Methylated lncRNAs suppress apoptosis of gastric cancer stem cells via the lncRNA–miRNA/protein axis

Background Long noncoding RNAs (lncRNAs) play essential roles in the tumorigenesis of gastric cancer. However, the influence of lncRNA methylation on gastric cancer stem cells (GCSCs) remains unclear. Methods The N6-methyladenosine (m6A) levels of lncRNAs in gastric cancer stem cells were detected by methylated RNA immunoprecipitation sequencing (MeRIP-seq), and the results were validated by MeRIP-quantitative polymerase chain reaction (qPCR). Specific sites of m6A modification on lncRNAs were detected by single-base elongation- and ligation-based qPCR amplification (SELECT). By constructing and transfecting the plasmid expressing methyltransferase-like 3 (METTL3) fused with catalytically inactivated Cas13 (dCas13b) and guide RNA targeting specific methylation sites of lncRNAs, we obtained gastric cancer stem cells with site-specific methylation of lncRNAs. Reverse transcription (RT)-qPCR and Western blot were used for detecting the stemness of treated gastric cancer stem cells. Results The site-specific methylation of PSMA3-AS1 and MIR22HG suppressed apoptosis and promoted stemness of GCSCs. LncRNA methylation enhanced the stability of PSMA3-AS1 and MIR22HG to suppress apoptosis of GCSCs via the PSMA3-AS1–miR-411-3p– or MIR22HG–miR-24-3p–SERTAD1 axis. Simultaneously, the methylated lncRNAs promoted the interaction between PSMA3-AS1 and the EEF1A1 protein or MIR22HG and the LRPPRC protein, stabilizing the proteins and leading to the suppression of apoptosis. The in vivo data revealed that the methylated PSMA3-AS1 and MIR22HG triggered tumorigenesis of GCSCs. Conclusions Our study revealed the requirement for site-specific methylation of lncRNAs in the tumorigenesis of GCSCs, contributing novel insights into cancer development. Supplementary Information The online version contains supplementary material available at 10.1186/s11658-024-00568-8.

Another hypothesis, which suggests that tumorigenesis results from the infinite proliferation of cancer stem cells, has become the dominant doctrine in recent years [4].Cancer stem cells (CSCs), a subclass of neoplastic cells within tumors, possess the ability to selfrenew and undergo aberrant differentiation [5].Over the past decades, there has been an increasing interest in cancer stem cells, because CSCs are not only responsible for tumor initiation, propagation, and maintenance, but also contribute to chemotherapy resistance and relapse of tumors [5,6].Therefore, exploring the factors that affect the biological activities especially the differentiation of CSCs becomes urgently required.It is found that some signaling pathways, such as the Hedgehog, Notch, Wnt/β-catenin, and NF-κB, contribute to the enhancement of the self-renewal and differentiation abilities of CSCs [7].The dysregulation of the genes of these signaling pathways has great effects on cancer stem cells.Long noncoding RNAs (lncRNAs), acting as regulatory factors in gene expression, play pivotal roles in the cellular activities of CSCs [8][9][10].However, the epigenetic regulatory mechanisms of RNAs, especially lncRNAs, have not been extensively explored.
To address this issue, the differences of lncRNA m6A modification between gastric cancer stem cells (GCSCs) and gastric cancer nonstem cells (GCNSCs), which were isolated from patients with gastric cancer (GC), was characterized in this study.The findings unveiled that the site-specific m6A modification of lncRNAs PSMA3-AS1 and MIR22HG promotes tumorigenesis in GCSCs by suppressing apoptosis and promoting proliferation through increased stability of miRNAs and proteins..

Sorting and culture of GCSCs and GCNSCs
GCSCs were sorted from the solid tumors of a patient with GC.The solid tumors were washed by phosphate buffered saline (PBS) and were subsequently sliced into 1 mm 3 fragments.Following this, they were subjected to a 5 h digestion using collagenase I (Gibco).The mixture was filtrated through a 200 mesh filter and then centrifuged at 100 × g for 10 min.Then the cell suspension was cultured in serum-free Dulbecco's modified Eagle's medium (DMEM)/F12 medium (Gibco) containing 2% B27 (Invitrogen), 5 μg/ml insulin (Sigma), 20 ng/ml basic fibroblast growth factor (bFGF; FSigma), 20 ng/ml epidermal growth factor (EGF; Sigma), and 1% penicillin-streptomycin solution (Gibco).After culture for 1 week, the cells derived from tumorspheres were identified as GCSCs.Tumorsphere formation assay and western blotting for stemness gene were used for identification of the separated GCSCs.The residual cells, which failed to form tumorspheres were regarded as GCNSCs and cultured in DMEM basic medium (Gibco) with 10% fetal bovine serum (FBS; Gibco) and 1% penicillin-streptomycin solution (Gibco).
Isolation of stem cells from the GC cell lines HGC-27 and MGC-803 has been described in our previous report [25].Briefly, ALDEFLUOR Kit (Cyagen) was employed to isolate GCSCs, which was based on the expression of ALDH1, a recognized marker for GCSCs.Cells suspended in buffer with ALDH1 fluorescent substrate BODIPY-aminoacetate (BAAA, 1 μmol/L) or diethylaminobenzaldehyde (DEAB), a specific inhibitor of ALDH1, were subjected to fluorescence activated cell sorting (FACS).

