Glucose-6-phosphate dehydrogenase maintains redox homeostasis and biosynthesis in LKB1-deficient KRAS-driven lung cancer

Abstract

Cancer cells depend on nicotinamide adenine dinucleotide phosphate (NADPH) to combat oxidative stress and support reductive biosynthesis. One major NADPH production route is the oxidative pentose phosphate pathway (committed step: glucose-6-phosphate dehydrogenase, G6PD). Alternatives exist and can compensate in some tumors. Here, using genetically-engineered lung cancer mouse models, we show that G6PD ablation significantly suppresses Kras^G12D/+;Lkb1^-/- (KL) but not Kras^G12D/+;P53^-/- (KP) lung tumorigenesis. In vivo isotope tracing and metabolomics reveal that G6PD ablation significantly impairs NADPH generation, redox balance, and de novo lipogenesis in KL but not KP lung tumors. Mechanistically, in KL tumors, G6PD ablation activates p53, suppressing tumor growth. As tumors progress, G6PD-deficient KL tumors increase an alternative NADPH source from serine-driven one carbon metabolism, rendering associated tumor-derived cell lines sensitive to serine/glycine depletion. Thus, oncogenic driver mutations determine lung cancer dependence on G6PD, whose targeting is a potential therapeutic strategy for tumors harboring KRAS and LKB1 co-mutations.

Fig. 1. G6PD promotes KL lung tumorigenesis but is dispensable for KP lung tumorigenesis.

a Scheme illustrating the induction of conditional tumoral G6pd knockout to investigate the role of G6PD in KP and KL lung tumorigenesis (Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license). b Representative immunohistochemistry (IHC) images of G6PD in KP and KL lung tumors at 12 weeks post-induction. n = 10 images for each genotype, scale bar = 200 μm. c. Representative gross lung pathology from mice bearing G6pd^WT;KP (n = 12 mice) and G6pd^KO;KP (n = 6 mice) lung tumors at 12 weeks post-tumor induction. Scale bar =1 cm. d Graph of wet lung weight from (c). e Representative H&E staining of scanned lung sections from (c). f, g Quantification of tumor number (f) and tumor burden (g) from (e). n is same with (c). h Representative IHC images and quantification of Ki67 in G6pd^WT;KP (n = 39 images) and G6pd^KO;KP (n = 16 images) lung tumors. Scale bar = 100 μm. i Kaplan-Meier survival curve of mice bearing G6pd^WT;KP (n = 34 mice) or G6pd^KO;KP (n = 22 mice) lung tumors. j Representative gross lung pathology from mice bearing G6pd^WT;KL (n = 8 mice) and G6pd^KO;KL (n = 8 mice) lung tumors at 12 weeks post-tumor induction. Scale bar = 1 cm. k Graph of wet lung weight from (j). l Representative H&E staining of scanned lung sections from (j). m, n Quantification of tumor number (m) and tumor burden (n) from (l). n is same with (j). o Representative IHC images and quantification of Ki67 (n = 31 images for G6pd^WT;KL, n = 10 images for G6pd^KO;KL), pERK (n = 24 images for G6pd^WT;KL, n = 12 images for G6pd^KO;KL), and pS6 (n = 18 images for G6pd^WT;KL, n = 15 images for G6pd^KO;KL) in KL lung tumors at 12 weeks post-tumor induction. Scale bar = 100 μm. p Kaplan-Meier survival curve of mice bearing G6pd^WT;KL (n = 28 mice) and G6pd^KO;KL (n = 13 mice) lung tumors. Data are presented as mean ± SEM, significance was calculated by Mann Whitney test (d, f, g, h, k, pS6 in o), two-tailed unpaired t-test (m), two-tailed unpaired t-test with Welch’s correction (n, Ki67 and pERK in o) or log-rank test (i, p). Source data are provided as a Source Data file.

