Identification and characterization of circPOLK in NSCLC cells
We collected blood exosomes from cancer patients and healthy people, and RNA sequencing results revealed that circPOLK was remarkably overexpressed in blood exosomes from cancer patients than those from healthy people (Fold change = 9.18, p = 1.45e-4; Fig. 1A and 1B, Table S3) [18, 19]. Meanwhile, the level of circPOLK was also higher in NSCLC cells than in normal lung epithelial cells (Fig. 1C). Sanger sequencing verified the presence of circPOLK head-to-tail junctions and its generation from host gene POLK in NSCLC cells (Fig. 1D). Furthermore, circPOLK could be amplified using divergent primers by complementary DNA (cDNA) but failed to be amplified from genomic DNA (gDNA), and host gene POLK could be amplified by convergent primers using cDNA or gDNA in NSCLC cells (Fig. 1E). These data proved the existence of circPOLK in NSCLC cells. Subsequently, the POLK mRNA could be degraded by RNase R, but circPOLK was resistant to the digestion of RNase R (Fig. 1F). In addition, actinomycin D were applied to suppress new RNA synthesis to determine the stability of circPOLK. As shown in Fig. 1G, circPOLK was more stable than linear POLK mRNA in NSCLC cells. Thus, we prove the existence of circPOLK in NSCLC cells, and circPOLK is a circular RNA and generates from host gene POLK by back-splicing.
circPOLK reinforces the metastasis of NSCLC cells
Then, we explored the function of circPOLK in NSCLC progression, we silenced the expression of circPOLK using sicircPOLK targeting head-to-tail junctions without affecting the level of POLK (Fig. 2A). As shown in Figure S1A circPOLK knockdown could not affect the proliferation of NSCLC cells. However, circPOLK silence statistically inhibited the migration and invasion in NSCLC cells (Fig. 2B and 2C). Similarly, would healing assay also supported that the knockdown of circPOLK restrained the migratory abilities of NSCLC cells (Fig. 2D). Subsequently, we constructed the circPOLK overexpressing plasmid, and the level of circPOLK was statistically upregulated in NSCLC cells transfected with circPOLK overexpressing plasmid compared with empty vector (Fig. 3A). In contrast, circPOLK overexpression statistically augmented the migratory and invasive capacities of NSCLC cells (Fig. 3B and 3C). Meanwhile, the ectopic expression of circPOLK could remarkably promote the migratory capacity of NSCLC cells as demonstrated by wound healing assay (Fig. 3D). Furthermore, sicircPOLK suppressed the EMT progression of via increasing E-cadherin and suppressing N-cadherin, and circPOLK overexpression reinforced the EMT progression of NSCLC cells (Fig. 3E and 3F). Additionally, NSCLC transfected with shcircPOLK exhibited less metastatic foci in the livers or lungs of nude mice as compared to NSCLC cells transfected with empty vector (Fig. 3G and 3H).
circPOLK functions as a sponge for miR-1204 in NSCLC cells
We next investigated the mechanism of circPOLK-mediated NSCLC metastasis. circRNA frequently functions as a miRNA sponge and regulate the levels of target genes. Thus, we first hypothesized that circPOLK might promote metastasis of NSCLC cells via sponging miRNA. Indeed, circPOLK was remarkably enriched by the AGO2 antibody compared with IgG using RIP assay, indicating that circPOLK might function as a binding platform between AGO2 and miRNAs (Fig. 4A). Furthermore, FISH analysis also demonstrated that circPOLK mostly located in the cytoplasm of NSCLC cells (Fig. 4B). We next predicted the potential target miRNAs of circPOLK based on miRanda, RNA hybrid, and circinteractome [20–22]. Through bioinformatical prediction programs, there were 12 overlapping target miRNAs of circPOLK (Table S4 and Fig. 4C). To further verify the potential targets of circPOLK, we established RNA pull-down to enrich circPOLK and its interacting partners. miR-4299 and miR-1204 could be statistically pulled down by circPOLK probe, and miR-1204 was efficiently pulled down by circPOLK in NCI-H1299 cells (Fig. 4D). The interaction between circPOLK and miR-1204 was predicted by circinteractome, and circPOLK and its mutant sequences were cloned as shown in Fig. 4E. Luciferase reporter assay showed that miR-1204 mimic could significantly sponge wild type circPOLK, but not mutant circPOLK (Fig. 4F). Furthermore, the level of extracellular miR-1204 was lower in the serum from multiple types of cancer patients than healthy persons (Fig. 4G) [23, 24]. Meanwhile, the expression of circulating miR-1204 was found to be significantly decreased in the serum of individuals diagnosed with lung cancer compared to that observed in healthy individuals (Fig. 4H) [25]. Additionally, low expression of miR-1204 predicted poor outcomes of lung cancer patients (Fig. 4I) [26]. Furthermore, the level of miR-1204 reduced in NSCLC cells compared to normal lung epithelial cells (Fig. 4J). Therefore, these data indicate that circPOLK may sponge miR-1204 in NSCLC cells, and serumal miR-1204 maybe a potential molecular marker for the diagnosis and prognosis of individuals with lung cancer.
