Synergistic drug combinations for cancer identified in a CRISPR screen for pairwise genetic interactions

Journal name:
Nature Biotechnology
Volume:
35,
Pages:
463–474
Year published:
DOI:
doi:10.1038/nbt.3834
Received
Accepted
Published online

Abstract

Identification of effective combination therapies is critical to address the emergence of drug-resistant cancers, but direct screening of all possible drug combinations is infeasible. Here we introduce a CRISPR-based double knockout (CDKO) system that improves the efficiency of combinatorial genetic screening using an effective strategy for cloning and sequencing paired single guide RNA (sgRNA) libraries and a robust statistical scoring method for calculating genetic interactions (GIs) from CRISPR-deleted gene pairs. We applied CDKO to generate a large-scale human GI map, comprising 490,000 double-sgRNAs directed against 21,321 pairs of drug targets in K562 leukemia cells and identified synthetic lethal drug target pairs for which corresponding drugs exhibit synergistic killing. These included the BCL2L1 and MCL1 combination, which was also effective in imatinib-resistant cells. We further validated this system by identifying known and previously unidentified GIs between modifiers of ricin toxicity. This work provides an effective strategy to screen synergistic drug combinations in high-throughput and a CRISPR-based tool to dissect functional GI networks.

At a glance

Figures

  1. Development of a CDKO platform to identify novel cancer drug combinations in high throughput.
    Figure 1: Development of a CDKO platform to identify novel cancer drug combinations in high throughput.

    (a) Drug combinations are modeled by developing a dual sgRNA system to simultaneously knock out the corresponding drug targets. (b) Design of i) a Csy4-based double-sgRNA expression system and ii) a dual-promoter-based system; flow cytometry analysis of GFP and mCherry knockout efficiency. RFU, relative fluorescence units (c) Schematic of double-sgRNA library generation. Two sets of pooled oligos were synthesized, amplified, and ligated into separate lentiviral vectors with either human or mouse U6 promoters. sgRNA cassettes were digested out from the hU6 library and ligated into the mU6 library as a pool to create the double-sgRNA CDKO library. (d) Design of drug-targeting sgRNA library. (e) Experimental strategy. Separate mU6-driven and hU6-driven single-sgRNA libraries were ligated together into a pooled dual sgRNA library. Cas9-expressing K562 cells were infected with the pooled high-coverage CDKO library, selected with puromycin, and grown in bioreactors at 1,000× coverage for ~14 d. The frequency of the dual sgRNA elements in the final population and original plasmid library were quantified by deep sequencing to calculate the growth phenotype and GI of each gene pair. Potent synergistic pairs were validated in vitro using pairs of drugs.

  2. 490,000-element DrugTarget-CDKO library can be efficiently assembled and delivered into cells and shows no positional bias.
    Figure 2: 490,000-element DrugTarget-CDKO library can be efficiently assembled and delivered into cells and shows no positional bias.

    (a) Cumulative distribution of sequencing reads for double-sgRNAs. Read counts were normalized by total reads of each sample and the cumulative sums of double-sgRNAs were plotted as relative percentages of the number of expected double-sgRNAs. (b) Histogram showing the number of double-sgRNAs per gene pair. 98.7% of the 42,319 detected gene pairs have more than 6 double-sgRNA combinations. (c) Growth (γ) phenotypes for two single genes and the corresponding gene pair were calculated from the double-sgRNA frequencies in the T14 sample and plasmid library. For γ-phenotypes of single genes, all possible double-sgRNA combinations of the 3 gene-targeting sgRNAs and 79 safe-sgRNAs were measured. For the γ-phenotype of a given gene pair, 9 double-sgRNA combinations were measured. Blue dotted lines mark the minimum threshold for read counts (50). (d) Minimal positional bias in the DrugTarget-CDKO library. γ-phenotypes of double-sgRNAs were compared between both orientations. PCC, Pearson correlation coefficient. (e) High reproducibility of measured γ-phenotypes between two experimental replicates.

  3. Strategy to calculate quantitative GI scores.
    Figure 3: Strategy to calculate quantitative GI scores.

    (a) Single-sgRNA γ-phenotypes plotted against corresponding double-sgRNA γ-phenotypes in combination with an sgRNA of interest. sgRNAs paired with a weak or moderate γ-phenotype sgRNA (sgFABP4_1 or sgKDM1A_1) showed a linear relationship between single-sgRNA and double-sgRNA γ-phenotypes. For sgRNAs paired with a strong γ-phenotype sgRNA (sgNAMPT_1), γ-phenotypes of double-sgRNAs quickly leveled off as γ-phenotypes of single-sgRNAs become more negative. Dotted lines mark polynomial fitting (n = 2). (b) Measuring GIs of double-sgRNAs as deviations from the expected double-sgRNA γ-phenotype. Observed and expected γ-phenotypes of all double-sgRNAs were plotted (black dots). Expected γ-phenotypes were calculated by sum of the two single-sgRNA γ-phenotypes. Medians of binned data were plotted (orange closed circle) and connected by a smooth median line (blue line). Deviations from the median line were defined as GIs. (c) Raw-GIs were plotted against the expected γ-phenotypes of double-sgRNAs (top panel). Positive raw-GIs are buffering while negative Raw-GIs are synergistic. Each Raw-GI was normalized by the s.d. of the 200 nearest neighbors in terms of the expected γ-phenotype. Normalized GIs (Norm-GI) were then plotted (bottom panel). In this plot, double-sgRNAs comprising only safe-sgRNA pairs (yellow dots) and all sgRNAs paired with safe-sgRNAs (purple dots) showed symmetric distributions centering around 0 Norm-GI, confirming that safe-sgRNAs generally do not interact with other sgRNAs. (d) T-value-based GIT scores plotted between two experimental replicates. GI scores of control pairs—genes paired with safe-sgRNAs (yellow dots) and Safe_Safe pairs (cyan closed circle)—showed negligible GIs as expected. Gene pairs comprising same genes are marked as pink dots. The five most synergistic pairs by rank-sum of GIT scores of two replicates are marked as dark blue dots. The Pearson correlation coefficient after same-gene targeting pairs were removed is reported in parentheses. PCC, Pearson correlation coefficient.

  4. A CRISPR-based GI map of ricin pathway regulators validates the CDKO platform.
    Figure 4: A CRISPR-based GI map of ricin pathway regulators validates the CDKO platform.

    (a) Correlations of GI profiles between two sgRNAs were compared between two experimental replicates. sgRNAs targeting the same gene are marked in pink. The Pearson correlation coefficient (PCC) after same-gene targeting pairs were removed is reported in parentheses. (b) The distributions of correlations of GI profiles for all sgRNA pairs (blue) and for sgRNAs pairs targeting the same gene (orange). Medians of the distributions are marked with dotted lines. (c) GI map of ricin modulators. GIM scores of all gene pairs were calculated and color-coded by a yellow-cyan heatmap. Genes were hierarchically clustered by their correlation of GI profiles. ρ-phenotypes of individual genes were marked in sidebars with a red-blue heatmap. Previously reported protein complexes are labeled. (d) Plot showing the percentage of the top N gene pairs that have corresponding protein interactions in the STRING database, sorted by correlation of GI profiles (orange), buffering GIs (blue), or synergistic GIs (brown) for known protein interactions. Randomly sorted gene pairs are marked in pink. Data were plotted for N > 10. (e) Relationship between the correlation of GI profiles and the predictive power for known protein interactions. Percentage of reported PPIs from STRING were averaged over a moving window of 20 data points against the correlation of GI profiles.

