IFI30 expression in GBM tissues was higher than in normal tissues
To explore the possible role of IFI30, we analyzed its expression in 33 human cancers. Compared with corresponding normal tissues, IFI30 mRNA was significantly up-regulated in 26 cancer types, including GBM, bladder urothelial carcinoma (BLCA), and breast invasive carcinoma (BRCA)(Fig. 1A). However, IFI30 was significantly down-regulated in four cancers, including lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), and thymic carcinoma (THYM). In addition, mesothelioma (MESO) and uveal melanoma (UVM) could not be compared due to the lack of normal tissue controls.
Using the TCGA database, cancer samples were grouped according to IFI30 expression levels and showed no significant differences (Table Ⅰ). Significant up-regulation of IFI30 in GBM was observed in a comparative study based on the TCGA database (Fig. 1B). Subsequently, differences in IFI30 transcript levels were further validated using the GSE116520 dataset (Fig. 1C) and the UALCAN database (Fig. 1D), and similar results were obtained.
After determining the transcriptional expression of IFI30 in GBM, we queried HPA database for representative immunohistochemical and immunofluorescence chemical images of IFI30, suggesting that IFI30 expression in GBM tissue was higher than in normal cerebral cortex (Fig. 1E-F). This result was consistent with our previous results regarding differential IFI30 mRNA expression. Meanwhile, immunofluorescence chemistry data suggested that IFI30 was mainly localized to the cytosol (Fig. 1G).
Table Ⅰ. Baseline Clinical Characteristics of GBM Patients in the TCGA Database
Characteristic | Low expression of IFI30 | High expression of IFI30 | p |
n | 84 | 84 | |
Gender, n (%) | | | 0.196 |
Female | 34 (20.2%) | 25 (14.9%) | |
Male | 50 (29.8%) | 59 (35.1%) | |
Race, n (%) | | | 0.391 |
Asian | 4 (2.4%) | 1 (0.6%) | |
Black or African American | 6 (3.6%) | 5 (3%) | |
White | 72 (43.4%) | 78 (47%) | |
Age, n (%) | | | 0.537 |
<=60 | 46 (27.4%) | 41 (24.4%) | |
> 60 | 38 (22.6%) | 43 (25.6%) | |
Karnofsky performance score, n (%) | | | 0.092 |
< 80 | 25 (19.5%) | 11 (8.6%) | |
>=80 | 47 (36.7%) | 45 (35.2%) | |
DSS event, n (%) | | | 0.116 |
Alive | 21 (13.5%) | 13 (8.4%) | |
Dead | 54 (34.8%) | 67 (43.2%) | |
IDH status, n (%) | | | 0.380 |
WT | 73 (45.3%) | 76 (47.2%) | |
Mut | 8 (5%) | 4 (2.5%) | |
Age, mean ± SD | 58.9 ± 13.54 | 59.54 ± 13.58 | 0.763 |
IFI30 methylation in GBM patients
The prognostic value of each CpG of IFI30 DNA methylation was investigated using the MethSurv database. Eleven methylated CpG sites were found, with cg00000029 and cg01783195 having the highest degree of DNA methylation (Fig. 2A). Eight CpG sites were associated with prognosis: cg01485548, cg01533387, cg04096365, cg07533630, cg15825970, cg17004101, cg26152923, and cg27142905 (p < 0.05) (Table Ⅱ). Patients with low IFI30 methylation at these CpG sites had worse overall survival (OS) than those with high IFI30 methylation. Subsequently, we found a significantly lower global methylation level of the IFI30 promoter in GBM tissues from the UALCAN database (Fig. 2B).
To further explore the role of IFI30 in the mechanism of GBM recurrence, we used Illumina 850K methylation chip to detect DNA methylation in primary and recurrent specimens of three GBM patients. Of the 605,192 probes that passed quality control, 62,546 (10.4%) were differentially methylated between treatment-naïve and relapsed samples (FDR q < 0.05, Fig. 2C). In total, 37.12% (23,220/62,546) of these differentially methylated cytosines (DMCs) were hypomethylated (Fig. 2D). Among the genes corresponding to all probes, 8,129 (41.6%) were found to be differentially methylated (FDR q < 0.05, Fig. 2E). Subsequently, cluster analysis was performed for CpG loci that met the screening criteria for differential loci. Interestingly, the DNA methylation signature of Case A with longer progression-free survival (PFS) in recurrent samples was similar to that of primary specimens, whereas the DNA methylation signature of Case C with shorter PFS in primary samples was similar to that of recurrent specimens (Table Ⅲ, Fig. 2F). Further analysis of the methylation of IFI30 gene revealed that among the 13 sites corresponding to IFI30, four sites were differentially methylated, namely, cg01485548, cg26152923, cg26638520, cg07533630 (FDR q < 0.05, (Fig. 2G). Cross-validation of IFI30 differential loci associated with GBM prognosis revealed that cg26152923, cg07533630, and cg01485548 were key prognostic loci (Fig. 2H). In conclusion, based on differences in methylation levels of the IFI30 promoter and the expression profile of IFI30, we speculate that IFI30 may play a key role in the tumorigenesis and recurrence of GBM.
