METTL3-mediated m6A methylation negatively modulates autophagy to support blastocyst development

Background: N 6 -methyladenosine (m6A) catalyzed by METTL3 regulates the maternal-to-zygotic transition in zebrash and mice. However, the role and mechanism of METTL3-mediated m6A methylation in blastocyst development remains unclear. Results: We found that reduced m6A levels triggered by METTL3 knockdown caused embryonic arrest during morula-blastocyst transition and developmental defects in trophectoderm cells. Intriguingly, overexpression of METTL3 in early embryos resulted in increased m6A levels and these embryos phenocopied METTL3 knockdown embryos. Mechanistically, METTL3 knockdown or overexpression resulted in a signicant increase or decrease in expression of ATG5 and LC3 (an autophagy marker) in blastocysts, respectively. m6A modication of ATG5 mRNA mainly occurs at 3’UTR, and METTL3 knockdown enhanced ATG5 mRNA stability, suggesting that METTL3 negatively regulated autophagy in an m6A dependent manner. Furthermore, single-cell analysis revealed that METTL3 knockdown only increased expression of LC3 and ATG5 in trophectoderm cells, indicating preferential inhibitory effects of METTL3 on autophagy activity in the trophectoderm lineage. Importantly, autophagy restoration by 3MA (an autophagy inhibitor) treatment partially rescued developmental defects of METTL3 knockdown blastocysts. Conclusions: Our results demonstrate that METTL3-mediated m6A methylation negatively modulates autophagy to support blastocyst development. WT1 Associated Protein; FTO, fat mass and obesity-associated protein; ATG5, Autophagy related 5; LC3, light chain 3; ICM, inner cell mass; TE, trophectoderm; mTOR, mammalian target of rapamycin; qPCR, CDS, coding sequence; PA, parthenogenetic activation; IVF, in vitro fertilization; PCA, principal component analysis; SOX2, SRY-box transcription factor 2; CDX2, caudal type homeobox 2; OCT4, octamer-binding transcription factor 4; GV, germinal vesicle; MII, metaphase II.


