N 6 -Methyladenosine Sequencing Highlights the Involvement of mRNA Methylation in Oocyte Meiotic Maturation and Embryo Development by Regulating Translation in Xenopus laevis

During the oogenesis of Xenopus laevis , oocytes accumulate maternal materials for early embryo development. As the tran-scriptionactivityoftheoocyteissilencedatthefullygrownstage and the global genome is reactivated only by the mid-blastula embryo stage, the translation of maternal mRNAs accumulated during oocyte growth should be accurately regulated. Previous evidence has illustrated that the poly(A) tail length and RNA binding elements mediate RNA translation regulation in the oocyte. Recently, RNA methylation has been found to exist in various systems. In this study, we sequenced the N 6 -methylad-enosine(m 6 A)modifiedmRNAsinfullygrowngerminalvesicle-stage and metaphase II-stage oocytes. As a result, we identified 4207 mRNAs with m 6 A peaks in germinal vesicle-stage or metaphase II-stage oocytes. When we integrated the mRNA methylation data with transcriptome and proteome data, we found that the highly methylated mRNAs showed significantly lower protein levels than those of the hypomethylated mRNAs, although the RNA levels showed no significant difference. We also found that the hypomethylated mRNAs were mainly enriched in the cell

. After that, the fully grown GV oocytes resume meiosis and develop to metaphase of the second meiosis (MII) stage. Then the oocyte is fertilized by sperm to form a zygote. The zygote starts mitosis, and embryo development is initiated. As embryo genomic transcription is not reactivated until embryo development to the mid-blastula stage for X. laevis (3) or the 2-cell to 16-cell stages for mammals (4 -7), oocytes or embryos need the maternal RNAs accumulated during oocyte growth to support new protein synthesis.
As the synthesis of new proteins such as cyclins is of importance for the meiotic and mitotic events, the translation of maternal mRNAs during oocyte meiotic maturation and early embryo development should be precisely controlled. Previous data have shown that there are mainly two methods for cells to control the maternal mRNA translation in oocyte or early embryo protein binding to the cytoplasmic polyadenylation elements (CPEs) in maternal mRNAs and controlling the polyadenylation of the mRNA poly(A) tails (8). In most cases, shortened poly(A) tails of mRNAs repress their translation, whereas elongated poly(A) tails activate translation (9). The poly(A) tail length of oocyte mRNA is mainly associated with cytoplasmic polyadenylation, which is controlled by RNA binding proteins and associated proteins. The maternal mRNAs, whose 3Ј UTR contains a cis-element CPE, could be bonded by CPE binding protein (CPEB) (10). When CPEB is phosphorylated under stimulation of progesterone, it recruits and binds to the cleavage and polyadenylation-specific factor. The cleavage and polyadenylation-specific factor then recruits poly(A) polymerase to the mRNA end and mediates poly(A) tail elongation and promotes mRNA translation (11).
In addition to the poly(A) tail and CPE-mediated mRNA translation regulation, recent evidence revealed a correlation between RNA translation and an RNA-specific modification called N 6 -methyladenosine (m 6 A) (12,13). It has been shown that m 6 A modifications have different effects on mRNA translation when the m 6 A modifications occur at different regions of the mRNAs (14 -16). RNA methylation at the 5Ј UTR could promote mRNA translation, whereas RNA methylation at the last exons participated in the 3Ј UTR regulation of mRNAs (15,16). By knocking out the m 6 Awriter Mettl3, mRNA methylation was also found to play important roles in other cellular events such as cell differentiation and RNA degradation (17).
As transcription is silenced in fully grown oocytes and early embryos, mRNA translation regulation is essential for biological events during this period. To investigate whether m 6 Amodification takes part in RNA translation regulation in the X. laevis oocyte and its potential roles in oocyte maturation and embryo development, we sequenced m 6 A-modified mRNAs in fully grown GV-stage and MII-stage X. laevis oocytes and compared the m 6 A-seq data with the transcriptome and proteome data.
Results m 6 A-Seq of GV-and MII-stage Oocytes in X. laevis-m 6 Amodified mRNAs of GV-and MII-stage oocytes from X. laevis (supplemental Fig. S1) were isolated and analyzed according to the methods described by Dominissini et al. (18). After mapping the methylated RNA fragments to the transcriptome of X. laevis (X. laevis v7.1 from Xenbase) (19), we found that 4207 mRNAs (4128 in GV oocytes and 3820 in MII oocytes) were methylated in GV-or MII-stage oocytes (supplemental Dataset S1).
