Obox4 promotes zygotic genome activation upon loss of Dux

Once fertilized, mouse zygotes rapidly proceed to zygotic genome activation (ZGA), during which long terminal repeats (LTRs) of murine endogenous retroviruses with leucine tRNA primer (MERVL) are activated by a conserved homeodomain-containing transcription factor, DUX. However, Dux-knockout embryos produce fertile mice, suggesting that ZGA is redundantly driven by an unknown factor(s). Here, we present multiple lines of evidence that the multicopy homeobox gene, Obox4, encodes a transcription factor that is highly expressed in mouse two-cell embryos and redundantly drives ZGA. Genome-wide profiling revealed that OBOX4 specifically binds and activates MERVL LTRs as well as a subset of murine endogenous retroviruses with lysine tRNA primer (MERVK) LTRs. Depletion of Obox4 is tolerated by embryogenesis, whereas concomitant Obox4/Dux depletion markedly compromises embryonic development. Our study identified OBOX4 as a transcription factor that provides genetic redundancy to preimplantation development.


Introduction
The mechanism by which zygotes acquire totipotency is a major question in Developmental Biology.Following fertilization, the zygote must build a de novo conceptus with a transcriptionally quiescent genome.The genome rapidly undergoes zygotic genome activation (ZGA), during which epigenetic reprogramming and expression of nascent transcripts enforce the replacement of parental infrastructures by their zygotic counterparts 1 .Completion of metazoan ZGA results in blastomeres, a collection of cells that are totipotent enough to reflect their potential to individually produce both embryos and extraembryonic appendages 2,3 .Totipotency lasts until the morula stage, where the first cell -2 -fate decision takes place, following which it is remolded into pluripotency in cells destined for the embryonic lineage 4 , to which embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs) correspond.In placental mammals, ZGA is characterized by the massive reactivation of transposable elements (TEs) and epigenome remodeling, predominantly implemented by a collection of ZGA genes that have adapted long terminal repeats (LTRs) of endogenous retroviruses (ERVs) as stage-specific cis-regulatory elements 5 .
In this study, we sought to identify the redundant transcription factor that drives ZGA in the absence of Dux.We discovered that the multicopy homeobox gene, oocyte-specific homeobox 4 (Obox4), promotes Dux-less ZGA.Obox4 is abundantly expressed in mouse 2C-embryos, 2C-like mESCs, and totipotent blastomere-like cells (TBLCs) 22 .
Mechanistically, OBOX4 promotes ZGA by binding to the LTRs of murine endogenous retroviruses with leucine tRNA primer (MERVL) and murine endogenous retroviruses with lysine tRNA primer (MERVK), and thereby affecting the deposition of active epigenetic modifications.Concomitant, but not respective, depletion of Obox4 and Dux severely compromises ZGA and preimplantation development.Taken together, our findings substantiate a preimplantation development model, in which the ZGA is redundantly promoted by Dux and Obox4.

Expression profiling identifies transcription factor candidates
Genetic redundancy is commonly observed among genes that share sequence homology 23 .Because Dux is a homeobox gene, we speculated that its redundant factors are homeobox genes.We examined the expression profiles of all mouse homeobox genes during pre-implantation embryogenesis using published single-cell RNA-seq (scRNA-seq) data 24 .Clustering analysis defined a collection of 45 homeobox genes, whose transcript levels reached a maximum before the late 2C stage (Fig. 1a).We then sought to identify bona fide ZGA factors from this collection by looking for genes whose transcripts were of zygotic origin, instead of ones that were parentally inherited.TBLCs have recently been established as in vitro counterparts of 2C-embryos, which are derived from splicing-inhibited mESCs, and hence, are free of parentally inherited transcripts 22 .Analysis of published RNA sequencing (RNA-seq) data showed that the expression of four homeobox genes, Duxf3, Emx2, Hoxd13, and Obox4, was ten times higher in TBLCs than in ESCs (Fig. 1b).We examined whether the expression of these genes was affected in the Dux knockout embryos.Using published RNA-seq data 18,20 , we found that the expression of all single-copy candidate genes and a subset of the multicopy gene Obox4 was unaffected in Dux knockout embryos (Fig. 1c).

