Bioinformatic Analyses and Integrated Machine Learning to Predict prognosis and therapeutic response Based on E3 Ligase-Related Genes in colon cancer

In summary, we identified novel molecular subtypes and developed a prognostic model, which will help a lot in the advancement of better forecasting and therapeutic approaches.

Gene expression profiles and clinical data of patients with colon cancer were acquired from the TCGA, GTEx, GSE17537 and GSE29621 databases. ERGs were identified by coexpression analysis. WGCNA and differential expression analysis were subsequently conducted. Consensus clustering identified two molecular clusters. Differential analysis of the two clusters and Cox regression were then conducted. A prognostic model was constructed based on 10 machine learning algorithms and 92 algorithm combinations. The CIBERSORT, ssGSEA and TIMER algorithms were used to estimate immune infiltration. The OncoPredict algorithm and The Cancer Immunome Atlas (TCIA) predicted susceptibility to chemotherapeutic and targeted drugs and immunotherapy sensitivity. CCK-8, scratch-wound and RT‒PCR assays were subsequently conducted.

Colorectal cancer is the third most common cause of cancer death worldwide. We probed the correlations between E3 ubiquitin ligase (E3)-related genes (ERGs) and colon cancer prognosis and immune responses.

Two ERG-associated clusters were identified. The prognosis and immune function of patients in cluster A were superior to those of patients in cluster B. We constructed a prognostic model with perfect predictive capability and validated it in internal and external colon cancer datasets. We discovered significant discrepancies in immune infiltration and immune checkpoints between different risk groups. The group with high-risk had a reduced half-maximal inhibitory concentration (IC50) for some routine antitumor drugs and reduced susceptibility to immunotherapy. In vitro experiments demonstrated that the ectopic expression of PRELP inhibited the migration and proliferation of CRC cells. Conclusions: In summary, we identified novel molecular subtypes and developed a prognostic model, which will help a lot in the advancement of better forecasting and therapeutic approaches.