Introduction

Mammalian life starts in transcriptional silence with an oocyte-to-embryo transition that encompasses the fusion of the egg and sperm, migration and fusion of the germ cell pronuclei, and a series of cleavage divisions culminating in the activation of the unique embryonic genome followed by compaction of blastomeres to form a morula and differentiation of the first cell lineages—the trophectoderm and inner cell mass (Rossant & Tam, 2009). Embryo development is remarkable in that the oocyte must provide all required resources to carry out the complex developmental pathways in the absence of transcription prior to embryonic genome activation (EGA) (Braude et al, 1988; Zhang & Smith, 2015).

DNA methylation reprogramming of oocyte and sperm genomes during early mammalian embryogenesis ensures the development of totipotent and pluripotent preimplantation embryos capable of contributing to diverse cell lineages (Messerschmidt et al, 2014; Eckersley-Maslin et al, 2018). Shortly after fertilization, a well-orchestrated and yet not fully understood combination of active and passive DNA demethylation, as well as de novo methylation reprograms the DNA methylation landscape of paternal and maternal genomes in the zygote (Ladstatter & Tachibana, 2019). Passive DNA demethylation during replication through non-maintenance of 5-methylcytosine (5mC) has been shown to have the biggest impact on global DNA methylation levels in preimplantation development (Guo et al, 2014; Shen et al, 2014) and is accompanied by selective maintenance and de novo methylation (Arand et al, 2015; Amouroux et al, 2016). Conversely, active DNA demethylation by enzymatic modification/removal of 5mC was reported to play an important role in the transcriptional regulation of specific genomic loci, but a rather moderate role in global DNA methylation reprogramming, and can involve ten-eleven-translocation (Tet) enzymes (Peat et al, 2014; Shen et al, 2014). Moreover, Tet enzyme activity in the zygote has been shown to protect certain CpGs from methylation buildup by DNA de novo methyltransferases (Peat et al, 2014). Tet proteins catalyze the enzymatic oxidation of 5mC to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC) in a consecutive manner (He et al, 2011; Ito et al, 2011). In mammals, three Tet enzymes (Tet1, Tet2, and Tet3) are differentially expressed during development and play distinct roles in different cell types (Ito et al, 2010; Gu et al, 2011; Koh et al, 2011; Moran-Crusio et al, 2011; Wossidlo et al, 2011; Moyon et al, 2021).

Tet3 is the highest expressed Tet protein in oocytes, and knockdown (KD) and knockout (KO) studies in mouse zygotes have shown that Tet3 mediates the conversion of 5mC to 5hmC (Gu et al, 2011; Wossidlo et al, 2011; Yu et al, 2013). Next to Tet3, Tet1 and Tet2 are also expressed during early preimplantation development (Wossidlo et al, 2011). Nevertheless, the importance of Tet enzymes and the oxidation of 5mC for preimplantation development is not clear, since so far, no severe phenotype in preimplantation development for different Tet-KO mouse models was reported. Observed Tet-KO phenotypes are obvious during postimplantation development, with preimplantation embryos showing mild transcriptomic changes or developmental impairment in different contexts. Here, heterozygous Tet3-KO mice displayed neonatal abnormalities, and Tet1−/− and Tet2−/− single KO embryos developed with postnatal malignancies (Dawlaty et al, 2011; Gu et al, 2011; Ko et al, 2011; Li et al, 2011). Tet1 + 3 double KO mice are characterized by abnormal early postimplantation phenotypes with variable gene expression and reduced developmental success during preimplantation development (Kang et al, 2015), and Tet1 + 2 double KO mice are characterized by smaller ovaries and strongly reduced fertility rates (Dawlaty et al, 2013). Furthermore, a conditional germline knockout of all three Tet1–3 enzymes (Tet-TKO) developed beyond the implantation stage, despite altered expression of a few hundred genes at the blastocyst stage (Dai et al, 2016). Derived Tet-TKO embryos showed severe gastrulation phenotypes and did not develop to term (Dai et al, 2016). These data so far point to an important role for Tet enzymes in later stages of development but not during the striking DNA reprogramming phase in early preimplantation embryos. Notably, the experimental design of the reported Tet-KO studies involved conditional knockout strategies targeting the growth phase of gametes during spermatogenesis and oogenesis. Loss of Tet enzymes in maturing gametes might already impact their epigenome, with the potential to influence and alter embryonic development. Moreover, the Tet-TKO and most other knockout studies were designed to create Tet-null mutants lacking the catalytic activities of Tet enzymes and are not complete genomic deletions of these enzymes. In this context, it has been shown that gene modifications, which aim to create catalytically dead enzymes by truncation of catalytical domains, like the conditional truncations of Tet enzymes in the growing oocyte (Gu et al, 2011; Dai et al, 2016), can trigger the expression of related genes, which can severely obscure observed phenotypes (El-Brolosy et al, 2019; Ma et al, 2019; Wilkinson, 2019). Interestingly, a study analyzing Tet-TKO mESCs revealed an important role of Tet enzymes in regulating the transition into the totipotent 2-cell embryo-like state in mESCs (Lu et al, 2014) suggesting an active role of Tet enzymes in the regulation of totipotent cells.

