Identification of a novel circRNA (cDOPEY2) downregulated in cisplatin-resistant ESCC cells and its characterization.
To explore the molecular mechanism underlying cisplatin resistance in ESCC cells, we initially established two cisplatin-resistant ESCC cell lines (Eca109-CR and TE1-CR) by consecutively culturing the cells in media containing increasing concentrations of cisplatin; their cisplatin resistance was then determined by conducting a clonogenic assay (16). As shown in Figure 1A, both of the cisplatin-resistant cell lines exhibited a higher survival rate after cisplatin treatment than their parental cells. To characterize candidate circRNAs mediating cisplatin resistance, we analyzed the expression profiles of circRNAs in ESCC from a GEO dataset (GSE131969) to identify dysregulated circRNAs (Figure 1B). Among the top 5 circRNAs with the most extensive differential expression, 7 were proven to be authentic circRNAs based on the results of RNase R treatment (Figure 1C). We further validated the expression of these circRNAs in Eca109-CR/Eca109 and TE1-CR/TE1 cells by qRT-PCR, revealing that only hsa_circ_0008078, a circRNA derived from the DOPEY2 gene, was consistently and significantly downregulated in cisplatin-resistant cells compared with their parental cells (Figure 1D). We thus further investigated the potential involvement of hsa_circ_0008078, named cDOPEY2, in ESCC cell cisplatin resistance.
Annotation from circBase (http://circrna.org) revealed that cDOPEY2 originated from the back-splicing of the DOPEY2 gene between the 20th and 24th exons (Figure 1E). We next utilized several approaches to characterize the nature of cDOPEY2. We first identified the back-splicing site of cDOPEY2 by Sanger sequencing (Figure 1F) and then used divergent primers to amplify cDOPEY2 from cDNA but not from gDNA (Figure 1G). Next, actinomycin D treatment was performed, confirming that cDOPEY2 was more stable than its parental host gene (Figure 1H). Moreover, cDOPEY2 was significantly enriched in samples that were reverse-transcribed using random hexamer primers compared with that in samples reverse-transcribed with oligo (dT) primers (Figure 1I). Ultimately, qPCR of subcellular fractions and fluorescence in situ hybridization (FISH) analyses showed that cDOPEY2 was mainly located in the cytoplasm (Figure 1J-K).
Moreover, assessment of cDOPEY2 expression in ESCC cell lines revealed that it was not only significantly downregulated in cisplatin-resistant cells but also markedly decreased in ESCC cell lines compared with the normal esophageal squamous epithelial cell lines HECC and Het-1a (Figure 1L).
Taken together, these results revealed a novel circular RNA (cDOPEY2) that was downregulated in cisplatin-resistant ESCC cells, indicating a potential suppressive effect of cDOPEY2 in mediating cisplatin resistance.
cDOPEY2 attenuates cisplatin resistance by enhancing cisplatin-induced apoptosis in ESCC
We next investigated the impact of cDOPEY2 on cisplatin resistance in ESCC by overexpressing cDOPEY2 in cisplatin-resistant cells and silencing cDOPEY2 in cisplatin-sensitive cells using lentiviral-based constructs (Figure S1A-B). Cell viability and colony formation assays revealed that high cDOPEY2 expression markedly sensitized cisplatin-resistant ESCC cells to cisplatin treatment, while depletion of cDOPEY2 exerted the opposite effects on parental ESCC cells (Figure 2A-B), suggesting an important role of cDOPEY2 in alleviating cisplatin resistance. We further assessed cisplatin-induced apoptosis by Annexin-V/PI staining, clearly showing that cDOPEY2 overexpression in cisplatin-resistant cells boosted cisplatin-mediated apoptosis (Figure 2C-D). Consistently, caspase-3 activity was enhanced by cDOPEY2 in cells treated with cisplatin (Figure 2E). Because cisplatin kills tumor cells by inducing replication fork arrest and subsequently causing replication stress (17), we next explored whether cDOPEY2 impacted the DNA damage in tumor cells by performing γ-H2AX staining. Although γ-H2AX staining was markedly decreased in cisplatin-resistant cells compared with parental cells, cDOPEY2 did not impact cisplatin-induced DNA damage (Figure 2F). We obtained similar results by determining the activation of the DNA damage repair (DDR) pathway (Figure 2G). The abundances of antiapoptotic regulatory proteins of the Bcl-2 family were further determined by western blotting, revealing that cDOPEY2 upregulation significantly decreased the expression of antiapoptotic Mc1-1. However, the other antiapoptotic proteins, Bcl-xL and Bcl-2, were unchanged following cDOPEY2 alteration (Figure 2H). Together, these results indicated that cDOPEY2 plays a vital role in alleviating ESCC resistance to cisplatin and that this effect is possibly exerted via the exaggeration of the cisplatin-induced apoptotic process.
