Article Text

Download PDFPDF

Original article
METTL3-mediated m6A modification of HDGF mRNA promotes gastric cancer progression and has prognostic significance
  1. Qiang Wang1,
  2. Chen Chen2,
  3. Qingqing Ding3,
  4. Yan Zhao4,
  5. Zhangding Wang4,
  6. Junjie Chen2,
  7. Zerun Jiang2,
  8. Yan Zhang1,
  9. Guifang Xu4,
  10. Jingjing Zhang5,
  11. Jianwei Zhou2,
  12. Beicheng Sun1,
  13. Xiaoping Zou4,
  14. Shouyu Wang1,2,6,7
  1. 1Department of Hepatobiliary Surgery, The Affiliated Drum Tower Hospital of Nanjing University Medical School, Nanjing, Jiangsu Province, People’s Republic of China
  2. 2Department of Molecular Cell Biology and Toxicology, Jiangsu Key Lab of Cancer Biomarkers, Prevention & Treatment, Cancer Center; School of Public Health, Nanjing Medical University, Nanjing, Jiangsu Province, People’s Republic of China
  3. 3Department of Geriatric Oncology, The First Affiliated Hospital of Nanjing Medical University, Nanjing, Jiangsu Province, People’s Republic of China
  4. 4Department of Gastroenterology, The Affiliated Drum Tower Hospital of Nanjing University Medical School, Nanjing, Jiangsu Province, People’s Republic of China
  5. 5Affiliated Hospital of Guangdong Medical University, Zhanjiang, Guangdong Province, People’s Republic of China
  6. 6Center for Public Health Research, Medical School of Nanjing University, Nanjing, Jiangsu Province, People’s Republic of China
  7. 7Jiangsu Key Laboratory of Molecular Medicine, Medical School of Nanjing University, Nanjing, Jiangsu Province, People’s Republic of China
  1. Correspondence to Professor Shouyu Wang, Department of Hepatobiliary Surgery, The Affiliated Drum Tower Hospital of Nanjing University Medical School, Nanjing, Jiangsu Province, People’s Republic of China; sywang{at}nju.edu.cn; Professor Beicheng Sun, Department of Hepatobiliary Surgery, The Affiliated Drum Tower Hospital of Nanjing University Medical School, Nanjing, Jiangsu Province, People’s Republic of China; sunbc{at}njmu.edu.cn; Professor Xiaoping Zou, Department of Gastroenterology, The Affiliated Drum Tower Hospital of Nanjing University Medical School, Nanjing, Jiangsu Province, People’s Republic of China; 13770771661{at}163.com

Abstract

Objective N6-methyladenosine (m6A) RNA methylation and its associated methyltransferase METTL3 are involved in tumour initiation and progression via the regulation of RNA function. This study explored the biological function and clinical significance of METTL3 in gastric cancer (GC).

Design The prognostic value of METTL3 expression was evaluated using tissue microarray and immunohistochemical staining analyses in a human GC cohort. The biological role and mechanism of METTL3 in GC tumour growth and liver metastasis were determined in vitro and in vivo.

Results The level of m6A RNA was significantly increased in GC, and METTL3 was the main regulator involved in the abundant m6A RNA modification. METTL3 expression was significantly elevated in GC tissues and associated with poor prognosis. Multivariate Cox regression analysis revealed that METTL3 expression was an independent prognostic factor and effective predictor in human patients with GC. Moreover, METTL3 overexpression promoted GC proliferation and liver metastasis in vitro and in vivo. Mechanistically, P300-mediated H3K27 acetylation activation in the promoter of METTL3 induced METTL3 transcription, which stimulated m6A modification of HDGF mRNA, and the m6A reader IGF2BP3 then directly recognised and bound to the m6A site on HDGF mRNA and enhanced HDGF mRNA stability. Secreted HDGF promoted tumour angiogenesis, while nuclear HDGF activated GLUT4 and ENO2 expression, followed by an increase in glycolysis in GC cells, which was correlated with subsequent tumour growth and liver metastasis.

Conclusions Elevated METTL3 expression promotes tumour angiogenesis and glycolysis in GC, indicating that METTL3 expression is a potential prognostic biomarker and therapeutic target for human GC.

  • METTL3
  • m6A
  • angiogenesis
  • glycolysis
  • gastric cancer
View Full Text

Statistics from Altmetric.com

Significance of this study

What is already known on this subject?

  • Gastric cancer (GC) is the fifth most common neoplasm and the third most deadly cancer worldwide.

  • Glycolysis and angiogenesis have been considered distinct hallmarks of cancer.

  • The m6A modification in RNA and m6A methyltransferase METTL3 have been reported to be associated with tumour progression.

What are the new findings?

  • The level of m6A RNA is significantly increased in GC due to the elevated expression of METTL3.

  • The expression of METTL3 is higher in GC due to P300-mediated activation of H3K27 acetylation, which is associated with poor GC prognosis.

  • METTL3 promotes the m6A modification of HDGF mRNA, and the m6A reader IGF2BP3 directly recognises and binds to the m6A site on HDGF mRNA and enhances HDGF mRNA stability.

  • Secreted HDGF promotes tumour angiogenesis, while nuclear HDGF activates GLUT4 and ENO2 expression, followed by an increase in glycolysis in GC cells, which subsequently causes tumour growth and liver metastasis.

How might it impact on clinical practice in the foreseeable future?

  • As METTL3 functions as an oncogenic factor promoting GC development and progression with poor prognosis, it may be a promising prognostic biomarker and therapeutic target for human GC.

Introduction

Gastric cancer (GC) is the fifth most common neoplasm and the third most deadly cancer worldwide, and almost half of all global GC cases are diagnosed in East Asia.1 2 Most patients with GC are first diagnosed at an advanced stage with malignant proliferation and metastasis.3 Unfortunately, the prognosis of patients with advanced-stage GC remains poor.4 5 Therefore, identifying novel biomarkers and therapeutic targets for GC diagnosis and treatment is an urgent need.