Despite many years of extensive research, it still remained unclear why mammals would invest in a potentially very dangerous mechanism of active DNA demethylation that could cause severe problems in the newly developing embryo. In this study, we specifically investigated the role of Tet enzymes in the oocyte-to-embryo transition, by performing knockdown experiments in fully grown and genetically normal oocytes to determine the biological significance of Tet-mediated DNA methylation reprogramming in mammalian preimplantation development. We also addressed the open question of whether Tet enzymes have overlapping roles in the enzymatic oxidation of 5mC and therefore can compensate each other interchangeably in the mammalian oocyte-to-embryo transition.

Results and Discussion Tet1–3-deficient mouse embryos arrest at the 2-cell stage

In mouse oocytes and zygotes, Tet3 is highly expressed and Tet1 and Tet2 transcripts are present at low levels (Wossidlo et al, 2011). To analyze the developmental importance of Tet enzymes specifically in the oocyte-to-embryo transition, we designed knockdown experiments to deplete Tet enzymes in genetically normal mouse oocytes. Therefore, we microinjected different combinations of morpholinos (MOs) targeting Tet1–3 mRNAs into mouse germinal vesicle oocytes (GVOs) followed by in vitro maturation (IVM) and in vitro fertilization (IVF; Fig 1A, and Appendix Table S1). Subsequently, we performed non-invasive time-lapse imaging to follow the development of control-MO-injected and Tet1–3-MO-injected embryos until the blastocyst stage (Figs 1B–D and EV1A, and Movie EV1). Embryos derived from control-MO injected oocytes developed into blastocysts with similar ratios as non-injected embryos at expected mouse in vitro culture ratios (83.5%, see Figs 1B and C, and EV1A and B). Strikingly, the triple KD of all three Tet enzymes (Tet-TKD) completely prevented blastocyst development (Fig 1B and D, see also Movie EV1). Moreover, the majority of Tet-TKD embryos arrested at the 2-cell stage (82%), and only a few advanced to the 4-cell stage (Fig 1D).

Figure 1. Tet enzyme-deficient mouse embryos arrest primarily at the 2-cell stage

A. Experimental setup: Mouse germinal vesicle oocytes (GVOs) were isolated and injected with Morpholinos (MOs) designed to target Tet1–3 mRNAs. After in vitro maturation (IVM), MII oocytes were fertilized by in vitro fertilization (IVF). Zygotes derived from knockdown (KD) and control-MO injected oocytes were analyzed by (i) non-invasive time-lapse imaging to monitor the developmental potential of KD embryos, (ii) immunofluorescence to analyze changes in DNA methylation reprogramming and (iii) by single 2-cell embryo RNA-Seq & BS-Seq to study the impact of Tet enzymes on embryonic gene activation (EGA) and DNA methylation reprogramming in the totipotent 2-cell embryo. B, C. Representative time-lapse images of (B) Tet1–3 triple KD (Tet-TKD), (C) Tet3-KD, and control embryos at 3.5 days post-fertilization. Scale bar = 100 µm. D, E. Developmental rates of (D) control (control-MO injected), Tet1-, Tet2- and Tet3-single KD and Tet-TKD embryos and (E) control (control-MO + Tet2 mRNA co-injected), Tet2-KD2 (Tet2 MO2 injected), Tet2-KD2 + GFP-mRNA and Tet2-KD2 + Tet2 mRNA embryos until blastocyst stage. Indicated P-values were calculated using log rank (Mantel-Cox) test against (D) control embryos or (E) Tet2-KD2 embryos or as indicated (ns = non-significant, *P < 0.05, ****P < 0.0001; numbers of analyzed embryos for each condition are indicated in parentheses). Click here to expand this figure.

