SCD1 Inhibition Blocks the AKT-NRF2-SLC7A11 Pathway to Induce Lipid Metabolism Remodeling and Ferroptosis Priming in Lung Adenocarcinoma

Abstract

Concurrent inactivating mutations in STK11 and KEAP1 drive primary resistance to therapies, leading to worse outcomes in KRAS-mutated lung adenocarcinoma (KRASmut LUAD), and are associated with metabolic alterations. Elucidation of the underlying biology of this aggressive LUAD subset is needed to develop effective treatments to improve patient outcomes. Our transcriptomic analysis of 5,498 "real-world" KRASmut LUADs demonstrated that STK11/KEAP1 co-mutation led to upregulation of fatty acid and redox signaling pathways and considerable enrichment of the metabolic genes SCD1 and SLC7A11. High expression of SCD1 and SLC7A11 predicted poor prognosis in KRASmut patients. Transcriptomics, lipidomics, and kinase arrays in preclinical models demonstrated that SCD1 inhibition promoted ferroptosis, altered fatty acid metabolism, and downregulated SLC7A11 via AKT-GSK3β-NRF2 signaling. SCD1 inhibition caused appreciable tumor regression in xenografts and augmented the efficacy of the ferroptosis inducer erastin. Overall, this study provides insights into the role of the SCD1-SLC7A11 axis in regulating metabolic programming and predicting poor patient outcomes in a genetically defined subset of KRASmut LUAD.

Significance: SCD1 and SLC7A11 are prognostic biomarkers and therapeutic targets for KRAS/STK11/KEAP1 co-mutant lung adenocarcinoma, which will refocus mechanistic studies and lead to treatment strategies for lung cancer.

Figure 1:. Characterization of “real-world” lung cancer cases demonstrates high expression of SCD1 and SLC7A11 in KRAS/STK11/KEAP1 co-mutant LUADs.

(A) Venn-diagram representing clinical dataset of real-world LUAD cases (Caris data). The cohort consisted of 5498 KRAS mutant LUAD tumors submitted for genomic and transcriptomic profiling. Of these patients, 3835 were KRAS^mut/STK11^wt/KEAP1^wt, 691 were KRAS^mut/STK11^mut/KEAP1^wt, 369 were KRAS^mut/STK11^wt/KEAP1^mut 603 were KRAS^mut/STK11^mut/KEAP1^mut. B) Expression of SCD1 in all four cohorts and C) Expression of SLC7A11 in all four cohorts, D) Expression of SCD1 in normal (N) vs LUAD (T) from publicly available data set from TNM plot analysis (p>0.0001) E) Expression of SLC7A11 in normal (N) vs LUAD (T) from publicly available data set (p>0.0001), F) Co- expression analysis of SCD1 and SLC7A11 in real-world LUAD cases, R value is 0.0.35; p>0.0001), G and H representing the most enriched pathway in co-mutant population, p<0.0001 I) and J) are the Kaplan-Meier plot for both SCD1 and SLC7A11 from Caris LUAD data set LUAD data set, p <0.0001.

Figure 2:. SCD1 inhibition augments the cytotoxic response of SLC7A11 inhibitor, erastin, in KRAS/STK11/KEAP1 co-mutant LUADs.

A) Bar graph representing the proliferation of H358 cells till 72H, +-/ Fer-1 treatment. B) and C) IC50 of FIN (erastin) on NSCLC lines with different mutation status D) and E) Loewe synergy score in NTC and DKO cells for Erastin and CVT-11127 drug combination. F-G) Flow-cytometry analysis of BODIPY C11 staining of H358 and H2122 cells, up on CVT-11127, Erastin, Fer-1 (10μM). H) bar graph representation of TBARS assay for the measurement of cellular MDA levels up on CVT-11127 treatment in H358 NTC, DKO and H2122 cells. I) and J) Cellular GSH measurement in NTC and DKO cells after SCD1 inhibition through CVT-11127. K) The bar graph represents the outcome from the Cysteine depletion experiment on NTC, DKO and DE novo co-mutant H2122 cells. L) Crystal violet staining image of H358 NTC and DKO cells in Cystine deprived) condition followed by the bar graph representation for quantification. For comparisons among multiple groups, one-way analysis of variance (ANOVA) and between two groups, student t-test followed by the Dunnett’s post-hoc test were performed. The p<0.05 (*), p<0.01 (**), p<0.001 (***) were considered as significant (p>0.05 was considered as non-significant (ns)). Error bar represents the ±standard error of the mean (SEM) and the results are representation of n= either 2 or 3 based up on the experiments

Figure 3:. SCD1 inhibition in KRAS/STK11/KEAP1 resulted in increased free fatty acid accumulation, leading to lipid peroxidation-mediated cell death.

A) NTC and B) DKO, volcano plots show differentially regulated lipid species in both cell lines after SCD1 is depleted (SCD1 KO) (p-value was adjusted). C) Venn diagram shows the overall upregulated lipid species and uniquely upregulated in DKO +/− SCD1 KO vs DKO. D) Heatmap for uniquely upregulated lipid species in DKO +/− SCD1 KO vs DKO. E-H) Box plot for important lipid species including saturated fatty acids (SFAs) with no double bond and poly unsaturated fatty acids (PUFAs) with two or more double bonds in NTC vs DKO cells +/− SCD1 KO. I) Summary or schematic representation of the events taking place before and after SCD1 KO in STK11/KEAP1 KO cells, D) For comparisons among multiple groups, one-way analysis of variance (ANOVA) and between two groups, student t-test followed by the Dunnett’s post-hoc test were performed. The p<0.05 (*), p<0.01 (**), p<0.001 (***) were considered as significant (p>0.05 was considered as non-significant (ns). The results are representation of n= 3 experiments. Error bar represents the ±standard error of the mean (SEM)

Figure 4:. SCD1 inhibition modulates the expression of metabolic pathways and leads to suppression of salt-inducible kinase genes in KRAS/STK11/KEAP1 co-mutant LUADs.