Fig. 2. G6PD is required to maintain cellular redox homeostasis in KL lung tumors.

a Pool size of NADPH and NADP⁺, and NADPH/NADP⁺ ratio in G6pd^WT;KL (n = 8 mice) and G6pd^KO;KL (n = 7 mice) lung tumors at 12 weeks post-tumor induction. b Pool size of GSH and GSSG, and GSH/GSSG ratio in G6pd^WT;KL (n = 8 mice) and G6pd^KO;KL (n = 7 mice) lung tumors at 12 weeks post-tumor induction. c Gene Set Enrichment Analysis (GSEA) of oxidative stress signaling for G6pd^KO;KL (n = 7 mice) and G6pd^WT;KL (n = 8 mice) lung tumors at 12 weeks post-tumor induction based on bulk-tumor mRNA-seq data. Gene set for “Oxidative stress” was downloaded from GeneCards (https://www.genecards.org/, accessed on April 09, 2023). d Representative IHC images and quantification of NRF2 and NQO1 in G6pd^WT;KL and G6pd^KO;KL lung tumors at 12 weeks post-tumor induction. n = 10 images for each quantification, scale bar = 100  μm. e Representative IHC images and quantification of 8-oxo-dG (n = 21 images for G6pd^WT;KL, n = 31 images for G6pd^KO;KL) and γ-H2AX (n = 32 images for both G6pd^WT;KL and G6pd^KO;KL) in G6pd^WT;KL and G6pd^KO;KL lung tumors at 12 weeks post-tumor induction. Scale bar = 10 μm. Data are presented as mean ± SEM, significance was calculated by two-tailed unpaired t-test (NADPH and NADPH/NADP⁺ in a, b), Mann-Whitney test (8-oxo-dG in e), two-tailed unpaired t-test with Welch’s correction (NADP⁺ in a, d, γ-H2AX in e). Source data are provided as a Source Data file.

Fig. 3. The maintenance of redox homeostasis by G6PD supports the survival of KL tumor cells under oxidative stress conditions.

a Scheme illustrating KL tumor-derived cell lines (TDCLs) generation (Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license). b Western blot of G6pd^WT and G6pd^KO KL TDCLs. Uncropped Western blot image is shown in Source Data file. c Pool size of NADPH and NADP⁺, and NADPH/NADP⁺ ratio of KL TDCLs in nutrient rich conditions (complete RPMI medium). n = 4 replicates for each clone. d Pool size of GSH and GSSG, and GSH/GSSG ratio of KL TDCLs in nutrient rich conditions. n = 4 replicates for each clone. e Basal ROS level of KL TDCLs in nutrient rich conditions. n = 3 replicates for each clone. f Proliferation rate of KL TDCLs in nutrient rich conditions measured by Incucyte for 4 days (left) with statistical analysis at day 4 (right). n = 3 replicates for each clone. g Relative proliferation rate of KL TDCLs treated with G6PDi-1 for 48 hours. n = 6 replicates for each clone at different G6PDi-1 concentrations, except 2489-2 at 40 μmol/L G6PDi-1 with n = 5 replicates. h Relative proliferation rate of KL TDCLs treated with H₂O₂ for 24 hours. For each clone, n = 6 replicates at 0 μmol/L H₂O₂, n = 3 replicates at 20, 40, 80 μmol/L H₂O₂. i ROS levels of KL TDCLs treated with 20 μmol/L H₂O₂ for 24 hours. For each clone, n = 3 replicates at 0 and  20 μmol/L H₂O₂. j Growth curve of KL allograft tumors from mice treated with or without high-dose Vitamin C (Vit C). n = 10 allograft tumors for each group. k Gross pathology of allograft tumors from (j). Scale bar = 1 cm. l Graph of allograft tumor weight from (k). m Representative IHC images and quantification of NRF2 (n = 15 images for each quantification), NQO1 (n = 10 images for each quantification), and 8-oxo-dG (n = 15 images for each quantification) of allograft tumors from (k). Scale bar = 100 μm. Data are presented as mean ± SEM, significance was calculated by two-tailed unpaired t-test (NADPH and NADP⁺ in c, GSSG and GSH/GSSG in d), two-tailed unpaired t-test with Welch’s correction (NADPH/NADP⁺ in c, GSH in d), one-way ANOVA followed by Bonferroni’s multiple comparisons test (e, f, g, h, i, m), or one-way ANOVA followed by t-test (l). Source data are provided as a Source Data file.

