ALDH1A3 promotes aggressive basal-like pancreatic cancer through an AP-1/RUNX2 enhancer network

Abstract

The basal-like transcriptional subtype of pancreatic ductal adenocarcinoma (PDAC) is linked to therapy resistance and poor prognosis. The cancer stem cell marker aldehyde dehydrogenase 1A3 (ALDH1A3) is a critical enzyme in acetaldehyde metabolism, but the interconnection to the basal-like subtype is poorly understood. Here, we identified ALDH1A3 as a key gene, which correlates with reduced survival and increased tumor growth. Functional studies revealed interaction of ALDH1A3 with genes like FAM3C, MCC, PMEPA1, and IRS2, forming a network driving PDAC progression. Chromatin profiling showed that ALDH1A3 affects acetylation of histone 3, mediating AP-1 activity, particularly via FOS family members, activating oncogenic pathways such as MAPK and TNF signaling. RUNX2 emerged as a therapeutic target within this network, as its knockdown disrupted MAPK signaling and reduced tumor growth. These findings emphasize the role of ALDH1A3 in linking nuclear metabolic-epigenetic programming in basal-like PDAC, highlighting it as a promising therapeutic target for novel treatment strategies.

Fig. 1. Enriched ALDH1A3-network signature in aggressive PDAC subtypes.

A Venn diagram showing differentially expressed genes between ALDH1A3-positive and -negative samples in patient-derived xenografts (PDXs), cell lines, and tissues. B Spearman´s correlation matrix of 8 genes. ***p < 0.001, **p < 0.01, *p < 0.05. C Interaction network analysis depicting relationships among the genes: ALDH1A3 (A), EMP1 (E), FAM3C (F), IRS2 (I), MAML2 (MA), MCC (MC), PMEPA1 (P), and SP100 (S). D Subcutaneous xenotransplantation of parental AsPC-1 cells (NC) to immunodeficient mice or genetically manipulated AsPC-1 cells with knockdown (KD) of ALDH13 (A), FAM3C (F), MCC (MC), or PMEPA1 (P), demonstrating effects on tumor 1size and weight. Left: images of tumor xenografts; Right: mean weights and standard deviations. Control (NC, n = 18), ALDH1A3^KD (n = 5), FAM3C^KD (n = 7), MCC^KD (n = 11), PMEPA1^KD (n = 7); p values via unpaired Student’s t test. E Metastatic lung colonization following tail vein injection in mice; NC (n = 20), ALDH1A3^KD (n = 22), FAM3C^KD (n = 13), MCC^KD (n = 12), PMEPA1^KD (n = 15); p values by unpaired t test. F Pie chart detailing weight parameters. A (ALDH1A3: 23.08%), F (FAM3C: 19.23%), E (EMP1:15.38%), P (PMEPA1: 11.54%), MC (MCC: 11.54%), I (IRS2: 7.69%), MA (MAML2: 7.69%), S (SP100: 3.85%). G Survival analysis for ALDH1A3^High versus ALDH1A3^Low patients in TCGA and Compass (stage I-III) datasets. H Heatmap demonstrating the ALDH1A3 network score and molecular subtype in TCGA and Compass datasets; High-risk PDAC subtypes are highlighted in purple. I Proportion of aggressive subtypes versus others in the ALDH1A3^High and ALDH1A3^Low groups from TCGA and Compass datasets. J ALDH1A3 network score in TCGA and Compass datasets, divided by aggressive subtype. p values via unpaired t-tests.

Fig. 2. ALDH1A3 regulates AP-1 activity through the FOS family.

A Screening results for AP-1 binding sites in the promoters of candidate genes. B Heatmap illustrating Pearson correlation coefficients between candidate genes and AP-1 subunits from the FOS and JUN families. ***p < 0.001, **p < 0.01, *p < 0.05. C Western blot analysis showing protein levels before (NC) and after ALDH1A3 knockdown (A^KD#1; A^KD#2) in HPAC and AsPC-1 cells. Key proteins detected include JNK activity markers (p-JNK^T183/Y185; p-c-JUN^S73) as well as expression of FOS subunits (FOSB, FOSL1, FOSL2). This panel shows one representative experiment out of three conducted. D TF enrichment analysis in PANC-1 and AsPC-1 cells affected by altered ALDH1A3 expression. Analysis based on RNA-seq data from three biological replicates. E Contingency table analysis for co-expression of ALDH1A3 and FOSL2 in PDAC sections; statistical significance assessed by Chi-square (χ2) test). IHC images displaying FOSL2/ALDH1A3 staining in PDAC sections, scale bars represent 50 μm. F Results from AP-1 luciferase reporter assays in cells subjected to dual or triple knockdown of FOS subunits. Data are presented as mean values from three independent experiments: p values calculated via unpaired t test. G RT-qPCR (left) and western blot analysis (right) performed on AsPC-1 cells before and after knockdown of multiple FOS subunits, examining the expression of the previously described 8 candidate genes. One representative result from three independent experiments is displayed.

