The organs of multicellular animals comprise highly organized aggregates of many cell types, each of which has differentiated from a multi-potent stem cell. Although we have learned much about the process of differentiation and organogenesis through studies of tissues such as bone marrow, tracing the lineages of cells in solid organs remains a challenge. New techniques of single-cell analysis that enable the simultaneous characterization of large numbers of cells should lead to substantial progress in identification of transitional cell populations. However, these techniques generate huge datasets that require appropriate computational approaches for their analysis. Thus far, most such approaches have been limited to the identification of single branch points in a differentiation pathway, and statistical methods for the comparative evaluation of multiple pathways have been lacking. This led Vanderbilt Basic Sciences investigator Ken Lau, in collaboration with Robert Coffey (Department of Medicine) and Michael Gerdes (GE Global Research) to develop p-Creode, a computational method for the evaluation of large datasets obtained from single cell analysis. Major advantages to p-Creode are that no prior assumptions about the nature, number, or arrangements of branch points in a differentiation pathway are necessary, multiple branch points can be identified, and robust statistical analysis enables comparison of multiple pathways. To test their approach, the investigators first used p-Creode to analyze publicly available single cell mass-cytometry data from human bone marrow. Hematopoiesis, the differentiation of blood cells in bone marrow is very well-defined, and p-Creode generated a differentiation hierarchy that matched the known process. The researchers then turned to a single cell multiplex immunofluorescence (MxIF) data set of intestinal and colonic epithelial cells. The results identified known mature cells in each case, as well as the expected lineages for the differentiation of those cells. However, results suggested that tuft cells were derived from different lineages in the intestine and the colon. This unexpected finding led to follow-up studies that confirmed that tuft cells arise from absorptive cell lineages, rather than secretory lineages as previously thought in the intestine, whereas in the colon, they branch off very close to progenitor cells. Analysis of a single cell RNA-seq dataset from mouse colon revealed additional new cell transition relationships. These findings support p-Creode as a robust and versatile approach for the elucidation of cell lineages in complex populations from single cell datasets. We look forward to the results of further applications of p-Creode in the future. The work is published in the journal Cell Systems [C.A. Herring, et al. Cell Systems, (2017) published online November 15, DOI: 10.1016/j.cels.2017.10.012].
Fluoresence image, overlay RGB Mouse Intestine.
Figure reproduced under the Creative Commons Attribution.