Circ0008399 Interaction with WTAP Promotes Assembly and Activity of the m⁶A Methyltransferase Complex and Promotes Cisplatin Resistance in Bladder Cancer

Bladder cancer is one of the most common tumors in the urinary system, and ranked 9th in global tumor incidence and 13th in global tumor-related mortality (1). In the United States, it is estimated that there will be 83,730 new cases of bladder cancer, and 17,200 Americans will die of this disease in 2021 (2). Cisplatin (CDDP)-based chemotherapy is the first-line treatment for muscle-invasive and metastatic bladder cancer. Although 60%–70% of patients have an initial response to CDDP, progression-free survival and overall survival (OS) of most patients with bladder cancer are not satisfactory due to the reduced responsiveness of CDDP within a short period of time (3). Therefore, the identification of key targets that mediate CDDP chemotherapy sensitivity is essential to provide progresses in precision diagnosis and treatment.

N⁶-methyladenosine (m⁶A) modification is the most abundant form in eukaryotic mRNA, and occurs in 0.1%–0.4% of the total adenosine residues in cellular mRNA (4). m⁶A methylation involves in the regulation of mRNA processing, including alternative splicing, translocation, stability, and translation, which affects a variety of biological processes of tumors (5). As a reversible epigenetic modification, m⁶A methylation relies on RNA methyltransferases “writers,” demethylases “erasers,” and m⁶A-binding proteins “readers” (6). Wilms' tumor 1–associating protein (WTAP), a vital component of the m⁶A RNA methyltransferase complex, recruits METTL3 and METTL14 to the corresponding mRNA targets to catalyze the formation of m⁶A (7). Accumulating lines of evidence have led to the view that m⁶A methyltransferases play a crucial role in CDDP chemosensitivity of various cancers (8). For example, METTL3 induces resistance to CDDP by increasing the extent of m⁶A modification of YAP in non–small cell lung cancer (NSCLC; ref. 9). Intervention of METTL3 could enhance the sensitivity to CDDP in pancreatic cancer (10). WTAP facilitates chemoresistance to CDDP in nasal-type natural killer/T-cell lymphoma by stabilizing DUSP6 mRNA in an m⁶A-dependent manner (11). However, the specific role and underlying mechanism of m⁶A methyltransferases in the regulation of CDDP chemotherapy in bladder cancer remain unclear.

Circular RNAs (circRNA) belong to a family of noncoding RNA (ncRNA) transcripts characterized by covalently closed continuous loops, which are more stable and more resistance to digestion with RNase R compared with linear transcripts (12). In 1979 years, circRNAs was observed in eukaryotic cells through electron microscopy, and was considered as a byproduct of reverse splicing of precursor mRNA with little functional potential (13). Following the development of next-generation RNA sequencing (RNA-seq) and bioinformatics technology, we recently found that circHIPK3, BCRC-3, circNR3C1 and has_circ_0001361 could play critical roles in cell proliferation, metastasis, and invasion during bladder cancer progression (14–17). Previous studies have shown that ncRNAs have been involved in modulating the expression and function of WTAP. For example, long noncoding RNA PCGEM1 promotes NSCLC proliferation, migration, and invasion via sponging miR-433-3p to upregulate WTAP expression (18). miR-550-1 impairs acute myeloid leukemia cell proliferation and tumorigenesis via targeting WTAP 3′-untranslated region to decrease WWTR1 mRNA stability (19). However, the role and mechanism of circRNAs in regulation of WTAP function still need to be clarified.

In this study, we focused on the upregulated circRNAs based on our high-throughput sequencing data of human bladder cancer and normal bladder tissues (14), and discovered that circ0008399, a novel circRNA generated from the circularization of exons 2,3, 4, 5, and 6 of RBM3 gene, was upregulated by Eukaryotic translation initiation factor 4A3 (EIF4A3) in bladder cancer tissues and cells. High expression of circ0008399 was associated with a poor OS. Circ0008399 suppressed cell apoptosis and decreased the chemosensitivity of bladder cancer to CDDP in vitro and in vivo. Mechanistically, circ0008399 bound with WTAP to facilitate the formation of the m⁶A methyltransferases complex, which promoted downstream target gene TNF alpha-induced protein 3 (TNFAIP3) expression via increasing its mRNA stability in an m⁶A-dependent manner, thereby decreasing caspase-8 activity. Taken together, this study delineates the novel mechanism of circRNA/WTAP-mediated m⁶A-dependent regulation of TNFAIP3 and highlights its role in regulation of CDDP chemosensitivity in bladder cancer.

A total of 72 pairs of bladder cancer tissues and adjacent normal (peritumor) tissues cases were obtained from patients undergoing radical cystectomy at the Department of Urology of Union Hospital affiliated of Tongji Medical College of Huazhong University of Science and Technology (Wuhan, P.R. China) from 2014 to 2018. All tissues were confirmed and classified by at least two experienced clinical pathologists independently according to the criteria of the 6th edition TNM classification of the International Union Against Cancer. This study was approved by the ethics review committee of Tongji Medical College of Huazhong University of Science and Technology (Wuhan, P.R. China) and written informed consent was obtained from all patients before the research started. Fresh tumor tissues were immediately frozen in liquid nitrogen, and then stored at −80°C. Detailed information was presented in Table 1 and Table 2. All of the patients were followed up regularly, and OS time was determined from the date of surgery to the date of death or the date of the last follow-up of the survivors.

