Classical Models of TE Formation
The TE is the first tissue type to be specified during mouse development and is morphologically distinct from the ICM by the early blastocyst stage (Fig. 1). The TE, which will form the fetal portions of the placenta, begins as an epithelial sheet enclosing the future embryo, the ICM. Molecularly, TE and ICM lineages can be distinguished by the expression patterns of several lineage-specific transcription factors. Cdx2, which is related to Drosophila Caudal, is expressed in the TE of the blastocyst where it is required for promoting TE fate (Beck et al.,1995; Strumpf et al.,2005). Two ICM-specific transcription factors are Oct4, a POU domain protein, and Nanog, a homeodomain protein (Palmieri et al.,1994; Chambers et al.,2003; Mitsui et al.,2003). Interactions among these factors during blastocyst formation are thought to reinforce TE and ICM fates. Although the molecular mechanisms regulating TE and ICM fates are beginning to be understood, the foundation of this field was provided by classic studies of early mouse development.
The Inside–Outside Model
The earliest models of mouse development suggested that, as in nonregulative embryos, inheritance of cytoplasmic determinants assigned cell fates (reviewed in Alexandre,2001). However, this proposal was rejected after the observation that cells of the early embryo can change fate if moved to a new position (Mintz,1965). These observations and others led to the development of a new model of early lineage determination. The Inside–Outside Model (Tarkowski and Wroblewska,1967) proposed that, instead of cytoplasmic determinants, cell fate is established by position in the late morula, the stage at which inside and outside cell populations are distinct (Fig. 2A). Position could be translated into cell fate in several ways. For example, the extent of cell–cell contact might inform a cell of its position and thereby determine cell fate: inside cells would symmetrically contact neighboring cells, whereas outside cells would asymmetrically contact other cells. Alternatively, inside and outside microenvironments could differ somehow chemically and thereby differentially affect cell fate.
Figure 2. Classical models of lineage formation. A: The Inside–Outside Model proposes that, at the 16-cell stage, cell position on the outside or inside of the late morula determines its fate in the blastocyst. B: The Cell Polarity Model proposes that, at the 8-cell stage, cell polarity and cleavage patterns lead to establishment of outside and inside cells at the 16-cell stage and later. Symmetric cleavage produces two outside cells (Ba), whereas asymmetric cleavage produces one outside and one inside cell (Bb).
Download figure to PowerPoint
The Cell Polarity Model
Efforts to understand how inside–outside positions are created led to the development of the Cell Polarity Model (Johnson et al.,1981). This model proposes that cell fate is established before the late morula stage (Fig. 2B). According to the Cell Polarity Model, cell fate is established at the eight-cell stage by the establishment of cell polarity along the radius of the early morula. Subsequent cell divisions lead to symmetric or asymmetric distribution of polarity information, based on the angle of cell division. Division parallel to the radial axis generates two identical, polarized cells that inherited the entire distribution of polarity information (Fig. 2Ba). In contrast, division perpendicular to the inside–outside axis creates two different cells: one that inherited all polarity information and stays outside, and one apolar cell that stays inside (Fig. 2Bb). Polar and apolar cells would then differ in their developmental potential, thereby setting up TE and ICM lineages.
The Inside–Outside and Cell Polarity models are not completely at odds with each other, because both rely on differences along the radial axis to break the symmetry of the early embryo rather than cytoplasmic determinants inherited from the oocyte. However, these models differ in an important way: the Inside–Outside Model predicts that cell position drives cell fate, whereas the Cell Polarity Model predicts that cell fate drives cell position. Extensive studies have lent support to the Cell Polarity Model and established a set of observations that have since guided molecular studies.
The Cell Polarity Model: The Evidence
Early experiments focused on determining whether cell polarity is affected by basic properties of the early embryo, such as mitosis, cell–cell contact, or protein synthesis. These experiments have provided a foundation for more recent efforts to identify the molecular mechanisms underlying the processes of polarization, compaction, and elaboration of cell fate. Cell polarity was first observed as the localization of a variety of subcellular components to the apical cell pole (outside-facing) around the time of compaction. These components included surface microvilli, cytoskeletal elements, endosomes, and microtubule organizing centers (reviewed in Johnson et al.,1986a; Johnson and McConnell,2004). Fluorescently labeled concanavalin A (Con A), a lectin that binds to certain proteoglycans, also becomes localized to the apical microvillus pole around the time of compaction (Handyside,1980; Ziomek and Johnson,1980; Johnson and Ziomek,1981a). Because inheritance of apical and basal regions is thought to lead to differences in cell fate, the stability of these domains is central to the Cell Polarity Model. It follows, therefore, that some aspect of cell polarity must be retained during mitosis. Although the cytoskeleton and endosomes are redistributed during cell division, mitosis does not disrupt Con A localization to the apical cell surface (Johnson and Ziomek,1981b).