Introduction
Mammalian blastocyst formation is accompanied with the formation of trophectoderm (TE) and inner cell mass (ICM) lineages [1][2][3]. Establishment of the functional TE lineage is an essential prerequisite for blastocyst and placenta development. TE cells require a larger amount of ATP than ICM cells to form and expand the blastocoel cavity [4,5]. Aberrant energy metabolism in TE cells is closely related to the impaired TE lineage speci cation and blastocyst development [6]. Thus, understanding metabolic regulatory mechanisms underlying TE cell formation is a fundamental question in mammalian blastocyst development.
Autophagy is an essential catabolic process that degrades cellular proteins and organelles into simple molecules to generate nascent energy. Autophagy process is negatively regulated by mTOR (mammalian target of rapamycin) and positively controlled through a set of autophagy-related (ATG) genes [7].
Although physiological levels of autophagy are required for mammalian early embryo development [8], excessive levels of autophagy frequently impair cell viability [9]. For instance, ATG5-deleted mouse embryos arrested at the 4-cell stage [10]. Reduced levels of autophagy prevented porcine blastocyst formation [11,12]. It was worth noting that autophagy activity was preferentially restricted to TE cells relative to ICM cells in mouse blastocysts [13]. Remarkably, autophagy induction by the inhibition of mTOR restored energy metabolism of TEAD4-de cient TE cells to allow TE lineage speci cation and blastocyst development [4]. To date, despite our current knowledge on the epigenetic mechanisms that modulate autophagy at the DNA and histone levels [14], little has been known about the ne-tuning mechanism of autophagy through post-transcriptional RNA modi cation. N 6 -methyladenosine (m6A), the most abundant internal modi cation on eukaryotic message RNAs (mRNAs), happens extensively at 5'untranslated region (5'UTR), coding sequence (CDS), and 3'UTR [15]. Recent studies showed that m6A methylation is a reversible dynamic modi cation that is mainly written by the core enzyme methyltransferase-like protein 3 (METTL3) and erased by the demethylase fat mass and obesity-associated protein (FTO) [16]. The m6A modi cation has been demonstrated at the molecular levels to regulate RNA stability, decay, translation, alternative splicing, and exportation [17]. It was recently reported that m6A methylation participated in controlling animal reproduction-associated developmental events, such as spermatogenesis [18,19], oogenesis [20], gamete maturation [21][22][23], oocyte-to-embryo transition [20,24], and blastocyst formation [25]. However, the underlying mechanism on the role of m6A in blastocyst development remains unclear. Of note, the phenomena of m6A modi cation negatively correlating with autophagy activity have been clearly documented in several different cellular contexts [26][27][28]. We thus hypothesized that METTL3-mediated m6A methylation might regulate blastocyst development via repressing autophagy activity.
In the present study, porcine embryos were used as a model to test this hypothesis. We showed that METTL3 negatively regulates expression of ATG5 mRNA in an m6A-dependent manner, thereby repressing autophagy activity to sustain porcine blastocyst development. Moreover, lineage-speci c analyses indicated that m6A modi cation preferentially inhibits autophagy activity in the trophectoderm lineage. These ndings provide new insights into the post-transcriptional epigenetic regulation of blastocyst development.
In vitro fertilization (IVF) MII oocytes were washed in the modi ed Tris-buffered medium (mTBM) containing 2 mg/mL BSA and 2 mM caffeine. Approximately 15 oocytes were incubated in 50 µL droplets of mTBM for 4 h at 38.5 °C in 5% CO 2 in air. Semen from two boars was mixed and centrifuged at 1900 g for 4 min in DPBS supplemented with 1 mg/mL BSA (pH 7.3). Then, sperm was resuspended with mTBM to a concentration of 1 × 10 6 cells/mL. Fifty microliters of sperm solution were added to the mTBM droplets containing oocytes. After co-incubation of oocytes and sperm for 6 h, excess sperm surrounding oocytes were washed out and presumptive zygotes were cultured in PZM-3 at 38.5 C in 5% CO 2 in air.
In vitro transcription METTL3-EGFP mRNA used for microinjection was synthesized in vitro. pIVT-METTL3-EGFP plasmid containing T7 promoter was linearized by digestion with BspQI. Linearized DNA templates were puri ed using a DNA clean & concentrator Kit (ZYMO RESEARCH, D4003, Tustin, CA, USA). According to the manufacture's manual, in vitro transcription of METTL3-EGFP mRNA was performed through using mMESSAGE MACHINE TM T7 kit (Ambion, AM1344, shanghai, China) and Poly (A) Tailing Kit (Ambion, AM1350, Shanghai, China). Then, mRNA was treated with TURBO Dnase to remove the DNA templates and was further puri ed using MEGAclear Kit (Ambion, AM1908, Shanghai, China). After dissolving mRNA in RNase-free water, mRNA concentration was determined by a Nanodrop instrument (Thermo Scienti c, Shanghai, China) and was then aliquoted and stored at -80 °C.
Microinjection siRNA species were designed to target three different sites of the porcine METTL3 coding region (GenePharma, Shanghai, China). Information on the siRNA sequences used in this study was listed in Table 1. Microinjection was performed in a T2 (TCM199 with 2% FBS) medium containing 7.5 µg/ml Cytochalasin B on an inverted microscope (Olympus, Japan). Approximately 10 pl of siRNA solution (50 µM) was microinjected into the cytoplasm of MII oocytes. Embryos were cultured in PZM-3 medium for 7 days.

Identi cation of TE and ICM blastomeres of blastocysts
Zona pellucida of blastocysts was removed by digestion of 3.3 mg/mL pronase for 3 min. Zona-free blastocysts were incubated in 0.25% trypsin for 40 min. Individual blastomeres were isolated by repeated pipetting of blastocysts with glass needle at 100 µm in diameter. Individual blastomeres were lysed using single cell quantitative kit. Samples were pre-ampli ed for 20 cycles according to the manufacture's protocol. The relative levels of SOX2 and CDX2 mRNA were detected by single-cell qPCR. Then, data were further analyzed by principal component analysis (PCA) using SIMCA14.1 software [29]. Blastomeres from TE and ICM were identi ed according to the clustering analyses.
mRNA stability Embryos at the morula stage were treated with 25 µg/mL α-amanitin (MCE, HY-19160) to inhibit global mRNA transcription. Embryos were collected at 0, 12, 24, 36, 48, and 60 h and total RNA was extracted for reverse transcription. The levels of ATG5 mRNA were determined using single-cell qPCR.

