GBP1 is overexpressed in gastric cancer
Expression of GBP1 in various types of cancer and normal tissues visualized as dot plots based on the RNA-seq data from TCGA and GTEx datasets through GEPIA. The database showed that the mRNA level of GBP1 was significantly higher in STAD (Stomach adenocarcinoma), DLBC (Lymphoid Neoplasm Diffuse Large B-cell Lymphoma), ESCA (Esophageal carcinoma), GBM (Glioblastoma multiforme), HNSC (Head and Neck squamous cell carcinoma ), KIRC (Kidney renal clear cell carcinoma), LGG (Brain Lower Grade Glioma), PAAD (Pancreatic adenocarcinoma), TGCT (Testicular Germ Cell Tumors), but it was significantly lower in KICH (Kidney Chromophobe), PRAD (Prostate adenocarcinoma), UCEC (Uterine Corpus Endometrial Carcinoma) and UCS (Uterine Carcinosarcoma) (Figure 1A). Meanwhile, the analysis of TCGA RNA-seq data using the TIMER database showed that GBP1 mRNA was significantly overexpressed in STAD, ESCA (Esophageal carcinoma), HNSC and KIRC, while it was significantly decreased in KICH, LUAD (Lung adenocarcinoma), LIHC (Liver hepatocellular carcinoma), LUSC (Lung squamous cell carcinoma), PRAD, SKCM (Skin Cutaneous Melanoma), THCA and UCEC (Figure 1B). To verify the protein profile of GBP1 expression level between tumor and normal tissue in gastric cancer, we used HPA for searching evidence of pathology immunohistochemistry (IHC), whose results showed that GBP1 was not found in the normal gastric tissue, but detected in the malignant tissue and mainly located on the cytoplasmic and membranous of cells that distributed in tumor stroma (Figure 1C). Collectively, the results consistently showed that GBP1 expression was varied in different tumor backgrounds, among which it was elevated in gastric cancer tissue compared to the corresponding normal tissue.
Prognostic significance of GBP1 expression in human cancers
In order to investigate whether the cancer patient survival rate was correlated with GBP1 expression level in various types of cancers, the Kaplan-Meier plotter database was used. Higher GBP1 expression was correlated with more favourable prognosis in gastric cancer (Overall Survival (OS) HR=0.53, 95%CI=0.43 to 0.66, P=5.2e-09; Progression-Free Survival (PFS) HR=0.49; 95%CI=0.36 to 0.66, P=1.90e-06) and HCC (OS HR=0.6,95%CI=0.4 to 0.92, P=0.018; PFS HR=0.69, 95%CI=0.49 to 0.98, P=0.036), while lower GBP1 expression corresponded with better prognosis in BRCA (PFS HR=1.44, 95%CI=1.12 to 1.83, P=0.036) and OV (PFS HR=1.4, 95%CI=1.21 to 1.63, P=4.7e-06), and it had no correlation with the survival of lung cancer (Figure 2). Additionally, high GBP1 expression associated with better OS in KIRC, KIRP, LUAD, LUSC, PAAD, and PCPG, while low GBP1 expression associated with better OS in ESCA, SARC, THYM, and THCA. Besides, the results showed that high GBP1 was positively correlated with better PFS in ESCA, KIRP, and HNSC (Additional file 1: Table S1). These results suggested that GBP1 was a potent prognostic biomarker depending on cancer types.
