Differentiation in the mammalian embryo is dependent upon spatial position - cells on the inside of the embryo remain pluripotent for a period until initiation of gastrulation while cells on the outer face of the embryo differentiate into TE and ultimately form much of the extraembryonic membranes. Here, using magnetic-assisted cell sorting and high-throughput next generation sequencing, we show the consequences of spatial differences between ICM and TE and subsequent divergence in lineage commitment for expression of genes regulating pluripotency and lineage commitment, cellular metabolism, and interactions with the maternal system.
Commitment towards the ICM lineage in the mouse is maintained by actions of Pou5f1 (Oct4), Sall4, Sox2 and Nanog; Cdx2 in the TE inhibits Pou5f1 expression and allows differentiation of extraembryonic membranes [3, 4, 7]. In the bovine, too, SOX2 and NANOG were overexpressed in ICM but expression of POU5F1 and SALL4 were not significantly different between ICM and TE. A high degree of expression of POU5F1 in the TE was expected because differences in the regulatory region of the POU5F1 gene in cattle as compared to the mouse gene make POU5F1 resistant to regulation by CDX2 . Nonetheless, POU5F1 expression is greater in the ICM of cattle [6, 42]. In the present study, expression of both POU5F1 and SALL4 were numerically greater for ICM; failure to find significant differences between ICM and TE may represent the small sample size. It should also be kept in mind that embryos produced in vitro have altered patterns of gene expression relative to embryos produced in vivo . Such alterations could change some of the differential gene expression between ICM and TE, as has been reported for the mouse embryo .
Analysis of genes upregulated in ICM provides some clues as to the signaling pathways required for specification, pluripotency, and other functions of the ICM. A total of 8 genes in the KEGG Jak-STAT signaling pathway were upregulated. In mice, LIF, which signals through the Jak-STAT pathway, can promote pluripotency of cells derived from the ICM . While LIF cannot cause bovine ICM cells to develop into stem cells , other molecules that signal through the Jak-STAT pathway are likely to be involved in regulation of the ICM. Several genes related to cellular migration were upregulated in ICM, as indicated by enrichment of the chemokine signaling pathway (10 genes) and axon guidance (7 genes) GO terms. In the mouse, blastomeres of the ICM can change position, at least in part to align position with subsequent formation of primitive endoderm [47–49]. Perhaps, movement is directed by guidance molecules such as chemokines.
Outer cells of the mouse blastocyst are committed towards the TE lineage through the actions of Yap1, Tead4, Gata3, and Cdx2 ([3, 4, 7]. We found no difference in CDX2 expression between ICM and TE using deep sequencing even though it is well established that the gene is expressed to a greater extent in TE of the bovine [6, 9, 42] and CDX2 expression was higher in TE than ICM in the qPCR experiment. CDX2 expression was very low in the deep sequencing experiment, especially compared to that of POU5F1. One possibility is that differences in CDX2 expression between TE and ICM at Day 7 (as detected by qPCR) become reduced at Day 8. Like seen earlier , other homologues of CDX2 were not detected (CDX1) or were nearly non-detectable (CDX4) (Additional file 2).
Another gene involved in TE lineage, GATA3, was expressed in higher amounts in TE. A similar but non-significant difference in expression between ICM and TE was noted earlier . There was no significant difference in TEAD4 or YAP1 expression between ICM and TE. Similar findings were observed in the bovine for TEAD4. A gene involved in development of extraembryonic ectoderm in mice, ELF5, was overexpressed in TE whereas another gene involved in development of extraembryonic membranes, EOMES, was barely detectable. In fact, there appears to be an absence or very low expression of EOMES in TE between day 7 and 15 of gestation in cattle . In addition, by Day 11 of gestation, trophoblast expression of ELF5 is inhibited and becomes limited to the epiblast .
It is notable that several genes characteristically expressed in ICM of mouse or human, DAB2, DSP, GM2A, SCD, SSFA2, and VAV3, [30, 32, 37] were significantly overexpressed in the TE of the bovine while CDH24, reported to be upregulated in the TE of the human , was expressed in higher amounts in the ICM of the bovine. Dsp and Dab2 are indispensible for embryonic development in mice and homologous recombination causes postimplantation embryonic failure [51, 52]. Clearly, as first shown by Berg et al. , divergent evolution in the control of early embryonic development means that study across a wide array of species is required to understand developmental processes fully.
By virtue of its position in the embryo, polarized morphology  and tight junctions between its member cells , the TE is fated to be the cell lineage through which the blastocyst interacts directly with the mother in terms of nutrient exchange, maternal-conceptus communication, and placentation. It appears that executing these functions places increased metabolic demands on the TE as compared to the ICM as indicated by upregulation of genes involved in metabolism, particularly those involved in lipid metabolism. Lipid accumulation in cultured bovine embryos is greater for TE than ICM, although the difference depends upon medium [54, 55].
It is through the TE that nutrients enter the embryo and from the TE that secretory products of the embryo must enter the uterine environment. Consistent with a role for the TE in uptake and delivery was upregulation of genes involved in endo- or exocytosis and membrane transport. Lysosomal-like structures have been reported to be more abundant in TE than ICM in cattle, at least for certain media [54, 55], and the mouse .
Molecules involved in signaling to the mother that were upregulated in TE include IFNT1, PAG2 and TKDP1. The role for IFNT1 is to act on the maternal endometrium to block luteolytic release of prostaglandin F2α[39, 56]. While this action is initiated later in pregnancy, between Day 15 and 17 of gestation, secretion of IFNT occurs as early as the blastocyst stage . TKDP1 is a member of the Kunitz family of serine proteinase inhibitors and may function to limit trophoblast invasiveness in species like the cow with epitheliochorial placentation . Little is known about the role of PAG2, which is the mostly abundantly expressed of at least 22 transcribed PAG genes . Unlike some PAG genes (the so-called “modern” clade), whose expression is limited to trophoblast giant cells formed later in development, PAG2 is expressed widely in the cotyledonary trophoblast and is predicted to be an active aspartic proteinase .
