EMILIN2 Correlated With Its Methylation and Immune Inltration Could Be an Independent Prognostic Biomarker in LGG

Low-grade gliomas (LGGs) are slow-growing brain cancer in central nervous system neoplasms. EMILIN2 is an extracellular matrix (ECM) protein which could inuence the progress of some tumour which is unclear in LGG. In our study, the methylation, expression, prognosis and immune value of EMILIN2 were analysed in LGG through bioinformatics analysis. we rst analysed the LGG data from TCGA and discovered that the EMILIN2 expression, negatively correlated to the EMILIN2 methylation could predict a poor prognosis and associated with different clinical parameters. Moreover, univariate and multivariate Cox regression were performed in CGGA showed that the EMILIN2 could be an independent prognostic biomarker in LGG. Finally, EMILIN2 expression showed a correlation with gene makers in some immune cells which identied the signicance of EMILIN2 in immune inltration. In conclusion, EMILIN2 could act as an independent prognostic biomarker which might be associated with the malignancy and development of gliomas and play a crucial role in glioma in immune inltration.


Introduction
Low-grade glioma (LGG) is a common primary malignant brain tumour that has great intrinsic heterogeneity in tumour biological behaviour [1]. Grade II and III gliomas, which have a median survival of more than 7 years, are collectively termed diffuse lower-grade gliomas [2][3]. DNA methylation, a major epigenetic modi cation that has been extensively studied in various cancers, is related to tumour heterogeneity [4]. An increasing number of studies have focused on the association between speci c genes and their methylation in tumours, but EMILIN2 has not been studied [5][6][7].
EMILIN2 was rst identi ed as a 116 kDa extracellular matrix glycoprotein with ve protein domains: a Cterminal C1q domain, a proline-rich domain, a collagenous domain, a coiled-coil domain, and an Nterminal cysteine-rich domain (EMI domain) [8][9]. A previous study reported that EMILIN2 could act as a key microenvironmental gene affecting vessel formation [10] or as a tumour suppressor gene in breast cancer. [11] However, there are fewer studies about EMILIN2 in glioma. Meanwhile, the clinical and prognostic value of EMILIN2 expression and methylation in glioma especially in LGG are still unknown.
In the present study, we discovered that EMILIN2 expression in WHO grade III LGG was signi cantly higher than that in WHO grade II LGG in the Chinese Glioma Genome Atlas (CGGA). Meanwhile, we identi ed the prognostic role of EMILIN2 for the rst time using LGG data from TCGA and the CGGA, which implied that high expression of EMILIN2 could predict a worse prognosis. Moreover, univariate and multivariate Cox regression analyses showed that EMILIN2 was an independent prognostic factor for LGG in the CGGA samples. Then, the correlation between EMILIN2 DNA methylation and EMILIN2 expression and the prognostic signi cance of EMILIN2 expression were also analysed by using LGG data from TCGA. Additionally, the biological processes of EMILIN2 were analysed by Gene Ontology (GO) and KEGG analysis, which showed that EMILIN2 might have a signi cant relationship to immunity. Furthermore, the Tumor Immune Estimation Resource (TIMER) database was used to evaluate the potential correlation between EMILIN2 expression and immune in ltration levels and its prognostic value in glioma. Finally, the correlation between EMILIN2 and marker genes of immune in ltration was analysed to con rm the vital role of EMILIN2 in immunity.

Gene Ontology and KEGG pathway enrichment analysis
To further study the potential function of EMILIN2, the Gene Ontology (GO) including cellular component (CC), molecular function (MF), and biological process (BP) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway were performed by using the R package with EMILIN2 related genes in CGGA which were identi ed by Pearson test (R > 0.05, p > 0.001). The p-value and q-value was no more than 0.05 as a cut-off point.

