CircNEIL3 was significantly upregulated in glioma tissues

To identify potential oncogenic circRNAs involved in the tumorigenesis and progression of glioma, we analysed the expression profiles of circRNAs in 39 glioma tissues (11 GBM cases and 28 low-grade glioma (LGG) cases) and 8 normal brain tissues (NBTs) using RNA sequencing (RNA-seq). Volcano plots showed systemic differences in circRNA expression between GBM and LGG, as well as between LGG and NBTs (Fig. 1A, B, Log2Fold Change (FC) ≥ 1, Padj≤0.05). We further analysed the differentially upregulated circRNAs in glioma tissues (Log2FC ≥ 4, Padj≤0.05) and identified hsa_circ_0001460, the only circRNA that was upregulated with increasing glioma grade (Fig. 1C). We named hsa_circ_0001460 as circNEIL3, which is generated from NEIL3. CircNEIL3 was markedly upregulated in glioma tissues compared with NBTs, and its expression increased with the increasing glioma grade (Fig. 1D). ROC (Receiver operating characteristic) curve was used to assess the diagnostic efficacy of circNEIL3 in glioma grade, the AUC (area under the curve) of which was 0.719, suggesting that circNEIL3 could predict poor prognosis in glioma patients (Fig. 1E).

Fig. 1
figure1

CircRNA expression profiles in glioma and characterization of circNEIL3. A Volcano plots of circRNAs that were differentially expressed between LGG tissues and NBTs. B Volcano plots of circRNAs that were differentially expressed between GBM samples and LGG samples. C Venn diagram showing the overlapping upregulated circRNAs (Log2 FC > 4) between LGG tissues and NBTs and between GBM and LGG tissues. D Relative expression of circNEIL3 in NBTs, LGG tissues and GBM tissues detected by high-throughput sequencing. E An ROC curve was used to assess the diagnostic value of circNEIL3 for glioma. F Schematic illustration of the genomic location and back splicing of circNEIL3 with the splicing site validated by Sanger sequencing. G RT-PCR with divergent and convergent primers and agarose gel electrophoresis analysis were performed to detect the presence of circNEIL3 and its maternal gene NEIL3 in cDNA and gDNA samples from GBM cells. H Actinomycin D treatment was used to evaluate the stability of circNEIL3 and NEIL3 mRNA in GBM cells. I RNase treatment was used to evaluate the stability of circNEIL3 and NEIL3 mRNA in GBM cells. J Nuclear-cytoplasmic fractionation assays showed that circNEIL3 was mostly localized in the cytoplasm of GBM cells. K RNA FISH assays showing the cellular localization of circNEIL3 in GBM cells. The circNEIL3 probe was labelled with Cy3 (red), while nuclei were stained with DAPI (blue). Scale bar, 25 μm. All data are presented as the means ± SD, ns, P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Through the UCSC Genomics Institute Bioinformatics site (http://genome.ucsc.edu/), we identified that circNEIL3 was backspliced between exon 8 and exon 9 of the NEIL3 gene with a length of 596 nt (Fig. 1)F. The back-splice junction site of circNEIL3 was amplified using divergent primers and was confirmed by Sanger sequencing (Fig. 1F). Furthermore, we designed divergent and convergent primers to amplify circNEIL3 and its linear form. The results of agarose gel electrophoresis analysis of the RT-qPCR products showed that circNEIL3 could only be amplified from cDNA, while its linear form could be amplified from cDNA and gDNA (Fig. 1G). Moreover, we confirmed that circNEIL3 was more stable than NEIL3 by treatment with RNase R and actinomycin D (Fig. 1H, I). The functions of circRNAs are mostly related to their intracellular localization. We then performed nuclear-cytoplasmic fractionation experiments and FISH assays. The results showed that circNEIL3 was mostly localized to the cytoplasm (Fig. 1J, K). In summary, these results indicated that circNEIL3 was significantly upregulated in glioma tissues, and its expression was highest in GBM, suggesting that it is involved in tumorigenesis and the malignant progression of glioma.

CircNEIL3 promoted glioma tumorigenesis in vitro and vivo

To investigate the potential biological role of circNEIL3 in the progression of glioma, we performed KEGG enrichment analysis on genes that were significantly positively correlated with circNEIL3 expression (Additional file Table S1) based on our transcriptome sequencing data. As shown in Fig. S1A, the genes were significantly enriched in many pathways involved in tumour pathogenesis, including cell proliferation regulation pathways (such as cell cycle and DNA replication), pathways associated with tumour invasion and migration (such as focal adhesion and ECM-receptor interaction), and classical signalling pathways involved in tumorigenesis, including p53 signalling pathways and pathways in cancer. Gene set enrichment analysis (GSEA) also showed that these pathways were significantly enriched in the high circNEIL3 expression group (Fig. S1B). Furthermore, to explore the biological behaviours among these distinct circNEIL3 expression samples, we used a single-sample GSEA (ssGSEA) algorithm to estimate pathway enrichment scores for each sample (see Methods). As shown in Fig. S1C, compared to the low circNEIL3 expression group, the high expression group was more markedly enriched in carcinogenic signalling pathways.