  5. Validation of predicted GIs using individual sgRNAs.
    Figure 5: Validation of predicted GIs using individual sgRNAs.

    (a) Reproducibility between γ-phenotypes in the primary and batch retest DrugTarget-CDKO screens. Data represent mean ± s.e.m. PCC, Pearson correlation coefficient. (b) Reproducibility between Norm-GIs in the primary and batch retest screens. Data represent mean ± s.e.m. The Pearson correlation coefficient (PCC) after same-gene targeting pairs were removed is reported in parentheses. (c) Schematic for dual sgRNA validation assay. Cas9-expressing K562 cells were infected with lentiviruses expressing one sgRNA in a GFP vector and another sgRNA in an mCherry vector. The abundances of the four resulting fluorescent populations (uninfected, GFP, mCherry, and GFP+mCherry) were measured after 7 d. (d) K562 cells infected with lentiviruses expressing PIM1-targeting (GFP) and PIM2-targeting (mCherry) sgRNAs. The growth phenotype is calculated by measuring the relative depletion of the single-infected and double-infected cells at day 7 vs. day 0. (e,f) Quantification of growth phenotypes and GI scores for indicated double sgRNA infections compared to single-sgRNA_safe-targeting controls. The PIM1_PIM2 sgRNA pair is synergistic as predicted while the BSG_GPI pair is buffering as predicted. Data represent mean ± s.d. from 3 replicate cultures. (g,h) Additional high-confidence synergistic and buffering gene pairs validated with predicted GIs. Data represent mean ± s.d. from 3 replicate cultures.

  6. Predicted synergistic gene pairs translate to synergistic drug combinations.
    Figure 6: Predicted synergistic gene pairs translate to synergistic drug combinations.

    (a) GIT scores and γ-phenotypes of gene pairs were plotted together. We set a γ-phenotype cutoff of −4 pZ for strongly toxic drug combinations and highlight the 30 most synergistic gene pairs by rank-sum of GIT and GIM scores (marked in pink). (bg) K562 cells were treated with APEX1 (CRT0044876) and ATM (KU-60019) inhibitors (bd) or BCL2L1 (A-1155463) and MCL1(A-1210477) inhibitors (eg) alone and in combination at the indicated concentrations for 72 h (b,e). Cell viability relative to the no drug control was determined by measuring live cell number using flow cytometry (FSC/SSC). (c,f) Drug synergy represented by excess over Bliss independence is calculated by subtracting the % expected inhibition from the % observed inhibition at each combination of drug doses. (d,g) Percent inhibition of cell viability upon treatment with the indicated inhibitors in a separate experiment with 6 replicate cultures. P-values represent significant differences by one-tailed t-test between the observed effect of the drug combination and the expected effect (dotted lines) based on the Bliss independence model. (h,i) K562 cells were treated with indicated drugs for 48 h and assessed by flow cytometry for Annexin V-FITC and propidium iodide (PI) staining. Plots in h are representative of three independent experiments and the percentages of Annexin V positive cells are quantified in i. (j,k) Cell viability plots for GM12892 (LCL) and CD34+ (bone marrow-derived hematopoietic stem-progenitor) cells in j or imatinib-resistant K562 cells in k treated with BCL2L1 (A-1155463) and MCL1(A-1210477) inhibitors alone and in combination at the indicated doses for 72 h. (l) BCL2L1 and MCL1 protein levels were compared by immunoblot analysis between lysates from parental and imatinib-resistant K562 cells. Data represent mean ± s.d. from 3 replicate cultures unless otherwise noted. Complete blots are shown in Supplementary Figure 13. Data represent mean ± s.d. from 3 replicate cultures unless otherwise noted.

  7. A scalable CDKO system with minimized recombination compatible with deep sequencing.
    Supplementary Fig. 1: A scalable CDKO system with minimized recombination compatible with deep sequencing.

    (a) Generation of PCR amplicons for deep sequencing (see methods). Double-sgRNA cassettes were directly amplified from genomic DNA and adapters were added during two rounds of PCRs. (b) Paired-end sequencing to directly read double-sgRNAs. PCR amplicons (around 640 bp) were clustered efficiently in flow cells and three custom sequencing primers used to read a front sgRNA, a rear sgRNA, and an index in order. (c) Double-sgRNA vectors with either two identical mU6 promoters, or mU6 and hU6 promoters driving expression of mCherry and GFP-targeting sgRNAs, were PCR-amplified using the primer pairs indicated. (d) The double-sgRNA cassettes in panel c were PCR-amplified from either purified plasmids or genomic DNA isolated from K562 cells infected with the corresponding double-sgRNA vectors. Expected size of the PCR amplicons are marked with red arrows and PCR amplicons from recombination-affected vectors are marked with blue arrows. (e) Flow cytometry analysis of GFP and mCherry knockout efficiency in cells infected with the vectors in panel. (f) Sequencing of the recombination-affected PCR amplicon from the double-sgRNA vector with two identical mU6 promoters shows that the recombination happens between two mU6 promoters.

  8. Assessment of the DrugTarget-CDKO library quality.
    Supplementary Fig. 2: Assessment of the DrugTarget-CDKO library quality.

    (a) Estimating the minimally required read count for a double-sgRNA. From the representation of single-sgRNAs in the hU6 and mU6 single-sgRNA library, the expected frequency of double-sgRNAs were calculated and compared to the observed frequency of double-sgRNAs in the DrugTarget-CDKO library. Ratios of the two frequencies showed that under ~50 read counts, the observed frequencies markedly fell below the expected. Based on this data, double-sgRNAs with less than 50 read counts were removed from further analyses. (b) Frequencies of double-sgRNAs were compared between the Plasmid library and the T0 sample. Frequencies of Safe_Safe sgRNAs were slightly enriched in the T0 sample since most double-sgRNAs have negative γ-phenotypes. (c) Minimal positional bias in DrugTarget-CDKO library. γ-phenotypes of gene pairs were compared between both orientations. PCC, Pearson correlation coefficient. (d) High reproducibility of measured γ-phenotypes of gene pairs between two experimental replicates. PCC, Pearson correlation coefficient. (e) Single knockout phenotypes of genes are highly correlated between the DrugTarget-CDKO screen and a previous genome-wide single-sgRNA screen10. PCC, Pearson correlation coefficient. (f) Sequencing depth affects the phenotypic plateau observed in Figure 3a.