Figure EV1. Developmental rates of embryos derived from microinjected oocytes

A. Summary of all analyzed samples (% blastocysts = % 2-cell embryos developed to the blastocyst stage). B, C. Developmental curves of selected sets of experimental groups starting from 2-cell stage. (B) Comparison of control groups. (C) Comparison of Tet2-MO1 and Tet2-MO2 groups and Tet1 + 2 combined KDs. Indicated significances were tested using log rank (Mantel-Cox) test (ns = non-significant, numbers of analyzed embryos per experimental group are indicated in parenthesis).

Our observation using a morpholino-based knockdown approach in fully grown oocytes stands in contrast to phenotypes observed upon germline deletion of all three Tet enzymes, where phenotypes are emerging in early postimplantation development (Dai et al, 2016). Both approaches have advantages and limitations and target different questions. The Tet-TKO approach diminishes Tet proteins during the growth phase of the oocyte—a phase where massive DNA remethylation is still occurring and compensation mechanisms can intervene, while the morpholino-based knockdown approach is effective in fully grown and transcriptionally silent oocytes, minimizing compensatory effects and only mimicking loss of Tet function specifically during the oocyte-to-embryo transition. Moreover, the Tet-TKO and most single Tet-KO models aimed to create genomic deletions of the catalytical domain of Tet enzymes leaving the possibility of the expression of a truncated product (Gu et al, 2011; Ko et al, 2011; Zhang et al, 2013; Dai et al, 2016), while the morpholino-based KD diminishes the expression of the complete gene product. Albeit Cre/lox-mediated KO approaches are highly specific and KD approaches bare possibilities of unspecific off-target effects, KO strategies can lead to unspecific phenotypes due to non-sense mediated decay of truncation constructs (El-Brolosy et al, 2019; Ma et al, 2019; Wilkinson, 2019), whereas in MO-based KD approaches, the full-length mRNA is still generated but blocked for translation.

To decipher individual roles of Tet enzymes during early embryogenesis, we generated single Tet enzyme-KD embryos. Tet3 is the most abundant Tet enzyme in mouse oocytes and mediates the conversion of 5mC to 5hmC (Gu et al, 2011; Wossidlo et al, 2011). Our Tet3 knockdown approach (Tet3-KD) completely diminished Tet3 protein in the zygote and confirmed the importance of Tet3 for the 5mC to 5hmC conversion by drastically reducing 5hmC levels in early zygotes (Figs EV2A and B). Moreover, Tet3-KD embryos demonstrated a decreased blastocyst formation (28% blastocysts, Fig 1C and D, and Movie EV2) suggesting that Tet3 is important for efficient preimplantation development, but not essential. Similar observations were made in Tet1 + 3 double KO mouse embryos (25% arrest before 8-cell stage (Kang et al, 2015)) and Tet3-KD bovine embryos (blastocyst rate decreases from 19 to 3% in Tet3-KD (Cheng et al, 2019)). Regarding the only partially impaired preimplantation development of Tet3-KD embryos, it is tempting to assume that passive DNA demethylation can compensate for Tet3 mediated demethylation. Here, while Tet3 might be targeted to specific genomic loci, this compensation process might be more unspecific, which can result in more variable expression profiles in early embryos (Kang et al, 2015) or, as observed in our experiments, in the early developmental arrest of few embryos. In line with that, hypermethylation in maternal Tet3-KO embryos was reported to be largely diminished by the blastocyst stage (Inoue et al, 2015). The successful, although inefficient formation of Tet3-KD blastocysts in contrast to the observed Tet-TKD phenotype, which completely inhibits blastocyst formation, suggested that Tet1/2 also contribute significantly to preimplantation development. Therefore, we next performed single and double KD (DKD) experiments for Tet1 and Tet2. Tet1-KD did not impact early embryogenesis, while Tet1 protein was successfully diminished at the 2-cell stage (Figs 1D and EV2C). Remarkably, Tet2-KD alone prevented normal preimplantation development to the blastocyst stage, with only a few embryos capable to develop to the morula stage (Fig 1D). The double KD of Tet1 and Tet2 (Tet1 + 2-DKD) did not further elevate this phenotype (Fig EV1C) suggesting a negligible role of Tet1 in preimplantation development (see also (Dawlaty et al, 2011)).

