ICR female mice aged 8 weeks, male mice aged 10 weeks were kept in controlled conditions of temperature (24°C) and light (12 h light:12 h dark) and had free access to food and water. All mice were approved by the Institutional Animal Care and Use Committee of China Agricultural University.

The female mice were superovulated by intraperitoneal injection of 5 IU of pregnant mare serum gonadotropin (PMSG, Ningbo, China) and a further intraperitoneal injection 48 h of 5 IU human chorionic gonadotrophin (HCG, Ningbo, China). The female mice were cocaged individually with male mice after the hCG injection. On the next morning, the females with vaginal plug were selected as mating successfully. Zygotes, the 2-cell, 4-cell, 8-cell embryos, morulae and blastocysts (16–20 h, 44–46 h, 54–56 h, 66–68 h, 74–76 h and 94–96 h post HCG, respectively) were recovered from donors by flushing the oviduct and uterus with M2 medium.

The IVF procedure were performed as previously described (Ren et al., 2015; Tan et al., 2016a; Tan et al., 2016b; Ren et al., 2017). In brief, sperm were released from the cauda epididymis and capacitated for 1 h in modified Krebs-Ringer bicarbonate medium (TYH), and the cumulus-oocyte complexes were transferred into modified human tubal fluid (mHTF) for 30 min, then inseminated for 4 h. After the insemination, zygotes were washed and cultured in potassium simplex optimized medium containing amino acids (KSOM + AA; Millipore, Darmstadt, Germany) at 37°C in 5% CO₂. IVF embryos at different preimplantation stages were collected based on their developmental progress and morphology.

Bovine oocytes were collected from ovaries obtained from a slaughterhouse, and matured in Tissue Culture Medium-199 (TCM-199, Thermo Fisher Scientific, Rockford, IL, United States) plus 10% (vol/vol) FBS (HyClone, Marlborough, United States), 1% antibiotic-antimycotic (Gibco BRL, Thermo Fisher Scientific), and 10 ng/ml epidermal growth factor (22–24 h). In vitro fertilization was conducted in Bracket and Oliphant’s (BO) medium. Briefly, matured oocytes with multiple layers of expanded cumulus cells were washed and then in BO fertilization medium supplemented with 6 mg/ml essential fatty acid-free (FAF)-BSA (Millipore, Billerica, MA, United States) and 10 mg/ml heparin. Fifteen to 20 cumulus-oocyte complexes (COCs) were placed in 50 μL BO medium, under mineral oil, containing frozen-thawed sperm (1–2 × 10⁶ sperm/ml) for 24 h in 5% (vol/vol) CO₂ in air at 38.5°C. Cumulus cells were removed by pipetting, and presumptive zygotes were cultured in 20-μl drops of Bovine VitroCleave (IVF Vet Solutions, North Adelaide, Australia) under mineral oil for 5 days. On day 5, embryos were transferred in groups of 5–10–20-μl drops of Bovine VitroBlast (IVF Vet Solutions) under mineral oil.

Oviduct and uterine fluids were collected from female mouse according to the protocol of a previous study (Harris et al., 2005). Briefly, once oviduct excised, the tissue was dried and placed under mineral oil, stabbed by needle. Then we collected the fluids with mouse pipette into 1.5 ml tube. Uterine was ligatured with nylon thread, gentle pressured from the thread end to another one, the fluids were flowed into 1.5 ml tube. All fluids were centrifuged for 5 min at 12,000 revolutions per minute (rpm) to obtain the supernatants, stored into -80°C. Remained tissue was washed twice with PBS solution, then 0.1 g tissue was put into 1.5 ml tube and immediately throwed in liquid nitrogen until assay.

Total RNA was extracted from embryos, the oviduct and uterus with TRIzol (Thermo Fisher Scientific) following the manufacturer’s instructions. Then the reverse transcription was performed by the Hiscript R Ⅱ Q RT Supermix (Vazyme, Nanjing, China) according to the manufacturer’s instructions. Quantitative real-time PCR (qRT-PCR) was performed with SoFast EvaGreen Supermix (BioRad, Hercules, California, United States) using a CFX96 real-time PCR machine (BioRad). All primers were listed in Supplementary Table S1.

