CD34hi subset of synovial fibroblasts contributes to fibrotic phenotype of human knee osteoarthritis

Abstract

Osteoarthritis (OA) shows various clinical manifestations depending on the status of its joint components. We aimed to identify the synovial cell subsets responsible for OA pathophysiology by comprehensive analyses of human synovium samples in single-cell resolution. Two distinct OA synovial tissue groups were classified by gene expression profiles in RNA-Seq: inflammatory and fibrotic. The inflammatory group exhibited high expression of inflammatory cytokines, histologically inflammatory infiltrate, and a more severe pain score. The fibrotic group showed higher expression of fibroblast growth factor (FGFs) and bone morphogenetic proteins (BMPs), showed histologically perivascular fibrosis, and showed a lower pain score. In single-cell RNA-Seq (scRNA-Seq) of synovial cells, MERTKloCD206lo macrophages and CD34hi fibroblasts were associated with the inflammatory and fibrotic groups, respectively. Among the 3 fibroblast subsets, CD34loTHY1lo and CD34loTHY1hi fibroblasts were influenced by synovial immune cells, whereas CD34hi fibroblasts were influenced by mural and endothelial cells. Particularly, in CD34hi fibroblast subsets, CD34hiCD70hi fibroblasts promoted proliferation of Tregs, potentially suppressing synovitis and protecting articular cartilage. Elucidation of the mechanisms underlying the regulation of these synovial cell subsets may lead to novel strategies for OA therapeutics.

Keywords: Bone biology; Inflammation; Osteoarthritis.

Figure 1. Classification of human knee OA synovium by RNA-Seq and histological evaluation.

(A) Hierarchical clustering of 41 synovium samples from 36 TKA and 5 arthroscopy patients. (B) GO analysis of the 2 groups of OA synovium. (C) DEGs of the 2 groups of OA synovium, classified as cytokines, chemokines, and growth factors. (D) H&E staining of synovial tissues. Representative images are shown. The right lower panel indicates the percentage of specimens with inflammatory infiltrates in the 2 types of OA synovium. Infla, inflammatory type; Fibro, fibrotic type. Infla, n = 21; Fibro, n = 15. Fisher’s exact test. *P < 0.05. Scale bar: 100 μm. (E) Masson’s trichrome staining of synovial tissues. Representative images are shown. The right lower panel indicates the percentage of positive areas in the 2 types of OA synovium. Infla, n = 21; Fibro, n = 15. Mann-Whitney U test (***P < 0.001). All data were expressed as dot plots and mean ± SEM. Scale bar: 100 μm.

Figure 2. Comparison of clinical scores between patients with knee OA with different types of synovium.

(A) Comparison of KOOS total scores and 6 subscale scores, including Symptom, Pain, ADL, Sports, and QOL, between patients with inflammatory and fibrotic groups of synovium. Infla, inflammatory group; Fibro, fibrotic group. Infla, n = 21; Fibro, n = 15. Mann-Whitney U test (*P < 0.05). (B) Subclustering of the 2 synovium types. (C) GO analysis of the 2 synovium subtypes within the inflammatory or fibrotic types. (D) KOOS between patients with 4 subtypes of OA synovium. Infla_1, n = 15; Infla_2, n = 6; Fibro_1, n = 8; Fibro_2, n = 7. Mann-Whitney U test. All data were expressed as dot plots and mean ± SEM.

Figure 3. Single cell–based transcriptional profiling of synovial cell subsets.

(A) UMAP of 11 synovial cell clusters identified by scRNA-Seq. (B) UMAP of 11 synovial cell clusters, split by each sample. (C) Percentage of immune cell and fibroblast subsets in the 4 synovium samples. Infla, inflammatory group; Fibro, fibrotic group. (D) Dot plots of representative cytokine and chemokine genes in MERTK^loCD206^lo and MERTK^hiCD206^hi macrophages. (E) Dot plots of representative genes in CD34^loTHY1^lo, CD34^loTHY1^hi, and CD34^hi fibroblasts. (F) Dot plots of stable state marker genes in CD34^loTHY1^lo, CD34^loTHY1^hi, and CD34^hi fibroblasts. (G) Dot plots of BMPs and WNT inhibitor genes in CD34^loTHY1^lo, CD34^loTHY1^hi, and CD34^hi fibroblasts. (H) GO analysis of 3 fibroblast subsets.

