A seven-gene signature diagnostic model for early-onset colorectal cancer and the tumor-suppressive role of SLC26A2 via redox modulation

Abstract

The rising incidence and poor prognosis of early-onset colorectal cancer (EOCRC) highlight an urgent need for effective diagnostic biomarkers. This study integrates RNA sequencing data with machine learning algorithms to identify EOCRC-specific genes and construct a diagnostic model to improve molecular diagnosis and explore the functional role of key genes. We analyzed RNA sequencing data from EOCRC patients (age ≤ 50) obtained from the Gene Expression Omnibus (GEO) database. The differentially expressed genes (DEGs) between EOCRC and normal tissues were identified and subjected to construct a diagnostic model through random forest analysis and artificial neural network modeling. The diagnostic performance of the model was evaluated in a training set and validated in an independent set. CIBERSORT was utilized to investigate the immune cell infiltration patterns. The expression levels of key DEGs were further verified in 15 paired clinical tissues. Moreover, RT-qPCR and western blotting were applied to detect the expression of solute carrier member 26 family 2 ( SLC26A2 ), which is one of the top-ranked genes, in CRC cell lines. The functional role of SLC26A2 was investigated in EOCRC cell lines (SW480 and HT29) through lentivector-mediated ectopic overexpression experiments assessing proliferation, migration and invasion. The association between SLC26A2 and the cytotoxicity of vanadium-containing compounds (BMOV, VO(acac)₂) was explored, focusing on redox modulation (ROS/GSH levels) using inhibitors (BSO) and antioxidants (NAC). We identified 22 DEGs between EOCRC tumor and normal tissues. A machine learning-based diagnostic model based on seven DEGs ( SLC26A2 , FCGBP , SLC7A5 , CA2 , STC2 , SLC4A4 and THBS2 ) for EOCRC was constructed. The model achieved an AUC of 0.987 (training) and 0.874 (validation). Immune profiling revealed altered infiltration of monocytes, M0 macrophages, and γδ T cells in EOCRC. Experimental validation confirmed the dysregulation of the seven key genes. SLC26A2 was downregulated in EOCRC tissues and its overexpression significantly inhibited cell proliferation, migration, and invasion in vitro. Transcriptomic analysis linked SLC26A2 overexpression to glutathione (GSH) metabolism. EOCRC cells exhibited sensitivity to vanadium compounds, which was enhanced by SLC26A2 overexpression. SLC26A2 expression correlated with compound-induced cytotoxicity alongside modulation of ROS/GSH levels, as evidenced by the synergistic effect of BSO (GSH inhibitor) and the attenuating effect of NAC (antioxidant). We developed and validated a robust machine learning-based diagnostic model for EOCRC using a 7-gene signature. The downregulated gene SLC26A2 acts as a tumor suppressor in EOCRC and is associated with cellular sensitivity to vanadium-containing compounds, potentially through a redox-dependent mechanism linked to GSH metabolism. These findings provide novel insights into the identification of diagnostic biomarkers and highlight the therapeutic potential for EOCRC.