Reprogramming of energy metabolism has been considered a distinct hallmark of cancer.6 Cancer cells undergo metabolic reprogramming to support malignant tumour initiation and progression.7 Aerobic glycolysis, characterised by an increase in glucose uptake and preferential lactate production even in the presence of oxygen, is the primary feature of metabolic reprogramming in cancers.8 Aerobic glycolysis could support the energetic and biosynthetic requirements of sustained proliferation and metastasis.9 Thus, targeting glucose transporters and glycolytic enzymes remains an attractive therapeutic intervention, and several potential inhibitors for blocking glycolysis have been applied in preclinical investigations.9 Emerging evidence also demonstrates that novel tumour drivers or suppressors modulate tumorigenesis and progression via the regulation of glycolysis in different types of cancers.10–12 In addition, glycolysis plays a critical role in the regulation of the tumour microenvironment, affecting aspects such as inflammatory factor secretion, immune evasion and tumour angiogenesis.13 14 Our previous study reported that angiogenesis is a key process involved in the proliferation and metastasis of GC.15 Thus, understanding the molecular mechanisms of glycolysis and angiogenesis in GC is essential for developing future diagnostic and therapeutic strategies.

N6-methyladenosine (m6A) methylation has been identified as the most abundant modification ubiquitously occurring in eukaryotic mRNAs.16 The effects of m6A modification include the regulation of mRNA stability, splicing, transport, localisation and translation; and RNA-protein interactions.17 The m6A modification is dynamic and reversible in mammalian cells, and it can be installed by m6A methyltransferases (also called writers: METTL3, METTL14 and WTAP) and removed by m6A demethylases (also called erasers: FTO and ALKBH5). Additionally, specific RNA-binding proteins (also called readers: YTHDF1/2/3, eIF3, IGF2BP1/2/3, HNRNPA2B1 and so on) can bind to the m6A motif directly or indirectly to affect RNA function.18 In the past few years, the biological functions of m6A modulators have been reported to be associated with stem cell differentiation, tissue development, circadian periods and tumour progression.18 However, the biological significance of m6A and the underlying regulatory mechanisms in human GC remain elusive.

In the present study, we revealed the biological role of m6A modification mediated by METTL3 in GC and proposed that METTL3 may be a novel predictive biomarker and therapeutic target for GC progression.

Materials and methods

The details are described in supplementary materials and methods.

Results

Elevated METTL3 expression correlates with poor prognosis of patients with GC

To elucidate the functional roles of m6A modification in GC, we first examined the m6A RNA levels in 28 GC tissues and paired normal gastric mucosa. We found that the m6A RNA levels were significantly higher in GC tissues via a dot blot assay (figure 1A). The results were also confirmed by colorimetric ELISA via the m6A RNA methylation quantification kit (figure 1B). The m6A modification is mainly catalysed by m6A writers and erasers, so we hypothesised that the increased m6A RNA levels in GC were caused by the dysregulation of writers and erasers. We then compared the mRNA levels of key m6A writers (METTL3, METTL14 and WTAP) as well as erasers (ALKBH5 and FTO) in 28 pairs of GC and paired normal gastric tissues. The core methyltransferase METTL3 was significantly upregulated in GC, while the expression of the others was not significantly different (figure 1C; online supplementary figure S1A-D). The m6A RNA level was also significantly correlated with METTL3 expression in the 28 GC tissues (R2=0.4647, p<0.0001, online supplementary figure S1E). The Cancer Genome Atlas (TCGA) data also confirmed that METTL3 mRNA was significantly upregulated in GC (figure 1D). Using the online bioinformatics tool Kaplan-Meier Plotter (http://kmplot.com/analysis/), we also found that patients with GC with increased METTL3 mRNA levels had worse overall survival (OS) (figure 1E).

Figure 1

Elevated METTL3 expression correlates with poor prognosis of patients with GC. (A) The mRNAs isolated from GC tissues and paired normal gastric mucosa were used in dot blot analyses with an anti-m6A antibody, and MB (methylene blue) staining served as the loading control (representative images in left panel). The relative m6A contents on mRNA in GC tissues and paired normal gastric mucosal tissues were calculated (right panel, n=28). (B) The m6A RNA levels in the same 28 GC tissues and paired normal gastric mucosa were detected by colorimetric ELISA-like assay via the m6A RNA methylation quantification kit. (C) The levels of METTL3 expression in GC and paired normal gastric mucosal tissues were measured by qRT-PCR (n=28). (D) The levels of METTL3 expression were analysed in GC (n=413) and normal gastric mucosal tissues (n=32) using TCGA data. (E) Kaplan-Meier survival curves of OS based on METTL3 expression using the online bioinformatics tool Kaplan-Meier Plotter. (F) METTL3 protein levels were measured in GC tissues and paired normal gastric mucosal tissues by western blotting (n=14). (G) Representative IHC images on the tissue microarray (TMA) probed with the anti-METTL3 antibody (scale bars=200 µm or 50 µm, respectively) are shown. (H) The distribution of the difference in METTL3 immunoreactivity score (IRS) (△IRS=IRST−IRSN). The IRS of METTL3 staining was available in 83 pairs of tissues. (I) Kaplan-Meier OS analysis of METTL3 expression in patients with GC (n=83, p<0.001, log-rank test). (J) Multivariable analyses were performed in the GC cohort. All bars correspond to 95% CIs. (K) The time-dependent receiver operating characteristic (ROC) analysis for the clinical risk score (TNM stage), the METTL3 risk score, and the combined METTL3 and clinical risk scores in the GC cohort. AUC, area under the curve; CI, confidence intervals; GC, gastric cancer; HR, hazard ratio; IHC, immunohistochemistry; OS, overall survival; TCGA, The Cancer Genome Atlas; TNM, tumour, node, metastases.