The role of cell–cell contact in setting up polarity is difficult to study in the intact embryo, because it is not possible to remove one cell without disrupting relationships between all the cells. Instead, disaggregated and reaggregated embryos have been used. These partial embryos compact essentially on schedule (Johnson and Ziomek,1981a) and have provided evidence that the development of polarity is initiated more effectively in the presence of cell–cell contact (Ziomek and Johnson,1980; Johnson and Ziomek,1981b). Also, the orientation of polarity may be influenced by regions of cell–cell contact. Couplets of cells isolated from the eight-cell embryo exhibit polarized Con A staining in a region opposite to their newly imposed region of cell contact (Ziomek and Johnson,1980; Johnson and Ziomek,1981b), suggesting that cell–cell contact gives orientation cues to the cell. This process has also been shown by disrupting cell adhesion in intact embryos by incubation in calcium-free medium (Pratt et al.,1982; Johnson et al.,1986b). The orientation of polarity is plastic for the first 5 hr of cell–cell contact, but becomes locked in thereafter (Johnson and Ziomek,1981b). Thus cell–cell contact is sufficient to define the angle of polarity, but is not necessary for the maintenance of polarity.
Remarkably, polarization and compaction not only proceed but occur prematurely after treatment with drugs that block protein synthesis (Levy et al.,1986) or activate protein kinase C (PKC; Winkel et al.,1990), offering three important clues about these processes: (a) that all components for polarization and compaction already exist in the early embryo, (b) that proteins made in the early embryo prevent precocious polarization and compaction, and (c) that posttranslational mechanisms, especially phosphorylation, oversee polarization and compaction. In more recent years, this last prediction has raised much interest in molecules such as PKC isoforms and their downstream targets.
Another important component of the Cell Polarity Model is that cell fate is influenced by the inheritance of polarity, as determined by the orientation of the cleavage plane (Fig. 2B). Cell division along the inside/outside axis produces two polar cells, both fated to occupy outside territory and become TE (Fig. 2Ba). On the other hand, division perpendicular to the inside–outside axis will produce one polar and one apolar cell, fated to occupy outside and inside space and to become TE and ICM, respectively (Fig. 2Bb). It follows, therefore, that every cell of the eight-cell stage will produce at least one TE cell during each ensuing division. In support of this proposal, isolated cells from the eight-cell stage give rise to polar–polar or polar–apolar couplets, but no apolar–apolar couplets (Johnson and Ziomek,1981a). Cell division planes occurring both parallel and perpendicular to the inside–outside axis have been observed using time-lapse microscopy (Sutherland et al.,1990), but evidence of a direct link between cleavage plane orientation and cell fate is still lacking.
Molecular Regulation of TE Vs. ICM Fate Choice
The link between cell polarity and cell fate suggests the exciting possibility that polarity-producing proteins may direct differentiation of the TE. The mature TE bears hallmarks of a polarized epithelium, with apical microvilli, and the polarized distribution of specialized protein complexes, such as tight and adherens junctions. As such, efforts in the field have focused on the examination of proteins known to be important for defining apical/basal polarity in other systems, such as Drosophila and cultured mammalian cells. Notably, because cells of the eight-cell embryo are initially unpolarized and nonepithelial, the preimplantion embryo offers a unique opportunity to study the process of mammalian epithelialization in vivo. Several proteins with compelling expression patterns as early as the eight-cell stage have been identified (Fig. 3A).
Figure 3. Localization of apical/basal proteins during blastocyst formation. A: In the compacted early morula, apical proteins (yellow) localize to a pole opposite areas of cell–cell contact, while basolateral proteins (purple) localize to regions of cell–cell contact and comprise adherens junctions. B: In the blastocyst, the apical protein complex assembles as tight junctions apical to the adherens junctions.
Download figure to PowerPoint
Based on timing of detection, adherens junction assembly appears to coincide with the assembly of an apical domain around the time of compaction in the eight-cell embryo. At this time E-cadherin, a cell adhesion molecule and a major component of adherens junctions, becomes greatly enriched at regions of cell–cell contact (Vestweber et al.,1987). In addition, the serine/threonine kinase EMK1 (Par1) is localized basolaterally (Vinot et al.,2005). Also at this time, several proteins localize to a small region of the apical cell cortex. These proteins include mouse homologues of Par3, Par6, and atypical PKCs (aPKCs; Pauken and Capco,2000; Thomas et al.,2004; Plusa et al.,2005; Vinot et al.,2005). Later, in the late morula, tight junctions begin to assemble just apical to the adherens junctions (Fleming et al.,1989; Sheth et al.,1997). Many apical proteins become localized to the tight junction around this time (Fig. 3B; Eckert et al.,2005; Vinot et al.,2005).
