Molecular maps of synovial cells in inflammatory arthritis using an optimized synovial tissue dissociation protocol

Summary In this study, we optimized the dissociation of synovial tissue biopsies for single-cell omics studies and created a single-cell atlas of human synovium in inflammatory arthritis. The optimized protocol allowed consistent isolation of highly viable cells from tiny fresh synovial biopsies, minimizing the synovial biopsy drop-out rate. The synovium scRNA-seq atlas contained over 100,000 unsorted synovial cells from 25 synovial tissues affected by inflammatory arthritis, including 16 structural, 11 lymphoid, and 15 myeloid cell clusters. This synovial cell map expanded the diversity of synovial cell types/states, detected synovial neutrophils, and broadened synovial endothelial cell classification. We revealed tissue-resident macrophage subsets with proposed matrix-sensing (FOLR2+COLEC12high) and iron-recycling (LYVE1+SLC40A1+) activities and identified fibroblast subsets with proposed functions in cartilage breakdown (SOD2highSAA1+SAA2+SDC4+) and extracellular matrix remodeling (SERPINE1+COL5A3+LOXL2+). Our study offers an efficient synovium dissociation method and a reference scRNA-seq resource, that advances the current understanding of synovial cell heterogeneity in inflammatory arthritis.

. Representative synovial biopsy images from a patient with psoriatic arthritis.The upper (lower) whisker extends from the hinge to the largest (smallest) value no further than 1.5 * interquartile range from the hinge.Individual dots represent data from different samples.Our data and data from Stephenson W et al.S2 is derived from unsorted synovial cells, while Wei K et al.S3 dataset includes sorted CD45 neg CD235 neg synovial fibroblasts, pericytes/mural cells and synovial endothelial cells.

SUPPLEMENTAL TABLES AND LEGENDS
Figure S18

Figure S2 .
Figure S2.Quality control of integrative protocol scRNA-seq analysis of 18 human synovial biopsies of patients with inflammatory arthritis.Related to Fig. 2. a) The number of genes vs. total counts per sample coloured by filter (left) and percentage mitochondrial counts (right).b) The number of cells per sample (left) and per protocol (right), coloured by the number of genes (top) and by the number of genes in at least 1% of cells (bottom).c) Distribution of counts per protocol (left) and per sample (right).d) Distribution of the number of genes per protocol (left) and per sample (right).

Figure S3 .
Figure S3.UMAPs of integrated scRNA-seq data from integrative protocol analysis showing main synovial cell types.Related to Fig. 3. Data derive from 18 synovial tissue biopsies of patients with inflammatory arthritis and are coloured by top marker genes of major synovial cell populations including a) T cells, B cells and plasmablasts; b) macrophages and plasmacytoid dendritic cells; c) myeloid dendritic cells, mast cells and neutrophils; d) endothelial cells, pericytes/mural cells and synovial fibroblasts.

Figure S4 .
Figure S4.UMAPs of scRNA-seq data from integrative protocol analysis showing synovial cell subsets.Related to Fig. 3. Data from 18 synovial tissue biopsies of patients with inflammatory arthritis.a) T cells/NK cells, color by the expression of key marker genes CD3, CD4, CD8, NKG7 or by the geometric

Figure S5 :
Figure S5: The heatmap of major neutrophil genes, detected in the integrative protocol scRNA-seq analysis.Related to Fig. 3. Data derive from 18 synovial tissue biopsies of patients with inflammatory arthritis.Expressions are aggregated by sample and cell type.

