Single nucleus RNASeq profiling of mouse lung: reduced dissociation bias and improved detection of rare cell types compared with single cell RNASeq

RATIONALE: Single cell RNA-sequencing (scRNASeq) has led to multiple recent advances in our understanding of lung biology and pathophysiology, but utility is limited by the need for fresh samples, loss of cell types due to death or inadequate dissociation, and the induction of transcriptional stress responses during tissue digestion. Single nucleus RNASeq (snRNASeq) has addressed these deficiencies in some other tissues, but no protocol exists for lung. We sought to develop such a protocol and compare its results with scRNA-seq. METHODS: Single nucleus suspensions were prepared rapidly (45 min) from two mouse lungs in lysis buffer on ice while a single cell suspension from an additional mouse lung was generated using a combination of enzymatic and mechanical dissociation (1.5 h). Cells and nuclei were processed using the 10x Genomics platform, and following sequencing of cDNA libraries single cell data was analyzed by Seurat. RESULTS: 16,656 single nucleus and 11,934 single cell transcriptomes were generated. Despite reduced mRNA levels in nuclei vs. cells, gene detection rates were equivalent in snRNASeq and scRNASeq (~1,750 genes and 3,000 UMI per cell) when mapping intronic and exonic reads. snRNASeq identified a much greater proportion of epithelial cells than scRNASeq (46% vs 2% of total), including basal and neuroendocrine cells, while reducing immune cells from 54% to 15%. snRNASeq transcripts are enriched for transcription factors and signaling proteins, with reduced detection of housekeeping genes, mitochondrial genes, and artifactual stress response genes. Both techniques improved mesenchymal cell detection over previous studies, and analysis of fibroblast diversity showed two transcriptionally distinct populations of Col13a1+ cells, termed Bmper+ and Brinp1+ fibroblasts. To define homeostatic signaling relationships among cell types, receptor-ligand mapping of was performed for alveolar compartment cells using snRNASeq data, revealing complex interplay among epithelial, mesenchymal, and capillary endothelial cells. CONCLUSION: Single nucleus RNASeq can be readily applied to snap frozen, archival murine lung samples, improves dissociation bias, eliminates artifactual gene expression and provides similar gene detection compared to scRNASeq.