A decision point between transdifferentiation and programmed cell death priming controls KRAS-dependent pancreatic cancer development

Abstract

KRAS-dependent acinar-to-ductal metaplasia (ADM) is a fundamental step in the development of pancreatic ductal adenocarcinoma (PDAC), but the involvement of cell death pathways remains unclear. Here, we show that key regulators of programmed cell death (PCD) become upregulated during KRAS-driven ADM, thereby priming transdifferentiated cells to death. Using transgenic mice and primary cell and organoid cultures, we show that transforming growth factor (TGF)-β-activated kinase 1 (TAK1), a kinase regulating cell survival and inflammatory pathways, prevents the elimination of transdifferentiated cells through receptor-interacting protein kinase 1 (RIPK1)-mediated apoptosis and necroptosis, enabling PDAC development. Accordingly, pharmacological inhibition of TAK1 induces PCD in patient-derived PDAC organoids. Importantly, cell death induction via TAK1 inhibition does not appear to elicit an overt injury-associated inflammatory response. Collectively, these findings suggest that TAK1 supports cellular plasticity by suppressing spontaneous PCD activation during ADM, representing a promising pharmacological target for the prevention and treatment of PDAC.

Fig. 1. TAK1 supports KRAS^G12D-driven ADM and PanIN formation resulting in PDAC development.

a, b Immunohistochemistry (IHC) of phospho-TAK1^Ser192 (p-TAK1 ^Ser192) in samples of a human PDAC tissue microarray (TMA) and quantification using the Allred scoring system (Scores 0-2 = negative, Scores 3-8 = positive). n = 173, biologically-independent samples. c, d Representative images after H&E staining, α-Amylase and cytokeratin 17/19 (CK17/19) IHC on pancreatic tissue sections from 18-week-old mice and quantification of healthy pancreas tissue (Normal), acinar-to-ductal metaplasia (ADM), pancreatic intraepithelial neoplasia 1 (PanIN-1) and PanIN-2 in these animals (n = 6 mice per genotype). e–g Experimental design to study the transdifferentiation capability of pancreatic acinar cell explants in 3D collagen matrices. Representative images by bright-field microscopy, after H&E staining and SOX9 IHC. Duct-like hollow structures formed at day 3 are highlighted with red arrowheads (quantified in g). Green arrowheads indicate SOX9⁺ duct-like structures. Results are expressed as mean ± SEM. The experiment was done with acinar explants from 3 different mice per genotype. P value was calculated by ordinary one-way ANOVA (two-tailed) with Tukey’s multiple-comparisons test. ****p < 0.0001. h–j Experimental setting to study the transdifferentiation capability of KRAS^G12D-expressing pancreatic acinar cell explants in 3D collagen matrices in the presence of various inhibitors. Representative images by bright-field microscopy, after H&E staining and SOX9 IHC of KRAS^G12D-expressing pancreatic acinar explants grown in normal medium (untreated), or treated with DMSO, 5Z-7-Oxozeaenol (10 µM) or TPCA-1 (10 µM). Duct-like structures formed at day 3 are highlighted with red arrowheads and their total amount per genotype is indicated. Green arrowheads indicate SOX9⁺ duct-like structures. Results are expressed as mean ± SEM. The experiment was done with acinar explants from 3 different mice per genotype. P value was calculated by ordinary one-way ANOVA (two-tailed) with Tukey’s multiple-comparisons test. ***p = 0.0003 untreated vs Oxo, ***p = 0.0003 DMSO vs Oxo, ***p = 0.0002 Oxo vs TPCA-1. k, l H&E staining, IHC of α-Amylase and CK17/19 on pancreatic tissue sections from the indicated 52-week-old mice and quantification of the different stages of pancreatic cancer development (n = 6 mice per genotype). Source data are provided in the Source Data file.

Fig. 2. Transdifferentiation sensitizes acinar cells to apoptotic and necroptotic PCD.