Fig. 4. G6PD suppresses p53 activation for KL lung tumorigenesis.

a GSEA (top) and heatmap of relative expression of genes (bottom) contributing to predicting positive regulation of intrinsic apoptotic signaling pathway by p53 class mediator for G6pd^KO;KL (n = 7 mice) and G6pd^WT;KL (n = 8 mice) lung tumors at 12 weeks post-tumor induction based on bulk-tumor mRNA-seq data. b Representative IHC images and quantification of p53 (n = 27 images for G6pd^WT;KL, n = 33 images for G6pd^KO;KL), p21 (n = 30 images for G6pd^WT;KL, n = 33 images for G6pd^KO;KL) and cleaved caspase-3 (n = 30 images for both G6pd^WT;KL and G6pd^KO;KL) of G6pd^WT;KL and G6pd^KO;KL lung tumors at 12 weeks post-tumor induction. Scale bar = 10 μm. c Scheme to induce conditional tumoral G6pd knockout to study the role of G6PD in KPL lung tumorigenesis (Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license). d Representative IHC images of G6PD in G6pd^WT;KPL and G6pd^KO;KPL lung tumors at 6 weeks post-tumor induction. n = 10 images for each genotype, scale bar = 200 μm. e Representative gross lung pathology from mice bearing G6pd^WT;KPL (n = 4 mice) and G6pd^KO;KPL (n = 4 mice) lung tumors at 6 weeks post-tumor induction. Scale bar = 1 cm. f Graph of wet lung weight from (e). g Representative H&E staining of scanned lung sections from (e). h, i Quantification of tumor number (h) and tumor burden (i) from (g). n is same with (e). j Representative IHC images and quantification of Ki67 in G6pd^WT;KPL and G6pd^KO;KPL lung tumors. n = 20 images for each quantification, scale bar = 100 μm. k Kaplan-Meier survival curve of mice bearing G6pd^WT;KPL (n = 26 mice) and G6pd^KO;KPL (n = 27 mice) lung tumors. l Representative IHC images and quantification of NRF2 (n = 15 images for G6pd^WT;KPL, n = 13 images for G6pd^KO;KPL) and NQO1 (n = 15 images for both G6pd^WT;KPL and G6pd^KO;KPL) in G6pd^WT;KPL and G6pd^KO;KPL lung tumors. Scale bar = 100 μm. Data are presented as mean ± SEM, significance was calculated by two-tailed unpaired t-test (f, h, i, NQO1 in l), two-tailed unpaired t-test with Welch’s correction (b, j, NRF2 in l), or log-rank test (k). Source data are provided as a Source Data file.

Fig. 5. G6PD depletion impairs KL lung tumor lipid metabolism.

a, b GSEA of lipid (a) and fatty acids (b) biosynthetic process for G6pd^KO;KL (n = 7 mice) and G6pd^WT;KL (n = 8 mice) lung tumors at 12 weeks post-tumor induction based on bulk-tumor mRNA-seq data. c Representative IHC images and quantification of pAMPK and pACC in G6pd^WT;K (Kras^G12D/+), G6pd^WT;KL and G6pd^KO;KL lung tumors at 12 weeks post-tumor induction. n = 10 images for each quantification. Scale bar = 100 μm. d Scheme of in vivo D₂O infusion to examine tumor de novo fatty acid synthesis (Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license). e, f C16:0 (e) and C18:2 (f) deuterium (²H) labeling fraction in G6pd^WT;KL (n = 9 mice) and G6pd^KO;KL (n = 8 mice) lung tumors at 12 weeks post-tumor induction. g C16:0 pool size of G6pd^WT;KL (n = 9 mice) and G6pd^KO;KL (n = 8 mice) lung tumors at 12 weeks post-tumor induction. h, i Principal Component Analysis (PCA) (h) and Heatmap (i) of saponified fatty acids pool size of G6pd^WT;KL and G6pd^KO;KL lung tumors (n = 5 mice for G6pd^WT;KL, n = 4 mice for G6pd^KO;KL) and serum (n = 5 mice for G6pd^WT;KL, n = 3 mice for G6pd^KO;KL) in fasted state at 12 weeks post-tumor induction. j Scheme to examine the impact of high-fat diet (HFD) on KL lung tumorigenesis (Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license). k Representative gross lung pathology from mice bearing G6pd^WT;KL (n = 4 mice for 7 weeks in normal diet (ND), n = 5 mice for 7 weeks in HFD, n = 8 mice for 11 weeks in ND, n = 6 mice for 11 weeks in HFD) and G6pd^KO;KL (n = 4 mice for 7 weeks in ND, n = 4 mice for 7 weeks in HFD, n = 11 mice for 11 weeks in ND, n = 10 mice for 11 weeks in HFD) lung tumors fed with ND or HFD. Scale bar = 1 cm. l Graph of wet lung weight from (k). m Representative H&E staining of scanned lung sections from (k). n, o Quantification of tumor number (n) and tumor burden (o) from (m). n is same with (k). p Representative IHC images and quantification of Ki67 in G6pd^WT;KL (n = 17 images for ND, n = 29 images for HFD) and G6pd^KO;KL (n = 17 images for ND, n = 21 images for HFD) lung tumors at 11 weeks post-tumor induction. Scale bar = 100 μm. Data are presented as mean ± SEM, significance was calculated by two-tailed unpaired t-test (g, i), one-way ANOVA followed by Bonferroni’s multiple comparisons test (c, p), two-way ANOVA followed by t-test (l, n, o). Source data are provided as a Source Data file.

