Morphogenetic analysis of peri-implantation development


Correspondence to: Jesse Mager, UMASS - Amherst, 661 North Pleasant Street, Amherst, MA 01003. E-mail:


Background:Although successful implantation is required for development in placental mammals, the molecular and morphogenetic events that define peri-implantation remain largely unexplored. Results: Here we present detailed morphological and immunohistochemical analysis of mouse embryos between embryonic day 3.75 and 5.25 of gestation, during the implantation process in vivo. We examined expression patterns of key transcription factors (Sox2, Oct4, Nanog, Cdx2, Gata6, Sox17, and Yy1) during pre- and postimplantation development. Additionally, we examined morphogenetic changes through analysis of ZO-1, Laminin, and E-Cadherin localization. The results presented reveal novel changes in gene expression and morphogenetic events during peri-implantation in utero. Here we show: (1) molecular and morphological changes in primitive endoderm cells as they transition from a salt and pepper distribution to a sheet covering the inner cell mass; (2) tissue-specific GATA6 levels; and (3) a striking pattern of SOX17 that is suggestive of a functional role either directing or permitting implantation at specific sites in the uterine epithelium. Conclusions: A growing number of knockout mice display peri-implantation lethality, and the data presented herein identify key morphogenetic landmarks that can be used to characterize mutant phenotypes, as well as further our basic understanding of peri-implantation development. Developmental Dynamics 242:1110–1120, 2013. © 2013 Wiley Periodicals, Inc.


Mammalian development begins following sperm and egg fusion in the ampulla of the oviduct. In the mouse, the first cleavage division occurs within 24 hr, with additional asynchronous cell divisions producing the ∼64-cell embryo approximately 100 hr post-human chorionic gonadotropin (hCG; Barlow et al., 1972). As the blastocyst forms several morphologically distinct features appear. The blastocoel cavity expands and two distinct cell populations emerge that have restricted fates: the inner cell mass (ICM) and trophectoderm (TE). The expanded blastocyst then undergoes a morphological change in shape from round to elongated, concurrent with or before implantation (Finn and McLaren, 1967). The embryo hatches from the zona pellucida, and extensive interactions between the embryo and maternal uterine tissue (involving epidermal growth factor, prostaglandins, cytokines, and other signaling pathways) lead to blastocyst activation, uterine receptivity, uterine epithelium extracellular matrix degradation, and implantation (Schlafke and Enders, 1975; Strickland et al., 1976; Copp and Rossant, 1978; Parr et al., 1987; Brenner et al., 1989; Yamazaki and Kato, 1989; Rohde and Carson, 1993; Kamijo et al., 1998; Galan et al., 2000; Paria et al., 2001; Zhang et al., 2013). In humans, it is estimated that as many as 75% of early embryo losses are thought to result from implantation failure (Wilcox et al., 1988; Norwitz et al., 2001; Cockburn and Rossant, 2010). Many important questions remain about mechanisms that control implantation, as well as the morphogenetic changes that occur during peri-implantation embryo development.

Beginning at the morula stage and throughout blastocyst formation, there are well-documented molecular events that influence and distinguish early cell fates. For example, OCT4, SOX2, fibroblast growth factor-4 (FGF4), and NANOG are restricted to the inner cell mass (ICM), while GATA3, CDX2, and EOMES specifically mark the cells of the TE (Feldman et al., 1995; Nichols et al., 1998; Ciruna and Rossant, 1999; Hancock et al., 1999; Avilion et al., 2003; Mitsui et al., 2003; Home et al., 2009). In the late blastocyst, a sheet of primitive endoderm (PrE) covers the ICM and can be identified by GATA4, GATA6, SOX7, and SOX17 expression (Cai et al., 2008; Plusa et al., 2008; Artus et al., 2011). Progress defining the mechanisms responsible for initial lineage specific gene expression and cell fate has been made largely through analysis of knockout (KO) embryos that have revealed genetic interactions and hierarchies of these critical proteins (reviewed in Rossant and Tam, 2009; Stephenson et al., 2012).