Since circNEIL3 was potentially involved in cell proliferation, invasion and migration, we further investigated the function of circNEIL3 in cell behaviour. Consistent with our bioinformatics analysis results, we confirmed that downregulation of circNEIL3 significantly inhibited cell proliferation, migration, and invasion. Overexpression of circNEIL3 remarkably promoted these cellular behaviours in vitro (Fig. 2A-D, Fig. S2A-F). We also demonstrated that circNEIL3 significantly promoted the tumorsphere expansion (Fig. S2G-H) and sphere formation ability of the GSCs (Figs. S2I-J). Moreover, in vivo experiments revealed that circNEIL3 downregulation could markedly inhibit the tumour growth and invasiveness and prolong the survival of tumour-bearing mice, while circNEIL3 overexpression elicited the opposite effects (Fig. 2E, F, and Fig. S2K). In addition, immunohistochemistry (IHC) of the excised tumour sections indicated that the expression of Ki67 (a proliferation marker) and CD44 (an invasiveness marker) in circNEIL3-knockdown tumour tissues was lower than that in the vector group, while circNEIL3 overexpression showed the opposite effects (Fig. 2G). Taken together, our data suggest that circNEIL3 is functionally important in regulating tumorigenesis and glioma progression.

Fig. 2
figure2

CircNEIL3 promotes proliferation and metastasis of GBM cells in vitro and vivo. CCK8 assays showing the proliferation ability of GBM cells transfected with (A) sh-NC or sh-circNEIL3, and (B) ov-NC or ov-NEIL3, n = 3. Representative transwell migration and invasion assays showing the migration and invasion ability of GBM cells transfected with (C) sh-NC or sh-circNEIL3, and (D) ov-NC or ov-NEIL3. Quantification histogram represented relative cell numbers, n = 3, scale bar, 200um. E Up, bioluminescent image showing the tumor size for animals in different groups in the indicated time. n = 5 for each group. Down, statistical analysis of bioluminescent tracking plots. F Kaplan-Meier survival curves for animals in different groups, n = 5 for each group. G Represented CD44 and KI67 immunohistochemistry images for a subgroup of animal sacrificed simultaneously in each group, n = 5 for each group, scale bar, 20um. All data are presented as the means ± SD, **p < 0.01, ***p < 0.001, ****p < 0.0001

EWSR1 promotes the biogenesis of circNEIL3 in glioma

The mechanism of cyclization of circRNAs is regulated by RNA-binding proteins (RBPs) binding to the upstream and downstream regions of pre-mRNA [23, 24]. We predicted three binding sites of EWSR1 in the upstream and downstream regions of circNEIL3 called sequence 1, sequence 2 and sequence 3 (Fig. 3A) by using the CircInteractome database (https://circinteractome.nia.nih.gov/) and starBase 3.0 (http://starbase.sysu.edu.cn/index.php) database (Fig. S3A, B). We found that the EWSR1 expression level increased with the grade of glioma, and patients with high expression of EWSR1 had a worse prognosis than those with low expression (Fig. 3B, Fig. S3D, E). Furthermore, we validated that knocking down EWSR1 could inhibit GBM cell proliferation, invasion and migration (Fig. 3C, D, and Fig. S3F). In our own database, we found that circNEIL3 is positively correlated with EWSR1 (Fig. 3E), and the expression of circNEIL3 is significantly deceased in U251 and A172 cells with EWSR knockdown (Fig. 3F). A recent study indicated that EWSR1 could be more active in hypoxic environments [25]. Thus, we subjected U251 and A172 cells to hypoxic conditions for 0 h, 6 h, 12 h and 24 h, and the expression of EWSR1 and circNEIL3 were upregulated as the exposure time increased (Fig. 3G, H). RIP-qPCR assays confirmed that EWSR1 could bind to sequence 1 and sequence 3 of NEIL3 pre-mRNA, and the binding ability was significantly upregulated under hypoxic conditions (Fig. 3I). To further confirm that the observed EWSR1-mediated phenotypes were facilitated by the dysregulation of circNEIL3 expression, we performed functional rescue assays. As shown in Fig. 3J-L, EWSR1 overexpression induced increases in proliferation, migration and invasion in U251 and A172 cells, which could be reversed by circNEIL3 downregulation. Our data demonstrated that circNEIL3 could be cyclized by EWSR1 and that this process was more active under hypoxic conditions.