Obox4 activates 2C-genes and TEs in mESC
Before examining whether the candidates were crucial for embryogenesis, we first confirmed the presence of endogenous OBOX4 protein, as the Obox4 loci are marked as pseudogenes in the current genome annotation GRCm39 25 .We generated mouse anti-OBOX4 monoclonal antibodies.Immunofluorescence staining using these antibodies confirmed that endogenous OBOX4 was expressed in zygotes and highly abundant in 2C-embryos and 2C-like mESCs (Fig. 1d-e), which is consistent with translatome profile captured by Ribo-seq that detects OBOX4 translation at 2C stage 26 (Fig. 1f).Mouse ZGA is characterized by the surging activity of genes that specifically express at 2C stage (2Cgene), many of which have co-opted MERVL LTRs as stage-specific promoters 27 .We then sought to examined whether the candidates had potential to promote ZGA by activating MERVL and 2C-genes (Supplementary Fig. 1a and Supplementary Table 1-2).We constructed a 2C::tdTomato reporter mESC line with transgenic tdTomato red fluorescence protein driven by MERVL 5'-LTR. 28(Fig. 2a).Candidate genes were cloned and overexpressed in the reporter cell line (Supplementary Fig. 1b-c).Interestingly, ectopic expression of Obox4 markedly induced tdTomato expression, similar to that induced by Dux (Fig. 2b and Supplementary Fig. 2a-b).To characterize the impact of Obox4 expression on 2C-genes, we established an ESC stable line bearing an inducible Obox4 transgene (Fig. 2c).Differentially expressed gene analysis revealed that Obox4 induction led to transcriptome changes in mESCs, characterized by upregulation of 2C-genes and TEs that were highly expressed in early-to-middle 2C-embryos 29 (Fig. 2d-e), naturally occurred 2C-like mESCs 30 , and Dux induced 2C-like mESCs 16 (Fig. 2f, Supplementary Table 3).A substantial fraction (159/164) of Obox4 induced 2C genes were also induced by Dux (Fig. 2g), while transcriptomic perturbance induced by Obox4 is milder comparing with Dux, characterized by less number and less fold-upregulation of 2C genes (Fig. 2g-h).
Interestingly, Dux and Obox4 were mutually inductive, where one promoted expression of the other (Fig. 2h).These results suggest that Obox4 is an inducer of 2C-like genes and potentially redundant factor of Dux.

Obox4 binds to 2C-gene promoters and LTR elements in mESC
Obox4 contains a homeobox, activates 2C-genes and TEs, and upregulates Dux-induced genes, which prompted us to examine whether OBOX4 is a transcription factor that activates 2C-gene associated loci through direct binding with cis-regulatory elements.Cleavage under targets and release using nuclease (CUT&RUN) leverages antibody-targeted cleavage of proximal DNA to identify the binding sites of DNA-associated proteins 31 .To examine this technique, we performed CUT&RUN against triple FLAG-tagged DUX (3×FLAG-DUX) expressed in mESCs using a high-affinity anti-FLAG antibody 32 , and confirmed that the DUX binding pattern revealed by CUT&RUN recapitulated the published HA-tagged DUX ChIP data 16 (Supplementary Fig. 3a-b).We then proceeded with characterizing genomic footprint of OBOX4 by performing CUT&RUN against 3×FLAG-OBOX4 expressed in mESCs, and discovered ~24,000 peaks, among which 39.5% located in the gene promoter regions that covered 26.8% (273/1,019) of the 2Cgenes (Fig. 3a).De novo motif discovery using the top 500 CUT&RUN signal peaks predicted CTGGGATYWRMR as top OBOX4 binding motif, which is enriched in the promoter regions of 2C-genes (Fig. 3b).Collectively, OBOX4 and DUX targeted 48.1% (490/1,019) of the 2C-genes with a considerable overlap (36.3% for OBOX4 and 31.3% for DUX), which covered many important ZGA genes including Dppa2, Sp110, and Zscan4d (Fig. 3c-d and Supplementary Table 4).The overlap was substantiated by the observation that MERVL LTR MT2_Mm was among the top 10 LTR targets of both OBOX4 and DUX, based on loci coverage (Fig. 3e-f and Supplementary Table 5).Notably, while DUX strongly prefers MT2_Mm, OBOX4 binding was biased toward MERVK LTRs, namely RLTR9 and RLTR13 elements (Fig. 3g), which also demonstrated a 2C-specific expression profile (Supplementary Fig. 3c-d).To examine whether these bindings were functional in the absence of Dux, we generated Obox4/Dux single and double knockout mECS lines, in which 14 of 15 protein-coding Obox4 copies were removed from the genome (Supplementary Fig. 4-5).tdTomato reporters driven by MT2_Mm and RLTR13B2 were co-transfected with Dux and Obox4 into Obox4/Dux double knockout ESCs (Fig. 3h).Analysis of the tdTomato-positive population showed that Obox4 activated both RLTR13B2 and MT2_Mm, whereas Dux only activated MT2_Mm (Fig. 3i).These observations demonstrated that OBOX4 binds and activates a subset of DUX targets in mESCs and redundantly drives their expression in the absence of DUX.

