Transcriptome dynamics in early in vivo developing and in vitro produced porcine embryos

Background The transcriptional changes around the time of embryonic genome activation in pre-implantation embryos indicate that this process is highly dynamic. In vitro produced porcine blastocysts are known to be less competent than in vivo developed blastocysts. To understand the conditions that compromise developmental competence of in vitro embryos, it is crucial to evaluate the transcriptional profile of porcine embryos during pre-implantation stages. In this study, we investigated the transcriptome dynamics in in vivo developed and in vitro produced 4-cell embryos, morulae and hatched blastocysts. Results In vivo developed and in vitro produced embryos displayed largely similar transcriptome profiles during development. Enriched canonical pathways from the 4-cell to the morula transition that were shared between in vivo developed and in vitro produced embryos included oxidative phosphorylation and EIF2 signaling. The shared canonical pathways from the morula to the hatched blastocyst transition were 14–3-3-mediated signaling, xenobiotic metabolism general signaling pathway, and NRF2-mediated oxidative stress response. The in vivo developed and in vitro produced hatched blastocysts further were compared to identify molecular signaling pathways indicative of lower developmental competence of in vitro produced hatched blastocysts. A higher metabolic rate and expression of the arginine transporter SLC7A1 were found in in vitro produced hatched blastocysts. Conclusions Our findings suggest that embryos with compromised developmental potential are arrested at an early stage of development, while embryos developing to the hatched blastocyst stage display largely similar transcriptome profiles, irrespective of the embryo source. The hatched blastocysts derived from the in vitro fertilization-pipeline showed an enrichment in molecular signaling pathways associated with lower developmental competence, compared to the in vivo developed embryos. Supplementary Information The online version contains supplementary material available at 10.1186/s12864-021-07430-7.

important to understand which molecular pathways are affected by the in vitro embryo production pipelines. In vivo, the embryo starts to rapidly elongate by day 11 and secretes estradiol-17β (E2) as primary recognition of pregnancy signal (6). The secretion of embryonic E2 coincides with the endometrial expression of E2-regulated genes (7). The transition of the hatched blastocyst to an elongated embryo takes place rapidly (8).
A dynamic and embryonic developmental stage-speci c mRNA expression has been shown in various species (9,10). Single-cell RNA sequencing of murine and bovine embryos revealed a transcriptional variation of single blastomeres (10,11). Single murine blastomeres showed an increasing transcriptional variation with developmental progression (10). Similar ndings have been reported for stem cell differentiation. Stem cells had a more uniform transcriptome pro le compared to differentiated cells (12).
The single cell reconstruction of murine preimplantation development showed distinct developmental stage-dependent clusters, i.e., 2-cell, 4-cell, 8-cell and 16-cell stage embryos, while single cells from the early, mid and late blastocyst clustered together (10). In pigs, the transcriptional changes of embryos around the time of EGA (2-and 4-cell stage embryos) have been investigated in both in vivo developed and in vitro produced whole embryos, aiming at gaining insights into the mechanisms that lead to reduced developmental potential of in vitro produced embryos (13). In vitro produced embryos displayed altered transcript levels for apoptotic factors, cell cycle regulation factors and spindle components, as well as transcription factors, collectively contributing to reduced developmental competence of in vitro produced embryos (13). To understand the species-speci c regulatory networks involved in EGA, the rst lineage commitment and the primitive endoderm differentiation, Cao (14). By comparing the transcriptome changes with those of mouse and human pre-implantation embryos, a unique pattern was found in pig embryos (14). In addition, the global gene expression pattern was different in somatic cell nuclear transfer (SCNT) embryos compared to in vivo developed embryos (14). The pig EGA was con rmed to take place at the 4-cell stage, while this only appeared at the 8-cell stage in SCNT embryos (14). The differentially expressed genes from the hatched blastocyst to tubular and lamentous embryos included glycolytic enzymes that are potentially regulated by estrogen (15,16).
To date, the developmental competence, as well as pregnancy rates after transferring in vitro produced porcine embryos remain low (17). This can, in parts, be attributed to aberrant chromatin dynamics (18).
Compared to in vivo produced embryos, in vitro produced embryos showed developmental stagedependent altered chromatin dynamics. Already at the two-cell stage, they displayed aberrant chromatinnuclear envelope interactions (18). In vitro produced embryos showed global chromatin remodeling imperfections and failed to establish a proper rst lineage segregation at the blastocyst stage (18). To improve the developmental competence of in vitro embryos, it is crucial to elucidate their transcriptional pro le during pre-implantation development. In this study, we aimed at furthering the understanding of early embryo development, and to identify molecular pathways that could explain lower developmental competence of in vitro produced hatched blastocysts.

