April 6, 2017, by Amy E. Blum, M.A.
Determining gene function is critical to understanding cancer, and new technologies are allowing researchers to systematically analyze gene expression changes with increased precision. One such technology uses catalytically inactive Cas9 proteins guided by single guide RNAs (sgRNAs), which together bind near the gene promotor to either interfere (CRISPRi) or activate (CRISPRa) gene transcription.
Two libraries of sgRNAs developed in 2014, one designed for CRISPRi and one for CRISPRa, can significantly modify the expression of target genes. The research team from the Cancer Target Discovery and Development Network (CTD2) center at the University of California, San Francisco, that developed these libraries recently revisited them. Their new work, published in eLife, increased the precision and the efficiency of their libraries using a machine learning approach. The model that the investigators developed for predicting the effectiveness of sgRNAs, and their next-generation CRISPRi and CRISPRa libraries, are impressive, publicly available gene manipulation tools that may help researchers better understand cancer.
Optimizing Guide RNAs
Building from a discovery in their lab that nucleosomes influence CRISPRi and CRISPRa systems, the researchers created a model that would take this and other variables into account when predicting sgRNA activity. To generate the predictive model for CRISPRi, the researchers used a training set of 30 CRISPRi screens and calculated scores for each sgRNA in the screens based on their impact on transcription of their target gene. The resulting model was highly predictive of sgRNA activity in a portion of the test set that was kept separate from the training model, and in an independent CRISPRi data set. The model also showed which parameters contributed most the prediction, indicating that the site of the target relative to the transcription start site (TSS), sgRNA sequence features, sgRNA secondary structure, chromatin accessibility, and other attributes contributed the likelihood of robust sgRNA activity. The researchers then used the same method to generate a model for predicting the activity of sgRNAs for CRISPRa screens.
A Next-Generation Library
Using these models, the researchers improved upon their genome-scale sgRNA libraries targeting the human protein-coding transcriptome. When tested in a screen for essential genes in the chronic myeloid leukemia cell line K562, the new libraries, which they called CRISPRi and CRISPRa version 2, were found to be more efficient and significantly enriched for highly active sgRNAs. The investigators increased the sensitivity and specificity of their original libraries while decreasing the number of sgRNAs from 10 per gene to five.
Furthermore, the researchers demonstrated that their method does not induce off-target toxicities, an advantage over many other CRISPR-based methods. Based on a test of CRISPRi in comparison to CRISPR nuclease, they concluded that CRISPR nuclease-induced double strand breaks may cause toxicities that weaken the sensitivity of these assays compared to CRISPRi, which does not damage DNA.
This study provides a framework for refining CRISPR methods using computational insights, an improved understanding of the factors that contribute to CRISPRi and CRISPRa precision, and a widely applicable sgRNA library for use by the research community.
Gilbert, L.A., Horlbeck, M.A., Adamson, B. Villalta, J.E., Chen, Y., Whitehead, E.H., Guimaraes, C., Panning, B., Ploegh, H.L. Bassik, M.C., et al. (2014) Genome-Scale CRISPR-Mediated Control of Gene Repression and Activation. Nature. 139:3 647-661. http://dx.doi.org/10.1016/j.cell.2014.09.029
Horlbeck, M.A., Gilbert, L.A., Villalta, J.E., Adamson, B., Pak, R.A., Chen, Y., Fields, A.P., Park, C.Y., Corn, J.E., Kampmann, M., et al. (2016) Compact and Highly Active Next-Generation Libraries for CRISPR-Mediated Gene Repression and Activation. eLife. 5:e19760. http://dx.doi.org/10.7554/eLife.19760