Abstract
Background: RNA-seq and scRNA-seq identify unannotated transcripts across cell types; whether they are coding or noncoding, normal or erroneous, and functional or junk remains largely uncharacterized. Unannotated noncoding transcripts (UNTs) are of special interest because ncRNAs can function by directly interacting with other molecules. Previous studies reported several functional UNTs in tumors, but none systematically examined their extent and mechanisms. Results: We first performed a pan-cancer analysis, including cell lines and tissues of four tumors, to identify unannotated genes and transcripts (including UNTs). Many unannotated genes and transcripts show significant differential expression. We then predicted potential DNA-binding domains (DBDs) within UNTs and their DNA-binding sites (DBSs) in the promoter regions of differentially expressed genes (DEGs). To reveal whether and how UNTs regulate transcription, we investigated three UNTs (labeled as MSTRG.2513.6, MSTRG.4401.1, and MSTRG.34636.7) whose genomic regions overlap BCAN-AS2, LNCAROD, and LINC00513. We deleted their predicted DBD in two cancer cell lines using CRISPR/Cas9, performed RNA sequencing and cell experiments before and after DBD deletion, identified DEGs using three methods, analyzed enriched signaling pathways, and analyzed transcriptional regulatory modules. DBD deletion causes not only differential gene expression but also altered cell phenotypes. Conclusions: The results suggest that many UNTs can transcriptionally regulate genes, and this regulation is highly cancer- and species-specific. Because UNTs are widely expressed in cancer and other diseased cells, they may be an important class of transcriptional regulators. Their cell and species specificity may help explain inconsistencies across studies and species and make them potential disease-specific targets.