Results
Samples and RNA sequencing RNA sequencing was performed using 50 single embryos ( Figure 1). Figure 1. Experimental set-up for single embryo RNA-sequencing. The arrows indicate the between group analyses.
A total of 1,405 million raw reads was obtained after RNA sequencing, with a duplication rate of 63 ± 7% (mean ± SD) and a GC content of 45 ± 1% (mean ± SD). The mapping rate after quality ltering was 84 ± 6% (mean ± SD). The number of detected transcripts increased with developmental progression for the in vivo produced embryos, while it decreased for the in vitro produced embryos (Additional le 1). The low number of detected transcripts for the 4-cell in vivo embryos might be the consequence of analyzing early 4-cell embryos shortly after EGA combined with a relatively low input and cDNA yield during library preparation (Additional le 2). Given the uncertainty of the embryonic developmental stages of in vivo embryos, different RNA quality, and read alignment at the 4-cell, as well as at the morula stage (Additional le 2 and 3), the in vivo developed and in vitro produced embryos were analyzed separately and not compared to each other. To identify in vitro fertilization pipeline-induced transcriptome differences, the hatched blastocysts were used for an in vivo developed versus in vitro produced comparison.

Developmental transcriptome dynamics
To provide a developmental stage-speci c overview, global developmental transcriptome dynamics were investigated. Principal component analyses (PCA) were performed separately for the in vivo developed and in vitro produced embryos and showed a clear developmental stage-speci c clustering of the embryos (Figure 2A and B). For the in vivo developed embryos, PC1 and PC2 explained 77.8% and 11.4% of the variance in transcript levels. For the in vitro produced embryos, PC1 and PC2 explained 71.8% and 17.3% of the variance. The in vivo 4-cell embryos displayed a larger degree of transcriptional heterogeneity than the in vitro 4-cell embryos. The morulae and hatched blastocysts were sexed based on the expression of Y-chromosome speci c transcripts. At the morula stage, male and female embryos clustered together, yet the clusters were not fully overlapping. At the blastocyst stages, the male and female clusters were fully overlapping. In vivo and in vitro embryonic developmental dynamics The developmental transcriptome dynamics were further analyzed by identifying differentially expresses genes (DEGs) between the 4-cell and morula stage, and the morula and hatched blastocyst stage for both the in vivo developed and in vitro produced embryos. The number of DEGs was higher between the 4-cell to morula stage, than for the morula to hatched blastocyst stage ( Figure 3). For the in vivo embryos, 10,089 and 2,347 DEGs were identi ed between the 4-cell to the morula stage and the morula stage to the hatched blastocyst stage, respectively ( Figure 3A). For the in vitro embryos, 8,152 and 4,023 DEGs were identi ed between the 4-cell to the morula stage and the morula stage to the hatched blastocyst stage, respectively ( Figure 3B). The developmental dynamics were assessed with a self-organizing tree algorithm ( Figure 4A

Biological functions of embryonic developmental dynamics
To gain insight into the biological functions of the DEGs, a canonical pathway enrichment analysis was conducted ( Figure 5). In both the in vivo and the in vitro produced 4-cell to morula stage embryos, there was a signi cant enrichment of oxidative phosphorylation, tRNA charging and EIF2 signaling. From the morula to the hatched blastocyst stage, the DEGs in the pathways 14-3-3-mediated signaling, signaling of Rho Family GTPases, and NRF2-mediated oxidative stress response were all higher expressed at the hatched blastocyst stage for both the in vivo and in vitro produced embryos. The ERK/MAPK signaling pathway was signi cantly enriched in vivo at the 4-cell to morula transition, and was predicted to result in a lower rate of transcription at the 4-cell stage (Additional le 4). In the in vivo embryos, the TNFR1 signaling predicted a lower degree of apoptosis and cell survival at the 4-cell stage compared to morulae stage (Additional le 4). The in vivo hatched blastocysts displayed a signi cant enrichment of estrogen biosynthesis compared to the morulae stage (Additional le 4).

