Poster M11 - Best practices for analyzing single-cell RNA seq in engineered tissues

Introduction: Single-cell RNA sequencing (scRNAseq) has become popular in recent years for evaluating native tissues during development, homeostasis, and disease. This technique has been applied to study cellular communities within engineered tissues as well, including cultured cells and organoids, up to whole engineered tissues. However, most scRNAseq software packages and analysis pipelines have been developed specifically for native tissues, so special considerations must be made at all steps of the process to yield the most accurate and powerful analyses possible for engineered tissues.

Materials and

Methods: Through many years of collective work analyzing scRNAseq of cultured cells, organoids, and engineered tissues, we have developed recommended “best practices” for generating and analyzing such data. The steps of this process include sample dissociation; library preparation and sequencing; data filtration and cleaning; clustering and cell type identification; and downstream analyses.

Results, Conclusions, and Discussions: Dissociating engineered tissues to single cells for library preparation requires protocols that have been optimized for the particular tissue type to maximize cell viability and minimize cell fragments and doublets. Organoids are particularly susceptible to doublets and often require specialized enzyme mixes to break cell-cell junctions. It is important to consider at all steps that cells from engineered tissues or cultured cells are often larger and more functionally active than cells from native homeostatic tissues, which may result in higher transcript counts in these cells. This affects several decisions during data generation and analysis, including library preparation, data cleaning, and downstream analyses, particularly comparisons to native scRNAseq data. In the data cleaning step, it is important to use informed filtration cutoffs on transcript and gene counts for each engineered or cultured cell type individually, as most default parameters in cleaning pipelines are based on native cells and may remove important signal from engineered datasets. In clustering and cell type identification, engineered and cultured cells often do not closely resemble their native counterparts due to differences in transcriptomic activity based on their culture environment (Greaney 2020). It is therefore important to leverage multiple modes of analysis to define cell types or archetypes on a functional basis, not just a short canonical marker list. We show how to apply gene ontology, trajectory inference, and ligand-receptor expression analysis to scRNAseq data from engineered tissues.
scRNAseq is a powerful tool for analyzing engineered tissues and cultured cells at high resolution. Special considerations must be made to adapt common pipelines built for native cells to engineered systems in order for them to be most accurate and useful.

Acknowledgements (Optional):