IGF2BP3 stabilizes AGAP2-AS1 through m6A modification in ccRCC cells
In total, 726 upregulated genes and 871 downregulated genes between ccRCC tumor tissues and normal tissues were identified from GSE36895 (Fig. 1A). Additionally, 20 widely recognized m6A regulators were obtained from published studies15, 17, 24. Venn diagram showed that IGF2BP3 was the only intersection between the differentially expressed genes from GSE36895 and the 20 m6A regulators (Fig. 1B). In addition, we further analyzed the TCGA data in the ENCORI database and found that IGF2BP3 was highly expressed in ccRCC (Fig. 1C). Moreover, we screened the differentially expressed lncRNAs positively correlated with IGF2BP3 from GSE36895 by Pearson analysis and compared them with the IGF2BP3 substrates obtained from the m6a2Target database. The AGAP2-AS1 was finally selected for further study. AGAP2-AS1 was found to be positively correlated with IGF2BP3 (r = 0.654, p < 0.001) (Fig. 1D). It is shown that AGAP2-AS1 is significantly related to the poor prognosis of ccRCC patients9. Therefore, we hypothesize that IGF2BP3 may stabilize AGAP2-AS1 through m6A modification in ccRCC.
To verify this hypothesis, the expression of AGAP2-AS1 and IGF2BP3 was first detected in 50 clinical samples of ccRCC. The results showed that, compared with normal tissues, the expression of AGAP2-AS1 (t = 13.95, p < 0.001) and IGF2BP3 (t = 16.56, p < 0.001) increased significantly in ccRCC tumor tissues (Fig. 1E). Their expressions were further verified in cell lines, which showed that compared with human normal kidney cell line HK-2, the levels of IGF2BP3 (p < 0.001) and AGAP2-AS1 (F = 48.38, p < 0.001) were significantly higher expressed in ccRCC cell lines of 786-O and ACHN (Fig. 1F). In addition, we used meRIP to detect whether AGAP2-AS1 is modified by m6A. The results showed that compared with IgG antibody, m6A antibody significantly enriched AGAP2-AS1 (t = 15.25, p < 0.001; t = 16.33, p < 0.001) (Fig. 1G).
The KH3-4 domain is the core region where IGF2BP binds to m6A 43. We constructed an IGF2BP3 plasmid with a mutation in the KH3-4 domain. Real-time PCR showed that overexpression of wild-type IGF2BP3 significantly increased the expression of AGAP2-AS1 (95%CIF = 104.4, p < 0.001), while overexpression of IGF2BP3-KH3/4-Mut did not significantly affect AGAP2-AS1 (F = 0.5678, p > 0.05) (Fig. 1H). There was no significant change in AGAP2-AS1 after overexpression of IGF2BP3 and the use of methylation inhibitor DAA (F = 0.5678, p > 0.05) (Fig. 1H). Meanwhile, the half-life test results showed that overexpression of wild-type IGF2BP3 (p < 0.001) significantly prolonged the half-life of AGAP2-AS1 in 786-O (Fig. 1I) and ACHN cells (Fig. 1J). Furthermore, RIP showed that IGF2BP3 antibody (t = 13.71, p < 0.001; t = 14.86, p < 0.001) enriched AGAP2-AS1, while DAA treatment eliminated this effect (t = 1.631, p > 0.05; t = 1.950, p > 0.05) (Fig. 1K).
Additionally, the potential binding sites of AGAP2-AS1 and IGF2BP3 was predicted through TargetScan and StarBase databases. Two IGF2BP3 binding sites (TGGAC) were obtained by contrasting the AGAP2-AS1 sequence, and then the m6A modification was abolished by replacing the adenosine base in the m6A consensus sequence with T (Fig. 1L). Based on this, we constructed corresponding luciferase reporter plasmids and then dual luciferase reporter assay was performed. The results showed that the co-expression of IGF2BP and AGAP2-AS1-WT significantly increased luciferase activity (t = 9.34, p < 0.001; t = 12.61, p < 0.001) in 786-O (Fig. 1M) and ACHN cells (Fig. 1N).
Taken together, the above results indicate that IGF2BP3 stabilizes the expression of AGAP2-AS1 through m6A modification in ccRCC cells.
