3’ UTR isoform usage is highly consistent among single cells at the same cell state

The 3’ untranslated region (3’ UTR) of genes plays an important role in post-transcriptional regulation. Its length is determined by the site of poly-adenylation. Multiple sites can give rise to different 3’ UTR lengths (alternative poly-adenylation – APA), producing differentially regulated isoforms. RNA-seq can determine the 3’ UTR isoforms present, and shows changes in isoform usage in different cell types. Identifying multiple isoforms of a gene could result from all cells having multiple isoforms, or each cell expressing a different isoform. Single-cell RNA-seq studies performed to date have not attempted to distinguish between these possibilities, as cells are pooled according to cell type before looking at APA.

Here we use the CEL-Seq2 protocol to determine the 3’ UTR isoforms in single-cells during spontaneous differentiation of human embryonic stem cells, and identify many APA events. The sensitivity of our protocol allows us to determine the isoform usage at the single-cell level without the need for pooling. To distinguish between biological changes in isoform usage and technical noise we also analyzed technical replicates of an RNA sample diluted down to single-cell amounts. We show that the biological variability in isoform usage between cells is low, and that the ratio between 3’ UTR isoforms in each individual cell is highly consistent with the overall ratio between the isoforms observed in bulk RNA-seq of each cell state. This suggests that 3’ end choice is tightly regulated at the single-cell level, and is highly consistent among individual cells in homogenous populations.