ITGA8 Had Lower Expression in Non-Small Cell Lung Cancer (NSCLC)
To comprehensively evaluate the role of ITGA8 in multiple solid tumors, we used the TCGA dataset to explore the expression of ITGA8 in cancer. In Figure 1A, the expression of ITGA8 was significantly downregulated in lung, renal, rectal, and prostate cancers, suggesting that ITGA8 may be a tumor suppressor gene for these tumors. To further analyze the transcriptional expression of ITGA8 in NSCLC, we compared the mRNA expression of ITGA8 in normal lung tissues and NSCLC, which showed that the expressions of ITGA8 mRNA in LUAD and LUSC were lower than that in normal tissues (Figure 1B, p ˂ 0.001). Consistently, the data in the Human Protein Atlas showed that the ITGA8 protein level was lower in LUAD (Figure 1C) and LUSC (Figure 1E) than that in normal tissues (Figure 1D, 1F).
Patients with LUAD with High ITGA8 mRNA Expression Had Better Prognoses
We analyzed the survival data associated with mRNA expression of ITGA8 in patients with LUAD. The group cutoff for high or low ITGA8 expression was set with a median. As shown in Figure 1G and 1H, patients with higher ITGA8 expression have longer DSS (Disease-Specific Survival, p < 0.001) and PFI (Progression-Free Interval, p = 0.05). Furthermore, as the disease developed, patients with higher ITGA8 expression had a longer OS (Overall Survival, p < 0.001, Figure 1I). In conclusion, ITGA8 may act as a good prognostic indicator for LUAD. Survival analysis results in patients with LUSC were not statistically significant (Supplementary Figure 1).
High Expression of ITGA8 mRNA in LUAD Samples Was Associated with More Immune Infiltration
Integrins play important roles in immune cells and the immune microenvironment [21]. We found that chemokines (such as CCL14 and CXCL12.1), chemokines receptors (such as CCR6, CX3CR1, and XCR1), and immune checkpoints (such as SELP, EDNRB, CD40LG, and C10orf54) were positively correlated with the ITGA8 expression (Figure 2A, 2B). We separately calculated stromal (Figure 2C, r = 0.56, p < 0.001), immune (Figure 2D, r = 0.33, p < 0.001), and ESTIMATE scores (Figure 2E, r = 0.47, p < 0.001) in each tumor for each patient based on gene expression level, which revealed the positive correlation between the expression of ITGA8 and immune infiltration. We then reassessed the B cell, CD4 + T cell, CD8 + T cell, neutrophil, macrophage, and DC infiltration scores in each tumor for each patient based on gene expression level, and the results were all positively correlated with the ITGA8 expression and revealed more immune infiltration (Figure 1F, B cell, r = 0.33, p < 0.001; T cell CD4, r = 0.38, p < 0.001; T cell CD8, r = 0.28, p < 0.001; neutrophil, r = 0.22, p < 0.001; macrophage, r = 0.35, p < 0.001; and DC, r = 0.37, p < 0.001). The correlation between ITGA8 and immune infiltration revealed that the gene may play a significant role in immunotherapy.
ITGA8 Expression Was Negatively Correlated with Tumor Mutational Burden (TMB), Neoantigen, and MATH
We analyzed the correlation of ITGA8 with TMB, neoantigen, and tumor heterogeneity to further explore the relationship between ITGA8 and immune microenvironment and immunotherapy. TMB, neoantigen, and tumor heterogeneity can be used to predict response to immune checkpoint blockade (ICB) therapy and have emerged as useful biomarkers across multiple cancer types for identifying patients who would benefit from immunotherapy [22, [23]. We conducted correlation analyses between ITGA8 and TMB, neoantigen, and MATH in LUAD, respectively, and the results showed that TMB (Figure 3A, r = −0.20, p < 0.001), neoantigen (Figure 3B, r = −0.13, p < 0.001), and MATH (Figure 3C, r = −0.14, p < 0.001) were negatively correlated with the ITGA8 expression. Subsequently, we created a mutation landscape map for the high and low expression groups of ITGA8 in patients with LUAD and used the chi-square test to evaluate the difference in gene mutation frequency in each group of samples. The following were the top five genes with mutations: TP53, TIN, CSMD3, RYR2, and USH2A (Figure 3D).
