Vimentin promotes glioma progression and maintains glioma cell resistance to oxidative phosphorylation inhibition

Abstract

Purpose

Glioma has been demonstrated as one of the most malignant intracranial tumors and currently there is no effective treatment. Based on our previous RNA-sequencing data for oxidative phosphorylation (OXPHOS)-inhibition resistant and OXPHOS-inhibition sensitive cancer cells, we found that vimentin (VIM) is highly expressed in the OXPHOS-inhibition resistant cancer cells, especially in glioma cancer cells. Further study of VIM in the literature indicates that it plays important roles in cancer progression, immunotherapy suppression, cancer stemness and drug resistance. However, its role in glioma remains elusive. This study aims to decipher the role of VIM in glioma, especially its role in OXPHOS-inhibition sensitivity, which may provide a promising therapeutic target for glioma treatment.

Methods

The expression of VIM in glioma and the normal tissue has been obtained from The Cancer Genome Atlas (TCGA) database, and further validated in Human Protein Atlas (HPA) and Chinese Glioma Genome Atlas (CGGA). And the single-cell sequencing data was obtained from TISCH2. The immune infiltration was calculated via Tumor Immune Estimation Resource (TIMER), Estimation of Stromal and Immune Cells in Malignant Tumors using Expression Data (ESTIMATE) and ssGSEA, and the Immunophenoscore (IPS) was calculated via R package. The differentiated expressed genes were analyzed including GO/KEGG and Gene Set Enrichment Analysis (GSEA) between the VIM-high and -low groups. The methylation of VIM was checked at the EWAS and Methsurv. The correlation between VIM expression and cancer stemness was obtained from SangerBox. We also employed DepMap data and verified the role of VIM by knocking down it in VIM-high glioma cell and over-expressing it in VIM-low glioma cells to check the cell viability.

Results

Vim is highly expressed in the glioma patients compared to normal samples and its high expression negatively correlates with patients’ survival. The DNA methylation in VIM promoters in glioma patients is lower than that in the normal samples. High VIM expression positively correlates with the immune infiltration and tumor progression. Furthermore, Vim is expressed high in the OXPHOS-inhibition glioma cancer cells and low in the OXPHOS-inhibition sensitive ones and its expression maintains the OXPHOS-inhibition resistance.

Conclusions

In conclusion, we comprehensively deciphered the role of VIM in the progression of glioma and its clinical outcomes. Thus provide new insights into targeting VIM in glioma cancer immunotherapy in combination with the current treatment.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s13402-023-00844-3.

Introduction

Glioma is the most aggressive and intrinsic tumor in the central nervous system. WHO termed gliomas into four categories from grade I to grade IV. Grade I and II were treated as low-grade glioma (LGG) and grade III and IV as high-grade glioma (HGG). Grades II, III, and IV were marked as malignancy grades. Among these, grade IV is the most invasive and called glioblastoma multiforme (GBM) [1]. Besides classification by grade, diffuse glioma can also be classified as astrocytomas and oligodendrogliomas based on histopathological analysis [2]. The traditional treatment of glioma including operations and followed by chemotherapy and radiotherapy or the combination, and the advanced treatment immunotherapy is under investigation. To find the prognostic biomarker and the corresponding suitable treatment is necessary for glioma treatment.

Targeting cancer cells’ mitochondria oxidative phosphorylation (OXPHOS) has emerged as a novel cancer therapy [3–6]. From our previous RNA-sequencing for two groups of cells: the OXPHOS-inhibition resistant cells and the OXPHOS inhibition-sensitive cells [7], we found that Vimentin (VIM) is highly expressed in the OXPHOS-inhibition resistant cell lines especially in the glioma cancer cells. This arouses our interest to study this gene. VIM encodes the intermediate filament protein and plays important roles in cell motility [8], cytoplasmic organization, cell development and tissue homeostasis under both physiological and pathological conditions [9]. It is dispensable in the process of epithelial-mesenchymal transition (EMT) as it regulates the plasticity of the mesenchymal cell. VIM is one of the most typical mesenchymal markers. This unique role in EMT paved the way for its function in cancer cell migration. VIM interacts with oncogenic signaling pathways such as the AKT pathway to promote cancer cell migration [10, 11]. The role of VIM in promoting cancer progression has been verified in breast cancer [11], colon cancer [12], bladder cancer [13], etc. But its role in glioma cancer is elusive.

Besides promoting cancer cell migration, VIM is also reported to play an important role in drug resistance. In colon cancer, it promotes the cancer cell resistance to butyrate and other histone deacetylase inhibitors [14], while in ovarian cancer, the downregulation of VIM increased cancer cells’ resistance to cisplatin [15]. VIM contributes to the chemotherapy or radiotherapy resistance of breast cancer by increasing cancer invasion [16]. The function of VIM in gliomas hasn’t been explored.

