METTL3 and IGF2BP2 coordinately regulate FOSL1 mRNA via m6A modification, suppressing trophoblast invasion and contributing to fetal growth restriction.

RC, GH, and HB performed study concept and design. RC, TW, and HT conducted the experiments. RC, JR, and XZ performed data analysis. RC wrote the paper. RC, TW, and HT drew the figures. GH, XL, HB, PB, and MK performed the editing and critical revision of the paper. All the coauthors have read and approved the final paper for publication.

Written informed consent was obtained from all participants. All the collection of samples and animal operations in this study were evaluated and approved by the Medical Ethics Committee of The First Affiliated Hospital of Chongqing Medical University (Approval number: 2022‐K524).

Submitted for publication.

This study was funded by National Key Research and Development Program of China (2022YFC2704702, 2021YFC2701501) and Sichuan Province Science and Technology Support Program (2023ZYD0117).

To explore whether there are any changes in m6A modification in FGR, the levels of m6A methylation were evaluated in chorionic villous tissues from the FGR group ( n  = 27) and the normal group ( n  = 23). The results demonstrated that m6A methylation levels were substantially elevated in FGR tissues (Figure  1A,B ). We further examined the expression of critical m6A regulators, including its writers (METTL3, METTL14, and WTAP) and erasers (FTO and ALKBH5) in FGR and normal chorionic villous tissues. The results indicated that METTL3 and FTO mRNA levels were significantly increased in the FGR group (Figure  1C ). However, linear correlation analysis showed that the METTL3 and FTO mRNA levels were positively correlated with the m6A level (Figure  1D and Figure  S1 ). Additionally, only METTL3 protein levels were significantly higher in the FGR chorionic villous tissues compared to the normal tissues, while no difference was observed in the FTO protein levels (Figure  1E ). These results indicated that the increase of METTL3 expression promoted m6A methylation in FGR placentae. Furthermore, a stronger positive signal for METTL3 was detected in FGR group by both immunofluorescence staining and Immunohistochemistry (Figure  1F,G ). Overall, these results illustrated that METTL3 expression and m6A modification were increased in placentae of patients with FGR. The levels of N6‐methyladenosine and METTL3 were elevated in fetal growth restriction. (A) The RNA m6A methylation levels in chorionic villous tissues from FGR ( n  = 27) and normal group ( n  = 23) were assessed using the EpiQuik m6A RNA methylation quantification kit (**** p  < .0001 vs. Normal group). (B) The m6A levels in chorionic villous tissues from FGR and normal group were detected via m6A dot blot assay. (C) qRT‐PCR analysis of METTL3, METTL14, WTAP, FTO, and ALKBH5 mRNA levels in chorionic villous tissues from FGR and normal group ( n  = 20 per group), (** p  < .01 vs. Normal group). (D) Linear regression analysis of METTL3 mRNA levels and relative m6A methylation levels in chorionic villous tissues. (E) Western blot analysis of METTL3 and FTO protein levels in chorionic villous tissues from FGR and normal group ( n  = 7 per group), (** p  < .01; ns = nonsignificant vs. Normal group). (F) IF staining of METTL3 (green) and CK7 (red) in frozen sections of term placentae from FGR and normal group; nuclei were stained with DAPI (blue) (scale bar: 50 μm). (G) IHC staining of METTL3 in term placentae from FGR and normal group (scale bar: 100 μm). The results are the mean ± SEM. To investigate the function of METTL3 in trophoblast cells, we first detected the expression level of METTL3 in three trophoblast‐derived cell lines via western blotting and qRT‐PCR. The level of METTL3 was higher in the HTR‐8/SVneo (HTR‐8) cells and JAR cells compared to the BeWo cells (Figure  S2A,B ). Based on these results, we established METTL3 overexpression and knockdown models in the HTR‐8 cells (Figure  2A,B ) and JAR cells (Figure  S3A,B ) via the lentiviruses transfection. The m6A level was dramatically increased in METTL3 overexpression and decreased in METTL3 knockdown cells (Figure  2C and Figure  S3C ). Additionally, immunofluorescence was performed to clarify the localization of METTL3 in HTR‐8 cells. We observed METTL3 was expressed in nucleus of HTR‐8 cells (Figure  S2C ). The transwell invasion and migration assays revealed that overexpression of METTL3 significantly inhibited migration and invasion of HTR‐8 cells and JAR cells. Conversely, METTL3 knockdown had the opposite effects in HTR‐8 cells (Figure  2D ) and JAR cells (Figure  S3D ). Wound‐healing assay also revealed that METTL3 overexpression repressed the migration of HTR‐8 cells, whereas METTL3 knockdown promoted the migration (Figure  2E ). Furthermore, cell proliferation in HTR‐8 cells and JAR cells was evaluated by EdU and CCK8 assays, which illustrated that METTL3 overexpression inhibited the proliferation but METTL3 knockdown enhanced cell growth (Figure  2F,G and Figure  S3E,F ). To study the impact of METTL3 on cell proliferation, signals of Annexin V were quantified by flow cytometry. the results showed that there were no differences in the apoptosis levels of METTL3‐overexpressed or knockdown cells except METTL3‐knockdown JAR cells (Figure  2H and Figure  S3G ). Taken together, these results demonstrated the crucial role of METTL3 in regulating trophoblast cell proliferation, migration, and invasion. METTL3‐mediated m6A methylation inhibited the proliferation, invasion, and migration of HTR‐8 cells. (A) qRT‐PCR analysis and (B) western blot analysis showed the expression levels of METTL3 in HTR‐8 cells