Concomitant loss of Obox4 and Dux impairs preimplantation development
We then sought to determine whether Obox4 is functionally required for pre-implantation development, particularly in the absence of Dux.Transient depletion of DUX has previously confused the field in that it caused an embryonic phototype while subsequent Dux knockout females were shown to be fertile 15,[18][19][20][21] .Therefore, it is critical to examine the functional requirement of OBOX4 and DUX in genetic knockout models.Somatic cell nuclear transfer (SCNT) is a technique to create embryos by transferring nuclei of somatic cells into enucleated oocytes (Fig. 4a), which recapitulates ZGA 33 .When the knockout mESCs were subjected to SCNT as nuclear donors, 54.4% of WT mESC derived embryos developed to blastocyst stage at 4 days post nuclear transfer (dpt).Obox4 and Dux single knockout donors led to reduced blastocyst formation rate at 38.5% and 39.7% respectively, whereas double knockout resulted in additive effect that led to significantly lowered blastocyst rate of 29.3% (Fig. 4b).Notably, all the blastocysts derived from double knockout mESC were of low quality, judging from severe morphological abnormality (Fig. 4c).
As an alternative genetic knockout model, we generated mice bearing Dux and Obox4 knockout allele by direct CRISPR-Cas9 editing in embryos (Fig. 4d and Supplementary Fig. 4a-b).Among 39 Obox4 copies, there are 15 copies that maintain intact open reading frame (ORF) for full-length OBOX4 (Supplementary Fig. 4c).While 14 of the 15 intact ORFs form tightly packed cluster (Obox4 cluster), a solo ORF (Obox4-ps33) remains distant from the Obox4 cluster and is interspersed with Obox1/2/3, a collection of Obox family members that are critical for ZGA 34 .To minimize collateral genetic toxicity and interference to experiment result caused by removal of other Obox members, only Obox4 cluster was knocked out and the solo ORF was retained (Supplementary Fig. 4d and Supplementary Fig. 6).The observed Obox4 KO frequency (7/29) in the progenies of Obox4 Het × Obox4 Het crossing is consistent with the expected 25% Mendelian ratio (Fig. 4e, Supplementary Fig. 7, and Supplementary Table 6).That Obox4 KO mice developed to adulthood without discernable abnormalities and are fertile when intercrossed suggest that development is compatible with loss of Obox4 (Fig. 4f-g).We next attempted to produce Dux/Obox4 double knockout (Dux KO /Obox4 KO ) mice to examine whether concomitant loss of Dux and Obox4 compromises embryogenesis.We genotyped 54 pups from three litters of Dux KO /Obox4 Het × Dux KO /Obox4 Het and six litters of Dux KO /Obox4 Het × Dux Het /Obox4 Het mating pairs (Supplementary Fig. 8 and Supplementary Table 6).One Obox4 KO /Dux KO pup was present, and the frequency of Obox4 KO /Dux KO and Obox4 KO /Dux Het were significantly lower than the expected 25% Mendelian ratio (Fig. 4h).
Development monitoring and genotyping of embryos produced by Dux KO /Obox4 Het × Dux Het /Obox4 Het mating pairs at 4.5 days post cotium (dpc) revealed that Dux KO /Obox4 KO was under-represented in blastocysts while over-represented in 2-cell arrest and degenerated embryos (Fig. 4i, Supplementary Fig. 9a-c, and Supplementary Table 7).Single blastomere genotyping and RNA-seq of Dux KO /Obox4 KO 2C embryos (Supplementary Fig. 9d-e) showed dysregulation of 2C-genes and TEs targeted by OBOX4 (Fig. 4j).The impaired development and transcriptome of Dux KO /Obox4 KO embryos at 2-cell stage showed that ZGA was defective in those embryos.
As a transient depletion model in addition to genetic knockout approaches, microinjection of antisense oligonucleotide (ASO) into male pronuclei of zygotes was performed to knockdown Obox4 and Dux (Fig. 4k and Supplementary Fig. 10a-c).Monitoring development until 4.5 dpc showed that Obox4 single knockdown resulted in moderate developmental retardation, with almost 50% of embryos reaching the blastocyst stage.Similarly, blastocyst formation was preserved in nearly 25% of Dux single knockdown embryos, which is consistent with previously reported Dux knockdown/knockout experiments 15,[18][19][20][21] .The Obox4/Dux double knockdown markedly compromised blastocyst formation, resulted in more than 80% embryos degenerated or arrested before the 4C stage, and less than 20% manifested morula-like morphology (Fig. 4k-l).The double knockdown embryos with morula-like morphology manifest dysregulated transcriptome characterized by failure of expressing morula specific genes and activation of apoptosis related pathways, suggesting that the development was dysfunctional instead of delayed (Supplementary Fig. 10e-g).
The consistency among constitutive knockout in SCNT embryos and living mice and ASO-mediated transient depletion demonstrated that the expression of Obox4 and Dux is collectively important for pre-implantation development, and that Obox4 is capable of promoting ZGA in a Dux-independent manner.Collectively, these data show that Obox4 promotes mouse preimplantation development in the absence of Dux.