Western blot
Proteins was released using lysis buffer with protease inhibitors, and were then loaded onto sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gels for separation.Next, the proteins were transferred onto a nitrocellulose membrane, which was successively blocked with 5% nonfat milk.Incubation with primary antibodies (Abcam) against the target proteins overnight at 4 °C, followed by washing and incubation with horseradish peroxidase (HRP)-conjugated secondary antibodies (Jackson ImmunoResearch).The chemiluminescence signal was examined using ECL kit (PerkinElmer).

Tumorsphere formation assay
A single cell was cultured in a 96-well ultra-low attachment microplate (Corning) for 15 days, and the status of cells was recorded by light microscope.At various time points after culture, the percentage of tumorsphere formation was calculated.

Methylated RNA immunoprecipitation sequencing (MeRIP-seq)
After removal of rRNAs by RiboMinus ™ Eukaryote Kit v2 (Invitrogen), total RNAs were interrupted into ~ 100 nt by RNA fragmentation reagents (Invitrogen).Subsequently the RNAs were incubated with anti-m6A antibody (Abcam) and Protein A/G Magnetic Beads (Thermo Scientific) at 4 ℃ overnight.The RNA fragments bound with beads were eluted and subjected to RNA sequencing or reverse transcription-quantitative polymerase chain reaction (RT-qPCR).

Transwell migration assay
Cell migration was assessed using a Transwell ® 24-well Permeable Support (Corning, USA).Cells were seeded in serum-free medium and cultured for 24 h.Then, 5 × 10 4 cells were placed into the upper chamber of the Transwell insert (8.0 µm pore size).The lower chamber was filled with medium containing 20% serum as a chemoattractant.
After incubation at 37 °C for 24 h, cells that had not migrated through the pores were removed from the upper side of the Transwell membrane with a cotton swab.The cells on the lower side of the membrane were fixed with 4% paraformaldehyde, stained with 0.1% crystal violet, and then imaged using a light microscope.Migrated cells were quantified by counting the cells in five fields per membrane under 10 × magnification, and the mean number of cells was calculated.

Detection of chemoresistance of cancer cells
Chemoresistance of cancer cells was assessed using MTT assay.Briefly, cells were seeded into 96-well plates and cultured overnight.The following day, cells were treated with cisplatin at different final concentrations (10,25,50, and 100 μg/mL).After 48 h of treatment, cell viability was assessed according to the protocol of the cell viability assay.The half maximal inhibitory concentration (IC 50 ) values were calculated from the doseresponse curves for each cell line.
To rescue the expression of METTL3 or METTL14 in the METTL3-silenced or the METTL14-silenced stem cells, the METTL3 gene or the METTL14 gene was amplified using primers (METTL3, 5′-CCC AAG CTT TTC GAG AGG TGT CAG GGC TGG G-3′ and 5′-GCGG CCG CTA AAC TAT CAG AGC CAT GGC TAT GGA TTCT; METTL14, 5′-CCC AAG C TTT CTG TTC GTA AGC TCC CGG TGA -3′ and 5′-GGA TCC CGA CAA CTG CAA GC AAG GCT CGCT-3′) and inserted into pcDNA3.1 + vector.To avoid the targeting of recombinant METTL3 or METTL14 by METTL3-shRNA or METTL14-shRNA, the METTL3 gene or the METTL14 gene was mutated from C to T at position 1278 or from C to T at position 254.Then METTL3-knockdown or /METTL14-knockdown GCSCs were subjected to transfection with the recombinant plasmids.

Examination of lncRNA stability
To assess the stability of lncRNAs, cells were exposed to actinomycin D (Sigma) at a final concentration of 80 nM.Cells were collected at different time points (0, 2, 4, 8, and 12 h) after treatment, followed by total RNA extraction and RT-qPCR.

Dot blot
Total RNAs extracted were heated at 98 °C for 3 min.After being chilled on ice immediately thereafter, RNAs were dropped onto a nitrocellulose membrane, followed by crosslink under ultraviolet.The membrane was incubated with blocking buffer (PBS, 0.02% Tween-20, 5% milk, pH7.4) for 1 h at room temperature.Successively, anti-m6A antibody was incubated with the membrane overnight at 4 °C.After washing with PBST, the membrane was incubated with the HRP-conjugated secondary antibody.The signals were examined using ECL kit (PerkinElmer).

Dual-luciferase reporter assay
The recombinant pmirGLO vector inserted fragments of lncRNAs and synthesized miR-NAs (GenePharma) were transfected together into GCSCs using Lipofectamine 2000.The scrambled sequence of a miRNA was used as a control.The luciferase activities were measured using the dual luciferase reporter assay kit (Vazyme) at 36 h post transfection.

