METTL3-Mediated DLGAP1 Antisense RNA 2 Promotes Gastric Cancer Tumorigenesis Through Facilitating C-Myc-Dependent Warburg Effect


 Background

The critical roles of N6-methyladenosine (m6A) modification have been demonstrated by more and more evidence. However, the cross-talking of m6A and long non-coding RNAs (lncRNAs) in gastric cancer (GC) tumorigenesis is still unclear. Here, our work focused on the functions and molecular mechanism of m6A-modified lncRNA DLGAP1 antisense RNA 2 (DLGAP1-AS2) in GC.
Methods

LncRNA expression profile data was derived from GEO. M6A profile was screened using Methylated RNA immunoprecipitation sequencing (MeRIP-Seq). The metabolism assays were conducted using quantitative analysis of glucose, lactate, ATP and extracellular acidification rate (ECAR). The m6A level of specific RNA was identified using MeRIP-qPCR. The molecular interaction was detected using RIP assay.
Results

Microarray analysis found that lncRNA DLGAP1-AS2 up-regulated in GC cells. Clinical data showed that DLGAP1-AS2 high-expression was correlated with advanced pathological stage and poor prognosis. Functionally, DLGAP1-AS2 promoted the Warburg effect (aerobic glycolysis) and knockdown of DLGAP1-AS2 suppressed the tumor growth of GC cells. Mechanistically, m6A methyltransferase METTL3 enhanced the stability of DLGAP1-AS2 via m6A site binding. Moreover, DLGAP1-AS2 interacted with YTHDF1 to enhance the stability of c-Myc mRNA through DLGAP1-AS2/m6A/YTHDF1/c-Myc mRNA.
Conclusions

In conclusion, our work indicates the functions of m6A-modified DLGAP1-AS2 in the GC aerobic glycolysis, disclosing a potential m6A-dependent manner for GC treatment.


Introduction
Gastric cancer (GC) is a worldwide digestive system malignant tumor with a signi cant high mortality and morbidity [1,2]. In spite of the achievement on the GC therapy, including chemotherapy, radiotherapy and surgery, the overall survival rate or survival time of GC sufferers remains unimproved [3]. The reason for the lethal trait of GC is metastases and recurrence, however, the molecular mechanisms for GC progression is still poorly understood [4]. Therefore, it is signi cantly critical to explore the underlying mechanism of GC development, covering prognostic biomarkers, diagnostic and therapeutic targets.
Long non-coding RNAs (lncRNAs) are novel noncoding RNAs longer than 200 nucleotides length without protein-coding potential [5,6]. Over the years of research, growing number of lncRNAs are identi ed in the GC occurrence [7,8]. For example, lncRNA MAGI2-AS3 promotes the tumor progression through sponging miR-141/200a to overexpress ZEB1 level, highlighting an ideal biomarker for GC potential therapeutic target [9]. SNHG11 post-transcriptionally promotes oncogenic autophagy progression through aggravating ATG12 and CTNNB1 through miR-483-3p and miR-1276 [10]. In summary, these ndings suggest that lncRNA may function as an essential regulator in GC progression. N 6 -methyladenosine (m 6 A) is the most prevalent internal modi cation connected with eukaryotic mRNAs fate, including mRNA metabolism, splicing, export and stability [11][12][13]. Besides, in various tumor physiological events, m 6 A participates in these tumor progression related progresses [14]. For example, major m 6 A methyltransferase METTL3 expression elevated in GC tissue and was predictive of poor prognosis. The stability of ZMYM1 mRNA was enhanced by through METTL3 relying on reader protein HuR [15]. Wilms' tumour 1-associated protein (WTAP) highly expressed in GC cells and tissue, which indicates a poor prognosis for GC patients and correlated with a cell-related immune response [16]. Thus, these evidences support that m 6 A modi cation signi cantly modulates the GC carcinogenesis.
Herein, our work focused on the cross-talking of m 6 A and lncRNAs in gastric GC tumorigenesis to uncover the functions and molecular mechanism of m 6 A-modi ed lncRNA DLGAP1 antisense RNA 2 (DLGAP1-AS2). We found that lncRNA DLGAP1-AS2 was highly expressed in GC, which was correlated with poor prognosis and advanced clinical stage. Moreover, m 6 A methyltransferase METTL3 enhanced the stability of DLGAP1-AS2 and then DLGAP1-AS2 interacted with YTHDF1 to enhance the stability of c-Myc mRNA in m 6 A-dependent manner. Overall, these ndings provided a potential m 6 A-dependent therapeutic strategy for GC treatment.

