Identification of DE‑miRNAs and their target genes
The GSE65071 dataset (20 osteosarcoma plasma samples and 15 normal plasma samples) was first downloaded and preprocessed (normalized) (Fig. 1A and B). After preprocessing and removing batch effects, we analyzed the DEmiRNAs of GSE65071 using the limma package using FDR (false discovery rate) P < 0.05 and |log2fold change (FC)| > 2 for upregulated genes and |log2fold change (FC)| < − 2 for downregulated genes as the cutoff criteria. Based on this analysis, a total of 149 miRNAs were found to be significantly differentially expressed in osteosarcoma samples more than twofold when compared to normal samples, including 136 upregulated and 13 downregulated miRNAs. For better visualization, the top three most upregulated miRNAs and top three most downregulated miRNAs are presented in Table 1. A total of 864 potential target genes were predicted for the three upregulated miRNAs (hsa-miR-663a, hsa-miR-31-5p, and hsa-miR-203a) and 449 genes were predicted for the three downregulated miRNAs (hsa-miR-199a-5p, hsa-miR-223-3p, and hsa-miR-191-5p) by using miRTarBase. A volcano plot of these DE-miRNAs is provided in Fig. 1C. Cluster analysis of the top three DEmiRNAs (up- and downregulation) also revealed significant differences between osteosarcoma samples and normal samples (Fig. 1D).
Enrichment analysis of the target genes
Tree categories of GO functional annotation analysis were performed on these potential target genes, including biological process (BP), molecular function (MF) and cellular component (CC). As shown in Fig. 1E, the enriched GO functions for target genes of the three upregulated miRNAs included the regulation of transcription from RNA polymerase II promoter, regulation of transcription, DNA templated, negative regulation of transcription, DNA templated, regulation of cell migration, and positive regulation of transcription, DNA templated in the BP category. For the MF items, these mRNAs were enriched in transcriptional activator activity, RNA polymerase II transcription regulatory region sequence-specific binding, RNA polymerase II transcription regulatory region sequence-specific DNA binding, and DNA binding (Fig. 1E). Regarding CC, the DEmiRNAs were significantly enriched in nuclear chromatin, nuclear chromosome part, chromatin, focal adhesion, and nucleolus (Fig. 1E). As shown in Fig. 1F, the enriched GO functions for target genes of the three upregulated miRNAs included the regulation of transforming growth factor beta2 production, positive regulation of transcription from RNA polymerase II promoter, regulation of transcription, DNA templated, positive regulation of transcription, DNA templated, and regulation of gene expression in the BP category. For the MF items, these mRNAs were enriched in transcription regulatory region DNA binding, protein kinase binding, RNA polymerase II transcription regulatory region sequence-specific binding, and RNA polymerase II transcription regulatory region sequence-specific DNA binding (Fig. 1F). Regarding CC, the DEmiRNAs were significantly enriched in nuclear chromatin, chromatin, RNA polymerase II transcription factor complex, and nuclear chromosome part (Fig. 1F).
Construction and analysis of the PPI network and miRNA‑hub gene network
To explore the interactions of the target genes, a PPI network based on the STRING database was constructed. The results showed that many of these target genes could interact with each other. The top ten hub genes listed in Table 2 were screened out according to the node degree. For the upregulated miRNAs, the hub genes were TP53, PIK3CD, HRAS, CDKN1A, TGFB1, H2AFX, GRB2, ABL1, CDK1, and GNAI2. For the downregulated miRNAs, the hub genes were TP53, VEGFA, IL6, PIK3CD, STAT3, CDK2, NFKB1, CDH1, ERBB2, and MDM2. Among these genes, TP53 node degree is the highest. The results suggest that TP53 may be an important target correlated with osteosarcoma.
Subsequently, Cytoscape software was used to construct the miRNA-hub gene network, as shown in Fig. 1G. From Fig. 1G, we found that ten hub genes (TP53, PIK3CD, HRAS, CDKN1A, TGFB1, H2AFX, GRB2, ABL1, CDK1, and GNAI2) could be potentially modulated by upregulated miR-663a. One hub gene each may be regulated by miR-31-5p and miR-203a. Additionally, miR-199a-5p could potentially target five (VEGFA, PIK3CD, NFKB1, CDH1, and ERBB2) of ten hub genes. Another five hub genes (TP53, IL6, STAT3, CDK2, and MDM2) may be regulated by miR-223-3p. These data indicated that miR-663a, miR-199a-5p, and miR-223-3p might be three potential regulators in the development of osteosarcoma.