We next explored the clinical significance of cDOPEY2 by examining its expression in clinical samples from ESCC patients. Confirming the previous microarray results, cDOPEY2 was significantly downregulated in 81% (39 of 48) of the ESCC samples (cohort I, containing 48 pairs of tumor/nontumor samples) compared to the adjacent nontumorous tissues (Figure 2I). In a second cohort of patients with advanced ESCC receiving cisplatin treatment, low cDOPEY2 expression in tumor samples was significantly associated with a lower response rate and worse progression-free survival (PFS) than those of patients with high cDOPEY2 tumor expression levels (Figure 2J-K). These findings indicate that cDOPEY2 has potential as a suitable marker for predicting the efficacy of cisplatin responsiveness in ESCC patients.
cDOPEY2 physically interacts with the RNA-binding protein (RBP) CPEB4
We further dissected the mechanism by which cDOPEY2 alleviates cisplatin resistance in ESCC cells. First, cDOPEY2 was determined to be mainly localized in the cytoplasm, as shown in Figure 1J-K. However, AGO2 RIP showed that cDOPEY2 did not coimmunoprecipitate with AGO2, indicating that it likely does not act as a competitive RNA (Figure S2A). Second, no potential open read frame (ORF) or internal ribosomal entry site (IRES) sequences were found in cDOPEY2. Third, cDOPEY2 was unable to influence the expression of its host gene, DOPEY2 (Figure S2B). Taking the above observations into consideration, we hypothesized that cDOPEY2 functions by acting as a protein scaffold in the cytoplasm.
To test our hypothesis, we first conducted an RNA pulldown assay by incubating the extracts of Eca109 cells with a biotinylated probe specifically targeting the junction site of cDOPEY2. The coisolated proteins were separated by SDS-PAGE, stained with Coomassie blue, identified and analyzed by mass spectrum (MS) (Figure 3A-B). We further performed MS analysis of extracts from cDOPEY2-overexpressing Eca109-CR cells, and compared with control cells, a total of 39 dysregulated proteins were identified (Figure 3D). Gene ontology (GO) analysis showed that these proteins were significantly correlated with responses to drugs and with the regulation of apoptosis (Figure 3C). Among the top 10 most abundant proteins enriched with cDOPEY2, only CPEB4 was significantly regulated by cDOPEY2 (Figure 3E). CPEB4 is a cytoplasmic RBP that maintains the translation of regulating mRNAs, and importantly, cDOPEY2 harbors a binding sequence (UUUUA) for CPEB4 (18) (Figure 3F). Meanwhile, CPEB4 exhibited the most significantly differential expression following cDOPEY2 overexpression (Figure 3D). These results strongly suggest that CPEB4 not only binds to but is also regulated by cDOPEY2. We thus confirmed the interaction between cDOPEY2 and CPEB4 using RNA pulldown and RNA immunoprecipitation (RIP) assays (Figure 3G-H). In addition, the RNA FISH immunofluorescence assay revealed that cDOPEY2 colocalized with the CPEB4 protein in the cytoplasm (Figure 3I).
The RNA recognition motifs (RRMs) of CPEB4 are essential for the interaction between CPEB4 and RNAs containing the UUUUA sequence (18). To further elucidate the binding mechanism between CPEB4 and cDOPEY2, we constructed several truncated fragments of CPEB4 lacking RRM1, RRM2 or both (Figure 3J). RIP demonstrated that the binding of CPEB4 with cDOPEY2 required both RRMs, as deletion of either RRM1 or RRM2 reduced the level of binding between cDOPEY2 and CPEB4, and the deletion of both RRMs simultaneously completely abrogated its ability to bind cDOPEY2 (Figure 3K). To confirm whether cDOPEY2 interacts with CPEB4 via its UUUUA sequence, we mutated the cDOPEY2 UUUUA sequence to GACCG, which significantly reduced the interaction between CPEB4 and cDOPEY2 (Figure 3L). These data confirm that CPEB4 physically binds with cDOPEY2.