Human bladder cancer cell lines UMUC3, RT4, and 5637 were purchased from ATCC. EJ cells were obtained from the Institute of Biochemistry and Cell Biology of Chinese Academy of Sciences (Shanghai, P.R. China). UROtsa cells were generously provided by Donald and Maryann Sens (University of North Dakota, Grand Forks, ND). The human metastatic bladder cancer cell line T24T was provided by Dan Theodorescu (Department of Urology, University of Virginia, Charlottesville, VA) as a gift and authenticated in our previous studies (20). CDDP-resistant EJ cell line (named EJ-CDDP), which we previously constructed via continuously exposing bladder cancer cells to stepwise escalating concentrations of cisplatin, as we described in recent research (21). T24T and UMUC3 cells were cultured at 37°C and 5% CO₂ atmosphere with DMEM (Thermo Fisher Scientific) plus 10% FBS (Gibco), 1% penicillin/streptomycin (Gibco), and EJ, RT4, and 5637 cells were cultured at 37°C and 5% CO₂ atmosphere with RPMI1640 medium (Gibco) plus 10% FBS (Gibco), 1% penicillin/streptomycin (Gibco). All bladder cancer cell lines were confirmed within 6 months before use by using a short tandem repeat profiling and were confirmed negative for Mycoplasma contamination.

Surgical specimens (paired normal and cancerous tissues) were obtained from 20 patients with kidney tumors in Department of Urology, Union Hospital, Tongji Medical College (Wuhan, P.R. China), and were freshly frozen in liquid nitrogen for RNA extraction. Human RCC cell lines (786-O, 769-P, A498, OSRC-2, and Caki-1) and the human renal proximal tubular epithelial cell line HK-2 were gifts from Yifei Xing (Union Hospital, Wuhan, P.R. China). These cells were cultured in high-glucose DMEM containing 10% FBS and 1% penicillin/streptomycin (Gibco). Human prostate cancer cell lines PC3, C4-2, DU145, LNCaP, 22RV-1 and immortalized human prostatic epithelial cell line RWPE-1 were kindly provided by Yifei Xing (Union Hospital, Wuhan, P.R. China). Prostate cancer cells were maintained in RPMI1640 medium (Gibco) supplemented with 10% of FBS (Gibco). Immortalized human prostatic epithelial cells were grown in Keratinocyte serum-free medium (Gibco) supplemented with bovine pituitary extract (0.05 mg/mL) and EGF (5 ng/mL).

Cisplatin (Sigma-Aldrich) was solubilized in DMSO or PBS. Caspase-8 inhibitor (HY-101297, MCE) was solubilized in DMSO.

Total RNA was isolated from tissues and cell lines by TRIzol reagent (Invitrogen) following the manufacturer's instructions. cDNA was synthesized by HiScript III RT SuperMix for quantitative PCR (qPCR; Vazyme). Genomic DNA (gDNA) was extracted from BC cells with DNA Mini Kit (QIAGEN). The PCR products of cDNA or gDNA were observed using 1.2% agarose gel electrophoresis. For circ0008399 detection, 2 μg of total RNA isolated from EJ and T24T cells was incubated with or without 3 U/μg RNase R (Epicenter) at 37°C for 15 minutes. The qRT-PCR was performed using SYBR Green Master Mix (Vazyme) and primers (Supplementary Table S1). The levels of circRNA and mRNA were normalized to those of GAPDH and determined by 2^−ΔΔC_t method. All results were analyzed using the Real-Time PCR System (Applied Biosystems).

Cy3-labeled circ0008399 probe was purchased from RiboBio. Meanwhile, U6 and 18S were also purchased from the RiboBio as nuclear internal control and cytoplasm control, respectively. RNA FISH was operated according to the manufacturer's instructions. Briefly, bladder cancer cells were fixed in 4% paraformaldehyde for 30 minutes and permeabilized with 0.5% Triton X-100 for 8 minutes at 4°. After prehybridization, cells were hybridized in hybridization buffer with specific probes at 37°C overnight. All data were obtained using Nikon A1Si Laser Scanning Confocal Microscope (Nikon Instruments Inc).

The siRNAs were synthesized by RiboBio, targeting circ0008399 splice sites (Supplementary Table S1). Short hairpin RNAs targeting WTAP, TNFAIP3, and EIF4A3 were synthesized by TSINGKE, and were cloned into the GV298/Cherry/Puro vector (BioVector; Supplementary Table S1). Human circ0008399 cDNA was cloned into pcDNA3.1(+) CircRNA Mini Vector (Geenseed Biotech), which contained a front circular frame and a back circular frame, to construct the circ0008399 overexpression plasmid. The human WTAP, TNFAIP3, and EIF4A3 cDNA were synthesized by TSINGKE, which were cloned into p3XFLAG-CMV-10 vector (Sigma-Aldrich) to construct overexpression plasmid. Stable cell lines were screened by administration of neomycin or puromycin (Invitrogen). The siRNAs were transfected into bladder cancer cells using RNAiMAX (Invitrogen) according to the manufacturer's instructions. To construct the circ0008399 stably transfected cell lines, plasmids were transfected into cells by using Lipofectamine 2000 (Life Technologies) according to the manufacturer's instructions, then the cells were screened with G418 (Invitrogen) for 4–6 weeks.

