OCT4/POU5F1 is indispensable for the lineage differentiation of the inner cell mass in bovine embryos

The mammalian blastocyst undergoes two lineage segregations, that is, formation of the trophectoderm and subsequently differentiation of the hypoblast (HB) from the inner cell mass, leaving the epiblast (EPI) as the remaining pluripotent lineage. To clarify the expression patterns of markers specific for these lineages in bovine embryos, we analyzed day 7, 9, and 12 blastocysts completely produced in vivo by staining for OCT4, NANOG, SOX2 (EPI), and GATA6, SOX17 (HB) and identified genes specific for these developmental stages in a global transcriptomics approach. To study the role of OCT4, we generated OCT4‐deficient (OCT4 KO) embryos via somatic cell nuclear transfer or in vitro fertilization. OCT4 KO embryos reached the expanded blastocyst stage by day 8 but lost NANOG and SOX17 expression, while SOX2 and GATA6 were unaffected. Blastocysts transferred to recipient cows from day 6 to 9 expanded, but the OCT4 KO phenotype was not rescued by the uterine environment. Exposure of OCT4 KO embryos to exogenous FGF4 or chimeric complementation with OCT4 intact embryos did not restore NANOG or SOX17 in OCT4‐deficient cells. Our data show that OCT4 is required cell autonomously for the maintenance of pluripotency of the EPI and differentiation of the HB in bovine embryos.


| INTRODUCTION
During preimplantation development, the mammalian embryo undergoes two consecutive lineage differentiations resulting in a blastocyst with three distinct lineages. The trophectoderm (TE) represents the first differentiated epithelium and envelopes the inner cell mass (ICM), which retains a pluripotent state. Subsequently, within the ICM, the primitive endoderm (PE) or hypoblast (HB) in human and bovine segregates from the epiblast (EPI), which contains the last pluripotent cells and gives rise to the embryo proper. The TE will contribute the embryonic portion of the placenta and the PE/HB develops into the yolk sac. 1,2 The fundamental mechanisms regulating these events have been studied extensively in the mouse, while advances in genome editing have enabled researchers to study the specific function of genes during preimplantation development in alternative model organisms. Given the substantial differences in the regulation of lineage differentiation and the maintenance of pluripotency between mouse and other mammalian species, this progress harbors the prospect of a deeper understanding of preimplantation development, also in human. Because in vitro embryo production techniques are highly advanced in bovine, this species offers great opportunities as a model for preimplantation development. [3][4][5] The second lineage differentiation, when the PE/HB and EPI segregate, is regulated by EPI precursor cells expressing FGF4, which via the MEK-pathway induces the differentiation of PE/HB precursor cells. Preimplantation embryos cultured with exogenous FGF4 develop an ICM entirely made up of PE/HB cells. 4 The transcription factor OCT4/POU5F1 plays a pivotal role in mammalian embryo development, as it regulates both the maintenance of pluripotency as well as differentiation events. 6 In mouse, loss of OCT4 prevents the development of the PE during the second lineage differentiation, while initial expression of the EPI marker NANOG is not affected. 7,8 On the contrary, the expression of NANOG fails in OCT4-deficient bovine blastocysts, while the early presumptive HB marker GATA6 is still present. Yet, it remains unclear if OCT4 has a role in the second lineage differentiation in bovine embryos, as GATA6 does not exclusively mark cells committed to the HB, but also cells in the TE. 9,10 Because data on the second lineage differentiation in bovine embryos are scarce, we first investigated expression patterns of lineage marker proteins and transcriptome dynamics of day 7, 9, and 12 embryos produced completely in vivo (thus representing bona fide samples of early bovine development). Studies of OCT4 knockout (KO) blastocysts generated by somatic cell nuclear transfer (SCNT) and zygote injection (ZI) showed that both EPI maintenance and HB differentiation are dependent on OCT4. Neither chimeric complementation with OCT4-intact blastomeres nor supplementation of exogenous FGF4 could rescue the OCT4 KO phenotype. Therefore, we conclude that-as in mouse-OCT4 is required cell autonomously during the differentiation of the HB in bovine blastocysts.

| Ethics statement
All animal procedures in this study were performed according to the German Animal Welfare Act and to a protocol approved by the Regierung von Oberbayern (reference number ROB-55.2-2532.Vet_02-20-73).

