Single cell transcriptome data from human lymph node stromal cells and immune cells

For stromal cell isolation from human LN, tissues were cut into small pieces and collected in RPMI 1640 medium containing 2% FCS, 20 mM HEPES pH 7.2 (Lonza), 0.12 mg/ml Dispase (Roche), 0.16 mg/ml Collagenase P (Roche) and 25 μg/ml DNaseI (Applichem). The tissue pieces were incubated at 37°C for 60 min, with resuspension and collection of the supernatant every 12 min. To enrich for stromal cells, hematopoietic cells and erythrocytes were depleted by incubating the cell suspension with MACS anti-CD45 and anti-CD235a microbeads (Miltenyi Biotec) and FcR Blocking Reagent (Miltenyi Biotec) and passing it through a MACS LS column (Miltenyi Biotec). The unbound single cell suspension was stained for cell sorting. For hematopoietic cell isolation from human LNs and palatine tonsils, tissues were gently smashed through a 26-gauge wire mesh and washed with phosphate buffered saline until further staining for cell sorting.

Non-haematopoietic cells from human LNs and haematopoietic cells from human LNs and palatine tonsils were sorted using a BD FACSMelody cell sorter (BD Biosciences). Single cell suspensions were prepared using the 10x Chromium droplet-based system (10x Genomics). Following established commercial protocols for the Chromium Single Cell 3' Reagent Kit (v3 or NextGEM Chemistry), cDNA libraries were generated and subsequently sequenced on an Illumina NovaSeq 6000 at the Functional Genomic Center Zurich. Gene expression was assessed from the sequencing files by running CellRanger (v5.0.1) count, using the Ensembl GRCh38.9 release as a reference. Additional quality control procedures were performed in R v.4.2.1 using the R/Bioconductor package scater (v.1.24.0) (McCarthy et al., 2017) to eliminate low-quality cells with very high or low UMI counts (>2.5 median absolute deviation from the median of all cells), very high or low total number of detected genes (>2.5 median absolute deviation from the median of all cells), and elevated mitochondrial gene content (>2.5 median absolute deviations above the median of all cells). Subsequent analyses were performed separately for samples containing non-hematopoietic cells and hematopoietic cells. Downstream analysis was performed using functions implemented in the Seurat R package (v.4.3.0) (Butler et al., 2018; Hao et al., 2021) and included data normalisation, integration to mitigate batch effects due to different processing sites, data scaling, dimensionality reduction by PCA and UMAP, graph-based clustering and identification of unbiased cluster markers. Clusters were characterised based on the expression of unbiased cluster markers and canonical marker genes described in previous publications.