Immune database analysis
Tumor Immune Estimation Resource (TIMER) database is a web server for comprehensive analysis of tumor-in ltrating immune cells [16][17]. We evaluated the correlation of EMILIN2 expression with the six kinds of immune cells including CD4 + T cells, CD8 + T cells, dendritic cells, B cells, neutrophils, and macrophages in LGG and the survival value was also analyzed. Moreover, the spearman coe cient was used in the website to analyse the correlation between EMILIN and the gene markers related immune in ltration including markers of CD8 + T cells, T cells (general), B cells, monocytes, tumor-associated macrophage(TAMs), neutrophils, natural killer (NK)cells, M1macrophages, M2 macrophages, dendritic cells (DCs), T-helper 1 (Th1) cells, T-helper 2(vTh2) cells, T-helper 17 (Th17)cells, Tregs, follicular helper T (Tfh) cells, and exhausted T cells which were reported in previous studies. [18][19] Statistical analysis A log-rank test was performed to calculate the difference in overall survival (OS) between high-and lowexpression groups with a median value of the expression or DNA methylation of EMILIN2 as a break point. The expression difference between different clinical features were detected by Wilcoxon text and Kruskal-Wallis text by R (version 4.0). The univariate and multivariate Cox regression were performed to identify the independent prognostic factors. The relationships between the series clinical factors and EMILIN2 expression or its DNA methylation were analysed by chi-square tests. Pearson correlation coe cient was used to measure the EMILIN2 related genes in CGGA (R > 0.05, p > 0.001) and the correlation between the EMILIN2 expression and its DNA methylation in TCGA. A p-value of less than 0.05 was considered statistically signi cant.

Results
The clinical and prognostic value of EMILIN2 expression and methylation in glioma We analysed the expression of EMILIN2 in LGG (n = 528) between WHO grade II (n = 258) and grade III LGG (n = 270) and found that EMILIN2 expression increased markedly with increasing grade level (P < 0.001), as shown in Fig. 1A. To further explore the function of EMILIN2, as shown in Fig. 1B, we observed a obvilously negative correlation (r = − 0.586, P < 0.0001) between EMILIN2 expression and EMILIN2 DNA methylation. Moreover, the distribution of 29 CpG sites related to EMILIN2 DNA methylation is displayed in Fig. 1C. The analyses of Kaplan-Meier curves showed that high expression (Fig. 1D) or low methylation of EMILIN2 (Fig. 1E) could predict poor overall survival (OS) (P < 0.001).
Additionally, Kaplan-Meier analysis was performed to evaluate the prognostic values of these EMILIN2 DNA CpG sites, and as shown in Fig. 2, these 15 different EMILIN2 DNA CpG sites (cg24276681, cg24948962, cg23405696, cg13765206, cg06769774, cg16181848, cg25813942, cg21266975, cg07483811, cg20679955, cg22349573, cg09009111, cg07012770, cg07431339, cg00093099) were signi cantly related to prognosis (P value < 0.05). Next, the patients in TCGA were divided into low or high EMILIN2 expression and EMILIN2 methylation subgroups with the median level as the cut-off value. We applied the chi-square test to detect the detailed relationship of EMILIN2 expression and EMILIN2 methylation with clinical information, which is shown in Table 1. The results showed that EMILIN2 expression was signi cantly related to age (P < 0.001), family history of cancer (P = 0.04), IDH mutation (P = 0.003), and WHO grade (P < 0.001). Meanwhile, the methylation of EMILIN2 was