WTAP facilitates progression of hepatocellular carcinoma via m6A-HuR-dependent epigenetic silencing of ETS1

Samples of tumor tissues and adjacent tissues were gathered from 69 HCC patients who had received curative surgery from 2016 to 2017 at First Affiliated Hospital of Zhejiang University. Frozen tissues were subjected to mRNA or protein extraction for reverse transcription quantitative real-time PCR (RT-qPCR) or western blotting analysis, and a cohort including 32 pairs of paraffin-embedded tissues (without follow-up data) (named as cohort1) was applied for immunohistochemistry (IHC) analysis. Another cohort was based on tissue microarrays (TMA) which was purchased from Shanghai Outdo Biotech (HLivH180Su14) containing 90 pairs of HCC tissues and matching normal tissues with complete clinicopathological information and follow-up data (listed detailedly in Table 1). Written informed consents were obtained from each enrolled subject according to the guidelines of the Declaration of Helsinki. And this study was approved by Institutional Ethics Committee in the hospital.

The human HCC cell lines Huh7, PLC/PRF/5, Hep3B, HCCLM3, MHCC97H, SMCC7721 and an immortalized hepatocyte cell line QSG7701 were purchased from Shanghai Institutes of Biological Sciences, Chinese Academy of Sciences (Shanghai, China). All cells were routinely cultured with Minimum Essential Media (MEM, Gibco) supplemented with 10% fetal bovine serum (FBS, Gibco) and incubated in 5% CO2, 37 °C incubator (Thermo Scientific, USA) with a humidified atmosphere.

Total RNA was isolated using ESscience RNA-Quick Purification Kit (YiShan Biotech, Shanghai, China), followed by cDNA synthesis with HiScript II Q RT SuperMix for qPCR (Vazyme Biotech, Nanjing, China). Expression of RNA was measured by SYBR Green (Vazyme Biotech, Nanjing, China) operated on the Bio-Rad QX100 Droplet Digital PCR system (USA), and relative RNA amount was calculated by 2^ΔΔCt method with the normalization to GAPDH. All premiers were derived from Tsingke Biological Technology (Beijng, China) and summarized in Additional file 1: Table S1.

Total protein was extracted from tissues or cells using pre-cooled RIPA buffer (Beyotime, Shanghai, China) containing protease and phosphatase inhibitors (Thermo Scientific, USA). Quantification of protein was conducted with Bicinchoninic Acid Protein Assay Kit (Thermo Scientific, USA). An equal amount of protein samples was separated by 4–12% SDS-PAGE (GenScript, Nanjing, China) and then transferred to 0.45 μm PVDF membranes (Millipore, USA). After being blocked by 5% non-fat milk in TBST for 1 h, membranes were incubated with corresponding primary antibodies at 4 °C overnight. Washed by TBST for three times, they would be incubated with HRP-conjugated secondary antibodies for 1 h at room temperature. The immunoblots were detected with an imaging system (Bio-Rad, USA) using enhanced chemiluminescence detection kit (Servicebio, Wuhan, China). GAPDH and β-actin were selected to be the loading controls. All the antibodies employed in this study were listed in Additional file: Table S2. 2

The TMA cohort was utilized to construct the connection between WTAP and prognosis of HCC patients, and the paraffin embedded specimens (cohort1) were applied to certify the relationship of WTAP and ETS1. They were subjected to IHC staining using two-step method of Dako Envision™ Detection System (DakoCytomation, Glostrup, Denmark). IHC score was computed via multiplying staining intensity grade (grade of 0, 1, 2 or 3 implied negative, weak-positive, moderate-positive or strong-positive, respectively) by positive rate score (score of 0, 1, 2, 3 or 4 represented positive areas of 0–5%, 6–25%, 26–50%, 51–75% or 76–100%, respectively). The score was assessed by two proficient pathologists independently.

