MNK Inhibition Sensitizes KRAS-Mutant Colorectal Cancer to mTORC1 Inhibition by Reducing eIF4E Phosphorylation and c-MYC Expression

Abstract

KRAS-mutant colorectal cancers are resistant to therapeutics, presenting a significant problem for ∼40% of cases. Rapalogs, which inhibit mTORC1 and thus protein synthesis, are significantly less potent in KRAS-mutant colorectal cancer. Using Kras-mutant mouse models and mouse- and patient-derived organoids, we demonstrate that KRAS with G12D mutation fundamentally rewires translation to increase both bulk and mRNA-specific translation initiation. This occurs via the MNK/eIF4E pathway culminating in sustained expression of c-MYC. By genetic and small-molecule targeting of this pathway, we acutely sensitize KRAS^G12D models to rapamycin via suppression of c-MYC. We show that 45% of colorectal cancers have high signaling through mTORC1 and the MNKs, with this signature correlating with a 3.5-year shorter cancer-specific survival in a subset of patients. This work provides a c-MYC-dependent cotargeting strategy with remarkable potency in multiple Kras-mutant mouse models and metastatic human organoids and identifies a patient population that may benefit from its clinical application. SIGNIFICANCE: KRAS mutation and elevated c-MYC are widespread in many tumors but remain predominantly untargetable. We find that mutant KRAS modulates translation, culminating in increased expression of c-MYC. We describe an effective strategy targeting mTORC1 and MNK in KRAS-mutant mouse and human models, pathways that are also commonly co-upregulated in colorectal cancer.This article is highlighted in the In This Issue feature, p. 995.

Figure 1. Oncogenic KRAS results in aggressive rapamycin-resistant Apc-deficient tumor models

(A) Left, BrdU incorporation in the small intestine of vehicle- or rapamycin-treated Apc^fl/fl and Apc^fl/fl Kras^G12D/+ short-term model mice at 4- and 3-days post induction, respectively. Mice were injected with BrdU 2hours prior to sampling. ≥25 half crypts were counted per mouse with each point representing a biological replicate. P values are from one-way ANOVA Tukey’s multiple comparisons tests. Right, micrographs show representative H&E and BrdU staining. Red line marks the extent of the proliferative zone. (B) Top, schematic of treatment timeline. Bottom, survival of Apc^fl/+ Kras^G12D/+ tumor model mice treated with rapamycin or vehicle daily from 5days post-induction. (C) Top, schematic of treatment timeline. BrdU incorporation scored in small intestinal tumors from Apc^1322T/+ mice treated with rapamycin for 10days, with and without Rosa^CreER driven KRAS^G12D activation 7days after starting rapamycin. The average % BrdU positivity was scored in ≥2 adenomas per mouse with each point representing a biological replicate. P values are from one-way ANOVA Tukey’s multiple comparisons tests. Representative images of tumors (dashed lines) stained for BrdU antibody are shown. Proliferation in untreated mice is shown for comparison (left bar). All data are represented as mean ± S.E.M. Scale bars, 50μm. See also Figure S1.

Figure 2. Translation and proliferation are uncoupled following Kras mutation

(A) Growth of Apc^fl/fl or Apc^fl/fl Kras^G12D/+ organoids, following 30hour treatment with 1μM rapamycin, relative to vehicle-treated organoids. Technical triplicates were performed and averaged to show the biological replicates as plotted. P values are from paired student t tests. The relative growth for untreated Apc^fl/fl and Apc^fl/fl Kras^G12D/+organoids is set as 0. (B) Protein synthesis in small intestinal organoids of the indicated genotypes following 6hours of treatment with 250nM rapamycin or vehicle. Values are from 3 independent biological replicates ±SEM, represented relative to Apc^fl/fl vehicle treated organoids. P values are from paired student t tests or a Mann Whitney test. See also Figure S2.

Figure 3. P-eIF4E is increased by Kras mutation and synergizes with mTORC1 to promote proliferation

