Therapeutic targeting of FOSL1 and RELA-dependent transcriptional mechanisms to suppress pancreatic cancer metastasis

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive cancer often diagnosed at an advanced stage, leading to a poor prognosis. The tumor microenvironment (TME) plays a crucial role in driving metastasis, with inflammatory signaling pathways contributing to tumor progression and therapy resistance. However, the combined effects of inflammatory and oncogenic signaling on the epigenetic regulation of PDAC metastasis are poorly understood. Here, we demonstrate that tumor necrosis factor-alpha (TNFα) and epidermal growth factor (EGF) signaling converge to regulate PDAC cell migration through the activation of NF-κB and AP-1 transcription factors. Using single-cell RNA sequencing, in vitro and in vivo models, we show that the simultaneous activation of these pathways with TNFα and EGF cooperatively induces the expression of genes associated with cell motility and migration. Consistently, combinatorial induced genes are co-regulated by the transcription factors FOSL1 and RELA. Remarkably, inhibition of NF-κB transcriptional activity with a glucocorticoid receptor (GR) mixed agonist significantly reduced PDAC cell migration by decreasing RNA polymerase II recruitment to target genes. These findings reveal a novel mechanism by which inflammatory and oncogenic pathways cooperate to drive PDAC metastasis and highlight the therapeutic potential of GR agonists in mitigating tumor cell migration. Our study offers promising avenues for developing mechanism-based therapeutic strategies in PDAC management.

Fig. 1. TNFα-producing macrophages in TME enhance the ability to promote cell migration in PDAC.

A, B Schematic of published scRNA-seq data (left) and uniform manifold approximation and projection (UMAP) (right) of six published scRNA-seq data from patients with PDAC (n = 136,163 cells from n = 71 donors). Each cell is colored by cell subset. C UMAPs showing the expression patterns of the TNF family gene and the TNF receptor TNFRSF1A (n = 136,163). D UMAPs showing cells that express TNF-induced NF-κB target genes CXCL1 and TNFAIP3 (n = 136,163). E Percentage fraction of macrophages over total number of cells per donor sample (n = 20) representing highest (n = 10; purple) and lowest (n = 10; gray). F UMAPs (left) and violin plots (right) comparison of the ductal cell type 2 cluster of CXCL1 and TNFAIP3 expression in the macrophage^high and macrophage^low groups. G Heatmap of scaled expression after differential expression analysis using DESeq of the top up- and downregulated genes in the high vs low macrophage cohorts. Some analysis methods were adapted from Chijimatsu et al. 2022 (reference in supplementary material. H Representative crystal violet staining and quantification (mean ± s.d.) of migrated cells (n = 3) from the transwell co-culture migration assay of AsPC-1 (top; unpaired t-test) and L3.6pl (bottom; unpaired t-test) cells with THP-1 cells (n = 3). Scale bars, ×4 magnification.

Fig. 2. EGF and TNFα cooperatively stimulate cell migration PDAC.

A UMAP of EGFR expression from the scRNA-seq dataset in Fig. 1a. B Western blot for MAPK (pERK, ERK, pMEK, MEK) and NF-κB (pIKKα/β, IKKα, pRELA, RELA) pathway proteins in AsPC-1 and L3.6pl cells following EGF and TNFα treatments for 30 min (right). Representative of n = 3 independent experiments. HSP70 and β-Actin serve as loading controls. C Heatmap of the differentially expressed genes following RNA-seq on AsPC-1 cells treated with EGF and TNFα for 48 h. Unbiased clustering analysis was done after differential expression with DESeq2. One TNFα-treated replicate (replicate 1) was excluded due to low read depth and outlier behavior in PCA and z-score heatmap analyses, which indicated a sequencing error. D Pathway analysis for biological processes, KEGG, and cellular components enriched for the genes in clusters 5, 6, and 7. The top pathways were selected based on FDR values. E Single-cell migration tracking migration assay captured every 15 min over 48 h in AsPC-1 cells (magnification = ×10). F Mean instantaneous speed calculated from migrated tracks for AsPC-1 and L3.6pl following vehicle, EGF, TNFα, combined (TNFα/EGF) treatments (one-way ANOVA; Dunnett’s multiple comparisons compared to vehicle). G Mean instantaneous speed calculated from migrated tracks for AsPC-1 and L3.6pl following vehicle, TNFα/EGF, Bay 117082, and trametinib treatments (one-way ANOVA; Dunnett’s multiple comparisons compared to TNFα/EGF).

Fig. 3. Combined TNFα/EGF co-stimulation induces IL1B expression in PDAC cells.

A, B Expression of IL1β (purple), CD68 (red; macrophages), cytokeratin (pan) (yellow; tumor cells), α-SMA (green; fibroblasts), and DAPI (blue; nucleus) detected by multiplex immunofluorescence in PDAC tumor samples (original magnification 700 μm; magnification of white boxes 100 μm; inserts are 40 μm). C TNFα expression (mean ± s.d.) in AsPC-1 and L3.6pl cells after co-culture with THP-1 cells (n = 3) (unpaired t-test). D IL1B expression (mean ± s.d.) in AsPC-1 and L3.6pl cells after co-culture with THP-1 cells (n = 3) (unpaired t-test).

Fig. 4. FOSL1 and RELA are the key transcription factors involved in the convergence signaling.

