The integrated stress response is tumorigenic and constitutes a therapeutic liability in KRAS-driven lung cancer

Cancer imposes proteotoxic and other stresses that activate the unfolded protein response and the integrated stress response (ISR). ISR converges on phosphorylation of eIF2α (p-eIF2α), which dampens global translation while promoting selective translation of stress-adaptive transcripts. Although ISR upregulation is common in tumors, its precise role in tissue-specific tumor initiation and progression is unclear. Lung adenocarcinoma (LUAD), comprising ~40% of lung cancers, shows poor survival and frequently harbors activating KRAS mutations (~30%). This study asks whether ISR—particularly the PERK/p‑eIF2α arm—drives LUAD pathogenesis and KRAS-driven tumorigenesis, and whether ISR constitutes a therapeutic vulnerability. The authors assess prognostic significance of p‑eIF2α in a cohort of 928 LUADs and interrogate mechanistic links between p‑eIF2α and ERK signaling in mouse and human KRAS-mutant models, testing whether pharmacologic ISR inhibition impairs tumor growth and improves survival.

Background work establishes that ER stress activates the UPR and ISR via kinases PERK, HRI, GCN2, and PKR, which phosphorylate eIF2α to reprogram translation toward stress adaptation. PERK and GCN2 predominantly promote survival, whereas PKR and HRI can promote apoptosis. Prior studies implicate ISR components in various cancers but their roles in LUAD progression and KRAS-driven carcinogenesis are incompletely defined. DUSP family phosphatases, especially DUSP6, inactivate ERK and have been linked to tumor suppression and prognosis in LUAD. KRAS oncogenic signaling elevates cellular stress and may select for adaptive ISR activation, suggesting a potential therapeutic angle by targeting ISR.

- Human specimens and TMA analysis: A continuous cohort of 928 primary resected LUADs (1998–2015) was assembled into tissue microarrays (3×1 mm cores/patient). Duplex and multiplex IHC quantified cytoplasmic p‑eIF2α and nuclear p‑ERK (with cytokeratin masking) using automated digital pathology pipelines (Visiopharm; Akoya inForm). H-scores were validated against manual scoring. Associations with WHO tumor type, growth pattern, and Ki67 were tested. Survival (overall, cancer-specific, recurrence-free) was analyzed by Kaplan–Meier, log-rank tests, and Cox regression (univariate and multivariate). - Mouse models: KRAS+/LSL‑KRASG12D mice were crossed to conditional eIF2S1 S51A knock-in (eIF2αA/A) or wild-type (eIF2αS/S). Lung tumorigenesis was induced by intratracheal delivery of CRE-expressing lentivirus carrying TP53 shRNA (CA2 promoter). Tumors were monitored by high-frequency ultrasound at serial time points and assessed by H&E and IHC (p‑eIF2α, p‑ERK, DUSP6, Ki67, cleaved caspase‑3). A urethane-induced KRASQ61R/L lung tumor model with eIF2αS/A vs eIF2αS/S was also used. - Primary tumor cells and cell lines: Primary KRASG12D eIF2αS/S and eIF2αA/A lung tumor cells were isolated and cultured. Human LUAD cell lines: H23, H358 (KRASG12C), H1299, H1703 (WT KRAS). KRASG12C/G12D/G12V or WT KRAS were overexpressed in H1299 and H1703 for isogenic comparisons. - Perturbations: PERK was inhibited by GSK2606414 or siRNAs; ISR was antagonized by ISRIB; DUSP6 was inhibited by BCI or knocked down by siRNAs. ER stress was induced with thapsigargin (TG). Effects on p‑PERK, p‑eIF2α, ATF4, DUSP6, and p‑ERK were measured by immunoblotting. - Translational control assays: Polysome profiling with qRT–PCR quantified Dusp6 mRNA distribution in total and polyribosomal fractions to assess translational regulation. - Transcriptomics: RNA-seq of primary KRASG12D eIF2αS/S vs eIF2αA/A tumor cells (n=4 each). Differential expression (anota2seq, FDR<0.05, |log2FC|>1). Upstream Regulator analysis (IPA) and GO GSEA. - Functional assays and in vivo therapy: Clonogenic assays measured survival under PERK inhibition or ISRIB. Subcutaneous xenografts: H1299 cells overexpressing WT KRAS or KRASG12C in nude mice treated by oral ISRIB (10 mg/kg) or GSK2606414 (150 mg/kg). Orthotopic model: LLC (KRASG12C) cells delivered intratracheally into C57BL/6 mice; treated with ISRIB or PERK inhibitor beginning day 12, monitored by ultrasound. In situ KRASG12D lung tumors were treated with ISRIB starting 10 weeks post-induction; tumor volume by ultrasound and survival recorded. - Statistics: Appropriate two-tailed tests (t-test, ANOVA with multiple comparisons), survival analyses (log-rank, Cox). Ethical approvals for human tissue and animal work were obtained.