  9. Comparison of GI scores.
    Supplementary Fig. 3: Comparison of GI scores.

    (a) GIT scores were calculated based on Raw-GIs and compared between two experimental replicates. PCC, Pearson correlation coefficient. (b) GIM scores were calculated and compared between two experimental replicates (see methods). GIM scores of gene pairs comprised of two same genes are marked in pink. The 5 most synergistic gene pairs are marked in orange. PCC, Pearson correlation coefficient. (c) Histograms of GIT scores showed that most gene pairs comprised of two same genes are buffering (red solid line). GIT score of Safe_Safe pair (purple dotted line). (d) Norm GIs of double-sgRNA combinations targeting the most synergistic (AKT1_AKT2) and most buffering (SKP2_SKP2) gene pairs by GIT score were plotted against the expected γ-phenotypes. These two pairs showed distinct GI distributions that were well-separated in opposite directions from 0. Safe_Safe double-sgRNAs are marked as yellow dots and all other double-sgRNAs are marked as grey dots. (e) Expected and observed γ-phenotypes of gene pairs were plotted and color-coded by their GIT score. The most synergistic (AKT1_AKT2) and buffering (SKP2_SKP2) pairs are highlighted.

  10. Two different sgRNAs targeting the same gene are more synergistic than identical sgRNAs.
    Supplementary Fig. 4: Two different sgRNAs targeting the same gene are more synergistic than identical sgRNAs.

    (a-c) Norm-GIs were measured for three groups - all double-sgRNAs, pairs of same guides, and pairs of two different sgRNAs targeting the same genes. Two sgRNAs targeting the same gene tend to be buffering. However, two different sgRNAs targeting the same gene tend to be less buffering than two identical sgRNAs. (a) Norm-GIs were plotted against the expected γ-phenotypes for all three groups (black: all double-sgRNAs, pink: double-sgRNAs comprised of two same guides, blue: double-sgRNAs comprised of two different guides targeting the same gene) (b) Plots in a were binned against the expected phenotype of double-sgRNAs. Data represent mean ± s.e.m. Data were binned in a way that each range includes at least 30 data points for any given group among the three. (c) Distribution of Norm-GIs for the three groups. Only double-sgRNAs with absolute γ-phenotypes greater than 3.7 were analyzed. The pink line indicates the mean, the blue box represents the 25th-75th percentile, and the dotted bar indicates the 10th-90th percentile. (d) The Norm-GI patterns in 3 x 3 double-sgRNA combinations of pairs targeting the same gene. Top panel shows the distribution of GI scores for pairs targeting the same gene. The blue dotted lines mark the 5 most synergistic pairs in the distribution. The brown dotted lines represent the 35th-39th buffering pairs, and the yellow dotted lines mark the 5 most buffering pairs in the distribution. All double-sgRNA combinations for each group were averaged over the 3 x 3 grid and color-coded by their Norm-GIs. (e) In the dual sgRNA validation assay, two different sgRNAs targeting TK1 showed synergistic GI whereas two identical sgRNAs did not show significant GI. Data represent mean ± s.d. (n=3) from replicate cultures.

  11. [gamma]-phenotype-based DrugTarget-CDKO GI map shows sparse genetic interactions.
    Supplementary Fig. 5: γ-phenotype-based DrugTarget-CDKO GI map shows sparse genetic interactions.

    (a) Correlations of GI profiles between two sgRNAs were compared in two experimental replicates: sgRNAs targeting the same gene are marked in pink. Due to the low GI frequencies, correlations of GI profiles between sgRNA pairs were very low. PCC, Pearson correlation coefficient. (b) The distributions of correlations of GI profiles for all sgRNA pairs (blue) and for sgRNAs pairs targeting the same gene (orange). Medians of the distributions are marked by dotted lines. (c) Sparse genetic interactions in the DrugTarget-CDKO GI map. GIM scores of all gene pairs were calculated and color-coded by a yellow-cyan heatmap. Genes were hierarchically clustered by their correlation of GI profiles. γ-phenotypes of individual genes are marked in sidebars with a red-blue heatmap.

  12. Ricin-CDKO screen.
    Supplementary Fig. 6: Ricin-CDKO screen.

    (a) Selection of genes modulating ricin-sensitivity and resulting Ricin-CDKO library. (b) Schematic of the Ricin-CDKO screen. Infected cells were split into two and one group was treated with 4 pulses of ricin for 14 days, while the other was cultured for 14 days without ricin treatment. (c) Cumulative distribution of sequencing reads for double-sgRNAs. (d) Histogram plotting the number of double-sgRNAs per gene pair. 99.6% of the 6,063 detected gene pairs have more than 6 double-sgRNA combinations. (e) Minimal positional bias in Ricin-CDKO library for ρ-phenotype. PCC, Pearson correlation coefficient. (f) Measuring Norm-GIs of double-sgRNAs. Expected and observed ρ-phenotypes of double-sgRNAs were plotted and deviations from the median line (blue) were processed from Raw-GIs into Norm-GIs as previously done for DrugTarget-CDKO screen. (g) t-value-based GIT scores calculated for ρ-phenotypes were plotted between two experimental replicates. The 5 most synergistic and buffering pairs by rank-sum of GIT scores of two replicates are marked in orange dots. The Pearson correlation after same-gene targeting pairs were removed is reported in parentheses. PCC, Pearson correlation coefficient.

  13. Comparison of a previous shRNA-based ricin GI map with the Ricin-CDKO GI map.
    Supplementary Fig. 7: Comparison of a previous shRNA-based ricin GI map with the Ricin-CDKO GI map.

    (a) Measurement of correlation of genetic interactions between the previously published shRNA-based Ricin GI map11 and the Ricin-CDKO map for the 91 common interactions present in both maps. GIs are scaled according to the system described in the corresponding manuscript. Two cases of genetic interactions which have opposite signs in the two maps are labeled in the plot. PCC, Pearson correlation coefficient. (b) Genetic interactions of the common protein complexes and genes between the two maps are used to generate two compact GI maps for comparison. Essential genes are marked in red.

  14. Gene pairs with similar GI profiles are enriched for known protein interactions (Online Methods).
    Supplementary Fig. 8: Gene pairs with similar GI profiles are enriched for known protein interactions (Online Methods).

    (a) Data in Figure 4d were plotted as AUC (Area Under the Curve). Gene pairs were sorted by the different features of GI map and cumulative sums of the number of STRING interactions identified in the sorted gene pairs were plotted. (b) A genetic interaction network of the 66 most correlated gene pairs (GI correlation > 0.5) in terms of GI patterns was generated. Genes are grouped and colored by their biological functions and their known PPIs. Each edge between two genes indicates that they have a correlation of GI patterns over 0.5. A red edge indicates that this gene pair also has known protein interactions reported in STRING. 40% of the 66 gene pairs have reported protein interactions whereas on average, only 5% of 66 randomly selected gene pairs from the Ricin-CDKO map have reported protein interactions in STRING.

  15. Validation of synergistic gene pairs with individual sgRNAs.
    Supplementary Fig. 9: Validation of synergistic gene pairs with individual sgRNAs.

