Research Square
Codon-Specific Divergence of FSP1 and ACSL4 Senescence Co-expression Decoupling in KRAS-Driven Cancers: Convergent Dual Escape in Colorectal Cancer, Enzyme-Selective Escape in Pancreatic Adenocarcinoma
By Ekmekci CG.
Abstract
Abstract Background. KRAS-mutant cancers upregulate FSP1 (AIFM2) through NRF2 and MAPK activation, conferring ferroptosis resistance parallel to ACSL4. Both enzymes connect to the ferro-aging senescence programme in wild-type tissue. Whether KRAS mutation disrupts their senescence co-regulatory connections, and whether the pattern holds across cancer types and codon variants, has not been examined. Methods. Spearman co-expression correlations between FSP1 or ACSL4 and seven senescence genes were computed in KRAS codon-stratified TCGA-COAD (n = 397; WT n = 337, G12D n = 53, G12V n = 38, G13D n = 30) and TCGA-PAAD (n = 151; WT n = 66, G12D n = 39, G12V n = 26, G12R n = 20). BH-FDR correction was applied within each variant group. The Orphan State Score (OSS) was defined as mean expression / (FDR-significant senescence edges + 0.5). Results. In TCGA-COAD, both FSP1 and ACSL4 lose all FDR-significant senescence co-expression in every KRAS-mutant group (OSS ˜1.1–1.3 in WT vs ˜9.2–9.6 in mutant groups). The decoupling is convergent: both enzymes lose senescence co-regulation simultaneously. In TCGA-PAAD, the pattern diverges by codon. FSP1 decouples in G12D (OSS 8.35, 0 edges) while ACSL4 retains three senescence edges (OSS 1.47). G12V and G12R show the inverse: ACSL4 decouples while FSP1 retains partial or full senescence co-expression. The FSP1–KEAP1 ferroptosis co-expression edge is preserved in WT, G12D and G12V in both cohorts. Conclusions. KRAS mutation disrupts FSP1 and ACSL4 senescence co-expression in both CRC and PAAD, but the enzyme specificity and codon selectivity differ by cancer type. CRC shows convergent dual escape; PAAD shows codon-matched enzyme-selective escape. PAAD G12D may represent a selective FSP1 inhibition opportunity, consistent with published organoid data. All findings are correlational and require functional validation.
