The study was approved by the First Affiliated Hospital of the Harbin Medical University and implemented by the principles of the Helsinki Declaration. Patients signed informed consent forms before blood draw and surgery. We collected blood samples and clinical dates from 124 glioma patients, including three WHO grade I patients, 30 grade II patients, 34 grade III patients, and 57 grade IV patients who underwent treatment in the Department of Neurosurgery at the First Affiliated Hospital of Harbin Medical University in China between July 2015 and July 2017. The patients were diagnosed with glioma by postoperative pathological section according to the WHO criteria, and pathological diagnosis was performed by two independent pathologists. Non-glioma patients who had no medical history of another cancer were recruited into this study as controls (n = 36), matched by sex and age with the glioma group. Clinical characteristics of the glioma patients are summarized in Table 1. All plasma samples for glioma and non-tumor patients were collected in EDTA-K3 tubes before the start of any treatment. After the first centrifugation for 10 min at 3,000 × g, the supernatants were carefully moved to a new tube and snap frozen at −80°C to isolate the exosomes.

Human glioblastoma cell lines (LN229 and U87) were obtained from China Infrastructure of Cell Line Resource (National Science and Technology Infrastructure, NSTI). They were maintained in DMEM medium, and 10% fetal bovine serum (Biological Industries, Israel) was added to the medium. The miR-2276-5p mimics and negative control (NC) were purchased from the General Biosystems (Anhui, China), and Lipofectamine 2000 was purchased from Invitrogen (United States). Before transfection, the glioma cells LN229 and U87 were cultured in six-well plates at a density of 6 × 10⁴ cells per well, and transfections were performed according to the manufacturer’s instructions.

Five hundred microliters of plasma samples was centrifuged at 4°C at 300 × g for 10 min, 1,000 × g for 10 min, and 10,000 × g for 30 min to remove cells and debris. Then, supernatants were subjected to ultracentrifugation at 100,000 × g for 70 min. Next, we discarded the supernatants and used phosphate-buffered saline (PBS) to wash exosomes. Finally, the exosomes were pelleted by ultracentrifugation at 100,000 × g for 70 min again and resuspended in 100 μl of PBS for RNA isolation and follow-up use.

The use of transmission electron microscopy was consistent with that of previous reports, and the experimental procedures used in previous experiments were used (Wang et al., 2018). The exosome resuspension was loaded onto a carbon-coated 300 mesh copper grid and dried at room temperature for 5 min. Then, the grid was dyed with the 2% phosphotungstic acid at room temperature for 10 min, and the morphology of exosomes was checked using an electron microscope (JEM-1220, JEOL Ltd., Japan).

Extraction of proteins and Western blot (WB) analysis was performed as reported earlier (Wang et al., 2018). CD63 antibody (25682-1-AP, 1:1,000), CD9 antibody (20597-1-AP, 1:1,000), and GAPDH (60004-1-LG, 1:5,000) were purchased from Proteintech. RAB13 antibody (DF9813, 1:1,000) was purchased from Affinity Biosciences. Polyclonal goat anti-rabbit antibody (SA00001-2, Proteintech) was used as the secondary antibody, and the WB detection system (Gene Sys) was used for generation of data.

Transfected glioma cells LN229 and U87 were seeded in 96-well plates at a cell density of 1 × 10⁴ cells per well, and 20 μl of MTT was added at 24, 48, and 72 h. The test conditions were similar to our previous studies (Zhong et al., 2019).

Total exosome RNA was isolated from 100 μl of exosome samples using Trizol and stored in a –80°C freezer until use. U6 was used as an endogenous control miRNA that is stably expressed in exosomes. qRT-PCR was performed as described earlier (Gareev et al., 2019). The relative expression of miR-2276-5p was calculated using the equation 2^–ΔCt, in which ΔCt = Ct miR-2276-5p-Ct U6. The primer sequences used in this study are reported in Supplementary Table 1.

We consulted GEPIA¹, GEO², and GlioVis³ to check the expression level of RAB13 in the glioma patients and the outcome survival of RAB13 expression. TargetScan⁴, mirDIP⁵, and mirtargetbase⁶ were checked to calculate the gene targets of miR-2276-5p.

Statistical analyses were analyzed by SPSS 13.0. Data were expressed as mean ± SEM. Kaplan–Meier analysis was used to generate and analyze survival time data. The univariate Cox proportional hazards regression was performed on SPSS 13.0. The differences were considered statistically significant at p < 0.05.

