META-ANALYSIS OF SINGLE-CELL RNA SEQUENCING DATA OF THE SYNOVIUM TO DEFINE SYNOVIAL FIBROBLAST PHENOTYPES ACROSS JOINT LOCATION AND DISEASE

Background: Up to now, three groups used single cell RNA sequencing (scRNA-seq) to analyse the synovium in arthritis using different methods and material to measure RNA expression on a single cell level: Ref. 1 used unsorted dissociated synovial cells and a droplet based method; Refs 2 and 3 performed scRNA-seq on sorted cell populations. Objectives: The aim of this study was to perform a meta-analysis of scRNA-seq data of the synovium in arthritis: 1) to define synovial fibroblast (SF) phenotypes, 2) to confirm differences across SF clusters between rheumatoid arthritis (RA) and osteoarthritis (OA) and 3) to analyse joint specific differences between SF phenotypes. Methods: In addition to the available count matrices [1-3], we used unsorted dissociated synovial cells from three patients with undifferentiated arthritis (UA) with a droplet-based method (10x Genomics). We followed a standard protocol [4] to integrate the datasets into a shared space, even in the presence of extensive technical and/or biological differences (“batch-corrected”). SF were selected as previously described (PDPN+, ISLR+, COL1A2+, PTPRC-) [1-3]. We used a minimum log2 FC of 0.25 for average expression of genes in a cluster relative to the average expression in all other clusters combined to define marker genes. R with Seurat, Monocle and clusterProfiler packages were used for scRNA-seq analysis, pseudotime trajectory analysis and pathway enrichement analysis, respectively. Quantitative PCR (qPCR) (n=6-14 per location and disease), immunohistochemistry (IHC) and Krenn synovitis score (n=5-15 per location and disease) were performed according to standard protocols. Results: Data from 29 RA, 3 UA and 6 OA patients were analysed. From a total of 29’448 cells, we identified 14’787 (50%) with a fibroblast phenotype. Of those, we determined 5 subpopulations (Fig. 1): 1) THY1-CD55hi fibroblasts with high expression of MMP1 and MMP3 (SF1), 2) THY1loCD34+ fibroblasts expressing high levels of PI16 (SF2) 3) THY1hi fibroblasts expressing high levels of periostin (POSTN) and collagens (e.g. COL1A1, COL3A1) (SF3), 4) THY1hi fibroblasts expressing CXCL12 (SF4) and 5) THY1lo fibroblasts expressing CXCL12, NR4A1 and CCL2 (SF5). Fig. 2 shows pathway enrichment map of all marker genes; it organizes enriched terms into a network with edges connecting overlapping gene sets. Pseudotime trajectory axis derived from Monocle indicated that SF4 represent a state between SF3 and SF5. Pseudotemporal expression dynamics of THY1 marked the progression of these three subtypes (Graph 1). SF1 and SF2 were proportionally underrepresented and SF3-5 overrepresented in RA (chi-squared = 37.18, p = 1.65e-07). The expression of POSTN, a signature gene of SF3, was not different between RA and OA tissues, but significantly correlated with the synovitis score (Spearman ρ = 0.55, p=0.02), in particular with pathological changes in the sublining. POSTN expression was higher in hand than in knee synovial tissues (mean ± SD IHC score: hand 8 ±2, knee 5 ±2) and in cultured SF (qPCR: 10-fold difference). Accordingly, SF3 was enriched in hand versus knee synovial tissues in the scRNA-seq dataset (chi-squared = 944.87, p Graph 1 Conclusion: In our meta-analysis, we found comparable subtypes of fibroblasts as in the individual analyses [1-3], showing the robustness of cell phenotype identification using scRNA-seq. The different SF phenotypes appear to be plastic cell states rather than fixed cell subtypes, whose development is controlled by an interrelation between pathological changes in the synovium and joint location. References: [1]Stephenson et al. Nat. Commun. 2018 [2]Mizoguchi et al. Nat. Commun. 2018 [3]Zhang et al. Nat Immunol. 2019 [4]Stuart, Butler, et al. Cell 2019 Disclosure of Interests: Raphael Micheroli: None declared, Mojca Frank-Bertoncelj: None declared, Sam G. Edalat: None declared, Kerstin Klein: None declared, Tadeja Kuret: None declared, Kristina Buerki: None declared, Adrian Ciurea Consultant of: Consulting and/or speaking fees from AbbVie, Bristol-Myers Squibb, Celgene, Eli Lilly, Merck Sharp & Dohme, Novartis and Pfizer., Oliver Distler Grant/research support from: Grants/Research support from Actelion, Bayer, Boehringer Ingelheim, Competitive Drug Development International Ltd. and Mitsubishi Tanabe; he also holds the issued Patent on mir-29 for the treatment of systemic sclerosis (US8247389, EP2331143)., Consultant of: Consultancy fees from Actelion, Acceleron Pharma, AnaMar, Bayer, Baecon Discovery, Blade Therapeutics, Boehringer, CSL Behring, Catenion, ChemomAb, Curzion Pharmaceuticals, Ergonex, Galapagos NV, GSK, Glenmark Pharmaceuticals, Inventiva, Italfarmaco, iQvia, medac, Medscape, Mitsubishi Tanabe Pharma, MSD, Roche, Sanofi and UCB, Speakers bureau: Speaker fees from Actelion, Bayer, Boehringer Ingelheim, Medscape, Pfizer and Roche, Caroline Ospelt Consultant of: Consultancy fees from Gilead Sciences.