Flotillin‐1 is a prognostic biomarker for glioblastoma and promotes cancer development through enhancing invasion and altering tumour microenvironment

Abstract Flotillin‐1(FLOT1) has long been recognized as a tumour‐promoting gene in several types of cancer. However, the expression and function of FLOT1 in glioblastomas (GBM) has not been elucidated. Here, in this study, we find that the expression level of FLOT1 in GBM tissue was much higher than that in normal brain, and the expression was even higher in the more aggressive subtypes and IDH status of glioma. Kaplan–Meier survival revealed that high FLOT1 expression is closely associated with poor outcome in GBM patients. FLOT1 knockdown markedly reduced the proliferation, migration and invasiveness of GBM cells, while FLOT1 overexpression significantly increases GBM cell proliferation, migration and invasiveness. Mechanistically, FLOT1 expression may play a potential role in the microenvironment of GBM. Therefore, FLOT1 promotes GBM proliferation and invasion in vitro and in vivo and may serve as a biomarker of prognosis and therapeutic potential in the fight against GBM.

In this study, we found that FLOT1 was upregulated in both GBM tissue and cells. GBM cells with FLOT1 knockdown displayed marked reductions in proliferation, migration and invasiveness. While overexpression of FLOT1 increased the proliferation, migration and invasiveness of glioma cells via MAPK signalling. Additionally, FLOT1 expression was associated with stromal and immune cell infiltration in GBM.

| Lentiviral transfection and Flot1 shRNA gene silencing
The lentiviral vector LV19-Flot1 containing the Flot1 overexpression gene and puromycin resistance gene, Flot1 shRNA(GCAGA GAA GTC CCA ACT AATT), contains the puromycin resistance gene, and the corresponding controls were transfected into U87 and U251.
Transfected cells were cultured with DMEM containing puromycin(2 μg/ml) for 14 days. Then, we used the transfected cells to extract protein.

| Western blotting
Protein samples mixed with 5Xloading buffer were boiled at 100°C for 5 min. Then, cooled down the samples and subjected them into sodium dodecyl sulfate-polyacrylamide gel, electrophoresed and transferred the sample to PVDF membrane. Then, blocked the membrane with 5% skimmed milk for 1 h, and then incubated the membrane with primary antibodies, anti-Flot1 (1:1000, Proteintech, Rosemont, IL, USA), anti-p-ERK1/2 (Thr202/Tyr204) (1:1000, Proteintech) at 4°C overnight. Washed the membrane, incubated it into secondary antibodies, and washed it again 3 times. At last, detected the signals with ChemiDoc (Bio-Rad).

| Cell wound healing assay
Inoculated cells (5 × 10 5 ) into a 12-well plate. Scratched the plate using 10 μl pipettes after cell adherence, washed the plate with PBS buffer, and then cultured cells in DMEM without serum.
Observed wound closure at 0 and 24 h. Image J (1.4.3.67) was used to evaluate cell migration ability after the images were taken in microscopic fields three times randomly (× horizon200) at both time points.

| Transwell invasion assay
Suspended U87 and U251 Cells (5 × 10 4 ) were placed in the upper chamber of 8μm-pore Transwells (BD Biosciences) precoated with Matrigel, and the lower chamber was filled with DMEM with 20% FBS to attract cells. Culture cells in cell incubator for 2 days. The 8μm hole was stained with 0.5% crystal violet for 30 min after fixation (4% paraformaldehyde), and three microscopic fields (× horizon 200) were selected randomly under the microscope and then count cells.

| Xenograft mouse model
Eight NOD-SCID male mice of 4 weeks old were divided into 2 groups randomly. One group was injected with U87 FLOT1-OE cells (1 × 10 6 ) on the back under skin, and the other group were injected with control cells. All the mice were monitored, and tumour size was measured every 6 days. We used the formula V = (Length × Width2) to calculate tumour volume.

| Datasets
We obtained clinical information and RNA-seq data for 143 GBM samples from TCGA cohort. By searching the Genotype-Tissue Expression (GTEx) cohort, we got RNA-seq data for 1151 normal  Table S1. In addition, we adopted immunohistochemistry images of FLOT1 protein expression in normal brain and GBM patients' tissue from The Human Protein Atlas. 17

| Gene enrichment analysis
In this study, FLOT1-related genes were identified in the TCGA and CGGA cohorts, respectively, based on Spearman correlation analysis

| Immune infiltration analysis
Using the methods from Yoshihara et al., 19 the ESTIMATE algorithm was used to estimate ImmuneScore, StromalScore and ESTIMATEScore in GBM samples. We did single-sample gene set analysis to quantify the relative abundance of 28 previously reported immune cells. 20 In addition, by using TIMER tool, we explored the relationship exists between FLOT1 expression and the abundance of six major immune cell infiltration (https://cistr ome.shiny apps.io/timer/). 21

| Statistical analysis
The Student's t-test was used to analyse the two groups of data.

