P53 suppresses the progression of hepatocellular carcinoma via miR‐15a by decreasing OGT expression and EZH2 stabilization

Abstract Existing literature has highlighted the tumour suppressive capacity of microRNA‐15a (miR‐15a); however, its role in hepatocellular carcinoma (HCC) remains relatively unknown. This study aimed to investigate the role of miR‐15a in HCC and the associated underlying mechanism. Initially, RT‐qPCR was performed to detect the expression of miR‐15a in HCC tissues and cells. Bioinformatics analysis, Pearson correlation coefficient, dual‐luciferase reporter assay, and molecular approaches were all conducted to ascertain the interaction between miR‐15a and O‐linked N‐acetylglucosamine (GlcNAc) transferase (OGT). PUGNAc treatment and cycloheximide (CHX) assay were performed to evaluate O‐GlcNAc and the stabilization of the enhancer of zeste homolog 2 (EZH2). Finally, gain‐ and loss‐of‐function studies were employed to elucidate the role of P53 and the miR‐15a/OGT/EZH2 axis in the progression of HCC, followed by in vivo experiments based on tumour‐bearing nude mice. Our results demonstrated that the miR‐15a expression was decreased in the HCC tissues and cells. P53 upregulated miR‐15a expression, which inhibited the proliferation, migration and invasion of HCC cells, while inducing apoptosis and triggering a G0/G1 cell cycle phase arrest. OGT stabilized EZH2 via catalysing O‐GlcNAc, which reversed the effect of P53 and miR‐15a. The results of our in vivo study provided evidence demonstrating that P53 could suppress the development of HCC via the miR‐15a/OGT/EZH2 axis. P53 was found to inhibit the OGT expression by promoting the expression of miR‐15a, which destabilized EZH2 and suppressed the development of HCC. The key findings of our study highlight a promising novel therapeutic strategy for the treatment of HCC.

in a considerable public health burden. [3][4][5] In the Asia-Pacific region, approximately 75% of HCC cases reportedly stem from Asia, highlighting a public health issue. 6 The risk factors contributing to the occurrence of HCC include viral infections including hepatitis B or C viruses, alcohol assumption-related cirrhosis, non-alcoholic steatohepatitis (NASH), non-alcoholic fatty liver disease (NAFLD), and so forth. 7 Recently, the next-generation sequencing (NGS) opens the door to understand the genetic landscape like mutations in liver cancer. Li  HCC treatments are comprised of liver resection, transplantation, radiation, chemotherapy, targeted therapy and immunotherapy. 6 Over the past decade, although significant improvement regarding prevention, diagnosis and treatment of HCC has occurred with positive outcomes often seen in the early stages of the disease, reoccurrence still occurs in many patients, highlighting the need for an improved understanding of the mechanism underlying the initiation and progression of HCC. 10 MiRNAs (miRNAs) represent endogenous small non-coding RNAs that are approximately 22 nucleotides in length and bind to the 3′-untranslated region (3′-UTR) of target mRNAs resulting in mRNA degradation or inhibition of mRNA translation. 11 MiRNAs have been implicated in the progression of a wide variety of diseases including numerous malignancies, Alzheimer's disease and cardiovascular disease [12][13][14] while the dysregulation of miRNAs been linked with the tumorigenesis of HCC. 15, 16 Liu et al. 17 reported that microRNA-15a (miR-15a), a tumour suppressor, plays a key role in the proliferation, invasion, apoptosis and drug resistance of cancer cells. Recently, miR-15a has been reported to inhibit migration and invasion of HCC via repressing cMyb, which is implicated as a target for HCC treatment. 18 Interestingly, miR-15a has been reported to be one of the downstream tumour suppressors by TP53. 19 TP53 mutations often occur in hepatitis B virus (HBV)-related HCC. 9 The dysregulation of miR-15a-5p and miR-26a-5p promotes the proliferation of ccRCC in regulation of OGT. 20 OGT represents a GlcNAc transferase, which catalyses serine and threonine residues of intracellular proteins with O-GlcNAc modification. 21 In the current study, following a bioinformatics analysis as well as a literature review, we set out to elucidate the role of miR-15a and the underlying mechanism associated with the regulation of HCC development both in vitro and in vivo to provides a novel therapeutic strategy for the treatment of HCC, through gain-and loss-of-function analyses, cellular and molecular approaches, and in vivo experiments.