OBOX4 promotes 2C-gene expression upon depletion of DUX
As Obox4 has been shown to activate 2C-genes and TEs in mESCs, and embryonic depletion of OBOX4 impairs 2C-genes and TEs activation, we questioned whether the impairment can be ameliorated by restoration of OBOX4.
First, we determined whether the presence of OBOX4 can rescue the developmental arrest of Obox4/Dux double knockdown embryos.Codon-optimized Obox4 mRNA without the ASO target motif was produced using in vitro transcription.The rescue experiment was performed by means of co-microinjection of Obox4 mRNA with Obox4/Dux double knockdown ASOs (Fig. 5a).Restoration of OBOX4 in Obox4 mRNA-microinjected embryos was confirmed using immunofluorescence staining at the 2C stage (Fig. 5b).As expected, development monitoring showed that Obox4/Dux double knockdown embryos failed blastocyst formation at 4.5 dpc, whereas among Obox4 mRNA-rescued embryos, blastocyst formation was retained at a similar level to that observed in the Dux single knockdown experiment (Fig. 5c-d and Fig. 4k-l).RNA-seq of 2C-embryos revealed exacerbated dysregulation of 2C-genes and TEs by Obox4/Dux double knockdown comparing to single knockdown, and the dysregulated transcriptome of double knockdown was rescued by resupplying OBOX4, revealed by number of up and down-regulated genes (Fig. 5e-f).
Differential expression analysis revealed that 58% (102/175) of the down-regulated 2C-genes in double knockdown embryos were rescued, including 2C stage markers and important ZGA factors (Fig. 5g-h).These results showed that that OBOX4 redundantly activates 2C gene to promote ZGA under DUX depletion.

Discussion
Starting with a transcriptionally inert genome, the eutherian ZGA engages in massive yet coordinated expression of genes driven by LTR TEs.In mice, it has been reported that the transcription factor DUX accesses and opens condensed chromatin of the MERVL LTR loci, which results in the activation of downstream ZGA genes [15][16][17] .
However, the establishment of fertile Dux knockout mice suggests that ZGA is redundantly driven by other transcription factor(s) [18][19][20][21] .In the present study, we found that Obox4, a cryptic multi-copy cluster gene, encodes a homeodomaincontaining protein that potently induces ZGA gene expression.Using multiple genetic knockout models, we showed that concomitant depletion of Dux and Obox4 was hardly compatible with embryogenesis, characterized by Dux/Obox4 double knockout mESCs compromised SCNT and mouse produced at sub-Mendelian ratio.Consistently, Obox4/Dux double knockdown markedly compromised pre-implantation development, which tolerated single knockdown of either gene.Double knockdown embryos exhibited severely dysregulated transcriptomes, characterized by MERVL and ZGA gene activation failure.We also characterized the molecular mechanisms underlying the biological significance of Obox4 as well as its relevance to Dux.OBOX4 directly binds to MERVL (the target of DUX) and MERVK loci and activates specific MERVL and MERVK elements in a Dux-independent manner.In summary, our results highlight that OBOX4 is a transcription factor that is functionally redundant to DUX during ZGA.
Obox transcripts were first discovered in gonads 35 , and were later found to be highly abundant in mouse 2Cembryos 36 .The Obox family has 67 members clustered in the sub-telomeric region of mouse chromosome 7 and is further divided into six subfamilies: Obox1, Obox2, Obox3, Obox4, Obox5, and Obox6, with Obox4 constituting nearly 60% (39 of 67) of the family population 37,38 .Although Obox4 dominates the Obox family population, members of the entire subfamily have been annotated as pseudogenes and have largely been neglected from investigation.Despite being abundantly transcribed in the gonads, the biological significance of Obox genes is poorly understood.The role of Obox4 is particularly mysterious, as it remains to be determined whether Obox4 is a functional protein-coding gene.
Transcription of Obox4 was originally reported to be testis-specific and contradicted by the detection of Obox4 transcripts in the ovaries and oocytes 39 .Obox4 is involved in oocyte maturation and ESC differentiation [40][41][42][43][44] .However, these observations are supported by limited evidence, with no evidence at the protein level.By detecting OBOX4 in various totipotent cell types using high-quality monoclonal antibodies, we provided decisive evidence that Obox4 is a functional protein-coding gene.Our data demonstrated that OBOX4 binds to genomic loci of MERVL-derived promoters and MERVK-derived enhancers and mediates their activation during ZGA.However, whether and how other Obox family members contribute to ZGA remains unclear.
It is worth noting that the ability of OBOX4 to promote ZGA appears to be weaker than that of DUX, characterized by lower fold upregulation of 2C-genes upon overexpression of OBOX4 than DUX, non-induction of MERVL Gag protein in mESCs, less dysregulated transcriptome in Obox4 depleted embryos, and less adversarial phenotype in Obox4 depleted embryos and mice.On the other hand, the severe cytotoxicity borne by DUX 45 cannot be observed in OBOX4, as ectopic expression of OBOX4 in mESCs and embryos did not show hindrance in cell viability.Whether the variance in potency to promote ZGA has additional biological significance other than functional redundancy warrants further investigation.
Unlike Dux, which is structurally and functionally conserved throughout placentalia 46 , the ancestral locus of the Obox family appears to have undergone mouse-specific duplication and generated a gene cluster that is collectively -7 -syntenic to the Tprx2 locus in other mammals 47 .Despite the distal homology between Obox4 and Tprx2, it has been recently reported that human TPRX2 is expressed in 8-cell embryos 14 .Interestingly, TPRX2 was shown to cause defective ZGA upon embryonic depletion and bind important ZGA genes in hESCs 48 , suggesting that the Tprx2 locus has undergone functionally convergent evolution despite its divergent genetic context.However, whether TPRX2 plays a similar role in humans to Obox4 in mice regarding the redundancy to Dux (DUX4 in humans) remains to be elucidated.Genetic redundancy is widespread in higher organisms and typically arises in signaling networks, in which multiple functionally overlapping factors commit to a shared teleological objective, to counteract sporadic mutations or defects 49,50 .This principle is of particular importance to embryogenesis, a process that gives rise to a reproducing organism.Indeed, genetic redundancy in pluripotency has been evidenced by the observation that naïve pluripotency can be elicited and maintained through various "independent inputs that operate through both unique and convergent targets" 51 .Although the molecular pathway that governs totipotency has been largely restricted to the scope that centers on Dux, it is reasonable to assume that the acquisition of totipotency also complies with the rule of redundancy.This assumption was validated by the generation of fertile Dux knockout mice; however, the panorama of ZGA redundancy remained hidden.Our work on Obox4 sheds light on this topic by discovering that Obox4 serves as a redundant gene to Dux during ZGA and is a bivalent activator of MERVL and MERVK elements.Yet that concomitant loss of Dux and Obox4 is viable reveals that ZGA is so strongly canalized that many differences in genes make very little difference to the phenotype.Although the function of MERVL has been extensively investigated, particularly during early embryogenesis, the role of MERVK remains unclear.Interestingly, recent studies have suggested that MERVK serves as a meiosis-specific enhancer of spermatogenesis 52 .Mammalian Y chromosomes have undergone rapid evolution and have been intensively inserted by TEs 53 .As Obox4 is highly expressed in the testes, it is worth investigating whether Obox4 plays a regulatory role during spermatogenesis, by activating MERVK-derived enhancers.
Pioneer factors possess the ability to specifically recognize and access DNA motifs that are inaccessible to other transcription factors 54 .In the scope of pluripotency-to-totipotency transition, OBOX4 and DUX are qualified pioneer factors for they are capable of accessing and activating silenced 2C-genes in ESCs.However, it is not likely that such pioneer activity demonstrated in vitro is accounted for ZGA initiation in vivo, considering that genome-wide demethylation precedes the timing of OBOX4 and DUX translation 55 , and neither individual nor collective depletion of OBOX4 and DUX caused strict 2C arrest.Recent studies have identified the maternally deposited orphan nuclear receptor NR5A2 that triggers ZGA by activating short interspersed nuclear element (SINE) B1 family 56,57 , while DNA damage induced p53 activation is important but dispensable for Dux activation in mouse zygotes 58 .These results suggest another layer of genetic redundancy at ZGA initiation level.Hence, how Dux and Obox4 are activated after fertilization and whether they share the same activation mechanism requires further interrogation.culture vessel area, before seeding mESCs into it.The cells were cultured in a humidified atmosphere containing 5% CO2 at 37°C, with the media changed every 48 h.
TBLCs were generated by culturing mESCs in an ordinary medium supplemented with 2.5 nM PLaB (catalog no. 16538, Cayman Chemicals).Cells were sub-cultured every 2-4 d, at a seeding density of 1×10 5 cm -2 culture vessel area.
After five rounds of sub-culture, the cells were deemed TBLCs.
SP2/O-Ag14 myeloma and primary clones of hybridomas were cultured in GIT medium (catalog no.63725715, Wako) supplemented with recombinant human interleukin-6 (IL-6) (catalog no.20006, PeproTech).Cells were cultured in a humidified atmosphere containing 5% CO2 at 37°C and sub-cultured every day, at a seeding density of 3×10 5 mL - 1 .For monoclonal antibody production, hybridomas were cultured in hybridoma serum-free medium supplemented with IL-6.The cells were cultured in a humidified atmosphere containing 5% CO2 at 37°C, until they were overconfluent.