The distribution of these proteins is consistent with current models of apical/basal polarity formation in both Drosophila epithelium and mammalian cultured cells (reviewed in Nelson,2003; Margolis and Borg,2005). According to one current model, segregation of apical and basal domains is achieved through mutual antagonism between apical and basal protein complexes. Par3, Par6, and aPKC form an apical protein complex (reviewed in Etienne-Manneville and Hall,2003). Formation of this complex controls microtubule organization and excludes basolateral proteins, such as Lethal giant larvae (Lgl), from the apical membrane by direct aPKC-mediated phosphorylation (Plant et al.,2003; Yamanaka et al.,2003). Disruption of the apical protein complex or inhibition of aPKC function prevents formation of a functional apical domain and ultimately disrupts apical/basal polarity. Although it is unknown what initiates formation of the apical protein complex, these relationships can be adapted to the Cell Polarity Model to provide a molecular mechanism of cell fate segregation (Fig. 4). Assembly of the apical domain in the eight-cell embryo during compaction would exclude basolateral proteins from the apical membrane. Inheritance of the apical domain would lead to the continued polarization of daughter cells, whereas asymmetric division would result in the inheritance of the apical domain by only one daughter cell, whereas the other daughter cell would become apolar. Reassembly of the apical domain in the apolar daughter cell would be prevented by proteins associated with the basolateral domains.
Figure 4. Molecular cell polarity model. In the early morula, asymmetric cleavage (dotted line) leads to asymmetric distribution of the apical protein complex, while symmetric cleavage (not shown) leads to symmetric distribution of the apical protein complex. Polar cells that have inherited the apical protein complex maintain polarity. The apical protein complex may influence expression of trophectoderm (TE) genes such as Cdx2. Apolar cells that have not inherited the apical protein complex fail to express Cdx2. The basolateral protein complex may suppress the formation of the apical protein complex in apolar cells.
Download figure to PowerPoint
Functional analysis of many of these proteins has been provided using a variety of strategies (Table 1). Of interest, overexpression of either dominant-negative PKCλ, an aPKC, or to a lesser extent, RNA interference-mediated knockdown of ParD3 at the four-cell stage causes injected cells to preferentially occupy the ICM position at the blastocyst stage (Plusa et al.,2005b). Although it is not clear whether injected cells adopted an ICM fate, these observations support the notion that a functional apical domain is essential for polarity, outside cell formation, and probably TE fate (Fig. 4). However, it is less clear how the apical domain is established initially. One compelling model is that cell–cell contact and adherens junction assembly help establish or orient the apical domain. Of interest, ParD6b and EMK1 localization is disrupted in eight-cell embryos incubated in calcium-free medium (Vinot et al.,2005), suggesting cell–cell contact is important for the localization of these proteins. The absolute requirement for E-cadherin mediated cell–cell contact in cell polarization is not yet clear. Loss of zygotic E-cadherin expression does not disrupt the initiation of compaction, probably owing to perdurance of maternally supplied E-Cadherin (Larue et al.,1994; Riethmacher et al.,1995). Maternal germline knockout of E-cadherin postpones compaction, but embryos are soon rescued by zygotic expression of the paternal allele (De Vries et al.,2004). Examination of loss of both maternal and paternal alleles of E-cadherin will be important for establishing the role of cell–cell contact in establishment of cell polarity.
Table 1. Functional Studies on Epithelial Polarity Proteins in the Early Mouse Embryo
|Gene/protein||Type of protein||Manipulation||Phenotype||Reference|
|ParD3||PDZ domain protein||Injection of dsRNA||Injected cells tend to occupy inside position in blastocyst||Plusa et al.,2005|
|PKCλ||Atypical protein kinase C||Overexpression of dominant/negative||Injected cells tend to occupy inside position in blastocyst|| |
|PKCζ||Atypical protein kinase C||Inhibitory peptide||Delayed cavitation||Eckert et al.,2004|
|PKCδ||Novel protein kinase C||Inhibitory peptide||Delayed cavitation|| |
| || ||Activating peptide||Advanced cavitation|| |
|Ezrin||ERM domain protein||Overexpression of non-phosphorylable form||Disrupted adherens junctions (16-cell)||Dard et al.,2004|
| || ||Overexpression of phosphomimetic form||Disrupted adhesion, polarity (8-cell)|| |
|Jam1||Adhesion molecule||Neutralizing antibody||Delayed cavitation||Thomas et al.,2004|
|E-cadherin||Adhesion molecule||Neutralizing antibody||Randomly oriented polarity||Fleming et al.,1989|
| || ||Zygotic null||Cells dissociate after compaction, no cavitation, aberrant polarity||Larue et al.,1994; Riethmacher et al.,1995; Oshugi et al.,1997|
| || ||Maternal germline null||Blastomeres do not associate until morula stage||de Vries et al.,2004|
|β-catenin||Cadherin-assoicated protein||Maternal germline expression of truncated protein allele||Blastomeres do not adhere until four-cell stage|| |
|Vezatin||Transmembrane adherens junction protein||Morpholino oligonucleotide-mediated knockdown||Arrest||Hyenne et al.,2005|
|Epithin||Membrane serine protease||Injection of dsRNA||Compaction inhibited||Khang et al.,2005|
|RhoA||Small GTPase||Overexpression of constitutively active form||Premature, disorganized polarization||Clayton et al.,1999|
|Cdc42||Small GTPase|| || || |
Acquisition of apical/basal polarity also precedes TE formation in experimentally manipulated embryos. First, forced localization of apolar cells to the outside eventually leads to polarization of these cells (Ziomek and Johnson,1982; Ziomek et al.,1982; Johnson and Ziomek,1983) and adoption of a TE fate. Second, TE regeneration from isolated ICMs follows the relocalization of polarity-associated proteins at the apical membrane in outer cells of ICMs (Eckert et al.,2005). These observations suggest that removal from symmetrical cell–cell contact allows spontaneous reassembly of the apical protein complex (Fig. 5). In other words, formation of the apical protein complex is repressed in apolar cells by cell–cell contact, probably by the basolateral protein complex (Figs. 4, 5).