transfected with sh‐METTL3 and METTL3‐overexpressing lentivirus after 48 h ( n  = 3 each), (**** p  < .0001 vs. WT, sh‐NC and vector). (C) The RNA m6A methylation levels were assessed in HTR‐8 cells transfected with sh‐METTL3 and METTL3‐overexpressing lentivirus after 48 h ( n  = 3 each), (** p  < .01 vs. WT and sh‐NC; **** p  < .0001 vs. WT and vector). (D) The invasion and migration ability of METTL3 overexpression and knockdown cells were confirmed by Transwell assay (scale bar: 200 μm) ( n  = 3 each), (** p  < .01 vs. vector; *** p  < .001 vs. sh‐NC and vector; **** p  < .0001 vs. sh‐NC). (E) The migration ability of METTL3 overexpression and knockdown cells were confirmed by wound‐healing assay (scale bar: 1 mm) ( n  = 3 each), (** p  < .01 vs. sh‐NC and vector). (F) EdU staining of METTL3 overexpression and knockdown cells; nuclei were stained with DAPI (blue) (scale bar: 100 μm) ( n  = 3 each), (*** p  < .001 vs. sh‐NC and vector). (G) Cell proliferation ability of METTL3 overexpression and knockdown cells were confirmed by CCK8 assay ( n  = 3 each), (* p  < .01 vs. sh‐NC; *** p  < .001 vs. vector). (H) Flow cytometry analysis of cell apoptosis of METTL3 overexpression and knockdown cells ( n  = 3 each), (ns = nonsignificant vs. sh‐NC and vector). The results are the mean ± SEM. To identify the RNA m6A modification targets of METTL3 in trophoblast cells, we performed both Methylated RNA immunoprecipitation sequencing (MeRIP‐seq) and RNA sequencing in METTL3‐knockdown and control HTR‐8 cells. The metagene profile of m6A distribution demonstrated that the m6A peaks were primarily enriched in the coding sequence (CDS) regions and 3′ untranslated regions (UTR) (Figure  3A ). A previous report showed that the m6A modification occurs specifically in RRACH (R = G or A, H = A, C or U) consensus sequence. 26 Here, the UGGAC motif was enriched in the HTR8 cells (Figure  3B ). Furthermore, RNA‐sequencing showed that 156 were upregulated and 153 were downregulated in METTL3‐knockdown HTR‐8 cells (Figure  3C ). METTL3‐mediated m6A modification enhanced the expression and mRNA stability of FOSL1. (A) Peak distribution of m6A modification across mRNA transcripts. (B) The consensus sequence motif for m6A methylation was identified by MeRIP‐seq. (C) Volcano plot showed differentially expressed genes in sh‐METTL3 and sh‐NC groups. (D) GO enrichment analysis reflected the distribution of differential genes (E) KEGG pathway analysis of genes differentially expressed. (F) Distribution of genes with significant changes in mRNA levels (up or down) and m6A levels (hyper or hypo) in sh‐METTL3 and sh‐NC groups (fc, |fold change| ≥1.2; p  < .05). (G) Venn diagram showed both m6A hypomethylated and RNA downregulated genes in METTL3 knockdown cells identified by RNA‐seq and MeRIP‐seq. (H) MeRIP‐qPCR analysis illustrated enrichment of m6A‐modified FOSL1, DDX10, TMEM158 and ALDH1A3 in HTR‐8 cells ( n  = 3 each). The results are the mean ± SEM. Next, the Gene Ontology (GO) analysis revealed that m6A‐modified genes were primarily enriched in signal transduction, placental blood vessel development, and female pregnancy (Figure  3D ). The KEGG pathway analysis also indicated the potential regulatory pathways, including MAPK signaling pathway and PI3K‐Akt signaling pathway (Figure  3E ). We further elucidated the variation of differential genes at both mRNA levels and m6A levels via the combined analysis of RNA‐seq data and MeRIP‐seq data. We identified 28 Hyper‐Up genes as defined by hypermethylated m6A peaks with increased mRNA expression, and 25 Hypo‐Down genes as defined by hypomethylated m6A peaks with decreased mRNA expression. Similarly, there were 26 Hyper‐Down genes, 19 Hypo‐Up genes (Figure  3F ). Since there were very few Hyper‐Up genes that we were interested in, and the significantly elevated genes were mainly related to folate metabolism, kinase, and collagen synthesis, we focused on Hyper‐Down genes and then combined them with previous GO and KEGG analyses. We screened out four potential METTL3 target genes (Figure  3G ). Finally, we performed MeRIP‐qPCR and verified the m6A modification of FOSL1, DDX10, TMEM158, and ALDH1A3 (Figure  3H ). Collectively, we examined the m6A modification profile and identified four potential downstream target genes regulated by METTL3 in trophoblast cells for further investigation. Based on previous experiments, we detected the mRNA levels of candidate genes in HTR8 cells with METTL3 overexpression and knockdown, including FOSL1, DDX10, TMEM158, and ALDH1A3. However, only FOSL1 and ALDH1A3 showed a consistent change trend (Figure  4A,B ). Next, the MeRIP‐qPCR was performed to verify the METTL3‐mediated m6A modification of FOSL1 and ALDH1A3 in METTL3 overexpression and knockdown cells. Interestingly, only FOSL1 but not ALDH1A3 m6A modification was significantly increased upon METTL3 overexpression, while it was decreased after METTL3 knockdown (Figure  4C,D and Figure  S4A,B ). Furthermore, western blot analysis showed that METTL3 overexpression elevated FOSL1 protein expression, whereas METTL3 silencing reduced it, which was consistent with the variation of mRNA levels (Figure  4E,F and Figure  S4C,D ). Correspondingly, the result via IGV viewer also demonstrated that the m6A peak level was decreased in METTL3 knockdown cells (Figure  4G ). Additionally, we determined that METTL3 had the strongest interaction with FOSL1 mRNA by performing immunoprecipitation qPCR (RIP‐qPCR), compared to other genes (Figure  4H ). Therefore, these evidences supported that FOSL1 was a downstream target of the METTL3‐mediated m6A modification. MeRIP‐Seq identified m6A modification profile and downstream