IFI30 expression is an independent unfavourable prognostic factor in glioma

Abstract Gamma‐interferon‐inducible lysosomal thiol reductase, the only known lysosomal thiol reductase, is encoded by gene IFI30 and expressed constitutively in antigen‐presenting cells. Our comprehensive study on IFI30 in gliomas found its expression to be high in glioblastomas and in gliomas with a mesenchymal subtype or wild‐type isocitrate dehydrogenase, all of which indicated the malignancy and poor outcomes of gliomas. Kaplan‐Meier survival analysis ascertained that high IFI30 expression conferred poor outcomes. The IFI30 expression levels also showed high efficiency in predicting 1‐, 3‐ and 5‐year overall survival. Univariable and multivariable Cox regression analyses were performed to define IFI30 as an independent prognostic marker. Biological process analysis suggested that IFI30 was involved in immune responses. ESTIMATE and CIBERSORT were applied to evaluate immune cell infiltration, with results indicating that samples with higher IFI30 expression had higher infiltration of immune cells, including regulatory T cells and M0 macrophages. Correlation analysis showed that IFI30 was significantly positively correlated with immune checkpoints that suppress effective antitumour immune responses. Immunohistochemical staining was also performed to confirm the association between IFI30 expression and the immune phenotype. The suggested correlation between high IFI30 expression and an immunosuppressive phenotype contributes to our knowledge about the glioma microenvironment and might provide clues for the development of novel therapeutic targets.

(which differentiate tumour cells from normal cells) are captured and processed by dendritic cells. Second, the captured antigens are presented to T cells through major histocompatibility complex (MHC) I and II molecules, resulting in T cell priming and activation.
Finally, the activated T cells migrate to and infiltrate the tumour where they recognize and kill the tumour cells. [5][6][7] However, tumours have adopted multiple strategies to attenuate the attack of the immune system, ranging from the down-regulation of immunogenic antigens and antigen-presenting cells (APCs), to the prevention of T cell infiltration via vasculature barrier and through the suppression of effector T cells. 8,9 A deeper understanding of the mechanisms by which tumours escape immune attacks would facilitate innovative therapeutic strategies. In addition to immune checkpoint inhibitors and chimeric antigen receptor T cell therapy, a growing number of immunotherapeutics are under clinical investigation for their safety and efficacy, including peptide vaccines, dendritic cell vaccines, and therapies targeting chemokines in the tumour microenvironment. [10][11][12][13][14] Gliomas are the most common type of primary malignant tumours of the central nervous system (CNS). With the substantial advance in the 2016 World Health Organization (WHO) classification of tumours of the CNS, the refinement of glioma classification on the basis of molecular biomarkers has led to a massive increase in the development of targeted therapies against this disease. [15][16][17][18] At the same time, persistent efforts have been made to reveal the immune characteristics of gliomas for the design of novel immunotherapeutic approaches. Immune cells, such as microglia, peripheral macrophages, leukocytes, and myeloid-derived suppressor cells, are reported to infiltrate gliomas wherein they create an immunosuppressive microenvironment to facilitate tumour cell growth and invasion and to dampen the efficacy of immunotherapy. [19][20][21] For example, tumour-associated macrophages (TAMs) secrete cytokines such as transforming growth factor-beta (TGF-β) and interleukin-10 (IL-10) to inhibit effector T cells. TAMs express ligand receptors for programmed cell death protein 1 (PD-1) and cytotoxic T-lymphocyte antigen 4 (CTLA-4) to suppress the cytotoxic functions of T cells. [22][23][24] Hence, to achieve a breakthrough in glioma immunotherapy, a detailed understanding of the specific immune system is needed.
It was reported in a recent study that gamma-interferon-inducible lysosomal thiol reductase (GILT) is overexpressed in gliomas, and knockdown of the enzyme suppresses glioma cell proliferation through the promotion of apoptosis and induction of cell cycle arrest. 25 GILT, encoded by the IFI30 gene, is the only known lysosomal thiol reductase. It is constitutively expressed in APCs, including dendritic cells, monocytes/macrophages, and B cells.
Interferon-gamma can induce the expression of GILT in other cell types, such as melanoma cell lines. GILT catalyses disulfide bond reduction and enhances the MHC II-restricted presentation of a subset of epitopes. [26][27][28] GILT-free mice were reported to be defective in antigen processing. 29 In this study, we systematically analysed IFI30 in 921 glioma samples sourced from the Chinese Glioma Genome Atlas (CGGA) and The Cancer Genome Atlas (TCGA) databases, including its expression in different tumour grades and subtypes, potential biological functions, and prognostic significance. We also evaluated the correlation between IFI30 expression and some important immune-related molecules. Immunostaining was performed to confirm the expression pattern of IFI30 and its correlation with immune-related factors. The results suggested that IFI30 was a novel independent prognostic factor with immune-related functions.