Fig. 3. Aldh1a3 is crucial for Jnk/AP-1 activation and pancreatic carcinogenesis in vivo.

A Scheme of treatment of KC mice with cerulein. Western blot analysis of pancreata from KC mice 0, 3, 96 h (h) and 14 days (d) after cerulein treatment and detection of Jnk activity (p-Jnk^T183/Y185, p-c-Jun^S73), expression of Fos subunits (Fosb, Fosl1, Fosl2), and Aldh1a3 ; n = 3/group. B IHC displaying Aldh1a3 expression in KC pancreata 14 days after cerulein treatment. Immunofluorescence revealing Aldh1a3/α-amylase and Aldh1a3/Krt19 co-staining; scale bar: 50 μm; n = 3. C KC mice and KC; Aldh1a3^-/- underwent cerulein treatment and subsequent analysis of protein expression in their pancreas by western blot, as described above. n = 3. D Pancreas proteins analyzed by western blot post-cerulein treatment, n = 5. E H&E- and α-SMA-stained sections from KC^ERT; Aldh1a3^OE or KC^ERT control mice 14 days post-treatment; scale bars: 50 μm, n = 5/group. F Differential gene expression analyzed in RNA-seq between KC^ERT; Aldh1a3^OE (n = 3) and KC^ERT pancreata, (n = 4). Among 987 upregulated genes the top ten enriched TFs were shown. Among 1, 610 downregulated genes five enriched TFs were shown. (G) H&E-stained pancreas sections depicting KC and KC; Aldh1a3^–/– pancreata at 30 weeks. IHC showing Krt19 and α-amylase-positive cells; scale bars: 50 μm, n = 3 (KC), n = 4 (KC; Aldh1a3^–/–). p values by unpaired t test. H IHC results of H&E, Krt19, or α-amylase-stained pancreas sections from one-year-old KC and KC; Aldh1a3^–/– mice. Scale bars: 50 μm, n = 5/genotype.

Fig. 4. ALDH1A3 promotes an oncogenic, basal-like specific transcriptional program by regulating AP-1-mediated enhancer activity.

A Heatmap showing differential ATAC-seq results that identify ALDH1A3-associated accessible or inaccessible chromatin regions in PANC-1 and AsPC-1 cells. B Identification of the top five TF motifs found at ALDH1A3-accessible or -inaccessible chromatin sites in AsPC-1 and PANC-1 cells, analyzed with n = 2 biological replicates. C Pie charts illustrating the genomic distribution of ATAC-seq peaks associated with ALDH1A3, highlighting enrichment of intergenic and intronic sites within open peaks, p = 4.0×10^-6, chi-squared test. D KEGG pathway analysis of genes corresponding to ALDH1A3 open peaks derived from ATAC-seq data. E Quality control analysis of H3K4me1 and H3K27ac CUT&Tag profiles in PANC-1 and AsPC-1 cells, detailing the distribution of all peaks relative to transcription start sites (TSS) in base pairs (bp), with n = 2 biological replicates. F Identification of the top five enriched TF motifs in H3K27ac CUT&Tag data comparing ALDH1A3^OE overexpressing and control PANC-1 cells, as well as ALDH1A3^KD knockdown versus control AsPC-1 cells, with n = 2 biological replicates. G Visualization of overlapping data showing upregulated genes alongside H3K27ac-associated peaks in specific cellular comparisons; accompanied by KEGG pathway analysis, with n = 2 biological replicates in each group. H GSVA displaying expression profiles of basal-like and classical markers in 22 patient-derived xenografs (PDXs) with high (ALDH1A3^High) and low (ALDH1A3^Low) ALDH1A3 expression. I H3K27ac ChIP-seq intensity profiles for basal-like and classical markers across ALDH1A3^High and ALDH1A3^Low PDXs groups. J Top five enriched TF motifs identified in H3K27ac CUT&Tag profiles comparing ALDH1A3^High versus ALDH1A3^Low PDXs groups. K Pie chart showing RNA-seq-based upregulated gene overlap with ALDH1A3-associated H3K27ac peaks in ChIP-seq data between ALDH1A3^High versus ALDH1A3^Low groups, including KEGG pathway analysis.