AGAP2-AS1 promotes the proliferation, invasion, and migration of ccRCC cells and affects the polarization of macrophages to M2
Through lentiviral transfection, we constructed cell lines with AGAP2-AS1 knockdown. The real-time PCR results showed that sh-AGAP2-AS1-1 (p < 0.001) had a higher knockdown efficiency (Fig. 2A and 2B) and thus sh-AGAP2-AS1-1 was used for further analysis. CCK-8 assay revealed that, the cell viability was significantly reduced after silencing AGAP2-AS1 (p < 0.001) (Fig. 2C and 2D). However, cell viability increased significantly after overexpression of AGAP2-AS1 (p < 0.001) (Fig. 2C and 2D). EdU cell proliferation assay showed that the number of EdU-positive cells was significantly reduced after silencing AGAP2-AS1 (p < 0.001), while the number of EdU-positive cells was significantly increased after overexpression of AGAP2-AS1 (p < 0.001) (Fig. 2E and 2F). These results indicate that AGAP2-AS1 promotes ccRCC cell proliferation.
Transwell invasion assay and wound healing assay also demonstrated that the ability of cell invasion and migration was significantly reduced after silencing AGAP2-AS1 (p < 0.001), while it was significantly increased after overexpression of AGAP2-AS1 (p < 0.001) (Fig. 2G-2J), suggesting that AGAP2-AS1 promotes the invasion and migration of ccRCC cells.
In addition, after using PMA to induce THP-1 to differentiate into M0 macrophages, we co-cultured AGAP2-AS1 knockdown ccRCC cells with M0 macrophages. Real-time PCR showed that the expression of M1 polarization markers iNOS (p < 0.001) and TNF-α (p < 0.001) increased significantly after being co-cultured with the supernatant from the AGAP2-AS1 knockdown ccRCC cells, while M2 polarization markers of Arg-1 (p < 0.001) and IL-10 (p < 0.001) was significantly reduced (Fig. 2K and 2L). On the contrary, after co-culture with supernatant from ccRCC cells overexpressing AGAP2-AS1, iNOS (p < 0.001) and TNF-α (p < 0.001) were significantly reduced, while the expression of Arg-1 (p < 0.001) and IL-10 (P < 0.001) increased significantly (Fig. 2K and 2L). Similarly, flow cytometry showed that the number of CD163 + macrophages increased after co-culture with the supernatant of RCC cells overexpressing AGAP2-AS1 (Fig. 2M and 2N). These indicate that AGAP2-AS1 promotes the polarization of M0 macrophages to M2.
AGAP2-AS1 up-regulates THBS2 expression by binding to miR-9-5p
LncRNAs function mainly by affecting the expression of downstream miRNAs44. We then predicted the miRNAs that bind to AGAP2-AS1 through the LncBase database and intersected them with the low-expressed miRNAs in the GSE37989 dataset. Finally, we screened out miR-9-5p as the target of AGAP2-AS1 (Fig. 3A). In addition, we further predicted the binding site of AGAP2-AS1 and miR-9-5p (Fig. 3B). The expression level of miR-9-5p in 50 ccRCC tumor tissues was significantly reduced compared with that in normal tissues (t = 12.46, p < 0.001) (Fig. 3C). Dual luciferase reporter assay showed that compared with mimic NC, miR-9-5p mimic significantly reduced the luciferase activity of wild-type AGAP2-AS1 (t = 14.41, p < 0.001), but there was no significant difference for mutant AGAP2-AS1 (T = 0.2415, p > 0.05) (Fig. 3D). RNA pull down results showed that compared with Bio-MUT-miR-9-5a and Bio-NC groups, AGAP2-AS1 bound to Bio-WT-miR-9-5a significantly increased (p < 0.001) (Fig. 3E). These results indicate that AGAP2-AS1 could bind miR-9-5p.
Furthermore, real-time PCR was used to detect the expression of miR-9-5p in cells with AGAP2-AS1 silencing or overexpression. The results showed that compared with the control group, the expression of miR-9-5p was significantly increased after silencing AGAP2-AS1 (p < 0.001), while the expression of miR-9-5p was significantly decreased after AGAP2-AS1 overexpression (p < 0.001) (Fig. 3F and 3G). Meanwhile, FISH displayed that AGAP2-AS1 and miR-9-5p co-localized in cells (Fig. 3H).