In this study, we will check the expression of VIM in glioma tissue and normal tissue, its relations with glioma patients’ survival, and parse out the potential working mechanisms from the different perspectives, thus, deciphering the diversified roles of VIM in the progression of gliomas.

Materials and methods

Data collection

The expression of VIM of glioma and normal tissues was obtained from The Cancer Genome Atlas (TCGA) https://portal.gdc.cancer.gov/, and verified from the Human Protein Atlas (HPA, http://www.proteinatlas.org/), and the Chinese Glioma Genome Atlas (CGGA) http://www.cgga.org.cn/.

Survival analysis

Kaplan–Meier (KM) survival curves were checked on KM website (https://kmplot.com/analysis/).

Univariate and multivariate logistic regression analysis

We used the univariate Cox regression analysis and the multivariate logistic regression analysis to check the relation between the VIM expression and the glioma patients’ survival.

VIM expression in scRNA-seq

Tumor Immune Single-cell Hub 2 (TISCH2) [17] is a scRNA-seq database, which aims to characterize tumor microenvironments at single-cell resolution. TISCH2 (http://tisch.comp-genomics.org) has collected 187 sets of high-quality tumor single cell transcriptome data and corresponding patient information from GEO and ArrayExpress [17]. We applied the GSE131928, GSE162631, GSE84465 and GSE89567 in this study.

Identification and enrichment analysis of DEGs in VIM-high and-low groups

Distinct VIM subtype-related differentially expressed genes (DEGs) were identified using the “limma” package in R (adj. p < 0.05 and |log₂Fold Change|> 2) [18]. The functional and enrichment pathways (GO and KEGG) of DEGs were further explored using the “ cluster profiler” package in R [19].

Function and pathway analysis by Gene Set Enrichment Analysis (GESA)

DEGs between VIM-high and low groups were identified by using the DESeq 2 R package. GSEA was performed to explore the significant functions and pathways between the two groups via the ggplot2 R package. A significant difference has been set via the below criterion: adjusted p-value (adj. p) < 0.05, normalized enrichment score(|NES|) > 1.5, and false discovery rate (FDR) < 0.05.

The combination analysis of GO /KEGG and LogFC

We get the relations between the expression of VIM and the biological process (BP), cellular component (CC) and molecular function (MF) via GO analysis, then we conduct KEGG pathway analysis. On the basis analysis of GO and KEGG analysis, with the data of log₂Fold Change (LogFC), we calculated the zscore of each item.

$$\mathrm{zscore}=\frac{\left(\mathrm{Up},-,\mathrm{Down}\right)}{\sqrt{\mathrm{Counts}}}$$

Immune cells infiltration analysis

We checked the relationship between the VIM and the quantity of infiltrating immune cells (CD4⁺ T cells, CD8⁺ T cells, B cells, neutrophils, macrophages, and dendritic cells) in glioma patients respectively by The Tumor Immune Estimation Resource (TIMER) database (https://cistrome.shinyapps.io/timer/).

The correlation between VIM expression and the infiltration of immune cells

The tumor immune infiltration analysis with more immune cells (total 22) was analyzed by ssGSEA by GSVA R package. Original data is from TCGA (https://portal.gdc.cancer.gov/).

Estimation of Stromal and Immune Cells in Malignant Tumors using Expression Data (ESTIMATE)

The Estimation of Stromal and Immune Cells in Malignant Tumors using Expression Data (ESTIMATE) is a package which uses gene expression data to predict the content of interstitial cells and immune cells in malignant tumor tissues [20]. Based on the enrichment analysis of a single sample gene set (ssGSEA), the algorithm generates three scores: stromal score (recording the presence of stroma in tumor tissue), immune score (representing the infiltration of immune cells in tumor tissue), estimated score (inferring tumor purity). The Stromal score, Immune Score and Estimate Score of VIM can be obtained at http://www.sangerbox.com/.

Immunophenoscore (IPS) analysis

The IPS infiltration scores were checked from the MHC, EC, CP, SC, AZ for each patient using IOBR package in R software [21, 22] (https://github.com/IOBR/IOBR). The correlation between the expression of VIM and these different immune cells infiltration scores was calculated by Spearman’s coefficient coefficient.

The correlation between the expression of VIM and the immune checkpoint

The pan-cancer data were downloaded from UCSC(https://xenabrowser.net/) including the TCGA TARGET GTEx, then we checked the expression of Pearson correlation between the expression of ENSG00000026025 (VIM) and 60 immune checkpoints including 24 inhibitory and 36 stimulatory immune checkpoints [23].

The DNA methylation of VIM

The DNA methylation of VIM in glioblastoma for 659 samples from TCGA was checked from the website: https://mexpress.be/. The cluster analysis of VIM DNA methylation for low grade glioma and glioblastoma multiforme was obtained from below website: http://bio-bigdata.hrbmu.edu.cn/diseasemeth/.