Cell viability assay
At various time points posttreatment, cells were collected by centrifuging.Next, collected cells were exposed to CellTiter 96 ® AQueous reagent (Promega, USA) and incubated at 37 °C for 2 h.The iMark Microplate Absorbance Reader (Bio-Rad) was used for detection of the absorbance of cells.

Cell cycle analysis
Cells were incubated with prechilled 70% ethanol at 4 °C for 30 min, and then incubated with RNase A (Beyotime) at 37 °C for 20 min.Finally, incubation with PI (Yeasen) at 37 °C for 10 min was used for cell staining.The fluorescence intensity was measured using flow cytometry.

Tumorigenicity assay in vivo
GCSCs with the site-specific methylation of PSMA3-AS1 or MIR22HG were constructed and then injected into mice.The guide RNA targeting PSMA3-AS1 (5′-G TAC AGG TTA TCT CAG AAT ATC TTT TTT GGC -3′) or MIR22HG (5′-GAT ATT GG GTC TTA TTT TTT TCT GTT ATGGT-3′) was cloned into the pLVX-IRES-Neo lentiviral vector (Addgene) behind the cPPT site.METTL3 was also cloned into the lentiviral vector as a fusion protein of Cas13b.The recombinant lentiviral vector was transfected into HEK293T cells to package the virus.At 48 h later, the viral suspension was collected for transfection into METTL3-silenced GCSCs and cultured for 24 h.The cells were then screened using G418 sulfate (Thermofisher) for 24 h.The living cells were collected and cultured for 48 h.Subsequently, 1 × 10 6 cells were mixed with matrigel and subcutaneously injected into five female BALB/c mice at a volume of 100 μL per mouse.After 4 weeks, the mice were sacrificed and the solid tumors were dissected for subsequent use.

Statistical analysis
All experiments were replicated a biological minimum of three times.One-way analysis of variation (ANOVA) and Student's t-test were used to analyze the numerical data.

N6-methyladenosine lncRNAs in GCSCs
To reveal the relationship between the stemness of GCSCs and m6A modification of lncRNAs, the m6A-modified lncRNAs of GCSCs sorted from the solid tumors of a GC patient were characterized.The results indicated a tumorsphere formation rate of 97.9% for the identified potential GCSCs, whereas no tumorspheres were observed in the case of the GCNSCs (Fig. 1A).Correspondingly, a single cell from a tumorsphere of potential GCSCs had the capability to form another tumorsphere (Fig. 1B).Western blot analysis demonstrated a notable upregulation of some stemness genes in GCSCs (Fig. 1C).The migration efficiency and the chemoresistance capacity of GCSCs were significantly increased compared with those of GCNSCs (Fig. 1D and E).These data confirmed a successful isolation of GCSCs from solid tumors.
The results of dot blot using m6A-specific antibody revealed a considerably elevated m6A level in RNAs within GCSCs compared with GCNSCs (Fig. 1F), suggesting that the m6A modification of RNAs play a crucial role in GCSCs.Therefore, the m6A-modified lncRNAs of GCSCs were sequenced.On the basis of MeRIP-seq, the m6A level of 128 lncRNAs were elevated in GCSCs, yet the m6A level of 1 lncRNA was decreased in GCSCs (Fig. 1G).The MeRIP-seq data showed that the m6A levels of six lncRNAs (PAPPA-AS1, PSMA3-AS1, MIR22HG, LINC00342, LINC01410, and LINC00680) were significantly upregulated in GCSCs compared with those in GCNSCs (Fig. 1G).To confirm the expression profiles of the methylated lncRNAs in GCSCs, the m6A levels of these six lncRNAs were analyzed by MeRIP-qPCR.The results indicated that the six lncRNAs were significantly upregulated in GCSCs compared with the control (Fig. 1H), confirming the MeRIP-seq data.
Collectively, the findings unveiled a substantial increase in m6A levels of lncRNAs within GCSCs compared with those in GCNSCs, implying the essential roles of m6A modification of lncRNAs in GCSCs.

