The landscape of alternative polyadenylation in CD8 T cells in single-cell transcriptome

Lixinyu Liu0, Jiayu Wen0
(0) Australian National University

Find me on Tues Nov 24th, 1:40-3pm AEDT in Remo, table 95

Abstract
Alternative cleavage and polyadenylation (APA) enables the production of different isoforms of mRNA and affect more than 70% eukaryotic genes. APA is a critical process as it diversifies the transcriptome in cells, and can dynamically affect transcript stability, cellular localisation, nuclear export, and translation efficiency. APA is cell-type specific and has been found to be abundant in immune cells. In particular, specific APA events have been detected in several pathological conditions, including immunological and autoimmune diseases. However, a comprehensive exploration of APA in immune cells at the single cell level has not been conducted previously. In this study, we profiled mouse CD8+ T cells, known as cytotoxic T cells acting as intermediaries of adaptive immunity, from eight tissues by single-cell sequencing (scRNA-seq) to uncover cell type and developmental trajectories of differential APA varying across cell types, developmental states and tissues.
We first comprehensively maps the APA sites in 13 cell types. At the cell-type level, naïve CD8+ cells show more lengthening isoforms than other cell types, and the comparatively shorter isoforms are associated with drug metabolism, regulation of lymphocyte-mediated immunity, and positive regulation of immune effectors. Effector CD8+ cells prefer proximal sites compared to others, and the shorter isoforms are associated with biosynthesis, endocytosis, cell to cell recognition, and upregulation of immune defences. By comparing differential APA usage at the single cell level, we detect cell type-specific APA markers. For example, Id2 prefers its proximal polyadenylation site in effector memory CD8+ cells and prefers its distal polyadenylation site in Naïve CD8+ cells. To correlate APA usage and gene expression, we further use a deep learning method, Autoencoder, to compress APA usage and expression features at a single-cell level. With the autoencoder model, the single-cell map of APA usage can be used to improve the expression-based clustering map. Finally, we develop a visualisation method to visualise the APA usage of CD8+ cells at single-cell resolution.