circPOLK promotes the metastatic capabilities of NSCLC through miR-1204
The subsequent investigation focused on elucidating the role of miR-1204 in the progression of NSCLC. As shown in Fig. 5A, miR-1204 inhibitor reinforced the migratory and invasive capabilities of NSCLC cells, and miR-1204 mimic remarkably restrained cell migratory and invasive capabilities (Fig. 5B). These data indicate that miR-1204 is a tumor suppressor in NSCLC progression. Furthermore, miR-1204 inhibitor successfully enhanced the metastatic potential suppressed by shcircPOLK in NSCLC cells (Fig. 5C and 5D). Meanwhile, miR-1204 mimic reversed the enhancement of metastatic potential induced by circPOLK in NSCLC cells (Fig. 5E and 5F). Collectively, circPOLK reinforces the metastatic potential of NSCLC cells by miR-1204.
SOX8 is a downstream target of circPOLK in regulating miR-1204
We then determined the target proteins of circPOLK/miR-1204, RNA-sequence was used to explore the differentially expressed genes (DEGs) regulated by circPOLK in NSCLC cells (Table S5). Meanwhile, the downstream targets of miR-1204 were predicted by miRmap and miRWalk (Table S6) [27, 28]. A Venn diagram was generated showing that there were eight overlapping genes among circPOLK regulated DEGs, miR-1204 potential downstream targets of miR-1204 predicted by miRmap and miRWalk, including NECAB2, MSI1, SLC30A3, GMNC, SOX8, RNF152, MITF, and GPR85 (Fig. 6A). Then, qRT-PCR was established to further verified the target genes of circPOLK/miR-1204, circPOLK significantly upregulated the expression of SOX8 in three NSCLC cells, while circPOLK downregulated RNF152 in A549 cells and upregulate RNF152 in NCI-H1299 and NCI-H460 cells (Fig. 6B - D). Furthermore, SOX8 overexpression, but not RNF152 overexpression, predicted poor outcomes of lung cancer patients (Fig. 6E) [29]. Thus, we assumed that SOX8 might be a potential target gene of circPOLK/miR-1204. Indeed, miR-1204 mimic could significantly sponge wild type SOX8, but not mutant SOX8 (Fig. 6F). Furthermore, knockdown of circPOLK suppressed the expression of SOX8, and circPOLK promoted SOX8 expression in NSCLC cells (Fig. 6G). Meanwhile, miR-1204 mimic restrained the SOX8 expression, conversely, miR-1204 inhibitor enhanced the level of SOX8 in NSCLC cells (Fig. 6H). More importantly, miR-1204 mimic reversed the expression of SOX8 enhanced by circPOLK, and miR-1204 inhibitor also enhanced SOX8 expression suppressed by shcircPOLK in NSCLC cells (Fig. 6I). In addition, SOX8 silence significantly suppressed the progression of EMT in NSCLC cells (Fig. 6J). Thus, SOX8 might be a downstream target of circPOLK/miR-1204 in regulating NSCLC progression.
circPOLK promotes angiogenesis via regulating TME
GO enrichment of circPOLK regulated DEGs demonstrated that circPOLK might play vital role in cell-cell adhesion and cell adhesion, indicating that circPOLK could participate in the progression of EMT. Furthermore, circPOLK might also be involved in regulation of hemopoiesis and cell activation, cell communication, external side of plasma membrane, and regulation of cytokine production (Fig. 7A and Table S7) [30, 31]. Besides, circPOLK was identified in the exosomes from human blood, we thus hypothesized that circPOLK might be involved in angiogenesis and TME in NSCLC. Then, we constructed coculture systems to mimic angiogenesis in TME of NSCLC (Fig. 7B). NSCLC cells and HUVEC cells were cocultured according to Fig. 7B upper panel, and circPOLK silence in cancer cells could significantly restrain the migratory abilities of HUVEC cells (Fig. 7C and 7D). In contrast, circPOLK overexpression in NSCLC cells could remarkably enhance the migratory abilities of HUVEC cells (Fig. 7E and 7F). Meanwhile, we also cocultured NSCLC cells and HUVEC cells as shown in Fig. 7B lower panel, NSCLC cells with circPOLK knockdown restrained the migratory abilities of HUVEC cells in TME, and NSCLC cells with circPOLK overexpression increased the migratory abilities of HUVEC cells (Fig. 7G and 7H). Thus, the expression of circPOLK in NSCLC cells affected the migratory abilities of vascular endothelial cells in TME, indicating that circPOLK secreted by NSCLC cells might promote angiogenesis of NSCLC.