Research

Predicting CRISPR/Cas9 on-target and off-target effects

The CRISPR/Cas9 system is a powerful tool for editing the target genome. The success of CRISPR/Cas9 applications is highly dependent on the on-target efficiency and off-target effects. The Xu Laboratory combines high-throughput experiments with computational approaches to systematically define the rules that govern and the molecular mechanisms that underlie these effects. These projects lead to the development of new computational algorithms for accurate prediction of CRISPR/Cas9 on-target and off-target effects, ultimately providing more cost-effective methods for research and biomedical applications by defining more optimal targets.

In the image below,  panel (A) depicts a  sequence logo representing the preferred nucleotides at the target DNA site for efficient CRISPR/Cas9 gene editing. The top portion of panel (B) shows an agarose gel image of DNA fragment analysis using the SURVEYOR method. This method rapidly detects DNA mutations and single nucleotide polymorphisms (SNPs), and is based on using mismatch-specific nucleases that cleave at sites of small insertions/deletions (indels) and SNPs. The bottom portion of panel (B) is a quantification of the gel data displayed as a bar chart revealing the indel rates of sgRNAs that were predicted to be either efficient or inefficient, based on the sequence model in (A). For more details see Xu H et al, Sequence Determinants of Improved CRISPR sgRNA Design, Genome Research, 2015.

Screening for synthetic lethality between epigenetic regulators

Epigenetic regulators form complex regulatory networks to execute their functions. Some regulators are synthetically lethal with others. That is, cells with a deficiency in either one of two single regulators are viable,  but cells with deficiencies in both regulators die.  Because many epigenetic regulators are potent drug targets, defining synthetically lethal interactions could lead to novel therapeutic approaches for targeting specific vulnerabilities in cancer cells.

The Xu Laboratory is developing CRISPR/Cas9-based, high-throughput perturbation screens as well as analytical methods to systematically identify synthetically lethal interactions among epigenetic regulators. For example, they recently discovered a synthetic lethality between PRMT1 and PRMT5, two arginine methyltransferases that are often expressed aberrantly in many cancer types, but that were not known to influence each other.

The image below illustrates how the PRMT1-PRMT5 synthetic lethality was discovered. Panel (A) shows the workflow used in the CRISPR/Cas9 screen using the PRMT5 inhibitor EPZ015666 to knock out the function of PRMT5. Panel (B) shows the results of a computational analysis revealing that the loss of PRMT1 sensitizes cells to PRMT5 inhibition. Several other genes were also identified, some had been previously identified as being important to the PRMT5 pathway (e.g. WDR77), but novel interactions were also identified (e.g. INO80B). Panel (C) depicts a heatmap displaying synergy between PRMT1 (MS023) and the PRMT5 inhibitor EPZ015666, in the treatment of MIA PaCa-2 pancreatic cancer cells, based on Bliss scores. For more details please see Gao G and Zhang L, et al, PRMT1 Loss Sensitizes Cells to PRMT5 Inhibition, Nucleic Acids Research, 2019.

Repurposing CRISPR/Cas9 screens for protein domain analysis

CRISPR/Cas9 frequently introduces small insertion-deletions (indels) that can create either frameshift or in-frame mutations. Consequently, in CRISPR/Cas9 viability screens, the sgRNAs that target DNA sequences coding for essential protein domains often result in a more significant dropout phenotype in CRISPR-mediated screens compared to the sgRNAs that target non-essential protein domains in the same gene.  

Based on this observation, we have developed ProTiler, an approach that combines CRISPR/Cas9 tiling-sgRNA screens with computational algorithms to identify putative essential protein domains. The Xu Laboratory has applied this technology to predict functionally important protein domains in both epigenetic regulators and transcription factors. Many of the domains identified through this approach were previously identified as drug targets but many other identified domains are novel domains that were previously not known to be important. This new method will have broad applications for helping define uncharacterized proteins and will aid in drug target discovery.

The image below illustrates the rationale behind, and one example of, identifying essential protein domains using CRISPR/Cas9 tiling-sgRNA screens. Panel (A) illustrates the rationale for using CRISPR/Cas9 in protein domain analysis: mutations in functional/essential domains are more likely to cause a loss-of-function effect. Panel (B) shows the identification of essential domains in SMARCB1, an ATP-dependent chromatin remodeler, using ProTiler. These domains are marked as CKHS, or CRISPR knockout hypersensitive regions. A novel domain, that is frequently mutated in cancer, was detected in the N-terminus of SMARCB1, as was the previously defined ATPase domain. For details, see He W and Zhang L, et al, De novo Identification of Essential Protein Domains from CRISPR-Cas9 Tiling-sgRNA Knockout Screens, Nature Communications, 2019.