The correlation between the expression of VIM and its methylation

Epigenome-Wide Association Study (EWAS) has been considered as a novel strategy to study DNA methylation level of different phenotypes [24]. The methylation levels of VIM gene in gliomas and the methylation changing with the expression of VIM were explored at EWAS by accessing https://ngdc.cncb.ac.cn/ewas/datahub/. All the relations between methylation of VIM and patients survival were checked from http://www.cgga.org.cn/.

The correlation between the methylation of VIM and patients’ survival

MethSurv (https://biit.cs.ut.ee/methsurv/) is a website for survival analysis based on the CpG methylation model. It uses 7358 methylation data from 25 different human cancers from the TCGA database and uses Cox proportional risk model to develop an interactive network for survival analysis. We checked the correlation between the DNA methylation of VIM and patients’ survival from this website.

Relationship between VIM gene expression and cancer stemness

We checked the Pearson correlation between the expression of VIM and cancer stemness in the pan-cancer atlas by accessing http://sangerbox.com. The stemness was calculated based on the RNA-based stemness scores derived by the stemness group, the DNA methylation-based stemness scores derived by the stemness group and other stemness probes (219 probes).

The expression of VIM in scRNA-seq

Tumor Immune Single-cell Hub 2 (TISCH2) is a scRNA-seq database, which aims to characterize tumor microenvironments at single-cell resolution. TISCH2 (http://tisch.comp-genomics.org) has collected 187 sets of high-quality tumor single cell transcriptome data and corresponding patient information from GEO and ArrayExpress [17]. The data covers 50 cancer types, including 6 million cells from more than 1500 patients. Among them, 40 sets of TISCH2 data are single cell transcriptome data under different treatment conditions, including immunotherapy, chemotherapy, targeted therapy and combination therapy.

The expression of VIM in different cancer cell lines

The expression of VIM in different cancer cell lines which we checked the drug resistance is acquired from Cancer Cell Line Encyclopedia (CCLE) (https://sites.broadinstitute.org/ccle).

The CRISPR result of VIM-knockout in different Glioma cell lines

The cell viability of different glioma cell lines by knocking out VIM was downloaded from the Achilles DepMap datasets (https://depmap.org/portal/). VIM knocking down (RNAi) or knocking out (CRISPR) results were checked in DEMETR2 and CERES, respectively. The scores evaluate the effect of knocking down or knocking out VIM while normalizing expression against the distribution of pan-essential and nonessential genes. Positive scores (> 0) indicate that the cell line would grow faster while the negative scores indicate that the cell line would grow slower after experimental manipulation.

Cell viability of the Oligomycin A treatment of glioma cells

The cell viability of different glioma cell lines treated by OXPHOS inhibitor Oligomycin A was obtained from DepMap database https://depmap.org/portal/.

Cell culture

G-401, NCI-H82, SW48, MDA-MB-453 were bought from ATCC, and WSU-DLCL2 was bought from Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ), SF-126 cell was bought from JCRB Cell Bank. 786-O, NCI-H82, WSU-DLCL2, G-401 were cultured in 1640 medium adding 10% FBS, 1% Pen/Strep. SF-126, SW48, and HEK293T were cultured in DMEM medium adding 10% FBS and 1% Pen/Strep. CFPAC-1 was cultured in IMDM medium adding 10% FBS, 1% Pen/Strep. MDA-MB-453 was cultured in L-15 medium adding 10% FBS, 1% Pen/Strep. MDA-MB-453 was maintained in 100% air without CO₂ at 37 °C and other cells were in 5% CO₂ at 37 °C. G98 glioma cancer cell line were cultured with DMEM F12 with adding HHNG, FGF and EGF under 5% CO₂ at 37 °C.

Cell viability assay

To check the cell viability of OXPHOS-inhibition resistant or sensitive cell lines with knocking down VIM or overexpressing VIM, 1000 cells per well were seeded in a 96-well plate and incubated with or without Gboxin for 72 h. Cell viability was tested by the CellTilter Glo Luminescent Cell Viability Assay (Promega, G7572).

Preparation of RNA-Seq library

Total RNA was extracted from 1 × 10⁶ cells for each sample. RNA integrity number (RIN) higher than 7.0 was used as the criteria for library preparation. TruSeq Stranded Total RNA with Ribo-Zero Gold kit (Illumina, 20020598) was used to prepare the sample; PCR was used to generate the library. Agilent Bioanalyzer 2200, 150 bp paired-end sequencing was conducted on an Illumina Noaseq 6000 sequencer for the following quantification.

RT-qPCR and primer design

Total RNA was extracted from SF-126 shvector, SF-126 shVIM, G98 oevector and G98 oeVIM. Briefly, cells were harvested from 6 cm plates and lysed by Trizol, total RNA was isolated. Then total RNA was reverse transcribed into cDNA by using TAKARA Kit (RR047A). The further quantitative PCR was conducted with ChamQ SYBR Green Master Mix kit ( Vazyme, Q411-V23.1). Tubulin was used to normalize VIM genes.

