KRASG12D-driven pentose phosphate pathway remodeling imparts a targetable vulnerability synergizing with MRTX1133 for durable remissions in PDAC

Abstract

The KRAS^G12D inhibitor MRTX1133 shows the potential to revolutionize the treatment paradigm for pancreatic ductal adenocarcinoma (PDAC), yet presents challenges. Our findings indicate that KRAS^G12D remodels a pentose phosphate pathway (PPP)-dominant central carbon metabolism pattern, facilitating malignant progression and resistance to MRTX1133 in PDAC. Mechanistically, KRAS^G12D drives excessive degradation of p53 and glucose-6-phosphate dehydrogenase (G6PD)-mediated PPP reprogramming through retinoblastoma (Rb)/E2F1/p53 axis-regulated feedback loops that amplify ubiquitin-conjugating enzyme E2T (UBE2T) transcription. Genetic ablation or pharmacological inhibition of UBE2T significantly suppresses PDAC progression and potentiates MRTX1133 efficacy. Leveraging structure advantages of the UBE2T inhibitor pentagalloylglucose (PGG), we develop a self-assembling nano co-delivery system with F-127, PGG, and MRTX1133. This system enhances the efficacy of PGG and MRTX1133, achieving durable remissions (85% overall response rate) and long-term survival (100% progression-free survival) in patient-derived xenografts and spontaneous PDAC mice. This study reveals the role of KRAS^G12D-preferred PPP reprogramming in MRTX1133 resistance and proposes a potentially therapeutic strategy for KRAS^G12D-mutated PDAC.

Keywords: KRAS(G12D); MRTX1133; metabolic reprogramming; pancreatic ductal adenocarcinoma; pentose phosphate pathway.

Graphical abstract

Figure 1

KRAS^G12D drives a PPP-dominant central carbon metabolism pattern in PDAC (A) Kaplan-Meier analysis with log-rank test showing OS for patients stratified by KRAS^WT and KRAS^G12D from TCGA database. (B) KEGG analysis for metabolic pathway using DEGs from patients with KRAS^WT or KRAS^G12D mutation in TCGA database. (C) t-distributed stochastic neighbor embedding (t-SNE) visualization of spatial metabolomics data from pancreatic tissues of KC mouse. (D) Gene set enrichment analysis (GSEA) of central carbon metabolism in normal pancreas and lesion based on spatial metabolomics data. (E and F) H&E staining and mass spectrometry imaging (MSI) of glucose-6-phosphate (G-6-P), ribose 5-phosphate (R-5-P), sedoheptulose 7-phosphate (S-7-P), 3-phosphoglyceric acid (3-PGA), and α-ketoglutaric acid (α-KGA), followed by statistical analysis (n = 3). (G) Metabolic pathway analysis of differential metabolites from targeted metabolomics on central carbon metabolism in KRAS^WT and KRAS^G12D PDO-1. (H) Heatmap displaying the indicated metabolite level from U-¹³C₆-labeled metabolic flux analysis in KRAS^WT and KRAS^G12D PDO-1 (n = 3). Metabolite levels are represented by different sizes and colors of the indicated values. G-1-P, glucose 1-phosphate; F-1-P, fructose 1-phosphate; GAP, glyceraldehyde 3-phosphate; DHAP, dihydroxyacetone phosphate; FBP, fructose 1,6-bisphosphate; PEP, phosphoenolpyruvic acid; Ru-5-P, ribulose-5-phosphate; E-4-P, erythrose-4-phosphate. See Table S1. (I) Schematic example of U-¹³C₆-labeled glucose metabolism in the glycolysis, PPP, and TCA cycle. Number represents the fold change of metabolites in KRAS^G12D compared to KRAS^WT PDO-1. (J) Ratio of lactate level (M1/M1+M2) from U-¹³C_1,2-labeled metabolic flux analysis in WT and G12D-mutant PDO-1 (n = 3). (K) Sensitivity to MRTX1133 in PDO-2, 3, 4, 5, and 6 (n = 6). (L) Linear regression analysis shows the correlation of G6PD enzyme activity and MRTX1133 sensitivity. Mean ± SD, Student’s t test. ∗∗p < 0.01; ns, not significant. See also Figures S1–S3.

Figure 2

G6PD inhibition reduces malignancy and resistance to MRTX1133 in KRAS^G12D-mutated PDAC (A) Heatmap illustrating organoid area fold changes and synergy indexes with MRTX1133 following treatment with the indicated inhibitors. Measurements taken 6 days post treatment. Organoid area fold changes and synergy indexes are represented by the indicated values of different colors and sizes. (B) Synergy analysis of RRx-001 and MRTX1133 using the Loewe model in KPC organoids. (C–E) Representative images of pancreatic tissues from KC mice stained with H&E, Alcian blue, and amylase/CK19 with or without RRx-001 treatment (5 mg/kg/day) (C). Quantification of the total (D) and differential-grade (E) area of precancerous lesions in the entire pancreatic tissue section (n = 6). (F) Representative images and quantification of pan-keratin and Ki67 staining in PDAC tissues from KPC allografts and PDX-1 models, with and without RRx-001 treatment (5 mg/kg/day) (n = 6). (G and H) Tumor growth of KPC allografts (G) and PDX-1 (H) models with or without RRx-001 (n = 6). (I–L) Tumor growth and survival analysis of KPC allografts (I and J) and PDX-1 (K and L) models treated with RRx-001 (5 mg/kg/day) and/or MRTX1133 (30 mg/kg/day) (n = 6). (M–P) Tumor growth and survival analysis of MRTX1133-resistant KPC allografts (M and N) and PDX-1 (O and P) models treated with RRx-001 and/or MRTX1133. (n = 6). Mean ± SD, Student’s t test. ∗∗p < 0.01, ns, not significant. See also Figures S4.

