Synovial cell cross-talk with cartilage plays a major role in the pathogenesis of osteoarthritis

Abstract

We elucidated the molecular cross-talk between cartilage and synovium in osteoarthritis, the most widespread arthritis in the world, using the powerful tool of single-cell RNA-sequencing. Multiple cell types were identified based on profiling of 10,640 synoviocytes and 26,192 chondrocytes: 12 distinct synovial cell types and 7 distinct articular chondrocyte phenotypes from matched tissues. Intact cartilage was enriched for homeostatic and hypertrophic chondrocytes, while damaged cartilage was enriched for prefibro- and fibro-, regulatory, reparative and prehypertrophic chondrocytes. A total of 61 cytokines and growth factors were predicted to regulate the 7 chondrocyte cell phenotypes. Based on production by > 1% of cells, 55% of the cytokines were produced by synovial cells (39% exclusive to synoviocytes and not expressed by chondrocytes) and their presence in osteoarthritic synovial fluid confirmed. The synoviocytes producing IL-1beta (a classic pathogenic cytokine in osteoarthritis), mainly inflammatory macrophages and dendritic cells, were characterized by co-expression of surface proteins corresponding to HLA-DQA1, HLA-DQA2, OLR1 or TLR2. Strategies to deplete these pathogenic intra-articular cell subpopulations could be a therapeutic option for human osteoarthritis.

Figure 1

Single-cell RNA-Seq of human OA synoviocytes. (a) Flowchart shows the experimental strategy for systematically identifying cell diversity of synovium and cartilage in the pathogenesis of knee OA. (b) uniform manifold approximation and projection (UMAP) plot of scRNA-seq show unsupervised clusters colored according to putative cell types among a total of 10,640 cells in OA synovia. 44.1%, 33.2%, 12.82%, 3.63%, 3.28%, 1.35%, 1.13%, 0.49% of total acquired cells were synovial subintimal fibroblasts (SSF), synovial intimal fibroblasts (SIF), HLA-DRA⁺ cells (including immune regulatory [IR-MΦ] and inflammatory macrophages [I-MΦ], dendritic cells [DC], activated pro-inflammatory (HLA-DRA⁺) fibroblasts [iFIB] and B cells), endothelial cells (EC), smooth muscle cells (SMC), T cells, mast cells and proliferating immune cells (ProIC), respectively. (c) Heatmap of unsupervised clustering analysis shows the top ten highly expressed genes per cell type as determined by Seurat analysis with the top three genes per cluster highlighted on the right. Expression level is scaled based on z-score distribution. (d) Expression of the selected top marker genes for each cell type is shown in UMAP plots.

Figure 2

Identification of specific HLA-DRA⁺ cell subtypes in OA synovia. (a) The HLA-DRA⁺ cell populations consisting of five distinct cell subtypes are shown in a UMAP plot. (b) Bar graphs demonstrate the top ranked canonical pathways associated with each cell subtype based on their percentage of highly expressed genes (right) and their significance (left). Pathways related to these cells include: Fcγ receptor-mediated phagocytosis; TREM1 signaling; dendritic cell maturation; hepatic fibrosis; B cell receptor signaling; and multiple cell populations associated with OA. (c) Heatmap shows the top ten highly expressed genes for each cluster with the top five highly expressed genes per cluster highlighted on the right. The top highly expressed genes in IR-MΦ were SEPP1 and FLOR2 that are known to play a role in macrophage polarization and specifically expressed in regulatory macrophages, respectively. The top highly expressed genes in I-MΦ were inflammatory mediators, including CCL3 and CCL4. The top highly expressed genes in iFIB were collagen genes including COL14A1 and COL1A1. Consistent with identification as B cells were cells with high expression of MZB1, TNFRSF17 and CD79A. (d, e) The dot plots depict the average expression level (color scale) and percentage of cells expressing the selected marker genes (dot size) for each cluster. (d) The dot plot shows expression of HLA-DRA in all five cell subtypes and co-expression with an additional 11 markers (d) and additional immune markers and cytokines (e). (d) Classic macrophage marker genes (CD14, CD163 and FCGR3A) were highly expressed in IR-MΦ, I-MΦ and iFIB. Immune regulatory genes, CD169, STAB1 and TXNIP, were highly expressed in IR-MΦ. DC marker genes, FCER1A and CD1C were exclusively expressed in DC. Fibrous matrix genes (COL1A1 and COL1A2) and stromal cell-derived factor 1 (CXCL12) were highly expressed in iFIB. (e) Surface maker genes (HLA-DQA1, HLA-DQA2, OLR1 and TLR2) and cytokines were highly expressed in I-MΦ and DC. (f) Representative immunofluorescence staining showing co-expression of IL1B and surface biomarkers of I-MΦ and DC, such as HLA-DQA1, HLA-DQA2, OLR1 and TLR2 in human OA synovium. Scale bar = 20 μm.

