3.1 METTL3 is highly expressed in patients with gastric carcinoma and Helicobacter pylori infection contributes to the upregulation of METTL3.
To evaluate the potential involvement of the m6A methyltransferase complex in GC, we initially examined the expression of different members of the m6A writer complex in TCGA datasets. Our analysis revealed that METTL3, RBM15, WTAP, KIAA1429, and CBLL1 exhibit a significant upregulation in GC tissues compared to adjacent non-cancerous tissues (Fig. 1A and S1A). We further analyzed their correlation with the overall survival of patients with GC using the Kaplan–Meier Plotter Database and found that only METTL3 is associated with poor prognosis among these upregulated genes (Fig. 1B and S1B). Then we further investigated the expression of METTL3 in the GEO database and found that METTL3 was also upregulated in the GSE54129 cohort (Fig. 1C). Subsequently, we detected the protein levels of METTL3 in 35 pairs of GC tissues and adjacent non-tumor tissues, followed by quantification using ImageJ software. The results showed that compared with non-tumor tissues, METTL3 protein was upregulated in 74.3% of the GC tissues (Fig. 1D), and the difference was statistically significant (Fig. 1E). Given the significant association between H. pylori infection and GC, we infected GC cells with H. pylori to explore whether H. pylori infection contributes to the upregulation of METTL3 in GC cells. Our results showed that H. pylori infection increased the expression level of METTL3 in a time-dependent manner but had no significant effect on the expression of demethylases FTO and ALKBH5 (Fig. 1F).
3.2 METTL3 promotes GC cell proliferation and migration depending on its RNA binding ability
Next, we investigated the biological role of METTL3 in GC. We used CCK-8 and EdU experiments to detect cell proliferation ability and Transwell assay to test cell migration ability. Our results showed that overexpression of METTL3 promoted the proliferation and migration of GC cells (Fig. S2 A-G). In contrast, upon METTL3 knockdown with small interference RNA, the proliferative and migratory capacity of GC cells was significantly attenuated (Fig. S2 H-N), which is consistent with previous reports29, 30.
Then, we generated a mutated form of METTL3 with an impaired RNA binding domain18 to investigate its relevance to the oncogenic function of METTL3 in GC cells. Our results revealed that the promoting effect of WT-METTL3 on cell proliferation and migration can be significantly attenuated by METTL3 mutant (Fig. 1G-K). These findings strongly support the tumor-promoting effect of METTL3 in GC, which depends on its RNA binding ability.
3.3 STAT5A acts as a downstream target of METTL3.
In order to identify the downstream target genes of METTL3-mediated m6A modification, we performed MeRIP-Seq analysis to detect differentially enriched mRNAs upon knockdown of METTL3.Consistent with previous studies17, the most prevalent m6A motif GGACU exhibited significant enrichment in both negative control(NC) and METTL3 interference (si-METTL3) group (Fig. S3A). The majority of these m6A peak distributions are predominantly located within the coding sequence (CDS) and 3′ untranslated regions (UTR), particularly near the stop codon regions (Fig. S3 B, C). The KEGG pathway analysis of differentially methylated genes revealed an enrichment of cancer pathways following inhibition of METTL3 (Fig. S3 D), thereby providing further confirmation of the correlation between METTL3-mediated m6A and the occurrence and progression of GC. To identify METTL3-mediated potential downstream target genes with m6A modification that promotes GC progression, we conducted an intersection analysis between the reduced m6A modification genes upon METTL3 interference in the MeRIP-Seq assay and the differentially expressed genes identified in GC datasets from TCGA and GSE54129 (Fig. 2A). Among all 44 MeRIP-seq peaks, five potential target genes (CDCP1, TBC1D7, LPIN2, VCAN, and STAT5A) exhibited significant differential expression in TCGA and GEO datasets. We further used Kaplan-Meier Plotter to analyze the correlation between the five potential target genes and patient prognosis. As shown in Fig.S4, the high expression of TBC1D7 and LPIN2 exhibited a favorable prognosis, while CDCP1, VCAN, and STAT5A exhibited a poor prognosis. It has been reported that CDCP1 was positively regulated by METTL3 and promoted the development