Next, we further evaluated the patients' survival prognostic information of these 10 hub mRNAs in osteosarcoma using the R2 database (https://hgserver1.amc.nl/cgi-bin/r2/main.cgi) as shown in Fig. 2. The analytic results revealed that the expression of four (HRAS, CDKN1A, TGFB1, and ABL1) of eight targets of miR-663a was significantly upregulated in osteosarcoma tissue compared with normal tissue whereas TP53, PIK3CD, H2AFX, and GRB2 were markedly downregulated. It has been widely recognized that there is a negative correlation between miRNA expression and target gene expression. Therefore, among these detected genes, the significantly downregulated TP53, PIK3CD, H2AFX, and GRB2 may be the most potential targets for the upregulated miR-663a. Likewise, these meaningfully upregulated genes including VEGFA and ERBB2 may be the most potential targets for the downregulated miR-199a-5p in theory. And STAT3 and IL6 may be the most potential targets for the downregulated miR-223-3p.
miR-199a-5p could be transported into HUVECs via exosomes from osteosarcoma cells
To evaluate the effect of osteosarcoma exosomes on tumor angiogenesis, we first needed to determine whether the tumor exosomes were taken up by HUVECs. The extracted exosomes of HOS cells were characterized using TEM, NanoSight and western blotting. TEM images showed that the majority of the particles exhibited a cup- or round-shaped morphology (Figure 3A). As measured by NTA (nanoparticle tracking analysis), the diameter of the exosomes was approximately 122 nm (Figure 3B). The expression of the CD9 and CD63 proteins was detected using western blotting (Figure 3D). These results suggested that the extracted exosomes had characteristics that met widely accepted standards.
Additionally, to illuminate whether tumor-secreted exosomes were delivered to endothelial cells, exosomes derived from HOS cells were labeled with PKH67, and then the conditioned medium was collected and processed for exosomal purification. We then observed the transport of exosomes from osteosarcoma cells to HUVECs. After incubation with PKH67-labeled exosomes derived from HOS cells, PKH67 signals were localized in the cytoplasm of HUVECs (Figure 3E).
VEGFA is an important factor in the regulation of angiogenesis, and VEGFA is a potential target gene of miR-199a-5p, so we also detected whether miR-199a-5p was transported into HUVECs. As shown in Figure 3F, we found that HOS-exosomal miR-199a-5p-mimics-cy3 were internalized by HUVECs. Collectively, these data clearly demonstrated that miR-199a-5p secreted in the exosomes of osteosarcoma cells can be effectively transmitted into HUVECs.
Exosomal miR-199a-5p derived from osteosarcoma cells targets VEGFA in HUVECs
RT-PCR was applied to detect the expression of miR-199a-5p in osteosarcoma cells and their exosomes. The data indicated that the expression level of miR-199a-5p in exosomes derived from HOS and MG63 cells was remarkably higher than that in HOS and MG63 cells (Figure 4A and B). Additionally, miR-199a-5p mimics and inhibitor were transfected into HUVECs respectably.
It has been proven that miR-199a-5p can be transferred to HUVECs through exosomes, so we tested the effect of miR-199a-5p on HUVECs. We found that the expression of miR-199a-5p in HUVECs was significantly increased or decreased after transfection with mimics or inhibitor (Figure 4C and D).
According to our previous bioinformatics analysis, VEGFA was a candidate target of miR-199a-5p, whose mRNA 3’-UTR contains the miR-199a-5p interactive sequence (Figure 4E). To further clarify that VEGFA is a direct target of miR-199a-5p, we amplified the corresponding wild-type (WT) and mutant (MuT) 3’-UTRs and cloned them into the PGL3 luciferase reporter vectors. The relative luciferase activity of wild-type (WT) vectors was significantly diminished upon cotransfection with the miR-199a-5p mimic. However, the luciferase activity of the vectors harboring mutant (MuT) 3’-UTRs of these genes remained unchanged (Figure 4F). In addition, overexpression of miR-199a-5p using a miR-199a-5p mimic significantly decreased the expression of VEGFA and two other angiogenesis markers ANG1 and FLK1, while downregulation of miR-199a-5p expression via a miR-199a-5p inhibitor dramatically increased the expression of these three targets (Figure 4G-I). These results support bioinformatics prediction and confirm that VEGFA was indeed a direct target of miR-199a-5p in the HUVECs.
miR-199a-5p inhibits HUVEC migration and tube formation
Endothelial cell migration and tube formation are crucial steps in angiogenesis. We thus attempted to investigate the effect of exosomal miR-199a-5p on the migration and tube formation of HUVECs. First, the proliferation ability of HUVECs was detected using EdU assay after transfection with miR-199a-5p mimics or inhibitor. The data showed that cell proliferation was decreased after HUVEC transfection with miR-199a-5p-mimics compared with the control group (Figure 5A and F1).