| Statistics
All data were analyzed with GraphPad Prism 5.04, mean values ± standard deviation (SD) are presented. Statistical tests were two-tailed unpaired t test for pairwise comparisons or one-way anova with Tukey multiple comparison test for analyses with more experimental groups. Level of significance was set to p < .05.

| Superstimulation of donors, transfer, and flushing of in vivo produced embryos
German Simmental heifers, 18-20 months old and 350-420 kg, served as embryo donors and recipients. Superstimulation and artificial insemination (AI) were performed as described previously 34 and the embryos were collected nonsurgically by flushing at day 7, 9, or 12 (day 0 = estrus) using a flushing catheter with an enlarged tipopening. For transfer of day 6 in vitro produced embryos to the uterus, the estrous cycle of recipient heifers was synchronized with a progesterone-releasing intravaginal device for 8 days (PRID-alpha, Ceva) and a single dose of PGF2α analog (500 µg cloprostenol, Estrumate, Essex) at removal of the PRID. After 48-72 h, the recipients showed signs of estrus. At day 6, embryos were transferred using a standard procedure 35 and collected at day 9 as described above.

| RNA-sequencing and data analysis
The generation of RNA-sequencing libraries, sequencing, and data analysis was performed as described previously. 9 Briefly, after isolation of RNA, cDNA and RNA sequencing libraries were generated using the Ovation RNA-Seq System V2 Kit (Tecan Genomics) and tagmentation technology of the Nextera XT kit (Illumina), respectively.
Libraries were sequenced on a HiSeq1500 machine (Illumina) and reads were mapped to the bovine reference genome ARS-UCD1.2 18 with STAR RNA sequence read mapper. 36 Differential gene expression analysis was performed with DeSeq2 37 ; heat map was generated from a mean-centered matrix using Heatmapper. 38

| In vitro fertilization and somatic cell nuclear transfer procedures
In vitro fertilization and SCNT were performed as described previously. 39 Presumptive zygotes and activated fused complexes from SCNT were cultured in synthetic oviductal fluid supplemented with 5% estrous cow serum, 2× of basal medium Eagle's amino acid solution 50× (Merck), and 1× of minimal essential medium nonessential amino acid solution 100× (Merck). For culture of embryos with exogenous FGF4, human recombinant FGF4 (R&D Systems) and heparin (Merck) were added at 1 µg/ml each. 25

| Immunofluorescence microscopy and image analysis
Before fixation, the zona pellucida (ZP) was removed enzymatically using pronase (Merck) 40 or mechanically for in vitro produced or flushed embryos, respectively. Embryos from in vitro culture were fixed in a solution containing 2% paraformaldehyde (PFA, Merck) for 20 min at 37°C 41 and flushed embryos were fixed in 4% PFA overnight at 4°C. After sequential blocking for each 1 h in 5% donkey and fetal calf serum (Jackson Immunoresearch) and 0.5% Triton X-100 (Merck), embryos were transferred to the first antibody solution and incubated overnight at 4°C. After washing, embryos were incubated in the second antibody solution for 1 h at 37°C and subsequently mounted in Vectashield mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI, Vector Laboratories) in a manner that conserves the 3D structure of the specimen. 42 The antibodies used and their dilutions are provided in Table S1. Stacks of optical sections were recorded with a Leica SP8 confocal microscope (Leica) at an interval of 1 µm using water immersion HC PL APO CS2 40X 1.1 NA or HC PL APO CS2 20X 0.75 NA objectives (Leica) and a pinhole of 0.9 airy units. DAPI, eGFP, Alexa Fluor 555, Rhodamine Red ™ -X, and Alexa Fluor 647 were excited with laser lines of 405, 488, 499, 573, and 653 nm, respectively, and detection ranges were set to 422-489 nm, 493-616 nm, 561-594 nm, 578-648 nm, and 660-789 nm, respectively. Cell numbers were counted manually using the manual counting plug-in of Icy bio-imaging software, 43 figures were produced using FigureJ software. 44

