The RNA m 6 A Reader YTHDF2 Is Essential for the Post-transcriptional Regulation of the Maternal Transcriptome and Oocyte Competence

1RNA N6-methyladenosine (m6A) is the most abundant internal mRNA modification that is a key determinant of post-transcriptional mRNA regulation, with proven functions in mRNA processing, translation, and degradation. m6A is found enriched within the METTL3/14 methyltransferase RRACH (where R = G/A and H = A/C/U) consensus in the 3′UTR near the stop codon. The outcome of RNA methylation is instructed through changes in tertiary structure that recruit or displace defined RNA-binding proteins or by directly increasing the affinity of the binding site for YTH-domain-containing proteins. The mouse genome encodes five YTH domain-containing proteins, one nuclear (YTHDC1) and four normally cytoplasmic (YTHDF1–YTHDF 3 and YTHDC2). YTHDC1 regulates splicing and nuclear export. While YTHDF1/3 recruitment enhances translation, the binding of YTHDF2 to m6A mRNA within the GACU/A consensus is associated with its destabilization and degradation. YTHDF2 is required for early zebrafish embryonic development. RNA m6A is required for mouse embryonic stem cell exit from pluripotency and preimplantation development; however, the in vivo and developmental contribution of the m6A mRNA-YTHDF2-mediated mRNA destabilization pathway in mammals remains unknown.

2Transcription and translation are uncoupled during defined stages of gametogenesis and early zygotic development; thus the regulation of gene expression occurs at the post-transcriptional level. In spermatogenesis, the lepto-zygotene stages of meiotic prophase as well as the latter part of spermiogenesis are transcriptionally inert. Indeed, ALKBH5, an m6A RNA demethylase, is required for normal mouse spermatogenesis. The maternal transcriptome is assembled during the growth phase of oogenesis and is completed with the cessation of transcription in full-grown prophase I germinal vesicle (GV) oocytes. Post-transcriptional regulation and utilization of the maternal transcriptome underpin meiotic maturation, fertilization, and early embryonic development. Oocyte maturation is hormonally triggered and occurs just prior to ovulation when GV oocytes complete meiosis I and advance to metaphase II (MII). In mice, these ovulated MII oocytes enter the oviduct, where they await fertilization. Oocyte maturation is accompanied by a wave of RNA degradation where approximately 20% of total maternal RNA is actively degraded. This absolute reduction in cellular mRNA results in relative changes to transcript dosage in the MII transcriptome, where some transcripts are stabilized, are destabilized, or remain unchanged. The MII transcriptome is a large determinant of oocyte competence, and the mechanisms ensuring correct gene dosage achieved through meiotic maturation are not known.

3To understand the in vivo function of YTHDF2 and m6A-mediated mRNA destabilization of transcripts, we generated an epitope-tagged and conditional allele of Ythdf2 (Ythdf2HA-Fl) in mice (Figures 1A, S1A, and S1B). The N-terminal tagging of YTHDF2 with GFP-His6-FLAG-HA did not affect the function of the protein, as mice homozygous for Ythdf2HA-Fl were viable and fertile (Figure S1C, related to Figure 1). The HA-YTHDF2 protein was detectable by western blotting and was expressed in all tissues analyzed (Figure 1B). However, the respective tissues express differential amounts of YTHDF2, with testis displaying the highest expression (Figure 1B). We next sought to understand the expression of YTHDF2 at the cellular resolution in the germline during gametogenesis. YTHDF2 is expressed at all stages of spermatogenesis, with elevated expression observed in pachytene spermatocytes (Figures 1C). Folliculogenesis is the growth phase of oocyte development where the biomaterial and the maternal transcriptome required for oocyte competence are assembled. This initiates when a clutch of primordial oocytes commence growth coincident with the expansion of surrounding somatic granulosa cells that collectively form the follicles with folliculogenesis, culminating in ovulation (Figure 1D). At all stages of folliculogenesis, YTHDF2 is expressed both in the oocyte and in somatic granulosa cells (Figure 1E). YTHDF2 is also expressed during oocyte maturation, with abundant YTHDF2 detected in GV as well as in MII oocytes (Figure 1F). Both during spermatogenesis and folliculogenesis, YTHDF2 is cytoplasmic in both the germ and the somatic cells (Figures 1C, 1E, and 1F).