The biotin-labeled probe of circ0008399 was designed and synthesized by RiboBio. The sequence of circ0008399 antisense probe was just complemented to the backspliced junction of circ0008399 and listed in Supplementary Table S1. Pulldown assay was performed as described in our previously studies (14). Briefly, 2 × 10⁷ bladder cancer cells were fixed with 4% paraformaldehyde, lysed, sonicated, and centrifugated. Supernatant (50 μL) was maintained as input and the remaining was mixed with biotin-labeled antisense probes or sense probes and incubated with streptavidin C1 magnetic beads (Invitrogen) at 4°C overnight. Then, the beads were washed thoroughly with lysis buffer at least three times. The RNA–protein binding mixture was boiled in SDS buffer and the released proteins were detected by Western blot analysis and mass spectrometry (MS; Novogene).

Cellular protein was extracted using RIPA Lysis Buffer (Thermo Fisher Scientific) according to the manufacturer's instructions. The concentration of total protein was measured by BCA protein assay kit (Beyotime). Total protein was separated by 10% SDS-PAGE gels and transferred onto polyvinylidene difluoride membranes (Millipore) after electrophoresis was completed. After blocking for 1 hour in 5% non-fat dried milk at room temperature, membranes were incubated with primary antibodies overnight at 4°C, followed by hybridized with specific horseradish peroxidase (HRP)-conjugated secondary antibody at room temperature for 1 hour. Finally, the membranes were visualized using ECL substrate kit (Millipore) and the images were obtained via Bio Spectrum 600 Imaging System (UVP). Antibodies used included primary antibodies against GAPDH (catalog no. 60004-1-Ig, 1:1,000, Proteintech), WTAP (catalog no. 10200-1-AP, 1:1,000, Proteintech), METTL3 (catalog no. 15073-1-AP, 1:1,000, Proteintech), METTL14 (catalog no. 26158-1-AP, 1:1,000, Proteintech), TNFAIP3 (catalog no. 5630, 1:1,000, Cell Signaling Technology), caspase-8 (catalog no. 9496, 1:1,000, Cell Signaling Technology), SF3A2 (catalog no. 15596-1-AP, 1:1,000, Proteintech), UCHL5 (catalog no. 11527-1-AP, 1:1,000, Proteintech), EIF4A3 (catalog no. 17504-1-AP, 1:1,000, Proteintech), HRP-conjugated secondary goat anti-mouse (catalog no. SA00001–1, 1:4,000, Proteintech), goat anti-rabbit (catalog no. SA00001–2, 1:4,000, Proteintech).

Cells were seeded in a 15 cm dish until 70%–80% confluency about 1 × 10⁷ per sample, then cells were cross-linked at 254 nm (200 J/cm²) by UV light with ice for 1 minute. All cells were harvested and lysed to extract total protein. RNA immunoprecipitation assay (RIP) assay was performed according to the instructions of Magna RIP Kit (Millipore), with antibodies specific for WTAP (10220-1-Ag, Proteintech), Flag (ab45766, Abcam), m⁶A (56593, Cell Signaling Technology), EIF4A3 (17504-1-AP, Proteintech) or IgG (CS200621, Millipore), and protein A/G magnetic beads (Life Technologies). Input and coprecipitated RNAs were detected by qRT-PCR.

EJ and T24T cells were grown on confocal dishes, and treated with antibody specific for WTAP (Proteintech, 1:200 dilution) at 4°C overnight. Then the dishes were treated with Alexa Fluor 488 goat anti-rabbit IgG (ABclonal) for 1 hour at room temperature and DAPI (Servicebio) staining for 30 minutes for 37°. The images were captured under a Nikon A1Si Laser Scanning Confocal Microscope (Nikon Instruments Inc).

Briefly, 10⁷ cells were harvested and lysed with RIPA supplemented with protease inhibitors. Five percent of cell lysate was used as input, remaining was incubated with protein A/G beads for 2 hours at room temperature and washed twice with PBST. Then, the mixture was divided equally into two parts and incubated with 5 μg WTAP or IgG antibody overnight at 4°C, respectively. The complexes were washed twice with PBST, and all liquid was removed, and the proteins were removed from the beads with loading buffer under standard denaturing conditions. Purified proteins were detected by Western blotting. The following secondary antibodies were used for immunoblotting: HRP-mouse anti-rabbit IgG heavy chain specific (catalog no. SA00001–7H, Proteintech) or HRP-mouse anti-rabbit IgG light chain specific (catalog no. SA00001-7L, Proteintech) antibodies.