Small interfering RNAs (siRNA) directed against WTAP, ETS1, HuR, YTHDF2, ETS2 and negative control RNAs (siNC) were synthesized by GenePharma Company (Shanghai, China). Transient transfection was performed by jetPRIME Polyplus kit (France) in accordance with the standard protocol. Cells were harvested after 48 h for qRT-PCR examination or after 72 h for immunoblot analysis and functional study. All the siRNA sequences were generalized in Additional file: Table S3. 3

Lentiviruses expressing shWTAP (#1, #3) or shNC, Flag-WTAP (WTAP-OE) or empty vector (WTAP-NC), and Flag-ETS1 (ETS1-OE) or empty vector (ETS1-NC) were purchased from Genecopoeia (USA). Huh7 and PLC/PRF/5 cells were chosen to establish stable WTAP knockdown models, while SMMC-7721 and HCCLM3 cells were used for stable WTAP overexpression experiments. 10⁵ cells were planted into a well and transfected with indicated lentivirus according to the instructions. Infected cells were selected using 3 μg/ml puromycin (MCE, USA) for 1 week or more, with the transfection efficiency determined by RT-qPCR and WB analysis. All the targeted sequences were listed in Additional file 3: Table S3.

Cell proliferation ability was measured by Cell Counting Kit-8 (CCK-8, Dojindo Laboratories, Kumamoto, Japan). 2 × 10³ cells were seeded into a 96-well plate per well with six duplications, followed by incubation for 2 h at 37 °C. Absorbance was detected at 450 nm daily for 4 consecutive days. For colony formation assay, 1.5× 10³ treated cells were coated into 6-well plates with three repetitions. After 14-day incubation, these plates were washed with phosphate buffered saline (PBS) twice, fixed by methanol for 10 min and stained with 0.1% crystal violet solution within 10 min for further analysis.

The treated cells were collected and fixed with chilled 75% ethanol at − 20 °C overnight or longer. After ethanol discarded, cells were washed twice with PBS and resuspended with 250 μl DNA staining solution (Multiscience, China) at room temperature for 30 min. Cell cycle analysis was performed on the flow cytometry (FACS LSRII, BD Bioscience, USA).

A 5-ethynyl-20-deoxyuridine (EdU) assay kit (Ribobio, Guangzhou, China) was adopted to inquire the cell proliferation ability. Cells were seeded into confocal plates with a density of 10 × 10⁵ cells each well. And they were incubated with 50 μM EdU buffer at 37 °C for 2 h, fixed with 4% formaldehyde for 0.5 h and permeabilized with 0.1% Triton X-100 for 20 min. EdU solution was added into culture followed by the staining of nuclei with Hoechst. Then the results were visualized by a fluorescence microscope.

Migration or invasion assays were performed using 24-well plates inserted by 8-μm pore size transwell filter insert (Corning, NY, USA) with or without pre-coated diluted Matrigel (1,10) (Becton Dickinson, San Jose, CA, US). 5 × 10⁴ HCC cells with Serum-free medium were placed into the upper chamber, and medium containing 10% FBS was added into the bottom chamber subsequently. After incubation in 37 °C for 24 h (migration) or 48 h (invasion), cells on the underside of membrane were immobilized and stained with crystal violet (Sangon Biotech, China). Then penetrated cells were counted in five random fields under the microscope.

All animal experiments were approved by the Ethics Committee for Laboratory Animals of the First Affiliated Hospital of Zhejiang University. Four-week old Balb/c male nude mice were purchased from Shanghai Experimental Animal Center of Chinese Academic of Sciences (Shanghai, China). 3 × 10⁶ treated HCC cells resuspended in 100 μl PBS were subcutaneously injected to the left flank of the mice, which were randomly divided into several groups. Tumor sizes were measured every 3 to 5 days. At the end of feeding (3 weeks or more), mice were sacrificed with the tumors removed for histology analysis. The tumor weight was recorded and the volume was estimated with the formula: 1/2*(length×width²). And extreme values (maximum and minimum) were eliminated.

Total RNA was extracted with ESscience RNA-Quick Purification Kit (YiShan Biotech, Shanghai, China). Then the library construction and RNA-sequencing (RNA-seq) were performed at Shanghai Sinomics Corporation (Shanghai, China) with Illumina NovaSeq 6000 (Illumina, USA), followed by the computational analysis they provided. The criteria for differential genes was set up with P value < 0.01 and fold change > 1.5 or < 0.5.