targets in trophoblast cells. (A and B) qRT‐PCR analysis showed the mRNA levels of FOSL1, DDX10, TMEM158 and ALDH1A3 in (A) METTL3 overexpression and (B) METTL3 knockdown cells ( n  = 3 each), (* p  < .05 vs. vector and sh‐NC; *** p  < .001 vs. vector and sh‐NC; **** p  < .0001 vs. vector). (C and D) MeRIP‐qPCR analysis illustrated m6A‐modified levels of FOSL1 in (C) METTL3 overexpression and (D) METTL3 knockdown cells ( n  = 3 each), (* p  < .05 vs. sh‐NC; *** p  < .001 vs. vector). (E and F) Western blot analysis showed the protein levels of FOSL1 in (E) METTL3 overexpression and (F) METTL3 knockdown cells. (G) IGV analysis illustrated peak distribution of FOSL1in MeRIP profiles of sh‐METTL3 cells compared to sh‐NC cells. (H) RIP‐qPCR assay was performed to detect the direct binding of METTL3 protein to FOSL1, DDX10, TMEM158, and ALDH1A3 mRNA in HTR‐8 cells ( n  = 3 each). (I) qRT‐PCR analysis of YTHDF1, YTHDF3, IGF2BP2, and HNRNPA2B1 mRNA levels in chorionic villous tissues from FGR and normal group ( n  = 9–10 per group), (** p  < .01 vs. Normal group). (J) RNA stability assay showed the FOSL1 mRNA half‐life in METTL3 overexpression cells after being treated with actinomycin D (5 μg/mL) ( n  = 3 each), (*** p  < .001 vs. vector). (K–M) RIP‐qPCR assay was performed to detect the direct interaction between IGF2BP2 protein and FOSL1 mRNA in (K) HTR‐8 cells, (L) METTL3 overexpression and (M) METTL3 knockdown cells ( n  = 3 each), (*** p  < .001 vs. vector; **** p  < .0001 vs. IgG and sh‐NC). The results are the mean ± SEM. It has been reported that “m6A reader” proteins play a regulatory role in m6A‐modified transcripts. 27 , 28 Hence, we examined the mRNA levels of several “m6A readers” that exert different functions in FGR and normal placentae. Only IGF2BP2 was increased significantly in chorionic villous tissues from the FGR group (Figure  4I and Figure  S4E ). Given that, the main function of IGF2BP2 is to regulate mRNA stability, we treated the HTR8 cells with actinomycin D, which could interfere with cellular mRNA synthesis by inhibiting RNA polymerase activity, and found that the mRNA stability of FOSL1 was significantly increased in cells with METTL3 overexpression (Figure  4J ). Moreover, RIP‐qPCR further elucidated the regulatory mechanism of METTL3, IGF2BP2, and FOSL1. IGF2BP2 directly binds to FOSL1 mRNA, and this binding interaction was enhanced by METTL3 overexpression but suppressed by METTL3 silence (Figure  4K–M ). In sum, these results suggested that IGF2BP2 binds to FOSL1 mRNA to regulate its stability in a METTL3‐m6A‐dependent manner. To further verify the role of FOSL1 in the pathogenesis of FGR, we evaluated the expression of FOSL1 in FGR placentae. Both mRNA and protein levels of FOSL1 were significantly upregulated in villous tissues from FGR placentae, which was consistent with our sequencing results (Figure  5A–C ). Meanwhile, immunofluorescence staining showed that the proteins transcribed by FOSL1 were mainly located in the cytoplasm (Figure  5C ). Next, we knocked down FOSL1 by siRNA in HTR8 cells to investigate its effect on trophoblast function (Figure  S5A,B ). The Transwell assays demonstrated that the trophoblast migration and invasiveness were significantly enhanced after FOSL1 disruption (Figure  5D ). Moreover, the knockdown of FOSL1 had no effect on cell proliferation and apoptosis of HTR8 cells (Figure  5E,F ). Correspondingly, a rescue experiment was carried out to examine whether FOSL1 is a major mediator of METTL3's impact on trophoblast invasion. We used siRNA to suppress FOSL1 in METTL3‐overexpressed HTR‐8 cells, which showed that FOSL1 interference partially restored the migration and invasion ability inhibited by METTL3 overexpression (Figure  5G and Figure  S5C,D ). Additionally, suppressed cell proliferation caused by METTL3 overexpression was not reversed by FOSL1 knockdown, and no difference was observed in cell apoptosis either (Figure  S5E,F ). Thus, these results suggested that FOSL1 functions as a downstream target gene of METTL3 to regulate trophoblast cells in FGR progression. FOSL1 inhibited invasion and migration of trophoblast cells. (A) qRT‐PCR analysis and (B) western blot analysis showed the expression levels of FOSL1 in chorionic villous tissues from FGR and normal group ( n  = 5–9 per group), (** p  < .01 vs. Normal group). (C) IF staining of FOSL1 (red) in frozen sections of term placentae from FGR and normal group; nuclei were stained with DAPI (blue) (scale bar: 50 μm). (D) The invasion and migration ability of FOSL1 silencing HTR‐8 cells were confirmed by Transwell assay ( n  = 3 each), (*** p  < .001 vs. si‐NC; **** p  < .0001 vs. si‐NC). (E) Cell proliferation ability of FOSL1 knockdown cells was confirmed by CCK8 assay ( n  = 3 each), (ns = nonsignificant vs. si‐NC). (F) Flow cytometry analysis of cell apoptosis of FOSL1 knockdown cells ( n  = 3 each), (ns = nonsignificant vs. si‐NC). (G) The invasion and migration ability of si‐FOSL1 after METTL3 overexpression in HTR‐8 cells were confirmed by Transwell assay (scale bar: 200 μm) ( n  = 3 each). The results are the mean ± SEM. STM2457, a novel and highly specific inhibitor of METTL3's catalytic activity, has been used in treating acute leukemia. 