Identification and verification of key miRNAs in M2 macrophage-derived exosomes
We obtained the dataset GSE51332 from GEO database, which included four cases of peripheral blood mononuclear cells and four cases of infiltrating macrophages in glioma patients. By analyzing the miRNA expression in the samples through GEO2R (Figure 1A), we identified seven miRNAs that were most significantly differentially expressed in infiltrating macrophages (Table 2). By visualizing the expression in the samples through ClusterVis, we obtained the heat maps of the expression profiles of these seven miRNAs (Figure 1B).
Table 2
Information of differentially expressed miRNAs
miRNA ID | Log2FC | P.Value |
hsa-miR-15a | -2.24935 | 1.27E-05 |
hsa-miR-223 | -2.21532 | 8.15E-03 |
hsa-miR-92a | -2.14024 | 7.60E-05 |
hsa-miR-142-3p | -2.11088 | 1.71E-04 |
hsa-miR-574-5p | 2.10026 | 1.35E-02 |
hsa-miR-4792 | 2.36518 | 2.96E-03 |
hsa-miR-34a | 2.65544 | 1.36E-07 |
Subsequently, we tested the expression levels of biomarkers in macrophages with different differentiation states. The results of western blot assays (Figure 1C) showed that CD206 (M2 macrophage phenotype marker gene) was significantly over-expressed in M2 macrophages, and the expression of CD68 (macrophage marker gene) was not significantly different in M0, M1 and M2 macrophages. By qRT-PCR detection, we found that TNF-a, IL-12 and iNOS (M1 macrophage phenotype marker genes) were up-regulated in M1 macrophages compared with M0 macrophages (Figure 1D). IL-10, TGF-β, and CD206 (M2 macrophage phenotype marker genes) were found to be highly expressed in M2 macrophages compared with M0 macrophages (Figure 1E). Then we induced macrophage differentiation according to the tested markers through in vitro cell experiments and measured the expression of these seven miRNAs using qRT-PCR. The results elucidated that miR-15a and miR-92a were significantly under-expressed in M2 macrophages (Figure 1F), consistent with the expression levels in GEO. However, no obvious upregulated expression levels were shown in the detection of miR-574, miR-4792 and miR-34a, which was not consistent with their expressed results predicted in GEO. Therefore, we selected the obviously downregulated miRNAs, miR-15a and miR-92a, as the next research objects.
In order to verify the expression of miR-15a and miR-92a in exosomes derived from M2 macrophages, exosomes need to be extracted from M2 macrophages. Thus, we collected the culture supernatant of M2 macrophages and THP-1 cells which were used as the control group. According to the TEM image obtained in JEM-2010 HT transmission electron microscope (Figure 1G), the supernatant of M2 macrophage cell culture contained exosomes, of which the shape was solid and dense. Besides, analysis of size distribution of exosomes was assayed by NTA. As shown in Figure 1H, the isolated exosomes had a predominant size of 70-120nm. Exosomes were then extracted and the expression levels of CD9, CD63, and TSG101 (biomarkers of exosomes) were determined by western blot assays. The results showed that the protein levels of TSG101, CD63 and CD81 were significantly increased in the Exo group (Figure 1I), while the protein levels of Calnexin was evidently diminish in the Exo group compared with control group (Figure S4), which further confirmed the successful extraction of exosomes. Then we collected the culture supernatants of macrophages with different differentiation states, extracted the exosomes, and examined the expression of miR-15a and miR-92a. As shown in Figure 1J and 1K, miR-15a and 92a were under-expressed in M2 macrophage exosomes, and their down-regulation multiples were 0.48 and 0.58, respectively. Therefore, miR-15a and miR-92a were selected as key miRNAs for subsequent research.
M2 macrophage-derived exosomes promote migration and invasion of glioma cells
With an aim to investigate the effect of exosomes derived from M2 macrophages on glioma cells, scratch wound healing and transwell assays were employed. We added the extracted exosomes of M0, M1 and M2 macrophages to glioma cells, respectively. The transverse migration ability of glioma cells was examined by scratch wound healing assays in T98 and U251 cell lines. It can be seen from Figure 2A that the scratch healing rate of cells with M2 macrophage exosomes was faster than that of the M0 and M1 groups (P<0.05). We obtained similar results in U251 cell line (Figure 2A), indicating that exosomes derived from M2 macrophages can promote transverse migration of glioma cells.
The transwell assays were applied to investigate whether M2 macrophage exosomes could promote the vertical migration and invasion of glioma cells. We found that the number of migrating and invasive cells was significantly increased in M2 group compared with M0 and M1 groups in T98 cells (Figure 2B). We obtained similar results in U251 cell line (Figure 2B), suggesting that exosomes derived from M2 macrophages can promote vertical migration and invasion of glioma cells. Taken together, these findings elucidated that M2 macrophage exosomes can promote migration and invasion of glioma cells.
Exosomal miR-15a and miR-92a inhibit migration and invasion of glioma cells
The foregoing results revealed that miR-15a and miR-92a were under-expressed in M2 macrophage exosomes and that exosomes could promote the migration and invasion of glioma cells. Therefore, we speculated that miR-15a and miR-92a might have an effect on glioma cells. After transfection of miR-15a and miR-92a into M2 macrophages, we collected cell culture supernatant and extracted exosomes. The results of qRT-PCR revealed that the expression of miR-15a was up-regulated about 4.9 times in M2 macrophage exosomes, and miR-92a was up-regulated about 7.0 times (Figure S1), indicating that we successfully over-expressed miR-15a and miR-92a in M2 macrophage exosomes.