Materials And Methods
Patient tissues specimens collection A total of 50 GC tissue samples in the cohort, which were formalin-xed and para n-embedded, were recruited between January 2016 and December 2018 from patients who had underwent operation in Fujian Provincial Hospital. All samples were staged according to the criteria of the 7 th Edition of the AJCC Cancer Staging Manual (Stomach, 2010) [17]. This study was approved by the Fujian Provincial Hospital Ethics Review Committee. Informed consent was written by each patient before the study.
Cell culture GC cell lines (AGS, SNU-216, MKN45, AGS) and the immortalized human gastric epithelial cell line (GES-1) were purchased from the Cell Bank of the Chinese Academy of Sciences. The cells were maintained in RPMI-1640 medium supplemented with 10% fetal bovine serum (HyClone, Logan, UT, USA) and antibiotics at 37 °C under 5% CO 2 .
Cell were grown in 1 μg/ml puromycin (Invitrogen) to select the stable transfected cells. Moreover, for the overexpression of DLGAP1-AS2, the cDNA of DLGAP1-AS2 was PCR-ampli ed ad then sub-cloned into pcDNA 3.1 vector (Invitrogen, Carlsbad, Calif, USA). The e ciency of silencing or overexpression was subsequently quanti ed by qRT-PCR. The siRNAs and their controls (si-NC) were provided by GenePharma (Shanghai, China) and transfected using Lipofectamine 2000 (Invitrogen) when 70 % con uent growth.
Quantitative real-time PCR (qRT-PCR) Total RNA from GC tissues or cell lines was extracted by TRIzol Reagent (Invitrogen, CA, USA). For the concentration and purity of RNA, RNA was detected with ultraviolet spectrophotometer using 260 nm and 280 nm. The real-time PCR was performed using SYBR Premix Ex Taq (Takara) on the 7900 Real-time PCR System. Reverse transcription was performed using PrimerScript RT Reagent Kit (TaKaRa, Kyoto, Japan). Fast SYBR green master mix (Life Technologies) applied to perform qRT-PCR. The relative expression was calculated using the 2 −△△CT method. The β-actin acted as the internal mRNA control. The primer sequences were displayed in Table S1.

Western blot assay
The procedure details were conducted according to previous study. Proteins from transfected GC cells were extracted using RIPA buffer (Beyotime, China) and authenticated using BCA Kit (Pierce, Rockford, IL, USA). Protein was resolved using the SDS-PAGE Electrophoresis System and then transferred to PVDF membranes (Millipore, USA). Speci c primary antibodies (anti-METTL3, 1:1000, ab195352, anti-c-MYC, 1:1000, ab32072) were incubated at 4 °C overnight. β-actin was the internal reference.

Methylated RNA immunoprecipitation (MeRIP) Sequencing and qPCR
All the speci c manipulations were performed according to the protocol of Magna MeRIP™ m 6 A Kit (Merck Millipore). In brief, total RNAs were isolated from AGS cells and chemically fragmented into 100-300 nt. Fragmented fragments were incubated with m 6 A antibody for immunoprecipitation. The MeRIP was performed according to the manufacturer's instructions. Samples were sequenced with the HiSeq PE150 platform. The enrichment of m 6 A containing mRNA was sent for quantitative RT-PCR. The RT-qPCR primers were listed in Table S1.
Actinomycin D assay AGS cells were seeded in 6-well plates (2×10 5 cells/well). 24 hours later, cells were exposed to Actinomycin D (Act D, 2 μg/ml, Sigma) and harvested at indicated time point. The RNA remaining level was analyzed using qRT-PCR and normalized to mock group (0 hour).

RNA immunoprecipitation (RIP) assay
RIP assay was performed using EZ-Magna RIP™ RNA-Binding Protein Immunoprecipitation Kit (Millipore, USA) according to manufacturer's protocol. Speci c antibodies (anti-YTHDF1, Abcam) was conjugated with magnetic beads and then dissolved in RIP buffer. Approximate 90% con uence, cells were lysed using complete RIP lysis buffer (100 μl) containing protease inhibitor and RNase Inhibitor (Millipore).
Mouse anti-IgG antibody (Cell Signaling Technology, USA) acted as negative control. Then, total RNA was retrieved and the relative expression was detected by qRT-PCR analysis.

