Background A number of susceptibility loci for rheumatoid arthritis (RA) map to noncoding genomic regions. Long noncoding RNA (lncRNA) represent the largest part of noncoding transcriptome and are key regulators of gene expression. We have recently shown that the expression pattern of noncoding RNA in synovial fibroblasts (SF) - one of the key local effector cells in arthritis – is largely determined by anatomic joint location.
Objectives The aim of the present study was to identify arthritis-specific lncRNA transcriptomes of SF at different anatomic locations, focusing on RA and osteoarthritis (OA).
Methods RNA was isolated from cultured SF, passages 5–8, obtained from hands, shoulders and knees of RA and OA patients and knees of non-arthritic patients experiencing joint pain (normal) (n=3 per each group). RNA sequencing (Illumina HiSeq2000) of SF was followed by read alignment to the reference genome and transcriptome (the Ensembl GRCh37, STAR version 2.3.0e_r291), identification of differentially expressed (DE) genes between diagnoses and locations (DESeq2, Bioconductor Version 3.0, log(foldchange) >|0.5|, p<0.01,) and pathway analysis (Metacore, Thomson Reuters) of DE protein-coding genes.
Results RNA sequencing revealed that ∼15.000 protein-coding RNA and 4000 lncRNA are expressed in SF at each joint location. Comparing different joints, 101, 182 and 155 of expressed lncRNA were DE between knee-hand, knee-shoulder and hand–shoulder SF, representing 2.4%, 4.1% and 3.5% of expressed lncRNA, respectively. Furthermore, comparing different diagnoses at each joint location 101 (2.4%) lncRNA were identified as DE between OA-RA hands (e.g. MALAT1, HOXA11-AS), 36 (0.8%) between OA-RA shoulders (e.g. DNAJC3-AS1, MIR29B1), 19 (0.5%) between OA-RA knees (e.g. TBX5-AS1, ITGB2-AS1), 136 (3.5%) between OA-normal knees (e.g. CTC-529P8.1) and 108 (2.6%) between RA-normal knees (e.g. IL21-AS1), respectively, with 51 lncRNA altered in OA and RA vs. normal knee SF. This showed that lncRNA transcriptional programmes of SF in particular differ between OA and RA hands with a difference comparable to that seen between healthy vs. arthritic knees and between anatomic locations. Only a minor number of joint-specific lncRNA were altered by arthritis. Pathway analysis revealed the enrichment of platelet-endothelial-leucocyte adhesion and ECM remodelling networks between RA and OA hands, inflammation – MIF signalling and cytoskeleton-intermediate filaments networks between OA and RA shoulders and chemotaxis and cadherin and integrin adhesion networks between OA and RA knees.
Conclusions Our study identifies joint-specific alterations in stromal synovial lncRNA profiles in RA and OA and shows that in the same type of arthritis different disease pathways are activated at different joint locations. In particular in hands, stromal noncoding transcriptional programmes significantly differ between different types of arthritidis, emphasizing the central role of anatomic location when studying arthritis pathogenesis.
Disclosure of Interest M. Frank-Bertoncelj Grant/research support from: Promedica, euroTEAM, BTCure, CABMM, IAR, G. Russo: None declared, A. Bratus: None declared, C. Kolling: None declared, A. Filer: None declared, B. Michel: None declared, R. Gay: None declared, C. Buckley: None declared, S. Gay: None declared, C. Ospelt Grant/research support from: euroTEAM, BTCure, CABMM, IAR