Targeting polyamine metabolism and ferroptosis enhances the efficacy of KRAS-targeted therapy depending on KEAP1 status

Abstract

The resistance to KRAS-targeted therapies, particularly due to co-occurring gene mutations, remains a significant challenge. Through a metabolite library screening, we reveal that polyamines sensitize KRAS inhibitors only in KRAS^MU/KEAP1^WT cells but not in KRAS^MU/KEAP1^MU cells. Transcriptome sequencing and metabolome profiling pinpoint SAT1, the key enzyme in polyamine metabolism, as essential for this divergence. In KRAS^MU/KEAP1^WT context, treatment of KRAS inhibitors activates JNK/c-Jun pathway and SAT1 expression, while the augmented SAT1 facilitates polyamine metabolism and KRAS inhibitors-induced ferroptosis. Conversely, in KRAS^MU/KEAP1^MU cells, activated JNK promotes the degradation of NRF2, thereby inhibiting SAT1 expression. Our results further demonstrate that polyamine supplementation enhances KRAS-targeted therapy in KRAS^MU/KEAP1^WT resistant cells, patient-derived organoids, xenografts, and spontaneously tumorigenic mice, while KRAS^MU/KEAP1^MU models require lentivirus or adeno-associated virus-mediated SAT1 overexpression prior to polyamine treatment, to augment ferroptosis and drug sensitivity. Our findings highlight SAT1-mediated polyamine metabolism as a promising target in precision treatments for KRAS-mutant cancers.

Fig. 1. Polyamines synergistically enhance the efficacy of KRAS inhibitors in a KEAP1-dependent manner.

A Flow chart of the metabolic library screening process. Created in BioRender. Bian, Y. (2025) https://BioRender.com/3p0ref1. B Relative viability of metabolic-treated vs vehicle (DMSO)-treated MIAPACA2 cells showed the metabolic library screening results. MIAPACA2 cells were treated with KRAS G12C inhibitors sotorasib or adagrasib (MIAPACA2, sotorasib, IC50 = 40 nM; adagrasib, IC50 = 60 nM), along with metabolics contained in the library (MCE HY-L030) or vehicle (DMSO) for 72 h. The algorithm for “Relative viability” showed that “X” means a specific metabolic, “V” means vehicle (DMSO). C Relative viabilities of MIAPACA2 and H23 cells treated with gradient concentrations of spermine and KRAS^G12C inhibitors (sotorasib or adagrasib) for 72 h were shown. 2D surface response for cell viability and 3D surface Bliss synergy score were shown. (n = 3 independent experiments). D Relative viability of 15 pan-cancer KRAS^G12C mutant cell lines treated with vehicle (DMSO), spermine (100 nM), sotorasib (IC50), and combination (spermine and sotorasib) for 72 h. The cell lines including pancreatic adenocarcinoma cell line MIAPACA2; bladder urothelial carcinoma cell line UMUC3; esophageal squamous cell carcinoma cell line KYSE410; lung adenocarcinoma cell lines H23, H2122, LU99, SW1573, H2030, H1792, H358, LU65 and HCC44; ovarian epithelial tumor cell line OV56; and colorectal adenocarcinoma cell lines JVE-015 and SW1463. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. E Venn plot displayed two genes (KRAS and KEAP1) from the intersection of mutant genes in non-synergy effect cell lines, including H23, H2122, H2030, H1792, and HCC44. F Workflow of KRAS^G12C mutant PAAD and LUAD patient-derived organoids (PDOs) establishment. G The number of PAAD and LUAD organoids, accompanied by an informational schematic that includes the patients’ type of mutation and differentiation stage were shown. H, I Representative brightfield images of PAAD PDOs (from PAAD patient #1 and #4) (H) and LUAD PDOs (from LUAD patient #1 and #2) (I) treated with KRAS^G12C inhibitors (sotorasib, 300 nm; adagrasib, 300 nM), spermine (1000 nM) or combination for 72 h (scale bars, 100 μm). J–K Luminescence measurement showed the relative viability of PAAD PDOs (from PAAD patient #1 and #4) (J) and LUAD PDOs (from LUAD patient #1 and #2) (K) treated with KRAS^G12C inhibitors (sotorasib, 300 nm; adagrasib, 300 nM), spermine (1000 nM) or combination for 72 h. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. Source data are provided as a Source Data file.