Click here to expand this figure.

Figure EV2. Analysis of Tet3- and Tet1-knockdown efficiency

A, B. Analysis of Tet3-KD efficiency. (A) Representative images of Tet3 and H3K4me3 indirect immunofluorescence (IF) of 7.5 hpf zygotes derived from control or Tet3-KD GVOs. Tet3 can be detected in the paternal pronucleus (the maternal pronucleus is marked by H3K4me3-IF) of control derived zygotes, whereas Tet3 knockdown zygotes are negative for Tet3-signal (n = 12). (B) Representative images of 5mC- and 5hmC-IF of 7.5 hpf zygotes. The knockdown of Tet3 reduces the loss of 5mC and the gain of 5hmC in the paternal pronucleus (n = 14). Paternal and maternal pronuclei are indicated. C. Analysis of Tet1-KD efficiency. Representative images of Tet1 IF of G2-phase 2-cell embryos (32 hpf) of control or Tet1-KD-derived GVOs (n = 12). Control embryos show nuclear Tet1 signal, whereas Tet1-KD embryos show greatly reduced Tet1 signal.

Data information: Paternal and maternal pronuclei are indicated, Pb = polar body, scale bar = 20 μm.

These observations for Tet2-KD were unexpected and are in contrast to published Tet2-KO models, similarly to the Tet-TKD/Tet-TKO discrepancy, with Tet2-KO mice derived from Tet2+/− or Tet2−/− crosses (Ko et al, 2011; Li et al, 2011). For Tet2, we were not able to test KD efficiency via immunostaining as we did for Tet1 and Tet3 (see Fig EV2A and C); likely because Tet2 expression levels are low in mouse oocytes and early embryos (Wossidlo et al, 2011). Thus, to validate the phenotype for Tet2, we analyzed a second, non-overlapping MO (Tet2-MO2, see Appendix Table S1). This recapitulated the developmental phenotypes of the first MO (Fig EV1C). Moreover, we further verified the Tet2-KD mediated arrest in rescue experiments to exclude potential off-target effects of the Tet2-MO. To this end, we co-injected in vitro-transcribed Tet2-mRNA along with Tet2-MO2, whose target sequence is not included in the injected mRNA sequence (Appendix Fig S1A and C). As controls, we co-injected control-MO together with Tet2-mRNA and Tet2-MO2 together with GFP-mRNA (Fig 1E). The control group developed similar to non-injected embryos (Fig EV1B) and embryos co-injected with Tet2-MO2 and GFP-mRNA resulted in arrested preimplantation embryos similar to Tet2-MO2-injected embryos (Fig 1E). Importantly, co-injection of Tet2-MO2 with Tet2-mRNA significantly rescued the Tet-MO2 phenotype, with 27% of the embryos developing to the blastocyst stage (Fig 1E). Thus, our rescue experiments validated the specific phenotype we observed in Tet2-KD embryos, which differs from the reported phenotype in Tet2-KO mice which develop to term and are fertile (Ko et al, 2011; Li et al, 2011).

In summary, our analysis of Tet-KD experiments revealed a striking early developmental arrest of Tet-TKD embryos primarily at the 2-cell stage upon acute depletion of Tet enzymes in the oocyte-to-embryo transition, with the most severe phenotype linked to Tet2. In contrast to Tet-KO models, our KD studies demonstrate the essential role of Tet enzymes during preimplantation development.

Tet1/2 generates 5hmC and 5caC in the zygote

Tet3 was the only family member of Tet enzymes shown to play an important role in the conversion of 5mC to 5hmC in the zygote, with decreased 5hmC levels in Tet3-depleted zygotes (Gu et al, 2011; Wossidlo et al, 2011; Shen et al, 2014), and Fig EV2B). Given the observed phenotypes of the single Tet enzyme KDs, we questioned whether Tet enzymes function interchangeably or whether they possess non-redundant roles in the enzymatic conversion of 5mC to 5caC. Therefore, we analyzed Tet1 + 2-DKD, Tet3-KD, and Tet-TKD-derived zygotes in G2-phase (12 h post-fertilization, hpf) by 5mC-, 5hmC-, and 5caC- immunostainings (Fig 2).

