LATS1 is downregulated in bladder cancer and interacts with circXRN2
LATS1 is the vital promoting molecule in the Hippo signaling pathway, playing significant roles in various malignancies33,34. As shown in Fig.1a, tissue microarrays and clinical samples indicated that LATS1 was aberrantly downregulated in bladder cancer tissues, when compared with adjacent normal ones. By performing RNA immunoprecipitation (RIP) together with the high-throughput sequencing, we obtained the database of circRNAs potentially interacting with LATS1 protein (Fig.1b and Supplementary File 2). To further discover the specific circRNAs probably participating in tumorigenesis and progression, we verified the expression level of circRNAs from the database in immortalized human normal urothelium cell (SV-HUC-1) and bladder cancer cell lines (5637, T24, EJ, TCCSUP and UM-UC-3), finally determining 14 circRNAs with remarkably low expression levels in bladder cancer cell lines (Fig.1c). Furthermore, we used RIP assay to validate the affinity between LAST1 and the circRNAs mentioned above, among which circXRN2 (hsa_circ_0001134) had the highest binding efficiency and affinity with LATS1 protein (Fig.1d). Besides, by qRT-PCR, we also verify the dysregulation of circXRN2 in clinical tumor tissues (Fig.1e-1g). Notably, the expression level of circXRN2 in tumor cell lines was associated with the activation of key Hippo pathway molecules, TAZ and YAP to some extent (Fig.1h and 1i). These findings suggested that circXRN2 might involve in the modulation of LATS1 and Hippo signaling pathway.
CircXRN2 inhibits cell proliferation and migration of bladder cancer cells in vitro and in vivo
CircRNA is known as long non-coding RNAs featuring covalently closed loop structure that protects it from degradation via exonuclease. Most of them are believed to consist of exons derived from existing protein-coding genes. CircXRN2 was formed by Exon14, Exon 15 and Exon 16 of XRN2 gene located in chromosome 28 through back-splicing and consisted of 296 bp.According to Sanger sequencing, we confirmed the conjunctive site sequence of circXRN2 (Fig.1j). Then we performed qRT-PCR by using divergent primers and convergent primers. The results of Fig.1k demonstrated that circXRN2 could only be amplified by divergent primers in cDNA, which confirmed the loop structure of circXRN2 instead of linear structure. Compared with linear RNAs, circRNA is more stable in endogenous environment so that it is not easily degradable.Because of this property, we treated the samples with RNase R and found that circXRN2 could tolerate the digestion by RNase R while linear XRN2 mRNA was almost completely degraded (Fig.1l). Since circXRN2 is aberrantly low in bladder cancer cells, we overexpressed circXRN2 to inspect its biological functions. Simultaneously, we ensured the alteration of XRN2 mRNA in circXRN2 overexpressing cells (Fig.1m).
In vitro experiments, we performed CCK8 assay for evaluating cell viability, and the results demonstrated the inhibitive effects of circXRN2 (Fig.2a). Similarly, colony formation assay was performed to testify the biological function of circXRN2 in cell proliferation, which manifested consistent trend with previous experiments (Fig.2b). Besides, overexpression of circXRN2 also triggered cell apoptosis (Fig.2c). In terms of cell migratory ability, as shown in Fig.2d and Fig.2e circXRN2 also undermined cellular migration capacity. Lastly, western blot showed the alternations of Fibronectin, N-cadherin and Vimentin (Fig.2f).
In vivo study, we constructed subcutaneous xenograft tumorigenesis model and tail vein lung metastasis model in nude mice.The results indicated that circXRN2 could reduce the growth rate and weight of subcutaneous tumors (Fig.2g-j). Similarly, the number of metastatic nodes in overexpressed circXRN2 group was much less than negative control group (Fig.2k-m).
All in all, our results elucidated that circXRN2 suppressed cell proliferation and migration of bladder cancer in vitro and in vivo.