Specification of the ICM and TE are closely intertwined. Oct4 null embryos fail to form pluripotent ICM and although trophoblast cells are observed, neither ICM nor PrE proliferate in outgrowth assays (Nichols et al., 1998). Sox2 KO embryo outgrowths express reduced levels of Oct4 and Fgf4, and an abundance of the TE marker Pl1 (Avilion et al., 2003). In contrast, Nanog knock out embryos make blastocysts, but outgrowths differentiate specifically into PrE (Chambers et al., 2003; Mitsui et al., 2003). More recent work has shown that SOX2 activates Fgf4 and positively regulates Oct4 expression in the ICM (Avilion et al., 2003), while the TE marker CDX2 acts to repress Oct4 and Nanog and promote Eomes expression in the TE (Strumpf et al., 2005). Other studies show that GATA3 activates Cdx2 specifically in the TE through interaction with an intronic GATA binding site (Home et al., 2009), but control of Gata3 expression remains unclear.

Development of the PrE is also closely connected to ICM and TE formation. PrE precursors identified by expression of GATA-6 and platelet-derived growth factor receptor-α (PDGFRα) first arise within the ICM in the 64-cell embryo in a “salt and pepper” heterogenic pattern (Chazaud et al., 2006; Plusa et al., 2008; Yamanaka et al., 2010; Grabarek et al., 2012). The primitive endoderm marker GATA6 is not required for formation of OCT4-positive ICM in vivo, but Gata6 KO embryos fail to develop GATA4-positive PrE, DAB2-positive visceral endoderm (VE), or parietal endoderm (PE; Cai et al., 2008). Sox7 and Sox17 (both PrE markers) are also expressed in a mosaic pattern in the 64-cell embryo, in the PrE of the late blastocyst and in the extraembryonic VE of the gastrulating mouse embryo (Artus et al., 2011). It has been suggested that the two proteins are functionally redundant; however, the absence of Sox17 in implantation-delayed blastocysts results in a loss of PrE epithelial integrity, including premature delamination and migration of PE (Artus et al., 2011). Although the initial expression of Sox17 and Gata6 are not reliant upon FGF4, the level and heterogeneity of Fgf4 expression is crucial for differentiation of both the ICM and PrE lineages (Nichols and Smith, 2009; Yamanaka et al., 2010; Kang et al., 2013). Shortly after the “salt and pepper” GATA6/NANOG gene expression pattern is observed in the ICM, a single sheet of PrE is observed overlaying the ICM of the ∼80-cell embryo and then clearly covering the surface of the ICM of the 100-cell embryo (Plusa et al., 2008; Rossant and Tam, 2009; Morris and Zernicka-Goetz, 2012). However, the dynamics of the morphogenetic sorting events that occur during these stages (64–80 cell embryo) are not well documented in vivo.

By embryonic day (E) 5.5, the embryo adopts a radially symmetric egg cylinder morphology that has been well described (Wiley and Pedersen, 1977; Gonda and Hsu, 1980; Copp, 1981; Skreb et al., 1991; Hishinuma et al., 1995; Perea-Gomez et al., 2007; Rossant and Tam, 2009). Although the initial dynamics of egg cylinder formation are not completely defined, the directional growth of the egg cylinder has been detailed; egg cylinder expansion first occurs antimesometrially into the blastocoel cavity and then mesometrially toward the primary uterine lumen (Copp, 1979; Skreb et al., 1991). The enlarged primitive endoderm cells differentiate into visceral (classically named proximal) endoderm, the proamniotic cavity forms within the epiblast (EPI), and the polar trophectoderm differentiates into the extraembryonic ectoderm (Snell and Stevens, 1966). Additionally, many details regarding morphogenetic movements and distinct tissue formation have been defined during gastrulation in the postimplantation embryo (Downs and Davies, 1993). Although there is a large body of preimplantation (E0.5–E3.5) and postimplantation (E5.5–E8.5) literature, little is known about peri-implantation development in utero. This critical developmental window encompasses the embryo's first interaction with and subsequent implantation into the uterine luminal epithelium (ULE) as well as the transition from the blastocyst to the egg cylinder. An informative implantation staging description was produced by Finn and McLaren in 1967, which includes informative data collected through electron microscopy techniques (Finn and McLaren, 1967). More recently, a study of ex vivo E5.0–E6.5 embryos has revealed a new model for formation of both the extraembryonic ectoderm and visceral endoderm that suggests that the anterior visceral endoderm arises from polyclonal origins during peri-implantation (Perea-Gomez et al., 2007).

