Improving Transcriptome Fidelity Following Synovial Tissue Disaggregation

Abstract

Objective: To improve the fidelity of the cellular transcriptome of disaggregated synovial tissue for applications such as single-cell RNA sequencing (scRNAseq) by modifying the disaggregation technique. Methods: Osteoarthritis (OA) and rheumatoid arthritis (RA) synovia were collected at arthroplasty. RNA was extracted from intact or disaggregated replicate pools of tissue fragments. Disaggregation was performed with either a proprietary protease, Liberase TL (Lib) as a reference method, Liberase TL with an RNA polymerase inhibitor flavopyridol (Flavo), or a cold digestion with subtilisin A (SubA). qPCR on selected markers and RNAseq were used to compare disaggregation methods using the original intact tissue as reference. Results: Disaggregated cell yield and viability were similar for all three methods with some viability improved (SubA). Candidate gene analysis showed that Lib alone dramatically increased expression of several genes involved in inflammation and immunity compared with intact tissue and was unable to differentiate RA from OA. Both alternative methods reduced the disaggregation induced changes. Unbiased analysis using bulk RNAseq and the 3 protocols confirmed the candidate gene studies and showed that disaggregation-induced changes were largely prevented. The resultant data improved the ability to distinguish RA from OA synovial transcriptomes. Conclusions: Disaggregation of connective tissues such as synovia has complex and selective effects on the transcriptome. We found that disaggregation with an RNA polymerase inhibitor or using a cold enzyme tended to limit induction of some relevant transcripts during tissue processing. The resultant data in the disaggregated transcriptome better represented the in situ transcriptome. The specific method chosen can be tailored to the genes of interest and the hypotheses being tested in order to optimize the fidelity of technique for applications based on cell suspensions such as sorted populations or scRNAseq.