Expression patterns and the prognostic value of the EMILIN/Multimerin family members in low-grade glioma

Oncomine analysis

Oncomine analysis (Rhodes et al., 2004) was used to explore EMILIN/Multimerin expression levels in various cancer types. In the first search, we used the following keywords: “EMILIN1”, “EMILIN2”, “EMILIN3”, “MMRN1” and “MMRN2”, the primary filter for the analysis type was “Cancer vs. Normal Analysis”, and the chosen cancer type was “Brain and CNS Cancers”. We set the following thresholds: p-value = 0.05, fold change “All” and gene rank Top 10%. To compare the expression of EMILIN/Multimerins in LGG subtypes and normal tissue, “EMILIN1” and “EMILIN2” were selected as keywords. “Diffuse Astrocytoma vs. Normal Analysis” was the chosen analysis type, and we set thresholds for p-value = 0.05, fold change “2” and gene rank Top 10%. Additionally, “EMILIN3”, “MMRN1” and “MMRN2” were selected as keywords, and the analysis type and threshold were the same as the previous search (Diffuse Astrocytoma vs. Normal Analysis; p-value = 0.05, fold change “2” and gene rank Top 10%).

UALCAN analysis

Using the UALCAN database (Chandrashekar et al., 2017), we analyzed the expression profiles of EMILIN/Multimerin expression patterns in normal brain tissue and low-grade glioma samples. We established the correlation between expression levels of candidate genes and LGG tumor grades. The unique transcription expression of each tumor grade was compared using student’s t-test, considering p < 0.01 statically significant.

Gene expression profiling interactive analysis

We conducted Gene Expression Profiling Interactive Analysis (GEPIA) using The Cancer Genome Atlas (TCGA) database (Tang et al., 2017) to investigate EMILIN/Multimerin expression levels on overall survival (OS) and disease-free survival (DFS) in patients with LGG. We tested our hypothesis using GEPIA’s log-rank test and included the hazard ratio (HR) and 95% confidence intervals (CI), considering p-values < 0.05 statistically significant.

TCGA and cBioPortal analysis

According to TCGA dataset, we selected 530 pathology reports for further analysis (http://www.cbioportal.org/) (Gao et al., 2013). Kaplan–Meier plots displayed the relationship between EMILIN/Multimerin genetic mutations, OS and DFS in patients with LGG. A log-rank test was performed to identify different survival curves. We used cBioPortal’s online instructions to calculate OncoPrint, cancer type summary and gene co-expression. Additionally, a network of EMILIN/Multimerins and neighboring genes was constructed using GeneMANIA software (www.genemania.org) (Warde-Farley et al., 2010).

TIMER analysis

Gene modules were used in TIMER (http://cistrome.shinyapps.io/timer) (Li et al., 2017) to analyze EMILIN expression levels in different cancer types. We studied the correlation between EMILIN and immune cells, including CD4+ T cells, B cells, CD8+ T cells, neutrophils, macrophages and dendritic cells.

Metascape analysis

Metascape (http://metascape.org) (Tripathi et al., 2015) was used to conduct pathway and enrichment analysis of EMILIN family members and neighboring genes associated with EMILIN alterations. Gene Ontology (GO) terms (biological process, cellular component and molecular function categories) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were used in enrichment analysis. We only considered terms with a p-value < 0.01, minimum count of three and enrichment factor >1.5 to be significant (Chen et al., 2019). The most statistically significant term within a cluster was chosen as the representative. A select subset of enriched terms was rendered as a network plot to further determine the relationship between terms; terms with a similarity >0.3 were connected by edges. Protein–protein interaction (PPI) enrichment analysis was performed using the following databases: BioGrid, InWeb_IM and OmniPath. Additionally, the Molecular Complex Detection (MCODE) algorithm was applied to identify densely connected network components.

Expression analysis of EMILIN/Multimerin family members in patients with LGG

To explore the prognostic and potential therapeutic value of varying EMILIN/Multimerin family members in LGG patients, the gene expression was analysed by ONCOMINE database. The expression of EMILIN/Multimerins in different types of cancers are shown in Fig. 1 and Table 1. In the research of Lee Brain (Lee et al., 2006), Gutmann et al. (2002), TCGA, Sun et al. (2006), French et al. (2005) and Murat et al. (2008), we all found the significantly over-expression of EMILIN1 in Brain and CNS Cancers in parallel to normal tissues. In the datasets of Lee et al. (2006) and Bredel et al. (2005), we observed highly increase in EMILIN2 expression in Brain and CNS Cancers than normal tissues. The results from TCGA dataset and Beroukhim dataset (Beroukhim et al., 2007) indicated that EMILIN3 was significantly up-regulated in Brain and CNS Cancers than in normal tissues. Furthermore, the expression of MMRN1 was shown to be considerably higher in Brain and CNS Cancers than normal tissues in the Beroukhim dataset (Beroukhim et al., 2007). The result from Lee dataset (Lee et al., 2006) showed that MMRN2 was over-expressed in Brain and CNS Cancers, respectively.

In contrast with normal tissues, Beroukhim dataset (Beroukhim et al., 2007) and Bredel research (Bredel et al., 2005) was founded that significantly lower expressions of EMILIN3 in Brain and CNS Cancers. And Beroukhim dataset (Beroukhim et al., 2007) and TCGA dataset showed that there were lower in MMRN2 expression in Brain and CNS Cancers compared to normal tissues.

Multiple datasets showed significantly higher expressions of EMILIN/Multimerin family members in Brain and CNS Cancers. The expression levels of EMILIN/Multimerins in the Diffuse Astrocytoma and Anaplastic Astrocytoma which were the subtypes of LGG.