While both preimplantation and postimplantation embryos can be readily dissected and cultured, analysis of peri-implantation stages presents unique challenges as development of the embryo is reliant on interactions with maternal tissue. Although current protocols allow for relatively successful ex vivo embryo culture during the blastocyst to early egg cylinder transition (Morris et al., 2012), the molecular dynamics within blastomeres have not been thoroughly detailed in utero. Adding to the difficulty of isolating E4.0–E5.0 embryos, the uterus displays no obvious external clues to guide embryo collection until egg cylinder stages when decidual swellings permit embryo localization (Snell and Stevens, 1966). As a result, detailed molecular knowledge of peri-implantation development remains a gap in the understanding of mammalian development. Study of these stages in vivo may provide distinct insight regarding mechanisms that control the initial formation of a primitive endoderm sheet, the ICM to EPI transition, and embryo/uterus interactions or signaling.

Here, we present morphological and gene expression data collected in a systematic manner by sequential immunofluorescence to examine peri-implantation development within the context of the uterus in vivo. We document expression of key lineage-specific transcription factors and markers of epithelial biogenesis. Our data support current models but also suggest that in vivo: (1) the first GATA6-positive primitive endoderm cells are sorted to the periphery of the ICM and still express OCT4; (2) the GATA6-positive primitive endoderm cells that form a sheet over the ICM are OCT4 and Laminin negative, but E-Cadherin positive; (3) parietal endoderm cells exhibit strikingly higher GATA6 expression when compared with visceral endoderm cells; (4) maternal SOX17 in the uterine epithelium may guide or mark implantation sites. Our results are largely consistent with current models of PrE formation/specification, and add insight to the sorting events occurring as the embryo transitions from the blastocyst to early egg cylinder stage (reviewed in Stephenson et al., 2012).


Morphological Peri-implantation Staging

To examine morphology and gene expression of cells in the peri-implantation embryo, we collected uteri from super-ovulated and mated females at incremental stages of embryonic development (listed as hours post-hCG [hp-hCG]). Although many key aspects of early embryonic development have been previously described (summarized in Hogan et al., 1994), we sought to carefully analyze in vivo gene expression during this developmental window to further define peri-implantation events. To this end, we dissected, fixed, embedded, and sectioned whole uteri at incremental stages and analyzed embryos at 103 hp-hCG (∼E3.75, n = 14), 115 hp-hCG (∼E4.25, n = 35), 121 hp-hCG (∼E4.5, n = 19), 127 hp-hCG (∼E4.75, n = 18), 133 hp-hCG (E5.0, n = 16), and 139 hp-hCG (∼E5.25, n = 11). This straightforward approach allowed us to perform a thorough in vivo morphogenetic analysis of peri-implantation development.

We first analyzed changes in embryo size and cell morphology simply by visualizing E-Cadherin (CDH1) and DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride; nuclei) in sectioned embryos and uteri. At 103 hp-hCG, a compact clump of ICM cells as well as a large blastocoel cavity within the pseudo-stratified TE identify the round blastocyst (Fig. 1A). The 103 hp-hCG embryos have an average of 14.9 OCT4+/CDX2- (ICM) cells. Twelve hours later, 115 hp-hCG embryos contain 24.9- ICM cells per embryo, and the blastocyst has elongated with a markedly expanded blastocoel cavity (Fig. 1B). Surprisingly, we noticed that 115 hp-hCG embryos consistently have one or two cells per section precisely at the ICM/TE/cavity intersection that are CDH1 negative (Fig. 1B, and discussed extensively below). In slightly older embryos (121 hp-hCG), a complete layer of primitive endoderm covers the ICM; however, the sheet of PrE is CDH1 positive (Fig. 1C). During the dynamic movements resulting in egg cylinder formation at 127 hp-hCG (∼E4.75) the ICM/epiblast cells have reorganize into a single epithelial sheet of epiblast, forming the well documented radially symmetric epithelial “cup,” initially lacking a cavity (Fig. 1D,E). Coincident with these dramatic changes, the polar TE cells proximal to the early egg cylinder converge to form the extraembryonic ectoderm (EXE), and the overlying PrE differentiates into visceral endoderm (VE, Fig. 1D–F). The VE is distinct from PrE through membrane bound CDH1 and increased size at 127 hp-hCG (Fig. 1D). By 133 hp-hCG (∼E5.0), a constriction is observed that clearly defines the epiblast/extraembryonic ectoderm border (arrowhead in Fig. 1F) and the proamniotic cavity begins to form within the epiblast, resulting in the familiar cup-shape of the egg cylinder stage mouse embryo (Fig. 1F).