Figure
Figure S6 Multicolor flow cytometry analysis of freshly isolated synovial tissue cells using protocol 2. Related to Fig. 3. Synovial cells were dissociated from a knee synovial biopsy of a patient with septic oligoarthritis.The cells were labeled with Zombie Green™ Fixable Viability Dye, APC anti-human CD45, BV421™ anti-human CD11b, PE anti-human CD64, BV785™ anti-human CD19, and Alexa Fluor® 700 antihuman CD3.a) Total synovial cells, excluding cell debris.b) Gating single cells and doublet exclusion.c) Gating live cells and excluding dead cells, with a high Zombie Green™ Fixable Viability Dye signal.d) Live synovial cells gated for CD45+ leukocytes (violet) and CD45-structural cells (green).e) Gating of CD45+ leukocytes into CD11b+ CD64+ macrophages.f) CD11b -CD64-cells were gated further into CD3+ CD19-T cells and CD3-CD19+ B cells.g) A total of 10121 events were analyzed, among which 74.2% were gated as synovial cells, consisting of 93.2% of single cells with 80.2% viability.

Figure
Figure S7 Spectral flow cytometry analysis of freshly isolated synovial tissue cells using protocol 2. Related to Fig. 3. Synovial cells were dissociated from a biopsied wrist synovial tissue of a patient with early RA.Fixed unsorted synovial cells were labeled with a cocktail of antibodies targeting surface leukocyte (CD45), myeloid (CD14), lymphocyte (CD3, CD4, CD8, CD19) and structural cell (CD31, PDPN) markers followed by analysis on spectral analyser Sony ID7000.a) Division of synovial cells based on the expression of CD45.CD45+ leukocytes included CD14+ macrophages, CD19+ B cells and CD3+ T lymphocytes, dividing further into CD4+ and CD8+ T cells.CD45neg synovial cells contained two structural cell clusters, including PECAM1+ endothelial cells and PDPN+ synovial fibroblasts.There was a considerable amount of cell debris, which might have been attributed to the sample shipment from Portugal to Germany.b) Unstained synovial cells were used for setting the flow cytometry gates.

Figure S8 .
Figure S8.Quality control of scRNA-seq analysis of the single cell reference map of fresh human synovium from 25 synovial biopsies of patients with inflammatory arthritis.Related to Figs. 4-8.a) A cell-level summary of the total number of counts, the number of detected genes and the percentage of mitochondrial counts.The number of genes vs. total counts per sample coloured by filter (left) and percentage mitochondrial counts (right).b) Sample summary statistics with the number of cells and number of detected genes after filtering of low-quality cells.The number of cells per sample is coloured by the number of genes (top) and by the number of genes in at least 1% of cells (bottom).c) Distribution of counts per sample.d) Distribution of the number of genes per synovial tissue sample.

Figure S9 .
Figure S9.Integrated scRNA-seq dataset from 25 synovial tissue samples from patients with inflammatory arthritis.Related to Figs. 4-8.a) UMAP of annotated main synovial cell populations coloured by main cell type.b) Bar plots of relative abundances of main cell types per sample coloured by main cell type.c) The variability of the proportion of cell types across patient synovial tissues.The box plot visualises 5 summary statistics: the median; two hinges, corresponding to the first and the third quartiles; two whiskers.The upper (lower) whisker extends from the hinge to the largest (smallest) value no further than 1.5 * interquartile range from the hinge.Individual dots represent data from different samples.See STAR Methods for details.

Figure S10 .
Figure S10.A heatmap of top marker genes of the main synovial cell types identified in the integrated scRNA-seq dataset.Related to Figs. 4-8.Data are derived from 25 synovial tissue samples from patients with inflammatory arthritis (see STAR Methods for details).Expressions are aggregated by sample and cell type.

Figure S11 .
Figure S11.T cell, natural killer (NK) cell and innate lymphoid cell sub clustering.Related to Fig. 4. The integrated scRNA-seq dataset consists of scRNA-seq profiles from 25 synovial tissue samples from patients with different types of inflammatory arthritis (see STAR Methods for details).a) UMAPs showing the small population of proliferating TOP2A+ CENPF+ T cells (cluster 2).b) A heatmap of the cluster-enriched genes and marker genes in synovial T cell, NK cell and innate lymphoid clusters in patients with inflammatory arthritides.Expressions are aggregated by sample and cell type.c) UMAPs demonstrating the expression of FOXP3, PCDC1 and CXCL13 genes in the TIGIT+ CTLA4+ T cells (cluster

Figure S12 .
Figure S12.Synovial fibroblast sub clustering.Related to Fig. 5.The integrated dataset from 25 synovial biopsies of patients with inflammatory arthritis (see STAR Methods for details) with a) violin plots showing the expression (log counts) of the lining marker PRG4 and the sublining marker THY1 across the seven synovial fibroblast clusters, and b) UMAPs showing the small population of proliferating TOP2A+CENPF+ synovial fibroblasts, co-clustering with the cluster four SERPINE1+ COL5A3+ synovial fibroblasts (see also Fig.5a, d).