Correlation between GBP1 expression and clinical characteristics of GC patients
To further understood the relevance and underlying mechanisms of GBP1 expression in cancer, we investigated the relationship between the GBP1 expression and clinical characteristics factors including Sex, Stage T/N/M, Lauren classification, and differentiation of gastric cancer patients in the Kaplan-Meier plotter databases. Overexpression of GBP1 was correlated with better OS and PFS in female and male patients as well as two types of Lauren classification (P<0.05), but no association was found between the characteristic of differentiation and OS or PFS. In addition, high expression of GBP1 was positively correlated with OS/PFS in stages except that it was independent of OS in stage 2 (P<0.05). TNM staging is an important basis for judging the treatment and prognosis of gastric cancer patients29, and we found that GBP1 was associated with OS and PFS of patients in stage T2 (OS HR=0.47, P=0.0059; PFS HR=0.48, P=0.0021) and T3 (OS HR=0.51, P=0.0021; PFS HR=0.59, P=0.01), but not in advanced stage T4 (OS HR=0.56, P=0.23; PFS HR=1.78, P=0.21). As for the stage N, which refers to lymph node involvement, GBP1 was positively significant with N1-N3 regional lymph node metastasis and higher GBP1 expression indicated better OS (HR=0.48, P=1.06e-05) and PFS (HR=0.42, P=2.1e-05); Notably, in patients at stage M0, GBP1 is positively correlated with OS (HR=0.5, P=7.10e-05) and PFS (HR=0.53, P=0.0002), even if the patients with distant metastasis, high expression of GBP1 still indicates better PFS (HR=0.5, P=0.0014) (Table 1). Collectively, these results demonstrated that GBP1 as a potent prognostic marker could impact the prognosis in gastric cancer with distinct clinical characteristics.
GBP1 expression positively related to the infiltration of activated T cells in gastric cancer
It has been reported that tumor-infiltrating lymphocytes play an important role in anti-tumor activity, which can be an independent factor of several cancer types 30-32. To explore the underlying mechanisms of GBP1 and good prognosis in gastric cancer, we focused on infiltrating lymphocytes in gastric tumor environment. To determine the relationship between the GBP1 expression level and infiltrating immune lymphocytes abundance, TIMER was utilized to analyze the correlation between GBP1 expression and the infiltration levels of 6 immune cell types. It was found that GBP1 expression was significantly negatively correlated with tumor purity in STAD (cor=-0.205, P=5.8e-05), indicating that GBP1 might be expressed by the non-malignant cells like stromal cells and (or) infiltrating immune cells in tumor tissue. We observed that GBP1 expression was correlated with infiltrating immune cells in different types of cancers whose GBP1 expression was significantly changed in TIMER and GEPIA simultaneously (Additional file 2: Figure S1A-L). In gastric cancer, it showed that high GBP1 markedly correlated with the infiltration of CD8+ T cells (cor=0.54,P=2.11e-29), macrophage (cor=0.204,P=8.02e-05), neutrophils (cor=0.595,P=7.52e-37), dendritic cells (cor=0.536,P=5.10e-29), but negatively correlated with B cells(cor=-0.278,P=5.44e-08), and there was no statistical significance with CD4+ T cells (cor=0.101,P=5.34e-02) (Figure 3A). For learning about more details of tumor immune microenvironment in gastric cancer, TISIDB was also used to analyze GBP1 expression and infiltration of different immune cells. We found that GBP1 was positively correlated with a variety of cell subtypes, among which the groups with higher correlation contained mainly CD8+ and CD4+ T cells. According to the status of cells, these two types of T cells were divided into activated/central memory (cm)/ effector memory (em) subsets. The result showed that activated cells have the highest correlation among CD8+ and CD4+ T cells (rho=0.701, P<2.2e-16; rho=0.681, P<2.2e-16, respectively) (Figure 3B, C). In addition, GBP1 is also positively associated with anti-angiogenic chemokines CXCL9, CXCL10, and CXCL11 in gastric cancer, it is similar to that in colon cancer where GBP1 was involved in the angiostatic immune reaction15(Additional file 3: Figure S2A, B). Taken together, these results suggested that GBP1 was an important gene in the tumor immune microenvironment, increasing the infiltration abundance of activated T cells.