    (a) Predicted synergistic gene pairs were validated using individual sgRNAs (using two separate vectors). Data represent mean ± s.d. from 3 replicate cultures. (b) 3 of the synergistic gene pairs were validated using a second pair of sgRNAs. Data represent mean ± s.d. from 3 replicate cultures. (c) Example of GI calculation for individual sgRNA validations using double-sgRNA vector. Double-sgRNA vectors were cloned containing two safe-sgRNAs, one safe-sgRNA and one gene-targeting sgRNA (PIM1_Safe and Safe_PIM2), or two gene-targeting sgRNAs (PIM1_PIM2) and infected into Cas9-expressing K562 cells. Growth phenotypes of single and double gene knockouts are calculated by measuring the depletion of GFP+ cells relative to uninfected cells (PIM1_Safe and Safe_PIM2 for single knockout phenotypes and PIM1_PIM2 for double knockout) from T0 to T7, normalized to Safe_Safe cells. GIs are determined by comparing the observed double knockout phenotype to the expected from the single knockout phenotypes. (d) Using the double-sgRNA vector system, synergy was validated for sgRNA pairs predicted to be synergistic (PIM1_PIM2, BCL2L1_MCL1, PRKDC_ATM, PRKDC_TSPO) while sgRNA pairs not predicted to be synergistic did not show synergy in dual-sgRNA retests (PRKDC_PIM2, PRKDC_MCL1, PIM1_TSPO, BCL2L1_TSPO, BCL2L1_ATM, PIM1_MCL1, BCL2L1_PIM2). Data represent mean ± s.d. from 3 replicate cultures. (e) TIDE indel analysis for sgRNAs against indicated genes.

  16. Combination APEX1_ATM drug treatment synergistically induces DSBs and apoptosis.
    Supplementary Fig. 10: Combination APEX1_ATM drug treatment synergistically induces DSBs and apoptosis.

    (a-c) K562 cells were treated with APEX1 (CRT0044876) and ATM (KU-60019) inhibitors for 48 h, fixed, and stained for γH2AX. Cells were analyzed by flow cytometry and representative histograms from each sample are plotted in a. Median FL-1 ± s.d. from 3 replicate cultures are plotted in b. Cells were additionally stained with Hoescht and representative images are shown in c. Scale bars,10 μm. (d,e) K562 cells were treated with indicated drugs for 48 h and assessed by flow cytometry for Annexin V-FITC and propidium iodide (PI) staining. Plots in d are representative of three independent experiments and the percentages of Annexin V-positive cells are quantified in e (mean ± s.d., n = 3).

  17. DrugTarget-CDKO genetic interactions predict drug synergy.
    Supplementary Fig. 11: DrugTarget-CDKO genetic interactions predict drug synergy.

    (a-k) Cell viability and Bliss drug synergy plots for drug pairs in K562 cells (a-j) and MV4;11 cells (k). Additional true positives are shown in a-c: (a) ATM (KU-60019) and PRKDC (NU7441), (b) APEX1 (CRT0044876) and PRKDC, (c) TSPO (PK-11195) and PRKDC. Examples of true negatives are shown in d-g: (d) TXN (PX-12) and XPO1 (KPT-330), (e) MCL1 (A-1210477) and PRKDC, (f) TSPO and XPO1, (g) CARM1 (1-benzyl-3,5-bis-(3-bromo-4-hydroxybenzylidene)piperidin-4-one) and XPO1. (h) NAMPT (FK866) and XPO1 (false positive). (i) BCL2L1 (A-1155463) and XPO1 (false negative). (j) The BCL2L1 and MCL1 pair was tested using a different MCL1 inhibitor (UMI-77) and (k) in the MV4;11 AML cell line (using A-1155463 and A-1210477).

  18. Normalized GI improves reproducibility of genetic interactions between replicates.
    Supplementary Fig. 12: Normalized GI improves reproducibility of genetic interactions between replicates.

    (a) GIT scores calculated from Raw-GI and Norm-GI are compared for the ρ-phenotype-based Ricin-CDKO map. GIT scores calculated from Norm-GI show higher correlation between replicates than those from Raw-GI. PCC, Pearson correlation coefficient. (b) GIT scores calculated from Raw-GI and Norm-GI are compared for the γ-phenotype-based DrugTarget-CDKO map. GIT scores calculated from Norm-GI show slightly higher correlation between replicates than those from RawGI. PCC, Pearson correlation coefficient. (c) Normalization of GIs improves the uniformity of variance across the range of expected γ-phenotypes. Variance of Raw-GIs and Norm-GIs for the same guide pairs across two experimental replicates in DrugTarget-CDKO map was measured with respect to the expected γ-phenotype. Data are binned across the expected γ-phenotypes (bin number = 22, bin size = 1 pZ) and average variance on each bin is calculated and plotted in the graphs.

  19. Complete blot images.
    Supplementary Fig. 13: Complete blot images.

    Complete blot images presented in Figure 6l. Protein bands shown in Figure 6l are indicated by arrows. (a) Blots for BCL2L1 (left panel) and tubulin loading control (right panel). (b) Blots for MCL1 (left panel) and tubulin loading control (right panel).