4To understand the in vivo function of YTHDF2, we converted the conditional allele to a null allele (Ythdf2−) (Figures 2A, S1A, and S1B). We observed that Ythdf2−/− mice are retrieved in sub-Mendelian ratios (Figure 2B), with approximately half of the expected Ythdf2−/− mice observed at weaning. The loss of Ythdf2−/− from heterozygous intercrosses increased as the allele was bred toward a C57Bl6 genetic background (Figure 2B). Nonetheless, the viable Ythdf2−/− mice were indistinguishable from their heterozygous or wild-type littermates. We next determined the fertility status of both male and female Ythdf2−/− mice by set-up crosses with wild-type mice. Despite the abundant YTHDF2 expression throughout spermatogenesis (Figure 1C), Ythdf2−/− males were fertile (Figure 2C), with normal seminiferous tubule histology (Figure 2D). However, Ythdf2−/− females were sterile (Figure 2E), with corpora lutea observed in the Ythdf2−/− ovaries, indicating that ovulation had occurred (Figure 2F). In summary, YTHDF2 deletion is partially permissive in mice and results in female-specific infertility.

5The female infertility in Ythdf2−/− mice could arise from either germline- or somatic-related defects. Indeed, YTHDF2 is expressed both in the oocyte and in somatic granulosa cells during folliculogenesis (Figure 1E). The somatic cells support and transmit key instructive signals to the growing oocyte. We therefore employed conditional genetics to test the intrinsic and maternal oocyte function of YTHDF2. To this end, we combined the Zp3Cre allele that deletes in growing oocytes with our Ythdf2HA-Fl allele to generate experimental Ythdf2HA-Fl/HA-Fl; Zp3Cre Tg+ (Ythdf2mCKO) and control Ythdf2HA-Fl/+; Zp3Cre Tg+ or Ythdf2+/+; Zp3Cre Tg+ (Ythdf2CTL) mice. This maternal conditional deletion (mCKO) strategy resulted in the oocyte-specific deletion of YTHDF2 without affecting its expression in somatic granulosa cells (Figure 3A). Crossing of Ythdf2mCKO females with wild-type males revealed that maternal expression of YTHDF2 is intrinsically required for female fertility (Figure 3B). Histological analysis of Ythdf2mCKO ovaries revealed the presence of corpora lutea, indicating that ovulation has occurred (Figure 3C). Hormone priming with pregnant mare serum gonadotropin (PMSG) and human chorionic gonadotropin (hCG) to induce oocyte growth and subsequent ovulation revealed that Ythdf2mCKO females produce normal numbers of MII oocytes that have completed meiosis I and arrested at metaphase II properly (Figures 3D and 3E). We next sought to understand if Ythdf2mCKO oocytes are competent to be fertilized. Ythdf2CTL and Ythdf2mCKO females were hormone primed and set up with wild-type males, and zygotes were collected 0.5 days later and examined for fertilization as evidenced by the progression to the 2 pronuclei (2PN) stage. A similar frequency of Ythdf2mCKO and Ythdf2CTL oocytes had reached the 2PN stage accompanied by the extrusion of the second polar body (Figure 3F); thus YTHDF2 is not required for the process of fertilization per se. Harvesting zygotes at 2.5 days after priming and mating revealed that development is derailed at or prior to the two-cell stage in Ythdf2mCKO zygotes (Figure 3G). In comparison to Ythdf2CTL, fewer Ythdf2mCKO zygotes made two-cell stage embryos of normal morphology, with many of the Ythdf2mCKO two-cell embryos presenting various cytokinesis defects such as micronuclei and enucleated cells (Figure 3G). In summary, YTHDF2 is maternally required for early zygotic mouse development.