Immuno uorescence staining
Oocytes and embryos were xed in 4% paraformaldehyde solution for 15 min, permeabilized with 1% Triton X-100 for 30 min at room temperature (RT) and then blocked with 2% BSA at RT for 1 h. Samples were incubated in solution containing primary antibodies overnight at 4 °C. After washing, samples were incubated for 1 h in solution containing secondary antibodies in the dark at 37 °C. Afterwards, samples were counterstained using 4, 6-diamidino-2-phenylindole dihydrochloride or propidium iodide for 10 min and were then loaded onto glass slides. Finally, samples were imaged using laser scanning confocal microscopy (Olympus, Japan). The average pixel intensity of embryos was determined using Image J. Information regarding primary and secondary antibodies used was listed in Table 3.

Statistical analysis
All experiments were carried out at least three times. Data were analyzed using one-way ANOVA or student's t test (SPSS 17.0) and were presented as mean ± standard error of mean (mean ± S.E.M). P < 0.05 was considered to be statistically signi cant.

Characterization of m6A methylation and its writers in early embryos
To investigate the kinetics and subcellular localization of m6A methylation in early embryos, immuno uorescence staining was performed to characterize dynamic patterns of m6A modi cation. The speci city of commercially available m6A antibody in porcine embryos had been veri ed by RNase A treatment (Additional le 1a and b). The results in PA embryos revealed that m6A levels in oocytes and blastocysts appeared to be higher than that in embryos from 2-cell to morula stage, and m6A methylation occurred in cytoplasm from GV oocyte to morula stage whereas m6A modi cation are present in nucleus and cytoplasm of blastocysts (Fig. 1a). Similarly, the highly dynamic patterns and localization of m6A modi cation were also observed in IVF embryos (Fig. 1b). To determine whether transcripts encoding m6A writers including METTL3, METTL14 and WTAP are expressed in early embryos, qPCR was performed to examine its relative abundance. We found a persistent expression of the three genes during meiotic maturation and subsequent early embryo development, suggesting its maternal and zygotic origins (Fig. 1c, d and e). However, the expression levels of genes were a little lower through morula to blastocyst stage relative to oocytes and early-cleavage embryos (Fig. 1c, d and e) (P < 0.05). Therefore, these results indicate m6A modi cation and its writers are dynamically present in early embryo development.

METTL3 knockdown impedes blastocyst development and trophectoderm lineage formation
To explore the biological role of endogenous METTL3 in early embryo development, siRNAs against METTL3 were microinjected into MII oocytes. Meanwhile, noninjected and negative control (NC) siRNAinjected MII oocytes served as control groups. qPCR results showed that METTL3 siRNA injection signi cantly reduced the levels of METTL3 mRNA at the 4-cell stage compared to the control groups ( Fig. 2a) (P < 0.05). Unfortunately, we did not directly analyze the in uence of METTL3 siRNA injection on METTL3 protein levels because of the lack of porcine speci c METTL3 antibodies. Alternatively, METTL3-EGFP mRNA and siRNA were coinjected into oocytes served as experimental group, noninjected, EGFP mRNA or METTL3 siRNA injection alone served as control groups. Fluorescence intensity analyses revealed that METTL3 siRNA injection signi cantly reduced the levels of METLL3 protein at the 4-cell stage compare to the control groups (Fig. 2b)(P < 0.05). Furthermore, a subset of embryos at 2-cell, 4-cell and blastocyst stage were isolated and subjected to immuno uorescence staining to detect m6A levels. METTL3 siRNA injection signi cantly reduced m6A levels at the aforementioned stages compared to the control groups (Fig. 2c, d and e)(P < 0.05). Of note, no differences in the levels of METTL3 mRNA, protein and m6A were observed between the NC siRNA injected and uninjected control groups.
To determine whether METTL3 knockdown (referred to as METTL3 KD) affected early embryo development, the developmental rates of METTL3 KD embryos were compared to NC siRNA and uninjected embryos. We found that METLL3 KD had no in uence on development to 2-cell, 4-cell, 8-cell and morula stage (additional le 2), but compromised the development to blastocyst stage (Day 5-7) compared to the control groups ( Fig. 2f and g)(P < 0.05). A small proportion of METTL3 KD embryos developed to the blastocyst stage (Fig. 2f), we then analyzed the lineage allocation in these blastocysts. Blastocysts were stained with a CDX2 antibody to label the TE cells (Fig. 2h). The number of inner cell mass (ICM) cells was indirectly determined by subtracting the TE number from the total cell number. The results revealed that ICM cell number did not change between METTL3 KD and the control groups (Fig. 2i). However, METTL3 KD resulted in a signi cant reduction in both total cell number and TE cell number (Fig. 2i)