Subcellular localization
Probes were obtained from Genepharma (Shanghai, China). Fluorescence in situ hybridization (FISH) was performed using uorescent in situ hybridization kit according to the manufacturer's protocols (Genepharma). Cy3-labeled probe for DLGAP1-AS2 and FAM-labeled probe for c-Myc, YTHDF1 were used for localization of DLGAP1-AS2, c-Myc and YTHDF1 in GC cells. Nuclei was stained with 4,6-diamidino-2phenylindole (DAPI) and images were obtained using confocal microscope (Olympus).

Animal models in vivo
Approximately 2×10 6 GC cells (AGS), transfected with sh-DLGAP1-AS2 or controls, were suspended in 100 μl PBS and then injected into ank of male BALB/c nude mice (5-week old). In the following observation, tumor size was recorded by vernier caliper. At the indicated time, mice were anesthetized and sacri ced for tumors neoplasm acquisition. All animal assays were performed in the light of institutional Fujian Provincial Hospital Animal Ethics Committee guidelines.

Statistical analysis
The data was expressed as mean ± S.D (standard deviation). All results were conducted using SPSS Statistics software (Armonk, NY, USA) or GraphPad prism 7.0 (La Jolla, CA, USA). Difference within groups was calculated using One-way ANOVA test, non-parametric Mann-Whitney test and student's ttest. Overall survival rate was calculated using Kaplan-Meier analysis and log-rank test. *P ≤ 0.05 or **P ≤ 0.01 was considered statistically signi cant. Results DLGAP1-AS2 was an up-regulated lncRNA in GC Using the microarray data (GSE147167), we found that numerous lncRNAs were dysregulated in the GC tissue samples ( Figure 1A). Moreover, we found that a potential up-regulated lncRNA (DLGAP1-AS2) might be correlated with the GC carcinogenesis. In GC cell lines, expression of DLGAP1-AS2) overexpressed as comparing to the normal cell lines ( Figure 1B). In GC samples, DLGAP1-AS2 levels signi cantly over-expressed as comparing to the normal adjacent tissue ( Figure 1C, Table 1). Kaplan-Meier analysis and log-rank test found that GC tissue samples with high DLGAP1-AS2 levels displayed a lower survival rate as the control cohort ( Figure 1D). Therefore, these data revealed an up-regulated lncRNA DLGAP1-AS2 in GC.

MeRIP-Seq revealed the m 6 A modi cation of DLGAP1-AS2
MeRIP-Seq in GC cells revealed the m 6 A peaks density, including 5'-UTR, coding region (CDS) and 3'-UTR ( Figure 2A, 2C). The signi cant m 6 A motif for the m 6 A peaks was GGAC ( Figure 2B). MeRIP-Seq demonstrated that there was a remarkable m 6 A modi cation site in the 3'-UTR of DLGAP1-AS2 ( Figure   2D). In the 3'-UTR sequences of DLGAP1-AS2, there was a remarkable m 6 A site (AGAC) ( Figure 2E). In the exploration which m 6 A enzyme catalyzes the m 6 A modi cation on DLGAP1-AS2, we found that m 6 A reader YTHDF1 was positively correlated to the level of DLGAP1-AS2 ( Figure 2F). Taken together, these MeRIP-Seq data indicated the m 6 A modi cation of DLGAP1-AS2.
METTL3 enhanced the stability of DLGAP1-AS2 in m 6 A-dependent manner In the MeRIP-Seq, we noticed a m 6 A modi cation of DLGAP1-AS2, thus, DLGAP1-AS2 might be modulated in m 6 A-dependent manner. To investigate the possibility which m 6 A enzyme catalyzes the m 6 A modi cation on DLGAP1-AS2, we found that m 6 A methyltransferase METTL3 was positively correlated to the level of DLGAP1-AS2 ( Figure 3A). In GC cells, METTL3 overexpression was transfected ( Figure 3B).
After METTL3 overexpression, the m 6 A modi cation levels were found to be up-regulated ( Figure 3C, 3D). RNA stability analysis found that the METTL3 overexpression accelerated the RNA stability of DLGAP1-AS2 in MKN45 cells and AGS cells ( Figure 3E, 3F). In summary, these ndings revealed that m 6 A-modi ed DLGAP1-AS2 demonstrated higher stability mediated by METTL3.