| Induction of OCT4 knockout in fibroblast cells and zygotes
For the transfection of female adult ear fibroblasts, sgRNA2b (5′-ACTCACCAAAGAGAACCCCC-3′) was cloned into pSpCas9(BB)-2A-Puro (PX459) V2.0, a gift from Feng Zhang (Addgene plasmid #62988), using the BbsI cutting site. 22 Transfection and clonal expansion after selection with 2 µg/ml puromycin for 48 h were performed as described previously. 45 The generation of OCT4 KO cells randomly tagged by eGFP was achieved by first transfecting somatic cells derived from a male fetus with a crown-rump length of 9 cm (FSC) with a linearized DNA construct and subsequently inducing OCT4 KO via lipofection using a ribonucleoprotein (RNP) containing the sgRNA2b. The linearized construct was produced by excising the CAG-eGFP-SV40pA sequence from a plasmid, generated by introducing a de novo synthesized CAG promoter and eGFP-SV40pA 46 into the pUC57-AmpR vector backbone. The RNP was produced by mixing the synthetic and modified sgRNA2b (Synthego) and TrueCut ™ Cas9 Protein v2 (Thermo Fisher Scientific) at equimolar concentrations of 8 µM in 10 mM Trisbuffer with 1 mM EDTA. Lipofection was performed in a six-well dish using CRISPRMAX ™ Cas9 Transfection Reagent (Thermo Fisher Scientific) according to the manufacturer's instructions. After 48 h of lipofection, eGFPpositive cells were sorted individually into 96-well dishes and clonally expanded as described above. Screening of single-cell clones for mutations at OCT4 and ETF1 was achieved by Sanger sequencing as described previously 9 with primers presented in Table S2. For ZI, RNPs with final concentrations of 2 µM sgRNA2b or sgRNA Ctrl (5′-GGTCTTCGAGAAGACCTGCG-3′) and 1 µM Cas9 in 10 mM Tris-buffer with 0.1 mM EDTA were produced as described above. After co-incubation of sperm and cumulus-oocyte complexes for 14 h, cumulus cells were removed by vortexing and approximately 10 pl of the RNP was injected into presumptive zygotes using a FemtoJet4i device (Eppendorf). After 7 days of culture, DNA was extracted by incubating the blastocysts in a buffer containing 25 mM MgCl 2 , 1 µl/ml TritonX-100, and 150 µg/ml Proteinase K (Carl Roth) at 37°C for 1 h and subsequently at 99°C for 8 min. For Sanger sequencing, a nested PCR amplification of the OCT4 locus was performed using 2 µl of the DNA extraction buffer directly as template. For the first PCR, we ran 25 cycles with Herculase II Fusion DNA Polymerase (Agilent) in a 25 µl reaction volume; the second PCR used 2 µl of the first reaction as template and HotStarTaq DNA Polymerase (Qiagen) in a 20 µl reaction volume for 15 cycles. All PCRs were performed using the buffers and instructions provided by the manufacturers; primer sequences are provided in Table S2. Extraction of DNA from fixed embryos after the imaging procedure was achieved by using the QIAamp DNA Micro Kit (Qiagen) according to the manufacturer's instructions regarding the isolation of genomic DNA from laser-microdissected tissues followed by 35 cycles of Herculase II PCR using 4 µl of template.

| Chimera aggregation
Embryos were produced via SCNT from OCT4 KO cells tagged with eGFP (OCT4 2bKOeGFP ) and FSC wild-type cells (NT Ctrl FSC ). At day 4, the ZP was removed enzymatically and each one morula from OCT4 2bKOeGFP and NT Ctrl FSC were aggregated to a chimera using phytohemagglutinin (Merck) and cultured as described previously. 40 Chimera formation was confirmed by time-lapse imaging (Primo Vision, Vitrolife) and by detection of both eGFP-positive and -negative cells in the developed blastocysts. Chimeric blastocysts were fixed and stained as described above.