Fig. 3
figure3

EWSR1 promotes the biogenesis of circNEIL3. A The binding sites of EWSR1 were predicted in the upstream and downstream region of NEIL3 pre-mRNA transcript using the starbase and circinteractome database. B Relative expression of EWSR1 in our local NBTs, LGG and GBM tissues detected by high-throughput sequencing. C CCK8 assays showing the proliferation ability of GBM cells transfected with sh-NC or sh-circNEIL3, n = 3. D Representative transwell migration and invasion assays showing the migration and invasion ability of GBM cells transfected with sh-NC or sh-circNEIL3. Quantification histogram represented relative cell numbers, n = 3, scale bar, 200um. E The correlation between circNEIL3 and EWSR1 expression in our local glioma high-throughput sequencing data (n = 39). F qRT-PCR assays showed the expression of circNEIL3 in GBM cells transfected with sh NC or sh EWSR1, n = 3. G Western blot assays showing the expression of EWSR1 under hypoxia treatment for 0, 6, 12, 24 h. H qRT-PCR assays showing the expression of circNEIL3 under hypoxia treatment for 0, 6, 12, 24 h, n = 3. I RIP and qRT-PCR assays were performed to verify the putative binding site of NEIL3 pre-mRNA with EWSR1 under normal and hypoxia environment, n = 3. (J) CCK8 assays, (K) Representative EDU assays showing the proliferation ability of GBM cells co-transfected with ov-NC, ov-EWSR1 and si-circNEIL3 as indicated. Quantification histogram represented relative cell numbers, n = 3, scale bar = 50um. L Representative transwell migration and invasion assays showing the migration and invasion ability of GBM cells co-transfected with ov-NC, ov-EWSR1 and si-circNEIL3 as indicated, Quantification histogram represented relative cell numbers, n = 3, scale bar, 200um. All data are presented as the means ± SD, and the ov-EWSR1 group is indicated as the control in (J-L), ns, P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

CircNEIL3 physically interacts with IGF2BP3 and inhibits the ubiquitin/proteasome-mediated degradation of IGF2BP3

Studies have shown that cytoplasm-retained circRNAs usually perform their functions by acting as ceRNAs or scaffolds for RBPs [5, 6]. To explore the molecular mechanism of the circNEIL3-induced progression of GBM cells, we first performed RNA pull-down assays and subsequent mass spectrometry assays to explore potential proteins binding to circNEIL3. A total of 50 proteins interacting with circNEIL3 were identified, which did not include AGO2, excluding the possibility that circNEIL3 functions as a ceRNA (Additional file Table S2). We then intersected our mass spectrometry data with the RBPs predicted in the CircInteractome database. We found that circNEIL3 bound to IGF2BP3 protein (Fig. 4A, Fig. S4A). Then, we further confirmed the interaction between circNEIL3 and IGF2BP3 by RNA pull-down and RIP-qPCR assays (Fig. 4B, C). We also performed RNA FISH-immunofluorescence (FISH-IF) analysis and found that circNEIL3 colocalized with IGF2BP3 in the cytoplasm (Fig. 4D). IGF2BP3 consists of 2 RNA recognition motifs (RRMs) and 4 K homology (KH) domains; therefore, we established six FLAG-tagged vectors, which were confirmed by Western blot, to test which domain interacts with circNEIL3 (Fig. 4E, F). The RIP-qPCR assay showed that circNEIL3 mostly bound to the region of KH3–4, suggesting that the KH3–4 di-domain is responsible for recruiting circNEIL3 (Fig. 4G). We then searched for the motif necessary for IGF2BP3 recruitment in circNEIL3. Moreover, to identify the sequence in circNEIL3 to which IGF2BP3 binds, we used the CircInteractome database and found that two sequences may bind to IGF2BP3 (Fig. S4B, the yellow filling sequences). Interestingly, a study by Markus et al. [26] showed that the sequence CAUH (H = A, U or C) was the only one that could be recognized by IGF2BP3. Therefore, we reviewed the sequence found by the CircInteractome database and found three possible binding sites (Fig. S4C red font sequences). Then, we designed three mutant sites and finally observed that IGF2BP3 binds to the third site of circNEIL3 (Fig. 4H, Fig. S4D red font sequences). These results suggested that circNEIL3 physically interacts with IGF2BP3 in the cytoplasm.

Fig. 4
figure4

CircNEIL3 physically interacts with IGF2BP3 and inhibits its ubiquitination. A Different protein bands detected by silver stain for mass spectrometry of the circNEIL3-protein complex pulled down by circNEIL3 junction sense or anti-sense in U251 cells, the arrow points to IGF2BP3 band. B RNA pull down and western blot assays showing the interaction between IGF2BP3 with circNEIL3 in U251 and A172 cells. GAPDH was used as a negative control. C RIP and qRT-PCR assays showing the interaction between IGF2BP3 with circNEIL3 in GBM cells, using Igg and IGF2BP3 antibodies, n = 3. D IF-Fish assay showing co-localization of circNEIL3 (red) with IGFB2P3 proteins (green) in U251 and A172 cells. Scale bar, 25 μm. E Schematic structures of IGF2BP3 proteins and five mutants (Delet KH1–2, Delet KH3–4, RRM 1–2, KH1–2, KH3–4). F Western blot assay showing the full length of IGF2BP3 and its mutants. G RIP and qRT-PCR assays showed that the KH3–4 fragment of IGF2BP3 was the region responsible for circNEIL3 binding. H RNA pull-down and western blot assays showed that the interaction site on circNEIL3 with IGF2BP3. I The mRNA expression of IGF2BP3 was detected by qRT-PCR in GBM cells transfected with ov-NC or ov-circNEIL3, n = 3. J Western blot assays showing the protein levels of IGF2BP3 and downstream genes, including CD44, CDK4, CDK6 and c-MYC in GBM cells transfected with si-NC or si-circNEIL3, and ov-NC or ov-circNEIL3. K Western blot assays showing the protein levels of IGF2BP3 in GBM cells transfected with ov-NC or ov-circNEIL3, treated with 20 μg/ml CHX for the indicated periods of time. L Co-IP assay showing the ubiquitination modification level of IGF2BP3 in GBM cells transfected with ov-NC or ov-circNEIL3. Ub ubiquitin. M Co-IP assay showing the ubiquitination modification level of IGF2BP3 in GBM cells co-transfected with ov-NC or ov-circNEIL3 and vetors expressing FLAG-tagged WT or IGF2BP3 mutants (K450) as indicated. N Conservation ability of the K405 ub site on IGF2BP3 protein. O Crystal structure of IGF2BP3 proteins with K405. All data are presented as the means ± SD, ns, P > 0.05, ***p < 0.001

