A role for the unfolded protein response stress sensor ERN1 in regulating the response to MEK inhibitors in KRAS mutant colon cancers

Abstract

Background: Mutations in KRAS are frequent in human cancer, yet effective targeted therapeutics for these cancers are still lacking. Attempts to drug the MEK kinases downstream of KRAS have had limited success in clinical trials. Understanding the specific genomic vulnerabilities of KRAS-driven cancers may uncover novel patient-tailored treatment options.

Methods: We first searched for synthetic lethal (SL) genetic interactions with mutant RAS in yeast with the ultimate aim to identify novel cancer-specific targets for therapy. Our method used selective ploidy ablation, which enables replication of cancer-specific gene expression changes in the yeast gene disruption library. Second, we used a genome-wide CRISPR/Cas9-based genetic screen in KRAS mutant human colon cancer cells to understand the mechanistic connection between the synthetic lethal interaction discovered in yeast and downstream RAS signaling in human cells.

Results: We identify loss of the endoplasmic reticulum (ER) stress sensor IRE1 as synthetic lethal with activated RAS mutants in yeast. In KRAS mutant colorectal cancer cell lines, genetic ablation of the human ortholog of IRE1, ERN1, does not affect growth but sensitizes to MEK inhibition. However, an ERN1 kinase inhibitor failed to show synergy with MEK inhibition, suggesting that a non-kinase function of ERN1 confers MEK inhibitor resistance. To investigate how ERN1 modulates MEK inhibitor responses, we performed genetic screens in ERN1 knockout KRAS mutant colon cancer cells to identify genes whose inactivation confers resistance to MEK inhibition. This genetic screen identified multiple negative regulators of JUN N-terminal kinase (JNK) /JUN signaling. Consistently, compounds targeting JNK/MAPK8 or TAK1/MAP3K7, which relay signals from ERN1 to JUN, display synergy with MEK inhibition.

Conclusions: We identify the ERN1-JNK-JUN pathway as a novel regulator of MEK inhibitor response in KRAS mutant colon cancer. The notion that multiple signaling pathways can activate JUN may explain why KRAS mutant tumor cells are traditionally seen as highly refractory to MEK inhibitor therapy. Our findings emphasize the need for the development of new therapeutics targeting JUN activating kinases, TAK1 and JNK, to sensitize KRAS mutant cancer cells to MEK inhibitors.

Keywords: Colon cancer; ERN1; Ire1; JNK; JUN; MEK inhibitor.

Fig. 1

Unfolded protein response (UPR) executors are synthetic lethal with mutant RAS in S. cerevisiae. a Venn diagram showing the overlap of the RAS synthetic lethal (SL) gene deletion strains identified in the RAS1(V19) and RAS2(V19) genetic screens. b Gene Ontology (GO) enrichment analysis on the SL gene deletion strains from the RAS2(V19) screen identifies a variety of biological processes, including endosomal transport and protein targeting. c List of genes coding for protein complexes among the validated list of RAS2(V19) SL gene deletion mutants. Higher values correspond to stronger growth arrest in the presence of mutant RAS. The pathways and complexes in which the genes are involved are indicated. d The effect of the deletion of the UPR stress sensor IRE1 (ire1Δ) in RAS2(V19) screen (top) and in the empty vector (EV) control background (bottom). e Control vs mutant growth ratios of the UPR genes IRE1 and HAC1. Higher values correspond to stronger growth arrest in the presence of mutant RAS. f Schematic representation of the evolutionary conserved mechanism of UPR execution in yeast (top) and humans (bottom). Ire1 is responsible for the editing of HAC1 mRNA which produces an active executor of the UPR. ERN1 is the human ortholog of yeast IRE1; XBP1 is a functional human homolog of HAC1

Fig. 2

Effects of ERN1 inhibition in KRAS mutant human colon cancers. a, b Western blot analysis of ERN1 expression in control cells expressing non-targeting (NT) gRNA and LoVo ERN1^KO clones 5B, 6B, and 7B (a) and HCT-116 ERN1^KO clones C1, C2, and C3 (b). c, d qPCR analysis of spliced XBP1 mRNA (XBP1s) in control cells expressing non-targeting (NT) gRNA and LoVo ERN1^KO clones 5B, 6B, and 7B (c) and HCT-116 ERN1^KO clones C1, C2, and C3 (d). Error bars indicate standard deviation calculated from three biological replicates. e Representative colony formation assays of three different ERN1^KO clones compared to the non-targeting (NT) gRNA expressing control cells in the KRAS mutant LoVo (top) and HCT-116 colon cancer cells (bottom). Cells were maintained in the indicated range of concentrations of the MEK inhibitor selumetinib (AZD6244) for 10 days, stained and photographed. f, g Live cell proliferation assay (IncuCyte®) of control (NT gRNA) and ERN1^KO cells following exposure to the MEK inhibitor AZD6244. Error bars indicate standard deviation of three replicate experiments. h qPCR analysis of spliced XBP1 mRNA (XBP1s) levels following exposure of LoVo cells to increasing concentrations of the ERN1 kinase inhibitor. Error bars indicate standard deviation calculated from three replicate experiments. i Colony formation assay showing the effect of ERN1 kinase inhibitor on the proliferation of KRAS mutant LoVo cells in the presence of the indicated concentrations of the MEK inhibitor AZD6244. j Quantification of spliced XBP1 mRNA (XBP1s) levels following 1 h treatment with 100 nM of ER stress inducer thapsigargin (Tg) in the presence and absence of the ERN1 kinase inhibitor. k Quantification of the mRNA levels of the RIDD target CD59 after 1 h treatment with 100 nM thapsigargin (Tg) in the presence and absence of the ERN1 kinase inhibitor