a, b Experimental design to study cell death induction in primary pancreatic acinar cells grown in 2D cell culture upon treatment with TNF (100 ng/ml). Immunoblotting analysis in lysates of pancreatic acinar cells isolated from the indicated mice. The experiment was done twice with one mouse per genotype. c–e Experimental design to examine the susceptibility of KRAS^G12D-expressing pancreatic acinar cell explants grown in 3D collagen matrices to PCD after transdifferentiation and formation of duct-like structures. The effect of DMSO, 5Z-7-Oxozeaenol (10 µM) alone or in combination with Z-VAD-fmk (25 µM), Nec-1s (50 µM) or both for 24 h was assessed. Representative images by bright-field microscopy, H&E staining and SOX9, Ki67 and cl. CASP3 IHC. Arrowheads highlight intact duct-like structures (red; quantified in e), SOX9⁺ duct-like structures (green), Ki67⁺ (pink) and cl. CASP3⁺ (yellow) cells. Results are expressed as mean ± SEM. The experiment was done with acinar explants from 3 different KRAS^G12D mice. P value was calculated by ordinary one-way ANOVA (two-tailed) with Tukey’s multiple-comparisons test. **p = 0.0016, n.s. = not significant. f qRT-PCR analysis of the mRNA expression of key apoptotic and necroptotic cell death mediators in 3D collagen matrices of KRAS^G12D pancreatic acinar cell explants before (d0) and after (d3) transdifferentiation, treated with or without MRTX1133 (200 nM or 400 nM). The experimental design is shown in Supplementary Fig. 4a. All values were normalized to Sdha expression. Results are expressed as mean ± SEM. The experiment was done using acinar explants from the same three KRAS^G12D mice. P values were calculated by ordinary one-way ANOVA (two-tailed) with Tukey’s multiple-comparisons test indicated in the figure. g, h Experimental design to examine the upregulation of RIPK3 after transdifferentiation of KRAS^G12D-expressing pancreatic acinar cell explants in 3D collagen matrices upon DMSO or 5Z-7-Oxozeaenol (10 µM) treatment. Representative images after RIPK3 IHC. The experiment was done with acinar explants from 3 different KRAS^G12D mice. i Representative images after RIPK3 IHC on pancreatic tissue sections from the indicated 18-week-old mice (n = 4 mice per genotype). Source data are provided in the Source Data file.

Fig. 3. Inhibition of necroptotic and apoptotic cell death restores transdifferentiation capability of pancreatic acinar cells from KRAS^G12D TAK1^ΔAc mice.

a, b Experimental design to study cell death induction in primary pancreatic acinar cells grown in 2D cell culture upon treatment with TNF (100 ng/ml). Immunobloting analysis in pancreatic acinar cells isolated from the indicated mice. The experiment was done once with one mouse per genotype. c, d H&E staining and IHC of α-Amylase and CK17/19 on pancreatic tissue sections from 18-week-old mice with the indicated genotype and quantification of healthy pancreas tissue (Normal), ADM and PanIN-1 of the same animals (n = 6 mice per genotype). e, f Experimental design to study the transdifferentiation capability of pancreatic acinar cell explants in 3D collagen matrices isolated from the indicated mice. Representative images are shown by bright-field microscopy and after H&E staining. Intact duct-like structures are highlighted with red arrowheads. The experiment was done with acinar explants from 2 different mice per genotype. g, h H&E staining and IHC of α-Amylase and CK17/19 of pancreatic tissue sections from 52-week-old mice with the indicated genotype and quantification of healthy pancreas tissue (Normal), ADM and PanIN-1 of the same mice (n = 5 mice per genotype). i, j Array comparative genomic hybridization (aCGH) analysis of pancreas tissue (ADM, PanIN and PDAC areas) from 52-week-old WT (n = 3), KRAS^G12D (n = 5) and KRAS^G12D TAK1/CASP8^ΔAC RIPK3^-/- (n = 5) mice. The pancreatic tissue of individual transgenic mice was hybridized against the pancreatic tissue of age-matched WT mice and analyzed by aCGH. The q-arm of each chromosome is shown and chromosome numbers are indicated. Dark horizontal bars within the symbolized chromosomes represent G bands. Chromosomal deletions (loss) are indicated in red and amplifications (gain) in blue. Five mice per analyzed genotype are labeled by horizontal-colored bars. Mice with PDAC are labeled with a green star. Map of pancreatic cancer associated genes located in the chromosomal gain and loss regions detected by aCGH analysis is depicted in (j). Vertical lines next to each gene locus represent individual mice. Source data are provided in the Source Data file.