29 We hypothesized that this small‐molecule inhibitor STM2457 might act as a treatment for FGR. As shown in Figure  6A , we first established an FGR mice model by feeding mice with a prenatal PR diet, 30 and then pregnant mice were intravenously administered 25 mg/kg of STM2457 or vehicle (150 μL) once a day from embryonic day 13.5 (E13.5) to E17.5. As expected, the fetal and placental weights at E17.5 were significantly reduced in the group with PR diet, but no significant difference in the average litter size was observed, as compared to the normal pregnant mice. In the dams treated with STM2457, we found that the weight loss of fetuses and placentae was partially reversed as compared to those in the dams with PR diet (Figure  6B–E ). Furthermore, an H&E staining of fetuses confirmed similar results (Figure  6F ). Glycogen trophoblasts are PAS staining positive. PAS‐stained histological sections of placentae displayed that the placental development in the FGR mice was compromised, especially a marked reduction in the junctional zone. The treatment with STM2457 partially restored the junctional zone development (Figure  6G ). These results indicated that the placental development is under insufficient nutrient supply, and METTL3 inhibition improves the development. The effects of METTL3 inhibitor on FGR progression in vivo. (A) Scheme of experimental design. (B and C) Representative photographs showed gross morphological appearance of (B) fetuses and (C) placentas from normal, low protein diet and STM2457 treatment groups on E18.5 ( n  = 5 per group of dams). (D) Fetal birth weights (ns = nonsignificant vs. Low protein; **** p  < .0001 vs. Normal and Low protein + vehicle), and (E) placental weights at E18.5 ( n  = 66–93 from 22 dams), (ns = nonsignificant vs. Low protein; *** p  < .001 vs. Low protein + vehicle; **** p  < .0001 vs. Normal). (F) H&E‐stained histological sections of isolated fetuses at E18.5 (scale bar: 1000 μm). (G) Full and high magnification images of PAS‐stained placental sections at E18.5 (scale bar: 500 μm of upper panel; 100 μm of lower panel). (H) The RNA m6A methylation levels in placentae from each group at E18.5 were assessed using the EpiQuik quantification kit ( n  = 5 per group), (* p  < .05 vs. Normal; ** p  < .01 vs. Normal). In order to examine the impact of STM2457 on METTL3‐mediated m6A modification in vivo, the levels of m6A modification and METTL3 expression in placentae were examined. The results showed that m6A levels were decreased in STM2457 treatment group compared to the control group, whereas increased in other two protein‐restricted diet groups (Figure  6H ). Additionally, METTL3 mRNA levels and METTL3 protein expression had no differences among the four groups (Figure  S6A,B ). Therefore, these results implied that the METTL3 inhibitor STM2457 reduced m6A modification to reverse fetal weight loss. Considering the potential of STM2457 for FGR treatment as shown in Figure  6 , we further investigated its impact on RNA modifications in vitro. First, HTR‐8 cells were treated with different concentrations (0.05, 0.5, 1, 5, 10, 25, and 50 μM) of STM2457 for 48 h. Interestingly, a dose‐dependent relationship between METTL3 protein expression and STM2457 was detected in HTR‐8 cells (Figure  7A ). A recent study in vitro also revealed that the application of STM2457 triggered compensatory METTL3 overexpression and protein accumulation. 31 Of note, the dose‐dependent association between STM2457 and levels of m6A modification was U‐shaped (Figure  7B ), perhaps influenced by inhibition of METTL3 catalytic activity and its compensatory mechanism. The effects of STM2457 on METTL3 expression and m6A modification in vitro. (A) Western blot analysis showed the expression levels of METTL3 in HTR‐8 cells treated with STM2457 (0.05, 0.5, 1, 5, 10, 25, and 50 mM) or DMSO (5 mM) after 48 h ( n  = 3 each), (* p  < .05 vs. DMSO; **** p  < .0001 vs. DMSO). (B) The RNA m6A methylation levels in HTR‐8 cells treated with STM2457 (0, 0.05, 0.5, 1, 5, 10, 25, and 50 mM) or DMSO (5 mM) after 48 h was assessed using the EpiQuik quantification kit ( n  = 3 each), (* p  < .01 vs. DMSO; *** p  < .001 vs. DMSO; **** p  < .0001 vs. DMSO). (C) Cell viability of HTR‐8 cells treated with STM2457 (0.05, 0.5, 1, 5, 10, 25, and 50 mM) or DMSO (5 mM) after 48 h were confirmed by CCK8 assay ( n  = 3 each), (ns = nonsignificant vs. DMSO; * p  < .01 vs. DMSO; *** p  < .001 vs. DMSO; **** p  < .0001 vs. DMSO). (D) Transwell assay revealed ability of HTR‐8 cells treated with STM2457 (0.05, 1, and 5 mM) or DMSO (5 mM) after 48 h in invasion ( n  = 3 each), (ns = nonsignificant vs. DMSO; * p  < .01 vs. DMSO; ** p  < .01 vs. DMSO) and migration ( n  = 3 each), (ns = nonsignificant vs. DMSO; * p  < .01 vs. DMSO; ** p  < .01 vs. DMSO) (scale bar: 200 μm). (E) Flow cytometry analysis of cell apoptosis HTR‐8 cells treated with STM2457 (0.05, 1, and 5 mM) or DMSO (5 mM) after 48 h ( n  = 3 each), (** p  < .01 vs. DMSO; *** p  < .001 vs. DMSO). The results are the mean ± SEM. Based on earlier studies, 31 , 32 STM2457 reduced the cell viability of tumor cells in a dose‐dependent manner. Therefore, we evaluated whether STM2457 restrains the cell viability in HTR‐8 cells. Interestingly, STM2457 promoted cell proliferation at lower concentrations (0.05 μM), while inhibited the proliferation at higher concentrations (≥ 5 μM) (Figure  7C ). Given that, STM2457 affected cell viability in a dose‐dependent manner, three concentrations (0.05, 1, and 5 μM) were selected to investigate cell invasion and apoptosis. Similar to the results of CCK8 experiment, STM2457 facilitated cell invasion and migration at 0.05 μM but suppressed it at 5 μM, and a concentration of 1 μM had no apparent effect (Figure  7D ). In contrast to its stimulation of tumor cell apoptosis, 32 , 33 STM2457 showed an inhibitory effect on trophoblast apoptosis (Figure  7E ). In conclusion, STM2457 regulated METTL3 expression and m6A modification in trophoblast in a dose‐dependent manner, thereby modulating trophoblast function. The complex molecular mechanism needs to be further explored.