Accession codes

Primary accessions

BioProject

Sequence Read Archive

References

  1. Jia, J. et al. Mechanisms of drug combinations: interaction and network perspectives. Nat. Rev. Drug Discov. 8, 111128 (2009).
  2. Ashburn, T.T. & Thor, K.B. Drug repositioning: identifying and developing new uses for existing drugs. Nat. Rev. Drug Discov. 3, 673683 (2004).
  3. Al-Lazikani, B., Banerji, U. & Workman, P. Combinatorial drug therapy for cancer in the post-genomic era. Nat. Biotechnol. 30, 679692 (2012).
  4. Sun, X., Vilar, S. & Tatonetti, N.P. High-throughput methods for combinatorial drug discovery. Sci. Transl. Med. 5, 205rv1 (2013).
  5. Collins, S.R., Weissman, J.S. & Krogan, N.J. From information to knowledge: new technologies for defining gene function. Nat. Methods 6, 721723 (2009).
  6. Costanzo, M. et al. A global genetic interaction network maps a wiring diagram of cellular function. Science http://dx.doi.org/10.1126/science.aaf1420 (2016).
  7. Schuldiner, M. et al. Exploration of the function and organization of the yeast early secretory pathway through an epistatic miniarray profile. Cell 123, 507519 (2005).
  8. Jonikas, M.C. et al. Comprehensive characterization of genes required for protein folding in the endoplasmic reticulum. Science 323, 16931697 (2009).
  9. Bandyopadhyay, S. et al. Rewiring of genetic networks in response to DNA damage. Science 330, 13851389 (2010).
  10. Collins, S.R. et al. Functional dissection of protein complexes involved in yeast chromosome biology using a genetic interaction map. Nature 446, 806810 (2007).
  11. Dixon, S.J., Costanzo, M., Baryshnikova, A., Andrews, B. & Boone, C. Systematic mapping of genetic interaction networks. Annu. Rev. Genet. 43, 601625 (2009).
  12. Frost, A. et al. Functional repurposing revealed by comparing S. pombe and S. cerevisiae genetic interactions. Cell 149, 13391352 (2012).
  13. Horn, T. et al. Mapping of signaling networks through synthetic genetic interaction analysis by RNAi. Nat. Methods 8, 341346 (2011).
  14. Roguev, A. et al. Quantitative genetic-interaction mapping in mammalian cells. Nat. Methods 10, 432437 (2013).
  15. Bassik, M.C. et al. A systematic mammalian genetic interaction map reveals pathways underlying ricin susceptibility. Cell 152, 909922 (2013).
  16. Srivas, R. et al. A network of conserved synthetic lethal interactions for exploration of precision cancer therapy. Mol. Cell 63, 514525 (2016).
  17. Wong, A.S.L., Choi, G.C.G., Cheng, A.A., Purcell, O. & Lu, T.K. Massively parallel high-order combinatorial genetics in human cells. Nat. Biotechnol. 33, 952961 (2015).
  18. Jinek, M. et al. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science 337, 816821 (2012).
  19. Mali, P. et al. RNA-guided human genome engineering via Cas9. Science 339, 823826 (2013).
  20. Cong, L. et al. Multiplex genome engineering using CRISPR/Cas systems. Science 339, 819823 (2013).
  21. Tsai, S.Q. et al. Dimeric CRISPR RNA-guided FokI nucleases for highly specific genome editing. Nat. Biotechnol. 32, 569576 (2014).
  22. Xie, K., Minkenberg, B. & Yang, Y. Boosting CRISPR/Cas9 multiplex editing capability with the endogenous tRNA-processing system. Proc. Natl. Acad. Sci. USA 112, 35703575 (2015).
  23. Wong, A.S.L. et al. Multiplexed barcoded CRISPR-Cas9 screening enabled by CombiGEM. Proc. Natl. Acad. Sci. USA 113, 25442549 (2016).
  24. Vidigal, J.A. & Ventura, A. Rapid and efficient one-step generation of paired gRNA CRISPR-Cas9 libraries. Nat. Commun. 6, 8083 (2015).
  25. Sack, L.M., Davoli, T., Xu, Q., Li, M.Z. & Elledge, S.J. Sources of error in mammalian genetic screens. G3 (Bethesda) 6, 27812790 (2016).
  26. Zhu, F. et al. Therapeutic target database update 2012: a resource for facilitating target-oriented drug discovery. Nucleic Acids Res. 40, D1128D1136 (2012).
  27. Wishart, D.S. et al. DrugBank: a comprehensive resource for in silico drug discovery and exploration. Nucleic Acids Res. 34, D668D672 (2006).
  28. Pawson, A.J. et al. The IUPHAR/BPS Guide to PHARMACOLOGY: an expert-driven knowledgebase of drug targets and their ligands. Nucleic Acids Res. 42, D1098D1106 (2014).
  29. Morgens, D.W. et al. Genome-scale measurement of off-target activity using Cas9 toxicity in high-throughput screens. Nat. Commun (in press). (2017).
  30. Deans, R.M. et al. Parallel shRNA and CRISPR-Cas9 screens enable antiviral drug target identification. Nat. Chem. Biol. 12, 361366 (2016).
  31. Kampmann, M., Bassik, M.C. & Weissman, J.S. Integrated platform for genome-wide screening and construction of high-density genetic interaction maps in mammalian cells. Proc. Natl. Acad. Sci. USA 110, E2317E2326 (2013).
  32. Gilbert, L.A. et al. Genome-scale CRISPR-mediated control of gene repression and activation. Cell 159, 647661 (2014).
  33. Collins, S.R., Schuldiner, M., Krogan, N.J. & Weissman, J.S. A strategy for extracting and analyzing large-scale quantitative epistatic interaction data. Genome Biol. 7, R63 (2006).
  34. Fischer, B. et al. A map of directional genetic interactions in a metazoan cell. eLife 4, 121 (2015).
  35. Blomen, V.A. et al. Gene essentiality and synthetic lethality in haploid human cells. Science 350, 10921096 (2015).
  36. Brazil, D.P., Yang, Z.Z. & Hemmings, B.A. Advances in protein kinase B signalling: AKTion on multiple fronts. Trends Biochem. Sci. 29, 233242 (2004).
  37. Saurabh, K. et al. The PIM family of oncoproteins: small kinases with huge implications in myeloid leukemogenesis and as therapeutic targets. Oncotarget 5, 85038514 (2014).
  38. Chen, J.L., Limnander, A. & Rothman, P.B. Pim-1 and Pim-2 kinases are required for efficient pre-B-cell transformation by v-Abl oncogene. Blood 111, 16771685 (2008).
  39. Johannes, L. & Popoff, V. Tracing the retrograde route in protein trafficking. Cell 135, 11751187 (2008).
  40. Moreau, D. et al. Genome-wide RNAi screens identify genes required for Ricin and PE intoxications. Dev. Cell 21, 231244 (2011).
  41. Morgens, D.W., Deans, R.M., Li, A. & Bassik, M.C. Systematic comparison of CRISPR/Cas9 and RNAi screens for essential genes. Nat. Biotechnol. 34, 634636 (2016).
  42. Wang, W. et al. Mannosidase 2, alpha 1 deficiency is associated with ricin resistance in embryonic stem (ES) cells. PLoS One 6, e22993 (2011).
  43. Bar-Peled, L. et al. A Tumor suppressor complex with GAP activity for the Rag GTPases that signal amino acid sufficiency to mTORC1. Science 340, 11001106 (2013).
  44. Margueron, R. & Reinberg, D. The Polycomb complex PRC2 and its mark in life. Nature 469, 343349 (2011).
  45. Szklarczyk, D. et al. STRING v10: protein-protein interaction networks, integrated over the tree of life. Nucleic Acids Res. 43, D447D452 (2015).
  46. Parnas, O. et al. A genome-wide CRISPR screen in primary immune cells to dissect regulatory networks. Cell 162, 675686 (2015).
  47. Chen, S. et al. Genome-wide CRISPR screen in a mouse model of tumor growth and metastasis. Cell 160, 12461260 (2015).
  48. Brinkman, E.K., Chen, T., Amendola, M. & van Steensel, B. Easy quantitative assessment of genome editing by sequence trace decomposition. Nucleic Acids Res. 42, e168 (2014).
  49. Lord, C.J., Tutt, A.N.J. & Ashworth, A. Synthetic lethality and cancer therapy: lessons learned from the development of PARP inhibitors. Annu. Rev. Med. 66, 455470 (2015).
  50. Sultana, R. et al. Synthetic lethal targeting of DNA double-strand break repair deficient cells by human apurinic/apyrimidinic endonuclease inhibitors. Int. J. Cancer 131, 24332444 (2012).
  51. Riabinska, A. et al. Therapeutic targeting of a robust non-oncogene addiction to PRKDC in ATM-defective tumors. Sci. Transl. Med. 5, 189ra78 (2013).
  52. Austin, C.J.D., Kahlert, J., Kassiou, M. & Rendina, L.M. The translocator protein (TSPO): a novel target for cancer chemotherapy. Int. J. Biochem. Cell Biol. 45, 12121216 (2013).
  53. Placzek, W.J. et al. A survey of the anti-apoptotic Bcl-2 subfamily expression in cancer types provides a platform to predict the efficacy of Bcl-2 antagonists in cancer therapy. Cell Death Dis. 1, e40 (2010).
  54. Bixby, D. & Talpaz, M. Seeking the causes and solutions to imatinib-resistance in chronic myeloid leukemia. Leukemia 25, 722 (2011).
  55. Delbridge, A.R.D., Grabow, S., Strasser, A. & Vaux, D.L. Thirty years of BCL-2: translating cell death discoveries into novel cancer therapies. Nat. Rev. Cancer 16, 99109 (2016).
  56. Aichberger, K.J. et al. Identification of mcl-1 as a BCR/ABL-dependent target in chronic myeloid leukemia (CML): evidence for cooperative antileukemic effects of imatinib and mcl-1 antisense oligonucleotides. Blood 105, 33033311 (2005).
  57. Dai, Y., Rahmani, M., Corey, S.J., Dent, P. & Grant, S. A Bcr/Abl-independent, Lyn-dependent form of imatinib mesylate (STI-571) resistance is associated with altered expression of Bcl-2. J. Biol. Chem. 279, 3422734239 (2004).
  58. Berns, K. et al. A large-scale RNAi screen in human cells identifies new components of the p53 pathway. Nature 428, 431437 (2004).
  59. Friedman, A. & Perrimon, N. A functional RNAi screen for regulators of receptor tyrosine kinase and ERK signalling. Nature 444, 230234 (2006).
  60. Mendes-Pereira, A.M. et al. Genome-wide functional screen identifies a compendium of genes affecting sensitivity to tamoxifen. Proc. Natl. Acad. Sci. USA 109, 27302735 (2012).
  61. Eichhorn, P.J.A. et al. Phosphatidylinositol 3-kinase hyperactivation results in lapatinib resistance that is reversed by the mTOR/phosphatidylinositol 3-kinase inhibitor NVP-BEZ235. Cancer Res. 68, 92219230 (2008).
  62. Corcoran, R.B. et al. Synthetic lethal interaction of combined BCL-XL and MEK inhibition promotes tumor regressions in KRAS mutant cancer models. Cancer Cell 23, 121128 (2013).
  63. Prahallad, A. et al. Unresponsiveness of colon cancer to BRAF(V600E) inhibition through feedback activation of EGFR. Nature 483, 100103 (2012).
  64. Opferman, J.T. Attacking cancer's Achilles heel: antagonism of anti-apoptotic BCL-2 family members. FEBS J. 283, 26612675 (2016).
  65. Leverson, J.D. et al. Potent and selective small-molecule MCL-1 inhibitors demonstrate on-target cancer cell killing activity as single agents and in combination with ABT-263 (navitoclax). Cell Death Dis. 6, e1590 (2015).
  66. Chonghaile, T.N. et al. Maturation stage of T-cell acute lymphoblastic leukemia determines BCL-2 versus BCL-XL dependence and sensitivity to ABT-199. Cancer Discov. 4, 10741087 (2014).
  67. Ko, T.K., Chuah, C.T.H., Huang, J.W.J., Ng, K.-P. & Ong, S.T. The BCL2 inhibitor ABT-199 significantly enhances imatinib-induced cell death in chronic myeloid leukemia progenitors. Oncotarget 5, 90339038 (2014).
  68. Hong, H.-Y., Choi, J., Cho, Y.-W. & Kim, B.-C. Cdc25A promotes cell survival by stimulating NF-κB activity through IκB-α phosphorylation and destabilization. Biochem. Biophys. Res. Commun. 420, 293296 (2012).
  69. Bassi, Z.I., Audusseau, M., Riparbelli, M.G., Callaini, G. & D'Avino, P.P. Citron kinase controls a molecular network required for midbody formation in cytokinesis. Proc. Natl. Acad. Sci. USA 110, 97829787 (2013).
  70. Chen, B. et al. Dynamic imaging of genomic loci in living human cells by an optimized CRISPR/Cas system. Cell 155, 14791491 (2013).
  71. Shalem, O., Sanjana, E.N., Hartenian, E. & Zhang, F. Genome-Scale CRISPR-Cas9 Knockout. Science 343, 8488 (2014).
  72. Langmead, B., Trapnell, C., Pop, M. & Salzberg, S.L. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 10, R25 (2009).
  73. Kampmann, M., Bassik, M.C. & Weissman, J.S. Functional genomics platform for pooled screening and generation of mammalian genetic interaction maps. Nat. Protoc. 9, 18251847 (2014).
  74. Shannon, P. et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 13, 24982504 (2003).
  75. Chatr-aryamontri, A. et al. The BioGRID interaction database: 2015 update. Nucleic Acids Res. 43, D470D478 (2015).
  76. Eisen, M.B., Spellman, P.T., Brown, P.O. & Botstein, D. Cluster analysis and display of genome-wide expression patterns. Proc. Natl. Acad. Sci. USA 95, 1486314868 (1998).
  77. Saldanha, A.J. Java Treeview--extensible visualization of microarray data. Bioinformatics 20, 32463248 (2004).