According to the height of the m 6 Apeaks, we divided these mRNAs into three classes: m 6 A high mRNAs, m 6 A medium mRNAs, and m 6 A low mRNAs. From these results we found that the m 6 A modification was maintained in 1674 mRNAs during oocyte maturation from GV to MII stage, but in 2400 mRNAs the m 6 A levels were decreased, and in 133 mRNAs the m 6 A levels were increased (Fig. 1A).
Using the m 6 A peak-corresponding mRNA sequences, we predicted the conservative m 6 Amotifs in X. laevis oocytes. As described for human and mouse cells, mRNA methylation in X. laevis oocytes also occurred at the GGACU motifs (Fig. 1B). By analyzing the m 6 A peak positions along the mRNAs, we found that the m 6 A peaks were mainly distributed downstream of the coding DNA sequence (CDS) start sites and around the CDS end sites (Fig. 1C).
To analyze the association between m 6 A peak height and mRNA level in oocytes, we integrated the m 6 A-seq data with the transcriptome data published by Charlier et al. (20) (supplemental Figs. S1 and S2). From the transcriptome data, we found 1030 genes with detected m 6 A modifications. The m 6 A levels of these genes decreased from GV stage to MII stage (p Ͻ 0.01, supplemental Fig. S2, A and B). In addition, we also found that there was no obvious relativity between the m 6 A peak heights and the mRNA levels (supplemental Fig. S2, C and D). Next we compared the ratios of m 6 A peak height/mRNA level in GV and MII oocytes and found that most ratios of m 6 A peak height/mRNA level were decreased from GV stage to MII stage (supplemental Fig. S2, E-H).
Gene Set Enrichment Analysis of the Methylated mRNAassociated Pathways in X. laevis Oocytes-To know whether RNA methylation participates in oocyte maturation and embryo development, we analyzed the m 6 A -modified, mRNA-enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways through the DAVID tools (21). From the results (supplemental Dataset S2), we could see that the highly or moderately methylated mRNAs mainly enriched in pathways like ErbB signaling, progesterone-mediated oocyte maturation, and cell cycle. However, the hypomethylated mRNAs mainly enriched in pathways like RNA degradation, DNA replication, ribosome, spliceosome, pentose phosphate pathway, and glycolysis/gluconeogenesis.
To further assess the biological functions of these m 6 A-modified mRNAs in Xenopus oocytes, we annotated the Xenopus mRNAs using BLAST on the Biocloud platform. From the gene ontology annotation results, we could find that the oocyte m 6 Amodified mRNAs were mainly associated with biological processes like transcription, protein phosphorylation, and cell division. Interestingly, in the m 6 A-modified mRNAs, there were 443 whose proteins had ATP binding functions, and 354 had zinc ion binding functions (supplemental Dataset S3).
The Relationship between mRNA Methylation and mRNA Translation-To investigate whether mRNA methylation is related to mRNA translation, we integrated our mRNA methylation data and the transcriptome and proteome data from Smits et al. (22). In the transcriptome and proteome data, Smits et al. (22) revealed the RNA and protein levels in the egg and embryo (supplemental Fig. S1) of X. laevis. Through BLAST, the information of 723 m 6 A-modified mRNAs were identified in the RNA/protein profile data. From the profile data, we found that the RNA levels (log10 RNA RPKM values) of high m 6 A peak mRNAs, detected in GV or MII stage oocytes, were higher than the levels of the total RNAs (in eggs or embryos); however, the protein levels (log10 protein amount/femtomole) were lower than the levels of total proteins (Fig. 2). Compared with the highly methylated mRNAs, the hypomethylated RNAs showed significantly higher protein levels ( Fig. 3; p values shown in supplemental Dataset S4). These results suggest that m 6 A modification may be associated with mRNA translation in oocytes or even in early embryos.
To analyze whether the mRNA positions (5Ј UTR, CDS, and 3Ј UTR) of RNA methylation are associated with translation, we extracted mRNAs whose m 6 A modification only occurred at the 5Ј UTR, 5Ј-terminal CDS, 3Ј-terminal CDS, or 3Ј UTR from the highly methylated RNAs in GV oocytes ( Fig. 4 and supplemental Fig. S3). In Fig. 4 we can see that the mRNAs whose 5Ј-terminal CDSs or 3Ј-terminal CDSs were methylated showed lower protein levels compared with the other sets. In addition, mRNAs whose m 6 A modification occurred at the 3Ј UTR, in contrast, showed higher median protein levels (p values are shown in supplemental Dataset S4). These results suggest that CDS region RNA methylation may suppress mRNA translation.