Finally, we predicted the target genes of miR-9-5p through the TargetScan database and performed KEGG analysis. The results showed that the target genes of miR-9-5p were enriched in the PI3K-Akt signaling pathway (supplementary Figure S1). Previous studies have revealed that THBS2 gene silencing inhibits the PI3K-Akt signaling pathway, which can affect macrophage polarization 45, 46. Here, by GSE40435 dataset analysis, we found that THBS2 was highly expressed in ccRCC tumor tissues (Fig. 3I). The high expression of THBS2 was further verified in clinical samples of ccRCC (t = 16.35, p < 0.001) (Fig. 3J).
The binding site between miR-9-5p and THBS2 was predicted by the TargetScan (Fig. 3K). The results of the dual luciferase reporter assay showed that miR-9-5p mimic significantly reduced the luciferase activity of wild-type THBS2 (t = 29.88, p < 0.001), but for mutant THBS2, there was no significant difference (t = 0.09252, P > 0.05) (Fig. 3L). We also detected THBS2 expression in each group of cells by real-time PCR. The results showed that the expression of THBS2 in the miR-9-5p mimic group was significantly reduced compared with the mimic NC group (p < 0.001) (Fig. 3M and 3N). However, compared with the inhibitor NC group, the expression of THBS2 in the miR-9-5p inhibitor group was significantly increased (p < 0.001) (Fig. 3M and 3N).
Furthermore, the cells were treated with AGAP2-AS1 overexpression and miR-9-5p mimics and then the expression of THBS2 was detected. We found that compared with the mimic NC + oe-NC group, the expression of THBS2 in the miR-9-5p mimic + oe-NC group was significantly reduced (p < 0.001) (Fig. 3O and 3P). However, compared with the miR-9-5p mimic + oe-NC group, the THBS2 expression in the miR-9-5p mimic + oe-AGAP2-AS1 group was significantly increased (p < 0.001) (Fig. 3O and 3P). Moreover, Western blot detection of THBS2 protein expression was consistent with real-time PCR results (p < 0.001) (Fig. 3Q).
The above results indicate that AGAP2-AS1 up-regulates the expression of THBS2 by competitively binding to miR-9-5p.
THBS2 promotes M2 polarization of macrophages by activating the PI3K/Akt pathway
We further silenced THBS2 expression in ccRCC cells. The sh-THBS2-1 had a better silencing efficiency (p < 0.001) (Fig. 4A and 4B) and thus was used in further analysis. CCK-8 assay showed that cell viability decreased significantly after silecening THBS2 (p < 0.001), but increased significantly after overexpression of THBS2 (p < 0.001) (Fig. 4C and 4D). In addition, after co-culturing THBS2 knockdown ccRCC cells with macrophages, we used Western blot to detect the expression of PI3K/AKT signaling pathway-related proteins in macrophages. The results showed that the expression of p-PI3K (p < 0.001) and p-AKT (p < 0.001) in the sh-THBS2 + M (PMA + THP-1) group was significantly reduced compared with the sh-NC + M group (Fig. 4E and 4F). Compared with the oe-NC + M group, the expression of p-PI3K (p < 0.001) and p-AKT (p < 0.001) in the oe-THBS2 + M group increased significantly (Fig. 4E and 4F). Moreover, compared with the sh-NC + M group, the expression of M1 polarization markers iNOS (p < 0.001) and TNF-α (p < 0.001) in the sh-THBS2 + M group increased significantly, while the expression of M2 polarization markers Arg-1 (p < 0.001) and IL-10 (p < 0.001) significantly reduced (Fig. 4G and 4H). In the oe-NC + M group, the expression of iNOS (p < 0.001) and TNF-α (p < 0.001) in the oe-THBS2 + M group were significantly reduced, whereas the expression of Arg-1 (p < 0.001) and IL-10 (p < 0.001) were significantly increased (Fig. 4G and 4H). The above results indicate that THBS2 may affect the M1/M2 polarization of macrophages.