of GC31.Therefore, we chose VCAN and STAT5A for further verification. Our real-time quantitative PCR results showed that compared with VCAN, the expression of STAT5A exhibited higher up-regulation in response to METTL3 overexpression (Fig. 2B). Moreover, correlation analysis of the gastric tissues from the GEPIA database showed that the expression of STAT5A is positively associated with the expression of METTL3 (Fig. 2C). Therefore, we focused on STAT5A for further investigation. We found that METTL3 knockdown resulted in a reduction of STAT5A mRNA in GC cell lines MKN-45 and SGC-7901 (Fig. 2D), while overexpression of METTL3 led to an increase in the mRNA level of STAT5A, which could be rescued by mutation of METTL3 (Fig. 2E). Then, we detected whether the regulation of STAT5A mRNA by METTL3 occurs at the transcriptional level. We constructed a dual luciferase reporter gene vector, containing the STAT5A promoter fragment and transfected the plasmid together with the empty vector or METTL3 expression vector into MKN-45. As shown in Fig. 2F, overexpression of WT-METTL3 or mutated METTL3 did not change the luciferase activity. Then, we further explored whether METTL3 affected the stability of STAT5A mRNA. We transfected GC cells with the empty vector or METTL3 expression vector and then treated the cells with transcription inhibitor actinomycin D (ACTD) at different durations. The results showed that METTL3 overexpression significantly increased the stability of STAT5A mRNA and attenuated its degradation (Fig. 2G). Afterward, we performed western blot experiments to investigate the regulation of METTL3 on STAT5A at the protein level. The results showed that METTL3 knockdown significantly reduced the protein level of STAT5A (Fig. 2H), whereas the overexpression of WT-METTL3 markedly enhanced the STAT5A protein expression. However, overexpression of METTL3 mutant had no discernible effect on STAT5A expression (Fig. 2I). Next, we explored whether METTL3 can cause the m6A modification of STAT5A mRNA. We first analyzed the MeRIP-Seq data using the Integrative Genomics Viewer (IGV) software and found that the m6A peak in the CDS and 3’ UTR region of STAT5A mRNA exhibited a significant decrease following the knockdown of METTL3 (Fig. 2J). We further used MeRIP-qPCR to detect the effect of METTL3 on the m6A modification of STAT5A mRNA. The results showed that the knockdown of METTL3 resulted in a significant decrease in m6A modification of STAT5A mRNA (Fig. 2K, L). In contrast, a significant increase in the m6A level of STAT5A was observed upon overexpression of WT-METTL3, while this increase can be counteracted by the METTL3 mutant (Fig. 2M, N). To further identify the specific m6A site, we utilized the SRAMP (http://www.cuilab.cn/sramp) and RMBase 2.0 (https://rna.sysu.edu.cn/rmbase/index.php) platforms to predict potential m6A sites on STAT5A and found there were five highly plausible sites according to both programs. (Fig. S5). Three of the predicted sites are found within the CDS region (sites 140,146 and 2318), while the remaining two are in the 3’UTR region (sites 2959 and 3258). Accordingly, we performed a dual-luciferase reporter assay to assess whether the effect of METTL3 on the stability of STAT5A is achieved by causing the m6A modification in the 3’UTR region of STAT5A mRNA. The results showed that compared with the control group, the alteration in the expression level of METTL3 does not exert any influence on the luciferase reporter activity (Fig. 2O), indicating that METTL3 does not methylate the .3’UTR region of STAT5A mRNA. To further evaluate whether METTL3 methylated the STAT5A mRNA at the CDS region, we constructed different STAT5A expression vectors with predicted site mutations in the CDS region (Fig. 2P, left panel). WT-STAT5A or different STAT5A mutant (CDS-Mut1-A140G, CDS-Mut2-A146G, CDS-Mut3-A2318G), was transfected with the empty vector or METTL3 expression vector into GC cells, respectively. Western blot results showed that METTL3 overexpression remarkably increased the expressionof STAT5A in MKN-45 cells transfected with WT-STAT5A, CDS-Mut1, and CDS-Mut2 vector but did not exert a significant impact in cells transfected with CDS-Mut3 (Fig. 2P, Right panel), thereby suggesting that A2318 site on STAT5A CDS region is a potential modification site. These findings indicate that METTL3 causes the m6A modification at the A2318 site on the STAT5A CDS region.