Fig. 3

A genetic screen for resistance to MEK inhibitors in ERN1 knockout colon cancer. a Schematic outline of the genome-scale CRISPR/Cas9 knockout screen for resistance to MEK inhibition. Two different MEK inhibitors, selumetinib and trametinib, were used, each in two replicates, and compared to the untreated control population. b, c MA plots of the selumetinib (b) and trametinib screens (c). Horizontal dashed line indicates an arbitrarily imposed threshold of log₂ (fold change of treated over untreated) of 7 and vertical dashed line indicates mean number of reads in untreated samples of 50. Highlighted in color are the sgRNAs targeting DUSP4, DET1, COP1, CBFB, RUNX2, and STK40, that are found above these two thresholds (with the p adjusted of ≤ 0.1) in both the selumetinib (b) and the trametinib (c) screen. d, e Functional validation of DET1 and COP1 in LoVo ERN1^KO background. d Colony formation assays of DET1 and COP1 KO cells in the presence and absence of the MEK inhibitor AZD6244 (selumetinib) are shown relative to control cells having NT gRNA. Shown is a representative example of at least three biological replicates. e Live cell proliferation assay of DET1 and COP1 KO cells in the presence and absence of 1 μM AZD6244 compared to control cells expressing NT gRNA. Error bars indicate standard deviation calculated from three replicate experiments. f Western blot analysis of DET1 and COP1 expression in DET1 and COP1 knockout cells using antibodies against ERN1, DET1, COP1, JUN, p-ERK, and HSP90 as control both in the presence and absence of the MEK inhibitor AZD6244. g Median-centered log(IC₅₀) of five different MEK1 inhibitors in high (top 25%) and low (bottom 25%) expressing DET1 (left) and COP1 (right) CRC cell lines in the GDSC100 data set [42]. Cell lines with high DET1 or COP1 expression have significantly lower IC₅₀s (p = 0.004 for both DET1 and COP1). Log(IC₅₀) estimates were median-centered over all cell lines to make them comparable between MEK inhibitors

Fig. 4

Effect of genetic and pharmacologic downregulation of JUN on response to MEK inhibition. a Five different JUN targeting shRNAs were used to downregulate JUN in LoVo cells. JUN protein levels were quantified by western blotting (top), and the response to increasing concentrations of the MEK inhibitor AZD6244 on JUN knockdown cells is shown in colony formation assay (bottom). Empty vector infected control (ctrl) cells are shown here for comparison. b Biochemical analysis comparing ERN1^KO cells with their control counterparts (ctrl) in the presence and absence of the MEK inhibitor AZD6244 for the indicated number of hours. One-hour thapsigargin treatment (Tg) at 0.1 μM was used as a control for p-JUN induction. c Quantification of spliced XBP1 mRNA (XBP1s) in the presence and absence of 1 μM AZD6244 at indicated time points. Error bars indicate standard deviation calculated from three replicate experiments. d Biochemical analysis of JUN phosphorylation in the presence and absence of increasing concentrations of the JNK inhibitor SR-3306. One-hour of thapsigargin treatment (Tg) at 0.1 μM was used for p-JUN induction. e A representative colony formation assay of LoVo cells grown in the increasing concentrations of the JNK inhibitor SR-3306 (horizontally) and the increasing concentrations of the MEK inhibitor AZD6244 (vertically). f Live cell proliferation assay for the combination of the MEK inhibitor AZD6244 and the JNK inhibitor SR-3306 (black), each inhibitor individually (red and blue), and vehicle-treated control cells (yellow line). Error bars indicate standard deviation calculated from three replicate experiments. g Schematic representation of the signaling from the endoplasmic reticulum (ER) embedded ERN1 to JNK and JUN via its binding factor TRAF2 and TAK1. Shown in yellow are resistance screen hits DUSP4, DET1, and COP1, which are all negative regulators of JNK and JUN, respectively. h A representative colony formation assay showing the effect of the TAK1 inhibitor (5Z)-7-oxozeanol (5ZO) on the proliferation of KRAS mutant LoVo cells in the presence of the indicated concentrations of the MEK inhibitor AZD6244. i Live cell proliferation assay for the combination of the MEK inhibitor AZD6244 and TAK1 inhibitor 5ZO over the course of 10 days (240 h). Yellow line shows vehicle-treated control cells. Error bars indicate standard deviation calculated from three replicate experiments