Total RNA was extracted by ESscience RNA-Quick Purification Kit (YiShan Biotech, China), followed by the purification of polyadenylated mRNA using GenElute™ mRNA Miniprep Kit (Sigma-Aldrich, Germany) according to manufacturer’s protocol. Then EpiQuik™ m6A RNA Methylation Quantification Kit (Colorimetric) (Epigentek, USA) was utilized to determine the total level of m6A in treated cells. Briefly, 200 ng RNA was added to each well with the capture antibody and detection antibody mixed in subsequently. After several incubations, the m6A content was quantified colorimetrically at the wavelength of 450 nm and calculated based upon the standard curve.

Total RNA or poly(A) + mRNA was isolated as described above. mRNA samples dissolved in 3 times volume of RNA incubation buffer were denatured at 65 °C within 5 min. Then the samples, divided into subgroups of 400 ng, 200 ng and 100 ng, were loaded to an Amersham Hybond-N+ membrane (GE Healthcare, USA) installed in a Bio-Dot Apparatus (Bio-Rad, USA) with the mixture of ice-cold 20*SSC buffer (Sigma-Aldrich, Germany). The membrane was UV crosslinked for 5 min and washed with PBST. Whereafter, it was stained with 0.02% Methylene blue (Sangon Biotech, China), followed by the scanning to indicate the total content of input RNA. After being blocked with 5% non-fat milk, the membrane was incubated with specific m6A antibody (1:1000, Millipore) overnight at 4 °C. Dot blots were hatched with HRP-conjugated anti-mouse immunoglobulin G (IgG) for 1 h before visualized by an imaging system (Bio-Rad, USA).

RIP assay was carried out in Huh7 and PLC/PRF/5 cells using Magna RIP Kit (17–700, Millipore, MA) following the manufacturer’s instructions. In brief, magnetic beads pre-coated with 5 mg normal antibodies against HuR (Santa Cruz) or mouse IgG (Millipore) were incubated with sufficient cell lysates (more than 2*10⁷ cells per sample) at 4 °C overnight. And the beads containing immunoprecipitated RNA-protein complex were treated with proteinase K to remove proteins. Then interested RNAs were purified by TRIzol methods (ThermoFisher Scientific) and detected by RT-qPCR with the normalization to input (fold change was calculated for comparison).

Cells with stable knockdown of WTAP were subjected to MeRIP assay using Magna MeRIP™ m6A Kit (17–10,499, Millipore, MA) in accordance with manufacturer’s recommendations. In short, 200 μg total RNA was isolated and randomly fragmented into 100 nucleotides or less, followed by the immunoprecipitation with 10μg m6A antibody (MABE1006, Millipore) or anti-mouse IgG which was linked to Magna ChIP Protein A/G Magnetic Beads. To determine the appropriate ratio between RNA and antibody, a dilution assay had been applied to optimize the MeRIP system and ensure the m6A antibody was not saturated. And one-tenth volume of fragmented RNA was saved as “10% input”. Elution of m6A-precipitated RNA was based on 6.7 mM N6-methyladenosine 5′-monophosphate sodium salt. And modification of m6A towards particular genes was determined by qPCR analysis with specific primers (All primers for MeRIP-qPCR were listed in Additional file 1: Table S1) Note: m6A sites of specific genes were predicted in RMBase v2.0 (http://rna.sysu.edu.cn/rmbase/↗) and SRAMP (http://www.cuilab.cn/sramp↗) (Additional file 5: Data S1). We focused on the potential m6A sites in 3′ UTR near the stop codon and designed primers to ensure that target sequence embodied all these sites with the limited length of 100 nt.

cDNAs containing partial CDS sequence near stop codon and full-length 3’UTR of ETS1 were cloned into pGL3-control vectors (Promega) which was comprised of firefly luciferase(F-luc). For mutant 1 or 2 reporter plasmids, 4 or 11 adenosine (A) in m6A motif were replaced by cytosine (C), respectively. The inserted sequences were listed in Additional file: Data S2. Pre-treated HCC cells were seeded into 24-well plate followed by co-transfection of 0.5μg of wild-type or mutated ETS1 reporter plasmids and 25 ng pRL-TK plasmids (renilla luciferase reporter vector) using jetPRIME Polyplus kit. After 24–36 h, cells were harvested to access the luciferase activity using Dual-Glo Luciferase system (Promega) with the normalization to pRL-TK. Each group was conducted in triplicate. 6