We found that the expression level of IGF2BP3 increased with tumour grade, and patients with high expression had a worse prognosis than those with low expression in The Cancer Genome Atlas (TCGA) and Chinese Glioma Genome Atlas (CGGA) datasets (Fig. S5A- D). In addition, with increasing IGF2BP3 expression, the cancer-promoting signalling pathways were significantly enriched in the TCGA and CGGA datasets, as estimated via the ssGSEA algorithm (Fig. S5E, F), which is consistent with the enrichment analysis of circNEIL3 (Fig. S1C). Moreover, similar to circNEIL3 enrichment analysis, the GSEA results also showed that classical pathways involved in tumour pathogenesis were significantly enriched in the high IGF2BP3 expression group in the CGGA and TCGA datasets (Fig. S5G, H). Recently, Dixit Deobrat et al. [27] stated that IGF2BP3 played key roles in glioblastoma maintenance, promoting tumour heterogeneity, which highlights its role in the malignant progression of glioma. Furthermore, we found that circNEIL3 did not significantly change the mRNA level of IGF2BP3 (Fig. 4I) but significantly promoted its protein expression and as well as that of its downstream targets, including CDK4/6, CD44 and c-MYC (Fig. 4J), which are involved in the biological functions of cell proliferation, invasion and migration [28,29,30]. Rescue assays showed that IGF2BP3 knockout compensated for the increased proliferation, invasion and migration capacity, as well as upregulation of its downstream targets, caused by exogenous overexpression of circNEIL3 (Fig. S6A-E), indicating that circNEIL3 may regulate the malignant progression of glioma by destabilizing IGF2BP3 proteins.

Our further investigation showed that circNEIL3 overexpression enhanced the protein expression levels of IGF2BP3, extended the half-life of IGF2BP3(Fig. 4K), indicating that circNEIL3 regulated the stability of IGF2BP3 protein via proteasomal activity. Consistently, circNEIL3 overexpression decreased the ubiquitination of IGF2BP3 (Fig. 4L). We then predicted IGF2BP3 functional ubiquitylation sites via the Ubibrowser database (http://ubibrowser.ncpsb.org.cn/). Only one ubiquitinated lysine (K) residue (K450), which is located in the KH3 domain of IGF2BP3, was identified (Fig. 4N, Fig. S7A). Furtherly, we used uniport database (https://www.uniprot.org/) to predict that the K450 site of IGF2BP3 was highly conserved in six species (Fig. 4N). Moreover, we predicted the structure of IGF2BP3 through the Swiss Model Online website (https://swissmodel.expasy.org/) and visualized the K450 ubiquitination site (Fig. 4O). We then mutated the predictive site from lysine (K) to arginine (R) to confirm its role as a ubiquitination target. Immunoprecipitation (IP) results showed that the mutation of K450 significantly reduced the ubiquitination level of IGF2BP3 compared to that of WT IGF2BP3, and the enhanced ubiquitination caused by overexpression of cirNEIL3 was also inhibited in cells expressing the mutant (Fig. 4M), highlighting K450 as the major ubiquitination site of IGF2BP3. These results demonstrated that circNEIL3 enhanced the stability of IGF2BP3 protein by inhibiting ubiquitin/proteasome-dependent degradation, thereby promoting malignant progression of glioma.

CircNEIL3 stabilizes IGF2BP3 protein by preventing HECTD4-mediated ubiquitination

The literature has shown that TRIM25 mediates the ubiquitination of IGF2BPs in NSCLC [31]. However, there is no report on the E3 ubiquitin ligase that mediates the ubiquitination of IGF2BP3 in glioma. Thus, we sought to identify the E3 ligase involved in proteasome-mediated degradation of IGF2BP3 in glioma. We performed co-immunoprecipitation (co-IP) experiments and mass spectrometry analysis and found that HECTD4, a protein belonging to the HECD family of E3 ubiquitin ligases that links polyubiquitin to the target protein to promote their ubiquitination [32, 33], could bind to IGF2BP3 (Fig. S7B, and Additional file Table S3). The expression level of HECTD4 decreased with increasing tumour grade, and patients with low expression had a worse prognosis than those with high expression in the TCGA datasets (Fig. S7C, D). Moreover, we validated that HECTD4 can inhibit GBM cell proliferation, invasion and migration in vitro (Fig. 5A, B, Fig. S7E, F). To confirm that HECTD4 could regulate the ubiquitination of IGF2BP3, we performed immunofluorescence (IF) experiments and confirmed that IGF2BP3 colocalized with HECTD4 in the cytoplasm (Fig. 5C). Furthermore, we found that the protein level of IGF2BP3 was dramatically increased (Fig. 5D), while its level of ubiquitination was obviously decreased, in HECTD4-knockdown GBM cells (Fig. 5E), indicating that HECTD4 acts as an E3 ubiquitin ligase to degrade IGF2BP3 via the ubiquitin–proteasome pathway in glioma.