3.4 IGF2BP2 increases the stability of STAT5A mRNA in an m6A -dependent manner
It has been reported that IGF2BPs (IGF2BP1/2/3) constitute a distinct group of m6A readers that specifically recognize the GG(m6A)C motif within mRNA transcripts,32 playing a pivotal role in augmenting the stability of targeted mRNAs through an m6A-dependent mechanism.24 Considering that METTL3-mediated m6A modification enhanced STAT5A mRNA stability, we speculate on the potential involvement of IGF2BPs as readers for m6A-modified STAT5A. We silenced IGF2BP1/2/3 to evaluate the impact on STAT5A expression using qPCR and western blot analysis. Our results showed that compared with IGF2BP1 and IGF2BP3, targeted knockdown of IGF2BP2 resulted in a more significant reduction in STAT5A expression at both the mRNA and protein levels (Fig. 3A, B). Then, we synthesized another IGF2BP2 siRNA and corroborated that IGF2BP2 siRNAs resulted in a significant reduction in STAT5A mRNA and protein level in different GC cell lines (Fig. 3C, D). On the contrary, overexpression of IGF2BP2 significantly increased STAT5A mRNA and protein levels in GC cell lines SGC-7901, MKN-45, and BGC-823 (Fig. 3E, F). Subsequently, we detected whether IGF2BP2 affects STAT5A mRNA stability. Our result demonstrated that IGF2BP2 overexpression significantly increased the half-life of STAT5A mRNA (Fig. 3G). To further investigate the interaction between STAT5A mRNA and IGF2BP2, we performed the RNA pull-down assay by using the STAT5A probe. As shown in Fig. 3H, IGF2BP2 can be pulled down by the STAT5A sense probe but not the antisense probe, indicating the binding of the STAT5A mRNA to IGF2BP2. To further investigate the specific IGF2BP2 domains that bind to STAT5A mRNA, we constructed several Flag-tagged WT and truncated IGF2BP2 expression vectors (Fig. 3I), which were subsequently transfected severally into MKN-45 for RNA pull-down assay. The results showed that The Flag-tagged WT-IGF2BP2 and the truncated IGF2BP2 mutant containing the KH3 and KH4 domain (T3 mutant) can be pulled down by the STAT5A probe, whereas the truncated mutant of IGF2BP2 containing either the RRM1-2 domain (T1 mutant) or KH1-2 domain (T2 mutant) cannot be pulled down (Fig. 3I), indicating that the KH3-4 domain of IGF2BP2 can selectively bind to STAT5A mRNA. Furthermore, The GXXG motif in the KH3 and KH4 domains of IGF2BP2 has been reported to play a key role in recognition and interaction with the target mRNA33. Therefore, to detect whether the GXXG motifs in IGF2BP2 participate in mediating the interactions with STAT5A, we introduced a GDDG mutation into the GXXG motif of either KH3 or KH4 domain of IGF2BP2, as well as both KH3 and KH4 domains, followed by conducting an RNA pull-down assay. As shown in Fig. 3J, single mutation in the KH3 or KH4 domain partially abolished the interaction between STAT5A mRNA and IGF2BP2, and the double mutations in the GxxG motif within the KH3 and KH4 domain of IGF2BP2 completely eliminated this interaction (Fig. 3J), indicating the GXXG motifs in both KH3 and KH4 domains mediated the interaction between STAT5A mRNA and IGF2BP2. We further transfected IGF2BP2 KH3/4 mutant into GC cells to detect whether IGF2BP2 KH3/4 mutant can affect the mRNA stability and protein level of STAT5A. As shown in Fig. 3K and 3L, the overexpression of the IGF2BP2 KH3/4 mutant no longer affects the mRNA stability and protein expression of STAT5A.
