WEE1 inhibitors trigger GCN2-mediated activation of the integrated stress response

Abstract

The WEE1 kinase negatively regulates CDK1/2 to control DNA replication and mitotic entry. Genetic factors that determine sensitivity to WEE1 inhibitors (WEE1i) are largely unknown. A genome-wide insertional mutagenesis screen revealed that mutation of EIF2A, a translation regulator, sensitized to WEE1i. Additionally, a genome-wide CRISPR-Cas9 screen revealed that inactivation of integrated stress response (ISR) kinase GCN2 or its co-factor GCN1 rescued WEE1i-mediated cytotoxicity. Conversely, loss of the collided ribosome sensor ZNF598 increased sensitivity to WEE1i. Mechanistically, WEE1i induced paradoxical GCN2 activation, ATF4 upregulation, and altered ribosome dynamics. ISR activation was independent of WEE1 presence, pointing at off-target GCN2 engagement by multiple chemically distinct WEE1i. ISR activation was observed in cancer cells as well as non-transformed cells, and required GCN1 and ongoing translation. Consequently, WEE1i induce multiple independent cellular effects: DNA damage, premature mitotic entry and sensitization to DNA-damaging chemotherapeutics in an ISR-independent fashion, as well as ISR activation independently of CDK1/2 activation. Importantly, low-dose WEE1 inhibition did not induce ISR activation, while it still synergized with PKMYT1 inhibition. Taken together, WEE1i trigger toxic ISR activation and translational shutdown, which can be prevented by low-dose or combination treatments, while retaining the cell cycle checkpoint-perturbing effects.

Fig. 1. Loss of EIF2A sensitizes cells to WEE1 inhibition.

a Experimental setup of the viral insertional mutagenesis screen in (b). b Sense/total insertion ratios from mutagenesis screens performed in HAP1 cells treated with AZD1775 or DMSO. Genes with significantly less sense insertions in AZD1775-treated cells are indicated. c Schematic representation of high-ranking gene mutations causing sensitization to WEE1i. d Pathway enrichment analysis, depicting the most highly enriched pathways involved in sensitization to WEE1i. e, f Dose-response curve of AZD1775 in control RPE1 TP53^KO and RPE1 TP53^KO EIF2A^KO cells (e) and OVCAR3, OVCAR3 EIF2A^KO #1 and OVCAR3 EIF2A^KO #2 cells (f). Cells were treated for 5 days and cellular viability was measured by MTT conversion. Data represent mean ± standard deviation (SD) (n = 3). g Immunofluorescence microscopy analysis of γH2AX in control RPE1 TP53^KO and EIF2A^KO cells treated with AZD1775 (500 nM) for 24 h. Mean foci intensity per cell is plotted. Medians of n = 218, 212, 228 or 206 cells, respectively, measured across 3 independent experiments. Paired t-test (two-sided) of medians, p ≤ 0.05 was considered significant. h scEdU-seq maximum normalized log counts are plotted for control RPE1 TP53^KO and RPE1 TP53^KO EIF2A^KO cells, ranked according to S-phase progression (y-axis) and binned per 400 kb (x-axis). A 40 megabase region of chromosome 2 is shown. All replicates are biological replicates unless indicated otherwise. Source data are provided as a Source data file.

Fig. 2. WEE1 inhibitors induce a translational attenuation mediated by GCN2 and the integrated stress response.