Cell transfection
Plasmid transfections were performed using jetOPTIMUS ® DNA Transfection Reagent (catalog no.101000025, Polyplus), according to the manufacturer's instructions.Briefly, mESCs were seeded at a cell density of 8×10 4 cm -2 culture vessel area.After 20 min, the jetOPTIMUS ® reagent and plasmids were diluted in jetOPTIMUS ® buffer, incubated for 10 min at room temperature, and then applied to the cell culture.After 48 h, the cells were collected for downstream experiments.

cDNA synthesis and cloning
Total RNA was extracted from TBLCs using ISOGEN (catalog no.311-02501, Nippon Gene), according to the manufacturer's instructions.Briefly, 500 μL ISOGEN was added to 1.5×10 Thermo Fisher) and, for 1 h at room temperature.Following three washes in PBS, the embryos were transferred into liquid paraffin-covered PBS drops on a glass-bottom dish (catalog no.D11130H, Matsunami).
For culturing, cells seeded on glass bottom chamber slides (catalog no.SCS-N02, Matsunami) were fixed with 4% paraformaldehyde in PBS, for 10 min.After three washes in PBS, the fixed cells were permeabilized with 0.1% Triton™ X-100 in PBS, for 15 min at room temperature.Following three washes in PBS, permeabilized cells were incubated with primary antibodies diluted with 2% skim milk in PBS, for 30 min at room temperature.Following three washes in PBS, the cells were incubated with 2% BSA in PBS containing 1:500 diluted Alexa Fluor™-conjugated secondary antibody and 1:200 diluted DAPI, for 1 h at room temperature.Following three washes in PBS, the chamber was removed and the slides were mounted with ProLong™ Glass Antifade Mountant (catalog no.P36982, Thermo Fisher).
Fluorescence images were taken using a confocal laser scanning microscope (catalog no.FV3000, Olympus).