In addition, we also integrated our data with the X. laevis embryo (supplemental Fig. S1) proteome data published by Peshkin et al. (23). As a result, we also found that the hypermethylated RNAs in eggs have a lower protein concentration distribution. As for the hypermethylated RNAs, these RNAs, whose m 6 A modification mainly occurred in CDS regions, showed a lower protein level distribution (supplemental Fig.  S4). From the protein profile data calculated by Peshkin et al. (23), we found that hypermethylated RNAs showed a lower protein synthesis rate (supplemental Fig. S5).

Discussion
When X. laevis oocytes develop to the fully grown GV stage, their transcription activity is silenced. After maturation (about 6 h), MII-stage oocytes are fertilized by spermatozoa, and the oocytes are transformed into fertilized eggs. The fertilized eggs then develop to mid-blastula stage embryos (50 -90 min), at which stage the embryonic genome is reactivated, and global transcription is initiated. From the fully grown GV oocytes to the mid-blastula embryos, oocyte maturation and embryo development mainly rely on the maternal factors accumulated in oocytes (24). For oocyte mRNAs, their translation is con-trolled strictly to satisfy the development requirements. Previous evidence showed that oocyte mRNA translation is controlled by two mechanisms: poly(A) tail length-correlated translation efficiency and 3Ј UTR element-mediated mRNA translation activation. Recently, evidence suggested a possible association of translation with RNA methylation (12,16,25), indicating that the 5Ј UTR m 6 A modification could initiate the translation of non-capped mRNAs. Through integrating the m 6 A-seq data and transcriptome/proteome data in X. laevis, we investigated m 6 A modification and possible roles of oocyte mRNAs. We found that mRNAs in which the CDS region was m 6 A-modified showed low protein levels, but mRNA with an m 6 A-modified 3Ј UTR showed high protein levels. Our results suggest that RNA methylation is involved in RNA translation regulation in X. laevis oocyte maturation and further embryo development. Thus, we reveal a mechanism regulating maternal mRNA dormancy or translation in early development. Our data also suggest that m 6 A may play different roles according to their positions. The correlation of CDS region m 6 A modification and low protein level may suggest that m 6 A modification represses the translation elongation of mRNAs.
Our results also showed a decrease in RNA methylation during oocyte development from the GV stage to MII stage. As there was no transcription activity in these oocytes, we inferred that there were two events that may induce this RNA methylation decrease: active RNA demethylation and RNA degradation in the MII stage. The RNA methylation pattern changes during oocyte meiosis may also be a mechanism for the oocyte to control mRNA translation. Recent evidence has shown that, in addition to Mettl3, there were other enzymes, like Mettl14 (26), Wilms tumor 1-associated protein (Wtap) (27), and KIAA1429 (28), also involved in the RNA methylation process. On the other hand, two m 6 A erasers, fat mass and obesity-associated (Fto) (29) and alkB homolog 5 RNA demethylase (Alkbh5) (30), have been found to be involved in m 6 A modification in mammalian cells (31). These findings suggest that the regulation of RNA methylation in eukaryotic cells is finely regulated through these m 6 A modification-associated factors. Microarray data about mouse oocyte development (32) have shown that both the m 6 A writers Mettl3 and Mettl14 and the m 6 A eraser Alkbh5 were expressed highly in growing oocytes, indicating that mRNA m 6 A modifications may exist in a dynamic pattern during oocyte growth. In any case, the functions of these m 6 Aassociated enzymes and cofactors should be further analyzed to reveal the regulation of RNA methylation in oocytes and their biological functions.
In our study, we found numerous oocyte meiosis-and early embryo development-associated genes whose mRNAs are methylated. For example, cyclin B2 (ccnb2) is required for X. laevis oocyte bipolar spindle formation, and evidence showed that its protein levels increased in MII-stage oocytes (33). mos is also essential for X. laevis oocyte meiosis resumption (34) and MII stage maintenance (35,36). Our data showed that the m 6 A modification in ccnb2 and mos mRNAs decreased in the 3Ј terminal of the CDS region but showed no obvious change at the 5Ј UTR (supplemental Fig. S6). cdt1, whose m 6 A modification decreased at the CDS region, has been proven to play roles in DNA replication in embryos (37)(38)(39). The m 6 A modification changes of these key factors indicate that RNA methylation is involved in meiosis regulation and early embryo development.