Fig. 4. Pharmacological TAK1 inhibition induces PCD in PDAC patient-derived tumor organoids.

a Schematic view of the generation of PDAC patient-derived tumor organoids (PDOs) from patient biopsies. b Representative examples and quantification of intact PDOs after being established in collagen matrix cultures for 7 days and treated with DMSO (solvent) or 5Z-7-Oxozeaenol (1 µM or 10 µM) for additional 24 h and 48 h is indicated (% over the total PDOs examined). A total of n = 21, n = 23 and n = 33 PDOs (24 h) and n = 108, n = 90 and n = 97 PDOs (48 h) treated with DSMO,1 µM 5Z-7-Oxozeaenol and 10 µM 5Z-7-Oxozeaenol were analyzed, respectively. Results are expressed as mean ± SEM. The experiment was done with PDOs established from the same three patients for all treatments (n = 3). P value was calculated by ordinary one-way ANOVA (two-tailed) with Tukey’s multiple-comparisons test with n.s. = not significant, *p = 0.0346 (24 h), ****p < 0.0001 (24 h), ***p = 0.0008 (24 h), **p = 0.0016 (48 h), *p = 0.0278 (48 h). c Representative images of PDOs established from three different patients as in b, treated with DMSO (solvent) or 5Z-7-Oxozeaenol (10 µM) for 6 days with media changes every 48 h, and visualized by bright-field microscopy and after H&E staining and Ki67 and cl. CASP3 IHC, are shown (n = 3). Intact duct-like hollow structures are highlighted with red arrowheads. Source data are provided in the Source Data file.

Fig. 5. TAK1 inhibition does not cause a pro-inflammatory immune response.

a Experimental approach to assess the immune response elicited by ubiquitous TAK1 inhibition on tumor spheroids generated by isolation of total cell populations from tumor tissue of one PDAC-patient. b Cell viability assay of PDAC-derived tumor spheroids (n = 4 per condition) treated with DMSO or 5Z-7-Oxozeaenol (25 µM) for 72 h. Results are expressed as mean ± SEM. P value was calculated by Mann–Whitney U test (two-tailed), *p = 0.0286. c UMAP of single-cell transcriptomes showing DMSO and 5Z-7-Oxozeaenol-treated immune cells (DMSO: n = 208; 5Z-7-Oxozeaenol: n = 78). d–f UMAP of CD8⁺ (DMSO: n = 67; 5Z-7-Oxozeaenol: n = 7), CD3E⁺ (DMSO: n = 117; 5Z-7-Oxozeaenol: n = 22) and CD44⁺ (DMSO: n = 122; 5Z-7-Oxozeaenol: n = 23) T-cells, isolated from PDAC-derived tumor spheroids, indicating T-cells as the main cell population. Color bar indicates log2-normalized expression. g, h Gene Set Enrichment Analysis (GSEA) in cluster 1 of T-cells isolated from DMSO vs. 5Z-7-Oxozeaenol-treated PDAC patient-derived tumor spheroids. Normalized enrichment score (NES) of significantly enriched (red), suppressed (blue) or non-significantly regulated (gray) (FDR > 0.05) pathways after 5Z-7-Oxozeaenol treatment (***FDR q  <  0.001) are presented. i Experimental approach to assess the immune response elicited upon TAK1 inhibition on CD45⁺ cell-depleted tumor spheroids generated by isolation and in vitro reconstitution from tumor tissue of one PDAC-patient. j Cell viability assay of PDAC-derived tumor spheroids (n = 6 per condition) treated with DMSO or 5Z-7-Oxozeaenol (1 µM, 10 µM and 25 µM) for 72 h. Results are expressed as mean ± SEM. P value was calculated by ordinary one-way ANOVA (two-tailed) with Tukey’s multiple-comparisons test with n.s. = not significant, *p = 0.0114, ****p < 0.0001. k UMAP of single-cell transcriptomes of PDAC-derived immune cells subjected with the supernatant of DMSO- or 5Z-7-Oxozeaenol (25 µM)-treated tumor spheroids (DMSO: n = 529; 5Z-7-Oxozeaenol: n = 427). l GSEA of PDAC-derived T-cells incubated with the conditioned medium from DMSO- or 5Z-7-Oxozeaenol-treated tumor spheroids. NES was non-significantly altered (FDR > 0.05) in all pathways that were affected after 5Z-7-Oxozeaenol treatment in g-h. Source data are provided in the Source Data file.

Fig. 6. Proposed model on the role of TAK1 inhibition in KRAS-driven ADM and PDAC development.

KRAS-dependent, and likely KRAS-independent, ADM induction leads to upregulated expression of PCD mediating molecules. Through its NF-κB-independent prosurvival functions, TAK1 prevents elimination of the PCD-primed transdifferentiated ductal cells, thereby enabling PanIN formation and PDAC development. In contrast, TAK1 deficiency/inhibition impairs cell survival during ADM and prevents PanIN establishment and progression to PDAC.