| Lineage marker and transcriptome dynamics during the second lineage differentiation of in vivo produced embryos
To investigate the expression patterns of lineage markers of EPI (OCT4, NANOG, SOX2) and HB (SOX17, GATA6), we stained embryos flushed from the uterus after superstimulation as bona fide samples at days 7, 9, and 12 from n = 7 (day 7 and 9) and n = 3 (day 12) different donor cows. At day 7, OCT4 was present in the ICM ( Figure 1) and in the TE (Figure S1), and by day 9, OCT4 was restricted to EPI cells and the percentage of OCT4 cells strongly decreased. At all examined stages, NANOG was only present in EPI cells and their precursors, that is, coexpressed with OCT4 and SOX2 but mutually exclusive with GATA6 and SOX17. SOX2 was expressed pan-ICM at day 7 and restricted to EPI by day 9, resulting in a decreased percentage of SOX2-positive cells. Together with NANOG, SOX2 cells were still present in the embryonic disk at day 12 and it has been shown previously that OCT4 is present in this lineage until day 17. 11,12 A total of 28 day 7 blastocysts were stained against SOX17, where this transcription factor was not present in eight embryos and only faint and restricted to the ICM in the remainder. By day 9, the HB began to form an inner lining of the blastocoel cavity consisting of the visceral and parietal HB, 13,14 which were both marked by SOX17 until day 12. GATA6 was expressed at day 7 and 9 in the TE and ICM, but not co-expressed with NANOG. At day 12, there was no GATA6 visible in any of the lineages (Figure 1).
In a global transcriptomics approach, we aimed to identify genes that are specific to the developmental stages at day 7, 9, and 12 and the respective embryonic cell lineages EPI, HB, and TE. From three different donor cows, we analyzed three day 7 and each four day 9 and day 12 embryos. Differential gene expression analysis using DESeq2 revealed 1890 and 2716 differentially abundant transcripts (DATs, p adj < .05) in day 9 versus day 7 and in day 12 versus day 9 blastocysts, respectively. DATs were categorized into eight different groups according to their gene expression pattern over the course of time, that is, steadily increasing or decreasing; peaking at day 7, 9, or 12; and showing no difference between day 7 and 9 but increase or decrease at day 12 and vice versa. Identified DATs were compared to gene sets, which have been reported to be specific for EPI, PE/HB, and TE in mouse, human, and bovine embryos ( Figure 2, Data S1). Transcripts from EPI-specific genes were generally more prominent at day 7 and day 9 than at day 12; consistent with the proportion of OCT4-positive cells at day 7 and 9 ( Figure 1), the abundance of OCT4 transcripts steadily decreased until day 12. NANOG and SOX2 mRNAs showed similar abundances at day 7 and 9 but decreased by day 12. NODAL, a member of the pluripotency maintaining TGFβ/ACTIVIN/NODAL signaling pathway, 15 increased 80-fold from day 7 to day 9 and again decreased 2.8-fold by day 12. Interestingly, the mRNA levels for the NODAL antagonist LEFTY2 16 followed the same pattern, while the transcript levels of the NODALactivating convertase FURIN 13 steadily increased. The only EPI gene showing its highest transcript abundance at day 12 was FGFR1, which in pregastrulation stage human embryos were reported to be enriched in HB cells. 17 HBspecific transcripts mostly increased until day 12, except GATA6 and HDAC1. While the decreasing abundance of GATA6 mRNA was consistent with the observed pattern in the immunofluorescence staining, SOX17 mRNA was not differentially abundant between day 7 and day 9 but increased later at day 12. CDX2 mRNA, encoding an early marker for TE, was not differentially abundant between day 7 and day 9 but increased 1.5-fold until day 12. Except group 2 (Figure 2), TE genes were present in every expression pattern, indicating that this lineage undergoes dynamic changes during the observed period.
A previously published global transcriptomic dataset covering in vitro cultured day 7 embryos from in vitro fertilization (IVP Ctrl) and SCNT with wild-type cells (NT Ctrl) or cells carrying an OCT4 KO mutation (OCT4KO tm1 ) 9 was reanalyzed using the current genome assembly ARS-UCD1.2 18 and compared to the transcriptome profile of in vivo produced day 7 embryos. By comparing the DATs of the three above-mentioned groups against in vivo produced day 7 embryos, we identified transcripts that were differentially abundant due to the SCNT procedure or in vitro culture. Five lineage-specific DATs appeared in all three groups and are therefore attributable to in vitro culture, causing reduced abundance of HAND1 (TE) and HDAC1 (HB) transcripts while mRNA levels of HSD17B11 (EPI), HMGCS1, and SLC2A3 (TE) were increased. Two DATs were specific to the SCNT procedure with increased levels of CLDN7 (TE) transcripts and a lower mRNA abundance of MAP2K6 (HB) ( Figure S2A). The remaining lineagespecific DATs in OCT4KO tm1 against in vivo produced day 7 blastocysts included six, two, and three with decreased and one, five, and eight with increased levels from the EPI, HB, and TE lineages, respectively ( Figure S2B), showing a shift of gene expression toward the differentiated lineages TE and HB in the absence of OCT4.