Experimental Procedures
X. laevis Oocyte Collection-Feeding and handling of X. laevis were conducted in accordance with the Animal Research Committee policies of the Institute of Zoology, Chinese Acad- emy of Sciences. Sexually mature X. laevis females were injected with 50 IU of pregnant mare serum gonadotropin 5 days before collection of GV-stage oocytes. Before the surgery, the X. laevis was buried under ice for about 30 min until it reached complete anesthesia. After making a small incision off the middle line of the belly, a part of the ovary was removed with forceps and scissors and placed in M199-HEPES medium. We then sutured the incision and let the X. laevis recover under a moist tissue until fully awake, and then individuals were submerged in water supplemented with penicillin. The excised ovary was cut into small pieces and transferred to a 50-ml tube containing 40 ml of M199-HEPES with 0.2% collagenase type I 4These oocytes were washed with fresh M199-HEPES and aliquoted in 1.5-ml tubes. After snap-freezing in liquid nitrogen, these oocytes were stored at Ϫ80°C until use. For the ovulation of MII eggs, the X. laevis was injected with 500 IU of human chorionic gonadotropin. After ovulation, we collected the healthy MII oocytes and aliquoted the oocytes in 1.5-ml tubes. After snap-freezing in liquid nitrogen, these oocytes were stored at Ϫ80°C until use.
RNA Preparation-Total RNAs of GV-or MII-stage oocytes were extracted using TRIzol reagent (Life Technologies). To avoid DNA contamination, all samples were treated with Turbo DNase (Life Technologies). RNA samples were chemically fragmented into about 100-nt fragments by incubating in fragmentation buffer (10 mM ZnCl 2 and 10 mM Tris-HCl (pH 7)) for 15 min at 94°C. The fragmentation reaction was stopped by adding 0.05 M EDTA, and the RNAs were collected by standard ethanol precipitation. m 6 A-Seq-m 6 A sequencing was performed as described previously Dominissini et al. (18) with little modification. Briefly, 200 l of protein A bead slurry (Thermo Fisher Scientific) was blocked by incubating at room temperature for 1 h with 1 ml of immunoprecipitation buffer (150 mM NaCl, 0.1% Igepal CA-630, and 10 mM Tris-HCl (pH 7.4)) supplemented with 50 mg/ml BSA and 200 units of RNasin (Promega). Then the fragmented RNAs (3 mg) were incubated at 4°C for 1 h with 50 l of blocked protein A beads to reduce nonspecific binding. After centrifugation at 4°C, 2500 ϫ g for 3 min, we collected the RNA supernatant and incubated it at 4°C for 2 h with 12.5 g of purified rabbit anti-m 6 A polyclonal antibody (Synaptic Systems). The RNA/antibody mixture was incubated with 50 l of blocked protein A beads for an additional 2 h at 4°C. After extensive washing, the bound RNAs were eluted by competition with 0.5 mg/ml N 6 -methyladenosine (Sigma-Aldrich). The eluted RNAs were precipitated with ethanol and resuspended  in nuclease-free water. Finally, the RNA library was generated with an mRNA sequencing kit (Illumina) and sequenced on an Illumina HiSeq2000 platform. The raw data of the sequencing reads were uploaded to the sequence read archive (accession no. SRR3667693). m 6 A-Seq Data Analysis-The X. laevis oocyte m 6 A-seq data were analyzed as described in the protocol by Dominissini et al. (18) with some modifications. Briefly, the RNA-seq reads were mapped to the mRNAs of X. laevis collected by Xenbase (19) using Bowtie software (40). The Bowtie-generated SAM files were treated using MACS software (41) to identify the m 6 A peak positions in the RNAs. The 100-nt RNA fragments including methylation sites were extracted and analyzed with MEME software (42) to construct the conservative m 6 A motif in X. laevis. To analyze the m 6 A sites distribution, we extracted the UTR and CDS information of X. laevis mRNAs from NCBI GenBank. To visualize the m 6 A peaks in RNAs, we wrote Python scripts and displayed the m 6 A peaks in SVG format. For the RNAs with m 6 A peaks, according to the height of the peak, we divided the m 6 A-modified RNAs into three classes: high (read depth Ͼ 60), medium (read depth between 30 and 60) and low (read depth Ͻ 30) methylated RNAs. To annotate the X. laevis mRNAs, we blasted the mRNA sequences to the GO dataset using Annofunction software on the Biocloud website.
Integration of m 6 A-Seq Data with Transcriptome and Proteome Data-The X. laevis oocyte and embryo omic data published by Smits et al. (22) and Peshkin et al. (23) were used for integration analysis of oocyte m 6 A data. As different reference genome or gene IDs were used in different research groups, we searched the corresponding gene ID using BLAST. All gene expression data we used were calculated by the authors themselves.