Fig. 2. KEAP1^WT-dependent SAT1 upregulation enhances the synergy between polyamines and KRAS inhibitors.

A Relative viabilities of MIAPACA2 with KEAP1 knockout (KO) and H23 cells with KEAP1 overexpression (OE) treated with gradient concentrations of spermine and sotorasib or adagrasib for 72 h. 2D surface response for cell viability and 3D surface Bliss synergy response were shown. (n = 3 independent experiments). B Schematic depiction of polyamine metabolism. Created in BioRender. Bian, Y. (2025) https://BioRender.com/3p0ref1. C The volcano plots showed differentially expressed genes (DEGs) in MIAPACA2 and H23 cells treated with KRAS^G12C inhibitors (sotorasib: MIAPACA2, IC50 = 40 nM; H23, IC50 = 500 nM; adagrasib: MIAPACA2, IC50 = 60 nM; H23, IC50 = 100 nM, 72 h) by RNA-seq analysis. ODC1, SAT1, SMOX, and PAOX are critical in polyamine metabolism. “DN” means downregulated in the KRAS inhibitors treatment group, and “UP” means upregulated in the KRAS inhibitors treatment group (|log2FC | > 0.4 and p < 0.05). D qRT-PCR and WB verified the mRNA and protein expression level of ODC1, SAT1, SMOX, and PAOX in MIAPACA2 and H23 cell lines treated with sotorasib or adagraisb (IC50) for 72 h. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. Source data are provided as a Source Data file. E Polyamine metabolites level in MIAPACA2 and H23 cells following KRAS inhibitors treatment (IC50, 72 h). (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. Source data are provided as a Source Data file. F Polyamine metabolites level in MIAPACA2 (NC and SAT1-KO) and H23 (NC and SAT1-OE) cells. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. Source data are provided as a Source Data file. G Relative viabilities of MIAPACA2 cells (NC and SAT1-KO group) and H23 cells (NC and SAT1-OE group) treated with spermine (100 nM) or spermidine (100 nM), KRAS inhibitors (IC50), or combination for 72 h. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. Source data are provided as a Source Data file. H qRT-PCR and WB showed the expression level of SAT1 in MIAPACA2 cells (NC, KEAP1-KO, KEAP1-WT, and KEAP1-MU group) and H23 cells (NC and KEAP1-OE group). (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. Source data are provided as a Source Data file. Source data are provided as a Source Data file.

Fig. 3. SAT1-mediated polyamine catabolism enhances KRAS inhibition in KRASMU/KEAP1WT through ferroptosis.

A Relative viabilities in MIAPACA2 cells treated with spermine (100 nM) or spermidine (100 nM) combined with or without DFO (5000 nM), fer-1 (500 nM), z-VAD (500 nM), or necrosulfonamide (250 nM) under treatment with sotorasib or adagrasib. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. B–D MDA (B), lipid peroxidation (C), and ferrous iron (D) level of MIAPACA2 and H23 cells treated with spermine (100 nM) or spermidine (100 nM), sotorasib or adagrasib (IC50) and combination for 72 h. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. E Transmission electron microscopy images of MIAPACA2 and H23 cells treated with spermine (100 nM) or spermidine (100 nM), sotorasib or adagrasib (IC50) and combination for 72 h (scale bars, 5 μm). F Number of damaged mitochondria per MIAPACA2 or H23 cell with cristae disappearance and mitochondrial shrinkage. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. G Relative enzymatic activity of SAT1 in H23 NC cells, SAT1-OE cells with or without critical enzymatic site mutation (Tyrosine 140 replaced by phenylalanine, Y140F).(n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. H, I MDA (H), lipid peroxidation (I) level of MIAPACA2 cells with SAT1-KO and H23 cells with SAT1-OE or SAT1-Y240F treated with spermine (100 nM) or spermidine (100 nM), sotorasib or adagrasib (IC50) and combination for 72 h. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. J Transmission electron microscopy images of MIAPACA2 cells with SAT1-KO and H23 cells with SAT1-OE cells treated with spermine (100 nM) or spermidine (100 nM), sotorasib or adagrasib (IC50) and combination for 72 h (scale bars, 5 μm). K Number of damaged mitochondria per MIAPACA2 SAT1-KO or H23 SAT1-OE Cell with cristae disappearance and mitochondrial shrinkage. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. L Ferrous iron level of MIAPACA2 cells with SAT1-KO and H23 cells with SAT1-OE treated with spermine (100 nM) or spermidine (100 nM), sotorasib or adagrasib (IC50) and combination for 72 h. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. Source data are provided as a Source Data file.