FGR, as a complication affecting 10% of pregnancies, 34 is associated with many chronic adult diseases, especially cardiometabolic disorders and psychological diseases. 35 , 36 Although the pathogenesis of FGR is diverse, the majority of FGR cases are considered to attribute to placental insufficiency besides fetal congenital malformations and fetal genetic abnormalities. Placenta is a supporting organ for material exchange between the fetus and the mother during pregnancy, which is indispensable to normal fetal development. Defects in placentation result in inadequate nutrient‐ and oxygen supply. Emerging evidence also suggests that repressed mammalian targets of rapamycin (mTOR) signaling 37 and activation of hypoxia‐inducible factors (HIFs) 38 are involved in FGR development. Epigenetic modifications have attracted increasing attention in placentation, of which DNA methylation is the most studied. 39 However, the epigenetic regulation of FGR development remains poorly understood. In this study, we revealed that epigenetic METTL3‐IGF2BP2‐FOSL1 axis was implicated in the pathogenesis of FGR. Increased expression of METTL3 elevated the m6A modifications of FOSL1 mRNA, and the m6A reader IGF2BP2 maintained the stability of FOSL1 mRNA by directly binding to its mRNA. Upregulated FOSL1 inhibited trophoblast invasion, which leads to abnormal placentation and promoted the progression of FGR (Figure  8 ). The graphic illustration of the mechanism by which METTL3‐mediated m6A modification regulates FOSL1 expression to inhibit trophoblast invasion in FGR progression. Recent studies have shown that m6A modification played important biological functions in female fertility and related reproductive diseases including endometriosis, premature ovarian insufficiency, and polycystic ovary syndrome. 19 , 40 Although the m6A modification has been reported to be involved in pregnancy‐related diseases including preeclampsia, 21 , 41 , 42 , 43 , 44 , 45 gestational diabetes mellitus, 20 , 46 and miscarriage, 22 , 23 , 47 , 48 , 49 , 50 its role in FGR remains unclear. Notably, the upregulations of METTL3 and m6A modification were ubiquitous in these diseases and suppressed the proliferation and invasion of trophoblast, 21 , 41 , 42 , 43 , 44 , 45 consistent with our results. Here, we are the first to provide the evidence that METTL3‐mediated m6A modification plays a key role in FGR. Among the main m6A writers, METTL3 was the only enzyme with consistent changes in both transcription and protein levels, whereas the inconsistent changes in FTO may be attributed to posttranslational modifications. 51 Given its increased levels in FGR, METTL3 was selected to explore how its dysregulation resulted in FGR. In the present study, MeRIP‐seq and RNA‐seq analysis demonstrated that FOSL1 was the downstream target of METTL3 in trophoblasts. Using RIP‐qPCR and MeRIP‐qPCR analyses, we further identified that METTL3 epigenetically activated FOSL1 mRNA via the m6A methylation, which was revealed for the first time. FOSL1 plays an important role in cancer progression, which was consistent with our KEGG analysis. Nevertheless, the downstream molecular pathways regulated by FOSL1 warrant further studies. FOSL1 is also involved in EMT. The previous studies have underscored the impact of EMT in the transformation of cytotrophoblast (CTB) to EVT. 52 Through this differentiation process, EVT acquires invasiveness to promote placentation. 53 Thus, whether FOSL1 is involved in regulating EMT leading to FGR needs to be further explored. Furthermore, FOSL1 is overexpressed in the majority of tumor tissues, promoting cell invasion and migration which was contrary to our results. 54 , 55 The plausible explanations could be attributed to differences in the regulation of cellular functions by target genes between trophoblasts and tumor cells. m6A modifications on target genes rely on m6A readers to exert their corresponding functions. IGF2BP2, characterized by increasing mRNA stability, 56 was identified to be upregulated in FGR. We hypothesized that IGF2BP2 was involved in the regulation of METTL3‐mediated m6A methylation to enhance FOSL1 expression by increasing the stability of FOSL1 mRNA. This perspective was confirmed by IGF2BP2‐RIP and RNA stability assay. We found that IGF2BP2 is physically associated with FOSL1 mRNA and accumulation of m6A modifications could enhance this binding, triggering an increase in mRNA half‐life. The mechanism of such gene interaction has been studied in colorectal carcinoma. 57 In the current study, we are the first to elucidate METTL3 and IGF2BP2 co‐regulated FOSL1 expression in an m6A‐dependent manner. Since elevated METTL3 expression correlated with the progression of FGR, we then explored whether METTL3 inhibition could reverse the FGR phenotype. STM2457 inhibits the catalytic activity of METTL3 and has anticancer effects against multiple cancers. 29 , 31 , 33 As expected, in vivo results showed that STM2457 markedly rescued placental and fetal weight loss in FGR mouse models. Moreover, STM2457 mainly restored the development of the junction zone, which is critical for normal placentation, 58 suggesting that STM2457 may improve the material transport to ensure the development of the offspring. Nevertheless, in vitro studies demonstrated that METTL3 protein expression was upregulated in STM2457‐treated group, which possibly attributed to compensatory protein expression. This mechanism is not rare in the action of small molecule inhibitors, such as the MET receptor tyrosine kinase inhibitors. 59 We detected an increase in m6A levels, whereas no differences in METTL3 protein and mRNA levels in STM2457‐treated group in vivo, indicating that STM2457 acted by inhibiting catalytic activity. Furthermore, the optimal inhibitory effect of STM2457 on m6A modification was at a concentration of 25 μM, which provides a basis for subsequent relevant studies. It is worth mentioning STM2457 regulated trophoblast proliferation, invasion, and migration in a dose‐dependent manner. In general, our findings shed light on the potential use of STM2457 in treatment of FGR, although the efficacy and safety for pregnancy require further validation. In conclusion, our study reveals a novel epigenetic mechanism in the pathogenesis of FGR. Mechanistically, elevated METTL3 upregulated FOSL1 expression in an m6A‐IGF2BP2‐dependent manner, resulting in compromised trophoblast invasion and migration, and ultimately led to the phenotype of FGR. To our knowledge, this is the first study testing the therapeutic potential of STM2457 against FGR, although the safety of its use in pregnancy requires further validation.

The authors declare no competing interests.