Consistent with these findings, the protein levels of METTL3 were found to be significantly higher in 12/14 (86%) human GC tissues than in their paired normal gastric tissues by western blot (figure 1F). In addition, the protein level of METTL3 was significantly increased in GC cells compared with that in human normal gastric mucosal tissues (online supplementary figure S1F). These results were confirmed in the tissue microarray of patients with GC, which showed that the expression of METTL3 was significantly increased in GC tissues compared with that in matched normal gastric tissues by immunohistochemistry (IHC) staining (n=83, p<0.001; figure 1G,H). The protein expression of METTL3 in the GC cohort was significantly correlated with clinicopathological features, such as lymph node metastasis (p=0.046) and tumour, node, metastases (TNM) stage (p=0.005); however, the METTL3 expression has no statistical significance between patients with diffuse-type and intestinal-type GC (p=0.171; online supplementary table S1). In addition, the data showed that patients with GC with increased METTL3 expression had poorer overall 5-year survival (n=83, p<0.001, log-rank test; figure 1I). Simultaneously, univariate Cox regression analysis revealed that lymph node metastasis, TNM stage and METTL3 expression were substantially associated with 5-year survival in patients with GC (online supplemental figure S1G), and multivariate Cox regression analysis showed that METTL3 was an independent predictive marker for the prognosis of patients with GC (HR=4.495, 95% CI (2.354 to 8.584); figure 1J). To further evaluate the predictive ability of METTL3 expression, we conducted a time-dependent receiver operating characteristics curve analysis, which indicated that the combination of clinical risk score (TNM stage) and METTL3 risk score contributed much more than either one alone in the GC cohort (figure 1K). For example, the area under the curve at year 3 was 0.711 for the clinical risk score, whereas it was significantly increased to 0.796 for combination of the clinical risk score and METTL3 risk score. Taken together, the above results reveal that the m6A modification and METTL3 levels are increased in GC and that METTL3 might be an independent prognostic factor for patients with GC.

P300-mediated H3K27ac activates METTL3 transcription in GC

To explore the mechanism of high METTL3 expression in GC, we first analysed the modification in the promoter of METTL3 by the UCSC genome bioinformatics site (http://genome.ucsc.edu/). As shown in figure 2A, abundant H3K27 acetylation (H3K27ac) signals were found in the promoter region of METTL3, suggesting that METTL3 might be regulated by chromatin acetylation. H3K27ac is known to be catalysed by the P300/CBP complex.19 By analysis of TCGA data, we found that P300 mRNA expression was significantly upregulated in GC (online supplementary figure S2A), while CREB binding protein (CBP) expression was not different between GC and normal gastric tissues (data not shown). We then treated the HGC-27 cells with C646, a histone acetyltransferase inhibitor targeting P300, and the results showed that METTL3 expression was significantly decreased in a time-depndent and dose-dependent manner, which did not show obvious cytotoxicity in HGC-27 cells (figure 2B; online supplementary figure S2B, C). This result was also confirmed in the other two GC cell lines (figure 2C,D; online supplementary figure S2D, E). Using a western blot assay, we found that the H3K27ac and METTL3 protein levels were decreased after C646 treatment (figure 2E). Next, we directly knocked down P300 using two specific siRNAs (figure 2F; online supplementary figure S2F), and found knockdown of P300 significantly reduced the mRNA and protein levels of METTL3 (figure 2G,H). Moreover, the chromatin immunoprecipitation (ChIP) assay results indicated that the promoter region of METTL3 was enriched in P300 binding and H3K27ac signals, and knockdown of P300 could significantly decrease the enrichment of H3K27ac signals in the promoter of METTL3 (figure 2I,J). These data confirmed that P300-mediated H3K27ac activation in the promoter of METTL3 may partly account for the upregulation of METTL3 (figure 2K).

Figure 2

P300-mediated H3K27ac activates METTL3 transcription in GC. (A) Data from the UCSC genome bioinformatics site (http://genome.ucsc.edu/) showed high enrichment of H3K27ac in the promoter of METTL3. (B–D) The mRNA levels of METTL3 in C646 (20 µM)-treated HGC-27 (B), NCI-N87 (C) and SGC7901 (D) cells at the indicated time points were measured by qRT-PCR. (E) The METTL3 and H3K27ac protein levels in HGC-27 and NCI-N87 cells were measured by western blotting after C646 treatment (20 µM) for 24 hours. (F) The P300 knockdown efficiency was verified at the protein levels in NCI-N87 cells by western blotting. (G) qRT-PCR analysis of the expression of METTL3 in NCI-N87 cells with P300 knockdown. (H) The METTL3 and H3K27ac protein levels in NCI-N87 cells with P300 deficiency were determined by western blotting. (I, J) ChIP assays were used to determine the level of P300 binding (I) and the enrichment of H3K27ac (J) at the promoter of METTL3 in P300 deficiency or control HGC-27 cells. (K) Graphical illustration of the mechanism by which P300-mediated H3K27ac activation induces METTL3 transcription. The data are the means±SEMs of three independent experiments. * p<0.05; ** p<0.01; *** p<0.001. H3K27ac, H3K27 acetylation; ChIP, chromatin immunoprecipitation; DMSO, dimethyl sulfoxide; GC, gastric cancer.

METTL3 promotes GC proliferation and angiogenesis

To examine the function of METTL3 in GC, we established stable METTL3-overexpressing GC cells (BGC823 and AGS) (figure 3A; online supplementary figure S3A). The upregulation of METTL3 significantly promoted GC cell soft agar colony formation efficiency (figure 3B) and clonogenic ability (figure 3C). We next examined whether METTL3 promoted GC cell proliferation dependent on its m6A catalytic activity, the plasmids expressing wild type METTL3 or its catalytic mutant (aa395-398, DPPW→APPA) were constructed as in a previous study,20 and the m6A level was dramatically decreased in BGC823 cells with catalytic mutant METTL3 compared with wild type cells (figure 3D). Moreover, it revealed that the m6A catalytic activity of METTL3 is indispensable for its role in promoting clonogenic ability of BGC823 cells (figure 3E).

