Evaluation of Computational and Experimental Parameters in RNA Bisulfite Sequencing Analysis and Applications in Brain Development Studies
Epitranscriptomics, the study of RNA modifications, has become a hotspot of research over the last decade. Over 170 unique modifications have been discovered with a widespread occurrence in a diverse range of RNAs. 5-methylcytosine, m5C, is an evolutionarily conserved and reversable modification that regulates the stability and export of tRNAs, rRNAs, and mRNAs. m5C has recently been implicated in many biological phenomena including tumorigenesis, embryonic cell expansion and differentiation, brain development, and neuronal functions. While we are just beginning to understand the functions of m5C, a gold standard of m5C detection has yet to be established due to the low signal-to-noise presence of m5C. In this work, we utilize RNA bisulfite sequencing as a transcriptome-wide approach to understand the computational and chemical parameters needed to optimize m5C discovery in the mitochondria and the developing brain. In Chapter 1, we systematically evaluate four preparation conditions of bisulfite sequencing to identify potential presence of m5C-mRNAs localized to the mitochondria in neuronal stem cells. In tandem, we utilize unique molecular identifiers and a consortium of control template transcripts to evaluate sources of false positive m5C sites that may emerge from sequencing errors, PCR amplification, and the inadequate bisulfite conversion of transcripts. While improvements to mitochondrial transcript bisulfite conversion and false positive filtering were observed, no mitochondrial mRNAs were identified to be methylated, indicating no or very few methylated cytosines in mitochondrial mRNAs and the need for improved chemical methods to detect mitochondrial m5C-mRNAs if any. In Chapter 2, we employ the computational approaches established in Chapter 1 to survey the m5C landscape of the developing mammalian brain. We discover a general increase in unique m5C sites in mouse whole brain tissue when compared to neuronal cell cultures. Of these sites, we found the post-natal day 0 and 17 brain time points to undergo significant methylation level changes in comparison to the 6-week-old brain. These differentially methylated sites were significantly enriched for brain development, synaptic development, and transcriptional control gene network pathways. In Chapter 3, we expand on our findings in Chapter 2 to understand the impact of m5C reader FMRP and m5C eraser TET1 loss in the mouse post-natal day 17 brain. Among a set of m5C sites identified in wildtype or knockout samples, few were differentially methylated after protein ablation, suggesting m5C may rely on compensatory enzymes. Using FMRP-RNA pulldown assays to validate FMRP binding positions, we identified Ralbp1 to be hypermethylated and overexpressed in Fmr1-KO brain tissues. RalBP1 is a binding protein responsible for the endocytosis of AMPA receptors, a process critical for neuronal long term depression and brain development.