Total RNAs were extracted from embryos, oviducts and uterine at different stages with TRIzol Reagent (Invitrogen, Carlsbad, CA, United States). Then the RNA was delivered to BGI (BGI, Shenzhen, China) for sequencing. Gene expression levels were measured in reads per kilobase of exon model per million mapped reads (RPKM). The RPKM was listed in Supplementary Table S2–4.

The database for Annotation, Visualization and Integrated Discovery (DAVID v6.7; http://david.abcc.ncifcrf.gov) was used to annotate biological themes (gene ontology, GO). The Kyoto Encyclopedia of Genes and Genomes (KEGG; http://www.genome.jp/kegg/) was used to determine the associated pathways. Phenotype annotations were analyzed based on the Mouse Genome Informatics (MGI; http://www.informatics.jax.org/phenotypes.shtml) database.

Collected embryos were washed three times with 0.1%PVA-PBS and then washed with acidic Tyrode’s solution to eliminate the zona pellucida. Then embryos were fixed in 4% paraformaldehyde in PBS overnight at 4°C. After permeabilized with 0.5% Triton X-100 in 0.1% PVA-PBS, embryos were blocked in 1% BSA (Millipore) in 0.1% PVA-PBS for 1 h, then sequentially incubated with primary antibody overnight at 4°C. Next, the embryos were washed three times with 0.1%PVA-PBS for 5 min and incubated with secondary antibodies for 1 h at room temperature. Finally, the samples were treated with DAPI for 5 min, mounted with coverslips. Images were recorded using fluorescence microscope (BX51TRF; Olympus, Tokyo, Japan) and processed using ImageJ software (Rawak Software Inc., Stuttgart, Germany).

For 5mC and 5hmC staining, nuclear DNA was denatured with 4M HCl for 10 min, neutralized with Ph8.0 Tris-HCl for 15 min, then embryos were blocked overnight, incubated with primary antibody for 2 h. Other steps were same as described above. The following antibodies used in this research were listed as follows: anti-5mC (1:200, Active motif 39,649), anti-5hmC (1:500 dilution, 39,791, Active motif, California, United States), anti-TET1 (1:200 dilution, GTX124207, GeneTex, Irvine, CA, United States), anti-TET2 (1:100 dilution, ab94580, Abcam, Cambridge, UK), anti-NANOG (1:500 dilution, ab80892), anti-CDX2 (1:500 dilution, BioGenex-MU392A-UC, BioGenex Laboratories, Fremont, CA 94538, United States).

Preimplantation embryos, the oviduct and uterus, as well as the oviductal or uterine fluid were prepared as described above. Each 50 oocytes, 50 embryos or total cumulus cells surrounding 150 oocytes as a biological replicate, CC indicates the total cumulus cells surrounding one oocyte. The ascorbic acid assay kit (ab65346, Abcam) was used for detecting the vitamin C content according to the manufacturer’s instructions.

Pseudo-pregnant female mice (recipients) were co-caged individually with vasectomized males 3.5 days before embryo transfer. The morning after mating, the recipients were checked for the presence of a vaginal plug. The day of plugging was considered as day 0.5 of the pseudo-pregnancy. 12 well-developed blastocysts were transferred into each recipient. At embryonic day (E)7.5 (4 days of embryo transfer), the conceptuses covered with decidual mass were gently teased away from the uterus, E7.5 embryos were E7.5 embryos were separated and washed in PBS, then imaged using a stereomicroscope (SZX16; Olympus, Tokyo, Japan) equipped with a digital camera.

Student’s t-test or one-way ANOVA were used to analyze the difference among groups by using SPSS 23.0 software (Statistical Package for the Social Sciences). Statistically significant differences were defined as p < 0.05.