Figure 1.

Peri-implantation embryo morphology. A–F: Immunofluorescence of sectioned paraffin-embedded embryos at times indicated. G: Inner cell mass (ICM; CDX2-negative) cell counts at 103 hr post-human chorionic gonadotropin (hp-hCG; n = 15), 115 hp-hCG (n = 36), and 127 hp-hCG (n = 18). E-Cadherin is red and nuclear DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride) counterstain is blue. ICM, Inner cell mass; TE, trophectoderm; ULE, uterine luminal epithelium; PPrE, peripheral primitive endoderm; VE, visceral endoderm; EXE, extraembryonic ectoderm. Arrowhead in F: embryonic ectoderm/extraembryonic ectoderm boundary. “hrs” indicates number of “hours post-hCG administration” until time of dissection. Scale bars = 50 μm.

Peri-implantation Embryonic Gene Expression

We next examined in vivo expression patterns of the lineage specific transcription factors Pou5f1 (OCT4), SOX2, NANOG, GATA6, and CDX2 as well as E-Cadherin, ZO-1, and Laminin. Our goal was to analyze expression of known markers of cell fate and function to assess morphogenetic lineage emergence of the epiblast, PrE, and VE. We first designed and validated a protocol that faithfully permitted sequential rounds of immunofluorescence (see the Experimental Procedures section) to visualize all of these proteins within the same embryo sections. Based on this data we compiled a matrix of gene expression for each cell of every embryo examined (Fig. 2A). This in vivo profiling revealed that CDX2-negative (non-TE) cells are either SOX2 and OCT4 positive, or OCT4 and GATA6 positive, but almost never express all three factors (OCT4, SOX2, and GATA6; Fig. 2A). Furthermore, at 115 hp-hCG (∼E4.25) there are equal numbers of these 2 populations (OCT4 and SOX2 positive or OCT4 and GATA6 positive), with an average of 13.9 and 14.5 cells/embryo, respectively (Fig. 2A, rows 2–3). Consistent with other reports, the OCT4-positive, SOX2-negative ICM (CDX2-negative) cells are almost always GATA6 positive (Fig. 2A and arrows in Fig. 2B,C). Also in agreement with other reports (Rossant et al., 2003; Chazaud et al., 2006), the GATA6-positive, SOX2-negative cells also lack NANOG (arrow in Fig. 2G). Importantly, at 115 hp-hCG (∼E4.25) these cells are E-Cadherin negative and appear consistently at the ICM periphery and do not cover the entire ICM, forming a ring of cells at the intersection of the ICM/TE/blastocoel cavity (arrow in Fig. 2B–D). For ease of discussion, we will refer to these cells as peripheral primitive endoderm (PPrE). These PPrE cells are distinct in that they are GATA6/OCT4 positive, and clearly SOX2/CDH1 negative, as documented by others (Chazaud et al., 2006; Plusa et al., 2008). The elongated shape of the PPrE cells suggests that they may be migrating cells. Therefore, we examined CDH1 in combination with the basement membrane component Laminin. At 115 hp-hCG (∼E4.25), the GATA6-positive/CDH1-negative PPrE cells are Laminin positive (Fig. 2D). Additionally, ZO-1 localization revealed bilateral projections on either side of the PPrE cells (asterisks in Fig. 2G,H).

Figure 2.