A) Venn diagram showing the differentially regulated genes in different NSCLCs with different de novo mutation status. Also showing the genes which are uniquely downregulated in KRAS^mutSTK11^mutKEAP1^mut H2122 cell line up on CVT-11127 treatment. B) Most notable downregulated pathway analysis. C) Heat map showing the most important downregulated genes in STK11/KEAP1 co mutation cell line H2122, after CVT-11127 treatment (10μM, 96 hours). D) Box plots depict the gene expression of SIK1 gene in H358 vs H2122 in +/− CVT-11127 condition (10μM, 96 hours). E) Box plots depict the gene expression of SIK1B gene in H358 vs H2122 in +/− CVT-11127 condition (10μM, 96 hours). F) Bar graph showing the mRNA expression levels SIK1, CREB, and SCD1 in H358 cells after +/− CVT-11127 treatment G) Bar graph showing the mRNA expression levels SIK1, CREB, and SCD1 in H2122 cells after +/− CVT-11127 treatment. H) mRNA expression of SIK1 gene in STK11/KEAP1 CRISPR KO H358 cell line. I) mRNA expression of SIK1 before and after CVT-11127 treatment for 24 to 96 hours of treatment (10μM). For comparisons among multiple groups, one-way analysis of variance (ANOVA) and between two groups, student t-test followed by the Dunnett’s post-hoc test were performed. The p<0.05 (*), p<0.01 (**), p<0.001 (***) were considered as significant (p>0.05 was considered as non-significant (ns)). The results are representation of n= 3 experiments. Error bar represents the ±standard error of the mean (SEM).

Figure 5:. SCD1 inhibitor modulates SLC7A11 expression via AKT/GSK3β/NRF2 pathway.

A) Differentially regulated kinases in NTC and DKO cell after CVT-11127 treatment (10μM, 96 hours). B) The bar graph represents the densitometry readouts of the Phospho proteome (kinase) assay. C) H358 NTC and DKO cells were treated with CVT-11127 (10μM, 96 hours). A representative western blot image is shown with indicated proteins. D) DKO cells were transfected with siSCD. A representative western blot image is shown with indicated proteins E) H358 NTC and DKO cells were treated with LY2090314 (GSK3β Inhibitor) for 24 hours and 30nM, a representative western blot image is shown with indicated proteins. F) Western blot for the major signaling changes in H358 NTC, DKO and H2122 cells due to CVT-11127 treatment for 24, 48, 72, 96 hours. G) Bar graph represents the lipid peroxidation event in the cells by BODPY/C11 staining after H358 NTC and DKO cells were treated with LY2090314 (GSK3β inhibitor) and SCD1 inhibitor CVT-11127. G-H) Crystal violet staining of NTC and DKO cells treated with either SCD1 inhibitor CVT-11127 or GSK3β inhibitor LY2090314. Bar graph represent the quantification. For comparisons among multiple groups, one-way analysis of variance (ANOVA) and between two groups, student t-test followed by the Dunnett’s post-hoc test were performed. The p<0.05 (*), p<0.01 (**), p<0.001 (***) were considered as significant (p>0.05 was considered as non-significant (ns)). Error bar represents the ±standard error of the mean (SEM) and the results are representation of n= either 2 or 3 based on the experiments. Error bar represents the ±standard error of the mean (SEM)

Figure 6:. SCD1 inhibition decreases tumor growth and augments the anti-tumor response of IKE in isogenic co-mutant models.

A) Schematic diagram showing the tumor development and treatment regimes in athymic nude mice. Vehicle (black) 5% DMSO+ 40% PEG300 + 5% Tween80 + 50% saline; A939572 (orange) (50mg/kg, P.O, QDx5), IKE (blue) (40mg/kg IP, every other day) and combination group (red). B) tumor regression curve in H358 NTC model treated with Vehicle, A939572, IKE and combination treatment C-F) individual tumor volume in different treatment groups. G) and H) liver function test for H358 NTC tumor xenograft. I) tumor regression curve for H358 DKO model treated with Vehicle A939572, IKE and combination treatment. J-M) change in tumor volume in each mouse. N) and O) liver function data for H358 NTC tumor xenograft. p-values were calculated using linear mixed-effects regression models (ns > 0.05, *p<0.05, **p<0.01, ***p<0.001). Error bar represents the ±standard error of the mean (SEM).

Figure-7. SCD1 inhibition decreases tumor growth and augments the anti-tumor response of IKE in de novo STK11/KEAP1 co-mutated LUAD model.

Tumor regression curve for (A) A549-GFP-Dox cells, (B) A549- LKB1-Dox cells, (C) A549-KEAP1-Dox cells, (D) H460 triple mutant cells treated with Vehicle, A939572, IKE and combination treatment. E and F) are the liver function test in H460 tumor xenograft treated with IKE with or without A939572. G) bodyweight data for H460 model. p-values were calculated using linear mixed-effects regression models (ns > 0.05, *p<0.05, **p<0.01, ***p<0.001).