a Gene set enrichment analysis of RNA-seq data of RPE1 TP53^KO control and EIF2A^KO cells (more detailed analysis in Supplementary Fig. 3a). Circle size represents gene set size (control: n = 3, EIF2A^KO: n = 1). b, c SILAC-based proteomics analysis of RPE1 TP53^KO control and EIF2A^KO cells, showing differential protein expression at baseline (b) and upon AZD1775 treatment for 6, 24 and 48 h (c) (n = 2). Red datapoint indicates eIF2A, green datapoints indicate other screen hits. d RPE1 TP53^KO control and EIF2A^KO cells were treated with AZD1775 or Debio 0123 (1 µM) for 24 h, followed by a 10’ puromycin pulse. Puromycin incorporation was visualized by immunoblotting (left) and quantified (right). Data represent mean ± SD; RPE1 TP53^KO EIF2A^KO treated with Debio 0123 (n = 4); control RPE1 TP53^KO treated with DMSO or AZD1775 (n = 7); Other conditions (n = 5). e RPE1 TP53^KO PAC^KO cells were treated with AZD1775 (250 nM) and ISRIB for 24 h and subsequently labeled with L-azidohomoalanine (AHA). Immunoblot of AHA-labeled proteins and quantification is shown. Data represent mean and individual datapoints (n = 2). f Schematic overview of the CRISPR/Cas9 genome-wide AZD1775 resistance screen. g MAGeCK scores for individual genes in the RPE1 TP53^KO AZD1775 positive selection screen (day 12 vs day 0). h Cell confluency analysis of parental RPE1 TP53^KO PAC^KO or GCN2^KO cells. Mean ± range, n = 3 for control and GCN2^KO #1, n = 2 for GCN2^KO #2. i Quantification of clonogenic survival assays of parental RPE1 TP53^KO PAC^KO and ATF4^KO cells (corresponding images provided in Supplementary Fig. 4h). Mean ± SD, n = 3 for all conditions. j, k Representative histograms (j) and quantification (k) of ATF4-mScarlet flow cytometry measurements in RPE1 TP53^KO ATF4-mScarlet reporter cells treated with thapsigargin (1 µM), AZD1775 (1 µM), Debio 0123 (1 µM) and/or A92 (1 µM) for 24 h. Data represent mean ± SD (n = 3). Statistical analysis of (d, k): unpaired t-test (two-sided) with p ≤ 0.05 considered significant. All replicates are biological replicates unless indicated otherwise. Source data are provided as a Source data file.

Fig. 3. Integrated stress response activation by WEE1i is cell cycle independent.

a RPE1 TP53^KO PAC^KO cells were treated with AZD1775 (250 nM) for 6 h in the presence or absence of ISRIB (1 μM), pulse labeled with EdU, and processed for quantitative image-based cytometry. Cell cycle stage was defined by DNA content and EdU positivity. Each point represents a cell. Two independent replicates are defined by two different colors. Bars represent means of each replicate. Experimental set-up (b) and flow cytometry analysis (c) of RPE1 TP53^KO ATF4-mScarlet-NLS reporter cells. Cells were treated overnight with nocodazole. Mitotic cells were isolated and replated in the presence or absence of palbociclib. After 2 h, AZD1775 (1 µM) and/or A92 (1 µM) was added for 20 h. Data represent mean ± SD (n = 3). Flow cytometry gating strategy (d) and analysis of γH2AX in RPE1 TP53^KO cells after treatment with AZD1775 (1 µM) and/or A92 (1 µM) in the absence (e, left) or presence (e, right) of nocodazole. Data represent mean ± SD (n = 4). f Flow cytometry analysis of MPM2-positivity in RPE1 TP53^KO cells after treatment with nocodazole, doxorubicin, AZD1775 (1 µM) and/or A92 (1 µM). Data represent mean ± SD (n = 3). g Flow cytometry analysis of MPM2-positivity in control RPE1 TP53^KO PAC^KO and ATF4^KO #1 or GCN2^KO #1 cells after treatment with nocodazole, doxorubicin and AZD1775 (1 µM). Data represent mean ± SD (parental conditions n = 4, KO conditions n = 3). h Immunoblot of resting PBMCs or PBMCs stimulated with anti-CD3/CD28 beads. PBMCs were treated with DMSO, AZD1775 (500 nM) or thapsigargin (500 nM) for 24 h. Representative blot of n = 3 independent experiments. i RPE1 TP53^KO PAC^KO and GCN2^KO cells were treated with AZD1775 and/or cisplatin for 5 days. Cell survival was analyzed by MTT conversion. Data represent mean ± SD (n = 4). Statistical analysis for c, e, f, g was performed using unpaired t-tests (two-sided), with p ≤ 0.05 considered significant. All replicates are biological replicates unless indicated otherwise. Source data are provided as a Source data file.

Fig. 4. WEE1i alter the response to translation perturbations.