The primers of VIM are as below:

• Forward Primer: GACGCCATCAACACCGAGTT

• Reverse Primer: CTTTGTCGTTGGTTAGCTGGT

Knocking down of VIM (shVIM) in SF-126 glioma cancer cell line

To generate the stable knockdown of VIM in SF-126 cell lines ( the OXPHOS-inhibition resistant cell line), two specific shRNA were designed. The shRNA was integrated into pLVX-shRNA Vector contained GF. SF-126 cells were then transinfected by the lentivirus containing shVector or shVIM, and the knockdown verification and ATP assay were done after two passages.

The two shRNA of VIM are as below:

ShRNA-1:

• Forward: 5’GATCCGCTAACTACCAAGACACTATTTTCAAGAGAAATAGTGTCTTGGTAGTTAGCTTTTTG 3’

• Reverse: 5’AATTCAAAAAGCTAACTACCAAGACACTATTTCTCTTGAAAATAGTGTCTTGGTAGTTAGCG 3’

Overexpression of VIM (oeVIM) in G98 glioma cancer cell line

To generate the stable VIM-overexpression in the OXPHOS-inhibition sensitive glioma cancer cell line, we purified the cDNA of VIM from SF-126 and over-expressed it into G98 glioma cancer cell line ( the OXPHOS-inhibition sensitive cell line). First, total RNA was isolated from SF-126 and reversely transcribed into cDNA, then primers were designed as below to amplify VIM gene, then the amplified cDNA was inserted into Plvx-pCMV-pPGK-Puro vector (Vazyme, Clonexpress Ultra One Step Cloning Kit V2, C116-01). Lentivirus were generated by the transfection of pMD2.G, psPAX2 and oeVIM with the proportion 2:4:8 into 293 T cells. Viruses were transfected into G98 and screened by Puromycin under 1 μM. Then the overexpression and ATP assay were conducted after two times passages.

The overexpression primers are as below:

• Forward: 5'- tcgagctcaagcttcgaattcATGTCCACCAGGTCCGTGTCC -3'

• Reverse: 5'- ttatctagagtcgcgggatccTTATTCAAGGTCATCGTGATGCTG -3'

Western blots

SF126 knocking down control cells (shVector) and two shVIM cells (shVIM-1, shVIM-2) were harvested each from 6 cm dish, then cells were lysed, protein liquid was boiled 5 min at 100 degree centigrade, then centrifuged for 5 min, the supernatant were loaded in gel and and SDS-page was conducted, then gel was transferred to nitrocellulose membrane, 5% milk blocked 1 h, then it was incubated by anti-VIM antibody (PTM Bio, PTM-5376) and anti-beta action antibody (Proteintech, Cat No. 81115–1-RR) overnight, then the membrane was washed 3 times each at 5 min, Anti-Rabbit IgG, HRP linked secondary antibody were incubated for 1 h, then the membrane were washed 3 times by PBST each for 15 min, then it was screened with the application of Hypersensitive ECL Chemiluminescence Kit (GBCBIO Technologies Inc. Lot:1903GC008).

Statistical analysis

The different expression levels of VIM in pan-cancers were compared by the Wilcoxon test. The survival curve was used by Kaplan–Meier plot, log rank test was applied to calculate log rank p value (p < 0.05). Spearman’s coefficient was used to check the correlation of gene expression. The receiver operating characteristic (ROC) curve and time-dependent ROC of glioma patients were calculated and generated with MedCalc in R version 4.0.2 (https://www.r-project.org/). In all the statistics, p < 0.05 was considered statistically significant.

Results

The different mRNA expression of VIM in glioma and normal samples

To explore the mRNA expression of VIM in all different cancers and normal samples, we checked The Cancer Genome Atlas (TCGA) dataset and GTex database, Fig. 1A shows the general atlas of VIM in pan-cancer. We then specifically focus on the gliomas, including TCGA- LGG, TCGA-GBMLGG (the combination data of both LGG and GBM patients) and TCGA-GBM database, and found that the expression of VIM is much higher in glioma than that in normal samples (Fig. 1B for LGG, Fig. 1C for GBMLGG and Fig. 1D for GBM). The protein expression of VIM was checked at Human Protein Atlas (HPA), as shown in Fig. 1E, F and G, the protein expression of VIM recognized by different antibodies in glioma patients is much higher than that in normal brain tissue. There is no significant difference in VIM expression of glioma patients of different ages and gender (Supplementary Fig. 1), but it increases as the pathological grades increase (Fig. 1H in LGG and Fig. 1I in GBMLGG). The single-cell sequencing data from TISCH2 also indicated that the VIM expression is higher in malignant glioma cells than that in other cells (Fig. 1J and K from GSE131928, Fig. 1L–M from GSE84465, and Fig. 1N–O from GSE89567). We also checked VIM expression in the Chinese Glioma Genome Atlas (CGGA) and found the same results (Supplementary Fig. 1).

