EXPLORING THE EFFECTS OF DIFFERENTIAL PROMOTER USAGE IN TRANSCRIPTIONAL NOISE GENOME-WIDE

IUNGMAN MARTÍN; SCHOR IGNACIO ESTEBAN

virtual

Congreso; LVI Annual Meeting of SAIB and XV Annual Meeting of SAMIGE; 2020

Sociedad Argentina de Investigación Bioquímica y Biología Molecular

In the last decades it has come to light that ?identical? cells, even in the same environment, are heterogeneous in terms of their gene expression patterns. At the RNA level, this variability between cells is mostly due to what we call transcriptional noise. Transcriptional noise can be caused by asymmetrical distribution of transcription factors or other molecules during mitosis, or by the nature of transcription, a temporally discontinuous process occurring in irregular bursts. Due to previous reports highlighting the role of gene promoters on this process, we question whether, in genes with multiple alternative promoters, transcriptional noise varies  when promoter usage changes, for example during cell differentiation. To explore this relationship we decided  to study differential promoter usage and transcriptional noise genome-wide in three related cell types from mouse bone marrow: hematopoietic stem cells, macrophages and megakaryocytes. To identify the genes which have differential usage of alternative promoters between these cell types, we took data published CAGE (Cap Analysis of Gene Expression) from the FANTOM project (https://fantom.gsc.riken.jp/5/), a bulk RNA technique that allows the identification of the 5? end of each transcript. We used a binomial generalized linear model to test changes in promoter usage for each pair of cell types. On the other hand, we analyzed single-cell RNA-seq from bone marrow, using the recently described VarID method (Grün, Nat Methods, 2020), combining a trimmed version of the K-nearest neighbours algorithm with noise modelling to define homogeneous cell groups, which later were assigned to cell types by the expression of cell-specific markers. Corrected RNA level variation between these homogeneous cell groups was used as a proxy for transcriptional noise, and differential noise between each pair of cell types was evaluated with a Wilcoxon Rank-Sum test. We detected 530, 427  and 381 genes with significant difference in promoter usage for the stem-megakaryocyte, stem-macrophage  and megakaryocyte-macrophage contrasts respectively. With respect to transcription noise, 25, 294 and 254 genes were found as significant for the same contrasts. Preliminary results integrating these analyses suggest  that there is not a strong association between these two phenomena, while showing instead a dependence between differential gene noise and differential gene expression. This latter result suggests that even after noise-expression modelling, we need to consider the changes on  expression level as a relevant variable in any noise-related analysis.  In addition, we plan to explore this behaviour on other cell types beyond the blood cells. Our work presents an effort to integrate different genomic datasets to build and test hypotheses on how gene expression robustness is achieved in eukaryotic cells, and how relevant to this robustness is the presence of multiple promoters.