| Induction of OCT4 knockout without targeting a known OCT4 pseudogene
Earlier studies on the function of OCT4 in bovine embryos 9,19 used a sgRNA-sequence, which also targets an OCT4 pseudogene present in intron 1 of ETF1. 20 Therefore, we adapted an sgRNA (sgRNA2b) known to be highly F I G U R E 1 The second lineage differentiation in embryos produced in vivo. (A) Representative confocal planes of day 7, 9, and 12 embryos stained for NANOG/SOX17 (n = 10, 5, and 5), NANOG/GATA6 (n = 10, 6, and 3), SOX2/SOX17 (n = 8, 6, and 4), and OCT4/SOX17 (n = 10 and 5). All scale bars represent 100 µm. Note that the confocal plane shown for OCT4/SOX17 in the day 7 blastocyst was selected to show an SOX17-positive cell in the inner cell mass. While OCT4 in the trophectoderm is not clearly visible in this particular plane, it was consistently present in this lineage ( Figure S1). (B) Total cell numbers and proportion of cells stained positive for lineage-specific markers at day 7 (D7) and day 9 (D9) relative to the total cell number. Data are presented as mean ± SD, asterisks indicate significant differences between D7 and D9 (two-tailed t test, *p < .05; ****p < .0001).  21 to the bovine ortholog sequence, where it spans an exon-intron junction at the 3′-end of exon 2 and thus does not target the pseudogene in ETF1, because the retrocopy does not contain intronic OCT4 sequences. 20 The sgRNA2b sequence was cloned into PX459 V2.0 to knock out OCT4 in somatic cells and single-cell clones were produced after selection with puromycin. 22 From 31 single-cell clones, three retained the wild-type sequence while 11 carried homozygous mutations that were confirmed by a single-nucleotide polymorphism 179 bp downstream the sgRNA2b cutting site. The remaining single-cell clones had bi-allelic heterozygous (n = 13) or mono-allelic (n = 4) mutations. None of the single-cell clones had any mutation at the OCT4 pseudogene, showing that sgRNA2b specifically targets OCT4. From the same transfection experiment, two single-cell clones with the same homozygous deletion of two base pairs (OCT4 2bKOX1 and OCT4 2bKOX4 ) and one where no mutation had occurred (NT Ctrl 2b ) were used to reconstruct embryos via SCNT. Embryos from OCT4 2bKOX1 developed to blastocysts by day 7, albeit at a much lower rate than NT Ctrl 2b embryos, while there was no difference between OCT4 2bKOX4 and NT Ctrl 2b (Table 1A). NT Ctrl 2b showed expression of OCT4 in both TE and ICM (n = 4), while blastocysts from OCT4 2bKOX1 (n = 5) and OCT4 2bKOX4 (n = 6) stained negative ( Figure S3). By day 8, we observed that NT Ctrl 2b embryos had expanded and started hatching through the incision in the ZP made during the SCNT procedure, while OCT4 2bKOX1 and OCT4 2bKOX4 were not able to exit the ZP and expanded to a lesser extent compared to NT Ctrl 2b embryos.