Fig. 1.

The different mRNA expression of VIM in glioma and normal samples. A The mRNA expression of VIM in pan-cancers. B The differentiated mRNA expression of VIM in TCGA-LGG, total 1675 samples (1152 normal + 523 tumor samples). C The differentiated mRNA expression of VIM in TCGA-GBMLGG, total 1846 samples (1152 normal + 689 tumor samples + 5 tumor adjacent samples). D The differentiated mRNA expression of VIM in TCGA-LGG, total 1323 samples (1152 normal + 166 tumor samples + 5 tumor adjacent samples). E The protein expression of VIM in normal brain tissues (left) and glioma (right) with antibody HPA01762. F The protein expression of VIM in normal brain tissues (left) and glioma (right) with antibody CAB000080. G The protein expression of VIM in normal brain tissues (left) and glioma (right) with antibody CAB058687. H The expression of VIM in different stages in LGG. I The expression of VIM in different stages of GBMLGG. J–K Single cell sequencing of GBM sample GSE 131928. L–M The single-cell sequencing data from TISCH2 of GSE84465 shows the expression of VIM in different clusters. N–O The single-cell sequencing data from TISCH2 of GSE89567 shows the expression of VIM in different clusters. * p value < 0.05; ** p value < 0.01; *** p value < 0.001

The correlation between VIM expression and glioma patients’ survival

To explore the effect of VIM on the glioma patients’ survival, we first checked the Kaplan–Meier (KM) survival curves, and found that the higher VIM expression in LGG (Fig. 2A), GBMLGG (Fig. 2B) and GBM (Fig. 2C) indicated shorter survival. The most significant difference is in GBMLGG (Fig. 2B). The receiver operating characteristics (ROC) and time-dependent ROC were used widely to check the quality of a prognostic biomarker. We checked the ROC (Fig. 2D in LGG, Fig. 2E in GBMLGG and Fig. 2F in GBM) and found that all the area under the ROC curve (AUC)scores were above 0.8, indicating that the high specificity and sensitivity of VIM as a prognostic marker in glioma. The Fig. 2G, H, and I shows the time-dependent ROC in LGG, GBMLGG and GBM respectively. Furthermore, clinical predictions require calibration, the calibration curve is a scatter plot of the actual occurrence rate and the predicted occurrence rate, we checked the calibration of the curve too. As shown in Fig. 2J (LGG), K (GBMLGG), the predicted 5-year survival is more similar to the ideal line. But in GBM (Fig. 2L), it deviates from the ideal line more than that in the previous LGG. The univariate and multivariate Cox regression analyses were conducted (Fig. 2M, N). The correlation between the expression of VIM and Chinese glioma patients’ survival were checked at CGGA too (Supplementary Fig. 2), consistent with the result in TCGA, the higher VIM expression indicates shorter survival in CGGA. In summary, VIM expression has close relations with glioma patients’ survival and can be a strong prognostic biomarker for GBMLGG.

Fig. 2.

The clinical correlation of VIM expression in glioma. A–C OS between and high-low expression groups of VIM gene in KM databases of LGG (A), GBMLGG (B) and GBM (C) patients. D–F ROC curve established the efficiency of VIM mRNA expression level on distinguishing LGG tumor(D), GBMLGG(E), and GBM (F) from non-tumor tissue. X-axis represents the false positive rate, and Y-axis represents true positive rate. G–I The ROC curve using VIM as an indicator of LGG (G), GBMLGG (H) and GBM(I) were explored. J–L The calibration curve using VIM as an indicator of LGG (J), GBMLGG (K) and GBM (L) was checked. M–N The univariate (Fig. M) and multivariate regression (Fig. N) analysis of VIM and other clinicopathologic parameters with OS in LGG patients were explored. *p value < 0.05; ** p value < 0.01;*** p value < 0.001. OS, overall survival, DFS, disease-free survival

The differentially expressed genes analysis in VIM-high and -low groups of glioma patients

To explore the function of VIM, we divided the glioma (TCGA-GBMLGG) into VIM-high and -low groups, the differentially expressed genes (DEGs) were shown in Fig. 3A and B. To check the functions of these DEGs, we then conducted Gene Ontology (GO) and KEGG pathway analysis, and found genes in the VIM-high groups were enriched in the “Hematopoietic cell lineage” pathway and genes are enriched in the “ immune response” (Fig. 3C). The genes in VIM-low groups are enriched in Neuroactive ligand-receptor interaction pathway (Fig. 3D). The Gene Set Enrichment Analysis of all the DEGs shown in Fig. 3E and F show the genes in VIM-high groups are enriched in “Reactome neutrophil degranulation”. The combination of GO/KEGG and logFC analysis (Fig. 3G, H, I) indicates that the DEGs most highly expressed in VIM-high groups contribute to immune response (GO annotation in Fig. 3I).

