NEW YORK (GenomeWeb) – A new software tool from researcher Rory Johnson of the University of Bern is helping scientists uncover the genome-wide function of long non-coding RNAs.
It's a web-accessible tool to design CRISPR/Cas9-based deletion screens for lncRNA promoter regions, called CRISPETa. The software designs a pair of guides to snip out a region in the promoter between 300 and 3,000 bases long, potentially for thousands of genes at a time, enabling the creation of enormous guide RNA (gRNA) libraries for high-throughput screening.
Rosella Pellegrino, a researcher at University Hospital RWTH Aachen who has used CRISPETa, said in an email that the tool was useful and well designed. "You don't have to spend much time to understand how to set up the tool and interpret the results," she said.
Johnson's lab has been using the tool internally, and alongside scientists from Barcelona's Centre for Genomic Regulation described the software and its validation in a paper published last week in PLoS Computational Biology.
The field of lncRNAs has undergone explosive growth in the last five years, Johnson said in an interview. "A lot of groups studying a particular disease have realized there could be lncRNAs in that process. The big question now is the question of function. This CRISPR tool will allow us to settle the question definitively."
While there are plenty of tools to score gRNA efficiency and accuracy, until last year there was not a tool to design them in pairs for deletion, an alternative strategy to perturbing lncRNAs by CRISPR-based transcriptional repression. "When we started, we had to do this by hand," Johnson said. With CRISPETa, he or any other researcher can scale up design of screening libraries with almost no upper limit.
"CRISPETa was the necessary step to begin performing genome-wide screens for the function of lncRNAs," he said. "Instead of working with one or two genes at a time, that scale enables us to screen hundreds or thousands of genes at a time." Previously, gene expression studies were the best way to interrogate lncRNA function in healthy or diseased tissues. "With this approach we can look at the cellular function of lncRNAs in complex human diseases," Johnson said.
The CRISPETa software tool followed the development of a plasmid vector Johnson's lab developed for its deletion experiments, called double excision CRISPR knockout (DECKO), published in 2015 in BMC Genomics. In that paper, Johnson saw the potential for phenotypic screens using CRISPR gRNA libraries, but knew that design would be the limitation without a bioinformatics pipeline capable of designing thousands of gRNAs at a time.
CRISPR-based screening has been a rapidly expanding field itself. For lncRNAs, DECKO and CRISPETa are far from the first tools available. Last October, scientists led by Dana Farber Cancer Institute researcher Shirley Liu published a study in Nature Biotechnology on a genome-scale, paired-guide deletion screen using CRISPR/Cas9. They too devised an algorithm for designing their library, published to online code repository Bitbucket.
While the Liu lab's program is run from the command line, CRISPETa is presented in a more user-friendly format online. "[CRISPETa] gives you the possibility to visualize directly where your guides are targeting the genome," Pellegrino said.
Like many gRNA design tools, CRISPETa considers both on-target efficiency and potential off-target accuracy. For on-target scoring, it uses an algorithm from Broad Institute researcher John Doench, considered the gold-standard in the field.
Off-targets, while important to consider, aren't so much of a problem for screens, Johnson said.
"We allow the user to select different thresholds of off-target similarity," Johnson said. "If you use stringent parameters, you often cannot design a gRNA pair within the region which you desire. We made everything very flexible."
Johnson said that how to handle off-targets is something his lab discussed a lot. "Our research is basic research in cell lines and therefore we're not so concerned about off-target problems," he said. "If you're working with pairs [of gRNAs], it reduces more the possibility you will have serious off-target effects. I don't see it as a major problem," though he added that off-targets become very important when considering CRISPR for therapeutic purposes.
After removing low-efficiency and likely off-target sites, CRISPETa generates a list of all possible pairs flanking the target sites, and returns a ranked set of pairs from which the researcher can choose one or several to add to the library. Pellegrino said that the software provides a pairwise ranking, which she called a "good parameter" for selecting guides.
The pairwise approach has been essential to the research Johnson is conducting. While many gRNA libraries use a single gRNA to target protein coding genes, that kind of approach doesn't work for lncRNAs. "You can use a single guide to knockout a protein because one gRNA introduces frameshifting indels," he said. But lncRNAs don't have an open reading frame. "A small indel not likely to have much impact on the function of the gene," he said.
Though gRNA libraries with one gRNA per construct won't work for lncRNA screening, Johnson thinks that his paired gRNA platform could be useful for research on protein coding genes and is working on a version for them.
"We can improve efficiency of knocking out a gene by hitting it with two independent guides in its protein coding sequence," he said. Some genes are difficult to knockout with a single guide; for these, DECKO and CRISPETa could help out, he said, adding that deletion-based screening is versatile and could help in research of other non-coding elements.
He's got other improvements in mind as well. On- and off-target scoring algorithms are their own fast-moving field of research.
"Already we want to implement a new and improved scoring system which was published just before our paper was accepted," Johnson said. Yet, he's unsure how much impact it will ultimately have. "I have a feeling it'll only be a small incremental improvement. The real improvement in these methods will come from experimentally improving the efficiency of DNA deletion and informatics may not yield so much improvement," he said.