a Schematic overview of the CRISPR screen in (b). b NormZ scores of individual genes in an RPE1 TP53^KO PAC^KO Cas9 negative selection AZD1775 CRISPR screen (day 15 vs. day 0). Positive normZ scores indicate suppression, and negative normZ scores indicate synergy. Genes are color-coded by category; Orange: cell cycle checkpoint kinases, Green: ribosome quality control, Blue: ISR kinases GCN1L1 (GCN1) and EIF2AK4 (GCN2). c Schematic representation of ribosome quality control pathway members with high-ranking screen hits sensitizing to WEE1i. d Schematic representation of single ribosome mRNA translation kinetics measured by socRNA-mediated GFP translation. e Representative image of socRNA-mediated GFP translation as analyzed in (f). f GFP foci intensity over time (n = 1, tracking of 67 (no drug) or 73 (AZD1775) socRNAs). Lines and shaded areas indicate mean ± SD. g Elongation speed of a single ribosome on a socRNA in control or AZD1775-treated cells. Data represent mean ± SD (no drug: n = 58, AZD1775: n = 54, from 2 experiments), unpaired t-test (two-sided) with p ≤ 0.05 considered significant. h Schematic representation of a socRNA containing an Xbp1 pause sequence. Puromycin addition triggers the release of ribosomes and nascent peptide chains from a socRNA, allowing quantification of ribosome copies per socRNA. i Visual representation of ribosome release upon puromycin addition to socRNA, as quantified in (j). j Fraction of analyzed socRNAs containing indicated numbers of translating ribosomes per socRNA after AZD1775 treatment (300 nM, 24 h) with or without GCN2i A92 (1 µM). Mean ± range (No drug n = 3; AZD1775 n = 4, AZD1775 + A92 n = 2). k Immunoblot of RPE1 TP53^KO cells after 30 min treatment with emetine (0.5 µg/ml) and puromycin pulse (n = 1). l Immunoblot of RPE1 TP53^KO cells co-treated with AZD1775 (250 nM) and indicated doses of emetine according to the indicated timeline. Representative experiment of n = 2. m Immunoblot of RPE1 TP53^KO cells treated with AZD1775 (250 nM) for 4.5 h, and during the last 30 min co-treatment with emetine (0.5 µg/ml). Representative experiment of n = 2. All replicates are biological replicates unless indicated otherwise. Source data are provided as a Source data file.

Fig. 5. AZD1775 and Debio 0123-mediated GCN2 activation is independent of WEE1.

a Relative ATF4-mScarlet MFI in RPE1 ATF4-mScarlet reporter cells after treatment with 125, 250, 500, or 1000 nM of AZD1775, RP-6306, VE-822 or AZD7762. Data represent mean ± SD (DMSO: n = 4; other conditions: n = 3). Statistical analysis was performed using a one-way ANOVA with Dunnett’s multiple comparisons with p ≤ 0.05 considered significant. b Immunoblot of RPE1 TP53^KO cells treated with siRNA targeting WEE1 for 72 h and AZD1775 (250 nM) for 2.5 or 5 h. Representative blot of n = 3 experiments. WEE1 and GCN2 engagement as measured by NanoBRET kinase target engagement assay. HEK293 cells were transfected with WEE1-NanoLuc (c) and NanoLuc-GCN2 (d) and incubated with K-10 tracer (0.5 μM), and indicated doses of CC1, GCN2iB, AZD1775, Debio 0123, ZNL-02-096, RP-6306 or neratinib for 2 h prior to substrate addition and BRET signal detection. Data represent mean ± SD (n = 3). Representative histograms (e) and quantification (f) of ATF4-mScarlet flow cytometry measurements in RPE1 TP53^KO ATF4-mScarlet reporter cells treated with siRNAs targeting GCN1 and AZD1775 or neratinib (1 µM) for 5 h. Data represent mean ± SD (n = 3), statistical analysis: two-way ANOVA with Tukey’s multiple comparisons test, with p ≤ 0.05 considered significant. g Model of WEE1 inhibitor mode of action and consequences. All replicates are biological replicates unless indicated otherwise. Source data are provided as a Source data file.

Fig. 6. WEE1i and PKMYT1i synergistically induce cytotoxicity without WEE1i-induced ISR activation.

a Relative ATF4-mScarlet MFI in RPE1 TP53^KO ATF4-mScarlet reporter cells after treatment with the indicated doses of Debio 0123. Data represent mean ± SD (n = 3). b Cell viability in parental RPE1 TP53^KO PAC^KO, GCN2^KO #1 and GCN2^KO #2 cells after treatment with Debio 0123 and RP-6306 at the indicated doses (n = 5). c ZIP synergy scores from cell survival matrices shown in (b, d). Data represent mean ± SD (n = 5). d Synergy plots with ZIP synergy scores of heatmaps shown in (b). e Relative ATF4-mScarlet MFI in RPE1 TP53^KO ATF4-mScarlet reporter cells treated with the indicated doses of AZD1775, Debio 0123 and RP-6306. Data represent mean ± SD (n = 4). Statistical analysis (a, c, e) was performed using paired t-tests (two-sided), with p ≤ 0.05 considered significant. All replicates are biological replicates unless indicated otherwise. Source data are provided as a Source data file.