Western blot
Freshly harvested cells were re-suspended in phosphate-buffered saline (PBS) (catalog no.14249-95, Nacalai Tesque) and lysed by means of sonication.Whole cell lysates were mixed with sample buffer containing reducing reagent (catalog no.09499-14, Nacalai Tesque) and boiled at 95°C for 5 min.Protein samples were loaded on a 10% trisglycine gel, run in AllView PAGE Buffer ® (catalog no.DS520, BioDynamics Laboratory), and then transferred to nitrocellulose membranes (catalog no.10600003, Cytiva) using a Power Blotter System (catalog no.PB0012, Thermo Fisher).The membranes were blocked with PBST containing 2% skim milk (catalog no.4902720131292, Morinaga), at room temperature for 15 min.Membranes were incubated with primary antibodies in PBST containing 2% skim milk, at room temperature for 30 min.After three washes with PBST, the membrane was incubated with secondary antibodies in PBST containing 2% skim milk, at room temperature for 15 min, with shaking.After three washes with PBST, the membrane was incubated with ECL reagents (catalog no.RPN2232, Cytiva) and exposed to an X-ray film (catalog no.28906839, Cytiva) or detected using a digital chemiluminescence imager (catalog no.17001402JA, Bio-Rad).
qPCR qPCR was performed using TB Green Fast qPCR Mix (catalog no.RR430A, TaKaRa) according to the manufacturer's instructions.The qPCR reactions were carried out and the signals were detected using a real time PCR system (catalog no.TP950, TaKaRa).For RT-qPCR, first strand cDNA of mESCs total RNA was produced as described in methods cDNA synthesis and cloning section.Primers targeting MERVL Gag consensus sequence were used to detect MERVL transcript.Primers targeting beta actin mRNA were used as internal control in all qPCR assays performed in this study.smFISH smFISH probes targeting MERVL were designed and synthesized by LGC Biosearch Technologies as previously described 59 .smFISH followed by immunofluorescence staining was performed according to the manufacturer's instructions.Briefly, Quasar 570 labeled probes were hybridized against MERVL RNA at 37°C for 16 h.Followed by immunofluorescence staining described in methods Immunofluorescence section, without blocking step to prevent RNase contamination.The slides were mounted with ProLong™ Glass Antifade Mountant (catalog no.P36982, Thermo Fisher).Fluorescence images were taken using a confocal laser scanning microscope (catalog no.FV3000, Olympus).
Data analysis was performed using the Sony SH800Z cell sorter software.

Genotyping and copy number examination
For mouse and culture cell, mouse right hindlimb toe or tail tip, or 1×10 supplemented with 1 µg RNase A (catalog no.131-01461, Nippon Gene).The solution was incubated at 37°C for 1 h, then phenol/chloroform/isoamyl alcohol extracted and ethanol precipitated again as described above, to obtain highly purified genomic DNA.The DNA was dissolved in 200 µL TE solution.
For single blastomeres and abnormal embryos, whole genome amplifications were performed on individual embryos by using REPLI-g Advanced DNA Single Cell Kit (catalog no.150363, Qiagen) according to the manufacturer's instructions.For blastocysts, crude genomic DNA from single blastocysts was prepared according to the method described by Sakurai et al. 62 with some modifications.Briefly, single blastocysts in 0.5 µL KSOM medium were transferred to the bottom of 0.1 mL PCR tubes, followed by addition of 10 µL blastocyst lysis buffer containing of 120 µg/mL recombinant proteinase K, 100mM Tris-HCl pH = 7.9, 100mM KCl, 0.45% NP-40, and 10 µg/mL yeast tRNA (catalog no.AM7119, Thermo Fisher).After brief vortex and pulse spin, the tubes were incubated at 55°C for 10 min followed by 95°C for 10 min.
For PCR based genotyping, 10 ng purified genomic DNA, 10ng whole genome amplified DNA, or 2µL of blastocyst lysate were used as template and amplified with target specific primers by KOD One ® PCR Master Mix (catalog no.KMM-101, TOYOBO) according to the manufacturer's instructions.Primers targeting mouse Tardbp were used as internal control when whole genome amplification product or blastocyst lysate were used as multiplex PCR template.For copy number examination, 5 ng genomic DNA template was amplified as described in methods qPCR section.