| SOX17 is lost in blastocysts lacking OCT4
To elucidate the effects of loss of OCT4 during the second lineage differentiation, we performed immunofluorescent staining of the lineage-specific markers GATA6, SOX17, NANOG, and SOX2 23,24 at day 8 blastocyst stage. In IVP Ctrl and NT Ctrl 2b embryos, we confirmed that at day 8, expression of SOX2 is restricted to the ICM and that NANOG and SOX17 are mutually exclusive markers of the EPI and HB, respectively. GATA6 is expressed in both ICM and TE, and F I G U R E 2 Differentially abundant transcripts (DATs) of in vivo produced blastocysts between day 7, 9, and 12 categorized into gene sets specific for epiblast (EPI), hypoblast (HB), and trophectoderm (TE). Three day 7 blastocysts and each four day 9 and 12 blastocysts were analyzed using DESeq2 (p adj < .05) GATA6-negative cells are present in the ICM. In contrast to the expression pattern in day 9 in vivo produced embryos, SOX17 and GATA6 are always co-expressed with SOX2, indicating that SOX2 is a late EPI marker (Figure 3, Figures  S4 and S5). In OCT4 2bKOX1 day 8 SCNT blastocysts, there were no GATA6-negative cells and SOX2 was expressed exclusively in the ICM ( Figure 3). As reported previously, 9 there was no expression of NANOG and we did not detect any SOX17-positive cells (Figures S4 and S5).
To validate our findings from SCNT experiments, we induced KO of OCT4 directly in zygotes from in vitro fertilization (IVF) by injection of a ribonucleoprotein (RNP) consisting of Cas9 protein and synthesized sgRNA2b. As control, we used an RNP with no target in the bovine genome (sgRNA Ctrl). Developmental data from 11 experiments with a total of 1224, 462, and 543 zygotes injected with OCT4 2bZI , sgRNA Ctrl, or non-injected, respectively, revealed that the injection procedure induced a decreased cleavage rate, but did not affect the percentage of blastocysts developed from cleaved zygotes ( Table 2). To determine the mutation rate after injection, DNA was isolated from individual embryos and the targeted site was amplified for subsequent Sanger sequencing. From four experiments, we analyzed putative mutations in a total of 57 day 7 blastocysts, of which 34 had expanded. There were no significant differences in percentage of wild-type, biallelic, homozygous, or monoallelic mutations between expanded and early day 7 blastocysts ( Figure 4D) and four expanded blastocysts carried homozygous mutations that induced a shift of the reading frame. Staining with antibodies against NANOG, SOX17, and OCT4 in OCT4 2bZI day 8 blastocysts in combination with genotyping after the imaging procedure enabled us to confirm the absence of OCT4 on the protein level and frame shift mutation on  could not be detected (n = 11, Figure 4A-C, Video S1 and S2). Developmental data from OCT4 KO embryos produced by both SCNT and IVF show that OCT4 is not essential for the formation of an expanded blastocyst by day 7. Expansion was present in SCNT embryos-although less pronounced-as well as in IVF embryos that carried biallelic OCT4 frameshift mutations. Yet, a decreased blastocyst rate in OCT4 2bKOX1 SCNT embryos and a higher percentage of expanded blastocysts in OCT4-intact embryos demonstrate that loss of OCT4 impedes development to the expanded blastocyst stage.

| Uterine environment cannot rescue the second lineage differentiation in OCT4 KO embryos
To evaluate if the phenotype of OCT4 KO embryos as described above is alleviated or rescued when the second lineage differentiation occurs in utero, we transferred each four day 6 OCT4 2bKOX4 early blastocysts to five synchronized heifers and collected the embryos at day 9. As controls, we used each four IVP Ctrl blastocysts transferred to two recipients. We collected three day 9 OCT4 2bKOX4 expanded blastocysts from three different recipients and five IVP Ctrl expanded blastocysts. The transferred OCT4 2bKOX4 and IVP Ctrl embryos showed total cell numbers similar to in vivo produced day 9 blastocysts ( Figure 1B) with 139.3 ± 20.5, 155.6 ± 50.28, and 161.5 ± 39.1 cells, respectively (mean ± SD, p > .05).
Staining of NANOG and SOX17 revealed a similar expression pattern in IVF blastocysts compared to embryos completely developed in vivo. HB precursor cells began to form an inner lining within the blastocoel, which was confirmed by a similar proportion of SOX17-positive cells (15.1 ± 5.8 vs. 18.9 ± 4.8, mean [%] ± SD, p > .05), while the proportion of NANOG-positive cells was markedly reduced in the IVF embryos (6.9 ± 3.8 vs. 16.5 ± 6.9, mean [%] ± SD, p < .05). All collected OCT4 2bKOX4 blastocysts stained negative for NANOG and SOX17. Although we were not able to recover the majority of OCT4 2bKOX4 blastocysts at day 9 (3/20), our data show that, in the absence of OCT4, bovine embryos survive until day 9 and expand in utero, but the second lineage differentiation cannot be rescued by a uterine environment ( Figure 5).