Localization of peripheral primitive endoderm. A: Summary of molecular profiling of non-trophectoderm cells at 115 hr post-human chorionic gonadotropin (hp-hCG; n = 24) reveals four categories of gene expression (left most column). “+” indicates expression, “−” indicates no expression. Average number of cells/embryo shown in right-most column. B–H: Immunofluorescence of sectioned embryos at 115 hp-hCG (B–D and G–H), 121 hp-hCG (E), and 127 hp-hCG (F). B: DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride) is blue, OCT4 is green, and SOX2 is red. C: DAPI is blue, CDX2 is green and GATA6 is red. D–F: DAPI is blue, Laminin is green, CDH1 is red, and GATA6 is white. G,H: DAPI is blue, NANOG is green, and ZO-1 is white. Arrows in B, C, and D indicate OCT4-positive and GATA6-positive peripheral primitive endoderm. Arrow in E and F, visceral endoderm. Arrowhead in E and F, parietal endoderm. Arrow in G, peripheral primitive endoderm progenitor. H is magnified inset as indicated in G. Asterisks in H indicate ZO1-positive projections on the peripheral primitive endoderm cell. Scale bars = 50μm. **Note the section shown in B and C are the same embryo section, as are the sections shown in D and G,H, following sequential immunofluorescence. “hrs” indicates number of “hours post-hCG administration” until time of dissection. Scale bars = 50μm.

We next examined localization of the same panel of markers in slightly older embryos. At 121 hp-hCG (∼E4.5), two distinct populations of GATA6-positive cells are present. GATA6-positive cells now cover the entire ICM and also begin to line the adjacent TE (Fig. 2E). The GATA6-positive cells in contact with the ICM are Laminin negative and express moderate levels of CDH1 (arrow in Fig. 2E). In contrast, the GATA6-positive cells that line the TE have high levels of Laminin and still lack CDH1 (arrowhead in Fig. 2E), and at this stage, the level of GATA6 is fairly uniform in both populations.

In contrast, at 127 hp-hCG (∼E4.75), the levels of GATA6 are quite distinct between the two populations: “GATA6 low” PrE cells remain in contact with the ICM/EPI (arrow Fig. 2F), and “GATA6 high” cells are located more distally along the TE (arrowhead Fig. 2F). The GATA6 high cells remain Laminin positive, and comprise the parietal endoderm (PE, arrowhead in Fig. 2F). This striking difference in GATA6 levels between PrE/VE and PE is retained in the 133 hp-hCG (E5.0) embryo (compare cells marked by arrows and arrowheads in Fig. 3). Additional changes occurring by 133 hp-hCG include expansion of the proamniotic cavity within the epiblast and the appearance of the exocoelomic cavity within the extraembryonic ectoderm (summarized in Hogan et al., 1994).

Figure 3.

GATA6/CDH1 expression. Immunofluorescence of sectioned paraffin-embedded embryos at 133 hr post-human chorionic gonadotropin. A–D: Transverse sections of a single embryo as indicated in the diagram on the left. E,F: Sagittal sections of another embryo. DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride) is blue, Laminin is green, CDH1 is red, and GATA6 is white. Arrow, parietal endoderm. Arrowhead, visceral endoderm. E,F: F is a magnified image of the tissue indicated in E. “hrs” indicates number of “hours post-hCG administration” until time of dissection. Scale bars = 50 μm.

Of interest, in all embryos that we examined, the two cavities appear to arise independently, as we observe formation of the proamniotic cavity before exocoelomic cavity, as well as an intact epithelial sheet at the embryonic/extraembryonic junction, shown clearly by transverse sections (Fig. 3, compare A, C, and D with B). As development progresses, “GATA6 high” parietal endoderm cells are found both adjacent to the embryo (Fig. 3C) as well as distally, surrounding the implantation site (Fig. 3D,E and arrow in F). All of the PE cells express high levels of both GATA6 and Laminin.