A wealth of evidence indicates that CD34 is expressed not only by MSCs, but also by a variety of other nonhematopoietic cell types, including vascular endothelial progenitor cells. Therefore, we used CD34 as a marker for the analysis of neovascularization. As shown in Figure 5B and F2, the percentage of CD34-positive HUVECs was remarkably higher in the miR-199a-5p inhibitor group than in the control group, indicating increased capillary density in the miR-199a-5p inhibitor group.
In addition, the data of the Transwell experiment showed that compared with the control group, the migration ability of HUVECs was significantly reduced after transfection with miR-199a-5p mimics (Figure 5C and F3). The in vitro endothelial tube formation assay revealed that miR-199a-5p mimics remarkably inhibited tube formation of HUVECs compared with the control (Figure 5D and F4).
Moreover, platelet endothelial cell adhesion molecule-1 (CD31) also participates in the process of angiogenesis. After HUVECs were transfected with mimics or inhibitors, CD31 immunohistochemistry was performed. The results suggested that miR-199a-5p mimics obviously reduced the expression of CD31 in HUVECs, while the inhibitor distinctly enhanced the expression of CD31 (Figure 5E). In summary, miR-199a-5p suppressed the proliferation, migration and neovascularization of HUVECs.
VEGFA siRNA attenuates the proliferation, migration and neovascularization of HUVECs induced by a miR-199a-5p-inhibitor
As VEGFA was proven to be the target gene of miR-199a-5p, VEGFA siRNA was utilized for the miR-199a-5p rescue experiment. The EdU assay revealed that the promotion of HUVEC proliferation by the miR-199a-5p inhibitor could be attenuated by VEGFA-siRNA (Figure 6A and E). In addition, the expression of CD34 and CD31 was significantly decreased in miR-199a-5p inhibitor and VEGFA siRNA transfected HUVECs compared with miR-199a-5p inhibitor transfected HUVECs (Figure 6B, D and E). These data suggested that the promotion of neovascularization by the miR-199a-5p inhibitor in HUVECs could be weakened by VEGFA-siRNA. Moreover, Transwell assay results indicated that VEGFA siRNA obviously reduced the migration of HUVECs transfected with miR-199a-5p-inhibitor (Figure 6C and E). Above all, suppression of the expression of VEGFA weakened the promotion of HUVEC proliferation, migration and neovascularization of induced by miR-199a-5p inhibitor, indicating that the proliferation and migration of endothelial cells promoted by miR-199a-5p was mediated by VEGFA.
MiR-199a-5p induces osteosarcoma tumorigenesis and angiogenesis in vivo
To further investigate the intrinsic role of miR-199a-5p in osteosarcoma tumorigenesis and angiogenesis, in vivo xenograft models were established. The mouse group injected with Lenti-miR-199a-5p had a lower proliferation rate, and formed evidently smaller tumors than the Lenti-miR-199a-5p negative control (Lenti-miR-199a-5p-NC) group, as shown in Figures 7A, B and E1. The tumor weight at the time of death in the Lenti-miR-199a-5p group was 194.24± 122.43 g, which was significantly less than that in the Lenti-miR-199a-5p-NC group (573.06± 343.14 g). These data suggest that Lenti-miR-199a-5p reduces the tumor volume and growth rate of osteosarcoma cells in vivo.
To further clarify the effect of miR-199a-5p on tumor angiogenesis in vivo, H&E staining was used to detect the number of vascular tissues in tumor sections. The results indicated that the number of tumor vascular tissues in the Lenti-miR-199a-5p group decreased (Figure 7B). In addition, western blotting assays and IHC were performed to determine whether VEGFA expression was suppressed by miR-199a-5p in vivo, we found that the expression of VEGFA was largely inhibited by Lenti-miR-199a-5p, and later, VEGFR2, CD31, and CD34 were also detected. VEGFA and CD31 staining revealed that HUVECs were significantly decreased in mice of the Lenti-miR-199a-5p group compared with those of the Lenti-miR-199a-5p-NC group, indicating that miR-199a-5p attenuated angiogenesis (Figures 7D and E2-3). Taken together, these results indicate that the overexpression of miR-199a-5p via lentiRNA treatment leads to an inhibitory growth and tumor angiogenesis in osteosarcoma.