In vivo and in vitro differences at the hatched blastocyst stage
The in vivo and in vitro hatched blastocysts were compared, as the embryos displayed similar cDNA pro les, library smears and alignment coverages for the most abundant transcripts at this developmental stage (Additional le 2 and 3). Embryos at this stage of development are thought to be more alike than at earlier stages, as time differences related to fertilization at earlier stages contribute more substantially to the actual developmental stage.
At the hatched blastocyst stage, the selection of developmentally competent embryos has already taken place. Yet, we unraveled in vitro fertilization pipeline-induced sex-speci c differences. The in vivo developed female and male hatched blastocysts clustered largely together ( Figure 6A). They were separated from the in vitro hatched blastocyst in a sex-speci c manner by principal component 1. While 33 DEGs were identi ed between the female in vivo and in vitro produced embryos, 241 DEGs were identi ed between the male in vivo and in vitro produced embryos. Figure 6B displays the difference between in vivo developed and in vitro produced embryos in a sex-independent manner. There were no DEGs when comparing male and female embryos for either in vivo developed or in vitro produced embryos. By comparing the female in vivo developed versus in vitro produced embryos, the DEGs inositol polyphosphate multikinase (IPMK) and Rac family small GTPase 1 (RAC1) were speci c to this comparison. The other 31 DEGs were also discovered by comparing the in vivo and in vitro male hatched blastocysts. These genes were involved in amino acids transport, synthesis and metabolism, and similarly expressed in both female and male embryos ( Figure 6C). Irrespective of the embryos' sex, the in vivo embryos had a lower expression of genes involved in amino acid transport, synthesis and metabolism compared to the in vitro embryos. When disregarding the sex of the embryos and emphasizing on the embryo source, the persistent difference between in vivo developed and in vitro produced embryos at the hatched blastocyst stage were illustrated by an enrichment of ve canonical pathways ( Figure 6D). Except for a higher expression in in vivo versus in vitro hatched blastocysts of DEGs involved in cyclins and cell cycle regulation, the DEGs involved in tRNA charging, cell cycle: G1/S checkpoint regulation, PEDF signaling, and neuro-in ammation signaling pathway were higher expressed in in vitro than in in vivo hatched blastocysts. The PEDF signaling pathway was inhibited in in vivo hatched blastocysts compared to the in vitro hatched blastocysts and was predicted to result in a lower rate of apoptosis in in vivo hatched blastocysts (Additional le 4). in amino acid metabolism, and D. Signi cantly enriched canonical pathways between the in vivo developed and in vitro produced hatched blastocysts. Red (-) dots represent canonical pathways of which genes were signi cantly less expressed in the in vivo embryos, while blue (+) represent canonical pathways of which genes were signi cantly higher expressed in the in vivo embryos. The GeneRatio indicates the proportion of DEGs that were identi ed in an enriched canonical pathway.

