Integrated analysis of the functions and clinical implications of exosome circRNAs in colorectal cancer

Abstract

Background: Exosome circRNAs (Exo-circRNAs) regulate cancer progression and intercellular crosstalk in the tumor microenvironment. However, their biological functions and potential clinical importance in colorectal cancer (CRC) remain unknown.

Methods: We used exoRBase 2.0 data to identify significant differentially expressed Exo-circRNAs (Exo-DEcircRNAs) in CRC patients and healthy individuals. The least absolute shrinkage and selector operation algorithm, support vector machine-recursive feature elimination, and multivariate Cox regression analyses were used to select candidate Exo-circRNAs and constructed a diagnostic model. Quantitative reverse transcription-polymerase chain reaction analysis was performed to confirm the expression of Exo-circRNAs in the serum samples of patients. Furthermore, we constructed an exosome circRNA-miRNA-mRNA network for CRC. Upregulated target mRNAs in the exosome competing endogenous RNA (Exo-ceRNA) network were used for functional and pathway enrichment analyses. We identified 22 immune cell types in CRC patients using CIBERSORT. Correlation analysis revealed the relationship between Exo-ceRNA networks and immune-infiltrating cells. The relationship between target mRNAs and immunotherapeutic response was also explored. Finally, using the Kaplan-Meier survival curve, a prognostic upregulated target mRNA was screened. We constructed a survival-related Exo-ceRNA subnetwork and explored the correlation between the Exo-ceRNA subnetwork and immune-infiltrating cells.

Results: The constructed diagnostic model had a high area under the curve (AUC) value in both the training and validation sets (AUC = 0.744 and AUC = 0.741, respectively). qRT-PCR confirmed that the Exo-circRNAs were differentially expressed in CRC serum samples. We constructed Exo-ceRNA networks based on the interactions among seven upregulated Exo-DEcircRNAs, eight differentially expressed miRNAs, and twenty-two differentially expressed mRNAs in CRC. Functional enrichment analysis revealed that the upregulated target mRNAs were significantly enriched in cytoskeletal motor activity and the PI3K-Akt signaling pathway. Co-expression analysis showed a significant correlation between the Exo-ceRNA networks and immune cells. The significant correlation was observed between target mRNAs and the immunotherapeutic response. Additionally, based on the prognostic upregulated target gene (RGS2), we constructed a survival-related Exo-ceRNA subnetwork (Exosome hsa_circ_0050334-hsa_miR_182_5p-RGS2). CIBERSORT results revealed that the Exo-ceRNA subnetwork correlated with M2 macrophages (P = 4.6e-07, R = 0.31).

Conclusions: Our study identified an Exo-diagnostic model, established Exo-ceRNA networks, and explored the correlation between Exo-ceRNA networks and immune cell infiltration in CRC. These findings elucidated the biological functions of Exo-circRNAs and their potential clinical implications.

Keywords: circRNA; colorectal cancer; competing endogenous RNA network; diagnostic model; exosome; immune-infiltrating cell.

Figure 1

The workflow implemented in this study.

Figure 2

Construction of a diagnostic model using Exo-DEcircRNAs. (A) Exo-DEcircRNAs between CRC patients and controls in the exoRBase 2.0 cohort. Blue nodes represent downregulation in CRC; Red nodes, upregulation; and black nodes, no significant difference from controls. (B) The 18 diagnostic Exo-circRNAs identified by the LASSO method. (C) The six diagnostic Exo-circRNAs identified by the SVM-RFE method. (D) The intersection of the diagnostic Exo-circRNAs from the two methods. (E) Diagnostic model ROC curves from the training set of the exoRBase 2.0 cohort. (F) Diagnostic model ROC curves from the validation set of the GEO cohort. DEcircRNAs, differentially expressed circRNAs; CRC, colorectal cancer; LASSO, least absolute shrinkage and selection operator; SVM-RFE, support vector machine recursive feature elimination; ROC, receiver operating characteristic.

Figure 3

Validation of the diagnostic model Exo-circRNA expression levels in CRC patient serums (n=24) and benign diseases patient serums (n=24). ns, no significance. ****P<0.0001; ***P<0.001; **P<0.01; CRC, colorectal cancer.

Figure 4

Construction and analysis of Exo-ceRNA network. (A) DEmiRNAs between CRC patients and control individuals in the GEO cohort. Blue nodes represent downregulation in CRC; Red nodes, upregulation; and black nodes, no significant difference from controls. (B) The intersection of target miRNAs in the MRE of Exo-DEcircRNAs and DEmiRNA. (C) DEmRNAs between CRC patients and controls in the GEO cohort. (D) The intersection of target mRNAs in the predicted mRNAs, from the miRDB database, TargetScan database and miRTarbase database, and DEmRNAs. (E) The Exo-ceRNA (Exo circRNA-miRNA-mRNA) regulatory network: Exo-circRNA, ovals; downregulation miRNA, inverted triangles; upregulation mRNA, triangles. DEmiRNAs, differentially expressed miRNAs; CRC, colorectal cancer; MRE, miRNA response elements; DEmRNAs, differentially expressed mRNAs.

Figure 5

The biological functions of upregulation mRNAs that Exo-circRNAs regulated through the ceRNA mechanism. (A) GO enrichment analysis of the upregulation mRNAs. (B) The KEGG pathway of the upregulation mRNAs. (C) A box diagram of the 22 types of immune cells. (D) The correlation between the immune cells and the 22 upregulation target mRNAs. *P<0.05; **P<0.01; ***P<0.001.

Figure 6

Construction and analysis of survival-related Exo-ceRNA subnetwork. (A) survival-related Exo-ceRNA subnetwork. (B) TCGA data indicating that the high expression levels of RGS2 are associated with poor prognostic survival. (C) The correlation between the immune infiltrating cells and RGS2. (D) The correlation between the M2 macrophages and RGS2.