DLGAP1-AS2promoted the aerobic glycolysis of GC
In the functional experiments, enhanced expression or silencing of DLGAP1-AS2 was constructed in MKN45 or AGS cells ( Figure 4A). Subsequently, the biological functions of DLGAP1-AS2 on GC aerobic glycolysis were investigated. Glucose uptake analysis illustrate that enforced DLGAP1-AS2 expression promoted the glucose consumption in MKN45 cells, and knockdown of DLGAP1-AS2 repressed the glucose consumption in AGS cells ( Figure 4B). Lactate production analysis unveiled that enforced DLGAP1-AS2 expression accelerated the lactate production in MKN45 cells, and knockdown of DLGAP1-AS2 inhibited the lactate production in AGS cells ( Figure 4C). ATP quantitative analysis illustrated that enforced DLGAP1-AS2 expression facilitated the ATP production in MKN45 cells, and knockdown of DLGAP1-AS2 restrained the ATP production in AGS cells ( Figure 4D). ECAR analysis for GC glycolytic ability illustrated that enforced DLGAP1-AS2 expression promoted the extracellular acidi cation accumulation ( Figure 4E), and knockdown of DLGAP1-AS2 repressed the acidi cation ( Figure 4F).
Moreover, the in vivo animal experiments indicated that knockdown of DLGAP1-AS2 inhibited the tumor neoplasm growth of GC cells ( Figure 4G). In conclusion, these ndings demonstrated that DLGAP1-AS2 promoted the aerobic glycolysis of GC

DLGAP1-AS2 targeted c-Myc via YTHDF1-dependent manner
On the basis of nding that DLGAP1-AS2 regulated the aerobic glycolysis, we assumed that DLGAP1-AS2 might regulate certain key enzyme of glycolysis. Firstly, the localization analysis using RNA-FISH found that the distribution of DLGAP1-AS2 was in cytoplasm, which was in accordance with the c-Myc and YTHDF1 ( Figure 5A). The coincident cytoplasmic distribution of DLGAP1-AS2, YTHDF1 and c-Myc indicated the potential interaction within them. MeRIP-Seq presented that there was a m 6 A site in the c-Myc mRNA, suggesting a potential regulation for DLGAP1-AS2 via m 6 A-dependent manner ( Figure 5B). In the GEPIA database, c-Myc was positively correlated with DLGAP1-AS2 and YTHDF1, indicating that c-Myc might act a target of DLGAP1-AS2 ( Figure 5C). Besides, the c-Myc levels were highly expressed in the stomach neoplasm (STAD, Stomach adenocarcinoma) ( Figure 5D). In summary, these data illustrated that c-Myc acted as a target of DLGAP1-AS2.

DLGAP1-AS2promoted c-Myc mRNA stability through m 6 A-YTHDF1-dependent manner
The MeRIP-Seq demonstrated that there was a m 6 A modi cation site in the 3'-UTR of c-Myc mRNA, which was in accord with the CAGAC motif ( Figure 6A). MeRIP-qPCR suggested us that the m 6 A enrichment on c-Myc mRNA was higher in the GC cells (MKN45, AGS) ( Figure 6B). Moreover, the RNA immunoprecipitation (RIP) unveiled that enhanced DLGAP1-AS2 expression promoted the c-Myc mRNA level precipitated by the anti-YTHDF1 antibody ( Figure 6B, left). Besides, knockdown of DLGAP1-AS2 inhibited the c-Myc mRNA level ( Figure 6B, right). RT-qPCR found that enhanced DLGAP1-AS2 expression ampli ed the c-Myc mRNA level ( Figure 6D, up). Moreover, the knockdown of DLGAP1-AS2 decreased the c-Myc mRNA level ( Figure 6D, down). RNA stability analysis found that DLGAP1-AS2 overexpression promoted the c-Myc mRNA remaining level, and the knockdown of DLGAP1-AS2 repressed the c-Myc mRNA remaining ( Figure 6E). In summary, these ndings illustrated that DLGAP1-AS2 promoted c-Myc mRNA stability through m 6 A-YTHDF1-dependent manner (Figure 7).