These data support that PPrE cells located on the surface periphery of the ICM at 115 hp-hCG are the precursors to both the PrE and the PE, and suggest that migration may be intimately involved in the formation of both primitive and parietal endoderm. These morphological and molecular developmental changes in the peri-implantation embryo are summarized in Figure 4, which also illustrates the first occurrence of PrE precursors documented by others (Chazaud et al., 2006; Plusa et al., 2008; Grabarek et al., 2012; Stephenson et al., 2012).

Figure 4.

Schematic of peri-implantation development. Schematic of molecular and morphological changes during peri-implantation development between 103 hr post-human chorionic gonadotropin (hp-hCG; ∼E3.75) and 127 hp-hCG (∼E4.75) that summarizes morphological changes and gene expression in published work and the presented data.

Characteristics of the Peri-implantation Uterus

In addition to documenting novel morphogenetic dynamics in the embryo, our analyses also revealed molecular events that occur throughout the maternal tissue during peri-implantation stages. We document changes in both the ULE and uterine glandular epithelium during peri-implantation stages. Additionally, we document dynamic expression of YY1 and SOX17 that suggests they may play a role in maternal tissue during implantation.

The surface of the ULE undergoes morphological changes as implantation progresses. Initially the luminal surface of the ULE is a smooth epithelium. Coincident with implantation, a zipper-like corrugated surface forms and opposing surfaces begin to contact each other (arrow in Fig. 5A; Hedlund and Nilsson, 1971; Nilsson, 1974). This zipper-like morphology appears irregularly throughout the surface of the ULE, as previously documented (Finn and McLaren, 1967). Although CDH1 is expressed throughout the ULE (Paria et al., 1999), we often observed increased CDH1 on the apical ULE membranes (arrowhead in Fig. 5A). At the time of implantation, the glandular epithelium also has high levels of both YY1 (arrowheads in Fig. 5B) and SOX17 (arrowheads in Fig. 5C). SOX17 is expressed in a seemingly random pattern with regions of high and low expression (compare asterisks and arrows in Fig. 6A). Surprisingly, we found that 89% of embryos implant at “SOX17 high” ULE. Of interest, a marked increase of SOX17 is also observed precisely in the ULE that directly interacts with the embryo (arrow in Fig. 6B), suggesting either that the ULE enhances SOX17 levels in response to contact with the embryo, or that embryos implant specifically at peak regions of SOX17 protein expression.

Figure 5.

CDH1, SOX17, and YY1 in the uterine luminal epithelium. Immunofluorescence of sectioned uterine tissue at 121 hp-hCG. A,B: DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride) is blue, YY1 is green, and CDH1 is red. C: DAPI is blue, ZO1 is green, and SOX17 is red. Arrow, uterine luminal epithelium. Arrowhead, uterine glandular epithelium. “hrs” indicates number of “hours post-hCG administration” until time of dissection. Scale bars = 50 μm.

Figure 6.

Uterine luminal epithelium characteristics. A–E: Immunofluorescence staining of sectioned uterine tissue at 127 hr post-human chorionic gonadotropin (hp-hCG; A,B), 103 hp-hCG (C), 121 hp-hCG (D), and 133 hp-hCG (E). A,B: DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride) is blue, ZO1 is green, and SOX17 is red. C–E: Laminin is green, CDH1 is red, GATA6 is in white, and DAPI is blue. F: Schematic of uterine luminal epithelium degradation during peri-implantation. Arrow in A, SOX17-positive uterine luminal epithelium (ULE). Asterisks in A, SOX17-negative ULE. Arrowhead in A, uterine glandular epithelium. Arrow in B, localized increase of SOX17 at the site of implantation. Arrow in C–E, basement membrane of the uterine luminal epithelium. Arrowhead in E, basal lamina juxtaposed to the apical edge of the maternal tissue. ULE, uterine luminal epithelium; Em, embryo; TE, trophectoderm; VE, visceral endoderm; PE, parietal endoderm. “hrs” indicates number of “hours post-hCG administration” until time of dissection. Scale bars = 50 μm in A,B; 20 μm in C–E.