Figure 5. Model of trophectoderm/inner cell mass (TE/ICM) lineage determination. A: TE/ICM segregation during normal development. B: TE regeneration from ICM after immunosurgery.
Download figure to PowerPoint
Maintenance of the ICM/TE Lineage Restriction
Several transcription factors are known to be important for ICM and TE fates. One gene important for ICM development is a POU domain transcription factor, Oct4. Oct4 is initially detectable in nuclei of all cells of the early embryo, but its expression becomes restricted to the ICM upon blastocyst formation (Palmieri et al.,1994). Oct4 null embryos implant, indicating formation of a functional TE, but they soon die and lack ICM derivatives such as epiblast and yolk sac (Nichols et al.,1998). Notably, Oct4 null embryos appear to form normal blastocysts with both TE and ICM-like cells present, suggesting that the initial generation of the ICM is not dependent upon Oct4. Furthermore, Oct4 protein is not detectable in nuclei of Oct4 null embryos at the eight-cell stage, arguing against rescue by maternally supplied Oct4. Of interest, the ICM of Oct4 mutants may be misspecified as TE, because half of the Oct4 mutant blastocysts expressed the TE marker, ENDO-A throughout the ICM.
A gene important for early TE development also has been identified. Cdx2 encodes a Caudal-like transcription factor and is expressed in the TE of the blastocyst (Beck et al.,1995; Strumpf et al.,2005). Cdx2 protein is first detectable in the nuclei of select cells at the eight-cell stage (Niwa et al.,2005) and becomes restricted to outer cells of the late morula (Niwa et al.,2005; Strumpf et al.,2005). Cdx2 mutant embryos form a normal-appearing blastocyst, although ICM markers such as Oct4 and Nanog fail to be repressed in the TE (Strumpf et al.,2005), indicating a defect in restriction of the ICM/TE lineage. Cdx2 mutant TE eventually collapses and undergoes apoptosis, leading to blastocyst lethality before implantation. The Cdx2 mutant phenotype thus parallels the Oct4 mutant phenotype: both mutants form morphologically normal blastocysts, but exhibit defects in segregation of the ICM and TE lineages. This finding and other data have led to a model in which mutual antagonism between Oct4 and Cdx2 reinforce the segregation of ICM and TE fates in the blastocyst (Fig. 8A). This model has been supported by studies using stem cells isolated from the blastocyst lineages and will be discussed more later.
Figure 8. Stem cells from the blastocyst. A: Trophoblast stem (TS) cells are derived from the trophectoderm (TE) lineage, embryonic stem (ES) cells from the epiblast (EPI) lineage, and extraembryonic endoderm (XEN) cells from the primitive endoderm (PE) lineage of the blastocyst. B: Stem cell fates are reinforced by lineage-specific transcription factors: Cdx2 in TS cells, Oct4 and Nanog in ES cells, and Gata4/6 in XEN cells. Many of these factors positively regulate their own expression (circular arrow), whereas negative interactions (bars) maintain lineage restrictions.
Download figure to PowerPoint
Recently, it has been suggested that localized maternal Cdx2 mRNA and protein can be detected before the eight-cell stage and that maternal Cdx2 is involved in TE formation (Deb et al.,2006). However, knockdown experiments using RNA interference strategies have led to conflicting results, with one group finding preblastocyst lethality (Deb et al.,2006) and another finding phenocopy of zygotic null Cdx2 mutants (Meissner and Jaenisch,2006). Ultimate proof of the functional role of maternal Cdx2 will be provided by analysis of germline null mutants.