Fig. 3.

The analysis of differentiated genes in VIM-High and -low groups in glioma patients. A The volcano plot shows the DEGs in VIM-high and -low groups. B The Rank of differentially expressed genes. C The GO and KEGG analysis of high expressed genes in VIM-high groups. D The GO and KEGG analysis of low expressed genes in VIM-low groups. E The GSEA analysis or the DEGs. F The GSEA analysis in ridge plot of DEGs. G–I The combinational analysis of GO/KEGG and LogFC of VIM-high groups

The correction between VIM expression and immune infiltration in glioma

According to the above analysis of DEGs, VIM expression has close relations with hematopoietic cell lineage and immune response. We then explored the relationship between VIM and immune infiltration. We calculated the immune cell infiltration first by Estimation of Stromal and immune cells in malignant tumour tissues using expression data (ESTIMATE), three scores including the stroma score, the immune score and the ESTIMATE scores were calculated. As shown in Fig. 4A–I, all three scores are very high, indicating the immune cells quantity is high in the tumor microenvironment, and all the scores have a positive correlation with VIM expression, suggesting that VIM is important for immune infiltration. We also checked Tumor Immune Estimation Resource (TIMER), macrophages and dendritic cells (Fig. 4J) have a higher correlation with VIM expression. We checked total of 22 immune cell infiltration via ssGSEA and found macrophages are the most infiltrated cell type (Fig. 4K, L, M). In addition, immunophenoscore (IPS) is an index to indicate of immunogenicity in solid tumors [22], we checked the IPS and found VIM expression has a significantly positive correlation with MHC (antigen presentation) and EC (effect cells) and a negative correlation with SC (suppressor cells) and CP (checkpoints) in TCGA-LGG and TCGA-GBMLGG (Fig. 4N). Moreover, immune checkpoints are critical modulators of immunotherapy [25, 26], we then explored the relation between VIM expression and checkpoint expression and found that it exhibits different characteristics in wildtype and different stages of gliomas (Fig. 4O). We further browsed the single-cell sequencing data posted on TISCH2, the expression of VIM is significantly shown in the Mono/Macro cluster in GSE162631 (Fig. 4P–Q). In summary, the VIM expression has close relations with immune infiltration and immunogenicity in gliomas, especially in LGG.

Fig. 4.

The expression of VIM has a positive correlation with the innate immune cell infiltration. A–C The correlation between the expression of VIM and stroma score. D–F The correlation between the expression of VIM and immune score. G–I The correlation between the expression of VIM and ESTIMATE score. J The correlation of the expression of VIM and immune cell infiltration, data from TCGA + GTEX. E The correlation of the expression of VIM and immune cells infiltration, data from GSEA via ssGESA analysis. K–M The correlation between the expression of VIM and the enrichment of macrophage in different grades of gliomas. N The relation between VIM expression and the IPS. MHC: antigen processing checkpoints/immunomodulators, EC: effector cells, SC: suppressor cells, IPS: immunophenoscore. O The correlation between VIM expression and the immune checkpoints. P–Q The single-cell sequencing data from TISCH2 GSE162631 shows the VIM expression in different clusters in glioma tumors

The DNA methylation analysis of VIM and its relations with glioma patients’ survival

DNA methylation has been widely used as a biomarker for cancers. We therefore check the methylation status of VIM. As shown in Fig. 5A, the DNA methylation of VIM in glioma samples is much lower than that in normal samples, and becomes lower in higher grades glioma (Supplementary Fig. 3). The DNA methylation of VIM plays diversified roles (Fig. 5B). There is a negative relation between the DNA methylation level of VIM and the expression of VIM (Fig. 5C), this result is consistent to the high expression of VIM in gliomas as its DNA methylation level is low. We then checked the DNA methylation level of VIM and the glioma patients’ survival, and found that patients with higher DNA methylation of VIM have longer survival (Fig. 5D), which supports the relation between VIM expression and glioma patients’ survival. All the above results were verified in the CGGA database for Chinese glioma patients. (Supplementary Fig. 3 and Fig. 4).

Fig. 5.

The DNA methylation status of VIM and its relations with patients’ survival time. A The DNA methylation level of VIM in glioma sample and normal sample. B The diversified roles of the methylation of VIM. C The dot plot shows the expression of VIM and its methylation status. D The correlation between the methylation of VIM (at the location of Shore, cg-10031015) and patients’ survival, methylation probes cg-1003015

VIM contributes to glioma progression

To explore the function of VIM in glioma progression and heterogeneity, we first checked the status of the genome stability and heterogeneity of VIM by indicators including MATH(Mutant-allele tumour heterogeneity), MSI(Microsatellite instability), HRD(homologous recombination deficiency), and LOH(Loss of heterozygosity) and their relations to gliomas. As shown in Fig. 6A, B, C and D, in TCGA-GBMLGG, the MATH and MSI of VIM has negative relations with glioma and the HRD and LOH of VIM has positive relations with gliomas. Therefore, the VIM gene itself has close relations with gliomas. Knocking out VIM in different glioma cancer cells shows different cell viability (Supplementary Table 2) in DepMap database, indicating that the function of VIM to cell viability varies in glioma cancer cells and glioma cancer cells have great heterogeneity.