| OCT4 is required cell autonomously during the second lineage differentiation
We performed a chimera aggregation experiment in order to investigate, if OCT4 is required cell-autonomously for the expression of NANOG and SOX17. Using fetal somatic cells (FSCs), we produced a single-cell clone, in which an eGFP expression vector was randomly integrated and OCT4 was knocked out by homozygous deletion of two nucleotides in frame (OCT4 2bKOeGFP ). Embryos from OCT4 2bKOeGFP developed to expanded day 8 blastocysts, ubiquitously expressed eGFP, and lacked expression of OCT4 (n = 7), NANOG (n = 8), and SOX17 (n = 7). As aggregation partner, we used embryos generated from wildtype FSCs (NT Ctrl FSC ), which at day 8 expressed SOX17 and NANOG as expected (n = 3, Figure 6, Figure S6).
In three experiments, we aggregated 25 chimeras and 12 showed contribution of both OCT4 2bKOeGFP and NT Ctrl FSC cells to the blastocyst. In none of these chimeras, we detected co-expression of eGFP with NANOG or SOX17 ( Figure 6). Therefore, we conclude that OCT4 expression in neighboring cells within the ICM cannot rescue NANOG or SOX17 expression.
F I G U R E 5 Day 9 expanded blastocysts collected from recipient heifers stained against NANOG and SOX17. Representative confocal planes of day 9 blastocysts transferred to a recipient at day 6 and collected at day 9 from in vitro fertilization (IVF) or produced by somatic cell nuclear transfer using OCT4 2bKOX4 cells. Scale bars represent 100 µm To further elucidate the role of OCT4 in the differentiation of the HB, we incubated OCT4 2bKOX4 and NT Ctrl 2b with exogenous FGF4, which induces pan-ICM expression of HBmarkers and ablates the expression of NANOG in wild-type embryos. 25 NT Ctrl 2b day 8 blastocysts showed full expression of SOX17 and no NANOG (n = 10), while in OCT4 2bKOX4 , 10 of 16 blastocysts showed no expression of NANOG or SOX17 ( Figure 7A). In two blastocysts, we found ectopic SOX17 expression in the TE and four blastocysts had positive cells in the ICM, albeit at a significantly lower proportion to the total cell number as FGF4 treated NT Ctrl 2b blastocysts (5 ± 2.4 vs. 16.3 ± 7.6, mean [%] ± SD, p < .05) and with a lower intensity of the fluorescent signal ( Figure S7). Pairwise comparisons of the total cell number revealed a significant reduction due to loss of OCT4, which was alleviated by exogenous FGF4, while in NT Ctrl 2b blastocysts, FGF4 had a detrimental effect on the total cell number ( Figure 7B). Because neither chimeric complementation nor treatment with exogenous FGF4 could restore a failing differentiation of the HB in cells without functional OCT4, we conclude that OCT4 is required cell autonomously to induce HB formation.