Fig. 6. G6PD ablation has no impact on KL lung tumor TCA cycle metabolism.

a Scheme of carbon contribution from glucose to glycolytic intermediates, TCA cycle intermediates, and serine (top) and in vivo [U-¹³C₆]-glucose tracing in the mice bearing G6pd^WT;KL and G6pd^KO;KL lung tumors (bottom). Schematic images are created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license. b Normalized ¹³C labeling fraction of glycolytic intermediates from glucose of G6pd^WT;KL (n = 6 mice) and G6pd^KO;KL (n = 6 mice) lung tumors in fasted state (food was removed from the mice at approximately 9:00 AM, and mice were euthanized with samples collected at 5:00 PM) at 12 weeks post-tumor induction. Glucose 6-phosphate (G6P), glyceraldehyde 3-phosphate (G3P), 3-phosphoglycerate (3-PG). c Normalized ¹³C labeling fraction of TCA cycle intermediates from glucose of tumors same with (b). α-ketoglutarate (α-KG). d Pool size of glycolytic intermediates of tumors same with (b). Data are presented as mean ± SEM, significance was calculated by two-tailed unpaired t-test (d), or two-way ANOVA followed by Bonferroni’s multiple comparisons test (b, c). Source data are provided as a Source Data file.

Fig. 7. G6PD ablation reprograms KL tumor serine metabolism.

a Normalized ¹³C labeling fraction from glucose to serine and glycine of KL lung tumors. b Pool size of serine and glycine of KL lung tumors. c Serine consumption of G6pd^WT;KL and G6pd^KO;KL TDCLs in nutrient rich conditions. d ²H labeling fraction of serine in G6pd^WT;KL and G6pd^KO;KL TDCLs after 4 hours [2,3,3-²H]-serine labeling in nutrient rich conditions. n = 6 replicates for each clone. e Scheme of hydrogen contribution from [2,3,3-²H]-serine to NADPH (Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license). f Scheme of in vivo [2,3,3-²H]-serine tracing (Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license). g ²H labeling fraction of serine in G6pd^WT;KL (n = 8 mice) and G6pd^KO;KL (n = 7 mice) lung tumors at 12 weeks post-tumor induction. h NADPH active-H labeling from [2,3,3-²H]-serine in G6pd^WT;KL (n = 8 mice) and G6pd^KO;KL (n = 7 mice) lung tumors at 12 weeks post-tumor induction. i NADPH/NADP⁺ ratio of G6pd^WT;KL and G6pd^KO;KL TDCLs cultured with RPMI medium with or without serine and glycine for 24 hours. RPMI denotes cells cultured in complete RPMI medium, RPMI w/o GS denotes cells cultured in complete RPMI medium without serine and glycine. n = 3 replicates for each clone under different conditions. j Relative GSH/GSSG ratio of same samples from (i). k ROS levels of G6pd^WT;KL and G6pd^KO;KL TDCLs cultured with RPMI medium with or without serine and glycine for 48 hours. n = 6 replicates for each clone under different conditions, except 2489-2 with n = 4 replicates. l Relative proliferation rate of G6pd^WT;KL and G6pd^KO;KL TDCLs cultured with RPMI medium with or without serine and glycine for 48 hours. n = 3 replicates for each clone under different conditions. m Model of G6PD-mediated KL lung tumor growth (Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license). Data are presented as mean ± SEM, significance was calculated by two-tailed unpaired t-test (a, b, d, g, h), or two-way ANOVA followed by Bonferroni’s multiple comparisons test (i, j, k, l). Source data are provided as a Source Data file.