Figure 2. Tet enzymes have distinct roles in DNA methylation reprogramming in the mouse zygote

A, B. Representative images of (A) 5hmC- and (B) 5mC and 5caC-IF of control (control-MO injected), Tet1 + 2-DKD and Tet-TKD zygotes at 12 hpf. Paternal and maternal pronuclei are indicated, Pb = polar body; Sp = sperm; scale bar = 20 μm. C–E. Quantification of paternal/maternal (C) 5mC-, (D) 5hmC-, and (E) 5caC-signal ratios of 12 hpf derived zygotes normalized against DNA signal. A total of 5–16 zygotes from two to three experiments per condition were analyzed. Significance to control zygotes was calculated using ordinary one-way ANOVA with Tukey's multiple comparisons test (C, E) or Welch ANOVA with Dunnett's T3 multiple comparisons test (D) and significant differences are indicated (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, data are represented as mean ± SD, with dots representing single zygotes).

As expected, control zygotes showed a loss of 5mC- and gain of 5hmC-/5caC-signals in the paternal pronucleus compared to the maternal pronucleus ([Fig 2A–E], and (Inoue et al, 2011; Wossidlo et al, 2011)). Loss of 5mC in the paternal pronucleus of Tet1 + 2-DKD zygotes is similar to control zygotes, indicating functional enzymatic oxidation of 5mC by Tet3 in Tet1 + 2-DKD zygotes and highlighting the important role of Tet3 in active DNA demethylation of 5mC. In contrast, Tet3-KD and Tet-TKD zygotes showed a reduced loss of 5mC-signal ratios between paternal/maternal pronuclei (Fig 2B and C), suggesting a loss of 5mC by passive DNA demethylation only during S-phase in these zygotes. This reduced loss of 5mC goes along with the complete abolishment of 5hmC- and 5caC-signals in Tet-TKD zygotes (Fig 2A and B). Notably, Tet1 + 2-DKD zygotes showed decreased levels of 5hmC- and 5caC-ratios compared to the control group, but still higher levels compared to Tet3-KD or Tet-TKD zygotes (Fig 2A, B, D and E). Interestingly, the Tet3-KD showed a significant higher loss of 5hmC-ratios compared to control and Tet1 + 2-DKD zygotes, but still higher ratios compared to the Tet-TKD (Fig 2A and D).

While these data confirm the important role of Tet3 in the early phase of zygotic DNA demethylation, they also indicate that the enzymatic oxidation of 5mC to subsequently 5caC in the mouse zygote is not only dependent on Tet3 activity but also in part on Tet1/Tet2. This points to non-redundancy and a specific role for individual Tet enzymes in the stepwise oxidation of 5mC during this phase of DNA methylation reprogramming. Specific contributions of Tet enzymes in the oxidation steps of 5mC were also recently reported during the differentiation of mESCs (Mulholland et al, 2020). As 5hmC and 5caC are persistent DNA modifications in the mouse 2-cell embryo (Inoue et al, 2011; Wossidlo et al, 2011) and Tet1 + 2 DKD embryos did not develop to the blastocyst stage, our results suggest important functions of specific DNA modifications in early preimplantation development. Factors that bind to 5hmC, 5fC, and 5caC are only beginning to be identified (Yildirim et al, 2011; Iurlaro et al, 2013; Song et al, 2013; Spruijt et al, 2013; Hashimoto et al, 2015; Nanan et al, 2019), and the function of these modifications regarding EGA and embryogenesis still remains unclear.

Tet-TKD 2-cell embryos are characterized by a transcriptional signature of pre-EGA embryos and cannot complete EGA

In mouse, the major wave of EGA occurs at the late 2-cell stage, and embryos failing to perform EGA do not develop beyond 2-cell stage (Aoki et al, 1997; Hamatani et al, 2004). Our findings that the vast majority of Tet-TKD embryos arrest at the 2-cell stage provided support for the concept that Tet enzyme function impacts EGA. Hence, we hypothesized that Tet enzymes and conversion of 5mC to 5caC are essential prerequisites for EGA. To test this hypothesis, we analyzed Tet-TKD and Tet3-KD-derived 2-cell embryos at G2-stage by single-embryo RNA-Seq (Dataset EV1). In addition, we also performed single-embryo RNA-seq on α-amanitin-treated control-MO derived 2-cell embryos, which selectively inhibits RNA polymerase II/III, to obtain a list of EGA genes (Dataset EV2).

