Development of drug-inducible CRISPR-Cas9 systems for large-scale functional screening

Abstract

Background: Large-scale genetic screening using CRISPR-Cas9 technology has emerged as a powerful approach to uncover and validate gene functions. The ability to control the timing of genetic perturbation during CRISPR screens will facilitate precise dissection of dynamic and complex biological processes. Here, we report the optimization of a drug-inducible CRISPR-Cas9 system that allows high-throughput gene interrogation with a temporal control.

Results: We designed multiple drug-inducible sgRNA expression vectors and measured their activities using an EGFP gene disruption assay in 11 human and mouse cell lines. The optimal design allows for a tight and inducible control of gene knockout in vitro, and in vivo during a seven-week-long experiment following hematopoietic reconstitution in mice. We next performed parallel genome-wide loss-of-function screens using the inducible and constitutive CRISPR-Cas9 systems. In proliferation-based dropout screens, these two approaches have similar performance in discriminating essential and nonessential genes. In a more challenging phenotypic assay that requires cytokine stimulation and cell staining, we observed similar sensitivity of the constitutive and drug-induced screening approaches in detecting known hits. Importantly, we demonstrate minimal leakiness of our inducible CRISPR screening platforms in the absence of chemical inducers in large-scale settings.

Conclusions: In this study, we have developed a drug-inducible CRISPR-Cas9 system that shows high cleavage efficiency upon induction but low background activity. Using this system, we have achieved inducible gene disruption in a wide range of cell types both in vitro and in vivo. For the first time, we present a systematic side-by-side comparison of constitutive and drug-inducible CRISPR-Cas9 platforms in large-scale functional screens. We demonstrate the tightness and efficiency of our drug-inducible CRISPR-Cas9 system in genome-wide pooled screening. Our design increases the versatility of CRISPR-based genetic screening and represents a significant upgrade on existing functional genomics toolbox.

Keywords: CRISPR; Functional genomics; Gene editing.

Fig. 1

Design and evaluation of drug-inducible sgRNA expression vectors. (a) Schematic for drug-inducible sgRNA expression vectors. Cas9 is constitutively expressed in the cells. EGFP reporter gene is used for the quantification of genome editing activity. PAC encodes puromycin N-acetyltransferase. (b) Representative flow cytometry histograms showing dose-dependent inducible EGFP knockout in MC-38 for tet- (left) and lac- (right) systems. (c) Evaluation of background activity and drug inducible gene knockout efficiency of the inducible sgRNA expression vectors in multiple cell lines. Data represent mean ± SD (n = 3). P values were derived from t tests: *P < 0.05; **P < 0.01; ***P < 0.001; NS, nonsignificant. (d) Calculation of leakiness score and activity score. (e) Heat map of leakiness scores. (f) Heat map of activity scores

Fig. 2

Further characterization of drug-inducible CRISPR platforms and benchmarking against literature designs. (a) Time-course of EGFP disruption activity after drug treatment. MC-38 cells with stable Cas9 expression were transduced with constitutive (grey lines), DOX-inducible (light and dark blue lines) and IPTG-inducible (light and dark red lines) sgRNA expression vectors and selected with puromycin. Parental MC-38 cells were transduced with an EGFP reporter as a control (green lines). 1 μg/mL DOX or 1 mM IPTG was used to induce the sgRNA expression. (b) RT-qPCR analysis of sgRNA levels in MC-38 cells. Chemical inducers were applied to cell culture medium on Day 0 and washed out on Day 2. 1 μg/mL DOX or 1 mM IPTG was used to induce the sgRNA expression. (c) RT-qPCR analysis of sgRNA levels with or without DOX treatment. 1 μg/mL DOX was used to induce the sgRNA expression. Data represent mean ± SD (n = 3). P values were derived from t tests: ***P < 0.001. (d) Comparison of background activity and inducible efficiency with existing (DD-Cas9 and DFHR.Cas9-DHFR) designs using the EGFP disruption assay. Induction conditions: 1 μg/mL DOX for 2xTetO system, 1 mM IPTG for 2xLacO system, 1 μM Shield-1 for DD-Cas9, and 10 μM TMP for DHFR.Cas9.DFHR. Data represent mean ± SD (n = 3). P values were derived from t tests: ***P < 0.001

Fig. 3

In vivo genome editing using the DOX-inducible sgRNA expression construct. (a) Schematic diagram of testing the DOX-inducible sgRNA expression vector in bone marrow chimeric mice. (b) Representative flow cytometry histograms showing DOX-inducible Cd44 knockout in splenic cells. (D) DOX-inducible Cd44 knockout in CD11b+, CD11c + and CD19+ cells in spleen. (d) DOX-inducible Cd44 knockout in CD11b + and CD19+ cells in bone marrow. (e) DOX-inducible Cd44 knockout in CD11b + and CD19+ cells in peripheral blood. Individual animals and mean (c-e) are shown with five mice per group. P values were derived from t tests: *P < 0.05; **P < 0.01; ***P < 0.001; NS, nonsignificant

Fig. 4

Genome-wide lethality screens using drug-inducible sgRNA expression constructs. (b) Schematic overview of proliferation-based negative-selection screening. (b) Normalized sgRNA read count distributions for essential and nonessential genes across different samples. CPM, counts per million. (c) Fold changes of normalized sgRNA read counts between cells after outgrowth and the original plasmid DNA. (d) Scatter plots showing the correlation between the lethality scores of essential (orange) and nonessential (green) genes in the screen using constitutive sgRNA library with those in the screens using inducible sgRNA libraries in the absence or presence of chemical inducers. Pearson correlation coefficient r values for essential genes are shown. (e) ROC curves indicating screen performance in identifying essential genes by comparing the library composition between the plasmid library and cells after > 10 population doublings. True positive rates and false positive rates were calculated using a gold-standard set of essential and nonessential genes. The ROC curves are based on the FDR. AUC, area under the curve

Fig. 5

FACS-based pooled CRISPR screens using drug-inducible sgRNA expression constructs. (a) Schematic overview of FACS-based CRISPR screening approach to identify PD-L1 expression regulators. (b) Regulation of PD-L1 expression by IFNγ signaling pathway. Known positive PD-L1 regulators including PD-L1 itself are shown in red. Known negative PD-L1 regulators are shown in blue. (c) Scatter plot for the result of constitutive CRISPR screen. Each dot indicates median log2 fold change of all sgRNAs for one target gene. (d) Scatter plots for the screening results using DOX- or IPTG-inducible sgRNA vectors. Each dot indicates median log2 fold change of all sgRNAs for one target gene. 1 μg/mL DOX or 1 mM IPTG was used to induce the sgRNA expression. (e,f) Frequency histograms of the changes of sgRNA abundance in PD-L1^low (e) and PD-L1^high (f) cells versus pre-sort. sgRNAs targeting known PD-L1 positive regulating genes are shown by red and sgRNAs targeting known PD-L1 negative regulating genes are shown by blue