Research ArticleCancer therapy

KRASG12C inhibition produces a driver-limited state revealing collateral dependencies

See allHide authors and affiliations

Science Signaling  28 May 2019:
Vol. 12, Issue 583, eaaw9450
DOI: 10.1126/scisignal.aaw9450
  • Fig. 1 Genome-scale CRISPRi screens reveal overlapping CDs that govern the cellular impact of direct KRASG12C inhibition.

    (A) Graphic delineating the concepts of SL and CD. (B) Schematic of cancer cell line selection criteria and genome-wide CRISPRi-based screening strategy for CDs. (C) Gene phenotypes from ARS-1620 CRISPRi screens in H358 and MIA PaCa-2 cells. Overlapping collateral dependent genes (hits determined by Δlog2 fold change < −0.5) that sensitize to KRASG12C inhibition are highlighted and functionally categorized: established RAS pathway (red) and extended processes (teal). Cells were grown in 2D adherent culture. Data represent two biological replicates. (D) Average essentiality scores (normalized Bayes factors) of hit CDs were determined by combining data from publicly available resources (PICKLES database and DepMap) for all available KRAS-WT and KRAS-mutant NSCLC cell lines. Color intensities portray higher (yellow) or lower (blue) essentiality scores. Grayscale intensities portray higher (black) or lower (white) SL scores, calculated by subtracting KRAS-WT from KRAS-mutant average essentiality scores.

  • Fig. 2 KRASG12C inhibition selectively potentiates the essentiality of collateral dependent genes.

    (A) H358 CRISPRi cells transduced with a negative control sgRNA or each of two sgRNAs targeting the indicated genes were grown in the presence or absence of ARS-1620 (300 nM) in a mixed population growth assay. Relative populations of sgRNA-expressing (fluorescent-positive) or non–sgRNA-expressing (fluorescent-negative) cells were determined by flow cytometry for the times shown. Cells were grown in 2D adherent culture. Data represent an initial value (day 0) as well as means of two biological replicates (days 6 and 12); error bars denote standard deviation (SD). (B) Correlation plot comparing screen phenotypes (Δlog2 fold change) to retest phenotypes (RI) in H358 CRISPRi cells. Retest phenotypes were determined from the endpoints of experiments as in (A). Points represent individual sgRNAs. Cells were treated with 300 nM ARS-1620 in experiment 1 (teal) and 1 μM ARS-1620 in experiment 2 (blue). Cells were grown in 2D adherent culture. ΔLog2 fold change and RI represent means of two biological replicates. (C) As in (B) in MIA PaCa-2 CRISPRi cells treated with 300 nM (experiment 1, teal) and 3.3 μM ARS-1620 (experiment 2, blue). (D) RT-qPCR in H358 CRISPRi and MIA PaCa-2 CRISPRi cells transduced with a negative control sgRNA or sgRNAs targeting FOSL1 and then treated with DMSO or ARS-1620 (1 μM) for 24 hours. Cells were grown in 3D spheroid culture. Data represent means of two biological replicates. (E) Immunoblots of H358 and MIA PaCa-2 cells as in (D). Cells were grown in 3D spheroid culture. Immunoblots are representative of two biological replicates (see also fig. S3).

  • Fig. 3 Critical signaling modules cooperatively sustain a KRASG12C-driven pro-growth and pro-survival program around KRAS-GTP.

    (A) Pathway map of CRISPRi screen gene phenotypes and Mann-Whitney P values from MIA PaCa-2 ARS-1620 sensitization (red) and resistance (blue). Color intensities portray phenotype strength, and circle diameters represent −log10 Mann-Whitney P values derived from the CRISPRi screen. (B) Waterfall plot of expression data from the Cancer Cell Line Encyclopedia (CCLE) and phenotype magnitudes of 58 human RTKs from the CRISPRi screen in MIA PaCa-2 cells. Z score on the x axis represents normalized mRNA expression data from RNA-seq, whereas color intensity and circle size represent phenotypes and Mann-Whitney P values from the CRISPRi screen, respectively. Cells in (A) and (B) were grown in 2D adherent culture. Phenotypes in (A) and (B) represent two biological replicates.

  • Fig. 4 Combination therapies targeting CDs with KRASG12C differ in their ability to promote S-IIP target engagement.