A ChIP-seq heatmap showing H3K27ac occupancy on genomic regions differentially bound in vehicle and combined treatment in AsPC-1 cells. B Dotplot for the top ten transcription factors enriched in H3K27ac upregulated regions in the combined treatment following ChIP-Atlas analysis. The top pathways were selected based on FDR values. Multiple dots indicate multiple datasets for a given factor. C Mean instantaneous speed calculated from migrated tracks for AsPC-1 and L3.6pl following knockdown of AP-1 and NF-κB transcription factors (one-way ANOVA; Dunnett’s multiple comparisons compared to siNT5+TNFα/EGF). D Schematic models for phosphomimetic mutants of FOSL1 and RELA. Western blot for FOSL1 and RELA in AsPC-1 and L3.6pl following transfections with plasmids containing mCherry and EGFP (for wildtype) and FOSL1^S252D/S265D and RELA^S536D. E Mean instantaneous speed calculated from migrated tracks for AsPC-1 and L3.6pl cells transfected with wildtype and mutant FOSL1 and RELA (unpaired t-test). F Immunofluorescence staining for FOSL1 (red) and RELA (green) showing co-localization following combined treatments in AsPC-1 and L3.6pl (magnification = ×63 magnification with oil immersion).

Fig. 5. FOSL1 and RELA co-localize to activate gene transcription.

A Genome browser view (IGV) of AsPC-1 cells at the IL1B, BIRC3, and CXCL1 loci showing FOSL1 and RELA ChIP-seq (red and blue, respectively) signals following vehicle, EGF, TNFα, and TNFα/EGF treatments. B Heatmaps for FOSL1 and RELA ChIP-seq signals showing FOSL1/RELA dominant (n = 2508), FOSL1 dominant (n = 24,357), and RELA dominant clusters (n = 331). C Heatmaps for FOSL1 and RELA ChIP-seq signals showing FOSL1/RELA dominant (n = 4523), FOSL1 dominant (n = 5195), and RELA dominant clusters (n = 24,085) in L3.6pl. D IGV tracks showing ChIP-seq signals for FOSL1, RELA, and H3K27ac after EGF, TNFα, and combined treatments with and without siRNA-mediated FOSL1 and RELA knockdowns. E PLA showing protein interactions between FOSL1 and RELA in AsPC-1 and L3.6pl (one-way ANOVA; Dunnett’s multiple comparisons compared to vehicle). Each red dot represents a single interaction, and DNA was stained with DAPI (magnification = ×63 magnification with oil immersion). Representative of n = 3 independent experiments; boxplots represent the number of PLA signals per 100 cells. F Metagene plots showing the binding profiles of RNAPII following vehicle, EGF, TNFα, and TNFα/EGF at the TSS, gene body, and TES of genes upregulated in TNFα/EGF (n = 334) and unregulated genes (n = 462). G IGV tracks of AsPC-1 cells at the IL1B gene showing RNAPII ChIP-seq signals following vehicle, EGF, TNFα, and TNFα/EGF treatments. H Metagene plots showing the binding profiles of RNAPII following FOSL1 and RELA knockdown at the TSS, gene body, and TES of genes upregulated in TNFα/EGF (n = 334) and unregulated genes (n = 462). I IGV tracks of AsPC-1 cells at the IL1B showing RNAPII ChIP-seq signals following FOSL1 and RELA knockdown.

Fig. 6. Treatment with glucocorticoid agonists inhibits FOSL1 and RELA binding activity.

A Mean instantaneous speed calculated from migrated tracks for AsPC-1 and L3.6pl following Dexamethasone and BI 653048 treatments. B Metagene plots showing the binding profiles of RNAPII following Dexamethasone and BI 653048 treatments of genes upregulated in TNFα/EGF (n = 334). Heatmaps showing GR (C), FOSL1, RELA, and H3K27ac (D) signals at genomic regions co-occupied by GR, FOSL1, and RELA (GR/F1RA; n = 2413) and at GR-only regions genome-wide (n = 40,795). E IGV showing FOSL1, RELA, H3K27ac, RNAPII, and GR signals at IL1B, BIRC3, and CXCL1 following Dexamethasone and BI 653048 treatments. F Scheme depicting the mechanism of action of the glucocorticoids Dexamethasone and BI 653048.

Fig. 7. Targeting FOSL1/RELA binding reverses cell migration in vivo.

A Nude mice were orthotopically implanted with luciferase-expressing L3.6pl cells. Tumor-bearing mice were treated with vehicle and BI 653048 (30 mg/kg; once weekly for 4 weeks) (n = 5/group). B Summary of the liver examination after harvest at the end of week four. C Immunohistochemistry for pan-cytokeratin in tumor sections from harvested liver samples. The sections represent 10% of the total liver size. Samples from vehicle-treated mice exhibit multiple micrometastases, which are absent in n = 4 BI 653048-treated mice. Magnification 2 mm; magnification of boxes 50 μm. D The graphical scheme describes how macrophages within the TME supply the TNFα needed to activate the EGF and TNFα convergence signaling pathway, driving cell migration in PDAC. This combined signaling leads to the co-binding of FOSL1 and RELA at specific genomic regions, which regulates the recruitment and release of RNAPII, promoting the transcription of migration-related genes. Targeting FOSL1 and RELA binding can inhibit this pathway, reducing the migratory behavior of PDAC cells.