Principal component analysis (PCA) of RNA-Seq results revealed segregation of Tet-TKD from control embryos along PC1, with Tet3-KD embryos clustering between the two samples (Fig 3A). 1,860 genes were downregulated in Tet-TKD 2-cell embryos, whereas only 337 genes were significantly decreased in Tet3-KD embryos (Fig 3B). 1,309 and 360 genes were upregulated in Tet-TKD and Tet3-KD embryos, respectively. Notably, when comparing differentially expressed genes to EGA genes, the vast majority of genes downregulated in Tet-TKD and Tet3-KD overlapped with EGA genes, 1,576 of 1,860 (85%) and 310 of 337 (92%), respectively (Fig 3C), while upregulated genes are depleted for EGA genes, indicating a significant contribution of Tet enzymes to EGA. In total, Tet-TKD embryos failed to activate 52% of EGA genes (1,576 of 3,049), whereas Tet3-KD had a minor effect on EGA genes (12%, 381 of 3,049). These findings were further corroborated by comparing hierarchical clustering of differentially expressed genes, in which Tet-TKD 2-cell embryos clustered with transcriptionally inhibited α-amanitin embryos and Tet3-KD with control-MO injected embryos (Appendix Fig S2A and B). Together, these results demonstrate that Tet-TKD embryos are deficient in completely activating the embryonic genome.

Figure 3. Tet enzymes are required for completion of embryonic genome activation

Principal component analysis of RNA-Seq data of genes from single Tet-TKD (purple), Tet3-KD (coral), and control (gray) 2-cell embryos. Each dot represents a single 2-cell embryo (control n = 11, Tet3-KD n = 11, Tet-TKD n = 11). MA plots of RNA-seq data from Tet3-KD/control (left) and Tet-TKD/control (right) 2-cell embryos. Significant differentially expressed genes are colored. The amount of up- and downregulated genes is indicated. Differentially expressed genes were determined using DEseq2 (Padj < 0.05, |log2FC| > 0.7). Venn diagrams visualizing the overlap of differentially expressed genes in Tet3-TKD, Tet-TKD, and embryonically activated genes (EGA) in 2-cell embryos. EGA genes were calculated with DEseq2 comparing alpha-amanitin-treated and control 2-cell embryos (Padj < 0.05, log2FC > 0.7). Comparison of transcriptional profiles of Tet3-KD and Tet-TKD 2-cell embryos to a published database of early embryonic transcripts (DBTMEE, see (Park et al, 2015)). Shown are overlaps of DBTMEE categorized genes with the differentially expressed genes in Tet3-KD or Tet-TKD embryos compared with control 2-cell embryos.

Next, we compared transcription profiles of Tet3-KD and Tet-TKD 2-cell embryos with a public database of early mouse embryonic transcriptomes (DBTMEE (Park et al, 2015)). This comparison revealed a clear transcriptional signature for an arrest of Tet-TKD embryos before the major wave of EGA (Fig 3D) and shows that genes, which are detected as upregulated in KD embryos, are in the majority transcripts which are maternally degraded or activated during minor EGA in control embryos. The transcriptome of Tet3-KD embryos revealed only a minor arrest signature, which was consistent with the only partly impaired developmental progression of Tet3-KD embryos (Fig 3D). Importantly, the transcriptional arrest of Tet-TKD embryos before the major wave of EGA is independent of cell cycle progression, as Tet-TKD embryos still undergo replication at the 2-cell stage (Fig EV3).

Click here to expand this figure.

Figure EV3. Tet-TKD embryos undergo S-phase in the 2-cell stage

Representative images of 2-cell embryos derived from non-injected, control-morpholino, or Tet-1–3 MO injected GVOs, which were incubated with EdU from 22.5 hpf until 27.5 hpf and analyzed for EdU incorporation. All three groups show similar incorporation of EdU indicating that Tet-TKD 2-cell embryos undergo replication in the 2-cell stage (Tet-TKD: n = 5, non-injected: n = 3, control-MO: n = 2; scale bar = 20 µm).