IFNT1, PAG2 and TKDP1 are all genes that are phylogenetically-restricted to ruminants. Another conceptus product that is produced more widely in mammals is estrogen. The role for embryonic estrogen is not known for most species but blastocyst estrogen has been suggested to be involved in hatching from the zona pellucida in hamsters  and in conceptus growth in the pig . The bovine blastocyst, too, produces estrogen  and the upregulation of genes involved in terpenoid backbone biosynthesis and steroid hormone biosynthesis suggest that the primary source of blastocyst estrogens is the TE.
Following blastocyst formation, the ruminant trophoblast undergoes a series of developmental steps that are dependent on changes in cell shape and spatial position, including hatching (which requires actin-based trophectodermal projections ), elongation (which leads to an increase in size of the conceptus from about 0.16 mm at Day 8 to as much as 100 mm or more at Day 16 ) and eventual attachment to the maternal endometrium (commencing around Day 20 in the cow . The upregulation of genes in the trophoblast for ontologies such as actin filament-based process, actin cytoskeleton organization, cell projection and cytoskeletal arrangement reflects the extensive changes in cell architecture required for these processes. In addition, three cathepsin genes, CTSB, CTSH and CTSL2, were upregulated in TE; these proteinases have been implicated in blastocyst hatching [59, 64].
Differences in gene expression between ICM and TE are probably due in large part to differences in transcription factor usage and to epigenetic modifications. Binding sites for the transcription factors PLAG1, RELA and RREB1 were enriched for genes overexpressed in ICM while binding sites for nine transcription factors (EGR1, GABPA, KLF4, MYF, SP1, MZF1, NHLH1, PAX5 and ZFX) were significantly enriched for TE. RELA is a subunit for NFκB, which in turn has been implicated in differentiation of trophoblast lineages from embryonic stem cells  and in function of trophoblast giant cells . Several of the transcription factors associated with genes upregulated in TE are involved in hematopoiesis, including EGR1 , GABPA , MZF1 , and ZFX . One of these transcriptional factors, GABPA, can enhance Pou5f1 expression in mouse embryonic stem cells  and another, KLF4, is a key regulator of maintenance and induction of pluripotency . The overall picture is one where hematopoiesis and stemness is under positive regulation in the TE. Another transcription factor associated with regulation of genes upregulated in TE was SP1. This protein exerts several actions to regulate trophoblast development and function, including activation of expression of other transcription factors such as Tfap2c and Id1. In the cow, SP1 becomes limited to binucleate cells of the trophoblast by Day 25 .
DNA methylation could be important for regulation of gene expression in the blastocyst because the promoter regions of over half of the genes that were upregulated in ICM or TE were classified as CpG positive. Indeed, the percent of genes classified as CpG positive for genes overexpressed in ICM or TE was higher than the percent that were classified as CpG positive for the entire bovine genome. Slightly fewer genes that were overexpressed in ICM were classified as CpG-positive than for genes that were overexpressed in TE, which might suggest more inhibition of gene expression by methylation in TE. It is noteworthy, however, that Niemann et al.  did not find a correlation between degree of CpG island methylation and amount of embryonic expression for eight genes examined. Recent evidence has been interpreted to signify that it is not the methylation state of individual CpG that determine gene expression but rather the methylation status of large regions of DNA that span multiple genes .
In cattle, there are conflicting data as to whether DNA methylation is less extensive for ICM or for TE in both embryos produced in vitro and by somatic cell nuclear transfer [78–80], Another epigenetic mark, H3K27me3, is similar for both cell types . Of the genes that were differentially regulated for ICM and TE, three were genes involved in epigenetic modification. Two were overexpressed in ICM: DNMT1, involved in maintenance of DNA methylation during succeeding cell divisions , and KDM2B, a lysine-specific histone dimethylase which catalyzes demethylation of H3K4 and H3K6 [82, 83]. In contrast, a DNMT3A like sequence, which establishes DNA methylation during development and also participates in methylation maintenance , was overexpressed in TE. The presence of increased transcript abundance for DNMT3A could be interpreted to mean that de novo DNA methylation occurs to a greater degree in TE, as is indicated by studies with embryos produced in vitro  and by somatic cell nuclear cloning . Further research is necessary to determine differences in DNA methylation between TE and ICM at the gene-specific and genome-wide level.
In general, analysis of a separate set of isolated ICM and TE by qPCR confirmed the results obtained for differences between cell types by deep sequencing. The exception was for CDX2, where there was no difference in expression as determined by SOLiD sequencing but where expression was greater for TE than ICM as determined by qPCR. The discrepancy could reflect either day of sampling differences (as discussed earlier) or, given the often-repeated observation that CDX2 is expressed to a greater extent in TE than ICM [6, 9, 42], an error induced by the deep sequencing procedure.
In conclusion, differentiation of blastomeres of the morula-stage embryo into the ICM and TE of the blastocyst is accompanied by differences between the two cell lineages in expression of genes controlling metabolic processes, endocytosis, hatching from the zona pellucida, paracrine and endocrine signaling with the mother, and genes supporting the changes in cellular architecture, stemness, and hematopoiesis necessary for development of the trophoblast. Much of the process leading to this first differentiation event seems to be under the control of genes such as NANOG and GATA3 that play central role in lineage commitment in the mouse. As found by others also [6, 42], there are fundamental differences from the mouse. Understanding the nature of the process of preimplantation development in mammals will necessarily require a comparative approach based on study of a variety of animal models.