Before implantation, the ULE is a stratified cuboidal epithelium with a continuous basement membrane (arrow in Fig. 6C). However, after the embryo contacts maternal tissue, the ULE begins to lose cuboidal epithelial structure and down-regulates CDH1 (Fig. 6D). Additionally, the thickness of the ULE basement membrane markedly decreases, as shown by Laminin (compare arrows in Fig. 6C and 6D). At 121 hp-hCG (∼E4.5), the embryo has hatched from the zona pellucida, and TE cells (identified by juxtaposition to the basal lamina of the blastocoel; “TE” in Fig. 6D) are in direct contact with the apical surface of the ULE. By 133 hp-hCG (∼E5.0), the cells of the ULE have completely lost their epithelial structure and there is no detectable Laminin (arrow in Fig. 6E). At the same stage, a basal lamina is observed juxtaposed to the apical edge of the maternal tissue (arrowhead in 6E), where Laminin-positive parietal endoderm cells (PE in Fig. 6E) and GATA6-positive VE are also visible (VE in Fig. 6E), as summarized in Figure 6F.


Although there is a significant body of literature on peri-implantation anatomy, there is relatively little in vivo molecular analysis available. We have examined peri-implantation developmental dynamics and integrated our results with known key morphogenetic events and models (Fig. 4). This molecular and morphogenetic map of peri-implantation will be useful for analysis of embryonic phenotypes that occur between blastocyst formation and gastrulation. Through analysis of gene expression at precise time points during peri-implantation development in utero, we have documented subtle changes in primitive endoderm morphogenetic differentiation.

In the 64-cell embryo, a “salt and pepper” pattern is observed which reflects a differentiation of primitive endoderm cells that lack both SOX2 and NANOG and express GATA6 and PDGFRα (Chazaud et al., 2006; Plusa et al., 2008; Yamanaka et al., 2010; Grabarek et al., 2012). In the 80-cell embryo, the primitive endoderm cells largely cover the ICM and the primitive endoderm cells clearly cover the surface of the ICM in the 100-cell embryo (Plusa et al., 2008). Our data revealed a consistent localization of primitive endoderm cells at the periphery of the ICM in vivo between these stages (between the 64-cell “salt and pepper” stage and the 80-cell stage “PrE covered ICM” stage). At 115 hp-hCG, these peripheral primitive endoderm cells express high levels of OCT4, GATA6, and Laminin, but are SOX2 and E-Cadherin negative. Slightly later at 121 hp-hCG, the GATA6-expressing PrE cells that cover the ICM are E-Cadherin positive and Laminin/OCT4 negative (Fig. 2). The PPrE are likely the descendants of the GATA6-positive primitive endoderm first observed in the “salt and pepper” ICM (Rossant et al., 2003; Chazaud et al., 2006). However, our data suggest that morphogenetic events result in precise localization of these cells at the periphery before covering the ICM, and that migration across the ICM may be the mechanism by which a single sheet of PrE is formed. Alternatively, a Cadherin-mediated sorting may be responsible for localization of the CDH1-negative/GATA6-positive PPrE; Cadherins are known to be involved in several developmentally regulated sorting events (reviewed in Nelson et al., 2013), and appropriate regulation of cell adhesion is essential for TE formation and implantation (Larue et al., 1994; Fleming et al., 2001; Jha et al., 2006). Additionally, several reports have suggested a sorting model of “directionally biased relocation and apoptosis,” which results in survival of those PrE precursors that are at the ICM surface and apoptosis of precursors within deeper the ICM (Rossant et al., 2003; Chazaud et al., 2006; Plusa et al., 2008). Our descriptive observations are not inconsistent with this model, and suggest that this sorting may direct cells precisely toward the intersection of the ICM, TE, and blastocoel cavity.

Recently, a model of sequential transcription factor activation in the primitive endoderm was suggested in which Gata6 is first expressed, followed by Sox17, Gata4, and finally Sox7 (Artus et al., 2011). It has also been suggested that Nanog expression in the ICM prevents ectopic expression of Gata6, which in turn silences Nanog in the differentiating PrE (Chambers et al., 2003; Mitsui et al., 2003). At the time of lineage allocation, this mechanism is also controlled by specific levels of heterogeneous FGF4, which plays an important role in acquisition of cell lineage bias through positively regulating Gata6 expression and negatively regulating Nanog (Kang et al., 2013). The mutually exclusive expression patterns of Gata6 and Nanog observed in utero are consistent with this model. Functional investigation of signaling between the ICM, TE and PPrE cells at this time may shed light on these important developmental events.