Figure 4. Estimated interactions between synovial cell subsets when three fibroblast subsets are set as the receptors.

Ligand-receptor analyses setting each of 3 fibroblast subsets (CD34^loTHY1^lo, CD34^loTHY1^hi, and CD34^hi fibroblasts) as recipients. The top ligands observed in at least 2 clusters were described as General.

Figure 5. Estimated interactions between synovial cell subsets when inflammatory immune cell subsets are set as the receptors.

Ligand-receptor analyses setting each of 3 immune cell subsets (MERTK^loCD206^lo macrophages, DCs, and T cells) as recipients. The top ligands observed in at least 2 clusters were described as General.

Figure 6. Estimated interactions between synovial cell subsets and articular chondrocytes using the merged datasets.

(A) UMAP of 18 synovial cell and chondrocyte clusters identified by scRNA-Seq. HomC, homeostatic chondrocyte; preHTC, prehypertrophic chondrocyte; HTC, hypertrophic chondrocyte; RepC, reparative chondrocyte; RegC, regulatory chondrocyte; preFC, prefibrotic chondrocyte; FC, fibrotic chondrocyte. (B) Dot plots of representative marker genes in 7 chondrocyte clusters. (C–E) Ligand-receptor analyses between each of 3 representative chondrocyte subsets and all synovial cell and chondrocyte subsets. Each of the 3 chondrocyte subsets — i.e., homeostatic (C), hypertrophic (D), and fibrotic (E) — was set as the recipient. Left: heatmaps of ligand-receptor interaction potential. Right: dot plots of the top 20 ligand gene expression profiles.

Figure 7. The expression profile of CD70 in OA synovium, especially fibroblast.

(A) Dot plots of representative DEGs from fibrotic group synovium (Figure 1C) in 11 synovial cell subsets (Figure 3). (B) Expression levels of CD70 in 2 OA synovium groups, determined by knee OA synovium RNA-Seq. All data are expressed as dot plots and mean ± SEM. Mann-Whitney U test. *P < 0.05. (C) Expression levels of CD70 in 4 OA synovium subgroups, determined by knee OA synovium RNA-Seq. All data are expressed as dot plots and mean ± SEM. Kruskal-Wallis test. *P < 0.05. (D) UMAP of 2 CD34^hi fibroblast clusters identified by scRNA-Seq. (E) Percentage of fibroblast subsets in the 4 synovium samples. Infla, inflammatory group; Fibro, fibrotic group. (F) Dot plots of stable state marker genes and CD70 in CD34^hiCD70^lo fibroblasts and CD34^hiCD70^hi fibroblasts. (G) Feature plots of stable state marker genes and CD70 in CD34^hiCD70^lo fibroblasts and CD34^hiCD70^hi fibroblasts. (H) Immunofluorescence images of Podoplanin (PDPN), CD34, and CD70 in human knee OA synovium tissue around vessels. Arrowheads indicate PDPN⁺CD34⁺CD70⁺ cells. Scale bar: 100 μm.

Figure 8. CD70 is crucial for T cell proliferation and contributes to suppression of synovitis by Tregs.

(A) T cell proliferation assay cocultured with CD70 transcript variant 1 or 2 or with GFP-transfected fibroblasts cell line (MRC-5). (B) The percentage of CD45⁺ cells and CD4⁺ T cells in live cells cocultured with CD70 transcript variant 1 or 2 or with GFP-transfected MRC-5 cells. (C) Graphical schema of OA synovial and cartilage tissues coculture experiments. (D) Time-course mRNA levels of IL1B, IL6, and TNF in cocultured synovial tissues and COL2A1, ACAN, and SOX9 mRNA levels in cocultured cartilage tissues with anti-CD70 antibody or isotype control. n = 6 biologically independent experiments. (E) The percentage of CD45⁺ cells and T cell subsets in cocultured synovial tissues with anti-CD70 antibody or isotype control. n = 10 biologically independent experiments. (F) The percentage of IL-1B⁺ DCs, macrophages (Mφ), and T cells in cocultured synovial tissues with anti-CD70 antibody or isotype control. n = 6 biologically independent experiments. Data are expressed as dot plots and mean ± SEM. Mann-Whitney U test. *P < 0.05, **P < 0.01.