The stress sensor GCN2 differentially controls ribosome biogenesis in colon cancer according to the nutritional context

Abstract

Nutrient availability is a key determinant of tumor cell behavior. While nutrient-rich conditions favor proliferation and tumor growth, scarcity, and particularly glutamine starvation, promotes cell dedifferentiation and chemoresistance. Here, linking ribosome biogenesis plasticity with tumor cell fate, we uncover that the amino acid sensor general control non-derepressible 2 (GCN2; also known as eIF-2-alpha kinase 4) represses the expression of the precursor of ribosomal RNA (rRNA), 47S, under metabolic stress. We show that blockade of GCN2 triggers cell death by an irremediable nucleolar stress and subsequent TP53-mediated apoptosis in patient-derived models of colon adenocarcinoma (COAD). In nutrient-rich conditions, a cell-autonomous GCN2 activity supports cell proliferation by stimulating 47S rRNA transcription, independently of the canonical integrated stress response (ISR) axis. Impairment of GCN2 activity prevents nuclear translocation of methionyl-tRNA synthetase (MetRS), resulting in nucleolar stress, mTORC1 inhibition and, ultimately, autophagy induction. Inhibition of the GCN2-MetRS axis drastically improves the cytotoxicity of RNA polymerase I (RNA pol I) inhibitors, including the first-line chemotherapy oxaliplatin, on patient-derived COAD tumoroids. Our data thus reveal that GCN2 differentially controls ribosome biogenesis according to the nutritional context. Furthermore, pharmacological co-inhibition of the two GCN2 branches and RNA pol I activity may represent a valuable strategy for elimination of proliferative and metabolically stressed COAD cells.

Keywords: GCN2; colon cancer; methionyl‐tRNA synthetase; nucleolar stress; ribosome biogenesis.

Fig. 1

Ribosomal dysfunction in colon adenocarcinoma is correlated with the upregulation of ATF4 activity. (A) Gene Ontology (GO) enrichment analysis on TCGA database highlights an upregulation of ribosomal biogenesis processes in colon adenocarcinoma (COAD) compared to normal tissue. Top enrichment GO terms in COAD tumors versus normal tissues are shown and significance is expressed as −Log10 adjusted P‐value. (B) Dysregulated expression of riboproteins (RP) is higher in COAD compared to the normal tissue. TCGA COAD data expressed as median ± SD (min to max), Mann–Whitney test (***P < 0.001). (C) The amino acid deprivation gene signature is enriched in COAD tumors with high dysregulation of RP expression. TCGA COAD data expressed as median ± SD (min to max) Mann–Whitney test (**P < 0.01). (D) ATF4 gene signature is higher in colon tumors compared to normal tissue. TCGA COAD data expressed as median ± SD (min to max), Mann–Whitney test (***P < 0.001). (E) ATF4 gene signature is enriched in COAD with high dysregulation of RP expression. TCGA COAD data expressed as median ± SD (min to max), Mann–Whitney test (*P < 0.05).

Fig. 2

GCN2 activity is critical to promote colon adenocarcinoma cells plasticity. (A) GCN2 (gene name EIF2AK4) expression levels are higher in colon tumor tissues compared to the normal counterparts. Colonomics cohort, Unpaired t‐test. (B) Immunohistochemistry analysis confirmed that GCN2 amount is augmented in human colon adenocarcinoma (COAD) tumors (n = 19) compared to adjacent normal (n = 10) tissues. Quantification is provided on the right panel. Data are expressed as expressed as median ± SD (min to max), unpaired Student's t‐test. (C) Live‐and‐dead assay in HCT116 spheroids treated for 3 days with vehicle (DMSO) or GCN2i. To facilitate the observation of death induction, staining of non‐viable cells is also represented in shades of gray (right panel). Quantifications are expressed as mean ± SEM of independent experiments (n = 5), unpaired Student's t‐test with the P‐value (***P < 0.001). (D) Live‐and‐dead assay in LoVo spheroids treated for 3 days with vehicle (DMSO) or GCN2i. To facilitate the observation of death induction, staining of non‐viable cells is also represented in shades of gray (right panel). Quantifications are expressed as mean ± SEM of independent experiments (n = 3), unpaired Student's t‐test with the P‐value (***P < 0.001). (E) Live‐and‐dead assay in patient's derived colospheres, grown with or without the ROCK inhibitor (Y‐27632), after 3 days of GCN2i treatment. To facilitate the visualization of cell death induction, staining of non‐viable cells is also represented in shades of gray (right panel). Quantifications are expressed as mean ± SEM of independent experiments (n = 3), unpaired Student's t‐test with the P‐value (*P < 0.05). (F) Evaluation of cellular compaction in the core of HCT116 spheroids treated for 3 days with DMSO or GCN2i via bright‐field microscopy and on hematoxylin‐phloxine‐saffron stained spheroid slices. The dotted line delimitates the dark and compact from the transparent areas used for the zonation ratio calculation. Data are expressed as mean ± SEM of independent experiments (n = 3), unpaired two‐tailed t‐test with the P‐value (***P < 0.001). (G) Measurements of immunohistochemical staining against cleaved PARP and cleaved caspase 3 on HCT116 spheroids treated with DMSO or with GCN2i for 3 days. Data are expressed as mean ± SEM of independent experiments (n = 3), unpaired two‐tailed t‐test with the P‐value (**P < 0.01, ***P < 0.001). (H) RT‐qPCR analysis of ALDH1A3 and c‐MYC mRNA levels in HCT116 spheroids treated for 48 h with DMSO or GCN2i. Data are expressed as mean ± SEM of independent experiments (n = 3), unpaired two‐tailed t‐test with the P‐value (**P < 0.01, ***P < 0.001).