Discussion
Transcriptome dynamics during early embryo development Early developing porcine embryos displayed a great adaptive capacity towards their environment, evidenced by largely similar transcriptome dynamics observed in both in vivo developed and in vitro produced embryos. In vitro produced embryos offer the opportunity to study molecular pathways of interest in a developmental-stage speci c manner, as there is a higher degree of certainty regarding the time of fertilization compared to in vivo developed embryos. However, developmental rates and embryo competence of in vitro produced embryos are still lower compared to their in vivo developed counterparts (5). A number of factors are known to contribute to embryo development. The presence of cumulus cells during maturation facilitates full oocyte maturation (19). In pigs, the presence of cumulus cells during oocyte maturation is essential for oocyte maturation, fertilization and subsequent embryo development (20). The discrepancy in embryo development between in vivo developed and in vitro produced embryos at early post-fertilization developmental stages might be explained by the use of a pool of non-selected oocytes of overall lower competence for in vitro maturation, compared to those selected for ovulation, and the effects of in vitro maturation on oocyte quality. A higher blastocyst rate has previously been shown after oocyte maturation under a 20% oxygen atmosphere (21). However, blastocyst quality assessed by the expression of genes related to metabolism (GLUT1 and LDHA), antioxidant response (SOD2 and GPX1), growth factors and apoptosis (IGF2R, BCL2 and BAX), methylation (DNMT3B), and blastocyst quality (AKR1B1, POU5F1 and CDX2) were not affected (21). In addition, the blastocyst rates of in vivo and in vitro matured rabbit oocytes did not signi cantly differ, while at earlier developmental stages, the in vivo embryo development rates were signi cantly higher than observed for embryos produced with in vitro matured oocytes (22). Thus, while oocyte quality and competence, and subsequent embryo development are affected by the maturation conditions, only minor transcriptional differences have been reported at the hatched blastocyst stage (23). In line with previous ndings, we found more similar transcriptome pro les at later developmental stages. At the hatched blastocyst stage, only limited transcriptional differences persisted. Additionally, the developmental-stage speci c differences were more pronounced than the sex-speci c differences, as previously described by Zeng et al. (2019), studying the transcriptome dynamics in in vivo developed day 8, 10, and 12 porcine embryos (16).

Early porcine embryo development
The early embryo development was studied at the 4-cell, morula and hatched blastocyst stage for both in vivo developed and in vitro produced embryos. Previously, porcine embryos after EGA have been shown to display an increased abundance of transcripts involved in, among others, transcription (13). In our in vivo developed embryos, the ERK/MAPK signaling pathway was signi cantly enriched during the 4-cell to morula transition, which is predicted to result in a lower rate of transcription at the 4-cell stage. This is in line with the increased number of detectable transcripts at the morula compared to the 4-cell stage, and indicates that the in vivo embryos might have been sampled at an earlier 4-cell stage, i.e., closer to EGA. In addition, the TNFR1 signaling pathway was enriched in the in vivo 4-cell to morula transition. A lower degree of apoptosis and cell survival was predicted at the 4-cell stage compared to in vivo morulae stage. An inverse pattern was observed in in vitro produced embryos, indicating a higher degree of apoptosis in early in vitro embryos. This is congruent with previous ndings that in vitro embryos around EGA have an increased cytoplasmic content of apoptotic factors and the low developmental rates observed for in vitro embryo production (13). This could indicate that in vitro produced embryos with compromised developmental competence are arrested at an early stage of development.
Both the in vivo developed and in vitro produced 4-cell to morula transition was characterized by an enrichment of oxidative phosphorylation, tRNA charging and EIF2 signaling. An increase in oxidative phosphorylation with developmental progression has previously been reported for mouse embryos (24). Oxidative phosphorylation accounted for 60-70% of consumed oxygen in blastocysts, compared to 30% of consumed oxygen in cleavage stage embryos (24). In addition, oxygen consumption of in vivo bovine blastocysts increased with increasing morphological quality and developmental stage (25). Yet, in vitro produced embryos displayed a higher oxygen consumption, which was related to lower pregnancy rates (25). Thus, after initial selection around the time of EGA, in vitro morulae seem developmentally competent, as they display increased transcription of genes related to oxidative phosphorylation, as observed for the in vivo embryos. EIF2 signaling has previously been shown to be downregulated in parthenogenetically activated expanded porcine blastocysts compared to in vivo developed embryos, evidencing a correlation between aberrant EIF2 signaling and reduced developmental competence (26). EIF2 signaling was upregulated in morulae compared to 4-cell embryos, irrespective of embryo source, evidencing cell growth and proliferation (27).
During the morula to the hatched blastocyst transition, both in vivo developed and in vitro produced embryos displayed an enrichment of the pathways 14-3-3-mediated signaling, signaling of Rho family GTPases, and NRF2-mediated oxidative stress response. The 14-3-3 signaling plays a role in normal growth and development (28), cell polarity (29), and cell fate (30). Signaling of Rho family GTPases in mice has recently been shown to be important in blastocoel formation and the regulation of trophectoderm-speci c marker genes (31). The upregulation of genes related to this signaling pathway indicates a physiological blastulation of embryos, with normally developing trophectoderm and expanded blastocoels. In bovine, the NRF-2 mediated oxidative stress response is enriched in competent blastocysts (32), and the functions and processes related to the NRF-2 mediated oxidative stress response and oxidative phosphorylation pathways have been suggested to be related to developmental competence (33). Thus, our data suggests an initial natural selection of in vitro produced embryos around the EGA.
Embryos that develop past the 2-to 4-cell stage display a higher developmental competence. The enrichment of the shared signaling pathways in both in vivo developed and in vitro produced embryos during further early embryo development appeared to be indicative of largely similar developmental transcriptional pro les, potentially related to embryo competence.

