High-dose ascorbic acid selectively induces pyroptosis in LKB1-deficient lung cancer and sensitizes immunotherapy

Abstract

Liver kinase B1 (LKB1)-deficient non-small cell lung cancers (NSCLCs) exhibit primary resistance to immune checkpoint inhibitors (ICIs). The redox imbalance inherent in these tumors may represent a potential therapeutic vulnerability. High-dose ascorbic acid (AA) could induce cell redox imbalance. Here, we uncover that LKB1 deficiency upregulates the transporter GLUT1, which enables the accumulation of AA, thereby exacerbating redox imbalance in NSCLC cells. This triggers pyroptosis in LKB1-deficient NSCLC cells via the H₂O₂/reactive oxygen species (ROS)-caspase-3-gasdermin-E (GSDME) axis. In pre-clinical models, high-dose AA reverses ICI resistance and remodels the immune microenvironment, characterized by T cell factor 1 (TCF1)⁺CD8⁺ T cell (progenitor-exhausted CD8⁺ T cell [Tpex]) infiltration. Pyroptosis-driven immunogenic cell death (ICD) promotes cross-presenting dendritic cell (DC) maturation, which drives Tpex proliferation. Crucially, in Batf3^-/- mice lacking functional CD103⁺ DC populations, both Tpex expansion and therapeutic benefits are abrogated, confirming DC dependence. In addition, GSDME is validated as a gatekeeper of pyroptosis-driven antitumor immunity. This work provides a rationale for clinical trials combining ICI with high-dose AA.

Keywords: GSDME; LKB1-deficient lung cancer; Tpex; caspase-3; dendritic cells; high-dose ascorbic acid; immunotherapy resistance; pyroptosis; reactive oxygen species.

Graphical abstract

Figure 1

LKB1 deficiency in NSCLC elevates GLUT1-mediated AA uptake and exacerbates redox imbalance (A and B) Intracellular reactive oxygen species (ROS) levels in A549 (A) and KP (B) cell lines with LKB1 deficiency or proficiency after AA treatment or not. Left: optical density (OD) quantification by microplate reader (A549: n = 6 per group; KP: n = 5 per group). Right: fluorescence microscope images. Scale bars: 50 μm. (C) Quantitative reverse-transcription PCR (RT-qPCR) of relative GLUT1 expression in A549, H1944, LLC1, and KP cells with LKB1 deficiency or proficiency. (D) GLUT1 expressions in A549, LLC1, and KP cells with LKB1 deficiency or proficiency were analyzed by immunoblot. (E) Heatmap illustrating the GLUT1 expression in murine lung cancer cell lines with or without LKB1 deficiency, based on scaled Gene Expression Omnibus (GEO) RNA sequencing data. (F) IHC staining of lung adenocarcinoma tissue microarray sections with anti-LKB1 and anti-GLUT1 antibodies. Representative images were displayed (left). The IHC-score of GLUT1 in LKB1-high and LKB1-low group was compared (right). Scale bars, 1 mm, 500 μm. Mann-Whitney test was used for statistical analysis. (G) The effect of AA on intracellular ROS levels in A549 cells with GLUT1 knocked down. (H) LC-MS analysis of the intracellular levels of AA in A549 cells with or without LKB1 deficiency. Cells were treated with 10 mM AA. (I) Intracellular levels of AA in A549 and H1944 cells with or without LKB1 was detected by phosphomolybdic acid colorimetry. (J) Schematic illustration showing schedules of high-dose AA treatment in the subcutaneous tumor model of LLC-shLkb1 (left). Levels of AA in subcutaneous tumors were detected by phosphomolybdic acid colorimetry (right). Data depict one representative experiment of three independent experiments. Unpaired t test (C, F, H, I, J), two-way ANOVA (A, B), and one-way ANOVA (G) were used for statistical analysis. Data are shown as mean ± SD. ns: p > 0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Figure 2

High-dose AA selectively induces pyroptosis in LKB1-deficient NSCLC cells via activating the cleavage of GSDME (A) Light microscope images showing morphological changes of LKB1-deficient and LKB1-proficient lung cancer cells after being treated with PBS (vehicle, top) or AA (bottom) for 4 h. Red arrows indicated cells with morphological characteristics of pyroptosis (swelling and bubble-like protrusions appearing on cellular membrane surface). Scale bars: 50 μm. (B) Transmission electron microscopy images showing morphological characteristics of pyroptosis. Red arrows showed a lack of intact cellular membrane. Scale bars: 4 μm (left) and 1 μm (right). (C) Single sample gene set enrichment analysis (ssGSEA) scores of different programmed cell death form signatures in A549 cells treated with PBS or AA (n = 3 per group), shown as a heatmap. (D) PI staining of A549 cells pretreated with Fer-1 or GSK’872 for 18 h, followed by AA for 4 h. Scale bars: 200 μm. (E) Release of ATP in A549 lung cancer cell supernatant after AA treatment or PBS (vehicle) tested by ENLITEN ATP assay system kit (n = 3 per group). (F) Release of IL-1β in A549 and H460 lung cancer cell supernatant after AA treatment or PBS (vehicle) tested by ELISA (n = 3 per group). (G) Relative expression of IL-1β and IL-18 in A549 cell lines after AA treatment or PBS (vehicle) tested by RT-qPCR (n = 3 per group). (H) Release of high-mobility group box-1 protein (HMGB1) in lung cancer cell supernatant after AA treatment or PBS tested by ELISA (n = 3 per group). (I) Release of HMGB1 in lung cancer cell supernatant from the cytoplasm after AA treatment or PBS tested by western blot. (J) Western blot showing the cleavage of gasdermin-E (GSDME) protein after indicated AA treatments in A549 lung cancer cells, with different time (left) and different concentrations (right). (K) Western blot showing the cleavage of GSDME protein after AA treatment or PBS in A549 and H460 lung cancer cells. (L) Western blot showing the cleavage of GSDME protein in lung cancer cells in vivo (left). Quantitative analysis of the gray value (n = 3 per group, right). Data depict one representative experiment of three independent experiments. Unpaired t test (E, F, H, L) and two-way ANOVA (G) were used for statistical analysis. Data are shown as mean ± SD. ns: p > 0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Figure 3

