Abstract 4503: Functional CRISPR screen towards identifying novel conserved long noncoding RNAs with oncogenic activity

Abstract

Large-scale transcriptome analysis has enabled the systematic profiling of long non-coding RNAs (lncRNAs) in cancer-or lineage-specific contexts. Although a significantly larger portion of the genome is transcribed into lncRNAs compared to proteins, only a small set of them appears to be evolutionally conserved across organisms. In general, functionally essential genes are subject to stronger selective pressure, and are thus more evolutionarily conserved than nonessential genes. However, it remains a question whether the evolutionally conserved lncRNAs play essential roles for cell growth and development. Recently, our lab described THOR, the first evolutionarily conserved lncRNA that is testis/cancer-specific and accelerates tumor growth in preclinical models. Based on these discoveries, we hypothesized that a subclass of conserved lncRNAs that are functionally essential for tumor growth and cancer progression should exist. To test this hypothesis, we first developed a top-down approach to generate a list of cancer-associated evolutionary conserved lncRNAs. By integrating base-pair conservation, best 200-bp conservation, and transcript expression information, we identified a total of 85 ultra-conserved lncRNAs that are expressed in cancer. We then set up a CRISPR-Cas9 based system to screen for functionally essential lncRNAs in various lineages of cancer. Unlike protein-coding genes, whose expression can be efficiently disrupted by conventional single guide RNA-based CRISPR-Cas9 system, lncRNA transcripts are insensitive to read-frame alterations. We thus utilized an lncRNA deletion strategy featured by paired guide RNAs flanking the conserved domains within the lncRNA transcript. We designed a total of 858 pairs of guide RNAs (gRNAs) targeting 242 conserved domains within the 85 ultra-conserved lncRNAs and protein-coding gene controls. Sequencing validation of the library confirms more than 90% coverage of the designed gRNA pairs. Using this library, we have performed functional CRISPR screens in cancer cell line models from various lineages, namely, MIA-PaCa-2 (pancreatic ductal adenocarcinoma), SK-N-Mc (neuroepithelioma), and VCaP (prostate cancer). Notably, we have successfully discovered both novel and previously reported oncogenic lncRNAs (e.g. ZNF503-AS1), which, upon knockout, significantly delayed cell growth (represented by the diminished count of guide RNAs targeting these genes in the cell population). Therefore, from these CRISPR screens, we discovered ultra-conserved lncRNAs with suggestive oncogenic roles in a lineage-specific or pan-cancer manner. Further interrogation of their structures, cellular locations, and interactomes is warranted to reveal their contributions to cancer progression.