Fetal growth restriction (FGR) is a common complication of pregnancy defined as the failure of the fetus to achieve its genetically determined growth potential and significantly increases the risk of neonatal morbidity and mortality. 1 , 2 Recently, although molecular pathways that contribute to FGR have been reported, including growth regulatory pathways and stress response pathways, most studies just focused on the pathologic mechanisms of preeclampsia complicated by FGR. 3 Current therapies for managing FGR are still limited. Therefore, it is of great value to explore the molecular mechanism of FGR for its specific treatments. N 6 ‐methyladenosine (m6A), as the most prevalent internal modification in eukaryotic RNA, has been extensively studied in the regulation of nuclear export, mRNA splicing, degradation, translation, and stability, serving as an important epigenetic regulatory mechanism underlying various bioprocesses. 4 Protein factors that dynamically regulate m6A modification are categorized into “writers,” “erasers,” and “readers.” The m6A methyltransferase complex (MTC), mainly composed of METTL3, METTL14, and WTAP, catalyzes m6A methylation as the “writers”. 5 , 6 , 7 Meanwhile, m6A modification is reversed by the “erasers” including demethylases FTO and ALKBH5. 8 , 9 , 10 The “readers” can directly bind to and recognize m6A to determine the target mRNA fate, such as YTH domain family, IGF2BP protein family, and so on. 11 , 12 , 13 , 14 , 15 Recent studies have demonstrated that conditional deletion of METTL3 by Pgr‐Cre resulted in complete implantation failure 16 , 17 and depletion of METTL14 led to embryonic lethality. 18 Moreover, dysregulation of m6A modification was associated with several female reproductive diseases, 19 and in terms of placenta‐associated disorders, aberrant m6A‐modified gene expression contributes to the pathophysiology of preeclampsia, recurrent miscarriage, and gestational diabetes mellitus. 20 , 21 , 22 , 23 Nevertheless, the relationship between m6A modification and FGR remains unclear. Fos‐like antigen‐1 (FOSL1), a transcription factor encoded by Fra‐1, regulates a variety of biological and pathological processes, including cell proliferation, epithelial‐to‐mesenchymal transition (EMT), extracellular matrix degradation, motility, and metastasis. 24 It is well known that deficient placentation results in fetal growth restriction and other pregnancy‐associated disorders. The early development of placenta is closely related to extravillous trophoblasts invasion contributing to remodeling of the spiral arteries and syncytiotrophoblasts formation, which provides nutrients and secretions stimulating placental cell proliferation. It is worth mentioning that Fra1 has been shown to play a crucial role in establishing normal vascularization of the placenta. 25 However, the epigenetic control governing FOSL1 in placentation and the effects of Fra‐1 on trophoblasts remain elusive. In the present study, we revealed FGR had significantly higher expression of METTL3 and METTL3‐mediated m6A modification in trophoblasts. Next, we identified the METTL3 and IGF2BP2 together upregulated Fra1 by stabilizing its RNA, thereby inhibiting trophoblast invasion and migration. Finally, we provided evidence that METTL3 inhibitors could reverse the FGR phenotype, suggesting that METTL3 could be a new therapeutic target in treating FGR patients.

The research protocol complied with the Declaration of Helsinki and was approved by the Ethics Committee of the First Affiliated Hospital of Chongqing Medical University (Approval number: 2022‐K524). All clinical samples were collected with written informed consent from the participating patients. Animal health monitoring, environmental supervision, welfare checks, and operational revision were regularly performed by qualified administrators. Twenty‐seven women whose pregnancies were complicated by FGR and 23 women with normal pregnancies were recruited. All human placental tissues in this project were taken from pregnant women who underwent cesarean section in the Department of Obstetrics of the First Affiliated Hospital of Chongqing Medical University. The diagnosis of FGR adhered to the standards of the American College of Obstetricians and Gynecologists. Patients with preeclampsia, diabetes, cardiovascular disease, chronic kidney disease, chronic hypertension, metabolic diseases, and fetuses born with congenital abnormalities or aneuploidies were excluded. The clinical characteristics of the cases are shown in Table  S1 . Each sample was divided into several parts as follows: (i) the samples were fixed in 4% paraformaldehyde and then embedded in paraffin; (ii) kept in RNA later (Thermo Fisher Scientific, MA, USA); and (iii) immediately frozen in liquid nitrogen and stored at −80°C for use. Sexually mature ICR mice aged 8–9 weeks in this study were purchased from Hunan SJA Laboratory Animal Co (Hunan, China). These mice were raised under standard laboratory conditions. Two maternal mouse models aimed at inducing or rescuing FGR during pregnancy were used: consumption of a protein restriction (PR) diet or STM2457 treatment. ICR virgin female mice were mated with male mice. The next morning, the presence of a vaginal plug confirmed that the mouse was pregnant, which was designated as embryonic day 0.5 (E0.5). In the prenatal PR diet model, pregnant mice were randomly divided into normal group, which was fed a standard diet (20% protein) until labor, and low‐protein group, which was fed a PR diet (8% protein) until labor. In the low protein with treatment model, pregnant mice were fed a low‐protein diet (8% protein) and intravenously injected with vehicle (40% (m/V) PEG300) or 25 mg/kg of STM2457 (150 μL) once a day from E13.5 to E17.5, for 5 consecutive days (5 treatments). There were five pregnant mice in each of the above four groups. All mice had unrestricted access to water and were sacrificed by cervical translocation under anesthesia to collect and weigh placentas and fetuses on E18.5. Livers, kidneys, and uteri were dissected for further analysis. Blood pressure of mice was measured every day from E12.5 to E17.5 by BP‐2000 blood pressure analysis system (Visitech, TX, USA). STM2457 (MedChemExpress, NJ, USA) is an efficient and selective METTL3 inhibitor. In in vitro experiments, STM2457 was dissolved in dimethyl sulfoxide (DMSO) according to the product instructions, forming a storage solution with a concentration of 10 mM. Subsequently, the storage solution was diluted with complete culture medium to generate a series of working solutions at different concentrations. In in vivo experiments, STM2457 was dissolved as a working solution in the 40% (m/V) PEG300 (MedChemExpress) carrier (diluted by sterile normal saline) at a concentration of 10 mg/mL. HTR8/SVneo cells, JAR cells, and BeWo cells were obtained from the Cell Bank of Type Culture Collection of Chinese Academy of Sciences, Shanghai Institute of Cell Biology. HTR8/SVneo cells