Fig. 6.

VIM expression is positively related to glioma cancer cell progression and stemness. A The mutant-allele tumor heterogeneity (MATH) of VIM in pan-cancer including glioma. B The Microsatellite instability (MSI) of VIM in pan-cancer including glioma. C The homologous recombination deficiency (HRD) of VIM in pan-cancer including glioma. D The loss of heterozygosity (LOH) of VIM in pan-cancer including glioma. E The relationship between the expression of VIM and CDH2 in GBMLGG. F The relationship between the expression of VIM and SNAIL1 in GBMLGG. G The relationship between the expression of VIM and TWIST1 in GBMLGG. H The relationship between the expression of VIM and ACTA2 in GBMLGG. I The relationship between the expression of VIM and EGF in GBMLGG. J The relationship between the expression of VIM and PDGFA CDH2 in GBMLGG

In addition, VIM has been demonstrated as an EMT biomarker in previous studies. We checked its expression with other EMT markers such as CDH2, Snail 1, Twist 1, ACTA2, epidermal growth factor (EGF) and platelet-derived growth factor subunit A (PDGFA), as shown in Fig. 6E–J, the expression of VIM has positive correlations with all these EMT biomarkers.

The high expression of VIM maintains the resistance of glioma cells OXPHOS- inhibition

To check the function of VIM in drug resistance, we rechecked our previous RNA-sequencing data after Gboxin treatment, Gboxin is an oxidative phosphorylation (OXPHOS) inhibitor [7, 27]. We found VIM expression is much higher in the OXPHOS-inhibition resistant cells than that in the OXPHOS-inhibition sensitive ones (Fig. 7A, B). We double checked the expression of VIM in CCLE for the cell lines we tested found that VIM expression in OXPHOS-inhibition resistant cells (Fig. 7C) is higher than that in the sensitive ones (Fig. 7D). Among all the cell lines we treated with Gboxin, two glioma cells SF-126 and GB-1 were included, their IC50s are the highest in all the 57 cell lines we tested (Fig. 7E and Supplementary Table 1). We then checked the relation between the expression of VIM and the IC50s of the cell lines we did RNA-sequencing, and found that the VIM expression is higher, the IC50 is higher in these cells (Fig. 7F), indicating that the expression of VIM has a positive relation with the OXPHOS inhibitor in these cancer cell lines, so the expression of VIM contributes to the drug resistance. We propose that high expression of VIM in gliomas maintains the resistance of glioma cell’s OXPHOS-inhibition. To further validate this hypothesis, we checked the cell viability of SF-126 after treatment with Gboxin and found its cell viability maintains at around 78% even the Gboxin is 10 μM (Fig. 7G), we checked more glioma cells after treatment with another classical OXPHOS inhibitor Oligomycin A from DepMap PRISM database and found their CERES scores are above zero (Fig. 7H), indicating they have good cell viability even after Oligomycin A treatment.

Fig. 7.

High expression of VIM renders the glioma cells OXPHOS inhibition. A–B Relative RNA expression of VIM in 8 cancer cells that we did RNA-sequencing, among which GB-1 and SF-126 are the human glioma cells. C The relative expression of VIM of OXPHOS-inhibition resistant cancer cell line, data from CCLE. D The relative expression of VIM of OXPHOS-inhibition sensitive cancer cell line, data from CCLE. E The IC50 of all 57 cancer cell lines after treatment of Gboxin (the OXPHOS inhibitor) for 72 h. Glioma cancer cells SF-126 and GB-1 were marked. F The correlation between the expression of VIM and the OXPHOS-resistant cancer cell’s IC50. G Cell Viability Assay shows the cell viability of glioma cell line SF-126 after treatment with Gboxin in different dose after treatment of 72 h. H Cell viability of more glioma cells after treatment with another classical OXPHOS inhibitor Oligomycin A, data is from DepMap PRISM database. I–J The knocking down efficiency of shVIM at transcription level and protein level in SF-126 cell line. K Cell viability of SF126 after transfected virus of shVector and shVIM, cells were treated by Gbxoin 72 h in different doses. L The relative expression of VIM in SF-126 and G98 at their transcriptional level. M The ATP result of G98 treated by Gboxin for 72 h at different doses. N–O The overexpression of VIM at transcriptional level and protein level in G98 cancer cell line. P Cell viability of G98 after transfected virus of oeVector and oeVIM, cells were treated by Gboxin 72 h in different dose. Q The correlation between the expression of VIM in LGG gliomas with key OXPHOS genes including mt-CO1, mt-CO2 and mt-CO3. R The correlation between the expression of VIM in GBMLGG gliomas with key OXPHOS genes including mt-CO1, mt-CO2 and mt-CO3. S The correlation between the expression of VIM in GBM gliomas with key OXPHOS genes including mt-CO1, mt-CO2 and mt-CO3