| Clinical samples
One hundred and fifty-three patients following radical hepatectomy at the hepatology department of The First Affiliated Hospital of Nanchang University from 06/2014 to 06/2016 were recruited for the purposes of the current study. All patients were yet to receive any anti-tumour therapy prior to their operations, including NASH-related HCC patients (n = 26), alcohol-related liver disease (ALD)-related HCC patients (n = 31), hepatitis C virus (HCV)-related HCC patients (n = 73) and HBVrelated HCC patients (n = 23). All HCC samples isolated from tumour centre were confirmed to be free of necrosis and bleeding while the corresponding adjacent normal tissue samples (no cancer cells confirmed by pathology) were stored at −80℃ for gene expression and histology analysis purposes. Informed consent documentation was signed by all participating patients, with the protocols involving humans performed in strict accordance with the 1964 Helsinki Declaration. Patients' information and follow-up were collected by medical record room. The outcome of patients after treatment was recorded and added to their clinical records, with the follow-up process initiating after operative procedures and continued for 36 months. Kaplan-Meier model was used to analyse the correlation between expression of miR-15a and patients' overall survival (OS) and disease-free survival (DFS). The cells were trypsinized and seeded into 6-well plates with a density of 1 × 10 5 cells/well upon reaching the logarithmic phase, followed by culturing. Upon reaching 75% confluency, cells were transiently transfected with Lipofectamine 2000 (Invitrogen) as per the manufacture's protocol. Transfection was performed in groups based on the following: miR-NC (miRNA control), miR-15a-mimic (miR-15a analog), oe-NC (overexpression control plasmid), oe-P53 (overexpression of P53 plasmid), si-NC (small interfering RNA control), si-P53 (P53 silencing plasmids, si1-P53, si2-P53 and si3-P53 (synthesized by Shanghai GenePharma Co., Ltd.), oe-NC + miR-NC (overexpression plasmid and miRNA control), miR-15a-mimic + oe-NC (miR-15a-mimic plasmid and overexpression control plasmid), oe-OGT + miR-15a-mimic (overexpression of OGT plasmid and miR-15a-mimic), oe-OGT (overexpression of OGT plasmid), sh-OGT (OGT silencing plasmids, sh1-OGT, sh2-OGT, and sh3-OGT (synthesized by Shanghai GenePharma Co., Ltd), si-EZH2 (OGT silencing plasmids, si1-EZH2, si2-EZH2 and si3-EZH2 (synthesized by Shanghai GenePharma Co., Ltd), oe-NC + miR-NC (overexpression control plasmid and miR-NC), oe-OGT + si-EZH2
Following a 48-h period of transfection, real-time quantitative polymerase chain reaction (RT-qPCR) and Western blot were performed to analyse the expression of the target genes as well as the corresponding proteins. Plasmids as well as the mimics were synthesized and purchased from Sino Biological Inc. The medium was renewed 6 h after transfection, after which the cells were cultured for an additional 48 h for subsequent experiment purposes.

| RNA isolation and RT-qPCR
Total RNA was extracted using a RNeasy Mini Kit (Qiagen). Reverse transcription (RT) of mRNA was performed using a RT kit (RR047A, Takara) for cDNA collection purposes. miRNA RT was performed using miRNA First Strand cDNA Synthesis (Tailing Reaction) kit (B532451-0020, Sangon Biotech Co., Ltd.). Samples for RT-qPCR were prepared with SYBR ® Premix Ex Taq™ II (Perfect Real Time) kit (DRR081, Takara) followed by RT-qPCR (ABI 7500, ABI). The twostep RT-qPCR program was performed using pre-denaturation at 95℃ for 30 s, then 95℃ for 5 s and 60℃ for 34 s with 40 cycles.
Each sample was prepared with triplicates. miRNA universe primer and U6 forward primer were provided in miRNA First Strand cDNA Synthesis (Tailing Reaction) kit; other primers were synthesized by Sangon Biotech Co., Ltd. listed in Table S1. Ct value was recorded in each well, glyceraldehyde 3-phosphate dehydrogenase (GAPDH) or U6 were employed as the internal control, with the 2 −ΔΔCt method applied to calculate the relative mRNA expression level.

