Identifying Repressive RNAs with Xist-Like Functions in the Mouse Transcriptome

Long non-coding RNAs (lncRNAs) play major roles in gene regulation and human health. Xist, the lncRNA responsible for X chromosome inactivation, has been studied for 30 years, but only a few repressive lncRNAs with similar functions have been discovered, including Airn and Kcnq1ot1. One notable obstacle in lncRNA discovery is a poor understanding of the lncRNA sequence-function relationship. To study this problem, we hypothesized that there exists a class of repressive RNAs that share 2 sequence features: k-mer content (the abundance of short motifs), and protein-binding profile (binding affinity with proteins important for Xist function). In my analysis of mouse trophoblast stem cells, I compared the k-mer content and protein-binding profile of expressed, chromatin-enriched RNAs to Xist, Airn, and Kcnq1ot1. I used the algorithm SEEKR to evaluate k-mer similarities and a computational pipeline I created to calculate protein-binding similarities. I found that Airn and Kcnq1ot1 share more similarity in protein-binding profile and k-mer content than Xist. Furthermore, k-mer content correlates with protein-binding positively for similarities to the unspliced Airn and Kcnq1ot1, but negatively to the spliced Xist. We infer that such contradiction to our hypothesis is due to the confounding effect of splicing, suggesting the need to further tailor the list of RNAs in search of gene repressors. Future directions include removal of RNAs with short half-life in the chromatin region, analysis of regional k-mer content, and biochemical testing of true functions of candidate RNAs.