NEW YORK (GenomeWeb) – A new CRISPR/Cas9-based screening method has identified thousands of functional long non-coding RNAs (lncRNAs) in human cell types.
Researchers at the University of California, San Francisco, led by co-first authors Max Horlbeck and John Liu and co-senior authors Daniel Lim and Jonathan Weissman, conceived a CRISPR transcription interference (CRISPRi)-based screen targeting more than 16,000 lncRNA loci in seven different human cell lines.
They identified 499 lncRNAs required for cell growth, of which almost 90 percent functioned exclusively in one cell type.
This figure greatly expands the number of lncRNAs previously known to have function in human cells, Lim told GenomeWeb, adding that while there are at least 100,000 known lncRNAs, which ones have function — and what that function is — is not well understood.
"The screen is a screen: It found things that have function," he said. "How they function is a mystery. Many of them will have important roles in and outside the nucleus, he said, adding that the field might be able to begin predicting lncRNA function as more examples are discovered.
The researchers published their results today in Science.
"We're very interested in what lncRNAs are doing and which are functional," said Horlbeck, a doctoral student at UCSF. "The only way to answer that in a comprehensive way is to do this systematically at genome scale and across cell lines."
Of the genomic transcripts that do not code for proteins, lncRNAs are defined as those greater than 200 bases. Research into the area has been ongoing, Lim said, but only a few lncRNAs have become well known for their function. Xist, which silences the X chromosome in female cells, may be the most famous.
"Some lncRNAs have function in the transcript itself," Lim said. "There's emerging data that suggest lncRNAs have functions that are enhancer-like activities."
But to figure out function, the researchers needed to expand the candidate list. Lim said his lab collaborated with Weissman's from the get-go.
They chose to use a CRISPRi-based screen — where a nuclease-null Cas9 blocks transcription — because using the DNA-cleaving function of Cas9 to induce indels doesn't block lncRNA biological activity, they said. In October, a study published in Nature used a deletion-based strategy to find 51 functional lncRNAs affecting cell growth. CRISPRi also empowered the UCSF to test both cis- and trans-acting capabilities of lncRNAs, Lim said.
Preliminary work involved inserting the CRISPRi machinery into all the cell lines and getting baseline levels of lncRNA expression in those lines. Liu said that when the team didn't do its own sequencing, it mined available data for the cell lines in question.
Designing the library of DNA-targeting guide RNAs (gRNAs) involved a machine learning algorithm designed by Horlbeck to predict highly active gRNAs that would suppress transcription of each of the 16,401 lncRNA targets. Nucleosomes directly block Cas9 access to DNA, so he incorporated data from other publications into his algorithm to predict which guide sequences would be most potent.
The researchers designed 10 guides against each target and used high-throughput oligo synthesis. Assaying for cell growth also allowed the gRNA sequence to be used as readout, since they were integrated into the cells' genomes.
"Extensive" validation confirmed the results of the cell proliferation assay, the researchers said. Moreover, repressing particular lncRNA led to specific changes in the transcriptome. In follow-up experiments looking at knockdown of 42 lncRNAs in three cell types, the researchers performed RNA sequencing. The transcriptional profiling suggested that lncRNA can regulate critical pathways involving genes associated with DNA replication, translation, and regulatory functions.
"What was surprising is that [lncRNAs] were incredibly cell type-specific in their function," Lim said. He pointed to one example, LINC00253, which triggered broad gene expression changes in U87 glioblastoma cells — both up- and down-regulation of specific transcripts, but not HeLa cells, where the data showed it was knocked down, but without the accompanying transcriptional changes.
"If you do a CRISPR screen for coding genes, 30 percent are absolutely essential in all the cell types," Lim said. "We did not find any examples like that. A given lncRNA seemed to have function in one cell type, maybe two, but not the others."
Lim said it was in an indication of the "rich biology" contained within the lncRNA genome. "There's tremendous diversity of function as well as cell type specificity," he said.
Lim said the labs will be working closely with nonprofit plasmid repository Addgene to make the gRNA library available for distribution. "What would be powerful for other researchers is to do the same screen in their cell lines of interest," he said. Looking at more complex cellular processes is another possible avenue of research.