Total RNA was isolated from cells using TRIzol reagent (Invitrogen) according to the manufacturer's instruction. Quantification of the m⁶A modification was performed using the m⁶A RNA Methylation assay Kit from Abcam (ab185912). First, 2 μL of the negative control and 2 μL of the diluted positive control were added to a 96-well plate, and then 200 ng RNA sample was incubated with capture antibody solution in a suitable diluted concentration. Signaling was measured after adding 100 μL diluted developer solution to each well at room temperature for 5 minutes away from light and adding 100 μL stop solution to stop enzyme reaction within 5 minutes using spectrometer (Thermo Fisher Scientific) at 450 nm. Finally, relative m⁶A RNA methylation status was calculated using the manufacturer's supplied formula.

The transcriptome profiles of circ0008399-overexpressed EJ and T24T cells and the corresponding control cells were determined by UID-RNA-seq (SeqHealth Tech). Sequencing results were deposited in the Gene Expression Omnibus database https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE173132 with accession number cpshcccgdlovbyv.

Bladder cancer cells were seeded into 6-well plates to get 60% confluency. Cells were treated with 5 μg/mL actinomycin D and collected at indicated timepoints. The total RNA was extracted by TRIzol reagent (Invitrogen) and analyzed by qRT-PCR. The turnover rate and half-life of mRNA was estimated according to our previously published article (21).

Cell viability was detected by Cell Counting Kit-8 assay according to the manufacturer's instructions (Dojindo). Cells transfected with plasmids were cultured in 96-well plates of 10,000 per well and cultured for different time periods, respectively (0, 24, 48, and 72 hours). Then, a volume of 10 μL of CCK-8 solution was added to each well to incubate for 2 hours at 37°, and the absorbance at 450 nm was captured using spectrometer (Thermo Fisher Scientific).

Anchorage-independent growth ability was evaluated in soft agar. Briefly, 1 mL 1.2% agar and 1 mL 2 × 1640 medium supplemented with 20% FBS was layered onto each well of 6-well tissue culture plates. A total of 1 mL 0.7% agar and 1 mL 2 × 1640 medium contained 20% FBS with cells (10⁴ cells) was then layered on top of the 1.2% agar layer. Plates were incubated at 37°C in 5% CO₂ for 2–3 weeks, and cell colonies with more than 32 cells were counted.

EJ and T24T cells transfected with plasmids were harvested to analyze cell apoptosis by flow cytometry (Becton Dickinson) after stained with FITC Annexin V and propidium iodide (PI) staining (BD Pharmingen). PI negative and FITC Annexin V positive can identify apoptosis at an earlier stage, and cells that are in late apoptosis or already dead are both FITC Annexin V and PI positive. The results were analyzed with FlowJo software.

We chose 4-week-old female BALB/c nude mice for tumor xenograft experiments, which randomly were divided into four groups (n = 5 per group). Bladder cancer cells (3 × 10⁶) were subcutaneously injected into the right axilla of the nude mice. The tumor volume and weight of mice were monitored and recorded. Four weeks after injection, the mice were sacrificed. Tumor volume was calculated according to the formula (Tumor volume = π/6 × length × width²). All animal experiments were allowed in the light of NIH Guidelines for the Care and Use of Laboratory Animals and approved by the Animal Care Committee of Tongji Medical College. For in vivo drug studies, 1 week after cells injection, CDDP or PBS was administered by intraperitoneal injection three times per week at the dose of 2 mg/kg.

TUNEL staining was performed with the Fluorescein (FITC) Tunel Cell Apoptosis Detection Kit (G1501, Servicebio) in xenograft tumor tissues according to the manufacturer's instructions. Olympus FSX100 microscope (Olympus) was used to capture images. Semiquantification was performed for three independent fields in each slide.

All data were presented as mean ± SD. Analysis was performed using GraphPad Prism 8.0 software. Student t test was used to evaluate the group difference. The χ² test and Fisher exact test were used to analyze the association of the expression level of circ0008399 or TNFAIP3 with the patient's clinicopathologic characteristics. Kaplan–Meier survival analysis and log-rank test were used to assess survival difference. All statistical tests were considered statistically significant when P values less than 0.05.

Our previous studies have found that hundreds of circRNAs were differentially expressed in bladder cancer by RNA-seq analysis (14). According to these data, we verified that circ0008399, which is named hsa_circ_0008399 in CircBase (http://www.circbase.org/), was upregulated in bladder cancer tissues compared with paired normal tissues. The genomic structure indicates that circ0008399 consists of the head-to-tail splicing of exons 2,3,4,5, and 6 (510 nt) from the RBM3 gene (Fig. 1A). Subsequently, circ0008399 was amplified by RT-PCR with divergent primers and confirmed by Sanger sequencing (Fig. 1A). Nevertheless, we cannot rule out the possibilities that the head-to-tail splicing might be the result of trans-splicing or genome rearrangement (22). Therefore, we took two steps to rule out these possibilities. First, we designed convergent primers to amplify RBM3 linear mRNA form and divergent primers to amplify circ0008399, and both cDNA and gDNA extracted from EJ and T24T cells were used as templates. The results indicated that divergent primers could only amplify circ0008399 in cDNA, but not in gDNA (Fig. 1B). Second, we used RNase R digestion assays to prove that the circular form was resistant to RNase R, while the linear RNA was significantly reduced after RNase R treatment (Fig. 1C). Consistent with the RNA-seq results, circ0008399 was significantly upregulated in bladder cancer tissues (Fig. 1D). Furthermore, Kaplan–Meier survival analysis showed that higher expression of circ0008399 in patients with bladder cancer was associated with lower survival rate (Fig. 1E). In addition, circ0008399 was also expressed at high level in bladder cancer cell lines (Fig. 1F). FISH assay indicated that circ0008399 mainly localized in the nucleus of EJ and T24T cells (Fig. 1G). These results indicated that circ0008399 was relatively high expressed in bladder cancer tissues and cell lines, and was localized in cell nucleus.