METTL3 negatively regulates blastocyst development and perturbs normal lineage allocation
We next investigated whether overexpression of METTL3 could affect embryonic development. METTL3-EGFP mRNA was introduced into the cytoplasm of MII oocytes. Uninjected and EGFP mRNA-injected MII oocytes served as control groups. qPCR and uorescence intensity analyses were performed to determine the relative expression of METTL3 mRNA and protein in the subset of embryos at the 2-cell, 4-cell and blastocyst stage. The results showed that METLL3-EGFP mRNA injection indeed induced a higher expression of METTL3 mRNA (Fig. 3a, b and c) and protein (Fig. 3d, e and f) at the aforementioned stages compared to the control groups (P < 0.05). To determine the effect of METTL3 overexpression (referred to as METTL3 OE) on early embryo development, the developmental rates of embryos were analyzed. Unexpectedly, METTL3 OE did not affect the development rate of 2-cell embryos ( Fig. 3g and  h), but signi cantly reduced blastocyst formation rate (Day 7) compared to the control groups ( Fig. 3g and i) (P < 0.05). In addition, METTL3 OE resulted in a signi cant reduction in the number of total cells, ICM cells, and TE cells (Fig. 3j and k) (P < 0.05). However, the ratio of ICM cells to TE cells in the METTL3 OE blastocysts did not change compared to the control groups (Fig. 3k). Altogether, these results demonstrate that overexpression of METTL3 impaired blastocyst development and normal lineage allocation.
METTL3 negatively regulates autophagy activity of early embryos in an m6A-dependent manner METTL3 was recently reported to negatively regulate autophagy in cardiomyocytes [26], which prompted us to investigate whether METTL3 regulates embryo development via autophagy. Morula was then subjected to qPCR analysis to examine the relative expression of several key genes involved in autophagy and apoptosis. METTL3 KD caused a signi cant increase in expression of ATG5, BECLIN1, and CASPASE3 ( Fig. 4a) (P < 0.05), whereas METTL3 OE induced decrease of ATG5 expression (Fig. 4b) (P < 0.05), This suggested a potential involvement of METTL3 in controlling AGT5 expression in porcine embryos. To verify the effect of METTL3 on ATG5 protein expression, we detected the ATG5 by immuno uorescence staining. The speci city of the commercially available ATG5 antibody was rst con rmed in porcine embryos (additional le 1c). Immuno uorescence staining showed that METTL3 KD signi cantly increased ATG5 protein levels ( Fig. 4c) (P < 0.05) whereas METTL3 OE decreased ATG5 protein levels in morula (Fig. 4d) (P < 0.05). In addition, TUNEL staining revealed that METTL3 KD did not increase apoptotic cell number (additional le 3a and b), but caused a signi cant increase in apoptosis ratio of the resulting blastocysts (additional le 3c) (P < 0.05).
Given that ATG5 acts as an upstream core regulator of autophagy pathway, we speculated that METTL3 might affect autophagy activity through regulating ATG5 expression. To address this, protein levels of LC3 (Light chain 3), an autophagy marker, were measured by immuno uorescence to determine autophagy activation in embryos. The speci city of LC3 antibody was validated in preliminary experiments (additional le 1d). Of note, METTL3 KD or OE led to a signi cant increase in protein levels of LC3 at the morula stage compared to the control groups, suggesting an elevation of autophagy activity ( Fig. 4e and f) (P < 0.05). Taken together, these data support that METTL3 KD or OE impaired the ATG5 expression and enhanced the autophagy activity in porcine embryos.
In the next set of experiments we examined whether METTL3 regulates ATG5 expression via m6A modi cation of ATG5 mRNAs. A web-based application tool called SRAMP was used to predict m6A sites in ATG5 mRNA. As shown in Fig. 4g, m6A modi cations are primarily present at 3'UTR of ATG5 mRNA. To examine whether METTL3-mediated m6A modi cation regulated ATG5 expression through modulating mRNA decay, we conducted the analysis of ATG5 mRNA stability in embryos. The results revealed that METTL3 KD enhanced stability of the preexisting ATG5 mRNA compared to the control group (Fig. 4h) (P < 0.05), suggesting that METTL3 regulates ATG5 expression via mediating its mRNA stability. Collectively, these data indicate that METTL3 negatively regulating autophagy activity of early embryos depends on decay of m6A-modi ed ATG5 mRNA.