Table Ⅱ. Effect of IFI30 methylation level on the prognosis of GBM.
CpG | HR | P-value |
TSS200-Island-cg00998146 | 0.744 | 0.26 |
TSS1500-N_Shore-cg01485548 | 0.642 | 0.036 |
1stExon-Island-cg01533387 | 0.658 | 0.046 |
TSS200-N_Shore-cg04096365 | 0.622 | 0.024 |
TSS1500-N_Shore-cg07533630 | 0.571 | 0.0097 |
Body-S_Shore-cg11431981 | 0.819 | 0.43 |
Body-Island-cg11777782 | 0.822 | 0.39 |
TSS200-N_Shore-cg13549667 | 0.811 | 0.38 |
3'UTR-S_Shelf-cg15577634 | 1.181 | 0.43 |
TSS1500-N_Shore-cg15825970 | 0.508 | 0.0018 |
Body-Island-cg17004101 | 0.623 | 0.039 |
TSS1500-N_Shore-cg26152923 | 0.574 | 0.0091 |
1stExon;5'UTR-Island-cg27142905 | 0.612 | 0.029 |
Table Ⅲ. Baseline clinical characteristics of three patients with GBM.
| Age (Years) | Sex | Histopathology (primary) | Removed Degree of Glioma | Standard RTl with concurrent TMZ | Adjuvant TMZ | PFS (months) | Histopathology (recurrent) | OS (months) |
Case A | 71 | male | Glioblastoma | All | Yes | Yes | 24 | Glioblastoma | 32 |
Case B | 49 | male | Glioblastoma | All | Yes | Yes | 10 | Glioblastoma | 18 |
Case C | 64 | female | Glioblastoma | All | Yes | Yes | 2 | Glioblastoma | 5 |
IFI30 expression was related to pathology and prognosis in GBM patients
After comprehensive analysis of the expression pattern of IFI30, we used CGGA database to further study the relationship between the expression of IFI30 and tumor subtype, WHO grade and recurrence status in GBM. First, it could be observed that IFI30 mRNA was up-regulated in MES subtype of primary and recurrent GBM, which was significantly different from CL and PN subtypes (Fig. 3A,C). In both primary and recurrent GBM tissues, IFI30 mRNA expression levels were significantly correlated with WHO grade (Fig. 3B,D). Further, we found a significant correlation between expression level of IFI30 mRNA in GBM and recurrence status. Compared with primary tumors, the expression level of IFI30 mRNA was higher in recurrent tumors (Fig. 3E). These findings were almost consistent with our previous results regarding IFI30 expression.
To investigate the prognostic value of IFI30 in GBM, we applied the TCGA database to analyze the correlation between differentially expressed IFI30 and clinical outcomes. GBM patients with higher IFI30 mRNA expression showed lower OS, worse disease-specific survival (DSS) and PFS compared with those with lower IFI30 mRNA expression level according to Kaplan-Meier survival curve (Fig. 3F-H). Therefore, IFI30 mRNA overexpression was associated with poorer prognosis and may be a valuable predictive biomarker.
From the diagnostic ROC curve, IFI30 mRNA expression could accurately identify tumors from normal tissues (AUC = 0.987) (Fig. 3I). IFI30 time-dependent survival ROC curves were created to predict 1-, 3-, and 5-year survival. AUC showed that IFI30 was suitable for predicting GBM outcomes (Fig. 3J). Subsequently, we integrated clinicopathological factors (including age, gender, and IDH status) and IFI30 mRNA expression levels, and established a nomogram model, which can be used to predict the 1-, 3- and 5-year survival probability of clinical patients (Fig. 3K). Model global statistical test situation: C-index: 0.621 (95%CI 0.594–0.648).