Fig. 5
figure5

CircNEIL3 blocked the binding between IGF2BP3 and HECTD4. A CCk8 assays showing the proliferation ability of GBM cells transfected with si-NC or si-HECTD4. n = 3. b. Representative transwell migration and invasion assays showing the migration and invasion ability of GBM cells transfected with si-NC or si-HECTD4. Quantification histogram represented relative cell numbers, n = 3, scale bar, 200 μm. C IF assays showing the co-localization of IGF2BP3 (green) with HECTD4 (red) in GBM cells. Scale bar, 25 μm. D Western blot assays showing the expression of HECTD4 and IGF2BP3 in GBM cells transfected with si-NC or si-HECTD4. E Co-IP assays showing the ub modification level in GBM cells transfected with si-NC or si-HECTD4. F Co-IP assays showing the binding between HECTD4 with IGF2BP3 proteins in GBM cells transfected with ov-NC or ov-circNEIL3. G Co-IP and western blot assays showing the interaction between HECTD4 and different IGF2BP3 domain in in 293 T cells transfected with vectors expressing the Flag-tagged FL or the truncated mutants (RRM1–2, KH1–2, KH3–4), using Flag antibodies. All data are presented as the means ± SD, **P < 0.01, ***P < 0.001, ****P < 0.0001

Mass spectrometry analysis of circNEIL3 showed that circNEIL3 did not bind to HECTD4 (Additional file Table S2), but we found that overexpression of circNEIL3 can block the binding between IGF2BP3 and HECTD4 (Fig. 5F), which explain how circNEIL3 stabilizes IGF2BP3. To further confirm that circNEIL3 can inhibit the binding of IGF2BP3 to HECTD4, we performed a Co-IP assay and observed that HECTD4 bound to the KH3–4 domains of IGF2BP3, which is the same site where circNEIL3 interacts with IGF2BP3 (Fig. 5G, and Fig. 4G). These data suggested that circNEIL3 blocked the binding between IGF2BP3 and HECTD4 via steric hindrance, ultimately inhibiting the ubiquitination of IGF2BP3. Collectively, these data showed that circNEIL3 stabilizes IGF2BP3 protein by preventing HECTD4-mediated ubiquitination.

CircNEIL3 facilitates macrophage infiltration in glioma

Recent evidence suggests that glioma cells can recruit immune cells to the TME, thereby maintaining glioma pathology and promoting malignant progression [12, 34]. Intriguingly, our ssGSEA enrichment results showed that compared to tumours with low circNEIL3 expression, tumours with high circNEIL3 expression were enriched not only in carcinogenic signalling pathways but also in immune-related pathways, such as inflammatory response and TGF-beta signalling (Fig. S1C). Likewise, patients in the TCGA and CGGA datasets with high expression of IGF2BP3 also presented consistent results (Fig. S5E, F), suggesting that circNEIL3 is involved in immune regulation. To explore the differences in the biological behaviours among samples with distinct circNEIL3 expression, we divided the glioma samples into high and low expression groups based on the median expression of circNEIL3. Differential expression analysis was performed using the Deseq2 package, and a total of 801 genes were differentially expressed in the high expression group compared with the low circNEIL3 expression group; 531 genes were upregulated and 270 genes were downregulated (fold change ≥2 and P-value< 0.05, Additional file Table S4). Furthermore, Metascape database [35] analysis revealed that the differentially upregulated genes were markedly enriched in the regulation of cell biological functions, stromal activation and immune-related pathways (Fig. S8A, Additional file Table S5). Enrichment analysis with PaGenBase [36] showed that these genes were almost completely specifically expressed in spleen, blood, bone marrow and other peripheral immune organs (Fig. S8B). We then stratified glioma samples from the TCGA datasets into high and low expression groups based on the median expression of IGF2BP3 and obtained similar results through differential expression and enrichment analysis (Fig. S8C, D, Additional file Tables S6, S7). GSEA also demonstrated that immune-related signatures, including inflammatory response, TGF beta signalling, protein secretion, and leukocyte migration, were highly enriched in circNEIL3-high samples compared with circNEIl3-low samples (Fig. S8E). GSEA of IGF2BP3 in TCGA glioma samples uncovered similar results (Fig. S8F), indicating that circNEIL3 may be involved in the regulation of TME cells.