| Patients and data collection
In total, 310 CGGA-sourced samples (105 grade II, 67 grade III and 138 grade IV gliomas, http://www.cgga.org.cn) and 611 TCGAsourced samples(214 grade II, 237 grade III, and 160 grade IV gliomas, http://cance rgeno me.nih.gov/) were studied. All analyses were performed using the CGGA data set and then validated with the cohort from TCGA. The clinical and molecular features of the patients are given in Table S1.

| Statistical analysis
Student's t test was used to determine differences between two groups. Kaplan-Meier survival analysis and the log-rank test were performed to assess the significance of IFI30 expression to sur-

| IFI30 expression was up-regulated in glioblastomas, and in gliomas with wild-type isocitrate dehydrogenase and mesenchymal subtype
To clarify the characteristics of IFI30 in gliomas, we first analysed its expression level in the CGGA and TCGA data sets stratified according to the tumour grade, isocitrate dehydrogenase (IDH) mutation status, and 1p/19q codeletion status. As shown in Figure 1A,D, IFI30 expression was significantly increased along with the grade of the tumour and was the highest in glioblastomas (GBM, glioma grade IV). In the gliomas that were stratified on the basis of the IDH and 1p/19q status, the IFI30 expression level was the highest in the IDH wild-type (IDH-wt) group, both in the lower-grade gliomas (LGG, grade II and III) and GBM ( Figure 1B,C for CGGA, Figure 1E,F for TCGA). The IFI30 expression level in different grades of gliomas was also confirmed by immunohistochemical staining ( Figure 1G). These results indicated that a high level of IFI30 expression implied malignant progression of the glioma. [30][31][32] Next, we explored the distribution of IFI30 in four TCGA molecular subtypes. In both the CGGA and TCGA data sets, the mesenchymal subtype showed the strongest expression of IFI30, followed by the classical, proneural, and neural subtypes successively ( Figure 2A,C). The gliomas were next divided into mesenchymal and non-mesenchymal subtype groups for ROC curve analysis, which showed the high efficiency of the IFI30 expression level in predicting the mesenchymal subtype. This was further suggested by the area under the ROC curve (AUC), which was 0.958 and 0.945 for the CGGA and TCGA cohort, respectively( Figure 2B,D). All the results demonstrated the association between IFI30 expression and the malignant phenotype of gliomas. LGG with IDH-wt, a higher level of IFI30 expression was associated with a shorter OS in both the CGGA ( Figure S1C) and TCGA ( Figure   S2C) cohorts. In addition, the IFI30 expression level could distinguish the prognosis of patients with LGG with a mutant IDH (IDHmut) and non-codeleted 1p/19q (1p/19q non-codel) in the data set from the CGGA (Figure S1A), and patients with GBM with IDH-wt in the data set from TCGA ( Figure S2E). Next, we compared the specificity and sensitivity of IFI30 expression, the patient age at Furthermore, we conducted univariable and multivariable analyses to confirm whether IFI30 expression was an independent prognostic biomarker for gliomas. As shown in Figure 4, IFI30 expression was still significantly associated with patient survival in the CGGA data set after multivariable Cox regression analysis, a characteristic that was validated with the cohort from TCGA ( Figure S4).