In vitro transcription
The Obox4 coding sequence was codon-optimized using GeneArt Instant Designer (Thermo Fisher), to remove the ASO target motif and improve translation efficiency.Codon-optimized DNA was synthesized using Prime Gene Synthesis Services (Thermo Fisher).Codon-optimized Obox4 mRNA was transcribed in vitro using the mMESSAGE mMACHINE™ T7 Transcription Kit (catalog no.AM1344, Thermo Fisher) and then polyadenylated using a poly(A) tailing kit (catalog no.AM1350, Thermo Fisher), according to the manufacturer's instructions.The polyadenylated were first generated using deepTools 73 v3.

Figure 1 .
Figure 1.OBOX4 is expressed during ZGA.a) Mouse homeobox genes that are specifically expressed during ZGA.The genes were identified by means of statistically determined k-means clustering based on their expression in pre-implantation embryos.Dux is shown in a bold italic font.

Figure 2 .
Figure 2. Obox4 and Dux induce 2C-gene expression in mESCs.a) Diagram of the 2C::tdTomato reporter assay.ESCs bearing the tdTomato expression cassette under the control of the MERVL LTR promoter showed red fluorescence upon entering the 2C-like state.The expression of the transcription factor increased the 2C-like population, as detected using FACS.An EGFP expression plasmid was co-transfected with a gene of interest, to normalize the transfection efficiency.b) Boxplot showing normalized 2C-like cell percentage in 2C::tdTomato reporter ESCs overexpressing candidate pioneer factors.Dux and Obox4 potently induced a 2C-like state.c) Upper panel: schematic of Obox4-inducible cell line construction.Lower panel: western blot showing OBOX4 level upon induction by different concentrations of doxycycline.Expression of OBOX4 was carried out in a dosedependent manner.d) Left panel: volcano plot of DEGs in mESCs with Obox4 induction for 48 hours.Representative 2C-genes are labeled with gene symbols.Right panel: expression profile of genes up-regulated by Obox4 during embryogenesis.e) Left panel: volcano plot of differentially expressed transposable elements in mESCs with Obox4 induction for 48 hours.MERVL and MERVK elements were highlighted.Right panel: expression profile of MERVL and MERVK elements during pre-implantation embryogenesis.f) Heatmaps of the expression of 2C-genes in preimplantation embryos, naturally occurred 2C-like mESCs, and induced 2C-like mESCs.g) Venn diagram showing overlap of 2C-genes with genes induced by ectopic expression of Dux and Obox4 in mESCs.h) Scatterplot showing per-gene expression changes in Dux induced mESCs versus Obox4 induced mESCs.2C-genes are highlighted in red.Obox4 and Dux are labeled.

Figure 3 .
Figure 3. OBOX4 binds and activates 2C-specific LTR elements.a) Left panel: pie-chart displaying proportions of annotated genomic regions of the OBOX4 binding sites.Right panel: heatmap showing the OBOX4 binding site distribution near 2C-gene promoters.b) Left panel: the predicted OBOX4 binding motif using the top 500 CUT&RUN peaks.Right panel: histograph showing the distribution of predicted OBOX4 binding motif near 2C and random gene promoters.c) Venn diagram showing the distinct and overlap 2C-genes targeted by DUX and OBOX4.d) Representative genomic track showing DUX and OBOX4 binding sites at the Dppa2, Sp110, and Zscan4d loci and their expression levels in Obox4 and Dux induced mESCs.Read counts were CPM normalized.The OBOX4 binding sites overlapped with those of DUX.Dppa2, Sp110, and Zscan4d expression was upregulated upon Obox4 and Dux induction.

Figure 4 .
Figure 4. Concomitant loss of OBOX4 and DUX severely hinders ZGA.a) Schematic representation of somatic cell nuclear transfer (SCNT) experiment.The nuclei of knockout mESCs were transferred to enucleated oocytes to generate zygotes with knockout genotype.b) Upper panel: percent SCNT embryos developed to different stages at 4 dpt.Lower panel: p-value of nonparametric analysis of variance (ANOVA) among different genotypes and stages comparing to wildtype SCNT embryos.Three independent experiments were conducted, with 150-200 embryos per condition.c) Representative picture of SCNT embryos at 4 dpt.Morphologically abnormal blastocysts are highlighted.Blastocysts generated by double knockout mESCs were severely defective.d) Schematic representation of CRISPR-Cas9 mediated Dux and Obox4 knockout mouse production.In vitro fertilized mouse zygotes were electroporated with pre-assembled CRISPR-Cas9 complex targeting Dux and Obox4 loci.e) Bar plot showing genotype percentage of the pups delivered by Obox4 Het intercrosses.4 litters delivered 29 pups, litter size 7.25 ± 1.26.ns p-value = 0.9146, chi-square goodness of fit test.f) Representative photos of Obox4 KO and WT adult mice analyzed in e. g) Photo of Obox4 KO intercross litter with live pups.h) Bar plot showing genotype of Obox4 allele in the pups delivered by crossing of Dux KO /Obox4 Het × Dux KO /Obox4 Het or Dux KO /Obox4 Het × Dux Het /Obox4 Het .Nine litters delivered 54 pups, Dux heterozygous and knockout allele were present in 31 and 23 pups, respectively.** p-value = 0.001306; * p-value = 0.02218; chi-square goodness of fit test.i) Bar plot showing observed percentages of different preimplantation stage embryos bearing Obox4 KO allele with Dux heterozygous or knockout allele at 4.5 dpc.Among the total 94 embryos assessed, two degenerated, ten 2-cell arrest, two 4-8 cell arrest, four morula arrest embryos were observed at 4.5 dpc, whereas 76 developed to blastocyst.*** p-value = 3.564×10 -5 ; for 2-cell * p-value = 0.04348; for blastocyst * p-value = 0.02549; chi-square goodness of fit test.