6The failure of Ythdf2mCKO oocytes to support early zygotic development could arise from the inability to process or utilize the transcriptome correctly. The utilization of the maternal transcriptome commences with the onset of oocyte maturation with an overall 20% reduction of cellular RNA in MII oocytes that results in relative changes to the MII transcriptome (Figures 4A and 4B). We decided to analyze the MII transcriptome, as this is the earliest stage where the maternal transcriptome is utilized, thus enabling us to determine the primary impact of YTHDF2 on the maternal transcriptome. Analysis of MII oocytes revealed deregulated gene expression within the Ythdf2mCKO transcriptome, with increased dosage of transcripts originating from 201 genes and decreased expression from 68 loci when applying a cut-off with a fold change greater or equal to two and a significance with a p value less than 0.05 (Figures 4C and S2). To exclude a function for YTHDF2 in the formation of the maternal transcriptome, we profiled gene expression in GV oocytes where transcription has ceased and the oocyte possesses a mature maternal transcriptome. This analysis revealed that Ythdf2mCKO GV oocytes contain a near normal transcriptome (Figure S3, related to Figure 4). In summary, the loss of YTHDF2 does not grossly impact oocyte growth or the formation of the maternal transcriptome but is required to instruct the correct gene dosage during oocyte maturation. The bias in deregulated gene expression in Ythdf2mCKO MII oocytes toward upregulation is consistent with the removal of a protein that potentiates RNA degradation. We next sought to understand what class of genes YTHDF2 regulates across oocyte maturation, whether those that are relatively stabilized, destabilized, or remain unchanged (Figure 4E). This analysis revealed that in the cohort of genes that are upregulated in Ythdf2mCKO MII oocytes, the majority (168/201) should remain relatively unchanged across oocyte maturation, and some (33/201) are destabilized (Figures 4E and S4). In summary, YTHDF2 function is intrinsically and maternally required to instruct the appropriate transcript dosage during oocyte maturation.

7We next sought to understand if some of the upregulated transcripts could be direct targets of YTHDF2. GACU/A is the most common YTHDF2-binding motif identified from PAR-CLIP. We therefore looked for enrichment of the GACU/A consensus ±400 bp around the stop codon in the top upregulated and downregulated genes as well as in genes whose dosage remains unchanged. To this end, we took the top 1,000 upregulated and downregulated transcripts; this selection is required for statistical power and corresponds to an approximate 1.4-fold change in expression levels. This analysis revealed a significant enrichment for the GACU/A motif only in upregulated genes in Ythdf2mCKO MII oocytes (Figure 4F). This enrichment was not observed around the 5′UTR (Figure 4F), where YTHDF2 is known not to occupy under steady-state non-stress conditions. The limiting amount of RNA that can be isolated from oocytes excludes the possibility of performing m6A-seq to determine if the consensus-containing upregulated genes in Ythdf2mCKO MII oocytes are methylated. We therefore sought to determine if we can find evidence for their methylation in other mouse tissues using the MeT-DB database. We found a stark enrichment for m6A adjacent to the stop codon in the transcripts that are upregulated in Ythdf2mCKO MII oocytes (Figure 4G). In summary, we find that the upregulated transcripts in the Ythdf2mCKO MII oocytes are enriched for the m6A/YTHDF2 consensus motif, and for some we find evidence for their m6A methylation in other cell types.

8The loss of METTL3 demonstrated an indispensable function for RNA m6A in embryonic stem cell exit from pluripotency and preimplantation development. Here we report the deletion of a mammalian m6A reader and identify the physiological importance of the m6A-YTHDF2-mediated mRNA destabilization pathway in mice. We identify and characterize an essential role for YTHDF2 in the female germline. We also find that loss of YTHDF2 is partially permissive in mice, indicating that YTHDF2 has other important developmental functions outside of the germline. The presence of Ythdf2−/− viable adult mice indicates that this branch of the RNA m6A regulatory pathway is not essential for mice under normal conditions. However, given that YTHDF2 has been shown to be involved in the cellular heat-shock response, it remains to be seen if YTHDF2 function is required in vivo to appropriately respond to various physiological and environmental perturbations. The fact that Ythdf2−/− mice show a genetic background effect indicates that mRNA m6A sites could constitute a basis for modifier alleles. If this is the case, it would demonstrate the power of mRNA m6A sites as potential modifiers of development and disease. This could be especially relevant to human development and disease given the increased frequency of RNA m6A observed in human versus mouse cell lines. While the mouse m6A demethylase ALKBH5 is required for normal spermatogenesis, we demonstrate that the YTHDF2-mediated destabilization of m6A-containing transcripts is dispensable for male gametogenesis, at least on a mixed genetic background. In yeast RNA, m6A is specific to meiosis; the loss of Ime4, an ortholog of METTL3, or the YTH-domain containing m6A reader Mrb1, affects meiotic progression. In mice, YTHDF2 does not regulate the dosage of meiosis-specific genes (Figure 4D), with meiosis I and likely II being completed normally (Figure 3E) in Ythdf2mCKO oocytes. Ythdf2−/− male mice are fertile, with no impact observed in spermatogenesis (Figures 2C and 2D). In summary, the YTHDF2-mediated regulatory RNA m6A pathway is not important for mouse meiosis.