Figure 3

METTL3 promotes GC proliferation and angiogenesis in vitro and in vivo. (A) The protein levels of METTL3 in BGC823 and AGS cells with METTL3 overexpression were measured by western blotting. (B) Upregulation of METTL3 enhanced GC cell (BGC823 and AGS) anchorage-independent growth in soft agar (scale bars=200 µm, left panel). Quantification of the soft agar colony formation assay results (right panel). (C) Overexpression of METTL3 increased the colony-formation ability of AGS cells (left panel). Quantification of the colony formation assay results (right panel). (D) The protein levels of METTL3 in BGC823 cells with wild type or catalytic mutant METTL3 overexpression were measured by western blotting (left panel). The mRNAs isolated from wild type or catalytic mutant METTL3-overexpressing GC cells were used in dot blot analyses with m6A antibody (right panel). MB (methylene blue) staining served as a loading control. (E) The clonogenic ability in BGC823 cells with wild type or catalytic mutant METTL3 overexpression was determined (left panel). Quantification of the colony formation assay results (right panel). (F) The protein levels of METTL3 in HGC-27 cells with METTL3 knockout were measured by western blotting (upper panel). The mRNAs isolated from METTL3 knockout or wild type GC cells were used in dot blot analyses with m6A antibody (bottom panel). MB staining served as a the loading control. (G) The anchorage-independent growth of METTL3 knockout HGC-27 cells or control cells in soft agar was determined (scale bars=200 µm, left panel). Quantification of the soft agar colony formation assay results (right panel). (H) Knockout of METTL3 impaired the colony formation ability of HGC-27 cells (left panel). Quantification of the colony formation assay results (right panel). (I) Overexpression of METTL3 effectively promoted GC subcutaneous tumour growth in nude mice (n=6). (J) The tumour volume was monitored every other day, and tumour growth curves were generated. (K) The tumours were extracted and weighed after 20 days. (L) Sections of tumours were stained with anti-Ki-67 and anti-CD31 antibodies by IHC (scale bars=100 µm). (M)Tube formation assay and cell growth in HUVECs cultured with medium collected from BGC823 cells with METTL3 overexpression or from the corresponding control cells. Tubes were imaged (scale bars=200 µm, left panel), and tube formation and cell growth were quantified (right panel). (N) Representative images of three different GC organoids transfected with METTL3 overexpression vectors or control lentivirus for 2 weeks (scale bars=250 µm, left panel) and quantification of organoid diameters (right panel). The data are the means±SEMs of three independent experiments. * p<0.05; *** p<0.001. GC, gastric cancer; HUVEC, human umbilical vein endothelial cell; IHC, immunohistochemistry; WT, wild type.

We also established stable METTL3 knockdown GC cells (HGC-27 and NCI-N87, online supplementary figure S3B, C), and generated METTL3 knockout HGC-27 cells through the clustered regularly interspaced short palindromic repeats/Cas9 gene editing system (figure 3F). As expected, the m6A level was dramatically reduced on METTL3 knockout or knockdown (figure 3F; online supplemental figure S3C). Knockout or knockdown of METTL3 obviously suppressed the soft agar colony formation efficiency and clonogenic ability (figure 3G,H; online supplemental figure S3D, E). We also performed tumour xenograft studies to verify the roles of METTL3 in vivo, and the results showed that overexpression of METTL3 significantly promoted tumour growth, as reflected by the tumour size and weight compared with those of tumours derived from the vector control cells (figure 3I–K). In addition, IHC results showed increased expression of Ki-67, a biomarker of proliferation, in the tumour tissues of the METTL3 overexpression group compared with that in the control group (figure 3L). Surprisingly, microvessel density as evaluated by staining of CD31, a marker of angiogenesis,15 showed a significant increase in tumour tissues of the METTL3 overexpression group compared with that in the vector control group (figure 3L). To further study the function of METTL3 in GC angiogenesis, human umbilical vein endothelial cell (HUVEC) growth and tube formation were investigated in vitro. Both tube formation and HUVEC growth were significantly increased by conditioned medium from BGC823 cells overexpressing METTL3 compared with conditioned medium from the vector controls (figure 3M); while conditioned medium from AGS cells overexpressing catalytic mutant METTL3 had no significant effect on HUVEC growth and tube formation (online supplementary figure S3F). Correspondingly, tube formation and HUVEC growth were significantly impaired by conditioned medium from HGC-27 cells with METTL3 knockout (online supplementary figure S3G). We also established a GC organoid model with organoids generated from three different patients with GC to further verify the potential clinical value of METTL3. The results showed that overexpression of METTL3 by lentivirus significantly promoted GC cell proliferation in the three different GC organoids (figure 3N). These data suggest that METTL3 may act as an oncogene that promotes GC proliferation and angiogenesis dependent on its m6A catalytic activity.

METTL3 promotes GC metastasis in vitro and in vivo

To determine the roles of METTL3 in GC metastasis, we first performed migration and invasion assays in vitro. The data showed that ectopic expression of METTL3 promoted the migration and invasion of BGC823 and AGS cells (figure 4A,B); however, overexpression of catalytic mutant METTL3 had no obvious effect on the migration and invasion of GC cell (online supplementary figure S4A). Conversely, METTL3 knockout or knockdown in HGC-27 and NCI-N87 cells produced the opposite effects (figure 4C; online supplementary figure S4B, C). We further assessed the effect of METTL3 on GC metastasis in vivo. BGC823-luciferase cells overexpressing METTL3 and the corresponding control cells were injected through the splenic portal vein of nude mice. After 8 weeks, upregulation of METTL3 significantly promoted GC liver metastasis, as shown by bioluminescence imaging (figure 4D) and the number and size of liver metastatic lesions (figure 4E,F) compared with those in the control groups. In contrast, NCI-N87 cells with METTL3 deficiency dramatically suppressed GC liver metastasis, as evidenced by the number and size of liver metastatic lesions compared with those in the corresponding control group (figure 4G,H). Collectively, these results indicate the critical role of METTL3 in promoting GC metastasis.

