Alarm Signal S100-Related Signature Is Correlated with Tumor Microenvironment and Predicts Prognosis in Glioma

Abstract

Glioma are the most common malignant central nervous system tumor and are characterized by uncontrolled proliferation and resistance to therapy. Dysregulation of S100 proteins may augment tumor initiation, proliferation, and metastasis by modulating immune response. However, the comprehensive function and prognostic value of S100 proteins in glioma remain unclear. Here, we explored the expression profiles of 17 S100 family genes and constructed a high-efficient prediction model for glioma based on CGGA and TCGA datasets. Immune landscape analysis displayed that the distribution of immune scores, ESTIMATE scores, and stromal scores, as well as infiltrating immune cells (macrophages M0/M1/M2, T cell CD4+ naïve, Tregs, monocyte, neutrophil, and NK activated), were significant different between risk-score subgroups. Overall, we demonstrated the value of S100 protein-related signature in the prediction of glioma patients' prognosis and determined its relationship with the tumor microenvironment (TME) in glioma.

Figure 1

Flow chart of this study.

Figure 2

The expression profiles of 17 S100 family genes in glioma. Heatmap depicting the expression profiles of 17 S100 family genes in the (a) TCGA cohort and (b) CGGA cohort. The differential expression of 17 S100 protein family genes between LGG and GBM was explored in the (c) TCGA cohort and (d) CGGA cohort, respectively. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗∗P < 0.0001.

Figure 3

Identification of S100 family genes related to overall survival of glioma. (a) In the CGGA cohort, 17 S100 protein family genes were analyzed by univariate cox analysis. (b) LASSO coefficient profiles of 9 S100 family genes. (c) Partial likelihood deviance of different variables revealed by the LASSO regression model. (d) Bar plot displaying the coefficients constructed using the LASSO method. K-M OS curves of 8 S100 family genes were drawn in the (e) CGGA cohort and (f) TCGA cohort, respectively.

Figure 4

Construction of the risk-score signature using 8 S100 family genes. (a) The expression of 8 signature S100 family genes, survival status, and risk score of each patient in the CGGA cohort. (b) K-M OS curves of different risk subgroups in the CGGA cohort. (c) ROC curves showing the sensitivity and specificity of risk score in predicting the OS of glioma patients at 1-, 3- and 5-year in CGGA cohort. (d) The expression of 8 signature S100 family genes, survival status, and risk score of each patient in the TCGA cohort. (e) K-M analysis of different risk subgroups in the TCGA cohort. (f) ROC curves showing the sensitivity and specificity of risk score in predicting the OS of glioma patients at 1-, 3-, and 5-year in TCGA cohort.

Figure 5

The prognostic analysis of the risk score in different subtypes. (a) Analysis of the risk scores in different IDH status, grade, gender, age, MGMT status, and 1p/19q status subtypes. (b) K-M OS curves of different risk subgroups in different subtypes compartmentalized by grade, gender, age, IDH status, 1p/19q status, and MGMT status. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001.

Figure 6

Immune and stromal signatures of gliomas. Ridge plot showing the immune, ESTIMATE, and stromal scores of different risk-score subgroups in the (a) CGGA cohort and (b) TCGA cohort, respectively. Correlation analysis of risk score with immune, ESTIMATE, and stromal scores in the (c) CGGA cohort and (d) TCGA cohort, respectively. Matrix bubble displaying the correlation analysis of the eight S100 family genes with the infiltration levels of T cell CD4+ naïve, Tregs, NK activated, monocyte, neutrophil, and macrophages M0/M1/M2 in (e) the CGGA cohort and (f) the TCGA cohort. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001.

Figure 7

The expression profiles of the immune signature genes in glioma. Violin illustration displaying the expression profiles of 39 immune signature genes between different risk subgroups in the (a) CGGA cohort and (b) TCGA cohort, respectively. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001.

Figure 8

Analysis of 22 infiltrating immune cells in different subgroups divided according to risk score. (a, b) The average frequencies of 22 immune cells of glioma patients in different subgroups were analyzed by TIMER2. Correlation analysis of 8 S100 family genes with the infiltrating immune cells in the (c) CGGA cohort and (d) TCGA cohort, respectively. Correlation analysis of the risk scores with 8 infiltrating immune cells in (e) the CGGA cohort and (f) the TCGA cohort, respectively. Boxplot displaying the infiltration levels of T cell CD4+ naïve, macrophages M0/M/M2, Tregs, monocyte, and neutrophil and NK activated between the different risk subgroups in the (g) CGGA cohort and (h) TCGA cohort, respectively. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001.

Figure 9

Risk score is an independent prognostic factor for glioma. (a) The clinical features in the CGGA cohort were analyzed by univariate cox regression analysis. (b) The positive clinical features and risk signature in the CGGA cohort were then analyzed by multivariate cox analysis. (c) The nomogram was used to predict the prognosis of patients in the CGGA cohort at 1-, 3-, and 5-year. (d) The calibration curve was drawn to displaying the effect of nomogram in predicting the OS of glioma patients in the CGGA cohort.

Figure 10

Validation the expression profiles of the prognostic S100 protein family genes. (a) Boxplot showing the expression profiles of eight prognostic S100 family genes between paracancerous tissue, tumor marginal tissues, and tumor core tissue in the GSE59612 dataset. (b) Immunohistochemical staining analysis of the protein levels of S100A2, S100A4, S100A10, and S100A11 between LGG and GBM. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001.

Figure 11

S100A4 affects proliferation and migration of glioma cells. (a) qRT-PCR and (b) western blot analysis of S100A4 knockdown efficiency in LN229 cells. (c) Analysis of proliferation of control and S100A4-deficient LN229 cells by CCK8 assay. (d) Representative images and (e) statistical analysis of EdU assay in control and S100A4-deficient LN229 cells. (f) Representative images and (g) statistical analysis of cell migration assay in control and S100A4-deficient endothelial cells at the indicated times. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001.