Many genes show highly heterogeneous expression across populations of cells, with high expression in some cells and low expression in other cells.
In development, this heterogeneity helps give rise to the diverse cell types in the body. Our hypothesis is that this structural feature of cell populations arising in development may be co-opted by cancer to form persister cell states that resist therapy. Gene expression variability has previously been attributed to “noise,” or random variation. Yet, not all genes show variable expression, with many stably expressed across populations. We combine bioinformatic analysis of single cell expression datasets with targeted experimental cell biology to understand what gives rise to gene expression variation at the single cell level. We focus on mouse embryonic stem cells and well characterized cancers to understand what common features drive heterogeneity in these populations.
Many genes such as the pluripotency regulator Esrrb show a broad range of expression, with some cells expressing little to no gene product and other cells expressing much larger amounts. In contrast, Gapdh shows relatively uniform expression across cells. Reproduced with permission from Garg & Sharp, Science 2016. Data plotted from Kumar et al, Nature 2014 and Klein et al, Cell 2015.