    (A) Chemical structure of the KRASG12C occupancy probe ARS-1323-alkyne. (B) Magnified immunoblots indicating the identities of different bands resulting from electromobility shift after treatment of H358 and MIA PaCa-2 cells with ARS-1323-alkyne (10 μM) and copper-catalyzed click reaction of lysate with TAMRA-N3. (C and D) Representative immunoblots (C) and relative densitometry of upper (KRAS + inhibitor) and lower (KRAS) bands (D) of H358 cells treated with ARS-1323-alkyne (10 μM) simultaneously with DMSO [same as shown in (B) to facilitate comparison], EGF (100 ng/ml), erlotinib (10 μM, EGFRi), SHP099 (10 μM, SHP2i), buparlisib (10 μM, PI3Ki), or palbociclib (10 μM, CDK4/6i) for the times indicated. Lysates were subjected to copper-catalyzed click reaction with TAMRA-N3. Relative densitometry was quantified using upper (KRAS + inhibitor) and lower (KRAS) bands. (E and F) As in (C) and (D), respectively, for MIA PaCa-2 cells. Instead of EGF and erlotinib, MIA PaCa-2 cells were treated with FGF2 (100 ng/ml), AZD4547 (10 μM, FGFRi), and bemcentinib (10 μM, AXLi). Other compounds were administered at the same dose as indicated in (C). Cells in (B) to (F) were grown in 3D spheroid culture. Immunoblots in (B), (C), and (E) are representative of three biological replicates. Data in (D) and (F) represent means of three biological replicates; error bars denote SD.

  • Fig. 5 CDs with KRASG12C cooperatively sustain downstream signaling outputs to promote survival and proliferation.

    (A) Immunoblots of H358 and MIA PaCa-2 cells treated with ARS-1323-alkyne (10 μM) simultaneously with DMSO, EGF (100 ng/ml), FGF2 (100 ng/ml), erlotinib (10 μM, EGFRi), AZD4547 (10 μM, FGFRi), bemcentinib (10 μM, AXLi), SHP099 (10 μM, SHP2i), buparlisib (10 μM, PI3Ki), or palbociclib (10 μM, CDK4/6i) for 2 hours. Lysates were subjected to copper-catalyzed click reaction with TAMRA-N3. Cells were grown in 3D spheroid culture. Immunoblots represent two biological replicates. (B) Combination indices (CIs) derived from drug synergism in three KRASG12C-mutant cell lines assessed in 3D spheroid culture. Cell viability was determined after 5-day treatment with ARS-1620, second compounds, or the combination in a 1:1 ratio (dilution series from 1.5 nM to 10 μM), and CI values were calculated using CompuSyn 1.0 from three biological replicates. CI < 0.75 indicates synergism with ARS-1620 (red), CI = 0.75 to 1.25 indicates additivity, and CI > 1.25 indicates antagonism. (C) Clonogenic assays of three KRASG12C-mutant cell lines cultured with indicated compounds or combinations thereof in 2D adherent culture. H358 (300 nM), MIA PaCa-2 (1 μM), and H23 (1 μM) cells were treated using ARS-1620 and second compounds at a 1:1 concentration. Data are representative of three biological replicates.

  • Fig. 6 CDK4/6 coinhibition enhances global antiproliferative effects of KRASG12C inhibition.

    (A) Heatmap displaying top significant (adjusted P < 5 × 10−14 by a Wald test) differentially expressed genes in H358 cells between DMSO and combination treatments, hierarchically clustered based on Pearson correlation distances. Gene expression values were Z score–normalized to the average expression value of each row. Total RNA was isolated following 24-hour treatment with DMSO, palbociclib (1 μM, CDK4/6i), ARS-1620 (1 μM), or their combination in H358 cells. Gene ontology enrichment terms for four major clusters are indicated. Data represent two biological replicates. (B) Volcano plot of log2 fold changes and P values from the experiment in (A). (C) Statistical analysis of differences in gene expression among clusters shown in (A). Data points represent individual genes. Biological replicates are shown as black and gray points. ***P < 0.001, ****P < 0.0001 by an unpaired t test. n.s., not significant. (D) Pairwise correlation matrix from the experiment in (A) between treatment conditions displaying global similarities in expression profiles based on Euclidean distances. (E) Immunoblots of H358 or MIA PaCa-2 cells cotreated with a dose range of ARS-1620 with DMSO or palbociclib (1 μM, CDK4/6i) for 24 hours. Cells in (A) to (E) were grown in 3D spheroid culture. Immunoblots are representative of two biological replicates (see also fig. S6).