9The maternal loss of zebrafish Ythdf2 has a modest impact on zygotic development, whereas maternal or zygotic Ythdf2 depletion profoundly impairs embryogenesis. Here we show a defining maternal function for YTHDF2 in regulating transcript dosage across oocyte maturation which is essential for generating MII oocytes that are competent to sustain early zygotic development. The loss of YTHDF2 during oocyte maturation results in the deregulation of approximately 270 genes, leading to an arrest prior to or at the two-cell stage, with various cytokinesis defects observed in the two-cell embryos (Figure 3G). The maternal transcriptome is essential for the first mitotic division. We posit that the deregulated gene expression in Ythdf2mCKO oocytes poisons the maternal transcriptome, rendering it incompetent to support the mitotic division. We have shown that YTHDF2 is expressed throughout oocyte growth and from GV through to MII oocytes but that the phenotype only arises after meiotic resumption (Figures 3E–3G). Furthermore, YTHDF2 does not majorly impact on the formation of the maternal transcriptome (Figure S3, related to Figure 4), although, present throughout oocyte growth, these observations beg the question as to why YTHDF2-mediated mRNA degradation is only active upon meiotic maturation. In somatic cells, YTHDF2 has been shown to function through the recruitment of deadenylases and subsequent mRNA decapping. RNA degradation factors are downregulated during oocyte growth, favoring the accumulation of RNA and the building of the maternal transcriptome. During meiotic maturation, the deadenlyase CNOT7 and the decapping enzymes DCP1A and DCP2 are translated from maternal transcripts that enable the resumption of active RNA degradation pathways. We believe this reactivation of RNA degradation machinery enables YTHDF2 to directly degrade the bound transcripts selectively during meiotic maturation. In summary, we demonstrate that the m6A reader YTHDF2 is an essential regulator of the mammalian maternal transcriptome and egg quality.

10Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Dónal O’Carroll (donal.ocarroll@ed.ac.uk).

11For the Ythdf2HA-Fl allele, a GFP-precission-His6-Flag-HA-HA epitope tag was inserted after the endogenous starting initiation ATG codon in exon 1 of Ythdf2. In addition, two loxP sites were placed flanking exon 2. The targeting construct was genetically modified so that it contained homology arms and FRT sites flanking a neomycin cassette 3′ of exon 1. Southern blotting of EcoRV-digested DNA extracted from ESC-derived clones with exogenous 5′probe was used for the validation of homologous recombinants. The wild-type Ythdf2 locus generates a ∼9 kb DNA fragment, whereas the integration of the second loxP site introduced an additional EcoRV site, thus decreasing the size of the EcoRV DNA fragment to 8 kb in the targeted allele. Flp-mediated recombination removed the FRT flanked neomycin cassette and generated the Ythdf2HA-Fl allele that can be identified by the 5′ probe as a 6 kb EcoRV DNA fragment. Cre-mediated deletion of the loxP flanked exon 2 resulted in 5.5 kb EcoRV DNA fragment, that can be identified by the 5′probe and validate the Ythdf2- allele. The Zp3Cre Tg allele was also used in this study for the generation of Ythdf2mCKO female mice. Fertility analysis with male and female Ythdf2−/− mice were performed on a mixed genetic background. The majority of the Ythdf2mCKO analysis was performed on mice that were backcrossed six times toward the C57BL/6 genetic background. All mice used in this study were on mixed or C57BL/6 genetic background and were bred and maintained in EMBL Mouse Biology Unit, Monterotondo, and subsequently at the Centre for Regenerative Medicine, Edinburgh. All procedures were done in accordance to the current Italian legislation (Art. 9, 27. Jan 1992, nu116) under license from the Italian health ministry or the UK Home Office regulations, respectively.

12The forward 5′-GCAGGTGACCTCTTCAGAAG-3′ and reverse 3′-CCAGTCCCTGTAGATTTTAGAG-5′ primers were used to generated an exogenous 5′ probe for detection of the targeted, Ythdf2HA-Fl and Ythdf2- alleles. Genomic DNA was restriction digested and run on a 0.8% agarose gel. The DNA fragments were then transferred to an Amersham Hybond-XL membrane (GE Healthcare RPN203S) through alkaline solution (0.4 M NaOH, 1.5 M NaCl) overnight. The membrane was neutralized in 2X SSC solution, UV-crosslinked with 150 mJ/cm2 and incubated in prehybridization solution (0.5 M Na2HPO4, 1 mM EDTA, 5% SDS, 3% BSA) for 2 hr at 65°C. DNA probe was synthesized with Random Primers DNA Labeling System (Thermo Fisher Scientific) in accordance to manufacturer’s protocol and was hybridized with the membrane overnight at 65°C. The membrane was washed in 40 mM Na2HPO4, 1 mM EDTA, 5% SDS and exposed on a phosphor screen (Fujifilm).