Discussion
As one of the most prominent characteristics of malignant tumors, aerobic glycolysis (also known as Warburg effect) is a complex process affected by genetic and epigenetic modi cations [18,19]. For the regulation of tumorigenesis, (Warburg effect) aerobic glycolysis exerts critical roles on the progress of tumor cell energy metabolism, providing the main energy supplement. In the current study, we demonstrated that m 6 A-modi ed DLGAP1-AS2 expression was increased in GC tissues and cell lines.
Furthermore, a high DLGAP1-AS2 expression implied a poor prognosis for GC.
Here, we found that DLGAP1-AS2 expression was signi cantly up-regulated in the GC and moreover promoted the glucose uptake, lactate production and ATP generation. Our results re ect this possibility that DLGAP1-AS2 may act as an energy metabolism-related oncogene in GC. Accumulating evidences have illustrated that lncRNA plays critical roles in cancer aerobic glycolysis. For instance, the overexpressed LINC00242 promotes the GC proliferation and aerobic glycolysis in vitro and relieves the tumorigenesis in vivo through directly targeting G6PD [20]. LINC01391 overexpression hampers the proliferation, invasion and aerobic glycolysis in vitro and tumor growth in vivo of GC, while LINC01391 knockdown shows the opposite results [21]. Overexpressed DLX6-AS1 in GC signi cantly promotes cell viability and colony formation, and then regulates aerobic glycolysis by targeting phosphoinositidedependent protein kinase 1 (PDK1) [22]. These results consistently emphasized the regulation of lncRNAs for GC aerobic glycolysis.
To further validate the biological function of lncRNA on GC, we utilized the MeRIP-Seq and found that there was a potential m 6 A site in the 3'-UTR of DLGAP1-AS2, suggesting the m 6 A modi cation for DLGAP1-AS2. The speci c m 6 A modi cation on DLGAP1-AS2 could signi cantly modulate its fate. In our research, we found that the m 6 A methyltransferase METTL3 could speci cally bind with the m 6 A site on the DLGAP1-AS2. Besides, METTL3 could assist the stability increasing for DLGAP1-AS2. DLGAP1-AS2, on the other hand, connected with its downstream targets through m 6 A binding. Mechanistically, we found that m 6 A-modi ed DLGAP1-AS2 functions through a METTL3-involved m 6 A mechanism. DLGAP1-AS2 integrated with c-Myc mRNA via the m 6 A-mediated combination, constructing a DLGAP1-AS2/m 6 A/YTHDF1/c-Myc mRNA axis.
On this basis of m 6 A-DLGAP1-AS2, we realize the importance that m 6 A-modi ed lncRNA provides an important regulatory pattern for human cancer. For instance, lncRNA LINRIS is upregulated in colorectal cancer, and LINRIS knockdown attenuates the degradation of IGF2BP2 and especially MYC-mediated glycolysis [23]. For another example, m 6 A is highly enriched on the lncRNA THOR transcripts, including GAACA, GGACU and UGACU motifs. Besides, m 6 A-modi ed lncRNA THOR regulates the cancer cells' proliferation via an m 6 A-reader-dependent manner [24]. In colorectal cancer, m 6 A-induced lncRNA RP11 triggers the progression and metastasis via post-translational Zeb1 upregulation through RP11/hnRNPA2B1/mRNA complex [25]. These results consistently demonstrated the m 6 A-lncRNA interaction regulation for human cancer.

Conclusion
In summary, our ndings in this study revealed that DLGAP1-AS2 promoted the tumorigenesis of GC through m 6 A/YTHDF1/c-Myc mRNA axis, thus triggering c-Myc mRNA stability and accelerating the aerobic glycolysis of GC cells. These m 6 A-dependent RNA-protein interactions could maintain the oncogenic characteristic of lncRNA DLGAP1-AS2. Based on the ndings, DLGAP1-AS2 may serve as potential target in the GC treatment.

Declarations
Ethics approval and consent to participate This study was approved by the Fujian Provincial Hospital Ethics Review Committee. Informed consent was written by each patient before the study.

Consent for publication
All authors agree with the publication.
Availability of data and materials Not applicable.

Competing interests
All authors declare no con icts of interest Funding Not applicable.

Authors' contributions
Changshun Yang and Yu Zhang performed the assays and were responsible for the writing. Xuefei Cheng was responsible for the statistic. Weihua Li was responsible for the funding and design.