Expression levels of EMILIN/Multimerins in diffuse astrocytoma and anaplastic astrocytoma subtypes of LGG

Oncomine database analysis explored the expression of EMILIN/Multimerins in LGG subtypes. We found that EMILIN1 and EMILIN2 were overexpressed in diffuse astrocytoma compared to normal brain tissue. Additionally, there were higher expression levels of EMILIN3 and MMRN1 in anaplastic astrocytoma compared to normal brain tissue. However, the expression of MMRN2 was not markedly higher in anaplastic astrocytoma compared to normal brain tissue (Figs. 2A–2E). Some datasets showed higher expression levels of EMILIN1, EMILIN2, EMILIN3 and MMRN1 in LGG subtypes.

Association between the expression of EMILIN/Multimerins and tumor grades in patients with LGG

We used UALCAN to analyze the expression of EMILIN/Multimerins in LGG tumors and demonstrate the value of predicting survival rates. There was a high correlation between the expression of EMILIN/Multimerin family members and patient tumor grades (Figs. 3A–3E); patients with a more advanced tumor grade had a higher expression of EMILIN/Multimerins. These results suggest that the expression of EMILIN/Multimerin family members was significantly associated with tumor grades in patients with LGG.

Association between the expression of EMILIN/Multimerins and prognosis of patients with LGG

We used GEPIA to analyze the role of EMILIN/Multimerin expression levels in predicting the prognosis of LGG. The expression of most EMILIN/Multimerin family members was significantly associated with LGG patient prognosis (Figs. 4 and 5). There was a significant association between the lower expression of EMILIN/Multimerins and longer OS in patients with LGG (Figs. 4A–4E). A lower expression of EMILIN1, EMILIN2, EMILIN3 and MMRN1 was highly correlated with DFS (Figs. 5A–5E). However, MMRN2 expression levels were not correlated with DFS in patients with LGG.

Higher EMILIN/Multimerin expression levels were observed in poor LGG prognosis. We used UALCAN to demonstrate any significant associations between EMILIN/Multimerin expression levels and a patient’s LGG tumor grade (Figs. 6A–6E). This association may be used as a biomarker for predicting LGG patient survival.

The relationship between EMILIN/Multimerin expression levels and immune infiltration levels in LGG

Immunity is closely related to the occurrence and development of tumors. TIMER analysis was applied to the relationship between EMILIN/Multimerin expression levels and immune infiltration levels in LGG. EMILIN/Multimerin family members in LGG infiltrated similar immune cell profiles, showing a clear link to neutrophil, CD8+ T cell, dendritic cell, macrophage B cell and CD4+ T-cell populations (Fig. 7; Table 2). These findings suggest a correlation between EMILIN/Multimerin expression levels and immune infiltration levels in LGG.

Exploring genetic alterations and neighboring genes of EMILIN/Multimerins in patients with LGG

The alteration frequency of EMILIN/Multimerin mutations in LGG was analyzed using cBioPortal and the results indicated five categories based on filtering (Fig. 8A). The ratios of genetic alterations in EMILIN/Multimerins range from 0.4% to 1.4% for each member (EMILIN1 0.4%, EMILIN2 1.4%, EMILIN3 0.4%, MMRN1 0.4% and MMRN2 1.4%; Fig. 8B). Additionally, we analyzed genetic alterations in EMILIN/Multimerins and their association with OS and DFS in patients with LGG. There was no discernable difference in OS and DFS cases with or without EMILIN/Multimerin alterations (Figs. 8C and 8D). The fifty most frequently altered neighboring genes were used to construct a network using cBioportal. These findings indicated that A2M, ACTN1, ACTN2, ALB and ALDOA were closely associated with EMILIN/Multimerin alterations and functions (Fig. 8E).

Co-expression and interaction analyses of EMILIN/Multimerins in patients with LGG

We performed bioinformatics analysis to explore the co-expression and interaction of EMILIN/Multimerins in patients with LGG. EMILIN1, EMILIN2 and EMILIN3 were positively correlated with each other. However, there was a negative relationship between EMILIN3 and MMRN2 (Fig. 9A). Using GeneMANIA, we constructed an EMILIN/Multimerin network using the structure and function of neighboring genes. The results revealed twenty genes, including F5, TGFBI, EGFL7 and IL33, that were significantly associated with EMILIN/Multimerin family members (Fig. 9B).

Functional enrichment analysis of EMILIN/Multimerins in patients with LGG

We used Metascape to analyze the functions of EMILIN/Multimerins and their neighboring genes. We performed pathway and process enrichment analysis on fifty neighboring genes (Fig. 10A; Table 3). Additionally, we constructed a network of enriched terms; a lower p-value was associated with terms containing more genes (Fig. 10B). We constructed a PPI network and MCODE components were identified in the gene list (Fig. 10C). We analyzed the functional enrichment of the three most significant MCODE components (Fig. 10D).

Regulating LGG using cBioportal and co-expression analysis of EMILIN/Multimerins

Using cBioportal, we performed co-expression analysis to predict the biological functions and potential mechanisms of EMILIN/Multimerins. We selected five co-expressed genes (PHTA2, CLCL1, FSTL1, IQGAP1 and MYL12A) from the resulting analysis that were significantly associated with EMILIN/Multimerins. GEPIA’s Kaplan–Meier curve was used to evaluate the function of five co-expressed genes and their impact on the survival of patients with LGG. Each gene was significantly related to the prognosis of LGG (Figs. 11–13).