Mechanism of lncRNA methylation in GCSCs
To elucidate the function of methylated lncRNAs in GCSCs, the GCSCs were further sorted from HGC-27 and MGC-803.Using flow cytometry, we sorted ALDH1-positive cells from HGC-27 and MGC-803, and the tumorsphere formation assay demonstrated that a tumorsphere could be formed by a single cell derived from the sorted ALDH1positive cells of HGC-27 or MGC-803 (Fig. 2A).Meanwhile, Western blot analysis showed that the genes related to stemness were upgraded in the ALDH1-positive cells (Fig. 2B).The capacities of migration and chemoresistance of the ALDH1-positive cells were significantly increased compared with those of the control (Fig. 2C, D).These results showed a successful isolation of GCSCs from HGC-27 and MGC-803.
To reveal the mechanism of lncRNA methylation in GCSCs, the enzymes responsible for lncRNA methylation, including METTL3 and METTL14, the key methyltransferases for the m6A modification of mRNAs and lncRNAs, were characterized in GCSCs, GCSC-HGC, and GCSC-MGC.Western blot analysis revealed a substantial upregulation of METTL3 and METTL14 in three distinct GCSC types (Fig. 2E), suggesting the significant roles of METTL3 and METTL14 in the GCSCs.
To evaluate the impact of METTL3/METTL14 on GCSCs, the protein levels of stemness genes were assessed in GCSCs with METTL3/METTL14-shRNA or METTL3/METTL14-shRNA rescue by western blot analysis.The findings demonstrated a significant reduction in protein levels of those genes in GCSCs with METTL3/METTL14-shRNA, whereas the expression levels of those genes in the METTL3/METTL14-shRNA rescued GCSCs were similar to those of the controls (Fig. 2H, K).These findings suggested that METTL3/METTL14 enhanced the stemness of GCSCs.
In summary, METTL3 and METTL14 have demonstrated their crucial roles in lncRNA methylation of GCSCs, which could promote the stemness of GCSCs.

Roles of methylated lncRNAs in GCSCs
On the basis of the MeRIP-Seq data, six lncRNAs (PAPPA-AS1, PSMA3-AS1, MIR22HG, LINC00342, LINC01410, and LINC00680) had a high m6A level in GCSCs.To further Fig. 2 continued elucidate the functions of methylated lncRNAs in GCSCs, the expression profiles of six lncRNAs were investigated.The results showed that two lncRNAs (PSMA3-AS1 and MIR22HG) were substantially elevated in all GCSCs compared with the corresponding GCNSCs (Fig. 3A), suggesting the crucial roles played by these lncRNAs in GCSCs.
To evaluate the impact of lncRNAs on GCSCs, PSMA3-AS1 or MIR22HG was knocked down in GCSCs, followed by examining the stemness of knockdown GCSCs.The RT-qPCR data demonstrated that PSMA3-AS1 and MIR22HG were respectively silenced in three types of GCSCs (Fig. 3B).The silencing of PSMA3-AS1 or MIR22HG caused a significant downregulation of the stemness markers in GCSCs (Fig. 3C), showing the important roles of PSMA3-AS1 or MIR22HG in GCSCs.And the results of the cell viability assay showed that the PSMA3-AS1 or MIR22HG downregulation significantly reduced the viability of GCSCs (Fig. 3D).The suppression of GCSCs' viability mediated by the PSMA3-AS1 or MIR22HG silencing resulted from cell cycle arrest in the G0/G1 phase, subsequently leading to apoptosis of GCSCs (Fig. 3E, F).These findings indicated that PSMA3-AS1 and MIR22HG exert positive influences on GCSCs.
To investigate the impact of m6A modification on lncRNAs in GCSCs, we initially predicted the m6A sites within the lncRNAs PSMA3-AS1 and MIR22HG by SRAMP (http:// www.cuilab.cn/ sramp).The analysis revealed the presence of 12 potential m6A sites in PSMA3-AS1 and 10 in MIR22HG (Fig. 3G).According a previous report, single-base elongation-and ligation-based qPCR amplification analysis (SELECT) was performed to determine the m6A sites of lncRNAs [27].The data of qPCR analysis confirmed that 11 and 8 of the predicted m6A sites of PSMA3-AS1 and MIR22HG were methylated in GCSCs, respectively (Additional file 1: Fig. S1A).Among the m6A sites of lncRNAs, the N6-methyladenosine modification level at position 1225 of PSMA3-AS1 (m6A1225) or at position 2041 of MIR22HG (m6A2041) was significantly increased in GCSCs compared with GCNSCs (Fig. 3H), implying that the site-specific methylated adenosine of PSMA3-AS1 and MIR22HG play a significant role in GCSCs.
To elucidate the impact of the site-specific m6A modification of lncRNAs on GCSCs, the A1225 site of PSMA3-AS1 and the A2041 site of MIR22HG were specifically methylated in the METTL3-silenced GCSCs and then the methylation on those lncRNAs were confirmed (Fig. 3I).Meanwhile, the results of the cell viability assay indicated that silencing of METTL3 obviously impaired the viability of GCSCs, while the methylation of A1225 of PSMA3-AS1 (METTL3-shRNA + methylated  PSMA3-AS1) or A2041 of MIR22HG (METTL3-shRNA + methylated MIR22HG) significantly increased the viability of METTL3-silenced GCSCs compared with METTL3-silenced GCSCs (METTL3-shRNA) (Fig. 3J), implying that the site-specific m6A modification of lncRNAs could promote the proliferation of GCSCs.The data of cell cycle analysis revealed that GCSCs with METTL3-shRNA induced cell cycle arrest in G1/G0 phase, and the methylation of A1225 of PSMA3-AS1 or A2041 of MIR22HG recovered the cell cycle of GCSCs (Fig. 3K).Correspondingly, the METTL3 Fig. 3 continued knockdown-mediated cell cycle arrest activated apoptosis of GCSCs, whereas the apoptotic activity of METTL3-silenced GCSCs with the methylation of A1225 of PSMA3-AS1 (METTL3-shRNA + methylated PSMA3-AS1) or A2041 of MIR22HG (METTL3-shRNA + methylated MIR22HG) was commensurate with that of the control (METTL3-shRNA-scrambled; Fig. 3L).Evidently, the site-specific m6A modification of lncRNAs was required for the proliferation of GCSCs.
Furthermore, the western blots analysis revealed a significant elevation of stemness in METTL3-silenced GCSCs with methylation of the A1225 site of PSMA3-AS1 or the A2041 site of MIR22HG (Fig 3M).Consistently, the methylation of PSMA3-AS1 or MIR22HG significantly promoted the tumorsphere formation of the METTL3silenced GCSCs (Fig 3N and O).These data unveiled a positive contribution of lncRNA methylation to stemness of GCSCs.
Together, these data demonstrated that site-specific m6A modification of lncRNAs (PSMA3-AS1 and MIR22HG) was required for the proliferation of GCSCs via suppressing apoptosis and promoting the stemness of GCSCs.