We performed gene ontology (GO) analysis of genes that are less abundant in Tet3- and Tet-TKD 2-cell embryos and found highly significant GO processes enriched for terms like “RNA metabolic process, RNA processing, ncRNA processing, translation”, with in general lower P-values for Tet3-KD than Tet-TKD embryos (Table EV1). Additionally, genes downregulated in Tet-TKD embryos were enriched for GO processes like “metabolic process, mRNA processing, RNA splicing, cell cycle”, mirroring the developmental phenotype of Tet-TKD embryos and suggesting failures in establishing biological processes that are vital for early development.

When performing ChIP Enrichment Analysis for downregulated genes, we found many prominent transcription factors and epigenetic modifiers essential for early development (Appendix Table S2). Most enriched factors appeared in both Tet3-KD and Tet-TKD downregulated gene sets. Targets of Tet1, the only present Tet enzyme in the reference dataset, were detected as significantly enriched in Tet-TKD embryos only. For this analysis, we analyzed published datasets from closely related mouse embryonic stem cells (mESCs) (Lachmann et al, 2010; Kuleshov et al, 2016), as ChIP-Seq experiments in preimplantation embryos are technically not feasible yet. This comparison to mESCs suggests that downregulated genes in Tet-TKD are, at least partially, direct targets of Tet enzyme activities; however, as Tet enzymes target DNA not directly, these correlation analyses need to be interpreted with caution.

Tet enzyme-depleted embryos dysregulate transposable elements and do not activate MERVL-driven genes

Since early mouse development is characterized by a transcriptional burst of (retro)transposable elements at the 2-cell stage, we analyzed the impact of Tet enzyme depletion on the expression of transposable elements (TEs). PCA of TE expression levels showed a tendency for segregation of Tet-TKD from control embryos along PC1, with Tet3-KD embryos scattered in between both groups (Appendix Fig S3A). Overall, Tet-TKD embryos were characterized by misregulation of many TEs, which was not as strongly pronounced in Tet3-KD embryos (Fig 4A, and Appendix Figs S3B and S4, Dataset EV3). The class of long interspersed nuclear elements (LINEs) was slightly upregulated in Tet-TKD embryos (Fig 4A), indicating that Tet enzymes were not required for the activation of LINE elements, as also shown before upon Tet3-KD (Inoue et al, 2012). Notably, short interspersed nuclear elements (SINEs) were tendentially downregulated in Tet-TKD embryos (Fig 4A and Appendix Fig S4). SINEs compromise about 8% of the mouse genome and are typically methylated to prevent transposition (Meissner et al, 2008). SINEs can cause hypermethylation of nearby genes (Estecio et al, 2012), which could play an important role for EGA and will need further investigation. Interestingly, long terminal repeat (LTR) elements showed up- and downregulation of specific class members (Fig 4A, and Appendix Fig S4). The class III of endogenous retroviruses (ERV3- including MERVL and MaLR elements) has been reported for its element-specific differential expression during embryonic development (Franke et al, 2017) and ERV3 elements are regulated via different LTR sequences. Notably, ERV3-LTRs that are expressed during late oogenesis showed higher expression levels in Tet-TKD embryos compared to control embryos (MTA, MTB; Fig 4B) and ERV3-LTRs, which are highly expressed specifically in two-cell embryos, are significantly downregulated in Tet-TKD embryos (MERVL-LTR, ORR1a2; Fig 4C). These observations further corroborated a characteristic transcription profile of embryos arrested pre-EGA and suggested that this developmentally important class of TEs are also regulated by Tet enzymes in the early embryo.