and JAR cells were cultured in RPMI‐1640 medium (Gibco, MA, USA) supplemented with 10% fetal bovine serum (PAN Biotech, Aidenbach, Germany), 1% penicillin–streptomycin (Gibco) at 37°C under humidified 5% CO 2 . The lentiviral vector for METTL3 overexpression, METTL3 knockdown, and the corresponding empty vector were constructed and synthesized by GenePharma (Shanghai, China). According to the manufacturer's instructions, 2 × 10 5 cells were infected with lentivirus with Polybrene for 48 h. RPMI‐1640 complete medium containing 1 μg/mL purinomycin (Beyotime, Shanghai, China) was added to screen the infected cells. Small interfering RNA (siRNA) and negative control (NC) siRNA were purchased from Tsingke (Beijing, China). In brief, 12 μL of TransIntro EL (Transgen, Beijing, China) and 10 μL of siRNA (20 μM) or negative control were mixed with Opti‐MEM medium (Gibco) at room temperature to a final volume of 200 μL/well for 20 min according to the manufacturer's protocol. Subsequently, the transfection mixture was added to HTR8/SVneo cells in a six‐well plate and incubated for 8 h. The transfection medium was then discarded and replaced with the serum‐containing medium for 48 h. Cells were collected to conduct qRT‐PCR and western blotting to validate the transfection efficiency and for further experiments. Total RNA was extracted from both cultured cells and placentae tissue using TRIzol reagent (Invitrogen, MA, USA). To ensure accurate and reliable results, the concentration, purity, and integrity of the RNA samples were meticulously assessed using a microvolume spectrophotometer (IMPLEN GmbH, Munich, Germany). To proceed with gene expression analysis, 1 μg of the total RNA was subjected to reverse transcription using the RT Master Mix (MedChemExpress). This step allowed for the conversion of RNA into complementary DNA (cDNA), facilitating subsequent amplification and detection. For quantification of specific gene expression, qRT‐PCR was performed using SYBR Green qPCR Master Mix (Selleck, TX, USA) according to the meticulously designed protocol provided by the manufacturer. Relative quantitation of these data was calculated and normalized to β‐actin with three repeats. The primers used are listed in Table  S2 . Total RNA of samples was extracted, and then Poly(A) RNA was purified from about 30 μg total RNA using Dynabeads™ mRNA Purification Kit (Thermo Fisher Scientific) by two rounds of purification. The m6A levels of Poly(A) RNA were evaluated by EpiQuik m6A RNA Methylation Quantification Kit (P‐9005, Epigentek, NY, USA), and 200 ng RNA was added to each assay well. After incubation in binding solution at 37°C for 90 min, capture antibody solution and detection antibody solution were then added to each well, respectively, and the relevant antibodies were added to each well in a suitable diluted concentration, respectively. The absorbance of each well at 450 nm was measured on a microplate reader (TECAN, Männedorf, Switzerland). Quantification was performed according to the standard curve to calculate m6A levels. Total RNA was harvested as previously described and denatured at 95°C for 10 min, and then immediately preserved on ice. Next, different quantities of mRNA were spotted on a Hybond‐N + membrane (LABSELECT, Shanghai, China). The membrane was incubated at 37°C for 30 min and blocked with 5% defatted milk for 1 h at room temperature. Then, the membrane was incubated with a m6A‐specifc antibody (1:1000; ab286164, Abcam, Cambridge, UK) at 4°C overnight. After three washes with PBST, the membrane was incubated with the HRP‐labeled goat anti‐rabbit IgG secondary antibody (1:5000; A0208, Beyotime) for 1 h at room temperature. Finally, the membrane was incubated with an ECL detection reagent (Advansta, CA, USA) and detected by the chemiluminescence system (Shenhua Science Technology, Hangzhou, China). Methylene blue staining was used to display equal quantity of RNAs added onto the membrane. Total RNA was isolated and purified using TRIzol reagent (Invitrogen) following the manufacturer's procedure. The RNA amount and purity of each sample were quantified using the microvolume spectrophotometer (IMPLEN GmbH), and the RNA integrity was assessed by Bioanalyzer 2100 (Agilent, CA, USA). Poly(A) RNA was purified from 50 μg total RNA using Dynabeads Oligo (dT)25–61 005 (Thermo Fisher Scientific) by two rounds of purification. Then the poly(A) RNA was fragmented into small pieces using Magnesium RNA Fragmentation Module (NEB, MA, USA) at 86°C for 7 min. The cleaved RNA fragments were incubated for 2 h at 4°C with m6A‐specific antibody (Synaptic Systems, Goettingen, Germany) in IP buffer (50 mM Tris–HCl, 750 mM NaCl and 0.5% Igepal CA‐630). The IP RNA was reverse‐transcribed to cDNA by SuperScript™ II Reverse Transcriptase (Invitrogen), which was next used to synthesize U‐labeled second‐stranded DNAs with E. coli DNA polymerase I (NEB), RNase H (NEB) and dUTP Solution (R0133, Thermo Fisher Scientific). An A‐base was added to the blunt ends of each strand, preparing for ligation to the indexed adapters. Dual‐index adapters were ligated to the fragments, and size selection was performed with AMPureXP beads. After the heat‐labile UDG enzyme (NEB) treatment of the U‐labeled second‐stranded DNAs, the ligated products were amplified with PCR. At last, we performed paired‐end sequencing (PE150) on an Illumina Novaseq™ 6000 platform (LC‐Bio Technology CO., Ltd., China) following the vendor's recommended protocol. Total RNA was extracted as described above. MeRIP assays were performed using the EpiQuik CUT&RUN m6A RNA Enrichment (MeRIP) Kit (P‐9018‐24, Epigentek), according to the manufacturer's instruction. Briefly, total RNA amount is 10 μg per reaction, which is separately incubated with 2 μg m6A antibody and 2 μg non‐immune IgG. After immunocapture and cleavage, enriched RNA was released from the affinity beads for further analysis by RT‐qPCR. Relative enrichment was normalized to the input. RIP assays were conducted with the RNA Immunoprecipitation Kit (Geneseed, Guangzhou, China) based on the manufacturer's protocol. One‐tenth of the cell lysates were used as input group, and the rest were used as the subsequent immunoprecipitation treatment. Briefly, 5 μg anti‐METTL3 (ab195352, Abcam), anti‐IGF2BP2 (ab128175, Abcam) and anti‐rabbit IgG (14678‐1‐AP, Proteintech, Wuhan, China) were conjugated to 200 μL protein A + G beads and then incubated with remaining lysates overnight at 4°C. After washing, bound RNA was extracted for RT‐qPCR and normalized to the input. HTR8/SVNEO cells after lentivirus transfection were seeded into 6‐well plates. 