Prognostic value of SLC4A4 and its correlation with the microsatellite instability in colorectal cancer

Abstract

Objective: To explore new biomarkers related to microsatellite instability in order to better predict prognosis and guide medication.

Methods: The "limma" R package was used to identify differentially expressed genes in GSE24514, and then weighted correlation network analysis was used to select key genes. Different cell types in the tumor microenvironment were identified and analyzed by single-cell sequencing, with a Lasso regression model used to screen prognostic variables. Furthermore, the correlation between microsatellite instability and potential prognostic variables was explored, as well as the expression characteristics and clinical characteristics of the prognostic variables in the TCGA, UALCAN, and HPA databases. PCR assay was used to investigate the expression of SLC4A4 in colorectal cancer cell lines. Finally, we further verified the expression of SLC4A4 by immunohistochemistry.

Results: First, 844 differentially expressed genes in GSE24514 were identified. Subsequently, weighted co-expression network analysis (WGCNA) of GSE24514 obtained all the genes significantly associated with microsatellite instability (MSI), a total of 1452. Analysis of GSE166555 single cell sequencing data set yielded 1564 differentially expressed genes. The gene sets obtained from the above three analysis processes were intersected, and 174 genes were finally obtained. The Lasso regression model revealed two potential prognostic genes, TIMP1 and SLC4A4, of which, there was a stronger correlation between microsatellite instability and SLC4A4. The mRNA and protein expression of SLC4A4 was significantly decreased in tumors, and patients with low SLC4A4 expression had a poor prognosis. In addition, SLC4A4 was specifically expressed in epithelial cells. In the microenvironment of colorectal cancer, malignant cells have a strong interaction with different stromal cells. PCR showed that SLC4A4 was significantly down-regulated in colorectal cancer cell lines Caco-2, HCT116 and HT29 compared with normal control NCM460 cell lines. Immunohistochemistry also showed low expression of SLC4A4 in colorectal cancer.

Conclusion: SLC4A4, as a tumor suppressor gene, is significantly downregulated and positively correlated with microsatellite instability, thus it may be combined with microsatellite instability to guide colorectal cancer treatment.

Keywords: Slc4a4; biomarker; colorectal cancer; microsatellite instability; single-cell sequencing.

Figure 1

The flow chart.

Figure 2

The sample quality was acceptable and many differential genes were identified. (A) PCA before quality control, (B) UMAP before quality control, (C) PCA after quality control, (D) UMAP after quality control, (E) sample normalization box diagram, (F) volcanic map, and (G) heat map.

Figure 3

Mining potential biomarker genes associated with MSI based on weighted gene networks. (A) scale independence, (B) average connectivity, (C) The sample clustering indicated no outlier samples, (D) gene clustering, (E) module feature vector clustering, and (F) heat map of module and phenotype correlation.

Figure 4

Cell heterogeneity in the microenvironment of colorectal cancer revealed by single-cell sequencing. (A) The UMAP map revealed 33 different cell clusters, (B) cell marker dot pattern, (C) 13 cell types were obtained after cell annotation, and (D) fan and stacked bar graphs show the proportion of the different cell types.

Figure 5

The correlation between the potential biomarker SLC4A4 and MSI is significant. (A) The Venn diagram indicates 174 intersecting genes, (B) the trace diagram of the different variables in the Lasso regression, (C) two undetermined variables were obtained by Lasso regression, (D) lollipop plot of the correlation between TIMP1 expression in different tumors and MSI, (E) the lollipop plot of correlation between SLC4A4 expression in different tumors and MSI, (F) risk factor diagram, and (G) correlation scatter plot.

Figure 6

SLC4A4 is significantly downregulated in tumors and contributes to poor prognosis. (A) Non-paired samples, (B) paired samples, (C) SLC4A4 transcriptional levels in UALCAN-TCGA, (D) SLC4A4 protein levels in UALCAN-CPTAC, (E) time-dependent AUC, (F) ROC tested the stability of SLC4A4 as a biomarker, (G) SLC4A4 protein expression in normal tissue, (H) SLC4A4 protein expression in tumor tissues, (I) the difference in SLC4A4 expression in different N stages, (J) low SLC4A4 expression is associated with poor prognosis, (K) the low SLC4A4 expression in the UALCAN database contributed to poor prognosis, and (L) grouping in combination with gender revealed a significantly poor prognosis for men with low SLC4A4 expression. *P<0.05; ***P<0.001.

Figure 7

SLC4A4 is specifically and highly expressed in epithelial cells. (A) The UMAP plot shows SLC4A4 expression in each cell type, (B) the violin diagram shows SLC4A4 expression in each cell type, (C) cell communication heat map, and (D) enrichment intensity and the fraction of different hallmark signaling pathways in each cell.

Figure 8

Experimental verification of SLC4A4. (A) PCR experiments in cell lines. SLC4A4 was significantly down-regulated in colorectal cancer cell lines Caco-2, HCT116 and HT29 (***P<0.001). (B) Immunohistochemical examination of 3 clinical samples. The expression of SLC4A4 is low in tumor tissue.