The discovery of the CRISPR‐Cas system in bacteria has initiated an impressive array of innovations that have enabled the use of the RNA‐guided Cas9 nuclease in functional genomic screens. At Horizon, we have embraced these developments, as they provide new opportunities for drug target identification and validation. The case studies presented in this below highlight how we use this technology to successfully conduct genome wide and focused sgRNA library screens and to verify whether specific genes are required for the survival and/or proliferation of cancer cell lines.
Case Study 1: 6-TG Resistance Screen
A proof of concept genome-wide positive selection screen in eHAP cells, a fully haploid cell line engineered from HAP1 fibroblasts cells, was carried out to identify genes whose loss of expression induces resistance to the purine antimetabolite 6 thioguanine (6-TG)
- Based on relative guide abundance, NGS analysis revealed several genes whose loss of expression conferred resistance, including MLH1, MSH2 and MSH6, which have been previously documented as having a role in 6-TG resistance (Figure 2A).
- To validate these targets as 6-TG resistance factors, MLH1 and MSH6 knockout HAP1 cells were generated using a CRISPR-Cas9 approach. These cells show reduced sensitivity to 6-TG compared with the parental cell line (Figure 2B)
Figure 1: Identification and validation of 6-TG resistance factors in HAP1 cells. (A) Ranking of screen hits by the MAGeCK hit calling algorithm. On the y-axis, genes are ranked by robust ranking aggregation values for their enrichment after 6-TG treatment. The mean log2-fold change in sgRNAs targeting the same gene are plotted on the x-axis. (B) Using CRISPR-Cas9, MLH1 and MSH6 knockout HAP1 cell lines were generated. Cells were seeded and treated with increasing concentrations of 6-TG, and the affect on cell proliferation assessed.
Case Study 2: Glucose Starvation Sensitivity Screen
A fully haploid cell line developed by Horizon was used to examine the power of CRISPR-Cas9 screens for identifying genes that increase sensitivity to a specific stimulus when knocked out (drop-out screening).
Loss of several genes increased the sensitivity of cells to glucose-depleted conditions. Use of an adjusted p-value for gene ranking (Li et al., 2015), identified several NADH-dehydrogenase enzymes and two stress-related tumour suppressors (TSC1 and TSC2) as potential hits. As expected, oxidative phosphorylation and electron transport chain encoding genes were also successfully identified in this screen.
Figure 2: Glucose Starvation sensitivity screen in eHAP cells. Ranking of screen hits by the MAGeCK hit calling algorithm, identifying genes that when lost sensitise eHAP cells to glucose starvation.
We also used the data from this screen to assess the penetrance of gRNA-Cas9 gene disruption. As anticipated, essential ribosomal genes were depleted. In contrast, non-essential ribosomal genes and the non-targeting control guides remained approximately constant.
Figure 3: Control guide performance during the glucose starvation screen. Mean LogFC between time zero and screen endpoint, aggregated by either gene cohort, individual gene or individual gRNA.
Case Study 3: Synthetic Lethal Targets in Colon Cancer
Identifying novel cancer-specific vulnerabilities or synthetic lethal interactions with proteins that are frequently mutated in cancer could yield alternative yet effective targets for cancer treatment. These types of interactions are difficult to predict, but using CRISPR-Cas9 we are able to perform unbiased large-scale functional genomic screens to identify such targets.
To perform these studies we have built a custom sgRNA library based on the druggable genome (Figure 4a), containing 10 sgRNA sequences per target gene and the requisite positive and negative control sequences.
Figure 4: The tools for identifying synthetic lethal targets in colon cancer. (A) Breakdown of Horizon Discovery’s druggable genome (DG+) sgRNA library, showing the target classes represented (control guides not shown). (B) Cancer and isogenic cell line panel selection for sgRNA screening
Following screen optimisation, each cell line was infected with the DG+ library and maintained at a minimum of 300-fold coverage for approximately 12 cell doublings.
Using the isogenic cell lines we can identify targets that show selective lethality in the presence of the activating PIK3CA E545K mutation found in DLD1 parental cells. To do this we followed the approach set out in Wang et al 2014, to identify genotype-specific essential genes.
We identified several genes that appear specifically essential in cells harbouring mutant PIK3CA, including PIK3CA itself and the downstream kinases AKT1 and AKT2, indicating an increased reliance on this pathway. Conversely, EGFR was found to be selectively essential in DLD1 isogenic cells that only contain a single wildtype PIK3CA allele, but is dispensable in cells with the PIK3CA E545K mutation, which is downstream of EGFR.
Figure 5: Comparison of gene essentiality between DLD1 parental (PIK3CA E545K/+) and isogenic (PIK3CA +/-) cells. To identify PIK3CA-essential genes, the log2 sgRNA fold change distribution of each line was mean-normalised to zero. For each sgRNA, a differential essentiality score was then defined as the average log2 fold change in the DLD1 PIK3CA E545K/+ parental line subtracted by the average log2 fold change in the PI3KCA +/- isogenic line.
The application of CRISPR-Cas9 technology to whole genome screens is revolutionising our ability to perform target identification experiments. Devoid of the caveats associated with siRNA and shRNA reagents, the hope is that novel targets can be uncovered and rigorously validated using CRISPR-Cas9, and that a series of innovative validated targets will enter the drug discovery pipeline.