To confirm the extraction of exosomes from the plasma of glioma and control patients, the markers of exosomes (CD63 and CD9) were evaluated by WB analysis. The result showed CD63 and CD9 expression in the samples, thus, establishing the presence of exosomes (Figure 1A). Then, we checked the diameter of exosomes using Nano-Sight (Figure 1B) and found the size of exosomes to be distributed below 100 nm. We further examined the morphology of exosomes and verified our findings using transmission electron microscope (Figure 1C). Thus, we concluded that we had successfully extracted exosomes from plasma samples.

As shown in Figures 2A,B, GSE139031, GSE113740, GSE113486, and GSE112246 reported a significantly reduced expression of miR-2276 in glioma. To further understand whether miR-2276-5p is also differentially expressed in the plasma exosomes of glioma patients, we checked miR-2276-5p in the exosomes isolated from the plasma of glioma patients and controls. The expression of exosomal miR-2276-5p was found to be differential between glioma and non-glioma patients’ plasma (Figure 2C). The results showed that plasma exosomal miR-2276-5p in glioma is significantly lower, compared with the non-glioma controls (Figure 2C). Another interesting revelation was that the exosomal miR-2276-5p expression levels were lower in patients with high-grade glioma (HGG, including grade III and grade IV) than in patients with low-grade glioma (LGG, including grade I and grade II) (Figure 2D). The result of univariate logistic regression analysis of factors associated with a risk factor for glioma (Table 2). Then a receiver operating characteristics (ROC) analysis curve was used to analyze the predictive diagnostic capability of exosomal miR-2276-5p for glioma patients. Admission exosomal miR-2276-5p had a good area under curve with an AUC value of 0.8107 (Figure 2E). These results suggested that the exosomal miR-2276-5p expression levels are markedly decreased in glioma patients and further correlate negatively with the tumor grade.

We further explored whether there existed a relationship between these differences in expression and the survival outcome of the patients. The relative expression of exosomal miR-2276-5p in patients was divided into a high-expression group and a low-expression group, and our analysis showed that glioma patients in the low-expression group of exosomal miR-2276-5p had lower overall survival (Figure 2F). We used the univariate and multivariable Cox proportional hazards regression to confirm our findings (Table 3). Based on these findings, we suggest that plasma exosomal miR-2276-5p expression can be used as an independent factor to predict the survival of patients with glioma.

To better understand which genes are the targets of miR-2276-5p in glioma, we used mirDIP, mirtargetbase, and TargetScan to predict the target genes and found that RAB13 and PLEKHG48 might be the target of miR-2276-5p (Figure 3A). However, PLEKHG48 had no difference in expression in glioma patients, and it was not necessarily related to the prognosis of patients in the TCGA database (Supplementary Figure 1). We transfected miR-2276-5p mimics and negative control (NC) miRNA into the LN229 and U87 human glioma cell lines and found that when miR-2276-5p was ectopically highly expressed in LN229 and U87 glioma cells, the expression of RAB13 mRNA in cells was decreased, as evaluated by RT-PCR (Figure 3B). We further confirmed these results with Western blotting (Figure 3C); the protein expression of RAB13 also decreased in both the glioma cell lines. Furthermore, the survival of miR-2276-5p-transfected LN229 as well as U87 cells was worse than in the NC group (Figure 3D). Moreover, we evaluated the tumor tissues obtained from our hospital and obtained the results that showed that the expression of RAB13 mRNA was negatively correlated with miR-2276-5p (Figure 3E). Based on these results, we speculate that RAB13 may be a target gene for miR-2276-5p. We used GSEA to analyze the TCGA-related data and analyzed the cell signaling pathways related to RAB13. Our analysis revealed that RAB13 was enriched in IL2-STAT5 signaling pathway, TGF-β signaling pathway, IL6-JAK-STAT3 signaling pathway, angiogenesis signaling pathway, TNF-α signaling pathway, and epithelial–mesenchymal transition (Figure 4A). Furthermore, in order to clarify that RAB13 was correlated with the overall survival of patients, the expression level of RAB13 was checked in the CGGA and TCGA databases. The expression level of RAB13 in HGG patients was significantly higher than in LGG patients (Figure 4B). Next, we classified the patients in the CGGA and TCGA database via IDH1 and 1p19q status; RAB13 was highly expressed in IDH1 wild-type and 1p19q non-codel patients (Figure 4B). We verified using the GlioVis database that the prognosis of patients with high RAB13 expression was significantly worse than that of patients with low RAB13 expression in both the CGGA and TCGA patients (Figure 4C). Thus, we conclude that RAB13 is a genuine target of miR-2276-5p, which is differentially expressed in glioma patients, relative to the controls, as well as in LGG vs. HGG patients, making it a putative indicator to predict the prognosis of glioma patients.