| DISCUSSION
In this study, we set out to further elucidate the role of OCT4 in bovine preimplantation development, especially during the second lineage differentiation. Because it has not been entirely clear how the expression of known markers of the different lineages evolves during the establishment of the HB lineage, we examined the patterns in embryos produced in vivo at the blastocyst stage (day 7), the expanded blastocyst stage (day 9), and the ovoid blastocyst stage (day 12). Complete HB migration by day 11 has been documented before by staining of SOX17. 26 We were able to show, that said migration begins at day 9 with an increase of SOX17 cells, which organize into the visceral HB and end the salt and pepper distribution of HB-and EPI-precursor cells within the ICM. Interestingly, despite an increase in SOX17 cell numbers between day 7 and day 9, we did not detect differences in SOX17 transcript abundance. At day 7, we found a subset of embryos where SOX17 was not present yet, while in embryos that expressed the marker, intensity was low, mutually exclusive with NANOG and restricted to the ICM. Other studies using in vitro produced embryos report co-expression of SOX17 with NANOG as early as the 16-32 cell stage 10 and ectopic expression in the TE until day 6.5. 27 In contrast to various reports using in vitro produced embryos, 9,10,25 we did not find co-expression of GATA6 and NANOG in day 7 in vivo produced embryos, indicating an earlier commitment to the HB or EPI lineage by reciprocal repression. The pluripotency factors OCT4 and SOX2 are restricted to the EPI by day 9, while at day 7, they are expressed throughout the blastocyst or the ICM, respectively, confirming SOX2 as a reliable marker for the ICM at day 7. 24 While OCT4 is not clearly visible in the TE of the day 7 blastocyst shown in Figure 1 (due to the selection of a confocal plain that provides optimal representation of SOX17), it was consistently present in this lineage in both day 7 in vivo produced ( Figure S1) and SCNT blastocysts ( Figure S3). The expression patterns described here may serve as a benchmark for assessing the quality of bovine embryos from long-term culture systems. 26 In vitro produced controls, that were transferred to recipients at day 6 and flushed at day 9, displayed the same total cell number and SOX17/NANOG expression pattern as embryos produced in vivo, showing that short-term incubation in vivo is sufficient for stage-adequate development of the embryo, as reported previously. 28 Consistent with the staining pattern and in line with a previous report, we observed a steady reduction of OCT4 transcripts from day 7 to day 12, while the abundance of NANOG and SOX2 transcripts did not change between day 7 and day 9 and eventually decreased by day 12. 29 Similar to human pregastrulation development, we found a decreasing abundance of transcripts associated with naïve pluripotency (KLF4, KLF17, PRDM8, TFCP2L1, ZFP42, UTF1), while markers for primed pluripotency, that increased in human (FGF2, DNMT3B, SOX11, SFRP2, SALL2), 17 did not change from day 7 until day 12. Van Leeuwen et al 13 detected NODAL transcripts in the EPI of Rauber's layer (RL) stage (day 10-11) embryos and suggested NODAL activation through the convertase FURIN, which they detected in the RL. We found a massive increase (80-fold) in NODAL transcripts between day 7 and 9 together with an increase of FURIN and LEFTY2 mRNAs, indicating that the NODAL/BMP/WNT pathway, that later regulates patterning, 30 is already activated by day 9. Comparing transcriptome profiles from day 7, 9, and 12 in vivo produced embryos enabled analysis of the dynamics of lineage-specific expression patterns during bovine development. Nevertheless, a greater sample size is still required to ultimately identify the stage-specific expression profiles at these respective stages.
Studies on the effects of in vitro culture on the transcriptome of bovine blastocysts have mainly identified pathways related to "energetic metabolism, extracellular matrix remodeling, and inflammatory signaling". 31 While we found a total of 463 DATs between in vivo and in vitro produced day 7 embryos, only five DATs were lineage specific, indicating that the in vitro culture system has no substantial effect on the mechanisms of lineage differentiation. The fact that we only found two lineage-specific DATs between NT Ctrl and in vivo produced embryos strengthens the use of embryos from SCNT to study the basic mechanisms of early lineage differentiation.
By generating OCT4 KO embryos with an sgRNA that exclusively targets OCT4 using both SCNT and ZI, we aimed to resolve existing conflicts regarding the OCT4 KO phenotype in bovine embryos. We confirmed that, regardless of the applied method, OCT4 is not essential for the expansion of the blastocyst and showed that OCT4 KO embryos survive until day 9 when transferred to a recipient cow. Daigneault et al 19 reported that targeting OCT4 using ZI prevented development to the expanded blastocyst stage, while a previous report from our laboratory showed an unchanged morphology of OCT4 KO blastocysts. 9 These two studies used the same sgRNA sequence, which also targets the pseudogene present in ETF1; therefore, it is unlikely that off-target effects caused the divergent phenotypes. Here, we also applied ZI to delete OCT4 and observed expansion of the blastocysts; therefore, we can exclude the effects of the different procedures used to knock out OCT4. We can only speculate that the conflicting results are caused by variables in the ZI procedure or the in vitro culture system.
As reported previously, 9 loss of NANOG was observed in all embryos without functional OCT4, but the pluripotency marker SOX2 was independent of OCT4, as reported previously 19 and similar to mouse embryos. 7 Although there was no reduction in expression of the early HB marker GATA6, expression of SOX17 failed in the absence of OCT4. As expression of SOX17 is not dependent on NANOG, 32 loss of SOX17 can be linked directly to the OCT4 KO phenotype. In mouse embryos, Oct4 KO prevents the differentiation of the PE, not only because FGF4-MEK signaling is reduced 33 but also because OCT4 is required cell autonomously. 7,8 Nevertheless, Frum et al 7 reported a modest induction of PE gene expression upon treatment with FGF4 in Oct4 KO mouse embryos, indicating that an alternative OCT4-independent pathway exists. We observed faint expression of SOX17 in a subset of OCT4 KO embryos treated with FGF4, which might be explained by such a pathway, but full expression of SOX17 was only possible when OCT4 was present. Using chimeric complementation, we were able to show that paracrine factors produced by OCT4 intact cells are not sufficient to restore the OCT4 deficiency. Therefore, and similar to mouse, the development of the HB requires OCT4 not only to produce paracrine factors, for example, FGF4, but also for the induction of differentiation in HB precursor cells, that is, OCT4 is required cell autonomously.
In conclusion, our data show that in the bovine preimplantation embryo OCT4 is required during the second lineage differentiation for the maintenance of pluripotency in the EPI and differentiation of the HB.