Subsequently, to further investigate whether the m6A modification of STAT5A mRNA is required for the interaction between IGF2BP2 and STAT5A mRNA, we performed RIP-qPCR experiments using IGF2BP2 antibody in MKN-45 cells overexpressing IGF2BP2 followed by either METTL3 knockdown or overexpression. Interestingly, we observed that IGF2BP2 overexpression led to an enhanced interaction between IGF2BP2 and STAT5A mRNA. However, the knockdown of METTL3 significantly attenuated this interaction, and overexpression of METTL3 substantially augmented the interaction (Fig. 3M), indicating that the presence of METTL3-mediated m6A methylation is crucial for the interaction between IGF2BP2 and STAT5A mRNA. Furthermore, we knocked down METTL3 in GC cells overexpressing IGF2BP2 and detected STAT5A mRNA stability and protein expression. As shown in Fig. 3N and 3O, the knockdown of METTL3 significantly abolished the IGF2BP2-mediated enhancement of STAT5A mRNA stability and protein expression. As expected, the knockdown of IGF2BP2 also abrogated the METTL3-mediated increase of STAT5A mRNA stability and protein expression (Fig. 3P, Q). These results suggest that both METTL3-mediated m6A modification and the recognition and binding of IGF2BP2 to STAT5A m6A site are crucial for the mRNA stability and protein expression of STAT5A.
3.5 STAT5A functions as an oncogenic gene in gastric carcinoma.
To further investigate the function of STAT5A in GC, we first used western blot to determine STAT5A expression in gastric immortalized epithelial cells GES-1 and four GC cell lines AGS, BGC-823, MKN-45, and SGC-7901. The results showed that the protein expression of STAT5A in immortalized GES-1cells and AGS cells was not detected by western blot assay, while it was relatively high in BGC-823 and SGC-7901 cells and showed a moderate expression level in MKN-45 cell (Fig. 4A). Therefore, AGS and MKN-45 cells were employed for the overexpression of STAT5A, while MKN-45 and SGC-7901 cells were utilized for the knockdown of STAT5A. qRT-PCR and western blot verified the overexpression efficiency (Fig. 4B, C). The results of cell proliferation experiments, including CCK-8, EdU, and colony formation ability as well as Transwell cell migration experiments indicate that overexpression of STAT5A in AGS and MKN-45 significantly promoted cell proliferation (Fig. 4D-F) and cell migration ability (Fig. 4G). In contrast, STAT5A knockdown with two distinct siRNAs inhibited cell proliferation and migration in SGC-7901 and MKN-45 cell lines (Fig. 4H-N). Altogether, these experiments provide evidence for the oncogenic role of STAT5A in GC occurrence and development.
3.6 Knockdown of STAT5A inhibits tumor growth in vivo
To further characterize the oncogenic role of STAT5A in GC, we conducted a subcutaneous implantation experiment in nude mice to assess the impact of STAT5A knockdown on the tumorigenesis ability of GC cells in vivo. The BGC-823 cells were infected with lentivirus harboring shRNA targeting STAT5A or its corresponding negative control, and the stable knockdown cells were screened and identified (Fig. 5A). The xenografts in the NC group exhibited significantly greater volume and weight compared to those in the STAT5A knockdown group (Fig. 5B-D). Western blot results confirmed a significant decrease in the expression of STAT5A protein in the tumors of the shSTAT5A group compared to that of the control group (Fig. 5E). Subsequently, to evaluate the impact of STAT5A on the metastatic potential of GC cells, we conducted a tail vein injection experiment by separately injecting the control group and shSTAT5A group of cells into the tail vein of nude mice. The mice in the shSTAT5A group exhibited a consistent upward trend in weight, whereas those in the control group displayed a significant decline starting from the 44th day in weigh (Fig. 5F). In addition, the results showed a significant decrease in both the size and weight of the lungs in mice injected with shSTAT5A cells, as compared to control cells (Fig. 5G, H). Furthermore, the STAT5A knockdown group exhibited a significantly lower incidence of lung metastasis nodules in comparison to the control group (Fig. 5I), and the HE staining assay confirmed the presence of solid tumors (Fig. 5J). In conclusion, the growth of GC cells was significantly inhibited by stable knockdown of STAT5A, and the incidence of lung metastasis was significantly decreased in the shSTAT5A group. Thus, these experiments provide evidence for the oncogenic role of STAT5A in GC tumorigenesis and metastasis.