ITGA8 Expression Was Negatively Correlated with Cancer Cell Stemness
ITGA8 was closely related to cancer stem cell (CSC), which plays an important role in immune evasion [3]. Some studies have shown that cancer cell stemness affects the efficacy of immunotherapy for various tumors [24, [25, [26]. We calculated cancer cell stemness scores and performed correlation analyses to further investigate the relationship between ITGA8 and cancer cell stemness. The cancer cell stemness score is a measure of the similarity of tumor cells to stem cells, which is associated with active biological processes in stem cells and a higher degree of tumor dedifferentiation. All six cancer cell stemness scores were negatively correlated with the expression of ITGA8 (Figure 4A, RNAss, r = −0.71, p < 0.001; Figure 4B; EREG-EXPss, r = −0.05, p = 0.31; Figure 4C, DNAss, r = −0.23, p < 0.001; Figure 4D, EREG-METHss, r = −0.25, p < 0.001; Figure 4E, DMPss, r = −0.23, p < 0.001; and Figure 4F, ENHss, r = −0.16, p < 0.001). Collectively, these results demonstrate that high expression of ITGA8 may contribute to good immunotherapy efficacy.
LINC01798 Regulated the Expression of ITGA8 through miR17-5p in LUAD
We determined regulatory genes upstream via online database to explore the regulatory mechanism of ITGA8 in lung cancer. The miRNAs that may affect the expression of ITGA8 were obtained through the online database, and the network map was created (Figure 5A). We performed correlation analyses between the expression of these miRNAs and ITGA8 in lung cancer, and three miRNAs (r ≤ −0.2, p < 0.05) were screened out (Figure 5B, miR-17-5p, r = −0.3, p < 0.001; Figure 5C, miR-20a-5p, r = −0.2, p < 0.001; Figure 5D, miR-93-5p, r = −0.28, p < 0.001). PFS-based Kaplan–Meier survival analyses were performed on the abovementioned three miRNAs, respectively (Figure 5E, miR-17-5p, p = 0.015; Figure 5F, miR-20a-5p, p = 0.107; Figure 5G, miR-93-5p, p = 0.152).
A total of 215 lncRNAs that may affect the expression of the abovementioned three miRNAs were obtained through the online database, and correlation analyses were performed between the expression of these lncRNAs and ITGA8 in lung cancer. Six lncRNAs (r ≥ 0.4, p < 0.05) with high correlation were screened out (Supplementary Figure 2A–2F), and Kaplan–Meier survival analyses were performed (Supplementary Figure 3A–3L). Finally, the ITGA8 regulatory network was obtained (Figure 6). According to the results of the Kaplan–Meier survival analysis, we selected LINC01798 (PFS, p = 0.0074 and OS, p = 0.00024) and has-miR-17-5p (PFS, p = 0.015) to conduct experimental verification.
The expression of ITGA8 is lower in LUAD than that in normal lung tissues, as shown in Figure 1B and Figure 1C–1F. To explore the aberrant expression of LINC01798, miR-17-5p, and ITGA8 in cellular levels, we selected human LUAD cell lines, A549, PC9, H1975, H1975/AR, and H1975/ABIR, and human bronchial epithelial cell line, BEAS-2B. The qRT-PCR assay revealed that the expressions of ITGA8 and LINC01798 in LUAD cell lines were significantly lower than that in normal lung tissue cell line (Figure 7A, 7B). In contrast, the expression of miR-17-5p in LUAD cell lines was significantly higher than that in normal lung tissue cell line (Figure 7C). si-LINC01798 was transfected into H1975 to investigate the effects of LINC01798 on LUAD cell functions. The results showed that the expression of LINC01798 was knocked down, and the expression of ITGA8 was downregulated in mRNA and protein levels, whereas the expression of miR-17-5p was slightly upregulated without statistical significance. When further transfected with the si-LINC01798 and miR-17-5p inhibitor, the expression of miR-17-5p decreased, and the downregulation of ITGA8 was reversed, even higher than those in the control group (Figure 7D, 7F, 7G, 7H). When LINC01798 was knocked down, the immune checkpoint genes, SELP, EDNRB, CD40LG, and C10orf54, which were positively correlated with the expression of ITGA8, were reduced in the mRNA level, and miR-17-5p inhibitor could reverse their reduction (Figure 7E).