To evaluate the RNA stability, RNA decay assay was conducted. HCC cells were cultured in four 6-well plates followed by the treatment of WTAP or HuR knockdown. Then Actinomycin D (MCE, HY-17559) was added into each well with a final concentration of 5 μg/mL. And cells were collected after 0, 1, 2, 4 h, respectively. Total RNA was isolated and subject to RT-qPCR subsequently to quantify the relative abundance of ETS1 mRNA (relative to 0 h).

ChIP assay was carried out in Huh7 and MHCC97H cells with Magna ChIP™ A/G kit (17–10,085, Millipore, MA) according to manufacturer’s protocols. In brief, 10⁷ cells fixed with formaldehyde were collected and subject to 500ul lysis buffer. Then lysate was sonicated for 25 cycles of 6-s power-on and 30-s interval with intensity of 200 W. Afterwards, the supernatant was diluted and fully mixed with Protein A/G magnetic beads. Then 5μg of IgG, ETS1 or anti-RNA Polymerase II antibody was added in respectively, followed by incubation at 4 °C overnight. The next day, after washing, the mixture was incubated with elution buffer at 62 °C for 2 h and then at 95 °C for 10 min. Then DNA was purified from the elution and was subject to RT-qPCR. Specific primers for p21 and p27 was designed with the help of JASPAR (http://jaspar.genereg.net/↗) and PROMO (http://alggen.lsi.upc.es/↗).

HCC cells were treated with 3-Deazaadenosine (DAA, B6121, APExBIO) in the concentration of 0, 100, 200uM, or with cyclolencine (A48105, Sigma-Aldrich) in the concentration of 0, 50, 100 mM for 24 or 48 h, followed by the RT-qPCR or western blotting analysis to examine the expression of ETS1 or ETS2.

Co-Immunoprecipitation (Co-IP) assay was performed in Huh7 cells using Dynabeads™ Co-Immunoprecipitation Kit (14321D, ThermoFisher Scientific, USA). According to the manufacturer’s manuals, the protein extracted with IP Lysis Buffer was subject to beads premixed with antibodies of WTAP or IgG. The immunoprecipitated protein complex was separated from beads after several washes, followed by the identification for partners of WTAP by immunoblots.

Cells (4*10⁴) were cultured on confocal dishes, fixed with 4% formaldehyde for 20 min, and permeabilized with 0.1% Triton X-100 for 20 min. After blocking with 5% BSA (in TBST), cells were incubated with the primary antibodies against WTAP (proteintech, 1:100) and ETS1 (CST, 1:200) at 4 °C overnight. The secondary antibodies were then incubated with the cells at room temperature for 30 min, and DAPI was applied to stain the nuclei. Immunofluorescence (IF) images were obtained with confocal microscopy (Olympus, Japan).

The Nuclear and Cytoplasmic Protein Extraction Kit (Beyotime, Shanghai, China) was used to isolate the nuclear and cytoplasmic components of HCC cells. The detailed procedures were operated in accordance with the protocols of manufacturer. GAPDH or LaminB1 were served as loading controls for nuclear or cytosolic fraction, respectively.

Statistical analyses were conducted with the SPSS 22.0 (SPSS, Inc., Chicago, IL, USA) and GraphPad Prism 7.0 software (GraphPad, Inc., La Jolla, CA, USA). Experiments were repeated at least for three times independently. Measured data were represented as the mean ± SD. One-way analysis of variance (ANOVA) or two-tail Student t test was applied to compare quantitative data, while the nonparametric χ2 test was used to analyze qualitative data. Cox proportional hazard regression model was employed for univariate or multivariate analysis to explore independent prognostic factors. The overall or disease-free survival was analyzed with Kaplan–Meier method, using the log-rank test to determine the difference. P-values for each analysis are marked on figures, and the level of statistical significance was defined to P < 0.05 (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