To further understand the exact role of circNEIL3 in the immunophenotype of glioma, we assessed glioma purity and stromal and immune scores using the ESTIMATE algorithm and analysed 31 gene sets representing different immune cell types, functions, and pathways via the gene set variation analysis (GSVA) algorithm (see Materials and Methods). As shown in Fig. 6A and Fig.S9, the circNEIL3 high expression group had greater cell infiltration into the TME, higher immune and stromal scores, and lower tumour purity than he circNEIL3 low expression group. Compared to gliomas with low circNEIL3 expression, gliomas with high circNEIL3 expression exhibited significantly increased macrophage infiltration, which constitute the most abundant cell population in the glioma TME. To identify specific macrophage populations associated with circNEIL3 expression, we analysed TAM MGs and TAM BMDMs in glioma samples using validated gene signatures [21]. The results showed that gliomas with high circNEIL3 expression exhibited significantly increased infiltration of TAM BMDMs (hereafter also referred to as macrophages), while the number of TAM MGs was decreased to some extent (Fig. 6A, Fig. S9). GSEA also showed that myeloid leukocyte migration and macrophage activation involved in immune response pathways were highly enriched in circNEIL3-high samples compared with circNEIl3-low samples (Fig. S8E). Similar to circNEIL3, high IGF2BP3 expression in glioma samples from the TCGA dataset showed similar results (Fig. S8F, Fig. S10A-D), suggesting that circNEIL3 is involved in the recruitment of macrophages. To further confirm the role of circNEIL3 in facilitating macrophage migration, we used the Transwell assay and found that conditioned medium (CM) from circNEIL3-overexpressing GBM cells significantly promoted THP1-differentiated macrophage migration compared to that of the NC group (Fig. 6B), while CM from circNEIL3-knockdown GBM cells showed the opposite results (Fig. 6C). Taken together, our results demonstrated that circNEIL3-overexpressing tumour cells could drive macrophage infiltration into the glioma microenvironment.

Fig. 6
figure6

CircNEIL3 facilitates macrophages infiltration in glioma. A The abundance of each TME infiltrating cells and regulators in circNEIL3 high and low groups. The statistical p-value was calculated using the nonparametric Wilcoxon test. Representative transwell migration assays showed the chemotaxis ability of human THP1-differentiated macrophages by exposing them to conditioned medium (CM) from GBM cells transfected with (B) ov-NC or ov-circNEIL3, and (C) sh-NC or sh-circNEIL3. Quantification histogram represented relative cell numbers, n = 3, scale bar, 200um. GSEA of CONRDENONsi YAP Conserved signature showed that glioma samples with high (D) circNEIL3 in our local glioma dataset, and (E) IGF2BP3 expression in TCGA glioma dataset were enriched in the YAP1 signaling compared to glioma samples with low expression, respectively. NES, normalized enrichment score; FDR, false discovery rate. F qRT-PCR assays showing the relative mRNA expression of CCL2 and LOX in GBM cells transfected with (left) sh-NC or sh-circNEIL3, and (right) ov-NC or ov-circNEIL3, n = 3. G qRT-PCR assays showing the relative mRNA expression of CCL2 and LOX in GBM cells co-transfected with ov-NC, ov-circNEIL3 and sh-IGF2BP3 as indicated, n = 3, and the ov-circNEIL3 group is indicated as the control. Western blot assays showing the protein expression of YAP1 and LOX in GBM cells (H) transfected with sh-NC or sh-circNEIL3, and (I) co-transfected with ov-NC, ov-circNEIL3 and sh-IGF2BP3 as indicated. All data are presented as the means ± SD, ns, P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Next, we explored the potential regulatory mechanism by which circNEIl3 promotes macrophage recruitment. Recently, Chen et al. [11] found that PTEN deficiency in GBM increases macrophage infiltration by activating YAP1 signalling, which directly upregulates the expression of lysyl oxidase (LOX), a powerful macrophage chemoattractant. The infiltrated macrophages in turn secrete SPP1, which is a member of the largest group of proteins specifically expressed by macrophages [14], to support GBM survival. In addition, C-C motif chemokine ligand 2 (CCL2), another powerful macrophage-secreted chemokine, is also a target of YAP1 [37]. Our GSEA results demonstrated that the YAP1 signalling gene signature was highly enriched in circNEIL3-high samples compared with circNEIL3-low samples (Fig. 6D). Further ssGSEA revealed that compared to tumours with low circNEIL3 expression, tumours with high circNEIL3 expression exhibited higher levels of CCL2 and LOX (Fig. S10E). Assessment of IGF2BP3 in TCGA glioma samples also showed the same results (Fig. 6E, Fig. S10E). Our qRT-PCR assays validated the results that circNEIL3 promoted the expression of CCL2 and LOX (Fig. 6F), and this enhanced expression induced by circNEIL3 overexpression could be rescued by IGF2BP3 knockdown (Fig. 6G). Finally, we confirmed that circNEIL3 could increase the protein expression of YAP1 and LOX in GBM cells (Fig. 6H). The increased expression caused by circNEIL3 overexpression also could be abrogated by IGF2BP3 knockdown (Fig. 6I). In summary, these results demonstrated that circNEIL3-overexpressing GBM cells might drive macrophage infiltration into the tumour-associated microenvironment by activating YAP1 signalling.