Figure 4

METTL3 promotes GC metastasis in vitro and in vivo. (A–C) Cell migration and invasion assays of BGC823 cells (A), AGS cells (B) and HGC-27 cells (C) after overexpression or knockout of METTL3. Representative images (scale bars=200 µm, upper panel) and quantification (bottom panel) of the cell migration and invasion assay results were shown. (D–F) Overexpression of METTL3 significantly increased GC liver metastasis in nude mice (n=6). (D) Representative bioluminescence imaging of mice 8 weeks after splenic portal vein injection of BGC823 cells with METTL3 overexpression or vector-transfected cells (left panel) and quantification of the results of bioluminescence imaging in the liver region (right panel). (E) Representative images of the metastatic nodes in the livers (scale bars=5 mm, left panel) and quantification of the metastatic nodes (right panel). (F) H&E-stained liver sections (scale bars=200 µm). (G, H) Knockdown of METTL3 in NCI-N87 cells markedly suppressed GC liver metastasis in nude mice (n=6). (G) Representative images of the metastatic nodes in the livers (left panel) and quantification of the metastatic nodes (right panel). (H) H&E-stained liver sections (scale bars=200 µm). The data are the means±SEMs of three independent experiments. * p<0.05; ** p<0.01; *** p<0.001. GC, gastric cancer; WT, wild type.

METTL3-mediated m6A modification of HDGF mRNA maintains its IGF2BP3-dependent stability

To identify the molecular mechanism by which METTL3 promotes GC progression, we performed RNA sequencing (RNA-seq) in GC cells with stable METTL3 overexpression and knockout and conducted m6A-modified RNA immunoprecipitation sequencing (MeRIP-seq) in GC cells with stable METTL3 overexpression and control cells (figure 5A). RNA-seq revealed that 3806 transcripts were significantly upregulated on METTL3 overexpression (fold change >2), while 2346 transcripts were significantly down-regulated (fold change <0.5) on METTL3 knockout. MeRIP-seq revealed that m6A peaks of 9930 transcripts exhibited increased abundance (fold change >1.2). Intriguingly, 3 transcripts were overlapped in the RNA-seq and MeRIP-seq data, and they were from three different genes, i.e., HDGF, C12orf45, and CDKN2A (figure 5A). The m6A abundance of these three transcripts increased by 1.47-, 1.57-, and 1.41-fold on METTL3 upregulation, and the P values were 7.94328E-35, 0.049, and 0.1, respectively. Next, we validated the mRNA level of these three candidate genes in METTL3-overexpressing GC cell lines (BGC823 and AGS) and METTL3 knockout or knockdown GC cell lines (HGC-27 and NCI-N87). Only HDGF but not C12orf45 and CDKN2A were consistently regulated by METTL3 in all four GC cell lines (figure 5B,C; online supplementary figure S5A-G). We also confirmed via a western blot assay that the HDGF protein levels were positively regulated by METTL3 in different GC cell lines (figure 5D). Meanwhile, we found that catalytic mutant METTL3 overexpression could not regulate HDGF on both mRNA and protein levels (figure 5E; online supplementary figure S5H). Additionally, the HDGF level was significantly increased in conditioned medium from BGC823 cells overexpressing METTL3, but not catalytic mutant METTL3, compared with that from the corresponding control cells (figure 5F). The HDGF level was also significantly decreased in conditioned medium from METTL3 knockout HGC-27 cells compared with that from the control cells (online supplementary figure S5I). As shown in figure 5G, HDGF was localised in both cytoplasm and nucleus, and the HDGF levels in both of these compartments were reduced on METTL3 knockout. Moreover, METTL3 overexpression resulted in appreciably increased HDGF staining in GC liver metastatic lesions but knockdown of METTL3 markedly reduced HDGF staining (online supplementary figure S5J).

Figure 5

METTL3-mediated m6A modification of HDGF mRNA maintains its IGF2BP3-dependent stability. (A) RNA-seq and MeRIP-seq identified differentially expressed genes in METTL3 stable overexpression and knockout cells when compared with their corresponding controls. (B–C) The mRNA levels of HDGF in METTL3-overexpressing (B), and METTL3 knockout (C) GC cells were detected by qRT-PCR. (D) The levels of the indicated proteins in METTL3-overexpressing and METTL3-deficient GC cells were detected by western blotting. (E) The protein levels of HDGF in wild type or catalytic mutant METTL3-overexpressing AGS cells were measured by western blotting. (F) The HDGF levels in conditional medium from wild type or catalytic mutant METTL3-overexpressing BGC823 cells were measured via ELISA assay. (G) Immunofluorescence (IF) assay was used to detect the expression and location of HDGF in METTL3 wild type and knockout GC cells (magnification: × 40). (H) Global profiling of m6A in BGC823 cells and the sequence motif identified from the top 1000 m6A peaks. (I) The m6A abundances on HDGF mRNA transcripts in BGC823 cells as detected by MeRIP-seq were plotted. (J) MeRIP-qPCR analysis was employed to demonstrate METTL3-mediated HDGF m6A modifications. The m6A modification of HDGF was increased on upregulation of METTL3, while it was depleted on knockout of METTL3. (K) The levels of HDGF expression in METTL3-overexpressing, METTL3 knockout and their corresponding control GC cells treated with actinomycin D (2 µg/mL) at the indicated time points were detected by qRT-PCR. (L) The mRNA levels of HDGF in IGF2BP3 knockdown GC cells were detected by qRT-PCR. (M) RNA immunoprecipitation (RIP)-qPCR assay using IGF2BP3-specific antibody and IgG control antibody to detect the enrichment of IGF2BP3 binding to HDGF m6A modification sites. (N) The levels of HDGF expression were detected in GC and paired normal gastric mucosa by qRT-PCR (n=28). (O) The levels of HDGF expression were analysed in GC (n=413) and normal gastric mucosa (n=32) using TCGA data. (P) The levels of IGF2BP3 expression were detected in GC and paired normal gastric mucosa by qRT-PCR (n=28). (Q) The levels of IGF2BP3 expression were analysed in GC (n=413) and normal gastric mucosa (n=32) using TCGA data. (R, S) METTL3 (R) and IGF2BP3 (S) expression were positively correlated with HDGF expression in GC (linear regression). (T) Knockdown of METTL3, HDGF, IGF2BP3 effectively inhibited NCI-N87 GC cells subcutaneous tumour growth in nude mice (n=6). (U) The tumour volume was monitored every other day, and tumour growth curves were generated. (V) The tumours were extracted and weighed after 21 days. The data are the means±SEM of three independent experiments, * p<0.05; ** p<0.01; *** p<0.001. GC, gastric cancer; MeRIP-seq, m6A-modified RNA immunoprecipitation sequencing; RNA-seq, RNA sequencing; TCGA, The Cancer Genome Atlas; WT, wild type.

