LKB1 regulates JNK-dependent stress signaling and apoptotic dependency of KRAS-mutant lung cancers

Abstract

The efficacy of molecularly targeted therapies may be limited by co-occurring mutations within a tumor. Conversely, these alterations may confer collateral vulnerabilities that can be therapeutically leveraged. KRAS-mutant lung cancers are distinguished by recurrent loss of the tumor suppressor STK11/LKB1. Whether LKB1 modulates cellular responses to therapeutic stress seems unknown. Here we show that in LKB1-deficient KRAS-mutant lung cancer cells, inhibition of KRAS or its downstream effector MEK leads to hyperactivation of JNK due to loss of NUAK-mediated PP1B phosphatase activity. JNK-mediated inhibitory phosphorylation of BCL-XL rewires apoptotic dependencies, rendering LKB1-deficient cells vulnerable to MCL-1 inhibition. These results uncover an unknown role for LKB1 in regulating stress signaling and mitochondrial apoptosis independent of its tumor suppressor activity mediated by AMPK and SIK. Additionally, our study reveals a therapy-induced vulnerability in LKB1-deficient KRAS-mutant lung cancers that could be exploited as a genotype-informed strategy to improve the efficacy of KRAS-targeted therapies.

Fig. 1. LKB1 loss confers sensitivity to combined MAPK + MCL-1 inhibition in KRAS-mutant NSCLC models.

A Schema for testing sotorasib drug combinations. B Relative increased efficacy of sotorasib + AMG 176 combination compared to sotorasib alone (ΔAUC—see Fig. S2A for explanation) against KRAS^G12C-mutant NSCLC cell lines. Each dot represents an independent biological replicate, N = 4). Comparison of ΔAUC between KRAS-mutant NSCLC cell lines stratified according to LKB1 status. *p = 0.029 (C), *p = 0.032 (D), unpaired-nonparametric t test, 2-sided. KRAS-mutant NSCLC cell lines were treated with 0.1 µM of trametinib or 1 µM of sotorasib in combination with 1 µM of AMG 176 for up to 72 h and apoptosis was assessed by annexin positivity by flow cytometry (E, data are mean and S.E.M. of N = 3 biological replicates) or live-cell imaging (F, data are mean and S.E.M. of 3 technical replicates). For annexin positivity, percentage of apoptotic cells in vehicle group was used as a control and normalized to 0. Source data are provided as a Source Data file.

Fig. 2. Modulating LKB1 status alters sensitivity to MAPK + MCL-1 inhibition in vitro and in vivo.

A, B Comparison of relative ∆AUC for isogenic LKB1-proficient and deficient KRAS-mutant cell line pairs (EV—empty vector, LKB1—LKB1 expression vector, sgGFP or LKB1—CRISPR KO of GFP or LKB1). Each dot represents an independent biological replicate (N = 3-6). For Sotorasib, H2030: **p = 0.002, H2122: *p = 0.012, MGH1112-1: *p = 0.014, MGH1114-1: *p = 0.037, H358: **p = 0.007. For Trametinib, H2030: **p = 0.003, H2122: *p = 0.026, H23: **p = 0.002, A549: *p = 0.007, MGH1114-1: *p = 0.011, MGH1112-1: *p = 0.015, H358: **p = 0.0014, H441: ***p = 0.0005, SW1573: **p = 0.002. Paired-parametric t test, 2-sided. Apoptotic response of isogenic KRAS-mutant NSCLC cell lines after treatment with 0.1 µM trametinib or 1 µM sotorasib in combination with 1 µM of AMG 176 (annexin positivity assessed by flow cytometry (C) or live-cell imaging (D). C Each dot represents an independent biological replicate, N = 3-5, H23: **p = 0.003, MGH1112: *p = 0.015, H2030: **p = 0.0017, MGH9019-2: *p = 0.011, SW1573: ****p = 0.00001, H358: ****p = 0.000015, unpaired-nonparametric t test, two-sided. D data are mean and S.E.M. of 3 technical replicates. E Subcutaneous xenograft tumors were established from H2030 EV and H2030 LKB1 cell lines, and mice were treated with vehicle, sotorasib (30 mg/kg daily), trametinib (3 mg/kg daily), AMG 176 (50 mg/kg daily) or combination. Data shown are mean and S.E.M of N = 5-6 mice per arm, statistical difference between single agent and combination arms was determined using mixed 2-way ANOVA effects model, *p = 0.01, **p = 0.0084. Source data are provided as a Source Data file.

Fig. 3. JNK activation in LKB1-deficient cells underlies dependency on MCL-1.