  • Fig. 7 Pharmacological targeting of CDs promotes response to KRASG12C inhibition in vivo.

    (A) Tumor volumes in mice bearing H358 xenografts and treated with vehicle, ARS-1620 (100 mg/kg), erlotinib (80 to 100 mg/kg, EGFRi), or their combination. Treatment was stopped 54 days after initial implantation to monitor the durability of observed responses. n ≥ 4 mice per group; error bars denote SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 comparing ARS-1620 and combination arms by a Mann-Whitney test. (B) Tumor volumes in mice bearing MIA PaCa-2 xenografts and treated with vehicle, ARS-1620 (100 mg/kg), palbociclib (100 mg/kg, CDK4/6i), or their combination. n ≥ 8 mice per group; error bars denote SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 comparing ARS-1620 and combination arms by a Mann-Whitney test.

Supplementary Materials

  • stke.sciencemag.org/cgi/content/full/12/583/eaaw9450/DC1

    Text S1. Chemical synthesis details and methods.

    Fig. S1. Genome-wide CRISPRi screening in the cancer cell line H358 identifies genes that influence cell growth and survival.

    Fig. S2. Genome-wide CRISPRi screening in the cancer cell line MIA PaCa-2 identifies genes that influence cell growth and survival.

    Fig. S3. FOSL1 minimally modulates p-ERK and p-S6 phosphorylation dynamics.

    Fig. S4. Unique genetic dependencies within common signaling modules are revealed by evaluating ARS-1620 CRISPRi selection screens from distinct cancer cell lines.

    Fig. S5. Cellular response to KRASG12C inhibition is mediated through phosphorylation changes in substrates of hit kinases.

    Fig. S6. CDK4/6i coinhibition minimally alters p-AKT and p-ERK phosphorylation dynamics.

    Fig. S7. EGFRi cotreatment enhances the suppression of oncogenic RAS signaling by ARS-1620 in vivo.

    Table S1. sgRNAs used for individual retesting.

    Data file S1. H358 ARS-1620 genome-wide CRISPRi sgRNA counts and phenotypes.

    Data file S2. H358 ARS-1620 genome-wide CRISPRi gene phenotypes.

    Data file S3. MIA PaCa-2 ARS-1620 genome-wide CRISPRi sgRNA counts and phenotypes.

    Data file S4. MIA PaCa-2 ARS-1620 genome-wide CRISPRi gene phenotypes.

    Data file S5. ARS-1620 global phosphoproteomics.

  • The PDF file includes:

    • Text S1. Chemical synthesis details and methods.
    • Fig. S1. Genome-wide CRISPRi screening in the cancer cell line H358 identifies genes that influence cell growth and survival.
    • Fig. S2. Genome-wide CRISPRi screening in the cancer cell line MIA PaCa-2 identifies genes that influence cell growth and survival.
    • Fig. S3. FOSL1 minimally modulates p-ERK and p-S6 phosphorylation dynamics.
    • Fig. S4. Unique genetic dependencies within common signaling modules are revealed by evaluating ARS-1620 CRISPRi selection screens from distinct cancer cell lines.
    • Fig. S5. Cellular response to KRASG12C inhibition is mediated through phosphorylation changes in substrates of hit kinases.
    • Fig. S6. CDK4/6i coinhibition minimally alters p-AKT and p-ERK phosphorylation dynamics.
    • Fig. S7. EGFRi cotreatment enhances the suppression of oncogenic RAS signaling by ARS-1620 in vivo.
    • Table S1. sgRNAs used for individual retesting.
    • Legends for data file S1 to S5

    [Download PDF]

    Other Supplementary Material for this manuscript includes the following:

    • Data file S1 (Microsoft Excel format). H358 ARS-1620 genome-wide CRISPRi sgRNA counts and phenotypes.
    • Data file S2 (Microsoft Excel format). H358 ARS-1620 genome-wide CRISPRi gene phenotypes.
    • Data file S3 (Microsoft Excel format). MIA PaCa-2 ARS-1620 genome-wide CRISPRi sgRNA counts and phenotypes.
    • Data file S4 (Microsoft Excel format). MIA PaCa-2 ARS-1620 genome-wide CRISPRi gene phenotypes.
    • Data file S5 (Microsoft Excel format). ARS-1620 global phosphoproteomics.

Navigate This Article