To further explore whether THBS2 affects the polarization of macrophages by affecting the PI3K/AKT signaling pathway, we used the PI3K/AKT inhibitor LY294002. Compared with the oe-THBS2 + DMSO + M group, the expression of p-AKT (t = 11.54, p < 0.001; t = 12.94, p < 0.001) in the oe-THBS2 + LY294002 + M group increased significantly (Fig. 4I and 4J). For analysis of M1/M2 polarization, real-time PCR showed that in oe-THBS2 + LY294002 + M group, iNOS (t = 10.59, p < 0.001; t = 10.77, p < 0.001) and TNF-α (t = 8.728, p < 0.001; t = 8.998, p < 0.001) levels increased significantly, while Arg-1 (t = 11.30, p < 0.001; 10.73, p < 0.001) and IL-10 (t = 12.53, p < 0.001; t = 16.21, p < 0.001) levels decreased significantly (Fig. 4K and 4L). Meanwhile, flow cytometry showed that compared with the oe-THBS2 + DMSO group, the number of CD163 + cells in the oe-THBS2 + LY294002 group was reduced (Fig. 4M and 4N). These results suggest that THBS2 may promote M2 polarization of macrophages by activating the PI3K/Akt signaling pathway.
AGAP2-AS1/miR-9-5p/THBS2 induces M2 polarization of macrophages and promotes the invasion and migration of ccRCC cells
We knocked down AGAP2-AS1 and overexpressed THBS2 in ccRCC cells to further verify the role of AGAP2-AS1/miR-9-5p/THBS2 axis in biological functions of ccRCC cells. We found that in the sh-AGAP2-AS1 + oe-NC group, the expression of AGAP2-AS1 and THBS2 was significantly reduced (p < 0.001), and that of miR-9-5p was significantly increased (p < 0.001), compared with sh-NC + oe-NC group (Fig. 5A and 5B). Compared with the sh-AGAP2-AS1 + oe-NC group, the sh-AGAP2-AS1 + oe-THBS2 group had significantly reduced AGAP2-AS1 (p < 0.001) and significantly increased miR-9-5p and THBS2 (p < 0.001) (Fig. 5A and 5B).
Cell proliferation by EdU assay showed that the sh-AGAP2-AS1 + oe-NC group had significantly reduced cell proliferation ability than the sh-NC + oe-NC group (p < 0.001) (Fig. 5C and 5D). The sh-AGAP2-AS1 + oe-THBS2 group had significantly increased cell proliferation ability than sh-AGAP2-AS1 + oe-NC group (p < 0.001) (Fig. 5C and 5D). Furthermore, Transwell invasion assay and wound healing assay showed that the cell invasion and migration ability of the sh-AGAP2-AS1 + oe-NC group was significantly reduced than the sh-NC + oe-NC group (p < 0.001) (Fig. 5E-5H). However, the cell invasion and migration ability of the sh-AGAP2-AS1 + oe-THBS2 group was significantly increased than sh-AGAP2-AS1 + oe-NC group (p < 0.001) (Fig. 5E-5H). The above data indicates that the AGAP2-AS1/miR-9-5p/THBS2 axis may promote the proliferation, invasion and migration of ccRCC cells.
Next, we co-cultured macrophages with ccRCC cells with AGAP2-AS1 konckdown and THBS2 overexpression. After that, changes of PI3K/Akt pathway key proteins and macrophage polarization were analyzed. As shown in Fig. 5I and 5J, compared to the sh-NC + oe-NC + M group, the expression of p-PI3K (p < 0.001) and p-AKT (p < 0.001) in the sh-AGAP2-AS1 + oe-NC + M group was significantly reduced. Their levels were significantly increased in the sh-AGAP2-AS1 + oe-THBS2 + M group than those in the sh-AGAP2-AS1 + oe-NC + M group (p < 0.001). Analysis of macrophage polarization markers showed that the expressions of iNOS (p < 0.001) and TNF-α (p < 0.001) increased significantly while those of Arg-1 and IL-10 decreased significantly in the sh-AGAP2-AS1 + oe-NC + M group than the sh-NC + oe-NC + M group (p < 0.001) (Fig. 5K and 5L). Compared with the sh-AGAP2-AS1 + oe-NC + M group, the expression of iNOS (p < 0.001) and TNF-α (p < 0.001) in the sh-AGAP2-AS1 + oe-THBS2 + M group was significantly reduced, but the Arg-1 (p < 0.001) and IL-10 (p < 0.001) expression increased significantly (Fig. 5K and 5L). Moreover, flow cytometry results showed that the number of CD163 + cells in the sh-AGAP2-AS1 + oe-NC + M group was lower than that in the sh-NC + oe-NC + M group. Compared with the sh-AGAP2-AS1 + oe-NC + M group, the number of CD163 + cells in the sh-AGAP2-AS1 + oe-THBS2 + M group was increased (Fig. 5M and 5N). The results demostrate that overexpression of THBS2 may reverse the decrease in M2 polarization of macrophages caused by silencing AGAP2-AS1.