Influence of m6A modification on the stability of lncRNAs
To investigate the effect of m6A modification on lncRNA stability, the RNA levels of PSMA3-AS1 and MIR22HG in METTL3-silenced GCSCs were determined.A considerable decrease of the RNA levels of PSMA3-AS1 and MIR22HG were exhibited in METTL3-silenced GCSCs (Fig. 4A), intimating that m6A modification could enhance the stability of lncRNAs.
To further validate the influence of m6A modification on lncRNAs stability, the METTL3-silenced GCSCs were exposed to actinomycin D, followed by the assessment of lncRNAs enrichment.The findings indicated elevated expression levels of PSMA3-AS1 and MIR22HG in METTL3-silenced GCSCs treated with actinomycin D Fig. 3 continued than those in GCSCs (Fig. 4B).It can be inferred that m6A modification could stabilize PSMA3-AS1 and MIR22HG in GCSCs.
To elucidate the roles of miR-411-3p and miR-24-3p in GCSCs, we examined the expression profiles of these miRNAs in both GCSCs and GCNSCs.The findings indicated an obvious upregulation of miR-411-3p and miR-24-3p in GCSCs (Fig. 5C), suggesting that these two miRNAs played important roles in GCSCs.
To further explore the roles of the miRNAs interacting with lncRNAs, the potential targets of miRNAs were predicted using microT, miRmap, and TargetScan algorithms.The results showed that four genes were the potential targets of miR-411-3p, while ten genes might be targeted by miR-24-3p (Fig. 5D).To confirm the target genes of miRNAs, the change of expression of target genes were examined after transfecting with miR-411-3p or miR-24-3p in GCSC.Upon overexpression of miR-411-3p or miR-24-3p in GCSC (Fig. 5E), the expression level of AP1G1 or TRPC4AP and SERTAD1 was substantially diminished (Fig. 5F), implying that miR-411-3p or miR-24-3p might target AP1G1 or TRPC4AP and SERTAD1.And the dual luciferase The expression profiles of miR-411-3p and miR-24-3p in GCSCs and GCNSCs.D Prediction of potential target genes of miRNAs using microT, miRmap, and TargetScan.E Validation for overexpression of miRNAs in GCSCs.F Detection of the expression levels of target genes in the miRNA-overexpressed GCSCs.G Validation of interaction between miRNAs and target genes by dual-luciferase reporter assay.H The expression levels of target genes of miRNAs in GCSCs and GCNSCs.I Western blot for target genes of miRNAs in GCSCs and GCNSCs.J Caspase-3/7 assay for AP1G1 or SERTAD1 silencing GCSCs.K Western blot for stemness genes in AP1G1 or SERTAD1 silencing GCSCs.L The expression level of AP1G1 or SERTAD1 in PSMA3-AS1 or MIR22HG silencing GCSCs.M Western blot analysis for target genes of miRNAs in PSMA3-AS1 or MIR22HG-silenced GCSCs.**, p < 0.01 assays were performed and the data demonstrated that AP1G1 or SERTAD1 was the gene targeted by miR-411-3p or miR-24-3p (Fig. 5G).
To explore whether the target genes of miRNAs had effects on GCSCs, the expression levels of AP1G1 and SERTAD1 in GCSCs and GCNSCs were assessed.The findings indicated a remarkable upregulation of AP1G1 and SERTAD1 in GCSCs (Fig. 5H and I), suggesting that AP1G1 and SERTAD1 might play essential roles in GCSCs.The AP1G1 or SERTAD1 silencing induced apoptosis of GCSC (Fig. 5J) and downregulated the protein abundance of stemness genes in GCSC (Fig. 5K), showing that AP1G1 and SERTAD1 were required for shaping GCSC characteristics.As reported, AP1G1 and SERTAD1 promote gastric cancer progression [28,29].In this context, AP1G1 and SERTAD1 could promote tumorigenesis of GCSCs.
The m6A sites of lncRNAs PSMA3-AS1 and MIR22HG were far from the base sequences bound to the seed sequences of miRNAs.Thus, the m6A modification of lncRNAs did not affect the interaction between PSMA3-AS1 and miR-411-3p, similarly to MIR22HG and miR-24-3p.To assess whether PSMA3-AS1 or MIR22HG functioned as a ceRNA, the expression profiles of AP1G1 or SERTAD1 were determined in the PSMA3-AS1 or MIR22HG-silenced GCSCs.The findings revealed that the PSMA3-AS1 or MIR22HG silencing noticeably decreased the expression level of miRNA-targeted mRNA of AP1G1 or SERTAD1 (Fig. 5L and M), indicating that lncRNAs PSMA3-AS1 and MIR22HG functioned as ceRNAs in GCSCs.