CircXRN2 prevents LATS1 from SPOP-mediated degradation
To unmask the interactions between circXRN2 and LATS1, we constructed plasmids expressing different fragments of LAST1 and transfected them into 293T cells (Fig.3a). RIP assay showed that Fragment 2 (279aa-373aa) of LATS1 protein could interact with circXRN2 (Fig.3b). Next, we found that overexpression of circXRN2 upregulated LATS1 at protein level but not mRNA level (Fig.3c and 3d). According to the above results, we supposed that circXRN2 might modulate LATS1 protein level via post-translational modification. To confirm this conjecture, we treated negative control and circXRN2 overexpressing cells with cycloheximide (CHX, an inhibitor of eukaryotic protein synthesis), and the results showed that circXRN2 could prolong the half-life of LATS1 protein (Fig.3e). Besides, Bortezomib (an inhibitor of proteasome) could restore the expression level of LAST1 protein in circXRN2-deficient tumor cells (Fig.3f). Next, we performed IP assays to detect the ubiquitylation of LATS1 protein in different cells. As shown in Fig.3g, under treatments with Bortezomib and N-Ethylmaleimide (NEM, an inhibitor of deubiquitinating enzyme), less LATS1 protein was ubiquitinated in circXRN2 overexpressing cells. Taken together, these results confirmed that circXRN2 modulated LATS1 protein level in bladder cancer cells by regulating posttranscriptional ubiquitylation.
A recent literature reported that LATS1 is a potential substrate of speckle-type POZ (SPOP, an E3 ubiquitin ligase adapter), and SPOP promoted ubiquitylation and degradation of LATS134. Therefore, to verify that SPOP was involved in the ubiquitylation and degradation of LATS1 in human bladder cancer cells, we firstly transfected HA tagged SPOP at different concentration into bladder cancer cells. The results showed that SPOP downregulated LATS1 protein at a dose-dependent manner (Fig.3h). Interestingly, SPOP knockdown significantly abolished the downregulation of LATS1 protein induced by circXRN2 deficiency (Fig.3i). Moreover, IP assay showed that SPOP knockdown remarkably decreased the ubiquitylation of LATS1 protein (Fig.3j). In the light of this, we wondered whether SPOP participated in the degradation of LATS1 protein regulated by circXRN2.
It has been well proved that the vast majority of SPOP substrates share a SPOP-binding consensus motif F-P-S-S/T-S/T (“F” represents nonpolar, and “P” represents polar)35. According to previous researches, as shown in Fig.3k, there are two putative SPOP-binding motifs or “degrons” located in the N-terminus of LATS1 protein (SBC1: 327- MQSSS- 341; SBC2: 429- PQSSS- 443)34. In our previous results, circXRN2 interacted with Fragment 2 (279aa-373aa) of LATS1, which contained SBC1. Therefore, we speculated that circXRN2 and SPOP might competitively interact with LATS1 protein to regulate its degradation. As shown in Fig.3l, we found that knockdown of circXRN2 promoted the interaction between SPOP and LAST1 in bladder cancer cells. For further verification, by transfecting wild or mutant types of LATS1 plasmid, we confirmed that mutation of SBC1 led to remarkable block of LATS1 degradation mediated by SPOP, while depletion of SBC2 had little effect (Fig.3m). Meanwhile, Co-IP results showed that wild type LATS1 could bind to the SPOP, but SBC1-mutant LATS1 almost lost the interaction with SPOP (Fig.3n). In summary, our results demonstrated that circXRN2 prevents LATS1 from SPOP-mediated degradation.
CircXRN2 activates Hippo signaling pathway to regulate biological functions
In previous experiments, we had confirmed that circXRN2 could bind to degron on LATS1 protein, inhibit the ubiquitination mediated by SPOP, thus stabilizing endogenous LATS1 protein.As the vital upstream molecule in Hippo pathway, LATS1 activates the signaling axis by phosphorylating TAZ/YAP36.Therefore, in order to verify that circXRN2 performs its biological functions through Hippo pathway, we firstly validated that when Hippo pathway was activated after overexpressing circXRN2, corresponding proteins level altered accordingly (Fig.4a). Besides, immunofluorescence results suggested that circXRN2 increased the retention of TAZ/YAP in the cytoplasm, which also indicated the activation of the Hippo signaling pathway (Fig.4b). Then we depleted TAZ and YAP by designing shRNA targeting corresponding sequence, and the efficiency of shRNA was verified by western blot analysis (Fig.4c). Functionally, TAZ/YAP deficiency severely suppressed cell proliferation and growth rates and led to cell apoptosis at the same time (Fig.4d).Transwell and wound healing assays indicated that knockdown of TAZ/YAP impaired cell migratory capacity (Fig.4e and Fig.4f).Proteins associated with cell migration, such as Fibronectin, N-cadherin and Vimentin were determined by immunoblot analysis, showing similar tendency with previous results (Fig.4h).Taken together, all these results supported that circXRN2 activates the Hippo signaling pathway and modulates the downstream biological behaviors of bladder cancer cells.