The m6A occurs mostly in RRACH (R=G or A, H=A, C or U) consensus sequence.21 Our MeRIP-seq results revealed that the AGACC motif was highly enriched in the immunopurified RNA on METTL3 overexpression (figure 5H), and the m6A abundance in the HDGF mRNA was significantly increased on METTL3 upregulation (p=7.94328E-35) (figure 5I). Interestingly, AGACC motif and GAACA motif were included in the exons of HDGF. The MeRIP-seq data were validated by MeRIP-qPCR, which showed that compared with the IgG control antibody, the m6A-specific antibody significantly enriched HDGF mRNA on METTL3 overexpression, while the m6A-specific antibody markedly reduced the enrichment of HDGF mRNA on METTL3 knockout (figure 5J). Considering that the m6A modification positively regulated the mRNA level of HDGF, we then investigated whether the m6A modification affected the stability of HDGF mRNA. The GC cells were treated with actinomycin D, an inhibitor of transcription, for the indicated times. The level of HDGF mRNA was shown to be highly stable on METTL3 overexpression, while the opposite effect was indicated on knockout of METTL3 (figure 5K). A recent study reported that IGF2BPs, including IGF2BP1/2/3, are a distinct family of m6A readers that target thousands of mRNA transcripts through the recognition of m6A motif, and HDGF was one of the targets of IGF2BP1/2/3 in their high-throughput sequencing results.22 Therefore, we explored the effect of IGF2BP1/2/3 on HDGF mRNA stabilisation. We designed two specific siRNAs against each target of IGF2BP1/2/3 and confirmed the knockdown efficiency of these constructs (online supplementary figure S5K-M). Only knockdown of IGF2BP3 markedly suppressed HDGF mRNA expression, while IGF2BP1/2 had no noticeable effect (figure 5L; online supplementary figure S5N, O). As shown in figure 5M, compared with the IgG control antibody, the IGF2BP3-specific antibody significantly enriched HDGF mRNA in the RIP-qPCR assay. Next, we compared the mRNA levels of HDGF and IGF2BP3 in 28 GC tissues and TCGA data. The results showed that the expression of HDGF and IGF2BP3 was significantly upregulated in GC tissues compared with that in normal gastric tissues (figure 5N–Q). We also found that the expression of METTL3 and IGF2BP3 was significantly correlated with HDGF expression (R2=0.232, p=0.009; R2=0.1709, p=0.0287) (figure 5R,S). We also performed tumour xenograft studies to verify the roles of HDGF and IGF2BP3 in vivo, and the results showed that knockdown of METTL3, HDGF and IGF2BP3 significantly inhibited tumour growth compared with the control group (figure 5T–V). Together, our findings indicate that METTL3-mediated m6A modification maintains HDGF expression via IGF2BP3-dependent HDGF mRNA stability.

METTL3 accelerates the GC malignant process by upregulating HDGF expression

To further characterise the oncogenic function of HDGF in GC, we designed three different siRNAs to target HDGF and confirmed the knockdown efficiency of these constructs by qRT-PCR and western blotting (figure 6A). Knockdown of HDGF dramatically suppressed colony formation (figure 6B), cell migration and invasion (figure 6C), and HUVEC growth and tube formation (figure 6D). Furthermore, HDGF recombinant protein (rHDGF) rescued the colony forming ability of METTL3 knockout GC cells (figure 6E). As expected, the expression of HDGF was knocked down in METTL3-overexpressing AGS cells using its specific siRNAs, which markedly suppressed METTL3-induced GC cell migration and invasion (figure 6F; online supplementary figure S6A), but rHDGF rescued the migration and invasion abilities of METTL3 knockout GC cells (online supplementary figure S6B). In addition, HDGF antibody could block the HUVEC growth and tube formation caused by METTL3 overexpression while rHDGF could rescue the HUVEC growth and tube formation caused by METTL3 knockout in GC cells (online supplementary figure S6C, D). Moreover, knockdown of HDGF markedly suppressed METTL3-induced GC growth and angiogenesis in vivo (figure 6G–J). Thus, our data suggest that METTL3 promotes GC malignant progression through the upregulation of HDGF expression.

Figure 6

METTL3 accelerates GC malignant progression by upregulating HDGF. (A) The HDGF knockdown efficiency was verified at the mRNA levels (upper panel) and protein levels (bottom panel) in HGC-27 cells by qRT-PCR and western blot assay, respectively. (B) Knockdown of HDGF impaired colony formation ability in HGC-27 cells (upper panel); quantification results of colony formation (bottom panel). (C) Knockdown of HDGF impaired cell migration and invasion ability in HGC-27 cells (scale bars=200 µm, upper panel); quantification results of cell migration and invasion (bottom panel). (D) Tube formation assay in HUVECs cultured with medium collected from HGC-27 cells with HDGF deficiency and corresponding control cells. tubes were imaged (scale bars=200 µm, upper panel) and quantified for tube formation and cell growth (bottom panel). (E) Representative images (left panel) and quantification results (right panel) of the colony formation abilities of METTL3 knockout HGC-27 cells added with human rHDGF or PBS. (F) Representative images (scale bars=200 µm, left panel) and quantification results (right panel) of the cell migration and invasion abilities of METTL3-overexpressing AGS cells transfected with the HDGF siRNAs or their corresponding controls. (G) Knockdown of HDGF inhibited METTL3-induced AGS GC cells subcutaneous tumour growth in nude mice (n=6). (H) The tumour volume was monitored every other day, and tumour growth curves were generated. (I) The tumours were extracted and weighed after 19 days. (J) Sections of tumours were stained with anti-Ki-67 and anti-CD31 antibodies by IHC (scale bars=100 µm). The data are the means±SEM of three independent experiments, * p<0.05; ** p<0.01; *** p<0.001. GC, gastric cancer; HUVEC, human umbilical vein endothelial cell; rHDGF, HDGF recombinant protein; IHC, immunohistochemistry; WT, wild type.