Fig. 5. Convergence of the AP-1-dependent enhancer network on the oncogenic MAPK pathway in ALDH1A3^High PDAC.

A Quality control (QC) analysis of FOSL2 CUT&Tag in PANC-1 and AsPC-1 cells, illustrating peak distances from the transcription start side (TSS) in base pairs (bp). Data shown for n = 2 biological replicates. B Overlapping pie charts depicting ALDH1A3-associated histone modifications (H3K4me1, H3K27ac) and FOSL2 peaks in PANC-1 cells with ALDH1A3 overexpression (ALDH1A3^OE) versus control, and AsPC-1cells with ALDH1A3 knockdown (ALDH1A3^KD) versus control. Analysis conducted with n = 2 biological replicates per group. C Top five enriched TF motifs identified in FOSL2 CUT&Tag analysis with open or closed chromatin states in PANC-1/ALDH1A3^OE versus PANC-1/control cells and AsPC-1/ control versus AsPC-1/ALDH1A3^KD cells. Performed with n = 2 biological replicates. D Overlapping charts illustrating upregulated genes co-localized with ALDH1A3-associated FOSL2 peaks across different cell line comparisons including KEGG pathway analysis of these genes. Performed with n = 2 biological replicates. E Heat map showing significantly upregulated genes and their association with FOSL2-binding sites linked to the MAPK pathway in PANC-1/ALDH1A3^OE versus PANC-1/control cells, and AsPC-1/ control cells versus AsPC-1/ALDH1A3^KD cells. IGV tracks showed open chromatin peaks of MAP2K3 and EREG in PANC-1/ALDH1A3^OE versus PANC-1/control cells, and NFκB1 and EGFR in AsPC-1/control versus AsPC-1/ALDH1A3^KD cells.

Fig. 6. RUNX2 as druggable target of ALDH1A3^High PDAC.

A Integrated analysis of RNA-seq up, ATAC-seq open, and FOSL2 peaks open under ALDH1A3 regulation in PANC-1 and AsPC-1 cells, identifying intersections with RUNX2 and CD55 as shared candidate targets. B ATAC-seq data revealing open chromatin sites at the RUNX2 locus in PANC-1 cells overexpressing ALDH1A3 (ALDH1A3^OE) compared to controls. C RUNX2 mRNA expression levels in ALDH1A3^High versus ALDH1A3^Low groups from TCGA and Compass datasets. D Western blot analysis showing RUNX2 protein levels in KC; Aldh1a3^–/– pancreata against KC controls, conducted with n = 3 per genotype. E Western blot indicating RUNX2 levels post-knockdown of dual or triple FOS subunits in AsPC-1 cells; representative of three similar experiments. F Quality control (QC) analysis of RUNX2 CUT&Tag in HPAC cells, showing all peak distances from the transcription start side (TSS) in base pairs (bp). Data from n = 2 biological replicates. G Top five enriched TF motifs in RUNX2-binding open chromatin sites in HPAC cells, identified from the RUNX2 CUT&Tag experiment with = 2 biological replicates. H Western-blot analysis demonstrating RUNX2 expression in HPAC cells transduced with negative controls (NC) or RUNX2-specific shRNAs; one of three independent experiments is shown. I Overlapping charts displaying upregulated genes in HPAC/NC versus HPAC/RUNX2^KD and RUNX2-binding genes in HPAC cells, accompanied by KEGG pathway analysis of overlapping genes. Performed with n = 2 biological replicates. J Western-blot analysis illustrating activation levels of oncogenic MAPK pathways (p-ERK^T202/Y204, p-c-JUN^S73, p-p38^T180/Y182 and p-JNK^T183/Y185) and expression of RUNX2 in HPAC/RUNX2^KD and control cells treated with FBS for 1 h or irradiated with UV for 30 min, representative of three independent experiments with similar outcome. K Xenograft model of HPAC cells demonstrating the effect of RUNX2 knockdown on tumor growth, with n = 4, p values by unpaired Student’s t test. L Tumor growth curves, treated with CADD522 (a RUNX2 inhibitor, n = 4) or control (n = 4), in a subcutaneous tumor model generated by HPAC cells; p values by unpaired Student’s t test.