Single cell analysis of immune cells from synovial tissues during the progression of osteoarthritis and obesity

Utilizing transcriptomic analysis we characterized the presence and fate of multiple populations of immune cells found in the knee joint capsule. Overall design: Single, live synovial CD45+ cells were isolated by flow cytometry sorting and subsequently analyzed by scRNA-seq. Single-cell capture was performed on a 10x Chromium Controller (10xGenomics, Pleasanton, CA). Approximately 1000 cells/µL were loaded into an eight-channel microfluidic chip, which encapsulated them with gel beads (GEMs) and barcoded with a unique molecular identifier (UMI) using 10xGenomics Single Cell 3' reagent kit following manufacturer instructions (10xGenomics, Pleasanton, CA). A single-cell RNA-seq library from each biological replicate was run independently on Illumina NovaSeq flow cell with 26x98 bp reads and generated approximately 38,000-1,000,000 mean reads per library capturing 700-16000 cells per mice (see details in sTable 3). The obtained reads were aligned to the mouse UCSC mm10 reference genome. The alignment results were used to quantify the gene expression level and to generate a gene-barcode matrix. Based on mitochondrial genes, the number of genes and UMIs expressed in each cell, low-quality cells, and potentially dead cells were removed. The Seurat R package (version 4.1) (44) was used to perform single-cell transcriptomic analysis and clustering. Briefly, for each dataset, we filtered out potential low-quality cells using the QC parameters of nUMI >= 500, nGene >= 550 per cell, and <10% of mitochondrial genes. All remaining variable genes were used for downstream analyses, and counts were normalized using the function NormalizeData. All 12 analyzed biological replicates were integrated using the SCT function with rpca reduction and CellCycle Scoring. For data alignment, we selected 3,000 highly variable genes in each data matrix and performed the 'FindIntegrationAnchors' and 'IntegrateData' functions in Seurat 4.1. Next, for clustering analysis, the FindNeighbors() and FindClusters() functions of the Seurat package were used. FindClusters() function was set to '0.3' granularity. For the identification of differentially expressed genes (DEGs), we used the FindMarkers, FindConservedMarkers, or FindAllMarkers function (test. use = 't,' logfc. threshold = log[0.25]) based on normalized data. Differentially expressed genes (DEGs) for each of the clusters are presented in Table 5.