Figure 3

KRAS^G12D drives PPP reprogramming through UBE2T-mediated p53 ubiquitination (A) Microscale thermophoresis (MST) curve displaying the interaction between p53 and G6PD. K_D, the equilibrium dissociation constant. (B) Detection of G6PD enzyme activity using NADPH/NADP⁺ ratio in KRAS^WT or KRAS^G12D PDO-1 with or without TP53 knockdown (n = 6). (C and D) Ubiquitination assay illustrating the degree of p53 ubiquitination in HEK-293T (C) and BxPC-3 (D) cells expressing the indicated plasmids. (E) Co-immunoprecipitation (coIP) assays reveal the interaction between p53 and G6PD in control (SgCtr) or UBE2T-knockout (SgUBE2T) BxPC-3 cells coexpressing the indicated plasmids. (F) G6PD enzyme activity in KRAS^WT or KRAS^G12D PDO-1 with or without UBE2T deletion (n = 6). (G) G6PD enzyme activity in SgCtr or SgUBE2T PDO-3 with or without TP53 knockdown (n = 6). (H) GSEA of differential metabolites from lesion tissues of KC or UKC mice based on spatial metabolomics data. (I and J) H&E staining and MSI of G-6-P, R-5-P, S-7-P, and 3-PGA (I), followed by statistical analysis (n = 3) (J). (K) Heatmap showing the indicated metabolites level from U-¹³C₆-labeled metabolic flux analysis in WT and G12D-mutant PDO-1 with or without UBE2T knockout (n = 3). See Table S1. Mean ± SD, Student’s t test. ∗p < 0.05, ∗∗p < 0.01, ns, not significant. See also Figures S5.

Figure 4

KRAS^G12D amplifies UBE2T transcription by Rb/E2F1/p53 axis-mediated positive feedback loops (A) Schematic diagram illustrating the identification of UBE2T transcription factors. (B) Volcano plot showing DEGs between KRAS^WT and KRAS^G12D PDO-1. (C) DNA pull-down assay showing the interaction of E2F1 with UBE2T promoter (top). Dual-luc assays detecting the transcriptional activity of the indicated UBE2T promoter with or without E2F1 overexpression (bottom) (n = 6). (D) Dual-luc assays detecting the transcriptional activity of UBE2T promoter (full length, −886 to −876 bp, and its mutant version) with or without E2F1 overexpression (n = 6). (E) CoIP assays showing the interaction between Rb and E2F1 in BxPC-3 cells. Green fluorescent protein (GFP) as control. (F and G) Immunoblotting (IB) analysis with the indicated antibodies in control or KRAS^G12D-overexpressed BxPC-3 cells with or without palbociclib treatment (F)/E2F1 knockdown (G). (H and I) Ubiquitination assay showing the degree of p53 ubiquitination using BxPC-3 cells expressing the indicated plasmids. (J) IB analysis with the indicated antibodies in control or KRAS^G12D-overexpressed BxPC-3 cells with or without TP53 knockdown. (K) Dual-luc assays detect the transcriptional activities of the UBE2T promoter (−886 to −876 bp) with or without E2F1 and/or p53 overexpression (n = 6). (L) IB analysis with the indicated antibodies in KRAS^G12D-overexpressed BxPC-3 cells with or without p53 overexpression and/or palbociclib treatment. (M) CoIP assays assess the interaction of E2F1 with Rb or p53 in BxPC-3 cells expressing the indicated plasmids. (N) MST curve showing the interaction between p53 and E2F1. (O) Schematic diagram of the generation of deletion-mutation p53. (P) CoIP assays detect the interaction between E2F1 and p53 mutants in HEK-293T cells expressing the indicated plasmids. (Q) Dual-luc assays detect the transcriptional activities of the UBE2T promoter (−886 ∼ −876 bp) with or without E2F1 and/or p53-mutant overexpression (n = 6). (R) IB analysis with the indicated antibodies in KRAS^G12D-overexpressed BxPC-3 cells expressing the indicated p53-mutant plasmids. (S) Schematic diagram of regulatory mechanism. Mean ± SD, Student’s t test. ∗∗p < 0.01, ns, not significant. See also Figures S6.

