Expression profiling by high throughput sequencing
Summary
Background: The planarian, Schmidtea mediterranea, is a master regenerator with a large adult stem cell compartment. The lack of transgenic labeling techniques in this animal have hindered the study of lineage progression and has made understanding the mechanisms of tissue replacement during regeneration a challenge. However, recent advances in single cell transcriptomics and analysis methods may allow for the discovery of novel cell lineages as differentiation progresses from stem cell to terminally differentiated cell. Here we apply Waterfall analysis and single cell transcriptomics to identify adult stem cells belonging to specific cellular lineages and identify novel candidate genes for future in vivo lineage studies. Results: Single-cell-RNA-deep sequencing (scRNAseq) of 168 purified single stem and progeny cells from the planarian head was used to identify a molecularly distinct, neural stem cell population (νNeoblasts). Hierarchical clustering revealed 10 distinct subgroups and pseudotime analysis with Waterfall predicted a neural lineage trajectory as well as a novel alternative lineage. Previously undescribed neural lineage genes identified in silico were characterized in vivo and were shown to have neural expression patterns. Using the predicted νNeoblast markers, we demonstrate that a novel piwi-2+piwi-1lo stem cell population exists adjacent to the brain and immediately takes up BrdU. Conclusions: scRNAseq coupled with in silico lineage analysis offers a new approach for studying lineage progression in planarians. The lineages identified here were extracted from a highly heterogeneous dataset with minimal prior knowledge of planarian lineages, demonstrating that lineage purification by transgenic labelling is not a prerequisite for this approach. The identification of the νNeoblast lineage demonstrates the usefulness of the planarian system for computationally predicting cellular lineages in an adult context coupled with in vivo verification.
Overall design
Cells from the heads of 30 animals were FACS sorted to obtain 96 single X1s and 72 single X2s and cDNA was prepared from poly-A+ mRNA for each cell using Smartseq2 and deep sequenced to a depth of 5M reads, as illustrated in Figure 1A of the manuscript. cDNA from two X1 and 1 X2 bulk tube controls of 200 cells each was prepared using the same protocol and sequenced to a depth of 10-20M reads each.