In vivo developed versus in vitro produced hatched blastocysts
The differences between in vivo developed and in vitro produced hatched blastocysts were investigated to understand persisting transcriptional differences and their relationship to embryo competence. Whitworth et al. (2005) previously reported DEGs in porcine blastocyst stage embryos by comparing in vivo developed and in vitro produced embryos (23). Unlike the difference in expression of HMGB1 they reported, we did not nd a difference in its expression between in vivo developed and in vitro produced hatched blastocysts. The expression of HMGB1 has been associated with the number of nuclei per embryo (23), suggesting that the stage of our hatched blastocysts is likely similar, thereby allowing the comparison between in vivo developed and in vitro produced embryos at this developmental stage. Likewise, there was no signi cant difference in the expression of ATP5A1 between in vivo developed and in vitro produced hatched blastocysts. The expression of ATP5A1 has previously been used to indicate differences in metabolic rates in in vivo developed and in vitro produced blastocysts (23). In addition, 71% of genes related to cellular metabolism were reported to be upregulated in in vivo developed compared to in vitro produced porcine blastocysts (34). The in vitro hatched blastocysts in this study displayed a signi cant increase in amino acid metabolism. Among the genes related to amino acid metabolism, the arginine transporter SLC7A1 has previously been reported to be signi cantly upregulated in in vitro produced embryos compared to in vivo developed embryos (35). Porcine embryos deplete arginine from the culture medium at a higher rate at the expanded blastocyst stage compared to early blastocysts (36). The arginine concentration in the embryo culture medium used in this study was at 0.1 mM (37). It has previously been shown that adding arginine to a nal concentration of 0.36 mM to the embryo culture medium decreased the SLC7A1 transcript level in in vitro produced embryos to a level comparable to the in vivo developed embryos (35). In our study, the in vitro produced hatched blastocyst displayed a higher transcript expression of genes related to tRNA charging, cell cycle: G1/S checkpoint regulation, PEDF signaling and neuroin ammation signaling pathway. The in vivo developed embryos displayed a higher transcript expression of genes related to cyclins and cell cycle regulation. In vitro produced porcine blastocyst have previously been reported to display a higher transcript expression of genes involved in, among others, mRNA transcription, nucleotide metabolism, DNA metabolism, amino acid metabolism, and lipid metabolism (35). The higher metabolic rate of in vitro produced embryos is evidenced in our in vitro hatched blastocysts by an enrichment in tRNA charging, the G1/S checkpoint in which DNA damage is usually repaired prior to replication, and the PEDF signaling which is related to an increased level of apoptosis. This transcriptional pro le is in line with the proposed quiet embryo hypothesis, where viability is highest for embryos with a low rate of metabolism (38). In addition, embryos with high DNA damage display an increased amino acid turnover (39,40). Thus, we propose that the transcriptome of in vitro produced hatched blastocysts is indicative of an increased level of DNA damage, as evidenced by the enrichment of the G1/S checkpoint regulation and the PEDF signaling, and the higher degree of amino acid metabolism. The effect of adding higher concentrations of arginine, i.e., 0.36 mM instead of 0.1 mM, to the embryo culture medium on the embryos' amino acid metabolism and DNA damage should be assessed and to provide a valuable improvement of the currently employed in vitro fertilization pipelines.