(A) Immunohistochemistry (IHC) staining for P-eIF4E in short-term model (left 2 panels) and tumor model colonic adenomas (right 2 panels) from Apc^{fl/fl or +} and Apc^{fl/fl or +} Kras^G12D/+ mice. Dashed line denotes adenoma boundary. Images are representative of ≥3 biological replicates. (B) Top, schematic of treatment timeline. Apc^fl/+ Kras^G12D/+ and Apc^fl/+ Kras^G12D/+ Eif4e^S209A/S209A tumor model mice were allowed to develop symptoms of intestinal tumors, treated with rapamycin for 5days, and then sampled 2hours after BrdU administration. Bottom, BrdU incorporation into colonic epithelial adenoma cells quantified from a minimum of 4 tumors per mouse, with each point representing a biological replicate P values are from one-way ANOVA Tukey’s multiple comparisons tests (C) Bottom, quantification of BrdU incorporation in Apc^fl/fl Kras^G12D/+ Eif4e^S209A/S209A short-term model mice treated with rapamycin or vehicle. Top, representative H&E, BrdU and P-eIF4E staining for each treatment, with red bar showing extent of proliferation. ≥25 half crypts were quantified per mouse, with the average for each mouse plotted in the figure. P value is from a Mann Whitney test. (D) Representative images of staining for P-eIF4E and P-RPS6 each duplexed with cytokeratin from a TMA of colorectal cancer patients. Examples of high and low staining are shown for each. (E) Correlation between P-eIF4E and P-RPS6 from the TMA shown in (D). Significance was tested using X ²-test. All data are represented as mean ± S.E.M. Scale bars, 50μm. See also Figure S3.

Figure 4. Mnk deletion suppresses mutant KRAS-driven tumor proliferation in combination with rapamycin treatment

(A) IHC for P-eIF4E in the small intestine of Apc^fl/fl Kras^G12D/+ short-term model mice with deletion of Mnk1, Mnk2 or both Mnk1 and Mnk2. Right, quantification of this P-eIF4E staining. (B) BaseScope analysis of Mnk1 and Mnk2 expression in Apc^fl/fl and Apc^fl/fl Kras^G12D/+ small intestines. (C) qPCR for Mnk1 and Mnk2 in Apc^fl/fl and Apc^fl/fl Kras^G12D/+ small intestines. Levels are expressed relative to Actb mRNA expression. (D) Top, schematic of treatment timeline. Apc^fl/+ Kras^G12D/+ and Apc^fl/+ Kras^G12D/+ Mnk1^-/- Mnk2^-/- tumor model mice were aged until symptoms of intestinal tumors, and then treated with rapamycin or vehicle until endpoint. Survival is plotted as the number of days on rapamycin or vehicle treatment. P value is from a Log-rank Mantel-Cox test. (E) BrdU incorporation scored from small intestines of Apc^fl/fl Kras^G12D/+ and Apc^fl/fl Kras^G12D/+ Mnk1^-/- Mnk2^-/- short-term model mice, treated with and without rapamycin. Apc^fl/fl Kras^G12D/+ data are reproduced from Figure 1A, for ease of comparison. ≥25 half crypts were quantified per mouse, with the average for each mouse plotted in the figure shown. P values are from one-way ANOVA Tukey’s multiple comparisons tests. Horizontal line represents the average for Apc^fl/fl Kras^G12D/+ mice. (F) Survival plot following orthotopic induction in the distal colon of Apc^fl/fl Kras^G12D/+ (n=3) or Apc^fl/fl Kras^G12D/+ Mnk1^-/- Mnk2^-/- mice (n=4). A cohort of Apc^fl/fl Kras^G12D/+ Mnk1^-/- Mnk2^-/- mice was also treated with rapamycin from day 16 (n=4), indicated by vertical dashed line. Colonic polyps were imaged by colonoscopy on the days shown. P value is from a Log-rank Mantel-Cox test. An Apc^fl/fl Kras^G12D/+ Mnk1^-/- Mnk2^-/- mouse on rapamycin was censored at 100days post induction not related to procedure and two were censored at the end of study at 202days post induction. All data are represented as mean ± S.E.M. Scale bars, 50μm. See also Figure S4.

Figure 5. Translation of C-myc requires P-eIF4E and active mTORC1 signaling

(A) Validation of immunoprecipitation (IP) of total eIF4E or P-eIF4E by western blotting (WB) for each respective target. (B) Top, schematic showing that total eIF4E antibody precipitates both phosphorylated and non-phosphorylated eIF4E, whereas P-S209 eIF4E antibody specifically precipitates the phosphorylated form. Bottom, qPCR for mRNA targets using RNA eluted from the eIF4E, P-S209 eIF4E IPs or IgG control precipitation which eluted no eIF4E. Data are plotted as the linearized Ct values for the RNA IP subtracted from the Ct value for the Input. The grey line in each graph is value for the Gapdh transcript. All data are the average of 4 biological replicates. Significance was tested by paired student t test (C) IHC for c-MYC in adenomas from Apc^fl/+ Kras^G12D/+ or Apc^fl/+ Kras^G12D/+ Mnk1^-/- Mnk2^-/- mice treated with or without rapamycin for 5days prior to sampling. Staining intensity was quantified using nuclear c-MYC H-scores shown in the adjacent graph. The average H-score from at least 7 adenomas per mouse, with each point shown a biological replicate. P values are from one-way ANOVA Tukey’s multiple comparisons tests. All data are represented as mean ± S.E.M. Scale bars, 50μm. See also Figure S5.