ARHGEF4 and P2RY2 are direct targets of circXRN2-Hippo pathway axis to promote tumor progression
Once the Hippo signaling pathway activated, core kinase cascade phosphorylates TAZ/YAP and restrains the binding of TAZ/YAP to transcriptional factors (such as TEAD family) in the nucleus, which negatively controls gene expression involved in multiple cellular processes. Therefore, to investigate the underlying molecular mechanism of circXRN2-Hippo pathway axis induced biological functions, we performed RNA sequencing at transcriptome level in both T24 and TCCSUP cells with or without overexpression of circXRN2 (Fig.5a, Supplemental file 3 and Supplemental file 4). Among the overlapping of RNA-seq results in T24 and TCCSUP, ARHGEF4 and P2RY2 were closely related to bladder cancer progression and were significantly downregulated by circXRN2 (Fig.5b and 5c). Next, we further verified whether ARHGEF4 and P2RY2 were regulated by the Hippo pathway and their biological functions in bladder cancer. First, remarkable decreased ARHGEF4 and P2RY2 at mRNA level were also observed in TAZ/YAP-deficient cells (Fig.5d). Moreover, within the 2000bp promoter region of both ARHGEF4 and P2RY2, there were several consensus TEAD-binding sequences (TBS) (Fig.5e). To further confirming the interactions between TAZ/YAP and TEAD and the promoters, we performed ChIP assay and the results indicated that both YAP and TEAD enriched TBSs in the promoter region of ARHGEF4 and P2RY2 (Fig.5f and 5g). In addition, the luciferase assay proved that mutation of TBS suppressed ARHGEF4 and P2RY2 promoter activities (Fig.5h and 5i).
Previous studies reported that ARHGEF4 and P2RY2 are involved in tumorigenesis in different human cancers37-40. Based on current evidence, we then knocked down ARHGEF4 or P2RY2 to evaluate their potential biological functions. First, we confirmed the elimination of ARHGEF4 and P2RY2 in T24 and TCCSUP (Fig.5). The results demonstrated that P2RY2 acted a vital role in cell proliferation and growth (Fig.5k and 5l) and ARHGEF4 promoted cell migration (Fig.5m-o). Subsequently, we further verified that circXRN2 regulates tumorigenesis by modulating expression levels of ARHGEF4 and P2RY2. We first determined the expression level of ARHGEF4 and P2RY2 in cells with different treatments (Fig.5p). Next, we replenished ARHGEF or P2RY2 in circXRN2 overexpressing cells, and the results in Fig.5q-u showed that ARHGEF and P2RY2 attenuated the inhibitory effects on cell migration and growth induced by circXRN2.
CircXRN2 serves as a suppressor of glucose metabolism in bladder cancer
A bunch of researches have reported the regulatory role of Hippo pathway in cell glycolytic metabolism32,41,42. Given the close relationship between circXRN2, LATS1 and Hippo pathway, we supposed that circXRN2 plays a key role in glycolysis and energy metabolism of bladder cancer cells.To validate this speculation, we overexpressed circXRN2 and performed 2-NBDG uptake detection, glucose uptake assay and lactate production determination. The above experiments proved that circXRN2 reduced the abilities of glucose uptake and lactate production in T24 and TCCSUP cells (Fig.6b-d).The results of Seahorse glycolytic rate analysis indicated that both glycoPER, basal glycolysis and compensatory glycolysis of bladder cancer cells overexpressing circXRN2 were lower than those of normal tumor cells (Fig.6e and 6f).The results of Fig.6g suggested that circXRN2 could significantly downregulate glycolysis-related genes at protein level. Of note, mitochondrial membrane potential, indicating early stage of cell apoptosis, also reflects the status of cell energy metabolism.Through JC-1 staining, we found that circXRN2 also had certain effects on mitochondrial membrane potential, suggesting that circXRN2 might modulated mitochondrial energy metabolism (Fig.6h and 6i).Taken together, circXRN2 served a negative role in glucose metabolism of T24 and TCCSUP cells.