Pyroptosis induced by high-dose AA depends on the H₂O₂/ROS-caspase-3-GSDME pathway (A) Light microscopy showing pyroptotic morphology (red arrows) in A549 and H1944 cells with or without catalase (CAT, 100 μM, 3–5 min) pre-treatment, followed by high-dose AA. Scale bars: 50 μm. (B) Flow cytometry showing the effect of CAT on AA-induced pyroptosis in A549 and LLC1-shLkb1 cells, with quantitative analyses shown in the right (n = 3 per group). (C and D) Intracellular ROS levels in A549 cells after indicated treatments (CAT, 100 μM; NAC, 50 μM; 3–5 min pre-treatment) detected by (C) fluorescence microscopy (scale bars: 100 μm) and (D) microplate assay (n = 6). (E) Relative released ATP in A549 supernatant after treatment tested by ENLITEN ATP assay system (n = 3 per group). (F) Light microscopy of A549 cells treated with caspase inhibitors (Z-DEVD-FMK or Z-VAD-FMK, 25 μM); red arrows indicate pyroptotic cells. Scale bars: 50 μm. (G) Western blotting showing the effect of CAT pre-treatment on the cleavage of caspase-3 and GSDME protein induced by high-dose AA in A549 and LLC1-shLkb1 cell lines. (H) Western blotting showing the effect of caspase inhibitors (caspase-3 inhibitor, Z-DEVD-FMK, 25 μM; pan-caspase inhibitor, Z-VAD-FMK, 25 μM) on the cleavage of caspase-3 and GSDME protein induced by high-dose AA in A549 cell lines. Data depict one representative experiment of three independent experiments. (I) IHC staining of cleaved caspase-3 in mouse tumors after high-dose AA or vehicle treatment (n = 3); representative images (left), OD quantification (right). Scale bars, 200 μm, 500 μm. (J) Levels of extracellular ATP, LDH, and HMGB1 in A549 supernatants with or without CASP3 knockout after AA treatment. ATP/LDH by microplate assay, HMGB1 by ELISA. Data depict one representative experiment of three independent experiments. Mann-Whitney test (I) and one-way ANOVA (B, D, E, J) were used for statistical analysis. Data are shown as mean ± SD. ns: p > 0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Figure 4

High-dose AA restores PD-1 blockade resistance of LKB1 deficiency in NSCLC in vivo (A–F) LLC1-shCtrl (A, B), LLC1-shLkb1 (C, D), and TC1-shLkb1 (E, F) cells were injected subcutaneously followed by different treatments (for control immunoglobulin G [vehicle], αPD-1 Ab, high-dose AA [AA], co-treatment with anti-PD-1 Ab and AA [αPD-1 Ab + AA], n = 7 per group in A–D and n = 5 in E and F). Tumor size and survival in different treatment arms were monitored. Tumor growth curves and survival curves were shown. Two-way ANOVA was performed to analyze the tumor growth curves. Log rank test was used to analyze the survival data. Data are shown as mean ± SE. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, ns: p > 0.05. (G–L) LLC1-shCtrl-luc cells (G–I) or LLC1-shLkb1-luc cells (J–L) were injected into the left chest of mice followed by indicated treatments (n = 5 per group). Tumor formation was detected using a bioluminescence imager every week. Representative bioluminescent images (H, K) and quantification of results (I, L). Two-way ANOVA was performed. ns: p > 0.05, ∗p < 0.05.