CPEB4 is an oncogene that augments cisplatin resistance in ESCC cells and is targeted by cDOPEY2
CPEB4 has a well-characterized role in oncogenesis (19, 20), and survival analysis of a TCGA data set confirmed the oncogenic role of CPEB4 in ESCC (Figure 4A). Moreover, the CEPB4 protein levels were substantially higher in ESCC samples than in adjacent nontumorous tissues (Figure 4B). To determine whether the chemosensitive effect of cDOPEY2 was mediated by the depletion of CPEB4, we first examined the CPEB4 expression in cisplatin-resistant cells. Western blotting showed that CPEB4 expression was significantly increased in cisplatin-resistant cells compared to their parental cells, while cDOPEY2 overexpression decreased CPEB4 expression, and cDOPEY2 silencing increased the expression of CPEB4 (Figure 4C). We further silenced CPEB4 in cisplatin-resistant ESCC cells and upregulated CPEB4 in parental cells (Figure S3A-B). cDOPEY2 knockdown in parental Eca109 cells markedly increased the cell viability and clonogenic formation ability after cisplatin treatment, whereas cDOPEY2 silencing-induced chemoresistance was almost impaired following CPEB4 knockdown. Conversely, the enhanced sensitivity to cisplatin after cDOPEY2 overexpression in TE1-CR cells was largely rescued by CPEB4 overexpression (Figure 4D-E). With respect to cisplatin-induced apoptosis, cDOPEY2 depletion led to the acquisition of cisplatin resistance in parental ECA109 cells, as reflected by Annexin-V/PI FACS analysis and caspase-3 activity, and these effects were largely abolished by CPEB4 knockdown. We observed similar results in TE1-CR cells, in which CPEB4 overexpression greatly restored the cisplatin resistance of cells overexpressing cDOPEY2 (Figure 4F-G). The increased expression of the antiapoptotic protein Mcl-1 induced by cDOPEY2 depletion was significantly blocked following CPEB4 knockdown, whereas cDOPEY2-suppressed Mcl-1 expression was completely reversed by CPEB4 overexpression (Figure 4H). Altogether, these data suggested that cDOPEY2 sensitizes ESCC cells to cisplatin by inhibiting CPEB4.
cDOPEY2 decreases the stability of the CPEB4 protein by enhancing its ubiquitination
As shown in Figure 4C, although cDOPEY2 negatively regulated the protein expression of CPEB4, its mRNA expression was not significantly changed by cDOPEY2 (Figure 5A). We also found that silencing cDOPEY2 increased the stability of the CPEB4 protein (Figure 5B). In addition, the reduced protein expression of CPEB4 in cDOPEY2-overexpressing cells was completely reversed by treatment with MG132, a proteasome inhibitor (Figure 5C). We thus speculated that cDOPEY2 might reduce the stability of CPEB4 by enhancing its ubiquitin/proteasome-dependent degradation. Consistent with our hypothesis, cDOPEY2 overexpression strongly promoted the ubiquitination of CPEB4, while cDOPEY2 knockdown significantly inhibited this phenomenon (Figure 5E). These data indicated that cDOPEY2 suppresses CPEB4 by inducing its ubiquitination.
The E3 ligase TRIM25 mediates the cDOPEY2 promotion of CPEB4 ubiquitination
We further explored the E3 ligase responsible for the cDOPEY2-mediated ubiquitination of CPEB4. Of all the proteins coprecipitated with cDOPEY2 as determined by MS, only one E3 ligase, TRIM25, was identified (Figure 3A-B, Figure 5D). Moreover, TRIM25 negatively regulated the expression of CPEB4 at the protein level but not at the mRNA level (Figure 5F, Figure S3C-D). The RNA-binding activity of TRIM25 is reportedly essential for the effective ubiquitination of its targeted proteins (21). RNA FISH immunofluorescence analysis revealed the colocalization of cDOPEY2 with TRIM25 in the cytoplasm (Figure 5G), and RIP and RNA pulldown assays confirmed the binding between cDOPEY2 and TRIM25 (Figure 5H). cDOPEY2 overexpression notably enhanced the effects of TRIM25 on the degradation of CPEB4, while knockdown of TRIM25 largely reversed this cDOPEY2-mediated effect (Figure 5I). These results indicated that cDOPEY2 might serve as a scaffold to enhance the interaction between TRIM25 and CPEB4. We then explored whether the ubiquitin ligase activity of TRIM25 toward CPEB4 was dependent on cDOPEY2. To do this, we constructed a myc-tagged TRIM25 fragment in which a portion of the RBD was deleted (ΔRBD, 470–508 aa deletion, Figure 5L). IP analysis confirmed that the RBD deletion drastically impaired the ability of TRIM25 to bind CPEB4, while RNase A (an RNA exonuclease blocking the interactions between cDOPEY2 and TRIM25) treatment and cDOPEY2 silencing also weakened the interaction between TRIM25 and CPEB4 (Figure 5K). Notably, transfection with TRIM25 lacking an RBD, RNase A treatment and cDOPEY2 knockdown all impaired the effects of TRIM25 on the degradation and ubiquitination of CPEB4 (Figure 5J, M). Taken together, our data strongly indicated that cDOPEY2 acts as a scaffold to enhance the interaction between TRIM25 and CPEB4, thereby potentiating TRIM25-dependent ubiquitination and proteasomal degradation.