CircNEIL3 can be packaged into exosomes by hnRNPA2B1

Past studies indicated that circRNAs could be packaged into exosomes and play an important role in the progression of tumours [38, 39]. Then, to investigate whether circNEIL3 can be loaded into exosomes, we collected exosomes from the supernatants of cultured GBM cells, which exhibited similar typical cup-shaped morphology, size, and number (Fig. S11A, B), and further confirmed their identity by detection of the exosome markers TSG101 and CD9 (Fig. S11C), indicating that we successfully isolated exosomes from GBM cells. Furthermore, we found that the expression of circNEIL3 was downregulated in cells treated a pharmacological inhibitor of neutral sphingomyelinase-2 (nSMase) GW4869, which blocks exosome formation (Fig. S11D), thus confirming the existence of circNEIl3 in exosomes. Finally, we showed that circNEIL3 overexpression in GBM cells led to increased levels of circNEIL3 expression in exosomes, while circNEIL3 knockdown in the same cell line produced the opposite results (Fig. S11E). Altogether, these results show that circNEIL3 could be packaged into exosomes.

We next investigated the mechanism by which circNEIL3 is packaged into exosomes. We first analysed the mass spectrometry data of the circNEIL3 pull-down experiments and found that circNEIL3 could bind to hnRNPA2B1, which can transport various RNAs into exosomes [40, 41] (Fig. S11F, Additional file Table S2). We then confirmed the interaction between circNEIL3 and hnRNPA2B1 by RIP and RNA pull-down assays in GBM cells (Fig. 7A, B). Moreover, we found that circNEIL3 expression was upregulated in cells and downregulated in exosomes in hnRNPA2B1-knockdown GBM cells (Fig. 7C). In summary, these results indicated that circNEIL3 could be packaged into exosomes by hnRNPA2B1.

Fig. 7
figure7

Exosomes could deliver circNEIL3 to TAMs, thereby enabling them to acquire angiogenic and immunosuppressive properties. A RIP and qRT-PCR assays showing the interaction brtween circNEIL3 and hnRNPA2B1, using Igg and hnRNPA2B1 antibodies, n = 3. B RNA pulldown and western blot assays showing the interaction brtween circNEIL3 and hnRNPA2B1. qRT-PCR assay showing the relative expression in (left) GBM cells transfected with si-NC or si-hnRNPA2B1, and (right) exosomes collected from GBM cells transfected with si-NC or si-hnRNPA2B1, n = 3. D Representative flow cytometry assay showing the proportion of CD11b + CD163+ in THP1 differentiated macrophages transfected with ov-NC or ov-circNEIL3. Quantification histogram represented the proportion of CD11b + CD163+ in THP1 differentiated macrophages, n = 3. E Western blot assays showing the expression of SPP1 in THP1 differentiated macrophages (Left) transfected with ov-NC or ov-circNEIL3, and (Right) treated with exosomes collected from GBM cells transfected with si-circNEIL3, blank or ov-circNEIL3. F Representative flow cytometry assay showing the proportion of CD11b + CD163+ in THP1 differentiated macrophages treated with exosomes collected from GBM cells transfected with si-circNEIL3, blank or ov-circNEIL3. Quantification histogram represented the proportion of CD11b + CD163+ in THP1 differentiated macrophages, n = 3. G RNA pulldown and western blot assays showing the interaction between circNEIL3 and IGF2BP3 in THP1 differentiated macrophages in the indicated group. H RIP and qRT-PCR assays showing the interaction between circNEIL3 and hnRNPA2B1, using Igg and IGF2BP3 antibodies, n = 3. I Western blot assays showing the protein levels of IGF2BP3 in THP1 differentiated macrophages transfected with ov-NC or ov-circNEIL3, treated with 20 μg/ml CHX for the indicated periods of time. J Co-IP assay showing the ubiquitination modification level of IGF2BP3 in THP1 differentiated macrophages transfected with ov-NC or ov-circNEIL3. Western blot assays showing the protein levels of YAP1 in THP1 differentiated macrophages transfected with (K) left, ov-NC or ov-circNEIL3, and right, sh-NC or sh-circNEIL3 as indicated, and (L) left, ov-NC or ov-IGF2BP3, and right, sh-NC or sh-IGF2BP3 as indicated. M Western blot assays showing the YAP1 protein expression in THP1-differentiated macrophages co-transfected with ov-NC or ov-circNEIL3 and si-IGF2BP3 as indicated. N Left, representative bioluminescent image showing the tumor size for animals in different groups in the indicated time. n = 5 for each group. Right, statistical analysis of bioluminescent tracking plots. O Kaplan-Meier survival curves for animals in different groups, n = 5 for each group. P Represented H&E staining images for a subgroup of animal sacrificed simultaneously in each group, n = 5 for each group, scale bar, 10 μm. Q Proposed working model of circNEIL3 functions in tumorgenesis and malignant progression of glioma. All data are presented as the means ± SD, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Exosomes deliver circNEIL3 to TAMs, thereby enabling them to acquire immunosuppressive properties by stabilizing IGF2BP3