Fig. 3

GCN2 is required for repressing 47S transcription and preventing the TP53‐proapoptotic program upon low glutamine. (A) Fibrillarin localization assessed by immunofluorescence in glutamine‐starved HCT116 cells treated with DMSO or GCN2i (representative images of three independent experiments). (B) Northern blotting of cellular 47S and 34S rRNAs after indicated treatments. The position of the investigated probe on the 47S rRNA and sub‐product is indicated (representative images of three independent experiments). (C) Western blot analysis of TP53 and MDM2 protein amounts in HCT116 TP53 WT and HCT116 TP53 KO treated with DMSO or GCN2i upon low glutamine condition (0.1 mm, −Gln) for the indicated period of time (representative images of three independent experiments). (D) Live‐and‐dead assay in HCT116 spheroids wild‐type (WT) or knockout (KO) for TP53 treated for 3 days with vehicle (DMSO) or GCN2i. To facilitate the observation of death induction, staining of non‐viable cells is also represented in shades of gray (right panel). Quantifications are expressed as mean ± SEM of independent experiments (n = 3), one‐way ANOVA Tukey's post hoc test with the P‐value (*P < 0.05). (E) Measurements of cell mass impairment in HCT116 TP53 WT and HCT116 TP53 KO treated with GCN2i for 48 h in medium containing 4 or 0.1 mm levels of glutamine. Data are expressed as mean ± SEM of independent experiments (n = 5). Results of unpaired two‐tailed t‐test are indicated with the P‐value (*P < 0.05).

Fig. 4

GCN2 is necessary for colon adenocarcinoma cancer cells proliferation independently of the integrated stress response pathway. (A) Cell proliferation assessed by the confluency index of HCT116 cells treated or not with GCN2i for 96 h. Data are expressed as mean ± SEM of independent experiments (n = 3). Unpaired two‐tailed t‐test with P‐value (**P < 0.01, ***P < 0.001). (B) Cell mass was assessed every day over the indicated period of time in HCT116 cells treated or not with GCN2i by SRB assay. A daily refreshment of the medium (indicated by the black rounded arrow) was performed to ensure a full supply of nutrients over the time course of the experiment. Data are expressed as mean ± SEM of independent experiments (n = 3). Unpaired two‐tailed t‐test with P‐value (**P < 0.01, ***P < 0.001). (C) Relative number of HCT116 cells transfected either with a siRNA control (CTRL) or against GCN2 (GCN2) following 4 days of culture. Medium were daily refreshed (black rounded arrow) to prevent nutrients exhaustion. Data are expressed as mean ± SEM of independent experiments (n = 3). Unpaired two‐tailed t‐test with P‐value (***P < 0.001). (D) Cell mass measurement of four patient‐derived primary cells cultured in 2D conditions and treated with DMSO or GCN2i for 48 h. Media were refreshed daily (black rounded arrow). (E) Spheroid formation assays of HCT116 treated with DMSO or GCN2i. ATP content was measured after 7 days of treatment. Treatment was initiated concomitantly to plating. Data are expressed as mean ± SEM of independent experiments (n = 3). Unpaired two‐tailed t‐test with P‐value (***P < 0.001). (F) Western blot analysis of P‐eIF2a, ATF4 and CHOP protein amounts in HCT116 starved for leucine for 8 h in the presence of ISRIB (200 nm) or not (representative images of three independent experiments). (G) Measurements of cell mass in HCT116 cells treated with ISRIB (200 nm) for 48 h. Medium were daily refreshed (black rounded arrow) to ensure complete nutrients supply. Data are expressed as mean ± SEM of independent experiments (n = 5). Unpaired two‐tailed t‐test. (H) Western blot analysis of ATF4 protein levels in HCT116 cells transfected either with a siRNA control (siCtrl) or against ATF4 (siATF4) following 2 days of culture with siCTRL and siATF4 upon 8 h of amino acid deprivation (leucine starvation). ‘ns’ means not significantly difference between the two conditions (representative images of three independent experiments). (I) Cell proliferation assessed by the confluency index of HCT116 cells transfected with a siCtrl and siATF4 was monitored over 48 h. Data are expressed as mean ± SEM of independent experiments (n = 3). Unpaired two‐tailed t‐test with P‐value (*P < 0.05).