CPEB4 upregulates the expression of the antiapoptotic protein Mcl-1 by promoting its translation
As shown in Figure 2H, cDOPEY2 decreased the expression of the antiapoptotic protein Mcl-1. We thus wondered whether the inhibitory effect of cDOPEY2 on Mcl-1 was mediated by the decreased expression of its downstream target CPEB4. Importantly, the up- or downregulation of Mcl-1 protein expression following CPEB4 overexpression or silencing was not confirmed at the mRNA level, suggesting that the regulation of Mcl-1 mediated by CPEB4 is modulated at the posttranscriptional level (Figure 6A-B). CPEB4 promotes cytoplasmic polyadenylation and the translational activation of mRNAs containing cytoplasmic polyadenylation elements (CPEs) (22). Notably, because we found several CPEs in the 3’UTR of Mcl-1 mRNA (Figure 6C), we speculated that CPEB4 upregulates Mcl-1 expression by promoting its translation. In support of this possibility, RIP analysis confirmed the interaction between CPEB4 and Mcl-1 mRNA (Figure 6D-E). Moreover, cordycepin (a known inhibitor of CPEB4-mediated polyadenylation) treatment decreased the protein expression of Mc1-1, accompanied by an increase in the expression of cleaved caspase-3, and these effects were dose-dependent (Figure 6F). The FISH immunofluorescence assay confirmed that CPEB4 colocalized with Mc1-1 mRNA in ESCC tissues (Figure 6G).
We performed a ribosome enrichment assay to validate the regulatory effect of CPEB4 on Mcl-1 translation. Extracts of Eca109 cells in which CPEB4 was overexpressed or silenced were subjected to 5–50% sucrose gradient centrifugation, followed by qPCR analysis of the Mcl-1 mRNA expression in each fraction. The distribution of Mcl-1 mRNA shifted from light polysomes to heavy polysomes in CPEB4-overexpressing cells compared with control cells, while CPEB4 silencing shifted the Mc1-1 mRNA-containing fraction from light polysomes to monosomes (Figure 6H), proving that CPEB4 upregulates Mcl-1 expression by promoting its translation. We further validated the involvement of Mcl-1 in CPEB4-mediated cisplatin resistance. Notably, Mcl-1 overexpression significantly rescued the effects of CEPB4 silencing on cisplatin resistance, as determined by cell viability, colony formation and apoptotic assays (Figure 6I-K). Moreover, depletion of Mcl-1 completely abolished the enhanced chemoresistance induced by CPEB4 (Figure 6K-M). Taken together, these results indicated that cDOPEY2 increases cisplatin resistance in ESCC by suppressing CPEB4-induced Mcl-1 translation.
cDOPEY2 inhibits tumorigenicity and enhances cisplatin sensitivity in ESCC xenograft cells
We further verified the biological impact of cDOPEY2 on ESCC in vivo. To do this, we constructed Eca109-CR cells with cDOPEY2 overexpression, cDOPEY2 knockdown and cDOPEY2/CPEB4 double overexpression and inoculated the indicated cells into BALB/c nude mice. The tumor-bearing mice were intraperitoneally injected with cisplatin (1 mg/kg) every 2 days (Figure 7A). The results demonstrated that silencing cDOPEY2 further enhanced the chemoresistance of Eca109-CR cells. In contrast, nude mice inoculated with cDOPEY2-overexpressing cells exhibited a marked sensitivity to cisplatin resistance, while the re-expression of CPEB4 in cDOPEY2-overexpressing cells completely abolished the effects mediated by cDOPEY2 (Figure 7B-C). Consistent with the in vitro results, Immunohistochemistry (IHC) confirmed that the expression of Mcl-1 and CPEB4 was increased in the sh-cDOPEY2 group but decreased in the cDOPEY2 overexpression group compared with the scramble group (Figure 7D).
These data verified the in vitro findings and support that cDOPEY2 has potential as a pharmaceutical intervention target to increase the efficacy of cisplatin treatment in patients with advanced ESCC.
Overall, our results show that cDOPEY2 acts as a scaffold to facilitate the ubiquitination and degradation of CPEB4 in a TRIM25-dependent manner, thereby boosting cisplatin-induced apoptosis by suppressing CPEB4-promoted Mcl-1 translation and alleviating cisplatin resistance in ESCC cells (Figure 8).