To test the impact of IVF processes on DNA demethylation during preimplantation development, we compared our previously published global MeDIP-seq data of IVO and IVF blastocysts (Ren et al., 2015; Ren et al., 2017). Focusing on promoter DNA methylation, which undergoes postfertilization demethylation (Smith et al., 2012) and participates in TET-induced transcriptional regulation (Ito et al., 2010; Gu et al., 2011; Blaschke et al., 2013), we found IVF blastocysts showed higher DNA methylation levels ( Figure 1A ) and a greater proportion of promoters were relatively hypermethylated in IVF blastocysts compared with their IVO counterparts ( Figure 1B ) Similarly, reanalysis of previously published DNA methylation array data also showed more hypermethylated regions in IVF bovine blastocysts (Supplementary Figure S1A). In addition, Gene ontology (GO) analysis and Mouse Genome Informatics (MGI)-based phenotype annotations suggested hypermethylated genes can participate in many basic molecular functions and cellular processes, and are essential for normal embryonic development and survival throughout the pregnancy (Supplementary Figure S1B, C). Venn diagram based on previously published methylome (Smith et al., 2012) showed that a substantial proportion of promoters that should be demethylated before blastocyst formation, were hypermethylated promoters in IVF blastocysts (Figure 1C). Next, we detected the chromosome-wide distribution of hypermethylated promoter in IVF blastocysts, and found hypermethylated promoters were globally distributed across all autosomes and sex chromosomes (Figure 1D). These results suggest that IVF preimplantation embryos may undergo extensive impairment in DNA demethylation.

Given results of an early study suggested that impaired DNA demethylation can be initially observed in paternal genome of IVF zygotes (Yoshizawa et al., 2010), we speculated that TET-mediated active demethylation may be impaired. To confirm this hypothesis, we tested 5mC and 5hmC levels at each stage of preimplantation development. Quantitation of 5hmC/5mC ratio indicated that TET-mediated active DNA demethylation was consistently impaired in IVF preimplantation embryos (Figures 1E,F).

Having confirmed the defects of TET-mediated active DNA demethylation in IVF embryos, we next asked if the gene expression level of Tet family members was inhibited in IVF preimplantation embryos. We found Tet1 and Tet2 expression, which should be increased during cleavage stages, were significantly inhibited in IVF embryos at the two- to 8-cell stage and the morula stage respectively (Figure 2A). In addition, Tet3 expression, although showed maternal deposition, was consistently inhibited in IVF embryos from the 4-cell stage onwards (Figure 2A). The IVF-induced expression inhibition of Tet family members was further confirmed using our RNA-seq data (Figure 2B). Moreover, the inhibition of TET1 and TET2 were also validated on the protein level. In line with the result of mRNA detection, quantitation of immunofluorescence signal indicated that TET1 and TET2 proteins were significantly deficient in IVF 8-cell embryos and morulae, respectively (Figures 2C,D). Of note, we also noticed that a proportion of blastomeres exhibit cytoplasmic localization of TET2, and the mislocation in IVF embryos were more evident (Figure 2D).

Next, we attempted to test if the inhibited expression of Tet family members in IVF embryos could be rescued by supplementing cytokines or small molecules that have been reported to upregulate Tet expression: retinoic acid (Hore et al., 2016), FGF2 (Choi et al., 2018), LIF (Tahiliani et al., 2009; Koh et al., 2011) and insulin (Lv et al., 2017). Given Tet3 is a well-known maternally deposited transcripts, we next focused on the embryo-expressed Tet1 and Tet2. However, neither these factors alone (Supplementary Figure S2A–D) nor combinations (Supplementary Figure S2E, F ) could upregulate expression levels of Tet1 and Tet2 in IVF embryos.

Having failed to rescue IVF-induced Tet inhibition, we next asked if impaired active DNA demethylation can be rescued by enhancing TET activity. To this end, we focused on vitamin C and α-ketoglutarate (α-KG) because TET enzymes are Fe(II)- and α-KG-dependent dioxygenases, and vitamin C can stimulate TET activity by maintaining reduced Fe(II) (Kohli and Zhang, 2013). Time-course expression profiling of genes related to vitamin C synthesis and transport in preimplantation embryos, as well as in temporally corresponding oviduct based on our RNA-seq data (Figures 3A,B), indicated that vitamin C synthesis and transport occurred in both preimplantation embryos and the oviduct, especially in the oviduct. These findings were further supported by results of qRT-PCR (Figures 3C,D). More importantly, we detected that vitamin C was highly enriched in the oviduct and uterus throughout the preimplantation stage (Figure 3E), and thus being detectable in the oviductal and uterine fluid (Figure 3F). In addition, we also detected low-level vitamin C in preimplantation embryos and found vitamin C might be pre-deposited in oocytes, and no significant difference can be detected between IVO and IVF embryos (Figures 3G,H). These results suggest that the requirement for vitamin C during preimplantation development may be satisfied by both oviductal paracrine and embryonic autocrine.