A Phosphoproteomic analysis of isogenic KRAS-mutant NSCLC cell lines treated with 0.1 µM of trametinib for 48 h. B Left: Differential enrichment of phosphopeptide signatures in trametinib-treated isogenic cell line pairs. NES – normalized enrichment scores. Right, individual phospho-sites of JNK1 downstream substrates are annotated. C Change in phospho-JNK in response to MAPK + MCL-1 inhibition in isogenic H2030 and H358 cells (data are mean and S.E.M., each dot represents an independent biological replicate, N = 3-4, H2030: **p = 0.0031, H358: ***p = 0.0008, ***p = 0.0009, paired-parametric t test, 2-sided). D Change in cell number of H2030 EV cells with siJNK1 + 2 or negative control (siNC) after treatment with 0.1 µM trametinib or 1 µM sotorasib in combination with 1 µM AMG 176 quantified by live-cell imaging. Data are mean and S.E.M. of 3 technical replicates. E H2030 EV or LKB1 cells were transfected with siRNAs targeting JNK1 and 2 or siNC and then treated with sotorasib (S) or trametinib (T) ± AMG 176 (A) and viability was determined after 3 days. Each dot represents an independent biological replicate (N = 3, TA: *p = 0.017, SA: *p = 0.028, unpaired-parametric t test, two-sided). F Schematic of siRNA knockdown of LKB1 effectors in H2030 LKB1 cells. G H2030 EV or LKB1 cells transfected with corresponding siRNAs were treated with sotorasib or trametinib in the absence or presence of AMG 176 (1 µM) and viability was determined after 3 days. Each dot represents an independent biological replicate (N = 3, ****p = 0.00001, unpaired-parametric t test, 2-sided). H, I Cells were transfected with the indicated siRNAs and then treated with trametinib (0.1 µM) or sotorasib (1 µM) for 48 h, AMG 176 for 4 h or trametinib (0.1 µM) or sotorasib (1 µM) for 48 h followed by AMG 176 for 4 h. J siPP1B restores sensitivity (∆AUC) to combined sotorasib or trametinib + AMG 176. Each dot represents an independent biological replicate (N = 3, ****p = 0.00001, 2-way ANOVA). K HA-tagged WT NUAK1 (WT) or GKKK NUAK were over-expressed in H2030 isogenic cells and the interaction of NUAK1 and PP1B was assessed by Co-IP. Source data are provided as a Source Data file.

Fig. 4. JNK phosphorylates BCL-XL to drive an MCL-1 dependent state.

A Co-IP of BIM bound to MCL-1 in H2030 EV (empty vector) and H2030 LKB1 cells after treatment with vehicle, trametinib (0.1 µM) for 24 h or trametinib for 24 h followed by AMG 176 (1 µM) for 4 h. B Time course of BCL-XL S62 phosphorylation in isogenic H2030 and H358 cells by western blot after treatment with 0.1 µM trametinib + 1 µM AMG 176. C Experimental approach for expressing MCL-1 & BCL-XL phospho-site mutants while suppressing endogenous MCL-1 and BCL-XL. Interrogated phosphorylation sites are designated in yellow, phosphomimetic sites in red. D MCL-1 phospho-site mutants do not reduce sensitivity to MCL-1 inhibition (∆AUC). After induction of mutant MCL-1 (or WT control) and knockdown of endogenous MCL-1, H2030 EV cells were treated with trametinib in the absence or presence of AMG 176 (1 µM) and viability was determined after 3 days. Each dot is an independent biological replicate (N = 3). E BCL-XL S62A mutant decreases MCL-1 dependence. After induction of BCL-XL S62A (or WT control) and knockdown of endogenous BCL-XL, H2030 EV cells were treated with sotorasib or trametinib alone or in the presence of AMG 176 (1 µM) and viability was determined after 3 days. Each dot is an independent biological replicate (N = 6, ****p = 0.000001, unpaired-parametric t test, two-sided). H2030 EV cells expressing inducible WT or S62A mutant BCL-XL S62A (F) or H2030 LKB1 cells expressing inducible WT or BCL-XL S62E phosphomimetic (G) were treated with 0.1 µM trametinib or 0.1 µM trametinib in combination with 1 µM AMG 176 and cell number was quantified by live-cell imaging. Data are mean and S.E.M. of 3 technical replicates. V vehicle, T trametinib, A AMG 176, TA trametinib + AMG 176. Source data are provided as a Source Data file. Western blots and immunoprecipitation images are representative of at least 2 independent biological replicates.