RNA-binding proteins (RBP), such as QKI, FUS, EIF4A3, MBL, NF90, and HNRNPL, play an essential role in pre-mRNA splicing and circRNAs biogenesis (23–28). Using Circinteractome (https://circinteractome.nia.nih.gov) database for circRNAs–protein interaction, we discovered 13 putative binding sites of EIF4A3 in the upstream and downstream region of the circ0008399 mRNA transcript (RBM3 pre-mRNA), while there were no binding sites of QKI, FUS, MBL, NF90, and HNRNPL (Supplementary Fig. S1A). RIP assay indicated that EIF4A3 could bind to RBM3 pre-mRNA in bladder cancer cells (Supplementary Fig. S1B). Importantly, we found that EIF4A3 was upregulated in bladder cancer tissues and cells (Supplementary Fig. S1C and S1D), and the positive correlation between the transcript levels of EIF4A3 and circ0008399 was observed in bladder cancer tissues (Supplementary Fig. S1E). Next, we established stable models of EIF4A3 overexpression and knockdown in EJ and T24T cell lines (Supplementary Fig. S1F and S1G), and demonstrated that ectopic expression of EIF4A3 facilitated the expression of circ0008399, while knockdown of EIF4A3 decreased the level of circ0008399 in bladder cancer cells (Supplementary Fig. S1H and S1I). These results showed that EIF4A3 could bind with RBM3 pre-mRNA to promote circ0008399 formation.

To explore the expression level of circ0008399 in other types of cancers, we then detected the expression of circ0008399 in clear cell renal cell carcinoma (ccRCC) and prostate cancer. However, we found that both circ0008399 and EIF4A3 were significantly downregulated in ccRCC tissues (Supplementary Fig. S1J and S1K), and the expression levels of EIF4A3 and circ0008399 were positively correlated (Supplementary Fig. S1L). Accordingly, as shown in Supplementary Fig. S1M and S1N, circ0008399 and EIF4A3 were also downregulated in ccRCC cells. On the other hand, the expression of circ0008399 and EIF4A3 in prostate cancer cell lines had no statistical difference compared with normal prostate epithelial cells (Supplementary Fig. S1O and S1P). Taken together, these results suggested that the differential expression of circ0008399 was tissue specific.

Similar to other ncRNAs, circRNAs that are located in the cytoplasm have been suggested to act as miRNA sponges or regulate the function of RBPs, while they mainly interact with RBPs when they are located in the nucleus (29). Given that circ0008399 was located in the nucleus, we tended to perform RNA pulldown assays to explore its role of interacting with RBPs. The proteins pulled down through biotin-labeled cir0008399 antisense or sense probes were detected by Western blotting and silver staining (Fig. 2A). Then, we performed MS analysis, and found that there were 27 different RBPs after overlapping the proteins pulled down in EJ and T24T cells. Three RBPs with protein molecular weight between 40–55 kDa were consistently pulled down by biotin-labeled circ0008399 antisense probe in both cell lines, including WTAP, SF3A2, and UCHL5 (Fig. 2B). In addition, MS analysis showed that AGO2 protein could not bind to circ0008399, which reminded us that circ0008399 exerted functions other than miRNA sponge. Validating biotin-labeled RNA pulldown assay further indicated that circ0008399 antisense probe was able to dose dependently interact with WTAP, but not with SF3A2 or UCHL5 (Fig. 2C). The identified peptides of WTAP from MS assay were shown in Fig. 2D. The interaction between circ0008399 and WTAP was further validated through RIP assay (Fig. 2E). Furthermore, dual RNA-FISH and immunofluorescence assay demonstrated the colocalization of circ0008399 and WTAP in EJ and T24T cells (Fig. 2F).

To delineate the structural determinants of the interaction between circ0008399 and WTAP, we constructed truncations of Flag -tagged recombinant WTAP protein. In vitro binding assay showed that removal of 101–240 amino acids region of FLAG-tagged WTAP protein, but not other domains, significantly abolished the interaction between WTAP and circ0008399, indicating that 101–240 amino acids region of WTAP was crucial for its interaction with circ0008399 (Fig. 2G and H). These results proposed that circ0008399 could interact with WTAP protein in the nucleus of bladder cancer cells.