13Protein extracts were prepared with dounce homogenizer in lysis buffer (50 mM Tris pH 7.8, 150 mM NaCl, 0.4% NP-40, 2 mM MgCl2, 1 mM DTT) supplemented with proteinase inhibitors. Extracts were collected after centrifugation at 14000 g for 10 min at 4°C, resolved on 7.5% SDS-PAGE gel and transferred on Immobilon-P membrane (Millipore) via wet transfer overnight. Membranes were blocked in 5% milk-PBST (PBS with 0.1% Tween20) and probed with anti-HA (Covance, MMS-101P, 1:1000) and anti-alpha-tubulin antibody (Sigma-Aldrich, T9026, 1:1000) for 4 hr at room temperature. Membrane was washed in PBST and incubated with appropriate horseradish peroxidase-coupled secondary antibody (Amersham) in 5% milk-PBST for 1 hr. Proteins were detected with ECL Western Blotting Detection Reagent (Amersham) and acquired on a ChemiDoc XRS system (BioRad).

14For the collection of GV oocytes, 3-8 weeks old females were injected with 10 U of pregnant mare serum gonadotropin (PMSG) (Henry Schein). After 44-48 hr GV oocytes were collected through puncturing the ovarian follicles in M2 media (Sigma-Aldrich). Subsequently, GV oocytes were released from the somatic cells via manual mechanical separation.

15For the collection of MII oocytes, 3-8 weeks old females were injected with 10 U of PMSG and after 46-48 hr with 10 U of human chorionic gonadotropin (hCG) (Intervet). MII oocytes were isolated from the oviduct of the hormone-stimulated females 14 hr after the hCG injection. MII oocytes were cleaned from the somatic cells with hyaluronidase (Sigma-Aldrich) in M2 media.

16For the collection of zygotes, PMSG and hCG stimulated females (as described above) were set up with a stud male immediately after the last injection. Zygotes were isolated from the oviduct of plugged females 0.5 and 2.5 days after the hCG injection. The collected 0.5 day zygotes were briefly cleaned from the somatic cells with hyaluronidase in M2 media.

17For the detection of YTHDF2 in GV and MII oocytes, the isolated oocytes of the respective genotypes were stained with mouse anti-HA antibody (Covance, MMS-101P, 1:100) and Hoechst33342 (5 mg/ml) (Sigma-Aldrich,) as previously described.

18For immunofluorescence on ovary sections, ovaries were dissected and fixed in 4% paraformaldehyde for 2 hr at 4°C and subsequently sucrose-OCT embedded. Ovary sections of 6 μm were cut on cryostat and subjected to antigen retrieval with antigen unmasking solution (Vector Lab H-3300). Next, sections were permeabilized with 0.1% Triton-X for 15 min at room temperature (RT). Blocking was performed with 10% normal donkey serum (NDS) (Sigma-Aldrich, D9663), 2% BSA and 0.1 M glycine for 30 min at RT. Sections were incubated with anti-HA mouse antibody (1:100) for 1 hr at 37°C in a humidified chamber. Appropriate secondary antibodies from Invitrogen were applied for 1 hr at RT, followed by Hoechst 33342 for DNA staining.

19For the microtubule staining, MII oocytes and zygotes were fixed in 2% formaldehyde, SB solution (0.01 M Pipes, 0.5 mM MgCl2, 0.25 mM EGTA) and treated with 2% Triton-X for 30 min at 37°C. The MII oocytes/zygotes were washed three times in 0.1% NDS and transferred to blocking solution (10% NDS and 0.01% Triton-X) for 1 hr. Then MII oocytes/zygotes were incubated with rabbit anti-beta tubulin antibody (Sigma-Aldrich) (1:200) at 37°C for 1 hr. They were washed three times in the blocking solution before incubation with secondary antibody in 5% NDS and Hoechst 33342 staining. MII oocytes/zygotes were mounted on teflon coated slides (Dutscher scientific) in a group of 10 per well.