We verified our hypothesis by knocking down VIM in SF-126 glioma cancer cell line ( the OXPHOS-inhibition resistant cell line), as shown in Fig. 7I–K, after knocking down VIM in SF-126, cells are much more sensitive to Gboxin treatment. Meanwhile, we over-expressed VIM in the VIM-low human glioma cancer cell line G98 (Fig. 7M–O), as shown in Fig. 7P, after overexpressing VIM in G98, cells become more resistant to Gboxin. This indicates that the expression of VIM maintains glioma cancer cell line’s resistance to OXHOS-inhibition. We further checked the relation between VIM expression and key genes in the OXPHOS pathway including mt-CO1, mt-CO2 and mt-CO3 in LGG, GBMLGG and GBM. As shown in Fig. 7Q–S, the expression of VIM has a negative correlation with all these OXPHOS genes. Therefore, the high expression of VIM in glioma maintains the resistance to OXPHOS inhibition.

Discussion

In this study, we comprehensively analyzed the role of VIM in glioma via analysis data from public resources and our RNA-sequencing data. We found that 1) The expression of VIM is expressed higher in glioma than that in normal tissues. 2) The high expression of VIM in gliomas predicts a poor prognosis and shorter survival. 3) VIM has signification relations with immune infiltration and immunotherapy. 4) The hypomethylation of VIM and its high expression have close relation with glioma patients’ survival. 5) VIM gene stability and heterogeneity have close relations with glioma and VIM promotes glioma progression and cancer stemness. 6) VIM maintains the glioma cancer cell OXPHOS-inhibition resistance.

VIM expression is highly and positively related with the EMT biomarkers and cancer metastasis biomarkers in glioma, indicating its role in the glioma, even the glioma doesn’t happen often to metastasis like other tumors due to its unique intracranial location. VIM has been verified to contribute to the function of elastic resilience under various stress conditions [9]. Cancer cells highly expressing VIM have advantages in EMT and in the process of invasion and migration [28].

VIM primarily functions in the EMT and thus promotes cancer invasion, but it also interacts with immune cells and contributes to the cancer cell immune evasion. In lung adenocarcinoma, VIM facilitates immune evasion by recruiting Smad2/3 to PD-L1 to promote cancer metastasis [29]. In cancer immunotherapy, human body uses the immune cells to attack cancer cells, but the cancer cells resist to the immune cytotoxicity by regulating VIM expression and remodeling cytoskeleton structure [30]. VIM also negatively regulates type I interferon production by interacting with TBK1 and IKKε [31]. Furthermore, macrophages in the tumor microenvironment secrete VIM into the extracellular matrix and contribute to the inflammation [32], the extracellular VIM also regulates the activation of dendritic cells and subsequently suppresses the pro-inflammatory adaptive immune [33]. Overall, the VIM expression can be applied to assess the immune response in the glioma clinical diagnosis.

VIM also contributes to cancer stemness. In the process of EMT, VIM will be highly expressed in the cytokeratin-positive epithelial cells and the EMT thus induces stem cell properties [9]. This Mesenchymal-like stemness characterized by the high expression of N-cadherin and vimentin and other stemness biomarkers promotes tumor progression and recurrence. VIM also protects the differentiating Embryonic Stem Cells (ESC) from aggerating toxicity [34]. EMT induces stem cell characteristics and drives tumor metastasis, drug resistance, and recurrence.

Conclusion

In conclusion, we deciphered the role of VIM in gliomas and found its high expression in gliomas affects the clinical outcomes by affecting immune infiltration, promoting cancer progression, and OXPHOS-inhibition resistance.

Supplementary Information

Below is the link to the electronic supplementary material.

Authors’ contributions

K.Z and X.J conceived and designed the study. Y.L analyzed the data, did experiments and wrote the manuscript. S.Z,Y.C, W,M and L.H performed some of experiments and did partial literature sourcing. S.L, J.C, Y.S, X.C and X,Z analyzed parts of the data and sourced literature. All authors read and approved the manuscript.

Funding

This study was funded by Discipline Climbing Scheme (2019YXK030) and Neuroscience Innovation and Development Research Project (YXJL-2022–00351-0183). This work was supported by grants from the National Natural Science Foundation of China (82073274, Y.S.), Science Technology Commission of Shanghai Municipality (20S11900700, Y.S.).

Data availability

The data underlying this study are freely available from TCGA data portal (https://portal.gdc.cancer.gov/projects/TCGA-LGG). The RNA-seq raw sequence data reported in this paper has been deposited into the Genome Sequence Archive (GSA) for humans under accession: HRA001452.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.