Conclusions
Taken together, we show that early developing in vivo and in vitro produced embryos display largely similar transcriptome pro les. Embryos with compromised developmental competence arrested at an early stage of development. At the blastocyst stage, only little differences persisted between in vivo and in vitro, and there was no transcriptional difference between male and female embryos. The in vitro produced hatched blastocysts displayed expression of transcripts indicative of a higher metabolic rate and the arginine transporter, suggesting a lower developmental competence compared to the in vivo developed embryos.

Embryo production
Porcine embryos were allowed to develop in vivo and were produced in vitro (Figure 1). The developmentspeci c transcriptome dynamics were investigated by analyzing 4-cell stage embryos, morulae and hatched blastocysts. At the hatched blastocyst stage, male and female in vivo embryos were compared to the respective in vitro produced embryos.

In vivo
The in vivo embryos were produced as described previously (16). In brief, twelve German Landrace × Pietrian crossbred gilts were kept at the Research station Thalhausen of the Technical University of Munich, Germany. The gilts were synchronized using Altrenogest ReguMate® for 12 days. Intergonan® (PMSG) was applied once on the following evening at 750 iU. Ovogest® (human chorion gonadotropin) was applied 3.5 days later at 750 iU. The next day (day 0), all animals were inseminated with sperm of the same Duroc boar, named SWIROC. On day 2, 4 and 6 post insemination, four gilts were randomly selected and slaughtered in a commercial slaughterhouse to retrieve the embryos. The reproductive tracts were collected immediately after slaughter and the embryos were recovered from the reproductive tracts by ushing. The day 2 embryos were ushed from the oviduct with 2 ml phosphate buffered saline (PBS), while on day 4 and 6, embryos were ushed from the uterus with 10 ml PBS per horn. The collected embryos were washed twice with fresh PBS and single embryos were transferred to a cryotube and snap frozen in liquid nitrogen. All samples were stored at -80°C until library preparation. At 2, 4 and 6 days after insemination, 4-cell embryos, morulae and hatched blastocysts were collected, respectively. Per group, n = 5-10 embryos were randomly selected, stemming from three to four gilts.

In vitro
The in vitro embryos were produced as previously described (41,42). In brief, antral follicles on the surface of ovaries obtained from a local abattoir with a size of 3-6 mm in diameter were aspirated for the collection of cumulus-oocyte complexes (COCs) (42). The maturation of COCs displaying more than three layers of compact cumulus cells took place by culturing them in FLI medium contained FGF2, LIF and IGF1 for 44-46 hours (37). During the rst 22 hours, the COCs were cultured in maturation medium supplemented with human chorionic gonadotropin and pregnant mare serum gonadotropin, followed by 22-24 hours of culture in hormone free maturation medium in a humidi ed atmosphere of 5% CO 2 , 5% O 2 and 90% N 2 at 38.5°C (42). The in vitro fertilization was performed using frozen sperm derived from the same Duroc boar as used for the in vivo developed embryos to reduce an in uence on genetic variation (41). A group of 20 matured oocytes was co-incubated for 7 hours with 1.0 × 10 6 cells/mL in a porcine fertilization medium (Functional PeptideCo., Yamagata, Japan) in a humidi ed atmosphere of 5% CO 2 , 5% O 2 and 90% N 2 at 38.5°C (41). After fertilization, the cumulus cells and excess sperm were removed from the presumed zygotes and were cultured in Porcine Zygote medium-5 (Functional Peptide Co., Yamagata, Japan) in a humidi ed atmosphere of 5% CO 2 , 5% O 2 , and 90% N 2 at 38.5°C (41). The embryos were produced in four independent experiments. Morphologically normal embryos of 4-cell stage, compacted morulae and hatched blastocysts were collected at the following time points after fertilization, respectively: 48h, 100h and 174h. 4-cell stage embryos and compacted morulae were especially collected from a population of preselected 2-cell embryos at 30h after fertilization to avoid sampling of abnormal embryos. Prior to freezing, the embryos were washed trice with PBS containing 0.1% PVA. The embryos were transferred to a 0.5 ml Eppendorf tube and snap frozen in liquid nitrogen. Samples were stored at -80°C until library preparation. Per group, n = 5-10 embryos were randomly selected, stemming from three to four experiments.