Fig. 5

GCN2 sustains the mTORC1 activity in proliferative colon cancer cells. (A) Gene set enrichment analysis in HCT116 cells silenced for GCN2 compared to the control. Top enrichment GO terms in siGCN2 transfected cells versus siCtrl are shown and significance is expressed as −Log10 adjusted P‐value. (B) Amino acid profiling of HCT116 cells transfected with a siRNA control or against GCN2 (n = 3). (C) Assessment of protein synthesis rate in HCT116 cells treated for 24 h with GCN2i (representative images of three independent experiments). (D) Western blot analysis of autophagic markers, LC3‐I, LC3‐II, and p62, in HCT116 cells treated with GCN2i for the indicated period of time (representative images of three independent experiments). (E) Autophagic flux analysis using HCT116 cells stably expressing the GFP‐LC3‐RFP‐LC3ΔG construct and treated with GCN2i for 48 h. Cleavage of GFP‐LC3 by autophagy released the RFP‐LC3 as an internal control, thus reduction of the GFP/RFP ratio illustrates induction of the autophagy flux. Data are expressed as the mean of quantification ± SEM of independent experiments (n = 3). Unpaired two‐tailed t‐test with P‐value (***P < 0.001). (F) Cell mass was measured in GCN2i‐treated HCT116 cells after 48 h in combination with chloroquine. Data are expressed relative to the vehicle as mean ± SEM of independent experiments (n = 7). Unpaired two‐tailed t‐test with P‐value (*P < 0.05). (G) Western blot analysis of mTORC1 pathway markers (P‐S6, P‐4EBP1, and P‐ULK1) amounts in HCT116 cells following 24 h of GCN2i treatment. Total S6, 4E‐BP1 and tubulin are provided as loading controls (representative images of three independent experiments). (H) Immunohistochemical staining against P‐S6 in HCT116 spheroids treated for 3 days with DMSO or GCN2i. Quantification of the IHC staining is provided on the right panel. Data are expressed as mean ± SEM of independent experiments (n = 3). Unpaired two‐tailed t‐test with P‐value (**P < 0.01). (I) Gene set enrichment analysis of colon adenocarcinoma (COAD) patients with low GCN2 expression (25%) compared to the rest of the COAD cohort (TCGA data).

Fig. 6

GCN2 inhibition inhibits the MetRS translocation and RNA polymerase I activity. (A) Time course analysis of cell cycle progression by monitoring HCT116 FUCCI cells upon DMSO or GCN2i for 48 h (T48h) and 72 h (T72h). Data are expressed relative to the vehicle as mean ± SEM of independent experiments (n = 3). Quantifications are expressed as mean ± SEM, unpaired t‐test. (B) Fibrillarin localization assessed by immunofluorescence in HCT116 cells treated with DMSO or GCN2i for 4 h. Nuclei were stained by Hoechst solution. Arrowheads designate enlarged nucleoli (representative images of four independent experiments). (C) Time course analysis of cellular rRNAs by northern blotting following GCN2i treatment (representative images of three independent experiments). (D) Cell proliferation assessed by the confluency index of HCT116 cells treated or not with GCN2i for 72 h. Data are expressed as mean ± SEM of independent experiments (n = 3). Unpaired two‐tailed t‐test with P‐value (**P < 0.01, ***P < 0.001). (E) Fibrillarin localization assessed by immunofluorescence in HCT116 cells, 48 h after transfection with a siRNA control (Ctrl) or against MetRS. Nuclei were stained by Hoechst solution. Arrowheads designate enlarged nucleoli (representative images of five independent experiments). (F) Western blot analysis of MetRS amounts in the cytoplasmic or nuclear fractions upon 2 h of GCN2i and GCNiB treatments. Histone H3 and tubulin are provided as loading controls for respectively nuclear and cytoplasmic samples (representative images of three independent experiments).

Fig. 7

GCN2 sensitizes colon adenocarcinoma cancer cells to inhibitors of the RNA polymerase I activity. (A) Cell death index of HCT116 treated or not with GCN2i combined to Actinomycin D (Act.D) or the vehicle (DMSO) for 48 h. Data are expressed relative to the vehicle as mean ± SEM of independent experiments (n = 3). One‐way ANOVA with Tukey's multiple comparisons test and P‐value (*P < 0.05). (B) Fibrillarin localization assessed by immunofluorescence in HCT116 cells treated with oxaliplatin (Oxa) in combination with DMSO or GCN2i for 4 h. Nuclei were stained by Hoescht solution (representative images of four independent experiments). (C) Cell death index of HCT116 treated or not with GCN2i combined to Actinomycin D (Act.D) or the vehicle (DMSO) for 48 h. Data are expressed relative to the vehicle as mean ± SEM of independent experiments (n = 3). One‐way ANOVA with Tukey's multiple comparisons test and P‐value (*P < 0.05). (D) Cell mass measurement of HCT116 treated for 48 h with Oxa in combination with GCN2i or ISRIB. Data are expressed relative to the vehicle as mean ± SEM of independent experiments (n = 4). One‐way ANOVA with Tukey's multiple comparisons test and P‐value (***P < 0.001). (E) Live‐and‐dead assay in patient‐derived colospheres, cultivated or not with ROCK inhibitor Y‐27632, after 5 days of treatment with Oxa and/or GCN2i. To facilitate the observation of cell death induction, staining of non‐viable cells is also represented in shades of gray (representative images of three independent experiments).