| Co-immunoprecipitation (co-IP) assay
Co-IP assay was performed using the Pierce co-IP kit (Thermo scientific 26149), as per the manufacturer's instructions. The total lysate extracted from HCC cells expressing OGT-FLAG was IP with FLAG Ab or EZH2 Ab, after which Western blotting was performed using the designated antibodies (Abs). The total lysate of HCC cells treated with TMG was IP with IgG or O-GlcNAc Ab (RL2), followed by Western blotting with EZH2 Ab, and Ig heavy chain (H chain) was used as loading control. The antibodies used included the following: EZH2 (5246, CST); O-GlcNAc (ab2739, abcam); Flag (SAB4200071, Sigma-Aldrich); IgG (NI01-100UG, Millipore).

| Western blot
Total proteins were extracted from cultured cells that were used for sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblot. A bicinchoninic acid assay (BCA) kit (20201ES76, Yeasen Biotech) was performed to quantify the protein concentration. The proteins were subsequently separated by SDS-PAGE, and transferred onto a polyvinylidene fluoride (PVDF) membrane. The membrane was blocked with 5% bovine serum albumin (BSA) for 1 h at room temperature, followed by incubation with primary antibodies including p53 (ab131442, 1: 1000, Abcam), OGT (ab96718, 1: 2000, Abcam), EZH2 (ab186006, 1: 25000, Abcam), O-GlcNAc (ab2739, 1: 1000, Abcam) and GAPDH (ab8245, 1: 5000, Abcam) at 4℃ overnight with shaking. The next day, PVDF was washed three times using Tris-buffered saline tween-20 (TBST) (5 min per wash) and incubated with horseradish peroxide-conjugated secondary antibody, followed by three additional TBST washes (5 min per wash). The PVDF was developed using enhanced chemiluminescence (ECL). ImageJ 1.48u software (National Institute for Health) was employed to quantify the relative expression level of proteins by calculating the ratio of each protein grey value/grey value of GAPDH loading control. Each experiment was performed 3 times.

| 3-(4,5-Dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT) assay
The cells were seeded into a 96-well plate at a density of 1 × 10 4 cells/well and cultured for 0, 1, 2 and 3 days, respectively. MTT was added to each well, followed by an additional 4 h incubation period at 37℃. Next, SDS-HCl was added to each well and incubated for 30 min. After the medium had been removed, an EL 800 Universal Microplate Reader (BioTek) was used to determine the optical density (OD) at 540 nm.

| Cell colony forming unit
The cells were seeded in a 6-well plate with a density of 600 cells/ well and cultured for 2-3 weeks with occasional observation. When colonies were visible, the cell culture procedure was terminated.
The colonies were washed using phosphate-buffered saline (PBS) and fixed with 1 ml methanol for 15 min at room temperature. Next, 1 ml of crystal violet was added to each well followed by incubation for 30 min. The colonies were washed with deionized water and air-dried at room temperature. The number of colonies was subsequently determined using a scanning plate reader.

| Cell scratch assay
Following a 48-h period of transfection, a sterile pipette tip was used to scratch the centre of the confluent cells. The detached cells and debris were washed away using PBS. The migration of cells was recorded with a microscope 0 or 24 h after scratching, and ImageJ (National Institute for Health) software was used to calculate the migration rate (MR). Each experiment was repeated at least 3 times.

| Transwell assay
The upper layer of the Transwell plate was embedded with Matrigel (BD), followed by incubation at 37℃ for 30 min to solidify the Matrigel. The cells were subsequently cultured in a serum-free medium for 12 h, after which they were collected and resuspended using serum-free medium to a density of 1 × 10 5 cells/ml. Medium with 10% FBS was placed in the bottom layer of the Transwell plate.
Next, 100 μl cell suspension was added to the bottom layer and incubated at 37℃ for 24 h. The cells with no invasion were wiped away using a cotton swap, while the invasive cells were fixed with 100% methanol and stained with 1% toluidine blue (Sigma). The stained invasive cells were observed under an inverted microscope with 5 fields randomly selected to record the number of invasive cells. Each experiment was repeated 3 times.