Download references

Author information

  1. These authors contributed equally to this work.

    • Kyuho Han &
    • Edwin E Jeng

Affiliations

  1. Department of Genetics, Stanford University, Stanford, California, USA.

    • Kyuho Han,
    • Edwin E Jeng,
    • Gaelen T Hess,
    • David W Morgens,
    • Amy Li &
    • Michael C Bassik
  2. Program in Cancer Biology, Stanford University, Stanford, California, USA.

    • Edwin E Jeng
  3. Chemistry, Engineering, and Medicine for Human Health (ChEM-H), Stanford University, Stanford, California, USA.

    • Michael C Bassik

Contributions

K.H. and M.C.B. conceived and designed the study. K.H. designed the CDKO system and the scoring systems for GI map. K.H. analyzed the screen data and performed the GI and PPI analyses. K.H., A.L., and E.E.J. performed the CDKO screens. G.T.H., A.L., and D.W.M. performed the genome-wide screens for ricin modulators. D.W.M. selected the best-working sgRNAs to design the CDKO libraries. K.H. and E.E.J. validated the hits from the CDKO screens. E.E.J. performed the drug validations and related experiments. K.H., E.E.J., and M.C.B. wrote the manuscript. All authors discussed the results and the manuscript. M.C.B. supervised the study.

Competing financial interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to:

Author details

Supplementary information

Supplementary Figures

  1. Supplementary Figure 1: A scalable CDKO system with minimized recombination compatible with deep sequencing. (123 KB)

    (a) Generation of PCR amplicons for deep sequencing (see methods). Double-sgRNA cassettes were directly amplified from genomic DNA and adapters were added during two rounds of PCRs. (b) Paired-end sequencing to directly read double-sgRNAs. PCR amplicons (around 640 bp) were clustered efficiently in flow cells and three custom sequencing primers used to read a front sgRNA, a rear sgRNA, and an index in order. (c) Double-sgRNA vectors with either two identical mU6 promoters, or mU6 and hU6 promoters driving expression of mCherry and GFP-targeting sgRNAs, were PCR-amplified using the primer pairs indicated. (d) The double-sgRNA cassettes in panel c were PCR-amplified from either purified plasmids or genomic DNA isolated from K562 cells infected with the corresponding double-sgRNA vectors. Expected size of the PCR amplicons are marked with red arrows and PCR amplicons from recombination-affected vectors are marked with blue arrows. (e) Flow cytometry analysis of GFP and mCherry knockout efficiency in cells infected with the vectors in panel. (f) Sequencing of the recombination-affected PCR amplicon from the double-sgRNA vector with two identical mU6 promoters shows that the recombination happens between two mU6 promoters.