Figure 6. Targeting the MNKs with eFT508 combines with rapamycin to suppress the APC KRAS phenotype

(A) BrdU incorporation in Apc^fl/fl Kras^G12D/+ short-term model mice treated with eFT508 or eFT508 in combination with rapamycin. ≥25 half crypts were quantified per mouse, with the average for each mouse plotted in the figure shown. P value is from one-way ANOVA Tukey’s multiple comparisons tests. (B) Apc^fl/+ Kras^G12D/+ tumor model mice were treated with eFT508 alone, or in combination with rapamycin for 5days, after presenting with symptoms of intestinal adenomas. BrdU was administered 2hours prior to sampling then the percentage of BrdU-positive adenoma cells was quantified. ≥5 adenomas were analyzed per mouse then the average of these plotted in the graph. Each point represents a biological replicate. P value is from one-way ANOVA Tukey’s multiple comparisons tests. (C) Micrographs showing representative staining for BrdU incorporation and P-eIF4E and c-MYC levels from (B). (D) Quantification of c-MYC staining in (C) by H-score. P value is from one-way ANOVA Tukey’s multiple comparisons tests. (E) Left, schematic of treatment timeline. Right, Apc^fl/+ Kras^G12D/+ tumor model mice treated as in (B) and allowed to age on treatment until clinical endpoint. P value is from a Log-rank Mantel-Cox test. (F) Apc^fl/fl Kras^G12D/+ organoids were treated with eFT508 (30nM) or eFT508 in combination with rapamycin (250nM) for 24hours, and the change in growth compared with vehicle-treated cells. Each data point represents an independent biological replicate, which is the average of 3 technical triplicates. P value is from a paired student t test. All data are represented as mean ± S.E.M. Scale bars, 50μm. See also Figure S6.

Figure 7. Expression of a human C-MYC transgene completely reverses the beneficial effect of eFT508/rapamycin treatment in vivo.

(A) Top, indicator of c-MYC variants expressed. Apc^fl/+ Kras^G12D/+ R26-lsl-MYC tumor model mice were treated with eFT508 in combination with rapamycin for 5days after presenting with symptoms of intestinal adenomas. BrdU was administered 2hours prior to sampling and the percentage of BrdU-positive adenoma cells quantified. Data for Apc^fl/+ Kras^G12D/+ control mice are reproduced from Figure 6B for ease of comparison. % BrdU positivity was averaged in at least 5 adenomas per mouse, with each point representing an individual mouse. P values are from one-way ANOVA Tukey’s multiple comparisons tests. (B) c-MYC staining (bottom) and H-score quantification (top) in adenomas from Apc^fl/+ Kras^G12D/+ and Apc^fl/+ Kras^G12D/+ R26-lsl-MYC tumor model mice after 5days of treatment with eFT508/rapamycin. C-MYC H-Score was determined for ≥7 adenomas per mouse then the average plotted on the graph shown. P values are from one-way ANOVA Tukey’s multiple comparisons tests. (C) Top, indicator of c-MYC variants expressed. BrdU incorporation in Apc^fl/fl Kras^G12D/+ R26-lsl-MYC short-term model mice treated with eFT508 in combination with rapamycin. BrdU incorporation values for Apc^fl/fl Kras^G12D/+ mice treated with eFT508/rapamycin are reproduced from Figure 6A. ≥25 half crypts were scored per mouse then the average plotted as an individual point on the graph. P values are from one-way ANOVA Tukey’s multiple comparisons tests. (D) Top, indicator of c-MYC variants expressed. Growth of Apc^fl/fl Kras^G12D/+ or Apc^fl/fl Kras^G12D/+ R26-lsl-MYC organoids treated with eFT508 (30nM) and rapamycin (250nM) for 24hours compared to growth of vehicle-treated counterparts. Technical triplicates were averaged for 3 independent organoid lines, with these average values represented by each point on the graph. P value is from paired student t test. (E) REACTOME pathway plot showing significantly positively enriched pathways in RNA sequencing between rapamycin/eFT508 treated Apc^fl/fl Kras^G12D/+ R26-lsl-MYC compared to Apc^fl/fl Kras^G12D/+ intestines. The gene ratio indicates the fraction of genes in a pathway that are regulated, with the absolute number shown by circle size. The circle color indicates the significance of enrichment. (F) Kaplan-Meier survival curve of cancer-specific survival for GMS1 patients staining low for one or both of P-RPS6 S240/4 and P-eIF4E S209 (n=46) compared to those staining high for both markers (n=42). Median survival is annotated on each line. Significance was determined by Log-Rank test. All data are represented as mean ± S.E.M. Scale bars, 50μm. See also Figure S7.