METTL3 enhances glycolysis in GC through HDGF-activated expression of GLUT4 and ENO2

Glycolysis is one of the primary metabolic signatures in cancers.23 RNA-seq data showed that different metabolic pathways, especially glycolysis/gluconeogenesis, were involved in GC progression on METTL3 overexpression (online supplemental figure S7A). Therefore, we investigated whether METTL3 regulated glycolysis to promote GC malignant processes. As shown in figure 7A,B, upregulation of METTL3 expression in BGC823 cells significantly increased glucose uptake and lactate production. In contrast, knockout of METTL3 in HGC-27 cells markedly reduced glucose uptake and lactate production (online supplementary figure S7B, C). To investigate whether HDGF could restore glycolytic activity in METTL3 knockout cells, HDGF overexpression plasmids were transfected into METTL3 knockout cells. Overexpression of HDGF was first confirmed (online supplementary figure S7D), and HDGF overexpression was found to restore glycolytic activity and significantly increase lactate production in METTL3 knockout GC cells (figure 7C). The extracellular acidification rate kinetic profiles further demonstrated the substantial increase in glycolytic activity in METTL3-overexpressing BGC823 cells and the decrease in METTL3 knockout HGC-27 cells (figure 7D,E). To further investigate the mechanism by which METTL3 regulates glycolysis, we examined the transcription of a panel of glucose metabolism-related genes in METTL3-overexpressing and METTL3 knockout GC cells (figure 7F; online supplementary figure S7E). We also examined whether these genes were regulated by HDGF, and we examined glucose metabolism-related genes in HDGF knockdown GC cells (figure 7G). Intriguingly, the expression levels of only GLUT4 and ENO2 were primarily enhanced (fold change >1.5) by upregulation of METTL3 but were consistently decreased (fold change <0.5) by knockout of METTL3 or knockdown of HDGF. Considering that HDGF can localise in the nucleus and act as a transcriptional regulator,24 we investigated whether HDGF could transcriptionally activate the expression of GLUT4 and ENO2. A ChIP assay was applied and it was shown that HDGF was directly bound to the promoter region of GLUT4 and ENO2 but not to the HK2 promoter (figure 7H). Moreover, IHC results showed that METTL3 overexpression resulted in appreciably increased HDGF, GLUT4 and ENO2 staining in AGS tumour xenograft, while GLUT4 and ENO2 expressions were significantly decreased when HDGF was knocking down (figure 7I). Consistently, HDGF, GLUT4 and ENO2 staining were obviously decreased in NCI-N87 tumour xenograft with deficiency of METTL3 (figure 7J). Collectively, these results suggest that METTL3/HDGF promotes GC tumourigenesis and liver metastasis partly through the glycolytic pathway.

Figure 7

METTL3-HDGF axis enhances glycolysis by targeting GLUT4 and ENO2 in GC cells. (A, B) Overexpression of METTL3-induced glucose uptake (A) and lactate production (B) in BGC823 cells. (C) Overexpression of HDGF rescues lactate production in METTL3 knockout HGC-27 cells. (D, E) The ECAR profile was monitored in METTL3-overexpressing (D) or METTL3 knockout (E) GC cells with a Seahorse XF24 analyser for 100 min. The metabolic inhibitors were injected sequentially at different time points as indicated. (F, G) GLUT4 and ENO2 were identified as METTL3/HDGF-regulated genes. The expression of a panel of glucose metabolism-related genes was detected by qRT-PCR in METTL3-overexpressing (F) or HDGF knockdown (G) cells and their corresponding control cells. (H) ChIP assay was used to detect the ability of HDGF binding to GLUT4, ENO2 and the HK2 promoter. (I) Sections of tumour xenografts from METTL3-overexpressing or METTL3-overexpressing with knockdown of HDGF AGS cells subcutaneously injected nude mice were stained with METTL3, HDGF, GLUT4 and ENO2 antibodies by IHC (scale bars=100 µm). (J) Sections of tumour xenografts from METTL3 knockdown NCI-N87 cells subcutaneously injected nude mice were stained with METTL3, HDGF, GLUT4 and ENO2 antibodies by IHC (scale bars=100 µm). The data are the means±SEM of three independent experiments, * p<0.05; ** p<0.01; *** p<0.001. ChIP, chromatin immunoprecipitation; ECAR, extracellular acidification rate; GC, gastric cancer; IHC, immunohistochemistry; WT, wild type.

To investigate the clinical significance of the METTL3/HDGF axis in promoting GC glycolysis and angiogenesis, we examined the expression of METTL3, HDGF, GLUT4, ENO2 and CD31 in cancerous tissues of patients with GC and subsequently categorised the tissues into the METTL3-low and METTL3-high groups and determined their relevance in GC (figure 8A). The expression of METTL3 was positively correlated with the expression of HDGF, CD31, GLUT4 and ENO2 in the 54 patients with GC (figure 8B). Furthermore, we found that patients with GC with increased GLUT4 and ENO2 mRNA levels had poorer OS via the online bioinformatics tool Kaplan-Meier Plotter (http://kmplot.com/analysis/; supplementary figure S7F). These data suggest that the METTL3/HDGF axis promotes human GC progression partly through inducing glycolysis and angiogenesis.