Correlation analysis between mRNA levels of GBP1 and markers of immune cells in tumor microenvironment
The correlation between GBP1 expression and the status of tumor infiltrating immune cells according to the levels of different subsets of immune cells marker gene expression in STAD tissues were identified using TIMER and GEPIA databases. The immune cells analyzed in gastric cancer tissues included CD8+ T cells, T cells (general), B cells, tumor-associated macrophages (TAMs), monocytes, M1 and M2 macrophages, neutrophils, DCs, and natural killer (NK) cells. Moreover, different subsets of functional T cells, namely, T helper 1 (Th1), Th2, follicular helper T (Tfh), Th17, regulatory T (Tregs), and exhausted T cells were also analyzed. Due to the tumor purity of clinical samples affects the analysis of immune infiltration, the correlation analysis was adjusted for purity. After adjustments, the expression level of GBP1 showed a positively correlated with most immune markers in divergent immune cell types and various functional T cells in STAD (Table 2).
Interestingly, we found that the expression levels of most marker sets of CD8+ T cells, T cells(general), dendritic cells, M2 macrophage, monocytes, Th1, Tregs and exhausted T cells have strong correlations with GBP1 expression in STAD using TIMER and GEPIA databases (Table 2, Table 3). Within these immune cells, CD8+ T cells as major cytotoxic lymphocytes could inhibit tumor progression through directly induce cell death, comprising an important part of TILs in tumor immune microenvironment. Thus, the abundance of CD8+ T cells in the tumor immune microenvironment might be associated with a favorable prognosis 33. As shown in Figure 4A and Table 3, GBP1 expression level strongly correlated with the markers of CD8+ T cells, CD8A (r=0.708, P=0.00e+00) and CD8B (r=0.538, P=0.00e+00). And the cytotoxic molecules granzyme B (GZMB), perforin (PRF1) and FasL (FASLG) which usually equipped with CD8+ T cells were also included in this study and were shown that higher expression of GBP1 markedly increased the level of GZMB (r=0.788, P=1.46e-81), PRF1(r=0.739, P=9.64e-67) and FASLG (r=0.731, P=1.33e-64) (Figure 4B). These findings indicate that GBP1 may result in a higher survival rate via increasing the infiltration abundance of CD8+ T cells and their effect of direct killing in STAD.
It suggested that GBP1 in colorectal cancer was mainly involved in the activities of Th1 cells for anti-tumor 15. Herein, in this study, we investigated the relationship between sets of markers of Th1 cells including T-bet, STAT4, STAT1, and IFN-γ, and we found that there was a strong correlation between GBP1 and markers of Th1 cells (Table2, Figure 4C). In addition, the expression of GBP1 correlated significantly with the expression of the marker genes of different subsets of T cells in GC, namely Tregs or exhausted T cells, such as FOXP3, CCR8, PD1, TIGIT, Tim3 and LAG3(P<0.0001; Table 2). High expression of GBP1 may also activate Tregs while the FOXP3+ Tregs is a major factor that could inhibit cytotoxic T cells from killing tumor cells. Interestingly, LAG3, as a pivotal gene that participates in T cell exhaustion, has a strong positive correlation with GBP1 expression, suggesting that high GBP1 expression plays an important role in LAG3 mediating T cell exhaustion.
We observed that the levels of expression in the majority of monocytes and M2 macrophage immunomarker genes were strongly correlated with GBP1 expression in STAD (Table 2). CD86 and CD115 of monocytes, and CD163, VSIG4, and MS4A4A of M2 phenotype greatly correlated with GBP1 expression level in GC (P<0.0001; Table 2). High GBP1 expression related to high infiltration level of DCs in STAD, while DC markers such as HLA-DPB1, HLA-DRA, HLA-DPA1, and CD11c also show significant correlations with GBP1 expression (P<0.0001; Table 2), it may reveal that GBP1 regulate macrophage polarization and DCs infiltration in GC. Furthermore, we assessed the interrelationship among GBP1 level and immune markers in GC in the GEPIA database, whose results were similar to those in TIMER. Strikingly, a significant correlation between GBP1 expression and markers of immune cells occurred in the tumor but barely found in the adjacent normal tissue (Table 3). Collectively, these findings suggested that GBP1 expression strongly correlated with infiltration of immune cells with different phenotypes which indicated it may involve in different immunological pathways in tumor microenvironment, but the underlying mechanisms remained to be further explored.