To clarify the role of WTAP, we first analyzed the mRNA expression of WTAP in human HCC samples from gene expression omnibus (GEO) datasets (Roessler liver) and the Cancer Genome Atlas (TCGA) data. The results showed conspicuously higher WTAP expression in tumor tissues (Fig. 1a and Additional file 7: Figure S1a). The WTAP protein level was also remarkably up-regulated in HCC tissues (Fig. 1b), which was further confirmed by IHC staining of TMA cohort (Fig. 1c, d). Moreover, we investigated the relationship between WTAP expression and clinicopathological features in 90 HCC patients. WTAP expression was significantly interrelated with tumor encapsulation and recurrence (Table 1). And Kaplan-Meier analysis revealed that patients with higher WTAP expression level were associated with poorer overall survival (OS) and disease-free survival (DFS) (Fig. 1e). Besides, the overexpression of WTAP was found to be an independent prognostic factor for OS (p = 0.008) and DFS (p = 0.013) in HCC patients (Fig. 1f). Taken together, we concluded that WTAP is up-regulated in HCC and is closely related to its poor prognosis.

Expression of WTAP in HCC cell lines was examined before functional experiments (Additional file 7: Figure S1b, c). Knockdown and overexpression of WTAP were carried out in Huh7/Hep3B/PLC/PRF/5 and SMMC-7721/HCCLM3 cells, respectively. The CCK-8 and colony formation assays indicated that WTAP deficiency inhibited the proliferation ability in all three cell lines (Fig. 2a-c), and the results were reversed when WTAP was overexpressed (Additional file 7: Figure S1d, e). In addition, the EdU assay implied that cell growth was decreased with WTAP knockdown (Fig. 2d). Furthermore, cell migration and invasion were substantially impaired by silencing of WTAP (Additional file 7: Figure S1f, g).

To confirm the role of WTAP in vivo, tumor xenograft models were constructed by subcutaneously injecting HCC cells with either stable knockdown (shWTAP #1, #3) or overexpression of WTAP (WTAP-OE) into nude mice. We found that WTAP depletion repressed tumorigenesis (Fig. 2e) with prominently lower tumor volumes and weights compared with negative control group (Fig. 2f, g). And it also contributed to the down-regulated level of Ki67 in xenograft tumor tissues (Additional file 8: Figure S2a) Meanwhile, forced expression of WTAP caused an inverse phenotype in xenograft mice (Fig. 2h-j; Additional file 8: Figure S2b-d). In summary, we believed that WTAP performed a tumor promoting function in HCC.

To understand the underlying mechanism by which WTAP exerted tumor promoting effects in HCC, we employed transcriptome sequencing to illustrate the transcriptional alterations in WTAP knockdown cells. Hierarchical clustering indicated 1636 up-regulated genes and 794 down-regulated genes after the deletion of WTAP (Fig. 3a). In order to narrow down the scope of downstream targets, we integrated the GEO data (GSE46705↗) published by Liu et al. [40] who performed MeRIP -seq and CLIP-seq for elaborating the transcripts regulated by WTAP, to establish the gene set overlaps. A total of 15 genes were unveiled by the Venn diagram (Fig. 3b). Then we conducted RT-qPCR to evaluate the impact of WTAP on each candidate. Among which, ETS1 and ETS2 were consistently significantly up-regulated following WTAP silencing, and they were both moderately down-regulated when WTAP was overexpressed (Fig. 3c-e). However, western blotting analyses revealed that the inactivation of WTAP notably led to the elevation of ETS1 at the protein level, rather than ETS2, (Fig. 3f, g). We further demonstrated that the ETS1 content was conversely affected by WTAP in Hep3B and HCCLM3 cells (Fig. 3h, i), accompanied by the parallel results from immunofluorescence staining and cytosolic/nuclear separation analysis (Additional file 9: Figure S3a, b). Therefore, we postulated that dysfunction of ETS1 probably accounted for the WTAP-mediated HCC proliferation signature.