| Cell apoptosis by flow cytometry
An Annexin V-FITC apoptosis kit (Invitrogen) was used to analyse cell apoptosis based on the manufacturer's instructions. Briefly, cell culture medium was renewed with serum-free medium after transfection. Cells were then collected, washed 3 times with PBS (pH 7.4) and resuspended in a staining buffer. Next, 5 μl Annexin-V-FITC and 5 μl propidium iodide (PI) were mixed with cells for 10 min followed by incubation at room temperature. The fluorescence-activated cell sorting (FACS) flow cytometry system (Scalibur-Becton Dickinson) was employed to analyse apoptosis. Annexin V-positive and PInegative stained cells were considered to be apoptotic cells.

| Cell cycle analysis by flow cytometry
The cells were collected from a 6-well plate and washed with PBS 3 times after 48-h transfection. The cells were subsequently resuspended with 0.5 ml PBS and fixed with 70% ethanol at 4℃ overnight.
The next day, cells were washed 3 times with PBS and incubated with 400 μl CCAA (PI, Engreen) solution and 100 μl RNase A (100 μg/ml) at 4℃ for 30 min in dark. Next, cells were resuspended and analysed using a BD flow cytometry machine. 20,000-30,000 cells were recorded, and the data were analysed using ModFit software.

| Dual-luciferase reporter assay
OGT wild-type (WT) and mutant (MUT) with putative binding sites of miR-15a were inserted into psiCHECK-2 plasmid to obtain luciferase reporters (Promega). miR-15a-mimic or NC was co-transfected into the cells according to the manufacture's protocols provided by Lipofectamine 3000 (Invitrogen). After 48-h transfection, luciferase activity was determined using a dual-luciferase reporter system (Promega), from which the values of the experiment were normalized to the luciferase activity of Renilla.

| Tumour xenograft experiment with nude mice
HCC-LM3 and Huh-7 cells that were transfected with oe-NC, oe-P53 and oe-P53 + oe-EZH2 plasmids were for inoculation purposes.
The cells were resuspended in serum-free DMEM (Gibco) with a density of 1 × 10 6 cells/200 μl. Thirty-six Balb/c mice were randomly separated into 6 group (6 mice of each) and housed under the same conditions. The mice were anaesthetized with ether, followed by transplantation with 200 μl cell suspension/mice into the soft region of the back of the right hind leg subcutaneously. Both the mice as well as tumour inoculation findings were recorded every day followed by euthanasia 4 weeks after inoculation. The tumour was isolated, photographed and weighed. The tumour volume was measured every 4 days and calculated using the following formula: Volume = (length × width 2 )/2. RT-qPCR was performed to measure the expression of miR-15, and Western blot was used to analyse the expression of P53, OGT and EZH2 proteins. All animal procedures were performed with the approval of the Animal Care and Use Committee of The First Affiliated Hospital of Nanchang University.

| Ki67 test
The tumours were fixed in 10% formaldehyde, embedded in paraf-

| Bioinformatics analysis
The GEO database was used to analyse HCC-related expression data, miRNA data set GSE41077 including 6 cancerous samples and 2 normal samples, mRNA data sets GSE46408 and GSE14520. GSE46408 includes 6 cancerous samples and 6 normal samples, while GSE14520 includes 22 cancerous sample and corresponding normal samples. R package 'limma' was used to do differential analysis. For GSE41077, the differential expressed miRNAs were screened based on |logFC| > 1 and p < 0.05. For GSE46408, the differentially expressed mRNAs were screened based on logFC| > 0.8 and p < 0.05.

| Statistical analysis
All experimental data were analysed using SPSS 21.0 (IBM Corp.).
Quantitative data were expressed as the mean ± standard deviation (mean ± SD). Data from both tumour and adjacent tissues were analysed by paired t test. Data between two groups were compared using an unpaired t test, while data among multiple groups were analysed via one-way analysis of variance (anova) and Tukey's post hoc test. Data at different time points were evaluated using repeated measures anova and Bonferroni post hoc test. Kaplan-Meier method was applied to calculate the survival rate, while a log-rank test was performed for single-factor analysis purposes. Pearson correlation coefficient was conducted to determine the correlation between samples. Statistically significant difference was indicated by p < 0.05.

