Identification and Characterization of the Copy Number Dosage-Sensitive Genes in Colorectal Cancer

Abstract

Dosage effect is one of the common mechanisms of somatic copy number alteration in the development of colorectal cancer, yet the roles of dosage-sensitive genes (DSGs) in colorectal cancer (CRC) remain to be characterized more deeply. In this study, we developed a five-step pipeline to identify DSGs and analyzed their characterization in CRC. Results showed that our pipeline performed better than existing methods, and the result was significantly overlapped between solid tumor and cell line. We also found that the top five DSGs (PSMF1, RAF1, PTPRA, MKRN2, and ELP3) were associated with the progression of CRC. By analyzing the characterization, DSGs were enriched in driver genes and they drove sub-pathways of CRC. In addition, immune-related DSGs are associated with CRC progression. Our results also showed that the CRC samples affected by high microsatellites have fewer DSGs, but a higher overlap with DSGs in microsatellite low instability and microsatellite stable samples. In addition, we applied DSGs to identify potential drug targets, with the results showing that 22 amplified DSGs were more sensitive to four drugs. In conclusion, DSGs play an important role in CRC, and our pipeline is effective to identify them.

Keywords: cancer immunity; colorectal cancer; copy number; dosage sensitive; driver gene; drug target.

Graphical abstract

Figure 1

Results of the Different DSG-Identifying Methods (A) The dosage effect score of PDSG and the linear model. (B and C) The sum of squares of residuals of PDSG and the linear model. (B) The overview of the sum of squares of residuals, and (C) was a part of (B). (D) The dosage effect score of PDSG, exponent, and S-curve. (E and F) The sum of squares of residuals of PDSG, exponent model and S-curve model. (E) Overview of the sum of squares of residuals, and (F) was a part of (E).

Figure 2

Overview of the DSGs in CRC (A) Overlap of DSGs between solid tumor data from TCGA and cell line data from CCLE with the dosage-sensitive relationship thresholds of 0.5, 0.6, 0.7, and 0.8. (B) Dosage-sensitive relationship scores in TCGA and CCLE. (C) Results of the disturbance experiment, indicating that the DSR of the top five genes is significantly higher than that in random genes. (D) Kaplan-Meier curves of the five DGSs with top DSR.

Figure 3

Microsatellite Analysis of CRC (A) Proportion of SCNA samples from different microsatellite states, of which MSI-H samples have a relatively low alteration ratio. (B) Dose-dosage sensitivity score in different microsatellite states, of which MSI-H samples have a lower score. (C) Venn diagram of DSGs under different microsatellite states. MSI-H has the least DSGs, but it has a high overlap with MSS and MSI-L. (D) Function enrichment of the overlap DSGs among MSS, MSI-H, and MSI-L samples.

Figure 4

DSGs That Are Driver Genes of the CRC Pathway (A) Mapping of DSGs in the CRC pathway from KEGG. The genes with a star mark represent DSGs, with one star mark representing the genes that were dosage-sensitive in either the data in solid tumor or the data in the cell line, and two star marks representing the genes in both the solid tumor and the cell line. (B) Kaplan-Meier curves of five DGSs in the CRC pathway from KEGG.

Figure 5

Kaplan-Meier Curves of Immunosuppressive DSGs

Figure 6

Application of DSGs in Drug-Target Discovery in CRC (A) Drug-target network. The target was the amplified DSGs, and when the target was more sensitive in amplified samples than in normal samples, an edge was added to connect the drug and the target. (B) Differential analysis of ln(IC₅₀) between amplified samples and normal samples in CRC.