The Hippo pathway modulates glycolysis in bladder cancer cells
For further consolidating that circXRN2/Hippo axis regulates glucose metabolism, we knocked down TAZ and YAP in T24 and TCCSUP cells, and used a series of experiments to evaluate glycolysis in differently treated groups. Similarly, we found a remarkable decrease in the uptake of 2-NBDG by tumor cells after knockdown of TAZ/YAP by flow cytometry (Fig.7a).Glucose uptake assay (Fig.7b) and lactate production detection assay (Fig.7c) also indicated that the Hippo signaling pathway is crucial for glycolysis of bladder cancer cells. In addition, through Seahorse glycolytic rate analysis, the effects of TAZ/YAP on glycoPER, basal glycolysis and compensatory glycolysis of bladder cancer cells also showed the same trend (Fig.7d and 7e). Glycolytic metabolism-related genes were significantly downregulated in TAZ/YAP-deficient cells (Fig.7f). Meanwhile, impairment of the Hippo pathway also decreased mitochondrial membrane potential (Fig.7g and 7h).
CircXRN2 activates the Hippo pathway through LATS1 to regulate bladder cancer cell proliferation, migration and glycolytic metabolism
To confirm that circXRN2/LATS1 axis regulates various biological functions and glucose metabolism of bladder tumor cells, we knocked down LATS1 protein in cells overexpressing circXRN2. Immunoblot analysis showed the protein levels of LATS1, TAZ and YAP (Fig.8a), and intracellular location of TAZ/YAP was confirmed by immunofluorescence (Fig.8b). The results of CCK-8 assay and colony formation assay showed that depletion of LATS1 reversed the inhibitory effects on cell viability and proliferation induced by circXRN2 (Fig.8c and 8d). Meanwhile, as shown in Fig.8e, depletion of LATS1 also promoted cell survival and prevented circXRN2-induced apoptosis. Cell migration was determined by transwell assay and wound healing assay. Compared with overexpressing-circXRN2 cells, those added with shLATS1 were more capable of migration (Fig.8f and 8g).
Additionally, glucose uptake capacity and lactate production of control group, circXRN2 overexpressing group, and circXRN2 + shLATS1 group were determined by 2-NBDG uptake measurement, glucose uptake assay, and lactate production detection, respectively. Fig.8i-k showed that shLATS1 could alleviate the suppression in glycolytic metabolism caused by circXRN2. Subsequently, glycoPER, basal glycolysis and compensatory glycolysis rates of different cells were analyzed and calculated by the Seahorse metabolic analyzer, and the results also supported the above conclusions (Fig.8l and 8m). Western blot was used to evaluate the alternations of genes related to cell migration and associated with glycolytic metabolism (Fig.8n). Finally, the mitochondrial membrane potential was evaluated by JC-1 probe. As shown in Fig.8o and 8p, circXRN2 overexpression enhanced the fluorescence of JC-1 monomer, while shLATS1 reversed the above effect.
CircXRN2/LATS1 axis inhibits tumor growth and metastasis of bladder cancer cells in vivo
To verify whether circXRN2/LATS1 axis inhibits tumor growth and metastasis of bladder cancer cells in vivo, we injected cells with different treatments (Vector, circXRN2 overexpression, circXRN2 + shLATS1) into nude mice subcutaneously. Similar to in vitro cell experiments, overexpressing circXRN2 could significantly reduce the tumor growth rate, and knockdown of LATS1 could significantly attenuate the inhibitory effect caused by circXRN2 (Fig.9a-d). In addition, we detected expression levels of LATS1 and TAZ/YAP in tumors of different treatment groups by immunohistochemistry, and the results were consistent with previous experiments (Fig.9e).We also established a nude mouse lung metastasis model by injecting different T24 cells into the tail vein to evaluate the ability of cells to metastasize in vivo. As demonstrated in Fig.9f-h, overexpression of circXRN2 significantly reduced lung metastatic nodules compared with the normal control group, while the tumor nodules were significantly increased in cells treated with circXRN2 overexpression along with knockdown of LATS1. In conclusion, our findings illuminated that circXRN2 suppresses tumor progression and glycolysis in human bladder cancer through activating Hippo signaling pathway.