Figure 3

Identification of chondrocyte cell types in OA and the potential upstream regulators of these phenotypes. a shows a similarity across patients and a predominance of preFibrochondrocytes (preFC), Fibrochondrocytes (FC), Regulatory chondrocytes (RegC), Reparative chondrocytes (RepC) and some preHypertrophic chondrocytes (preHTC) in diseased areas designated ‘D’ corresponding to the medial tibial plateau (MT); a also shows a predominance of Homeostatic chondrocytes (HomC), some PreHTC and Hypertrophic chondrocytes (HTC) in non-diseased areas designated ‘N’ corresponding to OLT or the lateral tibial plateau. Potential upstream cytokines (b) and growth factors (e) were identified based on all highly expressed genes from each cluster and scaled by activation z-score; asterisks indicate the upstream mediators that were expressed by > 1% of acquired chondrocytes. Nearly all cytokines (c) and growth factors (f) were expressed by synoviocytes: 22 of 31 cytokines and 14 of 30 growth factors were only expressed by synoviocytes and not by chondrocytes (the proportion of acquired synoviocytes expressing each cytokine and growth factor are shown in the pie charts). (d) Violin plots show expression levels of representative cytokines (TNF, IL6, IL1B and IL1A) and a growth factor (IGF1) across all cell types in OA synovia and the HLA-DRA⁺ cell subtypes. (g) Representative immunofluorescence staining of synovium and cartilage (intact and damaged regions of a single patient specimen) for IL6 and CD14. As expected, the damaged cartilage is hypercellular compared to intact cartilage due to the development of multicellular chondrons in damaged cartilage with osteoarthritis. IL6 was highly expressed by CD14⁺ synovial fibroblasts (SSF, SSF and iFIB), I- MΦ, and SMC, but not chondrocytes from either intact or damaged regions of cartilage.

Figure 4

The cross-talk model of osteoarthritis. We identified 12 subpopulations from OA synovia and 7 distinct chondrocyte subpopulations from OA articular cartilage. We predicted potential upstream regulators of chondrocyte gene expression during OA progression to infer molecular cross-talk networks between cartilage and synovium. Genes expressed by OA chondrocytes and identified as potential mediators of chondrocyte phenotypes in OA are indicated by blue solid dots with arrows; the preponderance of growth factors, such as TGFB among them, is consistent with upregulation of anabolic processes to maintain cartilage homeostasis in OA. However, genes identified as potential mediators of chondrocyte phenotypes in OA that were exclusively expressed (in > 1%) by synoviocytes but not by any of the chondrocyte subtypes are indicated by red solid dots with arrows; among these were genes for several key pro-inflammatory cytokines implicated in the pathogenesis of OA, including IL1A, IL1B, IL6 and TNF that were specifically expressed by inflammatory macrophages (I-Mϕ), dendritic cells (DC) or other synoviocytes. I-Mϕ and DC expressed HLA-DQA1, HLA-DQA2, OLR1 or TLR2; these cells appeared to be the primary cytokine producing cells.