Glutamine—fructose-6-phosphate transaminase 2: A Key Biomarker and Regulator of Chemotherapy Response in Colorectal Cancer

Introduction: Colorectal cancer (CRC) ranks among the world’s most prevalent and lethal malignancies, with an estimated 1.9 million new cases and 904 000 deaths reported in 2022. The efficacy of 5-fluorouracil and oxaliplatin resistance is limited in these patients caused by altered metabolism and tumor microenvironment (TME) interactions. Cancer-associated fibroblasts (CAFs) within the TME induce chemoresistance via chemokine-mediated signaling and extracellular matrix remodeling under hypoxia. We aimed to identify key mediators of CAF-tumor crosstalk in CRC chemoresistance, focusing on GFPT2 as a candidate regulator. Materials and methods We analyzed bulk RNA-seq data from oxaliplatin- (GSE119603) and 5-FU-resistant (GSE196900), a cohort of 110 post-treatment CRC patients (GSE107422), and 398 TCGA-COAD tumors. Differential expression (|Log₂FC|>2, P < 0.05) was identified via DESeq2, and machine learning methods created prognostic chemoresistance-associated signature. We utilized two scRNA-seq datasets (GSE144735, GSE132465) to investigate the CAF-tumor interaction. We found the molecular mechanism of GFPT2-mediated CAF-tumor interaction in CRC cells. For experimental validation, total RNA from 64 paired CRC and normal tissues (2021–2023) was extracted and GFPT2 levels quantified by real-time RT-PCR. Association between all clinicopathological features and GFPT2 expression was assessed in our cohort. Results Differential expression analysis and LASSO regression yielded 17 5-FU/oxaliplatin-associated genes with robust predictive performance (AUC = 0.873). Subsequent Cox-LASSO modeling identified eight genes, including GFPT2, which defined a high-risk signature that stratified patients by survival. High-risk tumors exhibited increased TMB and immune checkpoint expression. Single-cell profiling localized GFPT2 to cancer-associated fibroblasts (CAFs). Co-regulatory network and enrichment analyses linked GFPT2-correlated genes to hypoxia pathways. Validation in CAF bulk RNA-seq confirmed strong GFPT2–hypoxia pathway association (r = 0.92, P < 0001). Moreover, cell–cell communication analysis highlighted GFPT2-dependent CAF ligands (e.g., TIMP1, SERPINE1) interacting with tumor receptors. RT-qPCR in 60 CRC tissues showed GFPT2 upregulation in CRC tissues compared to normal tissues. Its expression level was correlated with diabetes, nodal involvement, and larger tumor size (P < 0.05). Conclusion Elevated GFPT2 expression was observed in our CRC patients and was linked to larger tumor size and diabetes. Furthermore, GFPT-2 was a predictive and prognostic marker in large CRC cohorts. Mechanistically, hypoxic CAFs upregulate GFPT2 to enhance N-/O-glycosylation capacity, promoting secretion of pro-survival ligands SERPINE1 and TIMP1. These ligands bind to the LRP1 and CD63 receptors on tumor cells, activating the PI3K/AKT cascade, thereby establishing a paracrine signaling axis that suppresses 5-FU/oxaliplatin-induced apoptosis. Therefore, GFPT2 may serve as a biomarker for chemotherapy response in CRC patients with diabetes.