2, Pkm, and Ldha) in RAW264.7 stimulated by ctr‐exo and inf‐exo for 6 and 12 hours. c) Venn diagram of the transcription factor for four glycolytic genes predicted by ChIP base 3.0 database. d) Histogram diagram for predicted transcription factor ranking using ChEA3 database. e) and f) Representative agarose gel electrophoresis images and quantification for PCR products of ChIP assay of three independent experiments. The RAW264.7 cells as indicated were immunoprecipitated with the antibody against isotype‐matched immunoglobulin or HIF1A respectively 6 h after exosome stimulation. Then, the genomic DNA input and the antibody‐bound DNA fragments were amplified using PCR with primers covering the predicted sites on promoters of the four glycolytic enzymes. g) Representative HIF1A immunofluorescence staining of mice joint subjected to sham or DMM operation. The synovium region is boxed and labeled as S. Scale bar, 100 µm. h) Representative HIF1A immunofluorescence staining of clinical synovium specimens from patients with or without OA. Scale bar, 50 µm. i) Representative images of immunocytochemical staining for HIF1A of RAW264.7 cells stimulated by ctr‐exo and inf‐exo for 24 h. Scale bar, 20 µm. j) and k) Western blot analysis and quantification of HIF1A in synovial tissues from patients with or without OA, using β‐actin as loading control. n = 6 per group. l) and m) Western blot analysis and quantification of HIF1A in RAW264.7 macrophages stimulated by inf‐exo and ctr‐exo for 24 h. n) and o) ECAR was measured using a XF96 Seahorse Analyzer after exosome stimulation and HIF1A inhibition, of which the glycolysis parameters were quantified. p) and q) Western blot analysis of HIF1A, M1 polarization‐related markers (iNOS, CD86, and NLRP3), and glycolytic enzymes (GLUT1, HK2, PKM2, and LDHA) in macrophages stimulated by ctr‐exo and inf‐exo following genetical or pharmacological blockage of HIF1A. β‐actin was used as loading control. r) and s) Heatmap demonstration of macrophage inflammation‐related genes (Il1b, Il6, Tnf, Ptgs2, Cd86, and Nos2) in macrophages aforementioned. t) Glucose consumption and lactate production in cell culture supernatant after si‐RNA or PX‐478‐mediated blockage of HIF1A. Data presented as mean ± SD. ** P < 0.01, * P < 0.05, versus the indicated groups, Student's t‐test.

Figure 7

Pharmacological inhibition of HIF1A blunts inf‐exo‐triggered pathological role of macrophages. a) Scheme of the co‐culture apparatus for macrophages (RAW264.7) and chondrocytes (ATDC5) and the treatment for macrophages before co‐culture. Macrophages were stimulated by ctr‐exo or inf‐exo, with or without PX‐478 treatment. b) and c) Western blot analysis and quantification of OA‐related markers (MMP3, MMP13, ACAN, and COL2A1) and proliferation‐related markers (cyclin D1, BAX, BCL‐2, cleaved caspase‐3) in whole cell lysates of ATDC5 chondrocytes after co‐culture with RAW264.7 macrophages receiving corresponding pretreatment. Color circles indicate the groups assigned in (a) to (e). d) Representative images of live‐dead staining and TUNEL staining for ATDC5 cells after co‐culture with RAW264.7 cells. Scale bar, 50 µm. e) Quantitative analysis of live‐dead staining and TUNEL staining using ImageJ software for three independent experiments. f) Scheme for in vivo study of the effect of PX‐478, the selective HIF1A inhibitor, on murine OA. n = 6 each group. g) and h) Synovitis and OARSI score for joints at 8 weeks after sham or DMM operation. i) and j) Quantification of osteophyte numbers and subchondral bone thickness. k) Heatmap for the quantification results of the IF and IHC staining of synovium and cartilage, representing as Z‐score of the average intensity measured using ImageJ software (n = 6 each group). l) and m) Representative synovial images of H.E. staining, IF staining of HIF1A and iNOS, and IHC staining of TNF‐α. Scale bar, 100 µm. n) and o) Representative cartilage images of H.E., T.B., and S.O. staining, and IHC staining for COL2A1 and MMP13. Scale bar, 100 µm. Data presented as mean ± SD. ** P < 0.01, * P < 0.05, versus the indicated groups, Student's t‐test.

Figure 8

The scheme for exosomes derived from inflammatory fibroblast‐like synoviocytes which exacerbate osteoarthritis through promoting macrophage M1 polarization.