| miR-15a expression is downregulated in HCC tissues and cells and correlated with patients' survival
The role of miR-15a has been highlighted in a wide variety of biological processes including cell proliferation, apoptosis, invasion and drug resistance in tumours through binding to the 3′-UTR of mRNA of target genes. 17 miR-15 has been reported to suppress the invasion of liver cancer cells, which makes it a potential therapeutic target of liver cancer. 18 The miRNA expression data set, GSE41077 was evaluated using |logFC| > 1 and p < 0.05 as the miRNA screening criteria R package 'limma' was applied to analyse the differentially expressed genes, which revealed 31 upregulated miRNAs ( Figure 1A), including miR-15a which was consistent with the findings of previous literature existing. However, the role of miR-15a in the development of HCC and the molecular mechanism remain unclear. Thus, we aimed to elucidate the potential mechanism by which miR-15a contributes to the progression of HCC.
Initially, RT-qPCR was performed to detect the expression of miR-15a in 153 cases of HCC tissues and corresponding adjacent healthy tissues, which revealed that the expression of miR-15a was significantly diminished in the cancer tissues relative to that of the healthy tissues ( Figure 1B). RT-qPCR demonstrated that the expression of miR-15a was significantly decreased in NASH-related HCC patients, ALD-related HCC patients, HCV-related HCC patients and HBV-related HCC patients, suggesting that HCC of different aetiologies did not affect the expression of miR-15a ( Figure 1C). We subsequently detected the expression of miR-15a in the HCC cell lines, HCC-LM3 and Huh-7 and normal liver cell line, THLE-2, which indicated that the results were consistent with the findings of the tissue experiment ( Figure 1D). Kaplan-Meier methodology was used to analyse the correlation between miR-15a expression and patients'

DFS and OS. The results revealed that the OS and DFS in patients
with high expression of miR-15a were notably longer compared with patients with low expression of miR-15a ( Figure 1E,F), indicating that low levels of miR-15a expression is correlated with poor prognosis in HCC.

| miR-15a overexpression inhibits the proliferation, migration and invasion of live cancer cells
To further ascertain the effect of miR-15a on HCC cells, we trans-

| P53 regulates the expression of miR-15a to suppress the proliferation, migration and invasion of live cancer cells
Next, to evaluate the upstream and downstream regulation of miR-15a, we conducted a literature review, which revealed that the expression of miR-15a was regulated by P53 in colon and breast cancers. 22,23 Based on these findings, we set out to verify whether the expression of miR-15a is regulated by P53 and determine whether it influences the proliferation, migration and invasion of live cancer cells, we overexpressed or silenced P53 in HCC-LM3 and Huh-7 cell lines. First, HCC-LM3 and Huh-7 cells were transfected with 3 si-P53 to silence the expression ofP53.
RT-qPCR and Western blot were performed to detect the mRNA and protein expression of P53, which demonstrated that compared with si-NC, the mRNA and protein levels of P53 by si-P53 were both downregulated (p < 0.05), whereby si1-P53 exhibited the best silencing efficiency and was selected for subsequent experiments ( Figure 3A,B).
The HCC-LM3 and Huh-7 cells were subsequently transfected with oe-P53 and si-P53. Western blot was performed to detect the expression of P53 protein, after which RT-qPCR methods were used to detect the expression of miR-15a. The results indicated that the expression of P53 protein and the mRNA expression of miR-15a were both markedly elevated in response to oe-P53 relative to oe-NC ( Figure 3C). Next, we analyse the cell viability via MTT assay and colony forming unit assays ( Figure S2A), which demonstrated that cell proliferation was notably decreased following oe-P53 treatment relative to that of oe-NC treatment; however, when compared with si-NC, cell proliferation was significantly increased in response to si-P53 treatment ( Figure 3D,E).
Cell scratch and Transwell assays demonstrated that oe-P53 treatment revealed notably decreased the cell migration and invasion in relative to oe-NC, while compared with si-NC group, cell migration and invasion was significantly increased in response to si-P53  Figure 3H and Figure S2D). The cell cycle data obtained suggested that oe-P53 triggered a G0/G1 cell cycle arrest, while in compassion to si-NC, si-P53 treatment led to a significant increase in the S phase ( Figure 3I and Figure S2E).
In summary, our data indicate that miR-15a is regulated by P53 in HCC cell, while the overexpression of P53 was found to elevate the expression of miR-15a leading to the inhibition of proliferation, migration and invasion of HCC cells and increased cell apoptosis and G0/G1 phase.