According to the MeRIP-seq data, modulation of ETS1 might occur in an m6A-dependent manner. To determine whether the m6A modification of ETS1 was mediated by WTAP, we first measured the global level of m6A in negative control group and stable WTAP knockdown group through two distinct methods (m6A dot blot and RNA methylation quantification assay). As expected, m6A levels were substantially decreased with the deletion of WTAP in two HCC cell lines (Fig. 4a, b). Furthermore, the MeRIP-qPCR assay was conducted to determine the enrichment of m6A in ETS1. Compared with IgG control, m6A-specific antibody robustly enriched ETS1 transcripts. Furthermore, we found a remarkably decreased amount of ETS1 modified by m6A following WTAP silencing (Fig. 4c). Therefore, we supposed that WTAP was able to influence the overall level of m6A, specifically for ETS1.

To substantiate the requirement of m6A modification for ETS1, luciferase reporter assays were conducted with a wild-type (WT) and two mutant (Mut) plasmids. For Mut reporters, several adenosine bases (A) in predicted m6A sites were replaced by cytosine bases (C) to eliminate the effect of m6A methylation, while WT reporter contained intact m6A sites (Fig. 4d). As expected, the luciferase activity of WT group moderately intensified under WTAP knockdown, while those of Mut groups definitely rendered resistance to the impact of WTAP silencing (Fig. 4e), implying that the regulation of ETS1 was under the control of WTAP-guided m6A modification.

Intriguingly, the direct protein interaction of WTAP and ETS1 seemed to be unwarranted (Additional file 9: Figure S3c). In addition, the IF staining failed to identify the colocalization of WTAP and ETS1, because WTAP was primarily located in the nucleus while ETS1 was mainly in the cytoplasm (Additional file 9: Figure S3a, b), which was in accordance with previous studies [9, 41]. Nevertheless, the WTAP antibody could lead to the enrichment of ETS1 mRNA (Additional file 9: Figure S3d), which was in agreement with the CLIP-seq data. Therefore, we further confirmed that the association of WTAP and ETS1 may not be at the protein level, but rather the RNA level.

To further elucidate the effect of m6A in regulation of ETS1, we introduced the methylation inhibitors, 3-deazaadenosine (DAA) and cycloleucine. Consistent with our hypothesis, treating HCC cells with DAA or cycloleucine led to a remarkable reduction of total m6A level (Additional file: Figure S3e), and markedly increased the expression of ETS1 meanwhile (Additional file: Figure S3f, g). 9 9

Collectively, these data implied that WTAP was sufficient to regulate the m6A modification of ETS1 mRNA.

It was indispensable to find appropriate readers of ETS1, since m6A-modified transcripts relied on the effector proteins to be functionally involved in biological processes. As the well-known m6A reader protein, YTHDF2 was reported to frequently participate in the regulation of mRNA degradation [18]. However, knockdown of YTHDF2 caused little impact on ETS1 expression (Additional file 10: Figure S4a), which precluded the potential role of YTHDF2 in ETS1 modulation. In addition, the RNA-binding protein (RBP) HuR was prone to binding with less m6A-modified RNA and stabilizing its bound counterparts [42], hence we supposed that HuR may regulate ETS1. Actually, the reduction of HuR prominently alleviated the expression of ETS1 (Fig. 4f). And we found that HuR-specific antibody dramatically enriched ETS1 mRNA compared to the IgG control, while repression of WTAP significantly intensified the enrichment of ETS1 mRNA as manifested by RIP-qPCR (Fig. 4g). Interestingly, HuR did not physically interact with WTAP according to Co-IP results (Additional file 9: Figure S3c), and HuR was rarely altered when WTAP was silenced (Additional file 10: Figure S4b, c). This signified that WTAP restrained ETS1 via interference of the conjunction between HuR protein and ETS1 mRNA, instead of altering the expression of HuR.

Moreover, the elevation of ETS1 induced by WTAP knockdown could be retrieved by attenuation of HuR (Fig. 4h and Additional file 10: Figure S4d). We found knockdown of WTAP would prolong the half-life of ETS1 RNA, while HuR silencing could reverse this effect (Fig. 4i). Additionally, we carried out luciferase reporter assays to determine the involvement of HuR in m6A modification. In WT groups, the enhancement of luciferase activity induced by WTAP silencing could be rescued by HuR. However, alteration of HuR expression seemed to be ineffective for luciferase activity when possible m6A sites of ETS1 were mutated (Fig. 4j).

Together, our findings suggested that WTAP repressed ETS1 through an m6A-HuR-dependent pathway.