  2. Supplementary Figure 2: Assessment of the DrugTarget-CDKO library quality. (148 KB)

    (a) Estimating the minimally required read count for a double-sgRNA. From the representation of single-sgRNAs in the hU6 and mU6 single-sgRNA library, the expected frequency of double-sgRNAs were calculated and compared to the observed frequency of double-sgRNAs in the DrugTarget-CDKO library. Ratios of the two frequencies showed that under ~50 read counts, the observed frequencies markedly fell below the expected. Based on this data, double-sgRNAs with less than 50 read counts were removed from further analyses. (b) Frequencies of double-sgRNAs were compared between the Plasmid library and the T0 sample. Frequencies of Safe_Safe sgRNAs were slightly enriched in the T0 sample since most double-sgRNAs have negative γ-phenotypes. (c) Minimal positional bias in DrugTarget-CDKO library. γ-phenotypes of gene pairs were compared between both orientations. PCC, Pearson correlation coefficient. (d) High reproducibility of measured γ-phenotypes of gene pairs between two experimental replicates. PCC, Pearson correlation coefficient. (e) Single knockout phenotypes of genes are highly correlated between the DrugTarget-CDKO screen and a previous genome-wide single-sgRNA screen10. PCC, Pearson correlation coefficient. (f) Sequencing depth affects the phenotypic plateau observed in Figure 3a.

  3. Supplementary Figure 3: Comparison of GI scores. (120 KB)

    (a) GIT scores were calculated based on Raw-GIs and compared between two experimental replicates. PCC, Pearson correlation coefficient. (b) GIM scores were calculated and compared between two experimental replicates (see methods). GIM scores of gene pairs comprised of two same genes are marked in pink. The 5 most synergistic gene pairs are marked in orange. PCC, Pearson correlation coefficient. (c) Histograms of GIT scores showed that most gene pairs comprised of two same genes are buffering (red solid line). GIT score of Safe_Safe pair (purple dotted line). (d) Norm GIs of double-sgRNA combinations targeting the most synergistic (AKT1_AKT2) and most buffering (SKP2_SKP2) gene pairs by GIT score were plotted against the expected γ-phenotypes. These two pairs showed distinct GI distributions that were well-separated in opposite directions from 0. Safe_Safe double-sgRNAs are marked as yellow dots and all other double-sgRNAs are marked as grey dots. (e) Expected and observed γ-phenotypes of gene pairs were plotted and color-coded by their GIT score. The most synergistic (AKT1_AKT2) and buffering (SKP2_SKP2) pairs are highlighted.

  4. Supplementary Figure 4: Two different sgRNAs targeting the same gene are more synergistic than identical sgRNAs. (95 KB)

    (a-c) Norm-GIs were measured for three groups - all double-sgRNAs, pairs of same guides, and pairs of two different sgRNAs targeting the same genes. Two sgRNAs targeting the same gene tend to be buffering. However, two different sgRNAs targeting the same gene tend to be less buffering than two identical sgRNAs. (a) Norm-GIs were plotted against the expected γ-phenotypes for all three groups (black: all double-sgRNAs, pink: double-sgRNAs comprised of two same guides, blue: double-sgRNAs comprised of two different guides targeting the same gene) (b) Plots in a were binned against the expected phenotype of double-sgRNAs. Data represent mean ± s.e.m. Data were binned in a way that each range includes at least 30 data points for any given group among the three. (c) Distribution of Norm-GIs for the three groups. Only double-sgRNAs with absolute γ-phenotypes greater than 3.7 were analyzed. The pink line indicates the mean, the blue box represents the 25th-75th percentile, and the dotted bar indicates the 10th-90th percentile. (d) The Norm-GI patterns in 3 x 3 double-sgRNA combinations of pairs targeting the same gene. Top panel shows the distribution of GI scores for pairs targeting the same gene. The blue dotted lines mark the 5 most synergistic pairs in the distribution. The brown dotted lines represent the 35th-39th buffering pairs, and the yellow dotted lines mark the 5 most buffering pairs in the distribution. All double-sgRNA combinations for each group were averaged over the 3 x 3 grid and color-coded by their Norm-GIs. (e) In the dual sgRNA validation assay, two different sgRNAs targeting TK1 showed synergistic GI whereas two identical sgRNAs did not show significant GI. Data represent mean ± s.d. (n=3) from replicate cultures.

  5. Supplementary Figure 5: γ-phenotype-based DrugTarget-CDKO GI map shows sparse genetic interactions. (206 KB)

    (a) Correlations of GI profiles between two sgRNAs were compared in two experimental replicates: sgRNAs targeting the same gene are marked in pink. Due to the low GI frequencies, correlations of GI profiles between sgRNA pairs were very low. PCC, Pearson correlation coefficient. (b) The distributions of correlations of GI profiles for all sgRNA pairs (blue) and for sgRNAs pairs targeting the same gene (orange). Medians of the distributions are marked by dotted lines. (c) Sparse genetic interactions in the DrugTarget-CDKO GI map. GIM scores of all gene pairs were calculated and color-coded by a yellow-cyan heatmap. Genes were hierarchically clustered by their correlation of GI profiles. γ-phenotypes of individual genes are marked in sidebars with a red-blue heatmap.

  6. Supplementary Figure 6: Ricin-CDKO screen. (179 KB)

    (a) Selection of genes modulating ricin-sensitivity and resulting Ricin-CDKO library. (b) Schematic of the Ricin-CDKO screen. Infected cells were split into two and one group was treated with 4 pulses of ricin for 14 days, while the other was cultured for 14 days without ricin treatment. (c) Cumulative distribution of sequencing reads for double-sgRNAs. (d) Histogram plotting the number of double-sgRNAs per gene pair. 99.6% of the 6,063 detected gene pairs have more than 6 double-sgRNA combinations. (e) Minimal positional bias in Ricin-CDKO library for ρ-phenotype. PCC, Pearson correlation coefficient. (f) Measuring Norm-GIs of double-sgRNAs. Expected and observed ρ-phenotypes of double-sgRNAs were plotted and deviations from the median line (blue) were processed from Raw-GIs into Norm-GIs as previously done for DrugTarget-CDKO screen. (g) t-value-based GIT scores calculated for ρ-phenotypes were plotted between two experimental replicates. The 5 most synergistic and buffering pairs by rank-sum of GIT scores of two replicates are marked in orange dots. The Pearson correlation after same-gene targeting pairs were removed is reported in parentheses. PCC, Pearson correlation coefficient.

  7. Supplementary Figure 7: Comparison of a previous shRNA-based ricin GI map with the Ricin-CDKO GI map. (118 KB)

    (a) Measurement of correlation of genetic interactions between the previously published shRNA-based Ricin GI map11 and the Ricin-CDKO map for the 91 common interactions present in both maps. GIs are scaled according to the system described in the corresponding manuscript. Two cases of genetic interactions which have opposite signs in the two maps are labeled in the plot. PCC, Pearson correlation coefficient. (b) Genetic interactions of the common protein complexes and genes between the two maps are used to generate two compact GI maps for comparison. Essential genes are marked in red.