Figure 8

The clinical significance of the METTL3/HDGF axis in inducing tumour angiogenesis and glycolysis in human GC. (A, B) METTL3 levels were significantly associated with the expression of HDGF, CD31, GLUT4, and ENO2 in 54 primary human GC specimens. (A) Two representative cases are shown (scale bars=50 µm). (B) The percentages of specimens showing low or high METTL3 expression relative to the levels of HDGF, CD31, GLUT4 and ENO2 are shown. (C) The graphic illustration of METTL3 modulating tumour glycolysis and angiogenesis promoting tumour growth and liver metastasis of GC. *** p<0.001. GC, gastric cancer.

Discussion

Over 100 types of chemical modifications are present in human RNA.17 25 Among the RNA modifications, m6A is the most prevalent in mRNA and non-coding RNA.26 A series of recent studies indicate that m6A modification is involved in various human diseases, including type II diabetes, cancer progression, viral infection and heart failure.27–31 The m6A modification is dynamically regulated via m6A methyltransferase, demethylases and readers, which regulate RNA biological functions.32 In the present study, we demonstrated that the m6A level is significantly increased due to the upregulation of methyltransferase METTL3 in GC. Mechanistically, P300-mediated H3K27ac activation in the promoter of METTL3 induces METTL3 transcription, which stimulates the m6A modification of HDGF mRNA, and the m6A reader IGF2BP3 then directly binds to the m6A site on HDGF mRNA and maintains the stability of this mRNA. Secreted HDGF promotes tumour angiogenesis, while nuclear HDGF activates GLUT4 and ENO2 expression, followed by an increase of glycolysis in GC cells, which promotes the progression of GC and leads to worse clinical prognosis (figure 8C).

Recent reports have shown that m6A modification plays important and diverse biological functions in the progression of various cancers.33 Among the m6A modulators, METTL3 is thoroughly and widely studied in various cancer types.34 Previous studies have reported that METTL3 promotes the progression of hepatocellular carcinoma,21 bladder cancer,35 lung cancer,20 breast cancer,36 glioblastoma29 and acute myeloid leukaemia.37 Here, we first found that H3K27ac activated METTL3 transcription and increased METTL3 expression in GC. Subsequently, METTL3 induced an increase in m6A modification in GC. Furthermore, our data indicated that METTL3 expression was associated with poor prognosis in patients with GC and that including METTL3 expression could enhance the predictive ability of clinical risk scores, suggesting that METTL3 may be a biomarker for GC diagnosis. Lauren’s criteria was the most widely used classification in GC including intestinal, diffuse and mixed type, which was associated with GC prognosis. However, the expression of METTL3 was not shown the statistical difference between diffuse and intestinal types in patients with GC. In vitro and in vivo studies show that METTL3 promotes GC tumour growth and liver metastasis dependent on its m6A catalytic activity. Thus, METTL3 could be used as a potential predictive biomarker and therapeutic target for GC.

Using RNA-seq and MeRIP-seq, we found the promising result that HDGF was the key downstream target of METTL3 in GC. HDGF is a novel growth factor that was initially purified from the conditioned medium of the HuH-7 liver cancer cell line.24 38 HDGF has been reported to be involved in numerous cancer processes, including cancer growth, apoptosis, angiogenesis and metastasis.39 HDGF can be shuttled between cytoplasm and nucleus under different conditions.40 Extracellular HDGF can stimulate tumorigenesis and progression via receptor-mediated pathways.39 Nuclear HDGF can regulate the transcription of downstream genes by binding to their DNA.39 41 By our screen of a panel of glucose metabolism-related genes, GLUT4 and ENO2 were identified as the target genes of METTL3/HDGF. Moreover, HDGF directly bound to the promoters of GLUT4 and ENO2, which induced their expression. It has reported that HDGF could act as a transcription repressor to regulate gene expression; however, some transcription factors or co-transcription factors could play two-sidedness in regulating gene expression, such as YAP,42 TAZ,42 YY1.43 Therefore, we consider that HDGF also could act an activator in some condition to regulate cell biological function. In addition, GLUT4 is a glucose transporter that promotes glucose uptake, while ENO2 is a cytoplasmic glycolytic enzyme that accelerates glycolysis.44 45 Here, the expression of METTL3 was positively correlated with the expression of HDGF, CD31, GLUT4 and ENO2 in GC tissues, indicating the clinical significance of the METTL3/HDGF axis in promoting GC glycolysis and angiogenesis. These data provide a new regulatory model of METTL3 in GC progression.

In summary, our findings reveal an oncogenic role of METTL3 in GC development. Mechanistically, the METTL3/HDGF/GLUT4/ENO2 axis promotes GC tumorigenesis and metastasis via an increase in glycolysis and angiogenesis. Moreover, METTL3 expression is significantly increased in GC and is correlated with poor prognosis of patients with GC. Therefore, METTL3 might be a potential predictor and therapeutic target for GC.

References

View Abstract

Footnotes

  • QW, CC and QD are joint first authors.

  • QW, CC and QD contributed equally.

  • Contributors QW, CC, QD, YZ, ZW, JC, ZJ and YZhao performed the experiments; QW and CC analysed data; GX, JZ, QD, BS and XZ provided the samples; QW and SW wrote the paper; BS and JZhang commented on the study and revised the paper; SW designed the research.

  • Funding This work was supported in part by the National Natural Science Foundation of China (81773383, 81370078, 31771628, 81903085); and the Science Foundation for Distinguished Young Scholars of Jiangsu Province (BK20170047); and the Fundamental Research Funds for the Central Universities (021414380439); and the Project funded by China Postdoctoral Science Foundation (2019M651808).

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Ethics approval The use of human gastric cancer tissues and the waiver of patient consent in this study were approved by the Clinical Research Review Board of the Institutional of the Nantong Cancer Hospital and Nanjing Drum Tower Hospital, respectively. The study was conducted according to the principles expressed in the Declaration of Helsinki.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement Data are available upon reasonable request.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.