| miR-15a directly targets OGT
To further investigate the anti-tumour mechanism of miR-15a in HCC, we analysed the mRNA expression data set, GSE46408 from the GEO database in addition to |logFC| > 0.8 and p < 0.05 employed as the differential miRNA screening criteria. R package 'limma' was applied to analyse the differentially expressed genes, which revealed 1754 upregulated miRNA and 1842 downregulated mRNA. Next, the starbase, miRDB and mirDIP databases were explored to predict the downstream targets of miR-15a, which showed top 70 genes that were intersected with 12,000 upregulated genes from differential mRNA data ( Figure 4A-C).
The data obtained indicated that CCNE1, MYB and OGT mRNAs existed in these 4 databases. Previous reports have suggested that CCNE1 and MYB are direct targets of miR-15a 24,25 ; however, the relationship between OGT and miR-15a remains exclusive.
Our results indicated that OGT was not only highly expression based on GSE46408 data ( Figure 4D) but could also bind with miR-15a ( Figure 4G). Evidence was obtained indicating that the expression of OGT in HCC sample was significantly higher than    Figure S3B). OGT upregulation also occurred in HCC samples compared with normal samples according to GSE14520 data set ( Figure S3C). In order to verify the aforementioned result, RT-qPCR was initially performed with 153 patient samples, with the results indicating that OGT was highly expressed in the HCC tissues relative to that of the normal live tissues, which was consistent with data set ( Figure 4E). Pearson correlation coefficient highlighted a negative correlation between OGT and miR-15a ( Figure 4F). Next, dual-luciferase reporter assay was performed to assess the relationship between OGT and miR-15a, the result of which suggested that miR-15a-mimic significantly decreased the luciferase activity of OGT-WT but not OGT-MUT ( Figure 4H).
Finally, Western blot and RT-qPCR results revealed that miR-15amimic significantly downregulated the expression of the OGT protein ( Figure 4I). Altogether, these results suggest that OGT is a direct target of miR-15a. were also significantly decreased in OGT silenced HCC-LM3 and Huh-7 cells ( Figure 5D). Next, to verify whether EZH2 was modified by glycosylation and whether glycosylation influences EZH2 in HCC cell lines, co-IP assay was performed and the results indicated that OGT could interact with EZH2 in HCC-LM3 and Huh-7 cells ( Figure 5E).

| OGT-mediated O-GlcNAc stabilizes EZH2 and increases its expression
Moreover, knockdown of OGT reduced the O-GlcNAc glycosylation level of EZH2 and increased the ubiquitination level ( Figure 5F).
Meanwhile, CHX analysis demonstrated that silencing OGT could

| OGT promotes proliferation, migration and invasion of HCC cells by regulating EZH2
Next, to evaluate whether OGT regulates the expression of EZH2 to influence live cancer progression, OGT was overexpressed or silenced EZH2 in HCC-LM3 and Huh-7 cells as follows, oe-NC + si-NC, oe-OGT + si-NC, and oe-OGT + si-EZH2. First, we transfected 3 si-EZH2 to silence EZH2 expression. RT-qPCR and Western blot were performed to detect the mRNA and protein expression of EZH2, which suggested that the mRNA and protein expression of EZH2 was markedly downregulated following the silencing of EZH2 silencing compared with si-NC, in which si1-EZH2 resulted in the lowest level of EZH2, which was selected for the subsequent experiments ( Figure 6A,B). Western blot analysis also demonstrated that the expression of OGT was significantly increased following OGT overexpression, while the expression of EZH2 was considerably decreased after treatment with si-EZH2 ( Figure 6C).
We subsequently applied MTT and colony forming unit assays to study the cell proliferation, cell scratch assay was employed to detect the cell migration, Transwell assay was conducted to de-