Similarly, our results also suggest that the requirement for α-KG, another cofactor for TET dioxygenases, may be also satisfied via synergistic effect of oviductal paracrine and embryonic autocrine, because α-KG synthetic and transporter genes were detectable in both embryos and the oviduct by preimplantation stage (Supplementary Figure S3A–C). Collectively, these results led us to test if supplementation of embryo culture medium with vitamin C and/or α-KG could rescue impaired active DNA demethylation in IVF embryos.

Next, we screened effective concentration of vitamin C supplementation by evaluating its efficacy in prompting preimplantation development, because TET deficiency would impair survival and growth of preimplantation embryos (Kang et al., 2015). We found 100 μg/ml vitamin C supplementation to culture medium significantly increased cleavage rate and blastocyst rate (Supplementary Figure S4A, B). Using this concentration, we found vitamin C supplementation significantly enhanced TET enzymatic activity in IVF embryos throughout the preimplantation development to the levels comparable to those of IVO embryos, as revealed by increased 5hmC/5mC ratio in both zygotes (Figure 4A and blastocysts (Figure 4B). Moreover, the beneficial effect of vitamin C supplementation on enhancing TET enzymatic activity was also confirmed using bovine IVF embryos as the model (Figure 4C), although the expression patterns of bovine TET enzymes were largely distinct from those in mouse embryos (Jiang et al., 2014; Salilew-Wondim et al., 2015) (Supplementary Figure S4F, G).

In contrast, however, α-KG supplementation using previously published effective concentration (Zhang et al., 2019) just tended to, but not significantly, enhance TET enzymatic activity in both IVF zygotes and blastocysts. In addition, combined supplementation of vitamin C and α-KG didn’t display synergistic effect on prompting TET enzymatic activity (Figure 4A, B), in line with their effect on preimplantation development (Supplementary Figure S4C–E).

Given previous studies have demonstrated that TET-mediated active DNA demethylation participate in inner cell mass (ICM) specification, and regulate embryonic growth and developmental potential (Ito et al., 2010; Kang et al., 2015), we next tested whether vitamin C-prompted TET enzymatic activity would affect total cell number and lineage differentiation of IVF blastocysts. In addition, we also evaluated the effect of vitamin C supplementation on embryonic developmental potential by detecting implantation rate and postimplantation survival rate following embryo transfer. Our results showed IVF blastocysts exposed to vitamin C displayed a significant increase in total cell number (Figures 5A,B), and ICM cell number (Figures 5A,C). Of note, vitamin C supplementation resulted in a changed lineage differentiation towards the ICM fate in IVF blastocysts (Figures 5A,D). Correspondingly, in comparison to their control counterparts, IVF embryos exposed to vitamin C had significantly higher implantation rate and survival rate shortly after implantation (Figures 5E,F). Similarly, the beneficial effects of vitamin C supplementation on preimplantation lineage differentiation and developmental potential, were also confirmed in bovine IVF embryos (Figures 5H,I).

Moreover, we found α-KG alone, but not in combination with vitamin C, enhanced total cell number and ICM specification of IVF blastocysts, as well as subsequent implantation success (Figures 5A–F). These beneficial effects, were not completely in line with quantification results of 5hmc/5mC ratio (Figures 4A–D), implying that functions of α-KG in improving IVF embryo development are complicated and may be partially independent of TET enzymatic activity.

Having confirmed the function of vitamin C on rescuing impaired active DNA demethylation in IVF embryos, we next attempted to determine if this beneficial effect was mediated by TET proteins. To this end, we used dimethyloxallyl glycine (DMOG), an inhibitor that blocks TET enzymatic activity (Zhang et al., 2017; Duforestel et al., 2019), to assess the role of TET enzymes in mediating vitamin C-induced active DNA demethylation and developmental advantages. We found DMOG significantly attenuated the active DNA demethylation-prompting effect of vitamin C in both IVF zygotes (Figure 6A) and blastocysts (Figure 6B). In addition, the beneficial effects of vitamin C on total cell number and lineage differentiation of IVF blastocysts were also attenuated by DMOG supplementation (Figure 6C). These results suggest that the beneficial effects of vitamin C are largely mediated by TET enzymatic activity.