3.7 STAT5A is involved in METTL3-mediated biological functions
We then asked whether STAT5A is required for METTL3-mediated biological function. We co-transfected the METTL3 expression vector, together with control siRNA or STAT5A siRNA into SGC-7901 and MKN-45 cell lines (Fig. 6A) and then performed functional restoration assays. The results showed that the overexpression of METTL3 led to a significant increase in cell proliferation and migration ability in SGC-7901 and MKN-45 cells, which was subsequently abolished upon depletion of STAT5A (Fig. 6B-F).
3.8 STAT5A binds to the promoter of KLF4 and regulates KLF4 expression
To further explore the downstream targets of STAT5A in GC, we performed RNA-seq in the AGS cells transfected with the STAT5A expression vector or the empty vector. The results revealed that a total of 1838 genes exhibited significantly differential expression following STAT5A overexpression in AGS cells (Fig. S6A). Subsequent enrichment analysis using GO, KEGG, and Reactome pathway gene sets demonstrated that these genes were closely associated with cell division and cell cycle (Fig. S6 B-D). Given that STAT5A acts as a transcription factor34, we utilized TRANSFAC and GTRD platforms to predict potential downstream genes regulated by STAT5A. At the same time, we screened the differentially expressed genes in GC tissues based on TCGA and GEO (GSE54129) databases. We intersected the RNA-seq results, TRANSFAC, GTRD, TCGA, and GEO databases and obtained 9 candidate genes (Fig. 7A). Since there are almost no reports on the correlation between CCDC71L gene and cancer in published literature, we conducted qRT-PCR to detect the expression of other 8 potential downstream genes following STAT5A overexpression in the AGS cells or STAT5A interference in the SGC-7901 cells. The results demonstrated that among the candidate genes, KLF4 showed the most significant change upon STAT5A overexpression or knockdown (Fig. 7B, C). We further validated the regulatory effect of STAT5A on KLF4 in both mRNA and protein level in different GC cell lines. Our results showed that STAT5A knockdown led to a significantly increased expression of KLF4 (Fig. 7D, E), while STAT5A overexpression resulted in a decreased expression of KLF4 in several cell lines (Fig. 7F, G). To further investigate whether the regulation of KLF4 by STAT5A occurs at the transcriptional level, we constructed the KLF4 promoter luciferase reporter vector and performed the Dual-luciferase reporter assay. As is shown in Fig. 7H, STAT5A knockdown increased luciferase activity linked to the KLF4 promoter, whereas overexpression of STAT5A resulted in a decline. According to the ConTraV3 website prediction, we introduced mutations within the predicted high-confidence STAT5A binding sites of the KLF4 promoter. We co-transfected these mutant vectors with empty vector or STAT5A expression vector into GC cells and performed the Dual-luciferase reporter assay. We observed that, overexpression of STAT5A did not significantly alter the dual luciferase activity of mutation site3 (-902~-895) (Fig. 7I). To further confirm the luciferase activity results, we performed ChIP assay. Consistent with the luciferase activity assay, we observed a higher enrichment of DNA fragments at site 3 through ChIP-qPCR experiments in MKN-45 cells overexpressing Flag-STAT5A (Fig. 7J). These findings suggest that site 3 may function as the putative binding region for STAT5A on the KLF4 promoter. It has been known that the Tyr-694 site of STAT5A is the key activation site for its phosphorylation and dimerization. 35 Therefore, we constructed a non-phosphorylatable mutation of the STAT5A (Y694A)35 and found that the expression of KLF4 was still decreased by the STAT5A mutant which fails to form an activated dimer (Fig. S7), indicating that the phosphorylation or dimerization of STAT5A is not necessary for regulating KLF4 expression.
3.9 STAT5A exerts its biological functions partially through KLF4
Having obtained the regulatory role of STAT5A on KLF4, we next aimed to investigate whether STAT5A exerted its biological role through KLF4. We first investigated whether KLF4 regulated cell proliferation and migration in cancer cells. As shown in Fig. 8A-F, overexpression of STAT5A inhibits gastric cancer cell proliferation and migration in vitro. Subsequently, we conducted functional recovery experiments by overexpressing STAT5A and KLF4 in SGC-7901 and MKN-45 cells. Our study demonstrated that the enforced expression of KLF4 can counterbalance the oncogenic impact of STAT5A (Fig. 8G-K).