associated with IDH mutation (P < 0.001), grade (P = 0.003) and KPS (P = 0.021). Identi cation of the prognostic and clinical value of EMILIN2 in LGG from CGGA by bioinformatic analysis We downloaded RNA-seq data and clinical information for 1008 patients from CGGA to further verify the prognostic signi cance of EMILIN2. As shown in Fig. 3A-F, the mRNA expression of EMILIN2 was signi cantly associated with clinical information in groups divided by histology (P < 0.001), cancer type (P < 0.001), WHO grade (P < 0.001), MGMT mutation (P = 0.045), 1p19q codeletion (P < 0.001) and IDH1 mutation (P < 0.001). There was no association found to be related to EMILIN2 mRNA expression in age, chemotherapy status, radio treatment, or sex (P > 0.05). These results showed that the expression of EMILIN2 was strongly related to the clinical parameters. In addition, we performed Kaplan-Meier analysis to con rm the prognostic value of EMILIN2, which showed that high EMILIN2 expression predicted poor OS (Fig. 3G).
EMILIN2 could be an independent prognostic biomarker in LGG The above results indicated that the expression of EMILIN2 was closely associated with OS. To further verify the independent prognostic signi cance of EMILIN2, we further analysed the LGG patient data from the CGGA database by univariate and multivariate Cox regression analysis. As shown in the forest diagram ( Fig. 4A-B EMILIN2-related Gene Ontology and signalling pathways in gliomas. To evaluate the function of EMILIN2 in glioma, we performed Pearson's test to identify EMILIN2coexpressed genes in LGG from the CGGA. There were 924 genes that were related to the expression of EMILIN2 with a p-value < 0.01 and R > 0.5 as cut-off points. Then, we performed GO and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses to detect the functions of the EMILIN2-correlated genes. Figure 4C-D showed the top ten most enriched biological process (BP), cellular component (CC), and molecular function (MF) terms and the top thirty most enriched KEGG pathways among the DEGs. The EMILIN2-correlated genes were involved in T cell activation, regulation of T cell activation, neutrophil activation involved in the immune response, growth factor binding, immune receptor activity, human T cell leukaemia virus 1 infection, Th1 and Th2 cell differentiation, and the Th17 cell differentiation signalling pathway. Based on these terms, we found that abnormal expression of EMILIN2 might lead to immune system changes in glioma.
The expression of EMILIN2 was correlated with tumour immune in ltration and related markers in glioma The TIMER database was used to analyse the correlations between EMILIN2 expression and immune cell in ltration. The expression of EMILIN2 was positively correlated with B cell (r = 0.391, P < 0.001), CD8 + T cell (r = 0.275, P < 0.001), CD4 + T cell (r = 0.417, P < 0.001), dendritic cell (r = 0.492, P < 0.001), macrophage (r = 0.51, P < 0.001) and neutrophil (r = 0.435, P < 0.001) in ltration and was negatively correlated with tumour purity (r = -0.279, P < 0.001) in LGG. Moreover, we also explored the prognostic value of EMILIN2 in LGG with different abundances of immune cells. The results showed that highexpressed EMILIN2 was signi cantly associated with the bad prognosis, as shown in Fig. 5, in different abundances of immune cells. The Cox proportional hazard model built in the TIMER database, as shown in Table 2, showed that EMILIN2 expression was an independent prognostic marker of LGG prognosis. Additionally, as shown in Table,3