Fig. 4. Overcoming drug resistance through SAT1-mediated polyamine metabolism and ferroptosis.

A Stepwise dose escalation of sotorasib (from 0 to 6400 nM) or adagrasib (from 0 to 2560 nM) throughout the process to generate KRAS^G12C inhibitors-resistant cell lines (MIAPACA2-Sotorasib-R, MIAPACA2-Adagrasib-R, H23-Sotorasib-R, and H23-Adagrasib-R). Created in BioRender. Bian, Y. (2025) https://BioRender.com/3p0ref1. B Relative viability of parental and four resistant cells incubated in increasing concentrations of sotorasib or adagrasib for 72 h (MIAPACA2-sotorasib-R, IC50 = 1020 nM; MIAPACA2-adagrasib-R, IC50 = 1420 nM; H23-sotorasib-R, IC50 > 5000 nM; H23-adagrasib-R, IC50 = 2338 nM). (n = 3 independent experiments) Data were analyzed by two-way ANOVA test. C Relative viabilities of MIAPACA2 and H23 resistant cells treated with gradient concentrations of spermine and sotorasib or adagrasib for 72 h. 2D surface response for cell viability and 3D surface Bliss synergy response were shown. (n = 3 independent experiments). D The volcano plots showed DEGs in MIAPACA2-Sotorasib-R, MIAPACA2-Adagrasib-R, H23-Sotorasib-R, and H23-Adagrasib-R cells versus parental cells by RNA-seq analysis. “DN” means downregulated in KRAS^G12C inhibitors resistant cells, and “UP” means upregulated in KRAS^G12C inhibitors resistant cells (|log2FC | > 0.4 and p < 0.05).E, F qRT-PCR (E) and WB (F) verified the expression level of ODC1, SAT1, SMOX, and PAOX in KRAS^G12C inhibitors-resistant MIAPACA2 and H23 cell lines. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. G The heatmap displayed the polyamine metabolites level in MIAPACA2 and H23 resistant cells following KRAS inhibitors treatment (IC50, 72 h). H Lipid peroxidation level of KRAS^G12C inhibitors-resistant MIAPACA2 and H23 cells treated with spermine (100 nM), sotorasib or adagrasib (IC50 in resistant cells), and combination for 72 h. (n = 3 independent experiments) Source data are provided as a Source Data file.

Fig. 5. KRAS inhibitors modulate SAT1 expression through JNK/NRF2/SAT1 and JNK/c-Jun/SAT1 pathways.

