Preview

PDF, English - main document Download (16MB) | Terms of use

Abstract

The usage of single cell sequencing techniques has grown extensively over the last years. One commonly utilized platform to conduct these experiments is the one from 10X Genomics. The applied workflow enables the detection of gene expression in single cells. Importantly, standard analysis tools count the gene but do not consider its exact mapping position within the transcript. This can be considered as a missed opportunity since mRNAs are captured by oligo(dT) primers and sequenced from the most 3’ position of the transcript (3’ capturing or tagging). These 3’ends are biologically relevant as one gene can have 3’ untranslated regions (3’UTRs) of alternative lengths, also referred as alternative polyadenylation (APA). From multiple polyadenylation singles within one gene, one signal is selected by the cleavage and polyadenylation machinery terminating mRNA transcription. This process yields mRNAs with shorter and longer 3’UTRs but the same protein coding sequence. In order to make use of this information I developed a bioinformatical pipeline to call 3’peaks in sequencing data with single cell resolution and analyzed alterations in APA by multinomial regression (MNR). I applied this method on the neural stem cell (NSC) lineage of the adult mouse. This system resembles the differentiation process from a quiescent stem cell into a neuroblast. Interestingly, genes altering their poly(A) site choice with lineage progression were enriched for risk genes in neurodevelopmental disorders, amongst others, Autism spectrum disorder (ASD). Analyzing 3’UTR usage in ASD patients and a control group revealed a trend for 3’UTR lengthening in individuals diagnosed with ASD compared to controls. Motif analysis in the murine and the human sequencing data further pointed to an enrichment of the cytoplasmic and polyadenylation element (CPE) in 3’UTRs. These motifs are recognized by CPE binding proteins. RNA immunoprecipitation showed that CPEB4 can bind this motif. Proteomics as well as ribosomal profiling data was utilized to estimate the effects of CPEB4 binding and 3’UTR shortening on mRNA translation. Intriguingly, co-expression of CPEB4 and amyloid beta precursor-like protein 1 (APLP1), a synaptic adhesion molecule which can bind CPEBs, was observable in the mouse and the human single cell sequencing data. Finally, comparing APLP1 knockout to wildtype mice revealed CPE-dependent alterations in 3’UTR usage between both genotypes. In summary, the results of the here developed method for 3’UTR calling shows that 3’peaks, single poly(A) signals, can be successfully detected in 10X Genomics data. In addition, the downstream analysis suggests a link between APA, neuronal differentiation, CPEB4 and the neurodevelopmental disorder ASD.