  8. Supplementary Figure 8: Gene pairs with similar GI profiles are enriched for known protein interactions (Online Methods). (130 KB)

    (a) Data in Figure 4d were plotted as AUC (Area Under the Curve). Gene pairs were sorted by the different features of GI map and cumulative sums of the number of STRING interactions identified in the sorted gene pairs were plotted. (b) A genetic interaction network of the 66 most correlated gene pairs (GI correlation > 0.5) in terms of GI patterns was generated. Genes are grouped and colored by their biological functions and their known PPIs. Each edge between two genes indicates that they have a correlation of GI patterns over 0.5. A red edge indicates that this gene pair also has known protein interactions reported in STRING. 40% of the 66 gene pairs have reported protein interactions whereas on average, only 5% of 66 randomly selected gene pairs from the Ricin-CDKO map have reported protein interactions in STRING.

  9. Supplementary Figure 9: Validation of synergistic gene pairs with individual sgRNAs. (138 KB)

    (a) Predicted synergistic gene pairs were validated using individual sgRNAs (using two separate vectors). Data represent mean ± s.d. from 3 replicate cultures. (b) 3 of the synergistic gene pairs were validated using a second pair of sgRNAs. Data represent mean ± s.d. from 3 replicate cultures. (c) Example of GI calculation for individual sgRNA validations using double-sgRNA vector. Double-sgRNA vectors were cloned containing two safe-sgRNAs, one safe-sgRNA and one gene-targeting sgRNA (PIM1_Safe and Safe_PIM2), or two gene-targeting sgRNAs (PIM1_PIM2) and infected into Cas9-expressing K562 cells. Growth phenotypes of single and double gene knockouts are calculated by measuring the depletion of GFP+ cells relative to uninfected cells (PIM1_Safe and Safe_PIM2 for single knockout phenotypes and PIM1_PIM2 for double knockout) from T0 to T7, normalized to Safe_Safe cells. GIs are determined by comparing the observed double knockout phenotype to the expected from the single knockout phenotypes. (d) Using the double-sgRNA vector system, synergy was validated for sgRNA pairs predicted to be synergistic (PIM1_PIM2, BCL2L1_MCL1, PRKDC_ATM, PRKDC_TSPO) while sgRNA pairs not predicted to be synergistic did not show synergy in dual-sgRNA retests (PRKDC_PIM2, PRKDC_MCL1, PIM1_TSPO, BCL2L1_TSPO, BCL2L1_ATM, PIM1_MCL1, BCL2L1_PIM2). Data represent mean ± s.d. from 3 replicate cultures. (e) TIDE indel analysis for sgRNAs against indicated genes.

  10. Supplementary Figure 10: Combination APEX1_ATM drug treatment synergistically induces DSBs and apoptosis. (127 KB)

    (a-c) K562 cells were treated with APEX1 (CRT0044876) and ATM (KU-60019) inhibitors for 48 h, fixed, and stained for γH2AX. Cells were analyzed by flow cytometry and representative histograms from each sample are plotted in a. Median FL-1 ± s.d. from 3 replicate cultures are plotted in b. Cells were additionally stained with Hoescht and representative images are shown in c. Scale bars,10 μm. (d,e) K562 cells were treated with indicated drugs for 48 h and assessed by flow cytometry for Annexin V-FITC and propidium iodide (PI) staining. Plots in d are representative of three independent experiments and the percentages of Annexin V-positive cells are quantified in e (mean ± s.d., n = 3).

  11. Supplementary Figure 11: DrugTarget-CDKO genetic interactions predict drug synergy. (138 KB)

    (a-k) Cell viability and Bliss drug synergy plots for drug pairs in K562 cells (a-j) and MV4;11 cells (k). Additional true positives are shown in a-c: (a) ATM (KU-60019) and PRKDC (NU7441), (b) APEX1 (CRT0044876) and PRKDC, (c) TSPO (PK-11195) and PRKDC. Examples of true negatives are shown in d-g: (d) TXN (PX-12) and XPO1 (KPT-330), (e) MCL1 (A-1210477) and PRKDC, (f) TSPO and XPO1, (g) CARM1 (1-benzyl-3,5-bis-(3-bromo-4-hydroxybenzylidene)piperidin-4-one) and XPO1. (h) NAMPT (FK866) and XPO1 (false positive). (i) BCL2L1 (A-1155463) and XPO1 (false negative). (j) The BCL2L1 and MCL1 pair was tested using a different MCL1 inhibitor (UMI-77) and (k) in the MV4;11 AML cell line (using A-1155463 and A-1210477).

  12. Supplementary Figure 12: Normalized GI improves reproducibility of genetic interactions between replicates. (122 KB)

    (a) GIT scores calculated from Raw-GI and Norm-GI are compared for the ρ-phenotype-based Ricin-CDKO map. GIT scores calculated from Norm-GI show higher correlation between replicates than those from Raw-GI. PCC, Pearson correlation coefficient. (b) GIT scores calculated from Raw-GI and Norm-GI are compared for the γ-phenotype-based DrugTarget-CDKO map. GIT scores calculated from Norm-GI show slightly higher correlation between replicates than those from RawGI. PCC, Pearson correlation coefficient. (c) Normalization of GIs improves the uniformity of variance across the range of expected γ-phenotypes. Variance of Raw-GIs and Norm-GIs for the same guide pairs across two experimental replicates in DrugTarget-CDKO map was measured with respect to the expected γ-phenotype. Data are binned across the expected γ-phenotypes (bin number = 22, bin size = 1 pZ) and average variance on each bin is calculated and plotted in the graphs.

  13. Supplementary Figure 13: Complete blot images. (58 KB)

    Complete blot images presented in Figure 6l. Protein bands shown in Figure 6l are indicated by arrows. (a) Blots for BCL2L1 (left panel) and tubulin loading control (right panel). (b) Blots for MCL1 (left panel) and tubulin loading control (right panel).

PDF files

  1. Supplementary Text and Figures (5.9 MB)

    Supplementary Figures 1–13 and Supplementary Text

Excel files

  1. Supplementary Table 1 (30 KB)

    Selected 207 genes for DrugTarget-CDKO library

  2. Supplementary Table 2 (30 KB)

    700 sgRNAs for DrugTarget-CDKO library

  3. Supplementary Table 3 (23 KB)

    Distribution of double-sgRNAs per gene pair after filtering

  4. Supplementary Table 4 (2,995 KB)

    GI scores of DrugTarget-CDKO screen

  5. Supplementary Table 5 (12 KB)

    Selected 79 genes for Ricin-CDKO library

  6. Supplementary Table 6 (18 KB)

    284 sgRNAs for Ricin-CDKO library

  7. Supplementary Table 7 (493 KB)

    GI scores of Ricin-CDKO screen

  8. Supplementary Table 8 (31 KB)

    246 STRING interactions between 79 Ricin hits

  9. Supplementary Table 9 (13 KB)

    STRING interactions in the 66 most correlated gene pairs

  10. Supplementary Table 10 (9 KB)

    Selected 79 genes for DrugTarget Batch retest

  11. Supplementary Table 11 (18 KB)

    287 sgRNAs for DrugTarget Batch retest

  12. Supplementary Table 12 (13 KB)

    sgRNAs used for the validation of individual sgRNA pairs

  13. Supplementary Table 13 (11 KB)

    Summary of sgRNA and drug validations

  14. Supplementary Table 14 (12 KB)

    30 most synergistic DrugTarget pairs

Additional data