Figure
Figure S13.A heatmap of the top cluster genes and known markers of synovial fibroblast clusters detected in the synovium of patients with inflammatory arthritides.Related to Fig. 5.The integrated scRNA-seq dataset consists of 25 synovial tissue samples from patients with different types of inflammatory arthritis (see STAR Methods for details).Expressions are aggregated by sample and cell type.

Figure S14 .Fig. 5 .
Figure S14.The expression of a selected set of genes in synovial fibroblast sub clustering.Related to Fig. 5.The integrated dataset from 25 synovial biopsies of patients with inflammatory arthritis (see STAR Methods for details) with a) UMAPs showing the highest enriched expression of the genes involved in MHCII class mediated antigen presentation in HLA-DRA high synovial fibroblasts (cluster 7), and b) UMAPs showing the enriched expression of IL6 and NOTCH3 genes primarily in sublining GGT5+ synovial fibroblasts (clusters 6).

Figure S15 .
Figure S15.Synovial macrophage and myeloid dendritic (DC) cell sub clustering.Related to Fig. 6.A heatmap of the top cluster genes and known marker genes of synovial macrophage and DK subclusters, detected in the integrated scRNA-seq dataset from 25 synovial tissue samples of patients with different types of inflammatory arthritis (see STAR Methods for details).Expressions are aggregated by sample and cell type.

Figure
Figure S16 Trajectory analysis of synovial endothelial cells.Related to Fig. 7. Inferred pseudotime versus gene expression in synovial endothelial cells.Start of the trajectory are GJA4+ CLDN5+ arterial

Figure S17 .
Figure S17.Endothelial cell sub clustering.Related to Fig. 7. Heatmap of the top cluster genes and known markers of synovial vascular and lymphatic endothelial cell subclusters detected in the integrated scRNA-seq dataset from 25 synovial tissue samples of patients with different types of inflammatory arthritis (see STAR Methods for details).Expressions are aggregated by sample and cell type.

Figure S18 .
Figure S18.Integration of our synovial scRNA-seq from fresh synovium with publicly available synovial scRNA-seq datasets.Related to Fig. 9.We integrated our scRNA-seq data from 25 synovial tissue samples of patients with inflammatory arthritis with published data from Stephenson W and colleagues S2 and Wei K et colleagues S3 (see STAR Methods for details).a) UMAP of annotated main synovial cell populations coloured by main cell type across the three studies.b) Bar plots of relative abundances of main cell types per sample coloured by main cell types across the 3 studies.c) The variability of the proportion of cell types across patient synovia in the 3 studies.The box plot visualises 5 summary statistics: the median; two hinges, corresponding to the first and the third quartiles; two whiskers.

Table S1 Detection of neutrophils in 18 synovial tissue biopsies included in integrative protocol analysis using histology and scRNA-seq analyses. Related to Figure 2.
For the detection of neutrophils in formalin-fixed, paraffin-embedded synovial biopsy fragments, tissue sections were labeled with CD15 antibodies (see STAR Methods for details).In scRNA-seq data, the quantity of neutrophils was estimated by calculating the proportion of neutrophils per sample.Data pertains to 18 samples included in integrative protocol analysis.Y: yes, N: no.

Table S2 . Demographic, clinical, therapy, histology and cell characteristics for 2 synovial tissue biopsy samples from patients with arthritis, included in the proof of principle flow cytometry analyses. Related to Figures 6
, 7. F: female, M: male, *Automatic cell counting with the Luna-FL Dual Fluorescence Cell Counter.