Figure 5

Genetic ablation of UBE2T inhibits malignant progression and potentiates MRTX1133 efficacy in KRAS^G12D-mutant PDAC (A–C) Pancreatic tissues from KC and UKC mice aged 2, 4, 6, 8, 10, and 12 months, stained with H&E and amylase/CK19 (A). Quantification of the total (B) and differential-grade (C) area of precancerous lesions in the entire pancreatic tissue section (n = 6). (D) Kaplan-Meier survival curves with log-rank test comparing overall survival between KPC and UKPC mice. (E) H&E, amylase/CK19, and Ki67 staining of PDAC tissues from 20-week-old KPC and UKPC mice (left). Quantification of Ki67 level (right) (n = 6). (F) H&E staining of liver tissues from 24-week-old KPC and UKPC mice. (G) G6PD enzyme activity measured by NADPH/NADP⁺ ratio in SgCtr or SgUBE2T PDO-3 and KPC or UKPC organoids with or without MRTX1133 treatment (10 μM, n = 6). (H and I) Representative images (H) and quantification (I) of the response of SgCtr or SgUBE2T PDO-3 response to MRTX1133 (10 μM, n = 6). (J) Sensitivity of SgCtr or SgUBE2T PDO-3 and KPC or UKPC organoids to MRTX1133 (n = 6). (K and L) Tumor growth (K) and tumor weight (L) of KPC or UKPC allografts models with or without MRTX1133 treatment (30 mg/kg/day) (n = 6). (M and N) Tumor growth (M) and survival analysis (N) of KPC or UKPC allografts treated with or without MRTX1133 (n = 9). Mean ± SD, Student’s t test. ∗∗p < 0.01. See also Figures S7 and S8.

Figure 6

UBE2T inhibitor PGG suppresses malignant progression and MRTX1133 resistance by regulating PPP reprogramming (A) Computational model and interactions of PGG and UBE2T. (B) Heatmap displaying the indicated metabolites level from U-¹³C₆-labeled metabolic flux analysis in PDO-3 with or without PGG treatment (10 μM) (n = 3). See Table S1. (C–E) Representative images of pancreatic tissues stained with H&E, Alcian blue, and amylase/CK19 with or without PGG treatment (40 mg/kg/day) (C). Quantification of the total (D) and differential-grade (E) area of precancerous lesions in the entire pancreatic tissue section (n = 6). (F) Representative images of PDAC tissues stained with H&E, amylase/CK19, and Ki67 in 20-week-old KPC mice with or without PGG treatment. (G) Quantification of the Ki67 level (n = 6). (H) Dual-luc assays detect the UBE2T promoter activity (positions −886 to −876 bp) with or without PGG (10 μM) and/or MRTX1133 (10 μM) treatment (n = 6). (I) IB analysis with the indicated antibodies in PDO-3 with or without PGG (10 μM) and/or MRTX1133 (10 μM) treatment. (J) G6PD enzyme activity in PDO-3 with or without PGG (10 μM) and/or MRTX1133 (10 μM) treatment (n = 6). (K) Synergy analysis of PGG and MRTX1133 in PDO-3 and KPC organoids using the Loewe, Bliss, HSA, and ZIP model. (L–N) Tumor growth of KPC allografts (L) and PDX-1 (M) and PDX-2 (N) models treated with PGG (40 mg/kg/day) and/or MRTX1133 (30 mg/kg/day) (n = 6). (O) Overall survival analysis of KPC allografts and PDX-1 and PDX-2 models treated with PGG and/or MRTX1133. (P–R) Tumor growth of MRTX1133-resistant KPC allografts (P) and PDX-1 (Q) and PDX-2 (R) models treated with PGG and/or MRTX1133 (n = 6). (S) Overall survival analysis of MRTX1133-resistant KPC allografts and PDX-1 and PDX-2 models treated with PGG and/or MRTX1133. Mean ± SD, Student’s t test. ∗∗p < 0.01, ns, not significant. See also Figures S9–S11.

Figure 7

MFP shrinks tumor volume and sustains long-term survival in PDAC with KRAS^G12D mutation (A) Schematic diagram of MFP nano-delivery system construction. (B–D) Tumor growth of KPC allografts (B) and PDX-1 (C) and PDX-2 (D) models treated with MFP (n ≥ 6). (E) Overall survival of KPC allografts and PDX-1 and PDX-2 models treated with MFP. NR, not reached. (F–H) Fold changes of tumor volume in KPC allografts (F) and PDX-1 (G) and PDX-2 (H) models treated with MFP at 120 days (n ≥ 6). mPD, progressive disease; mSD, stable disease; mPR, partial response; mCR, complete response. (I) PFS of KPC allografts and PDX-1 and PDX-2 models treated with MFP. (J–L) Tumor growth of MRTX1133-resistant KPC allografts (J) and PDX-1 (K) and PDX-2 (L) models treated with MFP (n ≥ 6). (M) Overall survival of MRTX1133-resistant KPC allografts and PDX-1 and PDX-2 models treated with MFP. (N–P) Fold changes of tumor volume in MRTX1133-resistant KPC allografts (N) and PDX-1 (O) and PDX-2 (P) models treated with MFP at 120 days (n ≥ 6). (Q) PFS of MRTX1133-resistant KPC allografts and PDX-1 and PDX-2 models treated with MFP. Mean ± SD, Student’s t test. ∗∗p < 0.01; ns, not significant. See also Figures S12–S14.