Target identification of small molecules using large-scale CRISPR-Cas mutagenesis scanning of essential genes

Unraveling the mechanism of action and molecular target of small molecules remains a major challenge in drug discovery. While many cancer drugs target genetic vulnerabilities, loss-of-function screens fail to identify essential genes in drug mechanism of action. Here, we report CRISPRres, a CRISPR-Cas-based genetic screening approach to rapidly derive and identify drug resistance mutations in essential genes. It exploits the local genetic variation created by CRISPR-Cas-induced non-homologous end-joining (NHEJ) repair to generate a wide variety of functional in-frame mutations. Using large sgRNA tiling libraries and known drug–target pairs, we validate it as a target identification approach. We apply CRISPRres to the anticancer agent KPT-9274 and identify nicotinamide phosphoribosyltransferase (NAMPT) as its main target. These results present a powerful and simple genetic approach to create many protein variants that, in combination with positive selection, can be applied to reveal the cellular target of small-molecule inhibitors.

Amino acid sequences of the corresponding Cas9 mutagenized and drug resistant HCT 116 cells as determined by CrispRVariants analysis of targeted amplicon sequencing of the edited locus. Residues known to provide drug resistance upon mutation are highlighted with blue columns. The Cas9 cut site is highlighted by the vertical dotted line. Only variants detected with a read frequency ≥1% are shown. c.
Overview of the abundance of the mutation types detected in mutagenized and drug resistant HCT 116 cells. The relative abundance of each mutation type is shown and categorized per sample. Only one experiment per sample was performed. Results were obtained by targeted amplicon sequencing analysis with CrispRVariants of the genome edited locus (ERCC3: locus S162, KIF11: locus A133, XPO1: locus C528).   Amino acid sequences of the corresponding Cas9 mutagenized and drug resistant HL-60 cells as determined by CrispRVariants analysis of targeted amplicon sequencing of the edited locus. Residues known to provide drug resistance upon mutation are highlighted with blue columns. The Cas9 cut site is highlighted by the orange, vertical dotted line. Only variants with a read frequency ≥1% are shown. c.
Overview of the abundance of the different mutation types in the mutagenized HL-60 cells. The relative abundance of each type is shown and categorized per sample and only 1 experiment was performed for each sample. Results were obtained by targeted amplicon sequencing analysis with CrispRVariants of the genome edited locus in selected cells (ERCC3: locus S162, KIF11: locus A133, XPO1: locus C528).   Sequencing chromatograms of HDR-edited HAP1 cells after treatment. Silent mutations were incorporated to control for HDR. c.

Supplementary
Cell viability assays showing the effect of selinexor, ispinesib or triptolide on wild-type (parental) and respective HDR-edited HAP1 cells after 72h. Data points are normalized to untreated cells and represent averages ± s.d. obtained from two experiments performed in triplicate.

d.
Confocal fluorescence imaging of parental and HDR-edited XPO1 HAP1 cells transfected with the NLS-AcGFP-NES XPO1 cargo (green). In addition, some cells were cotransfected with SpCas9 and a sgRNA pool targeting XPO1 to induce knockout of XPO1 (panel b). Two days after transfection, cells were treated with DMSO (panel a) or 2 μM selinexor (panel c) for 3 hours and then imaged. Scale bars: 25 μm. e.
Pulldown of XPO1 from wild-type and mutant XPO1 HDR-S528VHI HAP1 cells. Cells (2.5 x 10 6 ) were collected and lysed in RIPA buffer after treatment with DMSO or 2 µM KPT-9058 for 3 hours. KPT-9058 was extracted overnight at 4°C by streptavidin affinity chromatography (Dynabeads MyOne Streptavidin T1). The total lysate (input) and extracted fractions (pulldown) were then separated on a Simple Wes system. Indicated proteins were detected by chemiluminescent immunoblotting using primary and secondary-HRP conjugated antibodies.  Design sgRNA CDS tiling library  Frameshift Figure 9: proof of concept for the identification of a drug's target protein from a pool of 9 genes using the ispinesib-KIF11 drug-target interaction a.
Overview of the workflow for the CRISPR/Cas9-based chemical target identification screen used for ispinesib. b.
Cell viability after 72h of parental, Cas9 + and the polyclonal pool of drug resistant mutagenized HAP1 cells treated with different concentrations of ispinesib. Data points represent means ± standard deviation obtained from two experiments performed in triplicate.

c.
Representation of the different sgRNAs in cells before (after puromycin selection) and after treatment with 8 nM ispinesib as determined by EdgeR analysis of next generation sequencing data. Each dot represents a different sgRNA and a value of 1 was added to the read count to facilitate log transformation. Fold change was determined by dividing the adjusted read count per million reads (RPM) of the sgRNAs enriched after ispinesib treatment by the adjusted read count per million reads of the sgRNAs detected after puromycin selection. d.
Overview of sgRNA hits with a fold change >100 identified in surviving cells 14 days after ispinesib treatment. All sgRNAs targeted KIF11. e.
Enriched sgRNAs were cloned individually and transfected into parental HAP1 cells stably expressing SpCas9 to validate the results. Three days after transfection, cells were treated with 10 nM ispinesib for 5 days and surviving cells were counted using trypan blue exclusion.  Chemical structure of FK866. c.
The  Sequencing chromatogram of NAMPT in HDR-edited HAP1 cells stably expressing SpCas9. Cells were transfected with a NAMPT targeting sgRNA and a HDR donor template containing the G383del mutation together with a silent mutation not identified in our screen to control for HDR. Two days after transfection, cells were treated with 300 nM KPT-9274 for 1-2 weeks. b.
Cell viability assay showing the effect of KPT-9274 on wild-type, Cas9+ (parental) and polyclonal G383del HDR-edited HAP1 cells after 72h. Data points are normalized to untreated cells and represent averages ± s.d. obtained from three experiments performed in duplicate.

c.
Cell viability assay showing the effect of FK866 on wild-type, Cas9+ (parental) and polyclonal G383del HDR-edited HAP1 cells after 72h. Data points are normalized to untreated cells and represent averages ± s.d. obtained from three experiments performed in triplicate.

d.
Knockout of NAMPT using CRISPR/Cas9-mediated genome editing shows that NAMPT is important for cell survival. HAP1 or K-562 cells stably expressing SpCas9 were transfected with sgRNA plasmid pools targeting the indicated gene. The following day, cells were treated for 24 hours with puromycin to select for transfected cells. Four days after selection, surviving cells were counted using trypan blue exclusion. KIF11 is an essential gene and served as positive control, while non-targeting sgRNAs were used as negative control. Values represent means ± s.d. obtained from two independent experiments. The electron density map of the ligand (yellow) in the NAMPT binding site at 1σ shows a perfect match between the ligand and the observed density.

b.
Data collection and refinement statistics (molecular replacement).  Schematic overview of AsCpf1-mediated mutagenesis near the ERCC3 D54 locus. Asp54, known to provide resistance upon mutation is highlighted in blue. The crRNA cutting site is denoted by red arrows.

b.
Cell viability after 72h of wild-type parental and polyclonal mutagenized HAP1 cells treated with different concentrations of triptolide. Signals are plotted relative to the untreated control and data points represent means ± s.d. obtained from three experiments performed in triplicate.

c.
Amino acids variants identified in the triptolide resistant HAP1 cells mutagenized at the ERCC 54 locus. Only reads with a read frequency above 3% are shown. Variants were obtained by CrispRVariants analysis of next-generation sequencing reads. Amino acids Ala41 to Ser49 are shown and no mutations were detected at Asp54.