Exosomes play a vital role in the communication between cancer cells and TAMs, and we found that compared to tumours with low circNEIL3 expression, glioma samples with high circNEIL3 expression exhibited significantly increased expression of a panel of macrophage-derived immunosuppressive genes [12, 13], including CD163, TGFB1, IL1RA, IL10, ARG1 and PD-L1 (Fig. S10E). Our qRT-PCR analysis data also confirmed that circNEIL3 overexpression could significantly upregulate the expression of these genes in THP1-differentiated macrophages (Fig. S11G). Flow cytometry results further demonstrated that circNEIL3 overexpression significantly upregulated the macrophage activation marker CD163 (Fig. 7D). We observed that circNEIL3 overexpression significantly upregulated the protein expression of SPP1, which sustains glioma cell survival and stimulates angiogenesis [11]. Therefore, we hypothesized that exosomes could deliver circNEIL3 to macrophages to regulate the immunosuppressive phenotype of macrophages. First, we performed IF to confirm that exosomes can be taken up by macrophages (Fig. S11H). We then collected exosomes from GBM cells with normal expression, overexpression, or knockdown of circNEIL3 and divided them into high, medium and low expression groups before using them to treat THP1-differentiated macrophages (Fig. S11I). We found that macrophage activation markers, immunosuppressive molecules, and SPP1 were significantly upregulated with the increased expression of circNEIL3 in exosomes (Figs. 7E, F, Fig. S11J). Overall, these results suggested that exosomes could deliver circNEIL3 to TAMs, thereby enabling them to acquire angiogenic and immunosuppressive properties.

To clarify the mechanisms by which circNEIL3 mediates macrophage immunosuppressive properties, we performed RNA pull-down and RIP assays and found that circNEIL3 bound to IGF2BP3 at the same site in THP1-differentiated macrophages as in tumour cells (Fig. 7G, H). Moreover, as observed in tumour cells, circNEIL3 increased IGF2BP3 protein expression and inhibited IGF2BP3 ubiquitination in differentiated THP1 macrophages (Fig. 7I, J). A recent study reported that IGF2BP2, a member of the IGF2BP family, could promote macrophage polarization [42]. Meanwhile, IGF2BPs have similar 4 KH domains and 2 RRMs and play a similar role in various types of cells [31, 43]. Therefore, we hypothesized that, like that in GBM cells, IGF2BP3 can act as the downstream effector of circNEIL3 in macrophages and mediate the polarization of macrophage toward an immunosuppressive phenotype. Similar to our observations with circNEIL3, the panel of macrophage-derived immunosuppressive genes was dramatically upregulated in IGF2BP3 high expression gliomas compared with IGF2BP2 low expression gliomas in the TCGA dataset (Fig. S10E). Data from the qRT-PCR and flow cytometry assays confirmed that IGF2BP3 overexpression could significantly upregulate the expression of these genes in THP1-differentiated macrophages (Fig. S11K, M). Furthermore, the enhanced expression of these genes induced by circNEIL3 overexpression could be abolished by IGFB2P3 knockdown (Fig. S11L, N). Several studies have found that high YAP expression in macrophages can also promote their polarization toward the M2 phenotype [44, 45]. Thus, we hypothesized that circNEIL3 might also promote polarization of the immunosuppressive phenotype in macrophages by stabilizing IGF2BP3 protein to promote the expression of YAP1. Our western blot assays validated the results that overexpression of both circNEIL3 and IGF2BP3 enhances YAP1 protein expression in THP1-differentiated macrophages, while knocking down of them showed the opposite result (Fig. 7K, L). Furthermore, the enhanced protein expression of YAP1 induced by circNEIL3 overexpression could be abolished by IGFB2P3 knockdown (Fig. 7M), suggesting that circNEIL3 enhanced YAP1 expression by stabilizing the IGF2BP3 protein, which in turn promoted immunosuppressive phenotypic polarization in macrophages.

To further validate the effect of circNEIL3 on macrophage immunosuppressive polarization in vivo, macrophages overexpressing cicrNEIL3 or negative control vector were co-implanted with glioma cells into the brains of nude mice in situ. We found that compared to the NC group, the circNEIL3 overexpression group displayed elevated tumour growth and prolonged survival of tumour-bearing mice (Fig. 7N-P). Moreover, to further demonstrate that circNEIL3 exerts its function through the IGF2BP3-YAP1 axis, we then performed immunofluorescence staining assay in our local clinical glioma patient’s tissues. As shown in Fig. S11O, the expression of IGF2BP3, YAP1, CD68 (human macrophage marker) was enhanced in circNEIL3-high glioma tissues compared to circNEIL3-low tissues. Overall, these results indicated that IGF2BP3 can act as the downstream effector of circNEIL3 in macrophages and enable them to acquire immunosuppressive properties, in turn promoting glioma progression.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Disclaimer:

This article is autogenerated using RSS feeds and has not been created or edited by OA JF.

Click here for Source link (https://www.biomedcentral.com/)