- Clinical cohort (n=928 LUADs): p‑eIF2α positivity associated with significantly poorer overall survival by Kaplan–Meier (HR≈1.368, p=0.0011). Univariate Cox: HR=1.427 (95% CI 1.162–1.752, p=0.00068); effect attenuated in multivariate models including stage, sex, performance status, and WHO type (HR=1.091, p=0.422). p‑eIF2α correlated positively with proliferation (Ki67; Spearman’s rho=0.361, p<2.2×10⁻¹⁶) and was highest in invasive/solid growth patterns (Kruskal–Wallis p<2.2×10⁻¹⁶). Mucinous tumors showed low p‑eIF2α. - Mouse KRASG12D model: Genetic prevention of eIF2α phosphorylation (eIF2αA/A) extended survival by ~18 weeks vs eIF2αS/S (log-rank p=0.0019), reduced tumor number and size by ultrasound and histology. - Mechanism: p‑eIF2α increased p‑ERK without increasing p‑MEK, implicating ERK phosphatases. DUSP6 protein and translational engagement were reduced by p‑eIF2α. Polysome profiling showed enrichment of Dusp6 mRNA in polyribosomes in eIF2αA/A vs S/S cells, indicating translational repression by p‑eIF2α. DUSP6 knockdown or pharmacologic inhibition (BCI) restored p‑ERK and survival in eIF2αA/A cells to levels of eIF2αS/S. - Transcriptomics: RNA-seq of KRASG12D tumors (eIF2αS/S vs A/A) identified 2,249 differentially expressed genes. IPA predicted activation of PERK (EIF2AK3), ATF4, DDIT3, CREBBP and ERK1/2 signaling; inhibition of HOXA10, ESRRA, ASXL1, MFN2. GSEA indicated enrichment of growth factor signaling and proliferation; decreased mitochondrial respiration gene sets. - Human LUAD cells: Mutant KRAS (G12C/G12D/G12V) upregulated PERK/p‑eIF2α and p‑ERK vs WT KRAS; PERK inhibition reduced colony formation more strongly in KRAS-mutant cells. - PERK dependency: PERK siRNA or GSK2606414 reduced p‑eIF2α and survival in KRASG12D eIF2αS/S but not eIF2αA/A tumor cells; PERK activation by TG decreased DUSP6 and increased p‑ERK only when eIF2α was phosphorylatable. In human KRASG12C cells (H23, H358), TG increased p‑PERK, p‑eIF2α, p‑ERK and decreased DUSP6; PERK inhibitor reversed these effects. - ISRIB effects: ISRIB antagonized p‑eIF2α translation, suppressing ATF4, increasing DUSP6, and decreasing p‑ERK in KRAS-mutant mouse and human cells, but not in WT KRAS cells. - In vivo therapy: ISRIB and PERK inhibitor significantly suppressed growth of H1299 KRASG12C xenografts; WT KRAS xenografts were less responsive, particularly to ISRIB. In orthotopic LLC (KRASG12C) tumors, ISRIB or PERK inhibitor reduced tumor volume (weeks 3–6); ISRIB increased DUSP6 and reduced p‑ERK in tumors, decreased Ki67, and increased cleaved caspase‑3. In endogenous KRASG12D lung tumors, ISRIB (10 mg/kg) initiated 10 weeks post-induction reduced tumor volumes at 24 and 38 weeks and significantly prolonged survival (p=0.0034).

The data demonstrate that the adaptive PERK/p‑eIF2α branch of the ISR promotes KRAS-driven LUAD by repressing translation of DUSP6, thereby sustaining ERK activation and tumor cell proliferation and invasiveness. Clinically, high p‑eIF2α correlates with aggressive histology, increased proliferation, and worse outcomes. Mechanistically, ISR fosters a pro-tumorigenic signaling milieu beyond ERK via ATF4/CHOP and other regulators, while dampening tumor suppressive and mitochondrial respiration programs. Mutant KRAS persistently activates PERK/p‑eIF2α, creating a dependency exploitable by ISR-targeted therapies. Pharmacologic ISR antagonism with ISRIB or PERK inhibition reduces p‑eIF2α function, restores DUSP6, lowers p‑ERK, curtails tumor growth, and improves survival in KRAS-mutant models, with selective efficacy over WT KRAS settings. These findings position ISR as a therapeutic liability in KRAS-mutant LUAD and support ISR-targeted strategies as a means to disrupt KRAS-addicted signaling.

This study identifies the ISR—specifically the PERK/p‑eIF2α axis—as a tumorigenic driver and actionable vulnerability in KRAS-driven lung adenocarcinoma. p‑eIF2α serves as a prognostic marker linked to aggressive tumor behavior. Mechanistically, p‑eIF2α represses DUSP6 translation, elevating ERK signaling to support tumor growth. ISR inhibition with ISRIB or PERK inhibitors suppresses KRAS-mutant tumor growth and extends survival in vivo, with minimal impact in WT KRAS contexts, highlighting therapeutic selectivity. Future work should evaluate ISR inhibitors clinically, explore combinations with KRAS G12C inhibitors or MAPK pathway inhibitors, delineate ISR-driven metabolic reprogramming, and assess efficacy across diverse KRAS mutations and LUAD subtypes.

- Prognostic significance of p‑eIF2α lost independence in multivariate models including growth pattern, indicating potential confounding by tumor histology. - Preclinical models (mouse and cell lines) may not fully recapitulate human tumor heterogeneity and microenvironment. - PERK inhibitors can have p‑eIF2α–independent effects; off-target or pathway-compensatory mechanisms were not exhaustively excluded. - The precise cis-regulatory features mediating DUSP6 translational repression by p‑eIF2α were not mapped. - Long-term safety and efficacy of ISR inhibition in humans remain unproven. - Generalizability to non-KRAS LUAD or other oncogenic contexts requires further validation.