Figure 5

High-dose AA remodeled the tumor immune microenvironment characterized by the infiltration of TCF1⁺CD8⁺ T lymphocytes (A) Left: t-distributed stochastic neighbor embedding (t-SNE) plot of 44,745 cells from lung tumor samples treated with αPD-1 Ab (n = 2) and high-dose AA + αPD-1 Ab (n = 2), colored by their 10 major cell types. Right: cell type composition of each cluster. (B) Cluster composition of T cells. (C) Volcano plot displaying the differentially expressed genes between cells within cluster 2 and cells outside cluster 2 (red: upregulated in cluster 2). The x axis represents log-fold changes, and the y axis represents log10 adjusted p values. A two-sided Wilcoxon rank-sum test was used. (D) Expression levels of Tcf7 in T cells treated with αPD-1 Ab or high-dose AA + αPD-1 Ab. ∗∗∗∗p < 0.0001. (E) Enrichment (log2 p values) of progenitor-like gene signature in each cell illustrated in t-SNE plots. (F) Trajectory manifold of T cells annotated by cell type using the Monocle 2 algorithm. (G) RNA velocity analysis of gene expression in T cells from αPD-1 Ab group and αPD-1 Ab+AA group, predicted by Velocyto/ScVelo. (H) Visualization and localization images (left) and quantification (right) of TCF1 (red)-expressing CD8⁺ (green) T cells through immunofluorescence (IF) from lung tumor samples of LLC1-shCtrl and shLkb1 mice with indicated treatments (for control immunoglobulin G [vehicle], anti-PD-1 Ab [αPD-1 Ab], high-dose AA [AA], or co-treatment with anti-PD-1 Ab and AA [αPD-1 Ab + AA], n = 5 for per group). Groups compared using one-way ANOVA. ∗p < 0.05, ∗∗p < 0.01, ns: p > 0.05.

Figure 6

Pyroptosis induces immunogenic cell death and enhances cross-presenting DC maturation, thereby driving Tpex expansion (A) Left: schematic of the vaccination protocol. Right: tumor incidence in vaccinated versus control mice (n = 7 per group). (B) Tumor growth in vaccine (n = 4) vs. vehicle (n = 6) groups. (C and D) The percentages of mature DCs (CD80⁺CD86⁺), CD103⁺ DCs among total CD45⁺CD11c⁺ cells (C), and TCF1⁺ TIM3⁻ T cells among total CD45⁺CD3⁺CD8⁺PD1⁺ cells (D) in tumor-draining lymph nodes isolated from TC1-shLkb1 mice models with indicated treatments (n = 5 per group). (E) C57BL/6J background Batf3^−/− mice (n = 4 per group) and Batf3^+/+ (wild-type [WT]) mice (n = 5 per group) were injected with TC1-shLkb1 tumor cells. Tumor size was monitored. Two-way ANOVA was performed. Data are shown as mean ± SE. (F) The percentages of TCF1⁺ TIM3⁻ T cells among total CD45⁺CD3⁺CD8⁺PD1⁺ cells in tumor DLNs. (G) Visualization and localization images (left) and quantification (right) of TCF1⁺ (red) CD8⁺ (green) T cells through immunofluorescence (IF) in tumors from Batf3^−/− mice (n = 4 per group) and WT mice (n = 5 per group). One-way ANOVA (C, D, F, G) and two-way ANOVA (B, E) were performed. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, ns: p > 0.05.

Figure 7

Inner blockage of GSDME reversed the synergistic effect of high-dose AA and anti-PD-1 therapy (A) LLC1-shLkb1-sgctrl or LLC1-shLkb1-sgGsdme cells were injected subcutaneously, followed by indicated treatments (vehicle: control immunoglobulin G, Comb: αPD-1 Ab + AA). Tumor growth curves and tumor appearance were shown (n = 5 per group). Two-way ANOVA was performed to analyze the tumor growth curves, ∗p < 0.05, ∗∗∗p < 0.001, ns: p > 0.05. (B) LLC1-shLkb1-sgctrl or LLC1-shLkb1-sgGsdme cells were injected into the left chest of mice followed by co-treatment with αPD-1 Ab and AA. Tumor formation was detected using micro-CT scan. (C) Waterfall plot showing tumor volume response to the treatment (related to B). Each column represents one mouse. One-way ANOVA was performed. ∗p < 0.05. (D) Left: Uniform manifold approximation and projection plot of 16, 217 cells from control (n = 2) and sgGsdme (n = 2) tumors treated with the combination of high-dose AA and αPD-1 Ab. Eight cell types were identified. Right: percentages of cells from each cluster in each sample. (E) Violin plot of pyroptosis score in tumor cells from the control group and the sgGsdme group. ∗∗∗∗p < 0.0001. (F) The expression level of Hmgb1 in tumor cells from the control group and the sgGsdme group detected from single-cell RNA sequencing and visualized through a feature plot. (G) Expression level of Tcf7 in T cells from lung tumor samples of the control group and the sgGsdme group after αPD-1 Ab + AA treatment. One-way ANOVA was performed. ∗∗∗∗p < 0.0001. (H) mIHC assay of LLC1-shLkb1-sgctrl (left) or LLC1-shLkb1-sgGsdme (right) orthotopic mouse models with αPD-1 Ab and AA treatment. The representative image with CD8 (green), TCF1 (purple), F4/80 (yellow), SPP1 (red), and DAPI (blue). n = 5 per group. Scale bars, 40 μm. (I) Quantitative analysis of the proportion of TCF1⁺ CD8⁺ T among total CD8⁺ T and SPP1⁺ macrophages in TAMs (n = 5 per group). Groups compared using one-way ANOVA. ∗∗p < 0.01, ∗∗∗p < 0.001.