Collectively, the AGAP2-AS1/miR-9-5p/THBS2 axis may induce M2 polarization of macrophages and promote the development of ccRCC cells.
AGAP2-AS1/miR-9-5p/THBS2 axis induces M2 polarization of macrophages and promotes the development of ccRCC in vivo
After intervention, ACHN cells were inoculated into nude mice to establish the xenograft model. The tumor growth was observed. As shown in Fig. 6A and 6B, compared with the sh-NC + oe-NC group, the tumor volume and weight of the sh-AGAP2-AS1 + oe-NC group were significantly reduced (p < 0.001). However, the tumor volume and weight of the sh-AGAP2-AS1 + oe-THBS2 group increased significantly than sh-AGAP2-AS1 + oe-NC group (p < 0.001). Moreover, we analyzed the expression of AGAP2-AS1, miR-9-5p and THBS2 in the tumor using real-time PCR and found reduced expression of AGAP2-AS1 and elevated expression of miR-9-5p in the sh-AGAP2-AS1 + oe-NC group compared to the sh-NC + oe-NC group. However, THBS2 expression was restored in the sh-AGAP2-AS1 + oe-THBS2 group compared to the sh-AGAP2-AS1 + oe-NC group (Fig. 6C). Immunohistochemistry was used to detect the expression of Ki67 in the tumor section. The results showed that compared with the sh-NC + oe-NC group, the Ki67-positive cells in nude mice tumors in the sh-AGAP2-AS1 + oe-NC group were significantly reduced (p < 0.001). Compared with the sh-AGAP2-AS1 + oe-NC group, there was a significant increase in Ki67-positive cells in nude mice tumors in the sh-AGAP2-AS1 + oe-THBS2 group (p < 0.001) (Fig. 6D).
For analysis of PI3K/Akt pathway key proteins in peritoneal macrophages, the expression of p-PI3K (p < 0.001) and p-AKT (p < 0.001) in the sh-AGAP2-AS1 + oe-NC group was significantly reduced than sh-NC + oe-NC group. However, there were increased expressions of p-PI3K (p < 0.001) and p-AKT (p < 0.001) in the sh-AGAP2-AS1 + oe-THBS2 group than the sh-AGAP2-AS1 + oe-NC group (Fig. 6E). We also used real-time PCR to analyze the expression of iNOS, TNF-α, Arg-1 and IL-10. The results showed that compared with the sh-NC + oe-NC group, there were significantly increased levels of iNOS (p < 0.001) and TNF-α (p < 0.001) while significantly decreased levels of Arg-1 (p < 0.001) and IL-10 (p < 0.001) in tumors of the sh-AGAP2-AS1 + oe-NC group. Compared with the sh-AGAP2-AS1 + oe-NC group, the changes of these markers were reversed in the sh-AGAP2-AS1 + oe-THBS2 group (Fig. 6F). Moreover, as revealed by flow cytometry, compared with the sh-NC + oe-NC group, the sh-AGAP2-AS1 + oe-NC group had fewer CD163 + cells, while the sh-AGAP2-AS1 + oe-THBS2 group had increased CD163 + cells (Fig. 6G).
Finally, H&E staining showed that compared with the sh-NC + oe-NC group, the sh-AGAP2-AS1 + oe-NC group had significantly fewer lung metastasis tumor nodules. The tumor nodules metastasized to the lung tissue of nude mice in the sh-AGAP2-AS1 + oe-THBS2 group increased significantly (p < 0.001) (Fig. 6H).
In summary, consistent with in vitro experiments, the AGAP2-AS1/miR-9-5p/THBS2 axis may also induce macrophage M2 polarization and promote the development of ccRCC in vivo.