Novel exosome-associated LncRNA model predicts colorectal cancer prognosis and drug response.

Background: Exosomes are extracellular vesicles that carry various biological substances and have potential as functional mediators in cancers. However, little is known about special molecules in colorectal cancer (CRC) exosomes and their immunological functions.

Aims: Using genomic data from the TCGA-CRC cohort, we constructed a prognostic model based on exosome-related lncRNA for the first time, and the biological role of MIR4713HG in CRC was deeply analyzed.

Method: In this study, we downloaded the gene expression data and clinical data of CRC from the TCGA database. The limma package, SVM-REF and univariate Cox analysis were used to screen out core ERG (CERG) in CRC. LASSO and multivariate Cox regression analyses were used to filter out CERG-related LncRNA and construct a risk score. We explored the distribution and expression levels of ERG in immune cell types by scRNA-seq data. xCell was used to calculate the infiltration levels of stromal cells and immune cells in CRC. KM plotter was used for immunotherapy evaluation of core ERG. Next, we further provide colony formation assay, Transwell assay and xenograft models to understand the carcinogenic effect of MIR4713HG.

Result: First, 43 differentially expressed ERG and 7 CERG were obtained. We explored the expression and distribution levels of CERG in 9 types of cells by scRNA-seq data. In addition, two key exosome-associated LncRNA (MIR4713HG and ZEB1-AS1) were obtained, and a risk score (EALncRI) was constructed. EALncRI could accurately predict the prognosis of CRC. Based on the EALncRI, we constructed a nomogram that is easy to use in clinical practice, which can more accurately and stably predict the prognosis of CRC patients. Furthermore, EALncRI was significantly correlated with the expression of 5 HLA molecules and 13 immune checkpoint molecules. MIR4713HG showed a good predictive effect in the overall survival of patients with immunotherapy evaluation. Knocking down the expression of MIR4713HG significantly inhibited proliferation and migration, and also impaired subcutaneous tumor growth in nude mice.

Conclusion: In this study, a variety of machine learning algorithms were used to construct the EALncRI based on ERG, which can effectively predict the prognosis and distinguish the immune landscape of CRC. More importantly, we conducted an in-depth study on MIR4713HG, which may become an important therapeutic target in CRC.