5 μg/mL Actinomycin D ( SBR00013 , Sigma, MO, USA) was administrated to the plates. Cells were collected for different time periods (0, 3, 6, and 9 h) for RNA extraction and analyzed by qRT‐PCR, normalized to β‐actin. All HTR8/SVNEO cells and JAR cells were treated and subjected to corresponding tests. For invasion experiments, 60 μL of 1 mg/mL Matrigel matrix (BD Biosciences, NJ, USA) solution was pre‐coated on cell culture inserts (Corning, NY, USA). Then, 8 × 10 4 cells were mixed with 300 μL of serum‐free 1640 medium (Gibco) and seeded on the upper layer of the cell culture inserts, while 600 μL of 10% fetal bovine serum‐containing 1640 complete medium was added to the lower layer. The cells were incubated at 37°C with 5% CO 2 in a cell culture incubator for 24 h. After that, the cells were fixed with pre‐cooled methanol for 15 min and stained with crystal violet (C0121, China) for 10 min. Finally, the non‐invaded cells on the upper layer were gently wiped off with a cotton swab. For migration experiments, 4 × 10 4 cells were seeded on the upper layer of the cell culture inserts with no pre‐coating of 60 μL of 1 mg/mL Matrigel matrix solution, and the rest of the procedure was the same as the invasion experiment. Images were captured using the EVOSTMFL Imaging System (Thermo Fisher Scientific). ImageJ software was used for cell counting in both invasion and migration assays. After lentivirus transfection, HTR8/SVNEO cells were seeded in a 6‐well plate for a suitable period of time until they fully covered the culture dish. A straight line was gently drawn at the bottom of each well using a 10 μL pipette tip. Then, the cell culture medium was removed, and the cells were washed twice with PBS before adding serum‐free culture medium. Finally, images were captured by microscope (Olympus, Tokyo, Japan) at time points (0, 24, and 48 h). The proliferative capacity of HTR8/SVNEO cells was examined using the EdU assay kit (RiboBIO, Guangzhou, China). Briefly, HTR8/SVneo cells, which had been treated accordingly, were seeded in a 96‐well plate. Following the manufacturer's protocol, the EdU solution was diluted in a complete medium at a ratio of 1000:1 and added to the wells, followed by incubation for 2 h. The cell nuclei were stained with DAPI. Subsequently, cell images were captured using a fluorescence microscope (Olympus), and cell counting was performed using ImageJ software. Cell viability was assessed using the Cell Counting Kit‐8 (CCK‐8) reagent (MedChemExpress). 2 × 10 3 cells were seed in a 96‐well plate, and 10 μL of CCK‐8 reagent was added to each well at different time points. After the plate was placed in a 37°C, 5% CO 2 incubator for 2 h, the absorbance at 450 nm was measured using a microplate reader (TECAN). HTR8/SVNEO cells were seeded in a 6‐well plate after appropriate treatment. The cell culture supernatant was collected in a 5 mL centrifuge tube after 48 h. Then, the cells were digested with trypsin without EDTA, and terminated with the cell culture supernatant. After treatment with the Annexin V‐FITC/PI Apoptosis Kit (Elabscience, TX, USA), the fluorescence intensity was detected using CytoFlex flow cytometer (BD Biosciences) and analyzed with the CytExpert 2.5 software. Total protein in placental tissues and HTR8/SVneo cells was extracted with RIPA lysis buffer (Beyotime) containing 1% protease inhibitor and measured by bicinchoninic acid (BCA) assay kit (Beyotime). The lysed proteins were resolved via SDS‐PAGE and transferred to the polyvinylidene difluoride (PVDF) membranes (Millipore, USA). After blocking, the membranes were further incubated with primary antibodies against METTL3 (1:1000; ab195352, Abcam), FTO (1:1000; 27 226‐1‐AP, Proteintech), FOSL1 (1:100; DF3096, Affinity, OH, USA), and β‐actin (1:1000; 81 115‐1‐RR, Proteintech) overnight at 4°C. Next, the membranes were incubated with an anti‐rabbit secondary antibody (1:5000; Beyotime) for 1 h at room temperature. Finally, the immunoreactive signals were exposed by the chemiluminescence system (Shenhua Science Technology) using an ECL assay kit and analyzed with ImageJ 1.50i software ( https://imagej.en.softonic.com ). Following the protocol of the immunohistochemistry kit (ZSGB‐BIO, Beijing, China), the dewaxed and hydrated tissue sections were heated in citrate buffer (pH 6.0) for 15 min to repair the antigens. The sections were incubated with hydrogen peroxide for 10 min to block endogenous peroxidase activity. Subsequent incubation with a blocking buffer occurred for 1 h. Slices were incubated overnight at 4°C with primary antibodies against METTL3 (1:100; ab195352, Abcam) and FOSL1 (1:100; DF3096, Affinity). After 1 h of secondary antibody incubation, the DAB working solution was applied to the tissue sections and reacted for 20–50 s, followed by staining with hematoxylin. After dehydration and drying, tissue sections were sealed with neutral resin for digital scanning and analysis. For cell staining, cells seeded on the surface of coverslips were fixed for 30 min at room temperature with 4% paraformaldehyde, and then permeabilized with 0.1% Triton X‐100 for 15 min at room temperature. For tissue sections, deparaffinized and hydrated tissue sections were heated in citrate buffer (pH 6.0) for 15 min to repair the antigen. Both cell slides and tissue sections were incubated with a blocking buffer for 1 h. Slices were incubated overnight at 4°C with primary antibodies against METTL3 (1:100; ab195352, Abcam), cytokeratin‐7 (CK‐7) (1:150; ZM‐0071, ZSGB‐BIO), and FOSL1 (1:100; ab252421, Abcam). Fluorescent‐conjugated secondary antibodies were incubated at room temperature for 1 h. The cell nucleus was then stained with DAPI for 10 min. Anti‐fluorescence quenching sealing tablets, refrigerated at 4°C or − 20°C. The placentae and fetuses of mice, fixed in paraformaldehyde, were embedded in paraffin and cut into 3 μm‐thickness tissue sections. After deparaffinization, fetus tissue sections were stained with hematoxylin (Servicebio, Wuhan, China) for 5 min. Differentiation was achieved using 1% hydrochloric acid alcohol for several seconds, then reverted to blue with 1% ammonia water to complete nuclear staining. Eosin (Servicebio) was used for cytoplasmic staining for 3 min. After dehydration and drying, the tissue sections were sealed with neutral resin for digital scanning and analysis. After dewaxing the placenta tissue sections, assays were performed using PAS staining solution (Servicebio) according to the manufacturer's instructions. In a dark environment, the sections were oxidized with periodate solution for 10–15 min. Stained with Schiff reagent for 20–30 min. Hematoxylin was used to stain cell nuclei for 5 min and rinsed with tap water for bluing. After dehydration and drying, the tissue sections were sealed with a neutral resin and digitally scanned and analyzed. Statistics were analyzed using GraphPad Prism 9.0 ( https://www.graphpad.com ). All values are presented as the means ± standard errors of the means (SEM). Unpaired t‐tests were applied to compare continuous variables between two groups. Statistical differences among multiple groups were evaluated by one‐way analysis of variance (ANOVA). Correlations were evaluated using simple linear regression analysis. Here, p values <.05 were considered as statistically significant. All tests were performed in triplicate.

Figure S1. Table S1.