METTL3-mediated m6A methylation exerts inhibitory effects on autophagy activity in the trophectoderm lineage
It was reported that autophagy was preferentially restricted to the TE lineage in mouse blastocysts [13] and METTL3 speci cally regulated TE cell proliferation in porcine blastocysts. Thus, we hypothesized that METTL3 might exert TE lineage-speci c inhibitory effects on autophagy activity in porcine blastocysts. To test this hypothesis, immuno uorescence was performed to examine the localization and levels of m6A and LC3 in the ICM and TE lineages. The results showed that METTL3 KD reduced m6A levels in both ICM and TE lineages (Fig. 5a). LC3 levels appeared to be only decreased in the TE lineage of METTL3 KD blastocysts whereas it seemed to be not changed in the ICM lineage between the control and METTL3 KD groups (Fig. 5b), suggesting a preferential role of METTL3 in autophagy in the TE lineage. To further con rm the differential effects of METTL3 on autophagy activity between ICM and TE lineages, blastocysts in each group were separated into individual blastomeres (Fig. 5c), which in turn were subjected to quantitative analyses of SOX2 and CDX2 mRNA to identify the ICM and TE cells (Fig. 5d). Meanwhile, single cell qPCR revealed that METTL3 KD resulted in a signi cant increase in expression levels of ATG5 and BECLIN1 in the TE cells (Fig. 5e) (P < 0.05), but did not affect expression of the two genes in the ICM cells (Fig. 5f). Interestingly, METTL3 KD led to a signi cant reduction in expression levels of BECLIN1 in the ICM cells (Fig. 5f) (P < 0.05). Together, these date imply that METTL3mediated m6A methylation preferentially inhibits autophagy activity in the trophectoderm lineage.

Autophagy inhibitor partially rescues development and quality ofMETTL3knockdown embryos
To clarify if autophagy mediates the effect of METTL3 on embryo development, we reduced autophagy level by autophagy inhibitor 3MA and examined if this could rescue the defects of embryo development.
3MA was added into culture medium. Untreated and METTL3 KD embryos served as the control groups. Immuno uorescence analyses showed that 3MA supplement reduced the autophagy in METTL3 KD blastocysts to a level similar to that in the untreated control group (Fig. 6a and b) (P < 0.05). Of note, 3MA supplement partially rescued the defects of blastocyst formation by METTL3 KD (Fig. 6c and d). Moreover, total cell number of blastocysts was signi cantly increased in 3MA treatment group compared to METTL3 KD group (Fig. 6e) (P < 0.05). Therefore, these data indicate that restoration of autophagy levels partially rescued development and quality of METTL3 KD embryos.