Subsequently, EJ and T24T cells with stable overexpression of circ0008399 were constructed (Fig. 3A). Overexpression of circ0008399 had no effect on both mRNA and protein levels of WTAP (Fig. 3B). Meanwhile, we found no significant difference of circ0008399 levels between WTAP-overexpressed cells and control cells (Fig. 3C). These results indicated that circ0008399 and WTAP do not affect the expression of each other. Considering that WTAP is an important component that regulating m⁶A methylation modification, we then detected the total m⁶A levels by the RNA m⁶A quantification assay, and we found that overexpression of WTAP and circ0008399 could respectively upregulate the total m⁶A levels (Fig. 3D and E). On the other hand, knockdown of WTAP and circ0008399 decreased global m⁶A modification levels in bladder cancer cells (Fig. 3F–I). Furthermore, ectopic expression of WTAP reversed the reduction of m⁶A modification levels mediated by silencing of circ0008399 (Fig. 3I). It has been reported that WTAP, METTL3, and METTL14 composed the m⁶A methyltransferase complex to catalyze m⁶A modification (7). Next, we performed coimmunoprecipitation (Co-IP) assay to confirm endogenous binding of WTAP, METTL3, and METTL14 in EJ and T24T cell lines (Fig. 3J). We further observed that knockdown of circ0008399 reduced the interaction of endogenous WTAP with METTL3 and METTL14 (Fig. 3K). Moreover, ectopic expression of WTAP reversed the disassembly of “writer” complex induced by silencing of circ0008399 (Fig. 3K). These results implied that circ0008399 could facilitate m⁶A methyltransferases complex assembly to increase global m⁶A modification levels.

To further investigate the target genes of circ0008399, transcriptome analysis was performed in EJ and T24T cells stably transfected with circ0008399 (Fig. 4A). Comprehensive analysis of WTAP, METTL3, and METTL14 correlating genes (Supplementary Table S2) from The Cancer Genome Atlas (TCGA) database with our RNA-seq results indicated that four targets might be regulated by circ0008399 (Fig. 4B), including TCHH, TNFAIP3, IFIT2, and PAPPA. Next, we used qRT-PCR to identify whether overexpression of circ0008399 could upregulate these candidate mRNAs. As shown in Fig. 4C and Supplementary Fig. S2A, TNFAIP3 was the only one that was abundantly detected in EJ and T24T cells stably transfected with circ0008399. Consistently, there was a positive correlation between circ0008399 and TNFAIP3 in bladder cancer tissues (Fig. 4D). Furthermore, m⁶A-RIP assay demonstrated that overexpression of circ0008399 enhanced the m⁶A modification level of TNFAIP3 mRNA, whereas the changes in the m⁶A modification levels of other genes were not significant in bladder cancer cells (Fig. 4E; Supplementary Fig. S2B). On the other hand, knockdown of circ0008399 significantly reduced the m⁶A modification level of TNFAIP3 mRNA, which was reversed by ectopic expression of WTAP (Fig. 4F; Supplementary Fig. S2C). In addition, overexpression of WTAP could reverse the reduction of the stability of TNFAIP3 mRNA induced by silencing of circ0008399 (Fig. 4G; Supplementary Fig. S2D).

Subsequently, we confirmed that overexpression of circ0008399 and WTAP could promote the protein expression of TNFAIP3 (Fig. 4H and I). Previous studies revealed that TNFAIP3(A20) could decrease caspase-8 activity (30, 31). In this study, we found that overexpression of TNFAIP3 significantly prohibited caspase-8 cleavage in bladder cancer cells (Fig. 4J; Supplementary Fig. S2E). Importantly, knockdown of circ0008399 significantly decreased TNFAIP3 protein expression level and increased caspase-8 cleavage, which could be reversed by WTAP overexpression (Fig. 4K). Meanwhile, overexpression of TNFAIP3 reversed the increasing of caspase-8 cleavage induced by silencing of circ0008399 (Fig. 4L). Hence, these results suggested that circ0008399/WTAP promoted the expression of TNFAIP3 via increasing its mRNA stability in m⁶A-dependent manner in bladder cancer cells.

Previous studies have indicated that WTAP plays an oncogenic function in the occurrence and development of cholangiocarcinoma, renal cell carcinoma, hepatocellular carcinoma, and others (32), and WTAP might act as a new tumor marker for the diagnosis of bladder cancer (33). We found that the expression level of WTAP in bladder cancer tissues was significantly increased (Fig. 5A). Besides, Kaplan–Meier survival analysis of 72 BC samples demonstrated that high expression of WTAP was associated with poor OS of patients (Fig. 5B). Functionally, we discovered that knockdown of WTAP significantly induced cell apoptosis (Fig. 5C and D), and attenuated the viability and anchorage-independent growth of bladder cancer cells (Supplementary Fig. S3A and S3B). To further confirm in vitro results, we observed the biological roles of WTAP in vivo. Consistently, knockdown of WTAP resulted in an obvious decreasing in tumor volume and weight of subcutaneous xenograft tumors (Fig. 5E–G). Meanwhile, knockdown of WTAP promoted cell apoptosis in vivo (Fig. 5H). Accordingly, interfering of circ0008399 significantly increased percentage of apoptotic cells (Fig. 5I and J), and suppressed the viability and anchorage-independent growth of bladder cancer cells (Supplementary Fig. S3C and S3D). Overexpression of WTAP could impair the promotion of bladder cancer cell apoptosis induced by circ0008399 knockdown (Fig. 5K and L), as well as reverse the reduction of viability and anchorage-independent growth ability (Supplementary Fig. S3E and S3F). Moreover, knockdown of TNFAIP3 also promoted cell apoptosis (Fig. 5M and N), and decreased the viability and the anchorage-independent growth ability of EJ and T24T cells (Supplementary Fig. S3G and S3H). Similarly, ectopic expression of TNFAIP3 reversed the promotion of cell apoptosis induced by silencing of circ0008399, and the reduction of viability and anchorage-independent growth ability in bladder cancer cells (Fig. 5O and P; Supplementary Fig. S3I and S3J). Altogether, these results demonstrated that circ0008399/WTAP complex could suppress the apoptosis via promoting TNFAIP3 expression in bladder cancer cells.