| IFI30 expression was associated with immunerelated functions
To determine the biological function of IFI30 in gliomas, Pearson's correlation analysis was carried out to evaluate genes that are strongly correlated with IFI30 (|R| ≥ 0.5, P < .01). The total numbers of positively related genes in the CGGA and TCGA data sets were 1485 and 2440, respectively. The two positively related gene sets were separately subjected to functional annotation analysis with DAVID, whereupon the genes were found to be involved mainly in the immune response, antigen processing and presentation, chemotaxis, extracellular matrix organization, and angiogenesis ( Figure 5 for CGGA, Figure S5 for TCGA). Similarly, 1025 and 1633 negatively related genes in the CGGA and TCGA data sets, respectively, were screened and found to be annotated mainly to normal biological processes, such as chemical synaptic transmission, neurotransmitter secretion, and nervous system development ( Figure   S6 for CGGA, Figure S7 for TCGA). These results suggested that

| IFI30 was associated with immunosuppressive phenotype in gliomas
Given that IFI30 participates in MHC II-associated antigen processing, the discovery that its high expression implied poor outcomes for patients with gliomas prompted us to explore the relationship between IFI30 expression and immune cell infiltration by applying the ESTIMATE algorithm. 33 The immune scores increased with increase in the IFI30 expression levels in both the CGGA and TCGA data sets ( Figure 6A). Next, CIBERSORT was used to evaluate the abundance of various immune cell types in the CGGA and TCGA samples. 34 As illustrated in Figure 6B 35,36 To confirm the association between IFI30 expression and the immune phenotype, immunohistochemical staining of CD163, PD-L2, and IL-10 was carried out on the glioma samples with low and high IFI30 expression levels. 22,23 As shown in Figure 7C, the samples with high IFI30 expression had higher levels of CD163, PD-L2, and IL-10 staining, which indicated an immunosuppressive microenvironment. These results might partially explain the poor outcome of patients with gliomas with a high level of IFI30 expression.

| D ISCUSS I ON
According to the WHO classification of tumours of the CNS, gliomas-the most common primary intracranial tumours-can be pression have poor disease-free survival. 41 The association of low GILT expression with poor survival has also been validated in patients with DLBCL. 42 These reports might seem contradictory to the results in gliomas, where high IFI30 expression indicated poor OS, but this may be due to the specific immune microenvironment of gliomas. Our results showed that gliomas with higher IFI30 expression were more infiltrated by M0 macrophages and Tregs, whereas their infiltration by CD8 T cells did not increase correspondingly. A recent study demonstrated that GBM-associated myeloid cells resembled the M0 macrophage phenotype, which is consistent with our results. 43 Undifferentiated M0 macrophages can polarize into classically activated macrophages (M1) with the pro-inflammatory/ antitumoural phenotype and can also polarize into alternatively activated macrophages (M2) with the immune-suppressive/protumoural phenotype. 44 Gliomas have long been reported to be infiltrated by macrophages. 45 Glioma cells release several factors, such as colony-stimulating factor 1 (CSF-1), glial-derived neurotrophic factor and granulocyte-macrophage colony-stimulating factor, to attract TAMs to the tumour site. [46][47][48] However, TAMs secrete a wide array of cytokines, including epidermal growth factor and TGF-β, to promote glioma migration and invasion. 46,49 Hence, the reduction of M2 macrophage polarization or the promotion of M1 macrophage polarization is a promising therapy in cancer treatment, and studies on this have been conducted in gliomas. 50 For example, M1-like macrophages combined with immune checkpoint antibodies could eradicate GBM in mouse models. 51 CSF-1R inhibition reduced M2 macrophage polarization and regressed established gliomas. 52 Chlorogenic acid repolarized macrophages from the M2 to the M1 phenotype and reduced glioma growth. 53 All these research reports underscore the possibilities and potentials of targeting TAMs in glioma treatment, and our results imply that patients with high IFI30 expression might benefit most from such therapy.
In summary, we have conducted a comprehensive research study on IFI30 expression in gliomas and ascertained through bioinformatic profiling that this gene would be an unfavourable prognostic predictor of this disease. The underlying molecular mechanisms involved and more applications of our findings in clinical practice should be further explored.

ACK N OWLED G EM ENTS
This work was supported by grants from the National Natural Science Foundation of China (81972338); Beijing Municipal Science & Technology Commission (Z181100001718127).

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

AUTH O R S CO NTR I B UTI O N
XL analysed the data and wrote the paper, CS wrote the paper, SY and QJ contributed to the statistical analyses, FC revised the paper, and WL approved the submitted and final version.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available in the CGGA and TCGA repositories.