we analysed the correlations between EMILIN2 expression and many immune markers of various immune cells, including CD8 + T cells, T cells(general), the different functional T cells, such as Th1 cells, Th2 cells, Tfh cells, Th17 cells, and Tregs, as well as exhausted T cells, NK cells, B cells, monocytes, M1 and M2 macrophages, neutrophils, TAMs, and Dendritic cell in LGG.
The result showed that most of the markers were strongly related to EMILIN2 expression, regardless of whether the data were adjusted by tumour purity.

Discussion/conclusion
Currently, an increasing number of studies are focused on using bioinformatics analyses of previously published data to identify factors that play a vital role in various cancers. In our previous study, we found that the expression of EMILIN2 increased with increasing LGG grade. A previous study identi ed that EMILIN2 could play an important role in angiogenesis, platelet activation, thrombus formation, clot retraction and breast cancer [10][11]18]. However, few studies have examined the function of EMILIN2 in glioma, especially in LGG.
In the present study, we rst discovered that EMILIN2 is more highly expressed in WHO grade III compared to grade II LGG in the data from TCGA and the CGGA, which revealed that the altered expression of EMILIN2 might signi cantly regulate the progression of LGG. Similar results were found for EMILIN2 expression in gastric tumours and ovarian serous tumours [19][20][21].
Moreover, we analysed the correlation between EMILIN2 expression and methylation and found that the expression of EMILIN2 was negatively associated with EMILIN2 methylation in LGG. EMILIN2 hypermethylation and low expression also predicted good OS in LGG. Then, the prognostic and clinical value of EMILIN2 was identi ed based LGG from the CGGA database, which was similar to the ndings from the TGGA database in terms of IDH, WHO grades and OS. We further identi ed the prognostic value of EMILIN2 DNA methylation. Almost all sites were signi cantly related to OS, which veri ed the prognostic relevance of EMILIN2. Many studies have indicated that the relationship between speci c DNA methylation sites and the expression of neighbouring genes ranges from weak to moderate, and there are few genes that are regulated by DNA methylation [1,[5][6][7]. In our study, we found that EMILIN2 was obviously and negatively regulated by its DNA methylation. EMILIN2 methylation status and its expression might be a potent biomarker of OS. A similar result was found in breast cancer [22]. Similarly, the prognostic value of EMILIN2 has been reported in lung adenocarcinoma [23].
Furthermore, the LGG database from CGGA was used to recon rm the prognostic and clinical value of EMILIN2 and showed a similar result: the expression of EMILIN2 was remarkably associated with clinical parameters such as histology, WHO grade, IDH1 mutation and even survival status. Moreover, univariate and multivariate Cox regression analyses were performed, and the results showed that EMILIN2 could be an independent prognostic biomarker (p = 0.018, hazard ratio = 1.180 (1.029 − 1.354)); PRS type, WHO grade, and 1p19q_codeletion could also be independent prognostic factors in LGG. Furthermore, we performed GO and KEGG analyses to reveal the function of EMILIN2 in CGGA. The results implied that EMILIN2 might be involved in immune-related biological functions.
A lot of studies have found that tumour-in ltrating immune cells play an important role in the progression and development of cancer [24][25][26][27]. Few studies have pointed out the relationship between EMILIN2 and immune in ltration. Therefore, the TIMER database was used to detect the relationship between immune in ltration and EMILIN2 expression. In our study, our results rst revealed that the levels of immune in ltration were signi cantly correlated with EMILIN2 expression in LGG. Furthermore, we observed that EMILIN2 expression was positively correlated with B cells in LGG. Additionally, EMILIN2 was signi cantly associated with OS at different levels of immune in ltration, which again suggested the prognostic signi cance of EMILIN2 in LGG.
Additionally, the correlation between EMILIN2 expression and the markers of immune cells indicated that EMILIN2 might regulate tumour immunology in LGG. Studies have reported a positive correlation between elevated CD8 + T cells in the tumour microenvironment (TME) and good prognosis in cancer [28]. The gene markers of CD8 + T cells, T cells (general), and B cells such as CD8A, CD8B, CD3D, CD3E, CD2, CD19, CD79A, CD86, and CD115 showed signi cantly positive correlations with EMILIN2 expression, which veri ed the prognostic and immune value of EMILIN2. Macrophages play an important role in maintaining tissue homeostasis and modulating the immune response against pathogens as scavengers [29]. Thus, we detected correlations between biomarkers of M1 and M2 macrophages and tumourassociated macrophages (TAMs), revealing the potential regulatory function of EMILIN2 in macrophages.
In addition, our results shown in the Table 3 indicated that EMILIN2 could activate Tregs and induce T cell exhaustion. Moreover, helper T cells were reported to have the strongest relationship with clinical outcome in early-stage vulvar cancer [30]. Our study showed that there were many signi cant correlations between EMILIN2 expression and various markers of T helper cells (Th1, Th2, Tfh, and Th17) in LGG, which indicated that EMILIN2 might regulate T cell functions and in uence the clinical outcome in LGG.
EMILIN2 expression negatively correlated with that of STAT5B, an important role in the maintenance of normal immune function and homeostasis [31], but positively correlated with the expression of Treg and T cell exhaustion markers (CCR8, FOXP3, TGF-β (TGFB1), PD-1, CTLA4, LAG3, GZMB, and TIM-3) in LGG. These results suggest that EMILIN2 plays a vital role in the regulation of immune in ltrating cells in LGG. However, additional experiments to identify the potential mechanisms of EMILIN2 are urgently needed in the future.
In summary, EMILIN2 expression negatively correlated with EMILIN2 methylation, potentially contributes to the regulation of immune in ltration and is a potential biomarker of OS in LGG.

Declarations
Ethics approval and consent to participate This article didn't need the Ethics approval and consent.

Consent for publication
All authors approved the publication.   The univariate and multivariate Cox regression in LGG from CGGA and the GO and KEGG analysis among EMILIN2 related genes. (A) univariate Cox regression showed the EMILIN2, PRS, Histology, grade, age, IDH-mutation and 1p19q-codeletion were related to OS (P<0.05). (B) multivariate Cox regression in CGGA showed the EMILIN2 could be an independent prognostic marker.