Interactions between m6A-modified lncRNAs and proteins
To reveal the proteins that interacted with lncRNAs, lncRNA pulldown assays were performed.The results exhibited that a specific protein from GCSCs, identified to be EFF1A1 or LRPPRC, was bound to PSMA3-AS1 or MIR22HG (Fig. 6A).However, the protein level from the METTL3-silenced GCSCs was lower than that from GCSCs (Fig. 1A), connoting that the methylation of PSMA3-AS1 or MIR22HG affected the EFF1A1-PSMA3-AS1 and LRPPRC-MIR22HG interactions.Western blots confirmed the interaction between EFF1A1 and PSMA3-AS1 or LRPPRC and MIR22HG and the lncRNA-protein interaction was weakened by lncRNA demethylation (Fig. 6B).These data indicated that lncRNA methylation promoted the interaction between lncRNA and protein.
To unveil the impact of lncRNA methylation on the proteins bound to lncRNAs, protein stability was detected and the findings showed that PSMA3-AS1 or MIR22HG silencing decreased the EFF1A1 or LRPPRC protein level in GCSCs compared with the controls (Fig. 6C).However, PSMA3-AS1 or MIR22HG silencing had no impact on the EFF1A1 or LRPPRC mRNA level (Fig. 6C).These data demonstrated that the methylation of lncRNAs enhanced the stability of the proteins bound to PSMA3-AS1 or MIR22HG.
To investigate the roles of EEF1A1 and LRPPRC in GCSCs, the differential expression of EEF1A1 or LRPPRC in GCSCs and GCNSCs was evaluated.The results showed that at both RNA and protein levels, abundance of EEF1A1 and LRPPRC were considerably elevated in GCSCs (Fig. 6D), suggesting that EEF1A1 and LRPPRC played important roles in GCSCs.To reveal the function of EEF1A1 and LRPPRC in GCSCs, EEF1A1 and LRPPRC were knocked down in GCSCs (Fig. 6E).EEF1A1 or LRPPRC silencing led to the inhibition of GCSCs' viability (Fig. 6F) and caused cell cycle arrest in G0/G1 phase of GCSCs, which triggered apoptosis of GCSCs (Fig. 6G, H).Consistently, the expression of stemness genes were impaired in EEF1A1-silenced or LRPPRC-silenced GCSCs (Fig. 6I).These data revealed the influential role of EEF1A1 and LRPPRC in maintaining stemness and proliferation of GCSCs.
Taken together, these results implied that the methylated PSMA3-AS1 or MIR22HG was interacted with EFF1A1 or LRPPRC protein to enhance the protein stability, leading to the suppression of apoptosis and the promotion of stemness of GCSCs.

Influence of m6A modification of lncRNAs on tumorigenesis of GCSCs in vivo
To investigate the influence of methylated PSMA3-AS1 and MIR22HG on tumorigenesis of GCSCs in vivo, METTL3-silenced GCSC with or without the m6A modification of PSMA3-AS1 or MIR22HG were injected into BALB/c mice (Fig. 7A).The tumor volume exhibited a more rapid growth in the mice injected with METTL3-silenced GCSCs containing the methylated PSMA3-AS1 or MIR22HG than mice without the lncRNA methylation (Fig. 7B).Uniformly, much bigger and heavier tumors were developed in the mice with methylated PSMA3-AS1 or MIR22HG (Fig. 7C and D).Thus, it can be inferred that m6A modification of PSMA3-AS1 or MIR22HG prominently contributes to the promotion of tumorigenesis in GCSCs in vivo.
Then m6A level of lncRNAs were detected for confirming the methylation of lncR-NAs in solid tumors (Fig. 7E).To assess the function of lncRNA methylation on solid tumors, the apoptosis and proliferation of solid tumors were examined using immunohistochemical analysis.The immunohistochemical analysis data exhibited that Caspase 3 was dramatically reduced in the solid tumors of mice with methylated PSMA3-AS1 or MIR22HG, whereas Ki67 was upregulated (Fig. 7F), implying that the methylation of lncRNAs promoted tumor growth by suppressing apoptosis in vivo.Furthermore, the expression of stemness genes in the tumors of mice with the treatment of methylated PSMA3-AS1 or MIR22HG were upregulated (Fig. 7G).These results illustrated the positive contribution of lncRNA methylation to tumor progression in vivo.
In conclusion, these results underscore the potential of m6A modification in lncR-NAs (PSMA3-AS1 and MIR22HG) to drive tumorigenesis in GCSCs through both miRNA-mediated and protein-mediated pathways (Fig. 7H).