Fig. 5. JNK activation drives an MCL-1 dependent state by modulating BIM:BCL-XL interactions.

A Scheme for approach to investigating BIM sequestration upon displacement from MCL-1. B Co-IP assessment of BIM bound to MCL-1 and BCL-XL in H2030 (LKB1-deficient) and SW1573 (LKB1 wild-type) cells after treatment with 0.1 µM trametinib for 24 h followed by 1 µM AMG 176 for 4 h. C Co-IP assessment of BIM bound to BCL-XL and MCL-1 in H2030 EV and LKB1 cells after treatment with 0.1 µM trametinib for 24 h followed by 1 µM AMG 176 for 4 h. D Co-IP assessment of BIM bound to BCL-XL and MCL-1 in H2030 EV (empty vector) with JNK knockdown after treatment with 0.1 µM trametinib for 24 h + 1 µM AMG 176 for 4 h. E Co-IP assessment of BIM bound to WT BCL-XL or BCL-XL mutants in H2030 EV (S62A) and H2030 LKB1 (S62E) cells after treatment with 0.1 µM trametinib for 24 h followed 1 µM AMG 176 for 4 h. HA-tag pull downs are specific for inducible constructs. F Quantification of Co-IP assessment of BIM bound to BCL-XL in H2030 and MGH1112 cells overexpressing BCL-XL WT or S62A mutants. Data are mean and S.E.M., each dot represents a biological replicate (N = 3-5, H2030: *p = 0.0433, MGH1112-1: *p = 0.0345, unpaired-parametric t test, 2-sided). G Effect of NUAK1/2 knockdown on BCL-XL S62 phosphorylation in response to treatment with 0.1 µM trametinib for 48 h (T) or trametinib for 48 h followed by 1 µM AMG 176 (TA) for 4 h. H Model depicting the mechanism by which LKB1 loss leads to an MCL-1-dependent state and sensitizes KRAS-mutant NSCLCs to combined KRAS or MEK + MCL-1 inhibition. Source data are provided as a Source Data file. Western blots and immunoprecipitation images are representative of at least 2 independent biological replicates.

Fig. 6. LKB1 loss in associated with MCL-dependence of KRAS^G12C-mutant NSCLC PDX tumors and patient tumor explants.

A KRAS^G12C-mutant NSCLC tumor cells were collected for BH3 profiling and assessment of BIM:MCL-1 interactions after ex vivo treatment with sotorasib or trametinib. B Change in MCL-1 (MS1 10 + 30 µM peptide) and BCL-XL (HRK 10 + 100 µM peptide) dependent priming of patient tumor cells after ex vivo treatment with 0.1 µM trametinib or 1 µM sotorasib treatment. Data is normalized to Vehicle control and error bar presents S.E.M. (N = 3,6. *p = 0.0476, unpaired-nonparametric t test, 2-sided). C Co-IP assessment of BIM:MCL-1 interaction in tumor cells isolated from pleural fluid after ex vivo treatment with 0.1 µM trametinib (T) or 1 µM sotorasib (S) for 16 h. Data is representatives of 2 independent biological replicates. D Mice bearing KRAS^G12C-mutant NSCLC patient derived xenograft (PDX) tumors were treated with sotorasib (100 mg/kg) for 3 days and harvested for BH3 profiling. Data shown is the difference in MCL-1 dependent priming (MS1 peptide) of sotorasib treated tumors normalized to vehicle control, each dot represents an independent mouse tumor with error bar as S.E.M. (N = 3–7, ***p = 0.0002, 2-way ANOVA). E Mice bearing KRAS^G12C-mutant NSCLC PDX tumors (LKB1-loss: MGH1112-1, MGH1138-1, MGH1196-2; LKB1 WT: MGH1062-1, MGH1145-1, MGH10199-3) were treated with vehicle, sotorasib (100 mg/kg), AMG 176 (50 mg/kg) or sotorasib (100 mg/kg) + AMG 176 (50 mg/kg) daily. Data shown are mean and S.E.M. of N = 6–10 animals per arm, statistical difference between single agent and combination arms was determined using mixed-effects model, ****p = 0.00001, *p = 0.013, 2-way ANOVA). F Survival curves of PDX tumor-bearing mice: the progression free survival (PFS) metric was determined by time to 20% increase in tumor volume from baseline measurement (****p = 0.0001, **p = 0.0087, Log-rank Mantel-Cox test). Source data are provided as a Source Data file. A–D Created in BioRender. Li, C. (2025) https://BioRender.com/p50l332.