Given that circ0008399/WTAP complex played an important role in inhibiting apoptosis of bladder cancer cells, its function in regulating of CDDP sensitivity was subsequently investigated. We found that knockdown of circ0008399 synergized with silencing of WTAP to enhance the effect of CDDP in EJ and T24T cells (Fig. 6A–C; Supplementary Fig. S4A–S4C). Next, we performed gene set enrichment analysis (GSEA) based on mRNA expression profile from TCGA database. The results indicated that high level of TNFAIP3 was positively associated with the upregulated gene set with CDDP resistance in bladder cancer cells (Fig. 6D). Moreover, it showed that CDDP-induced cell apoptosis was significantly weakened by TNFAIP3 overexpression (Supplementary Fig. S4D and S4E), and was enhanced by knockdown of TNFAIP3 in bladder cancer cells (Fig. 6E; Supplementary Fig. S4F). Importantly, overexpression of TNFAIP3 significantly rescued the promotion of CDDP chemosensitivity in EJ and T24T cells induced by silencing of WTAP (Fig. 6F; Supplementary Fig. S4G). We further confirmed that knockdown of circ0008399 induced cell apoptosis upon CDDP treatment, which was partially reversed by caspase-8 inhibitor (Fig. 6G; Supplementary Fig. S4H).

Next, we performed experiments with CDDP-resistant EJ cell line (named EJ-CDDP), which exhibited a high level of resistance to CDDP (Supplementary Fig. S5A). The expression level of circ0008399, WTAP, and TNFAIP3 in EJ-CDDP cells was upregulated compared with its parental control (Supplementary Fig. S5B–S5D). Importantly, silencing of circ0008399 could sensitize EJ-CDDP–resistant cells to CDDP-induced apoptosis (Supplementary Fig. S5E). Furthermore, knockdown of WTAP or TNFAIP3 could also promote apoptosis in EJ-CDDP–resistant cells upon CDDP treatment (Supplementary Fig. S5F and S5G). Collectively, these results suggested that silencing of circ0008399/WTAP/TNFAIP3 axis overcame acquired resistance of bladder cancer cells to CDDP in vitro.

To further confirm the in vitro findings, T24T cells stably transfected with circ0008399 were injected subcutaneously into BALB/c nude mice, followed by intraperitoneal injection with CDDP. Supporting the results obtained in vitro, xenograft experiments showed that circ0008399 increased the tumor formation and weakened the sensitivity of CDDP in vivo (Fig. 6H). On the other hand, as shown in Fig. 6I, knockdown of WTAP decreased the tumor weights, and promoted the sensitivity of CDDP chemotherapy. Meanwhile, knockdown of TNFAIP3 could also inhibit the tumor formation and enhance the cytotoxicity of CDDP in vivo (Fig. 6J).

In general, the current study indicated that circ0008399/WTAP complex facilitated TNFAIP3 expression via increasing its mRNA stability in an m⁶A-independent manner, highlighting that therapeutic targeting of the circ0008399/WTAP/TNFAIP3 pathway shed light on improving the sensitivity of CDDP chemotherapy in bladder cancer.

With the development of biochemical enrichment strategies and deep sequencing, numerous and evolutionarily conserved circRNAs have been identified in mammalian cells and tissues (22). However, the mechanisms of circRNAs biogenesis are still quite elusive (29). Herein, we discovered that EIF4A3, an important component of RNA splicing, could combine with the region of RBM3 pre-mRNA and promote circ0008399 expression at the posttranscription level in bladder cancer. Previous studies show that circRNAs play a variety of important roles in cellular physiology via acting as miRNA sponges, RBP-binding molecules, transcriptional regulators, or protein translation template (34). In our previous studies, we have identified that circHIPK3 and circ0001361 can act as “miRNA sponge” to regulate the progression of bladder cancer (14, 17). Recently, we reported that circPTPRA could interact with IGF2BP1, a key m⁶A reader protein, and suppress the growth and aggressiveness of bladder cancer through downregulating IGF2BP1 regulation of MYC and FSCN1 expression (35). In this study, we discovered that circ0008399 could interact with WTAP, a pivotal m⁶A writer protein, and mediated m⁶A modification of downstream mRNA target. These findings for the first time expound that circRNA could bind to the key m⁶A writer, which revealed the novel role of circRNA in regulation of mRNA m⁶A modification.