Discussion
The dysregulation of lncRNAs has been increasingly recognized as a significant contributor to tumorigenesis [30].Although the precise mechanisms through which lncRNAs exert their functions are still being elucidated, it is established that lncRNAs play a role in modulating gene expression at diverse levels [31,32].Comprehending the complex interplay between lncRNAs and tumorigenesis is essential for unraveling the underlying mechanisms driving cancer progression [32].Owing to the importance of lncRNAs, the epigenetic modifications of lncRNAs, including RNA methylation, have emerged as crucial mechanisms modulating lncRNAs' roles in tumorigenesis, shedding light on the regulatory approaches that contribute to cancer pathogenesis [33,34].The methylation of lncRNAs, specifically m6A, is a prominent posttranscriptional modification of lncRNAs [33,34].Accumulating evidence highlights the profound influence of m6A modification of lncRNAs on the regulation of the stability, subcellular localization, and interactions with miRNAs and RNA-binding proteins [33,34].The m6A modification of lncRNAs can facilitate the recognition by specific reader proteins, leading to either enhanced or reduced stability, thereby shaping lncRNA abundance and availability [35].The m6A modification landscape can impact the transport and spatial distribution of lncRNAs within distinct cellular compartments, orchestrating their interactions with target genes or proteins [35].However, the involvement of m6A modification in lncRNA-mediated tumorigenesis remains largely unexplored.In this study, based on the global screening and characterization of the methylated lncRNAs in GCSCs, the site-specific m6A modification was found in two lncRNAs PSMA3-AS1 and MIR22HG of GCSCs.The methylated PSMA3-AS1 and MIR22HG, which were upregulated in GCSCs sorted from the solid tumors of a patient with gastric cancer and two distinct gastric cancer cell lines (HGC-27 and MGC-803), suppressed apoptosis and promoted the stemness of three types of GCSCs, thus promoting tumorigenesis.In this context, our findings revealed the requirement of the site-specific methylation of lncRNAs in tumorigenesis, shedding light on potential therapeutic targets for gastric cancer treatment.
Numerous studies have provided compelling evidence to demonstrate the popular roles of lncRNAs as ceRNAs in cancer development [36].In this study, the results indicated that the site-specific m6A modification of PSMA3-AS1 and MIR22HG enhanced the stability of lncRNAs in GCSCs, thus having a promotive role on PSMA3-AS1 and MIR22HG as ceRNAs in tumorigenesis of gastric cancer.Accordingly, our discoveries unveiled an innovative mechanism of lncRNA-mediated ceRNAs.In our study, it was found that the methylated but not the unmethylated PSMA3-AS1 or MIR22HG could interact with EEF1A1 or LRPPRC protein in GCSCs.As reported, lncRNAs can exhibit diverse functional roles in the process of interactions with proteins, acting as protein decoys to recruit or sequester proteins, or functioning as scaffolds to facilitate the assembly of protein complexes [37,38].Our study revealed that lncRNA methylation facilitated the interaction between lncRNA and protein to increase protein stability and further promoted tumorigenesis of GCSCs.In this context, our discoveries provide novel insights into the methylation of lncRNAs, emphasizing the significance of m6A modification in tumorigenesis of GCSCs.It should be noted that the mechanism was validated in three gastric cancer cell lines, including one sample from a patient with gastric cancer and two different typical gastric cancer cell lines.Given the heterogeneity of cancer cells, the conclusions we summarized require further validation with additional samples from patients with gastric cancer.In addition to gastric cancer, the roles of methylated PSMA3-AS1 and MIR22HG in other cancers merits further exploration in future.

Conclusions
Our findings demonstrated that the methylation level of PSMA3-AS1 and MIR22HG, mediated by METTL3 and METTL14, was significantly elevated in GCSCs.This study revealed that the site-specific m6A modification of PSMA3-AS1 and MIR22HG suppressed apoptosis and promoted the stemness of GCSCs.LncRNA methylation increased the stability of PSMA3-AS1 and MIR22HG to promote proliferation and suppress apoptosis of GCSCs via the PSMA3-AS1-miR-411-3p-AP1G1 or MIR22HG-miR-24-3p-SERTAD1 axis.At the same time, the methylation of PSMA3-AS1 or MIR22HG promoted the interaction between PSMA3-AS1 and EEF1A1 protein or MIR22HG and LRPPRC and stabilized the interacted proteins, leading to the suppression of apoptosis and the promotion of proliferation of GCSCs.