| P53 inhibits the proliferation, migration and invasion and increases the apoptosis of HCC cells through miR-15a/OGT/EZH2 axis
Based on the aforementioned findings, we asserted that P53 regu- Quantitative data were presented as mean ± SD. Data from tumour and the adjacent tissues were analysed by paired t test. Data of two groups were processed using unpaired t test, and data among multiple groups were analysed via one-way anova and Tukey's post hoc test. Pearson correlation coefficient was carried out to analyse the correlation between samples. *p < 0.05 compared with adjacent, or miR-NC, or oe-NC + mimic-NC, and # p < 0.05 compared with oe-NC + miR-15a-mimic. Experiments were repeated 3 times ( Figure 7A). The RT-qPCR data revealed that relative to oe-NC, miR-15a expression was considerably increased following treatment with oe-P53 or oe-EZH2 alone or in combination ( Figure 7B).
MTT and colony forming unit assays revealed that treatment with oe-P53 alone led to a notable reduction in the proliferation of HCC cells, while the cell proliferation was recovered by further treatment of oe-EZH2 ( Figure 7C,D). Cell scratch and Transwell assays demonstrated that the migration and invasion of HCC cells were significantly decreased in response to treatment with oe-P53 alone ( Figure 7E,F).
Cell apoptosis and cell cycle analyses uncovered that compared with oe-NC, cell apoptosis was increased ( Figure 7G) while the cell cycle was arrested at the G0/G1 phase in response to oe-P53 alone ( Figure 7H).
However, oe-P53 in combination with oe-EZH2 brought about a decrease in cell apoptosis ( Figure 7G) while increasing the S phase of the cell cycle ( Figure 7H). These results indicate that P53 suppresses OGT expression by promoting the expression of miR-15a, inhibiting the stability of EZH2 to inhibit proliferation, migration and invasion while elevating apoptosis and G0/G1 phase arrest in HCC.  Figure 8A,B), and miR-15a expression was also considerably elevated ( Figure 8C), while the expression levels of OGT and EZH2 were markedly decreased ( Figure 8D). Further treatment with oe-EZH2 increased the tumour volume in mice ( Figure 8A,B) and led to a notable upregulation in the expression of EZH2 ( Figure 8D). Ki67 immunohistochemistry results revealed that in comparison to oe-NC, cell proliferation was significantly decreased in response to oe-P53, while the cell proliferation was considerably increased by the further treatment of oe-EZH2

|
( Figure 8E). TUNEL assay revealed that compared with oe-NC, cell apoptosis was significantly increased in response to oe-P53, while apoptosis was notably decreased following further treatment with oe-EZH2 ( Figure 8F). Taken together, these results suggest that P53 promotes the expression of miR-15a to inhibit OGT expression, which inhibits EZH2 stability to inhibit liver tumour growth in nude mice. continue to emphasize the involvement of miRNAs in the repression of cancer development. [30][31][32][33] In liver cancer, miR-15a has been reported to inhibit the cancer cell proliferation and metastasis, which supports our results showing miR-15a as a tumour suppressor in liver cancer. 18  Finally, to expand our research in vivo, liver tumour-bearing nude mice models were established. The animal experiments provided evidence verifying that P53 could inhibit liver tumour growth in nude mice by regulating the miR-15a/OGT/EZH2 axis.
In conclusion, the key findings of our study elucidate a novel molecular mechanism whereby miR-15a functions as an anti-tumour miRNA in HCC, highlighting a promising therapeutic strategy for the treatment of HCC (Figure 9). Future study is required to demonstrate the mechanism by which P53 regulates the expression of miR-15a and to ascertain whether miR-15a exerts anti-tumour effects on other types of cancers.

CO N FLI C T S O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Research data are not shared.

O RCI D
Xiaodong Peng https://orcid.org/0000-0002-8441-0490 F I G U R E 9 P53 inhibits OGT expression by promoting the expression of miR-15a, which destabilizes EZH2 and inhibits the proliferation, migration and invasion and increases the apoptosis of HCC cells