| FLOT1 is highly expressed in GBM tissue and associated with clinicopathology characters and patients' outcome
Given that FLOT1 was overexpressed in several types of cancer, we investigated the expression of FLOT1 in human GBM samples and normal brain tissue by querying TCGA and GTEx databases. We found that the expression level in GBM tissue is much higher than that in normal brain (p < 0.0001) ( Figure 1A). Meanwhile, in contrast to normal brain tissue, GBM samples exhibited an increased staining intensity ( Figure 1B, Figure S3A,B).
As is well known, IDH1 and MGMT status affected GBM patients' outcome. According to the data in CGGA and TCGA databases, FLOT1 was highly expressed in IDH-WT samples ( Figure 1C), but no significant difference was found between methylated and unmethylated MGMT samples ( Figure 1D). Furthermore, we investigated the expression of FLOT1 in different GBM transcriptional subtypes.
Although there was a distinct trend between these three subtypes according to the TCGA and CGGA databases, we found that FLOT1 was highly expressed in the mesenchymal subtype ( Figure 1E).
To further study the effect of FLOT1 on GBM, we also did Kaplan-Meier survival analysis, and the results showed that patients with high FLOT1 expression showed shorter overall survival time (Figure 2A

| FLOT1 promotes proliferation, migration and invasion in glioma cells
To investigate the role of FLOT1 in GBM, U87 and U251 cells were transfected with FLOT1-OE and FLOT1-SH lentivirus,

| FLOT1 promotes GBM tumour growth in xenograft mouse model
To further investigate the role of FLOT1 in GBM development, we inoculated FLOT1-OE and control U87 cells into NOD-SCID mice on the back under skin. Mice were sacrificed, and tumours were collected and weighed 40 days later. We found that the tumours from FLOT1-OE group were larger in size and weight than that in control group, which indicated that FLOT1 overexpression dramatically increased GBM xenograft tumour development ( Figure 4A-C).

| Functional enrichment of FLOT1 in GBM
To investigate the mechanism of FLOT1 in GBM, we first identified FLOT1-related genes in TCGA cohorts, using Spearman correlation analysis with a threshold of |R| >0.6 and p < 0.05. The results showed that 180 genes were related to TCGA (Table S2). Next, we did GO enrichment and KEGG pathway analyses (Tables S3   and S4)

| FLOT1 expression is associated with stromal and immune cell infiltration in GBM
According to molecular studies, infiltrating stromal and immune cells had a great impact on cancer biology and tumour signalling perturbation. 19 Initially, we explored the association between FLOT1 expression and immune, stromal and ESTIMATES scores. According to TCGA database, the results showed a positive correlation between FLOT1 expression and immune, stromal and ESTIMATES scores ( Figure 6A).
Results from the CGGA database showed that FLOT1 expression was significantly associated with immune, stromal and ESTIMATES scores

| DISCUSS ION
Although recent studies have helped understand the biological features of glioma, further investigation is required to explore the specific molecular mechanisms and therapeutic targets of this disease. In this study, we first reported FLOT1 as a potential molecular marker in the development of glioma, as evidence suggested that FLOT1 is closely associated with clinicopathological features and immune surveillance of gliomas and promoted glioma cell proliferation, invasion and migration.
FLOT1 is a molecular marker of lipid rafts, which serves as a physical platform for assembling functional complexes. Therefore, FLOT1 is involved in many biological processes, including endocytosis, adhesion, actin cytoskeleton reorganization and cell-signalling events. [10][11][12][13] In addition, FLOT1 has been reported to be upregulated in many tumour types [14][15][16] and is involved in cancer biology by initiating receptor kinase signalling. 23 In this study, we found that FLOT1 is markedly upregulated in glioma tissue, compared with normal brain tissues. Moreover, according to the data from TCGA and CGGA databases, we documented that the level of FLOT1 expression was negatively correlated with the aggressiveness types and IDH status of glioma and overall survival of glioma patients, which indicated that FLOT1 was a prognostic marker for glioma patients.   30 Based on these findings, FLOT1 may play a potential role in the tumour microenvironment. However, the specific underlying mechanism requires further investigation.
Currently, there is no inhibitor targeting FLOT1. However, aside from GBM, FLOT1 was reported to play a role in neurological disorders, for example Alzheimer's disease, 31 Parkinson, 32 multiple sclerosis and biological ageing. 33 Therefore, FLOT1 may be a potential target for brain-related disease including tumour.

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial interests or conflicts concerning the work described.

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 from the corresponding author upon reasonable request.