Fig. 1 Fig. 2
Fig. 1 N6-methyladenosine lncRNAs in GCSCs.A Percentage of tumorsphere formation of GCSCs and GCNSCs.B Tumorsphere formation assay of a single cell from GCSCs; C detection of the expressions of stemness genes in GCSCs and GCNSCs with Western blot.D Efficiency of the migration of GCSCs.GCSCs and GCNSCs were subjected to Transwell migration assays to examine the cell migration.The representative images of the migrated cells are shown on the left.The percentage of the migrated cells is indicated on the right (**, p < 0.01).Scale bar, 50 μm.(E) Chemoresistance capacity of GCSCs.GCSCs or GCNSCs were treated with different concentrations of cisplatin.At 48 h after treatment, the cells were subjected to cell viability assays (**, p < 0.01).The half maximal inhibitory concentration (IC 50 ) values were calculated.F The m6A modification of RNAs in GCSCs and GNCSCs.G Heatmap of the differentially expressed m6A-modified lncRNAs in GCSCs and GNCSCs.H Expression profiles of methylated lncRNAs in GCNSCs and GCSCs.**, p < 0.01

(Fig. 3
See figure on next page.)Roles of methylated lncRNAs in GCSCs.A Expression profiles of six lncRNAs (PAPPA-AS1, PSMA3-AS1, MIR22HG, LINC00342, LINC01410, and LINC00680) in GCSCs.B Validation of silencing of lncRNAs in GCSCs.C Detection of the expressions of stemness genes in GCSCs with lncRNAs-siRNA.D Cell viability for PSMA3-AS1 or MIR22HG silencing GCSCs.E Cell cycle analysis of PSMA3-AS1 or MIR22HG-silenced GCSCs.F Caspase-3/7 assay for PSMA3-AS1 or MIR22HG silencing GCSCs.G Prediction of the methylation sites of lncRNAs.H The m6A levels of lncRNAs between GCSCs and GCNSCs.I The m6A levels of METTL3-silenced GCSCs with site-specific methylation of lncRNAs.J Cell viability for METTL3-silenced GCSCs with site-specific methylation of lncRNAs.K Cell cycle analysis of METTL3-silenced GCSCs with site-specific methylation of lncRNAs.L Caspase-3/7 assay for METTL3-silenced GCSCs with site-specific methylation of lncRNAs.M Western blot for stemness genes of METTL3-silenced GCSCs with site-specific methylation of lncRNAs.N The percentage of tumorsphere formation of METTL3-silenced GCSCs with site-specific methylation of lncRNAs.O Tumorsphere formation assay of a single cell from METTL3-silenced GCSCs with methylated lncRNAs.**, p < 0.01

Fig. 6
Fig. 6 Interactions between m6A-modified lncRNAs and proteins.A SDS-PAGE for lncRNAs pulldown assay.The arrows indicate the proteins identified.M, protein marker.B Western blot analysis of the proteins bound to lncRNAs.C Detection of the stability of proteins bound to lncRNAs in PSMA3-AS1 or MIR22HG silencing GCSCs.D The levels of mRNAs and proteins of EEF1A1 and LRPPRC in GCSCs.E Validation of silencing of EEF1A1 or LRPPRC in GCSCs by RT-qPCR and Western blot assay.F Cell viability for EEF1A1 or LRPPRC silencing GCSCs.G Cell cycle analysis of EEF1A1 or LRPPRC silencing GCSCs.H Caspase-3/7 assay of EEF1A1 or LRPPRC silencing GCSCs.I Western blot for stemness genes of EEF1A1 or LRPPRC silencing GCSCs.**, p < 0.01

Fig. 7
Fig. 7 Influence of m6A modification of lncRNAs on tumorigenesis of GCSCs in vivo.A Schematic diagram for model mice with lncRNA-methylated GCSCs.B The tumor volume of the xenograft mice injected with METTL3-silenced GCSCs alone (control) or METTL3-silenced GCSCs with methylated PSMA3-AS1 or MIR22HG.C The size of solid tumors of the mice with METTL3-silenced GCSCs alone (control) or METTL3-silenced GCSCs with methylated PSMA3-AS1 or MIR22HG.D The weight of solid tumors of mice with METTL3-silenced GCSCs alone (control) or METTL3-silenced GCSCs with methylated PSMA3-AS1 or MIR22HG.E m6A levels of lncRNAs in the solid tumors of the mice injected with METTL3-silenced GCSCs alone (control) or METTL3-silenced GCSCs with methylated PSMA3-AS1 or MIR22HG.F Immunohistochemical analysis of the expressions of Ki67 and Caspase 3 in the solid tumors of the mice injected with METTL3-silenced GCSCs alone (control) or METTL3-silenced GCSCs with methylated PSMA3-AS1 or MIR22HG.G Detection of the expressions of stemness genes in solid tumors of the mice treated with the methylated PSMA3-AS1 or MIR22HG using Western blot.H Model for the roles of the methylated PSMA3-AS1 and MIR22HG in GCSCs.**, p < 0.01