Discussion
A recent study reported an essential role of METTL3-mediated m6A methylation in blastocyst formation in mice [25], but its molecular mechanisms underlying blastocyst development are yet to be known. Our data in porcine embryos indicate that METTL3-mediated m6A methylation sustains blastocyst development via repressing autophagy activity. METTL3 negatively regulates expression of ATG5 mRNA in the TE lineage in an m6A-dependent manner. Therefore, we propose a working model in which METTL3 supports blastocyst development through negatively regulating expression of m6A-modi ed ATG5 mRNA that is required for autophagy activation (Fig. 7). To our knowledge, this work represents the rst report characterizing the epitranscriptomic regulatory mechanism underlying mammalian blastocyst development.
The physiology levels of autophagy are indispensable for blastocyst development in mice [30] and pigs [11]. On the contrary, excessive levels of autophagy are detrimental to blastocyst development [31][32][33]. In this study, we showed that KD or enforced expression of METTL3 severely impaired porcine blastocyst development. In addition, METTL3 loss or gain of function resulted in a signi cant increase or reduction in ATG5 expression and autophagy levels, respectively. It is thus possible that METTL3 mediated steady state of autophagy to allow blastocyst development in pigs. However, restoration of autophagy levels by the inhibitor treatment only partially rescued blastocyst development of METTL3-knockdown embryos, suggesting the existence of redundant mechanisms of METTL3 regulating blastocyst development.
Interestingly, previous studies in mice showed that maternal ATG5 was required for development to 4-cell embryo stage [10] whereas zygotic ATG5 was dispensable for normal embryo development [34]. In pigs, METTL3 overexpression dramatically decreased ATG5 expression and blastocyst rate, but did not affect development to 2-cell and morula stage. This discrepancy could be due to the diverse functions of ATG5 between different species.
Previous study in mice indicated that TE cells in blastocysts possessed a higher autophagy activity than ICM cells [13]. The high autophagy of TE cells met its requirement for much more energy [5]. Inhibition and induction of autophagy resulted in defects in the TE and ICM, and failure to separate the ICM and TE cells in mice [30]. Similarly, in pigs, repression of basal autophagy decreased total and TE cell number in blastocysts [33] and increased autophagy levels also disrupted proliferation and differentiation of TE cells [35]. Consistent with these studies, we found in this study that MELLT3 KD or OE led to an increased or decreased autophagy, and caused fewer total cells and TE cells. Importantly, correction of autophagy levels could partially restore the total cell number in METTL3 KD blastocysts. Therefore, we reasoned that abnormal autophagy levels at least partially accounted for TE lineage defects induced by METTL3 dysfunction in porcine blastocysts.
In this study, we observed that the levels of ATG5 mRNA were increased or decreased upon METLL3 KD or OE. ATG5 binds to ATG12 and ATG16L to facilitate the conjugation of autophagy marker LC3 and autophagosome [36]. Thus, we identi ed a corresponding increase or decrease in the LC3 levels in METLL3 KD or OE embryos. Studies indicated that m6A modi cation has been strongly linked to increased mRNA degradation [37]. Our data further showed that m6A modi cation mostly happened at 3'UTR of ATG5 mRNA and m6A levels were increased or decreased upon METTL3 KD or OE. Thus, these data demonstrate that METTL3 negatively regulates expression of ATG5 mRNA in an m6A-dependent manner. Based on these analyses, METTL3 suppresses autophagy likely by destabilizing ATG5 at the transcript level.

Conclusions
Our ndings show that METTL3-mediated m6A methylation negatively regulates autophagy to support blastocyst development in pigs. Our results may provide new insights into the function and potential mechanisms of METTL3 and m6A modi cation in regulating autophagy and mammalian blastocyst development.     Prediction of m6A methylation sites in ATG5 mRNA. m6A methylation sites in full-length sequence of ATG5 mRNA was predicted using a web tool (called as SRAMP) (f) Lifetime of ATG5 mRNA in control and MEKKL3 KD embryos. The relative abundance of ATG5 mRNA was determined by qPCR. Asterisks indicate signi cant differences (p < 0.05). (h, i) Expression of LC3 protein in blastocysts. Blatocysts from the indicated groups were stained for LC3 and DNA. The uorescence intensity was quantitatively analyzed by image J (right panel). The experiment was independently repeated three times with at least 20 blastocysts per group. Scale bar: 50 µm. All data are shown as mean ± S.E.M and different letters on the bars indicate signi cant differences (P < 0.05).   Working model illustrating how METTL3-mediated m6A methylation negatively regulates autophagy to support porcine blastocyst development In the normal TE cells, METTL3 generates proper m6A levels of ATG5 mRNA, thereby maintaining basal translation of ATG5 mRNA and normal ATG5 protein levels, sustaining appropriate expression of LC3 protein, physiological levels of autophagy and normal blastocyst development. On the contrary, METTL3 knockdown or overexpression separately caused inadequate or excessive m6A levels of ATG5 mRNA, leading to high or low translation of ATG5 mRNA, thereby resulting in high or low LC3 protein expression and autophagy levels and hindered the morula-toblastocyst transition. Collectively, METTL3-mediated m6A methylation negatively modulates autophagy in TE cells to sustain normal blastocyst development.