Figure 4. Tet enzyme deficiency dysregulates specific classes of transposable elements (TEs) and 2C genes

A. Expression changes of different classes of TEs in Tet3-KD and Tet-TKD. Shown are log2-fold changes of KD embryos compared to control 2-cell embryos of different repetitive elements subdivided into classes of TEs. Indicated P-values were calculated using a paired two-tailed t-test (***P < 0.001, ****P < 0.0001, data are represented as medium smoothed violin plots with indicated median and quartiles as dotted lines). B, C. Expression levels of (B) maternally expressed ERV3-LTRs and (C) 2-cell embryo specific ERV3-LTRs in control, Tet3-KD and Tet-TKD 2-cell embryos (P-values were calculated using DEseq2; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, data are represented as mean ± SD, dots represent single embryos). D. Volcano plots showing expression changes of genes in Tet3-KD and Tet-TKD 2-cell embryos. Chimeric genes, as defined by the presence of a nearby MERVL element and upregulation in 2-cell embryos, are indicated (see (Macfarlan et al, 2012)). Dashed lines represent the cutoff for differential expression: |log2FC| > 0.7 and Padj < 0.05. E. Expression levels of important candidate factors in control, Tet3-KD and Tet-TKD 2-cell embryos, implicated in the generation of 2C-like cells and as important regulators of EGA (P-values were calculated using DEseq2; *P < 0.05, **P < 0.01, ****P < 0.0001, data are represented as mean ± SD, dots represent single embryos).

MERVL elements have been reported to facilitate the expression of a set of genes during EGA—often as new chimeric transcripts using their LTR sequences as alternative promoters (Macfarlan et al, 2012). Remarkably, 51% of MERVL-driven genes, which were shown to be activated in 2-cell embryos, were significantly downregulated in Tet-TKD embryos (72 of 141 detected genes, Fig 4D; see (Macfarlan et al, 2012), Table EV2). In contrast, only 9% of MERVL-driven genes were significantly downregulated in Tet3-TKD embryos (12 of 141, Fig 4D). In addition, hierarchical clustering for chimeric genes revealed the misregulation of these genes in Tet-TKD embryos and to a lower extent also in Tet3-KD embryos (Appendix Fig S5). We also observed a significant decrease of chimeric junction usage in MERVL-driven chimeric genes in Tet3-KD and Tet-TKD compared with control 2-cell embryos (Fig EV4A–C, Table EV3).

Click here to expand this figure.

Figure EV4. Expression analysis of MERVL-driven chimeric transcripts

Normalized ratio of reads spanning the repeat-exon junctions of “MERVL-chimeric” genes according to MacFarlan et al (2012). Statistical significance was tested using Friedman test with Dunn's multiple comparisons test (**P < 0.01). Data from all embryos of one condition (11 embryos each) were merged for this analysis. Data are represented as box plots with indicated mean and quartiles with whiskers representing the min and max values; dots represent different repeat-exon junctions. Percent of analyzed “MERVL-driven chimeric” genes, which show a 2-fold decrease of the usage of the chimeric junction compared to control embryos. RNA-seq data from all embryos of one condition were merged for this analysis. Note: In α-Amatinin (α-AM) 2-cell embryos for 85% of genes no junctions could be observed (only genes which had more than 1,000 reads mapping to the whole transcript were considered). Screenshots of IGV genome browser showing RNA-seq data of control, Tet3-KD, Tet-TKD, and α-AM treated 2-cell embryos at chimeric repeat-exon junction sites of the “2C-genes” Gnpnat1 and Ube2e3.

Recently, several factors have been reported as master regulators to induce EGA and the expression of MERVLs. While quite a few factors are known to activate the expression of MERVL elements and “2C genes” in mESCs, which induce a so-called “2C-like” state (Macfarlan et al, 2012; Eckersley-Maslin et al, 2019), two prominent transcription factors, Dux and Nfya, stand out as potential master regulators or pioneering factors in 2-cell embryos. Three studies suggested that Dux, an early-induced gene in mouse preimplantation development, acts as a pioneering factor activating promoter regions of MERVL elements and 2C genes (De Iaco et al, 2017; Hendrickson et al, 2017; Whiddon et al, 2017). Interestingly, Dux expression was upregulated in Tet-TKD embryos (Fig 4E). While some targets of Dux, like Zscan4, a prominent factor expressed in 2C-stage embryos and a regulator of the 2C-state in mESCs (Rodriguez-Terrones et al, 2018; Srinivasan et al, 2020), were still activated in Tet-TKD embryos, the MERVL-driven chimeric genes were not activated to full levels (Fig 4D and E). Next to this, the maternal factor Nfya, which is implicated in chromatin opening during the early phase of EGA (Lu et al, 2016), did not show expres