Gene alteration and functional analysis of IFI30 in GBM patients
Genetic mutations in IFI30 in GBM were explored using the TGCA-PanCancer Atlas dataset in cBioPortal (n = 592 GBM patients). The IFI30 gene was altered in five samples (0.8%) (Fig. 4A). We found that IFI30 gene mutation had no significant effect on PFS (p = 0.665) and OS (p = 0.214) of GBM patients (Fig. 4B, C).
IFI30 expression was associated with immune cell infiltration and immune checkpoints
The relationship between IFI30 expression and immune cell infiltration adjusted for purity was investigated using TIMER2.0. The results showed that IFI30 expression level in GBM was positively correlated with CD8+ T cells, CD4+ T cells, Treg, neutrophils, macrophages, cancer-associated fibers, DCs, MDSCs and other immune cells, but negatively correlated with tumor purity (Fig. 5A-C). These results demonstrated that IFI30 was positively associated with immune cell infiltration. Subsequently, we assessed the association of IFI30 with immune checkpoints in the TIMER database. The results suggested that IFI30 in GBM was significantly positively correlated with PD-1, CTLA-4, CD274, and HAVCR2 (Fig. 5D).
IFI30 affects tumor immune microenvironment through antigen presentation
To explore the functions of IFI30 and co-expressed genes, 20 co-expressed genes were obtained using the GEPIA2 database, with PPC values ranging from 0.87 to 0.90. A PPI network of IFI30 was constructed using the Genemania database (Fig. 6A). The top 10 functional partner genes (PCC > 0.89) were selected as highly correlated. These genes were HK3, CTSS, MS4A6A, SIGLEC7, C1QC, TYROBP, FTLP3, LAIR1, CTSL, and SLC7A7. The results showed that CTSS, CTSL and C1QC were highly expressed in antigen processing and presentation (Fig. 6B-C). Subsequently, we performed gene correlation analysis using the TGCA database, which showed that CTSS, CTSL, C1QC, and IFI30 transcript levels were positively correlated (Fig. 6D). GO enrichment analysis included three main functions of biological process, cellular component, and molecular function (Table Ⅳ) (p < 0.05). KEGG analysis mainly included "antigen processing and presentation", "lysosome", "apoptosis". KEGG enrichment items showed that the high expression of IFI30 was mainly associated with Treg development, Toll-like receptor signaling pathway, T cell receptor signaling pathway, PPAR signaling pathway, NOD signaling pathway, NK cell-mediated cytotoxicity, JAK/STAT signaling pathway, Chemokine signaling pathway and antigen processing and presentation. GSEA analysis was performed to identify functional enrichment with high and low expression of IFI30 (Fig. 6E-G). Low expression of IFI30 was associated with disruption of postsynaptic signaling by CNV, synaptic vesicle pathway, GABA receptor signaling, neurotransmitter release cycle, neurofilament and neurogenic proteins, and protein interactions at synapses.(Figure 6H)
Table Ⅳ. GO and KEGG enrichment analyses of IFI30 and functional partner genes in GBM.
ONTOLOGY | ID | Description | pvalue |
BP | GO:0097067 | cellular response to thyroid hormone stimulus | 1.92e-05 |
BP | GO:0043312 | neutrophil degranulation | 2.90e-05 |
BP | GO:0002283 | neutrophil activation involved in immune response | 2.97e-05 |
BP | GO:0042119 | neutrophil activation | 3.22e-05 |
BP | GO:0002446 | neutrophil mediated immunity | 3.24e-05 |
CC | GO:0036019 | endolysosome | 3.50e-05 |
CC | GO:0031904 | endosome lumen | 1.03e-04 |
CC | GO:0062023 | collagen-containing extracellular matrix | 6.64e-04 |
CC | GO:0043202 | lysosomal lumen | 8.09e-04 |
CC | GO:1904813 | ficolin-1-rich granule lumen | 0.001 |
MF | GO:0001968 | fibronectin binding | 6.24e-05 |
MF | GO:0043394 | proteoglycan binding | 1.12e-04 |
MF | GO:0005518 | collagen binding | 3.90e-04 |
MF | GO:0004197 | cysteine-type endopeptidase activity | 0.001 |
MF | GO:0008234 | cysteine-type peptidase activity | 0.003 |
KEGG | hsa04612 | Antigen processing and presentation | 0.001 |
KEGG | hsa04142 | Lysosome | 0.004 |
KEGG | hsa04210 | Apoptosis | 0.004 |
KEGG | hsa04145 | Phagosome | 0.005 |