Expression of ETS1 in tumor tissues was identified to be lower compared with paratumor tissues in HCC (Fig. 5a, b). Moreover, we discovered a markedly higher positive rate in adjacent tissues based on the IHC staining of cohort1 (Fig. 5c, d). And high level of ETS1 implied better prognosis according to the TCGA data (Fig. 5e). Functional validation in vitro including CCK-8 and colony formation demonstrated that knockdown of ETS1 enhanced the proliferation capability of MHCC97H, HCCLM3 and Huh7 cell (Fig. 5f, g and Additional file 11: Figure S5a). Moreover, ETS1 silencing was sufficient to rescue the inhibitory effect of WTAP knockdown on HCC cell growth and viability (Fig. 5h, i). Hence the inactivation of ETS1 may lead to the progression of HCC via the WTAP-ETS1 axis.

WTAP was reported to be involved cell cycle regulation, thus we intended to explore the related mechanism in HCC. In accordance with the results above, WTAP knockdown caused substantial G2/M arrest (Fig. 6a, b), while overexpression moderately contributed to the promotion of G2/M phase transition (Additional file 11: Figure S5b). To clarify the underlying mechanism, we examined several elements associated with the cell cycle. Intriguingly, the expression of p21 and p27, the tumor suppressors that played vital roles in cell cycle and proliferation, were strongly increased when WTAP was inactivated at transcriptional level (Fig. 6c, d and Additional file 11: Figure S5c). Meanwhile, western blotting analysis identified that WTAP knockdown upregulated p21 and p27, and simultaneously downregulated CDC25C, CDK1, cyclin-A2 and cyclin-B1 (Fig. 6e). In contrast, overexpression of WTAP alleviated the expression of p21 and p27 and reduced the magnification of all the indicated checkpoint proteins in the G2 phase (Fig. 6e).

To further investigate whether ETS1 participated in WTAP-induced effects on cell cycle, we conducted a series of rescue assays. Although ETS1 deficiency did not independently intervene in the cell cycle distribution (Additional file 11: Figure S5d, e), the G2/M arrest vcaused by WTAP silencing was significantly inversed by suppression of ETS1 (Fig. 6f, g; Additional file 12: Figure S6a, b). And WTAP deficiency followed by ETS1 overexpression led to a more remarkable G2/M arrest (Additional file 12: Figure S6c, d). Additionally, knockdown of ETS1 induced the decrease of p21 and p27 (Additional file 12: Figure S6e), and western blotting analysis revealed the same conclusion (Fig. 6h). We speculated that ETS1 may enhance the transcription of p21 and p27 since the ChIP assay indicated that ETS1 could bind to their promoter (Fig. 6i). Moreover, the repression of ETS1 reinstated the elevated expression of p21 and p27 caused by WTAP inhibition (Fig. 6j, k), while epitopic expression of ETS1 contributed to further up-regulation of p21 and p27 upon WTAP silencing (Additional file 12: Figure S6f). These data suggested ETS1-p21/p27 axis was essential for WTAP-dependent cell cycle regulation in HCC.

To evaluate the clinical correlation between WTAP and ETS1, we performed IHC assays of WTAP and ETS1 staining within the same HCC specimens from cohort1. Nearly 65.2% of samples where WTAP was more highly expressed presented weaker ETS1 staining, while approximately 55.6% of those with lower WTAP expression exhibited stronger ETS1 staining (Fig. 7a, b). We also confirmed that high WTAP expression or low ETS1 expression was independently associated with poor prognosis of HCC patients (Figs. 1e, f and 5e). Then the Kaplan-Meier analysis based on the combination of these two elements further demonstrated that HCC individuals with the expression of WTAP^highETS1^low had an even worse overall survival rate than any other groups (P = 0.0002), especially compared with those who were in the state of WTAP^lowETS1^high (Fig. 7c). To conclude, WTAP and ETS1 were inversely interrelated in clinical samples and the co-expression pattern of WTAP and ETS1 might be regarded as an efficient prognostic factor of HCC.

All data generated or analyzed during the current study are included in this published article (and its supplementary information files) or available on published datasets (TCGA or GEO).