RNA sequencing
Single 4-cell stage embryos (n = 5/production method), morulae (n = 10/production method) and hatched blastocysts (n = 10/production method) were obtained from in vivo ushing or were in vitro produced ( Figure 1). The library preparation for RNA-sequencing was conducted as previously described (43). Single embryos were lysed in 1 µl lysis buffer containing dNTPs and tailed oligo-dT oligonucleotides (30 nt poly-dT stretch and 25 nt universal 5'anchor sequence) plus 3.1 µl PBS (43). The lysed embryos were subjected to cDNA synthesis and library preparation with the smart-seq2 protocol as described previously (43). The libraries were pooled and sequenced on the NovaSeq6000 with a sequencing depth of 14 ± 4 million reads per sample (mean ± SD).

Data analyses and bioinformatics
Raw sequence reads (Fastq les) were analyzed on a locally installed Galaxy system (44). Basic read statistics and read quality was evaluated based on FastQC reports (45), and a MultiQC overview report of all samples was generated (46). Adaptors were clipped, sequences shorter than 20 bp were removed, and a low-quality end score of 20 was applied with the Trim Galore! tool (47). The trimmed reads were aligned against the porcine genome (Sus scrofa 11.1) with HISAT2 (48). The mapping rate was 84 ± 6 % (mean ± SD). An additional sequencing quality control was included. The reads of three representative and most abundant transcripts were aligned and visualized with the Integrative Genomics Viewer (IGV, version 2.8.2). The sex of both morulae and hatched blastocysts was assigned based on the expression of DDX3Y, EIF1AY and EIF2S3Y (16,49). Even though the morulae still had sperm attached to their zona, females were identi ed based on the absence of expression of the Y-chromosome speci c genes. The 4cell embryos were not sexed, as they were sampled around the time of EGA and as sperm were still attached to the zona. A between group analysis was conducted in R (version 3.6.1) (50). A self-organizing tree algorithm was ran for both embryo production methods to visualize the developmental dynamics (51). Differential gene expression analyses was conducted with EdgeR (52). A false discovery rate (FDR) of < 0.1% and an absolute log 2 FC > 1 was applied to identify the differentially expressed genes (DEGs).
The identi ed DEGs were used for pathway enrichment analyses (53). The functional analysis was conducted with the Qiagen Ingenuity Pathway Analysis (IPA) software. Human orthologues of DEGs were identi ed with the Mammalian Annotation Database for improved annotation and functional classi cation of Omics datasets from less well-annotated organisms (54). For a mean ± SD of 88 ± 1% of the DEGs, human orthologues were identi ed. To conduct canonical pathway analyses, different log 2 FC cut-offs were set to prevent an enrichment of redundant and overly general pathways (53). To prevent overly general pathway enrichments, a maximum of 3,000 DEGs should be used, while allowing the inclusion of as many DEGs as possible (Qiagen IPA user manual). An absolute log 2 FC cut-off of 6 was applied to obtain 3,064 DEGs for canonical pathway analysis for the in vivo 4-cell to morula stage, while a log 2 FC cut-off of 1.9 was applied to obtain 1,877 DEGs for canonical pathway analysis for the in vivo morula to hatched blastocyst stage. An absolute log 2 FC cut-off of 4 was applied to obtain 2,638 DEGs for canonical pathway analysis for the in vitro 4-cell to morula stage, while a log 2 FC cut-off of 2 was applied to obtain 2,860 DEGs for canonical pathway analysis for the in vitro morula to hatched blastocyst stage. An absolute log 2 FC cut-off of 1 was applied to obtain 1,329 DEGs for canonical pathway analysis for the in vivo versus in vitro hatched blastocysts. Canonical pathways were considered statistically signi cant with a p < 0.05 and an absolute z-score > 2.