A WB showed the expression level of SAT1 and the involved pathway factors, including JNK and c-Jun, NRF2, as well as KEAP1 in MIAPACA2 (NC and KEAP1-KO group) and H23 (NC and KEAP1-OE group) cells treated with vehicle (DMSO) or KRAS inhibitors (IC50) for 72 h. (n = 3 independent experiments). B WB displayed the change in the expression level of SAT1 and the involved pathway factors, including JNK and c-Jun, NRF2 and KEAP1 in MIAPACA2 (NC and KEAP1-KO group) and H23 (NC and KEAP1-OE group) cells treated with vehicle (DMSO) or KRAS inhibitors, under treatment of JNK inhibitor JNK-IN-8 (10 nM). (n = 3 independent experiments). C Genome-wide data of SAT1 from the ENCODE database showed the c-Jun and NRF2 binding peak within the promotor region of SAT1. D ChIP-qPCR experiments showed the binding intensities of c-Jun and NRF2 to the promotor region of SAT1 in MIAPACA2 and H23 cells (n = 3 independent experiments). Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. E WB showed the results of DNA pull down assays of SAT1. F Dual-luciferase assays showed the fluorescence intensity in four groups: a wild-type control lacking any mutations, a mutant with disrupted putative NRF2 binding sites, a mutant with disrupted putative c-Jun binding sites, and a double mutant with both NRF2 and c-Jun putative binding sites disrupted in MIAPACA2 and H23 cells treated with vehicle (DMSO) or KRAS inhibitors (IC50) (n = 3 independent experiments). Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. G The immunofluorescence images of KEAP1, p-JNK, p-c-Jun, NRF2, and SAT1 in MIAPACA2 cells (NC and KEAP1-KO group) treated with KRAS inhibitors (IC50) for 72 h (scale bars, 50 μm). H Relative fluorescence intensity of KEAP1, p-JNK, p-c-Jun, NRF2, and SAT1 in MIAPACA2 cells (NC and KEAP1-KO group) treated with KRAS inhibitors (IC50) for 72 h. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. Source data are provided as a Source Data file.

Fig. 6. Precisely targeting SAT1-mediated polyamine metabolism to enhance KRAS inhibitors in PDOs, xenograft and spontaneous lung cancer mouse models.

A Workflow of the establishment of PDOs with SAT1 KO or SAT OE. Created in BioRender. Bian, Y. (2025) https://BioRender.com/3p0ref1. B WB showed the effect of SAT1 KO and SAT1 OE in PAAD and LUAD PDOs. C Luminescence measurement showed the relative viability of PAAD and LUAD PDOs (NC and SAT1-KO or SAT1-OE group) treated with spermine (100 nM), sotorasib or adagrasib (300 nM), and combination for 72 h. (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. D MDA level of PAAD and LUAD PDOs (NC and SAT1-KO or SAT1-OE group). (n = 3 independent experiments) Data were analyzed by two-tailed Student’s t-test and were presented by mean ± SD. E Schematic design of xenograft tumor assay. Created in BioRender. Bian, Y. (2025) https://BioRender.com/3p0ref1. F–H Tumor images (F), growth curves (G) (Data were analyzed by two-way ANOVA test.), and tumor weight at the end of the treatment (H) (Data were analyzed by two-tailed Student’s t-test.) of resected tumors from MIAPACA2 cells. (n = 5 mice/group, one experiment) (I–K) Tumor images (I), growth curves (J) (Data were analyzed by two-way ANOVA test.), and tumor weight at the end of the treatment (K) (Data were analyzed by two-tailed Student’s t-test.) of resected tumors from H23 cells. (n = 5 mice/group, one experiment) (L) Flow chart of spontaneous lung cancer mouse model with KRAS^G12C mutation. Created in BioRender. Bian, Y. (2025) https://BioRender.com/3p0ref1. M Survival analysis of spontaneous lung cancer mice. N CT scan images of lung cancer in mice models treated with vehicle, spermine, sotorasib, or combination (prior group and post-treatment group). Data were analyzed by log-rank test. Source data are provided as a Source Data file.

Fig. 7. Targeting SAT1-mediated polyamine metabolism and ferroptosis to enhance the efficacy of targeted therapy in KRAS-mutant cancer with differential KEAP1 status.

A Mechanism diagram for KRAS inhibitors regulate SAT1 expression through JNK/c-Jun or JNK/NRF2 pathway and mediated polyamine and ferroptosis in KRAS-mutant cancer with differential KEAP1 status. Created in BioRender. Bian, Y. (2025) https://BioRender.com/5ji2693.