20Immunofluorescence assay on testis sections was performed as previously described. The above mouse anti-HA antibody and anti-γH2AX (ICH) (1:250) rabbit antibodies were used.

21Images were acquired on SP5 and SP8 Leica TSC confocal microscope. The max projection images were cropped and modified in Photoshop with equal settings for control and experimental samples.

22Ovaries and testis from adult mice were fixed overnight in Bouin’s solution and embedded in paraffin. Sections of 7 μm were cut and stained with periodic acid Schiff reagent (Sigma-Aldrich) and Hematoxylin (Sigma-Aldrich) as per manufacturer’s instructions.

23Total RNA was isolated from 50-90 GV and MII oocytes with QIAzol lysis reagent (QIAGEN) following the manufacturer instructions. Total RNA was in vitro transcribed and biotinylated with the Ovation Pico WTA Systems V2 (NuGEN) and fragmented and labeled with Encore Biotin Module (NuGen). Hybridization of the cDNA was done with GeneChip Mouse Gene 2.0 ST Array (Affymetrix) for 16 hr at 45°C. Affymetrix Fluidics Station 450 was used for washes and staining of the GeneChips. For this analysis 3-4 biological replicates were used for GV and MII Ythdf2CTL and Ythdf2mCKO oocytes.

24For RNA expression analysis, robust multi-array average (RMA) was used for the raw data normalization and limma package to determine differential expression. Moderate t-statistics was done with adjusted p values. Gene ontology analysis was done with Gene ontology enrichment analysis and visualization tool Gorilla.

25Affymetrix identifiers were mapped to Ensembl transcripts using Biomart (32). Biomart was then used to obtain 400 nt 5′ and 3′ of both the start and stop codon. Sequences obtained that were shorter than 400 nt were padded to this length with the addition of ’N’s. Each transcript may only be present once in the Affymetrix gene list. Where multiple transcripts are present, the one with the highest absolute fold-change is retained. Only one isoform is retained when multiple transcripts have the same sequence. Motif occurrences were directly computed using Perl regular expression matching for “GCA[UA]” for sequences assigned to the top 1000 most upregulated, downregulated and for 1000 transcripts from the center of the gene list. A matrix of 1000x800 elements is hence obtained for each of the three sets, indicating which sequence and at which nucleotides motifs occur. This matrix is column summed and plotted using a cubic spline smoothing function smooth.spline from R/Bioconductor. Statistics for the motif enrichment was done with a hypergeometric test.

26Public m6A peak data was obtained from MeT-DB (http://compgenomics.utsa.edu/methylation/) for 12 mouse (mm9) samples. Peaks were translated into BED files with enrichment scores, lifted-over to mm10 via UCSC liftover, sorted, filtered and overlapping regions condensed into bigwig files. These bigwig files were provided to the DeepTools package v1.5.9.1 (ComputeMatrix and Heatmapper) to explore peak enrichments around gene-bodies from the start to the stop codon with 400 nt either side. The Ensembl GTF corresponding to GRCm38 version 79 was used for gene body coordinates. Statistics for the m6A peak enrichment was done with two-sample t test.

27Total RNA from 50-80 MII oocytes per biological replicate was reverse-transcribed using SuperScript IV and random hexamers (both Invitrogen) according to manufacturer’s instructions. qRT-PCR was performed using the LightCycler 480 SYBR Green I Master mix (Roche), and samples were run in technical triplicates on a Roche LightCycler 480 instrument. Ct values were normalized against the internal controls Gapdh, Sod1 and Bmp15. Fold differences in expression levels were calculated according to the 2−ΔΔCT method.

28Original images, immunoblots and radiography can be found in Mendeley Data (http://dx.doi.org/10.17632/zb7zyfghg3.1).

29I.I. contributed to the design, execution, and analysis of most of the experiments. D.O. conceived this study. C.M. designed, generated, and validated the Ythdf2HA-Fl allele and qRT-PCR experiments. M.D.G. and C.A. contributed to histology and imaging experiments. I.I., C.C., J.M., and M.M. performed the bioinformatic transcriptome analysis. P.N.M. aided in analysis of mouse zygotes. A.J.E. performed the consensus enrichment analysis and oversaw all bioinformatic analyses performed. D.O. and A.J.E. supervised this study. I.I., A.J.E., and D.O. wrote the final version of the manuscript.