Figure 5 . 1 a)
Figure 5. OBOX4 promotes ZGA in the absence of DUX.a) Schematic of the double knockdown rescue experiment.Male pronuclei of zygotes were injected with ASO targeting Obox4 and Dux transcripts as well as in vitro-transcribed codon-optimized Obox4 mRNA.b) Immunofluorescence staining of OBOX4 in early 2-cell embryos microinjected with scrambled ASO, double ASO (Obox4 and Dux), or double ASO with codon-optimized Obox4 mRNA.c) Representative picture of knockdown and rescue embryos at 4.5 dpc.Codon-optimized Obox4 mRNA injection rescued blastocyst formation in ASO knockdown embryos.d) The percentages of embryonic stages observed at 1.5 dpc, 2.5 dpc, 3.5 dpc, and 4.5 dpc.The plot represents the sum of three independent experiments, with 80-100 embryos per condition.e) Upper panel: volcano plot showing the results of DEG analysis of 2C-genes in knockdown and rescue embryos, as compared to that in scramble ASO-injected embryos.Standard deviations of log2(fold-change) were used to represent the degree of transcriptome dysregulation.Stdev, standard deviation.Lower panel: MA plot of differential expression of TEs in knockdown and recue embryos, as compared to that in scramble ASO-injected embryos.MERVK elements and MERVL are highlighted.f) Heatmaps of the expression of 2C-genes in preimplantation embryos, knockdown 2C-embryos, and rescue 2Cembryos.g) Rain plot displaying the expression change distribution of rescued 2C-genes in double knockdown and recue 2Cembryos.h) Bar plots showing the expression of representative 2C-genes in knockdown and recue embryos; n=3 biological replicates.

5 . 1
bamCoverage function (--binSize 10 --normalizeUsing CPM --smoothLength 30), and then normalized by subtracting the signal from non-immune IgG and wildtype mESCs using the deepTools bamCompare function (--scaleFactorsMethod None --operation subtract --binSize 10 --smoothLength 30).ChIPseeker 74 v1.28.3 was used to annotate the peaks.The MEME suite 75 v5.4.1 was used to identify the binding motifs.Heatmaps were generated using deepTools computeMatrix and plotHeatmap functions.Visualization of genomic tracks was performed using trackplot 76 v1.3.10.Raw ChIP-seq reads of DUXs in mESCs were downloaded from Hendrickson et al. (GSE85632).The published data were analyzed using the same method.Quantification and statistical analysisDescriptive and comparative statistics were employed in the manuscript as described in the figure legends with the number of replicates indicated.Significance is defined as a p-value less than 0.05 indicated with asterisk (* p-value < 0.05, ** p-value < 0.01, *** p-value < 0.001).Error bar represents the standard deviation (SD) of the mean of the replicates.
Tsukuba Institute and conducted in compliance with the Keio University Code of Research Ethics and the RIKEN's guiding principles.
6freshly harvested TBLCs.After 5 min of incubation at room temperature, 100 μL of chloroform (catalog no.03802606, Wako) was added to the cells and mixed by means of vigorous shaking.After 2 min of incubation at room temperature, the samples were centrifuged at 12,000 × g for 15 min at 4°C.The upper aqueous phase was then transferred to a new tube and mixed with 240 μL isopropanol (catalog no.15-2320, Merck), to precipitate the RNA.After 5 min of incubation at room temperature, the samples were centrifuged at 12,000 × g for 15 min at 4°C.The supernatant was then discarded, following which the RNA pellets were washed with 70% ethanol (catalog no.057-00456, Wako).The RNA pellets were then air-dried and dissolved in RNase-free water.The RNA solution was subjected to DNase treatment using TURBO DNA-free™ kit (catalog no.AM1907, Thermo Fisher), according to the manufacturer's instructions, to remove genomic DNA carryover.RNA (1 μg) was reverse transcribed to cDNA using the Transcriptor First Strand cDNA Synthesis Kit (catalog no.04379012001, Roche), according to the manufacturer's instructions.PrimeSTAR ® Max DNA Polymerase (catalog no.R045A, TaKaRa) and ProFlex™ PCR System (catalog no.4484073, Thermo Fisher) were used to amplify the sequence of interest from the cDNA.NEBuilder ® HiFi DNA Assembly Kit (catalog no.E2621L, NEB) was used to clone the sequence of interest into plasmid backbone.