Subsequently, we reconnoitered the effects of exosomal miR-15a and miR-92a on the migration and invasion of glioma cells. We transfected miR-15a, miR-92a and mimic NC into M2 macrophages, collected cell culture supernatant 48 hours later, extracted exosomes, and added them to T98 and U251 cell lines, respectively. The results from scratch would healing and transwell assays indicated that miR-15a over-expression inhibited the transverse migration of T98 and U251 cells (Figure 3A, P<0.05), as well as vertical migration and invasion (Figure 3C, P<0.05). We obtained similar results of miR-92a over-expression in T98 and U251 cell lines (Figure 3B, 3D, P<0.05). Collectively, exosomal miR-15a and miR-92a were found to inhibit migration and invasion of glioma cells.
Target gene prediction
To identify the target genes of miR-15a and miR-92a, picTar, miRanda, targetScan and PITA databases were used for prediction. A total of 304 target genes of miR-15a were obtained by intersecting the predicted target genes from the four online databases (Figure S2A). Through STRING and Cytoscape, we found ten hub genes from the 304 target genes of miR-15a (Figure S2B). Then we checked the expression of these ten hub genes in glioma and adjacent tissues in GEPIA. The results showed that CCND1, CDC42, RAF1, and CHEK1 were highly expressed in gliomas (Figure 4A). Therefore, these four genes were selected for subsequent research. For miR-92a, we obtained 240 target genes (Figure S2C), of which ten were identified as hub genes (Figure S2D). According to the expression in glioma and adjacent tissues in GEPIA, only RAP1B was highly expressed in gliomas (Figure 4B), which was chosen for following study.
Target gene validation
In order to verify the target genes obtained from database prediction, we performed qRT-PCR, western blot and dual luciferase reporter gene assays in vitro. We transfected miR-15a, miR-92a and mimic NC into T98 and U251 cells, respectively. The qRT-PCR results showed that the over-expression efficiency of miR-15a in T98 and U251 cells was 11.67 and 8.4 times (Figure S3A), and that of miR-92a was 15.92 and 14.59 times (Figure S3B), respectively. In addition, we verified the effect of target genes in cells through two different transfection methods of transfecting miR-15a and miR-92a into M2 type macrophages and collect exosomes or directly transfecting miR-15a and miR-92a into T98 cells. The results indicated that there was no evidently difference (Fiure S5).
After the transfection of miR-15a, only CCND1 was down-regulated in T98 and U251 cell lines among the four predicted target genes (Figure 5A). Western blot analysis also revealed that miR-15a over-expression could down-regulate the protein expression level of CCND1, so CCND1 was selected as the next research object (Figure 5B). The relative luciferase activity was decreased in co-transfection of pGL3-CCND1-WT with miR-15a, compared with the control of mimic NC (p<0.05), and there was no significant difference in luciferase activity in co-transfection of pGL3-CCND1-Mut with miR-15a (Figure 5C), indicating that miR-15a was bound to the CCND1 gene. For miR-92a, qRT-PCR and western blot assays confirmed that miR-92a over-expression can down-regulate the gene and protein expression levels of RAP1B (Figure 5D, 5E). In addition, the dual luciferase reporter gene assay confirmed that RAP1B was a target of miR-92a (Figure 5F).
RAP1B and CCND1 can activate the PI3K/AKT/mTOR signaling pathway
Studies have shown that PI3K/AKT/mTOR signaling pathway is closely related to the regulatory mechanism of glioma. We wondered whether CCND1 and RAP1B could affect the PI3K/AKT/mTOR signaling pathway in glioma. Therefore, si-CCND1 and si-RAP1B were transfected into T98 and U251 cell lines to study the effect of CCND1 and RAP1B on the PI3K/AKT/mTOR signaling pathway. As shown in Figure 7, CCND1 knockdown in T98 and U251 cell lines reduced the phosphorylation level of AKT and mTOR compared to the NC group. For RAP1B, we obtained similar results by western blot analysis (Figure 6A and 6B). Collectively, these results revealed that RAP1B and CCND1 could activate the PI3K/AKT/mTOR signaling pathway in glioma cells. To demonstrate whether macrophage-derived exosomal miR-15a and miR-92a have any effect on PI3K/AKT/mTOR signaling pathway in glioma cells. The relevant rescue experiments are performed. As shown in Figure 7A, miR-15a could block the PI3K/AKT/mTOR signaling pathway. The expression levels of p-mTOR and p-AKT were down-regulated in the miR-15a group. The over-expression of CCND1 in the miR-15a group could reverse the blocking effect of miR-15a on the signaling pathway and rescue the protein levels of p-mTOR and p-AKT to a certain extent. Similarly, miR-92a could block the PI3K/AKT/mTOR signaling pathway. The expression levels of p-mTOR and p-AKT were down-regulated in the miR-92a group, and overexpression of RAP1B in the miR-92a group could reverse the blocking effect of miR-92a on the signaling pathway, and rescue the protein levels of p-mTOR and p-AKT to a certain extent (Figure 7B).