circRNAs have unique covalently closed loop structures and specific tertiary structures and play crucial roles in circRNA–protein interactions (36). However, the underlying mechanism by which circRNA modulates protein–protein interaction is not clearly understood. Previous studies have shown that CDR1as blocks the interaction of p53 with MDM2 by binding to the DBD domain in the middle of the p53 protein, thereby reducing the ubiquitination of p53 and protecting cells from DNA damage (37), while MDM2 binds to the p53 N-terminus distal to the DBD domain (38). On the other hand, circ-Foxo3 binding to the C-terminal RING-finger domain of MDM2 protein promotes the interaction between MDM2 and p53, which increases Foxo3 and its downstream gene Puma expression and subsequently enhances cell apoptosis (39), while p53 binds to the N-terminal domain of MDM2 (40). In this study, circ0008399 promoted the binding of WTAP to the other important m⁶A methyltransferases METTL3 and METTL14. Notably, we demonstrated that circ0008399 bound the 101–240 amino acids regions of WTAP protein to enhance m⁶A methyltransferases complex assembly, facilitating m⁶A modification of TNFAIP3 mRNA. Objectively, nuclear localization signals of WTAP protein' N terminal is the key domain combined with METTL3/14 heterodimer (41). From these, we speculate that the binding of circRNAs with RBPs might change the spatial distance of these proteins, expose or mask their active sites, as well as might change their spatial conformation, which in turn affects the interaction between RBPs and other proteins.

TNFAIP3, also known as A20, is described as an anti-apoptotic protein that inhibits TNF-induced apoptosis (42). It is also an NFκB-responsive gene that plays a role in negative feedback mechanism to prevent sustained NFκB activation (43). Previous study has shown that high level of TNFAIP3 has a positive correlation with the tumorigenesis in bladder cancer (44). In the current study, we demonstrated that knockdown of TNFAIP3 significantly increased the percentage of apoptotic bladder cancer cells. Furthermore, we found that circ0008399 interacted with WTAP to promote the expression of TNFAIP3 via enhancing its mRNA stability in an m⁶A-dependent manner, which provided further evidences for the posttranscriptional regulation of TNFAIP3. It has been reported that m⁶A methyltransferases could not only enhance downstream genes expression through modulating their mRNA stability (45), but also promote protein expression via m⁶A-dependent translation process (46). Thus, circ0008399/WTAP-induced TNFAIP3 protein expression might also partially depend on the m⁶A-dependent translation process. Considered as one of the most common epigenetic RNA modifications, m⁶A can regulate drug sensitivity in human cancers (47). Emerging evidences indicate that m⁶A modification can lead to drug resistance by modulating ATP-binding cassette (ABC) transporters and cancer stem cells (48). For example, IGF2BP3 promotes the stability and expression of ABC transporter ABCB1 mRNA via binding to its m⁶A-modified region, thus inducing multidrug resistance in colorectal cancer (49). Besides, METTL3-mediated m⁶A modification could upregulate CBX8 expression through enhancing the stability of its mRNA, which promotes stemness and attenuates chemosensitivity of colon cancer (50). In the current study, we elaborated that m⁶A modification induced by circ0008399/WTAP could promote the stability and expression of TNFAIP3 mRNA, which resulted in suppressing cell apoptosis and attenuating CDDP chemosensitivity in bladder cancer. Nevertheless, whether circ0008399/WTAP-mediated m⁶A modification regulates CDDP chemosensitivity of bladder cancer by modulating ABC transporters and cancer stem cells still remains to be further elucidated.

In summary, we find that circ0008399, a novel circRNA, is upregulated in bladder cancer. High expression of circ0008399 is associated with poor survival of patients. Functionally, circ0008399 suppresses apoptosis and decreases the chemosensitivity of bladder cancer to CDDP. Mechanistically, circ0008399 cooperates with WTAP to promote target gene TNFAIP3 expression via enhancing its mRNA stability in an m⁶A-dependent manner. Importantly, our study reveals that therapeutic targeting of the circ0008399/WTAP/TNFAIP3 pathway is a potential strategy to sensitize bladder cancer to CDDP chemotherapy.

No disclosures were reported.

W. Wei: Data curation, formal analysis, methodology, writing–original draft. J. Sun: Resources, software. H. Zhang: Formal analysis, validation, methodology. X. Xiao: Supervision, funding acquisition. C. Huang: Data curation, methodology. L. Wang: Validation. H. Zhong: Formal analysis. Y. Jiang: Data curation. X. Zhang: Supervision, project administration. G. Jiang: Supervision, funding acquisition, project administration, writing–review and editing.

The authors owe their thanks to the patients. The authors are grateful to Union Hospital of Tongji Medical College of Huazhong University of Science and Technology (Wuhan, P.R. China).

This work was supported by the National Natural Science Foundation of China (nos. 81874091, 81772724, 81974396, 82072840, 82102734).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.