In vitro preimplantation analyses have revealed that Laminins (1, 2/4 and 10/11) and their receptor integrin α7β1 may be key players in trophoblast adhesion and invasion (Klaffky et al., 2001). Importantly, deletion of either Lamb1 or Lamc1 (genes that encode two of the three chains that comprise Laminin 1) results in peri-implantation lethality (Smith and Wilson, 1971; Miner et al., 2004). Our analyses of in utero gene expression confirm that, as development progresses, Laminin is produced by the primitive endoderm (Chazaud et al., 2006; Niakan et al., 2010), but also show that Laminin production is highest in the PPrE and parietal endoderm cells at 115 hp-hCG (∼E4.25), and down-regulated in the cells of the PrE when they cover the ICM at 121 hp-hCG (Figs. 23). These results raise the possibility that the Laminin knockout phenotypes may be due to defects in PPrE cell function.

The mouse uterus is bifurcated with two fluid filled uterine horns. Sites of embryo implantation have been shown to be random in both mouse and rat models (Krehbiel and Plagge, 1962). However, we show that the great majority of embryos implant specifically at SOX17-expressing regions of the ULE. Consistent with the assertion of random implantation sites (Krehbiel and Plagge, 1962), these areas of SOX17 expression are not uniform, appearing in patches throughout the ULE (Fig. 6). Recently, Sox17 was identified as a target of progesterone in uterine tissue (Rubel et al., 2012), and progesterone is known to control implantation in all mammals examined (Zhang et al., 2013). Although functional experiments are needed to test this hypothesis, our data raise the possibility that progesterone is required to elevate SOX17 in regions of the ULE, permitting implantation at these sites. The initiation of the ULE degradation has been suggested to be independent of embryonic signaling, as decidualization can be induced by the presence of glass beads (Blandau, 1949a, 1949b). However, molecular crosstalk between the embryo and maternal tissue has been demonstrated for several signaling pathways (reviewed in Zhang et al., 2013). Our data indicate that uterine SOX17 may play a functional role during implantation and should be added to the growing list of genes and pathways known to be involved.


The data presented herein have revealed novel information about both embryonic and maternal morphogenetic events and advance our understanding of mammalian implantation. Our results are largely consistent with current models of early lineage specification, but support a model of in vivo peri-implantation development, in which: (1) following the “salt and pepper” ICM stage, GATA6/OCT4/Laminin-positive peripheral primitive endoderm are sorted to the intersection of the ICM, TE and blastocoel cavity; (2) GATA6-positive primitive endoderm cells (that are OCT4/Laminin negative) progressively form a sheet over the ICM; (3) parietal endoderm cells exhibit higher GATA6 levels than visceral endoderm cells; (4) maternal SOX17 in the uterine epithelium either guides or marks implantation sites. These data provide subtle additional details regarding the molecular mechanisms that govern peri-implantation development.


Embryo Production

Eight- to 10-week-old B6D2F1 female mice (Jax #100006) were superovulated with 10 IU pregnant mare's serum gonadotropin (Sigma) at 5 pm followed by 10 IU hCG (Sigma) 48 hr later. Copulation was determined by presence of a vaginal plug, and fertilization (E0.0) was assumed to be 12 hr post-hCG administration. Whole uteri collections were performed at noon and midnight. Embryo were collected and staged as follows: 103h post-hCG (∼E3.75/∼14.9-ICM cell embryo); 115h post-hCG (∼E4.25/∼24.9-ICM cell embryo); 121h post-hCG (∼E4.5); 127h post-hCG (∼E4.75/∼41-ICM cell embryo); 133h post-hCG (∼E5.0); 139h post-hCG (∼E5.25). Natural timed mating generally produces litters with as much as 6–12 hr developmental differences among embryos. We, therefore, chose to collect embryos from superovulated females to produce more consistent developmental stages at the desired time points. Even with this approach, embryo-to-embryo variation was still observed, as shown by the data points populating Figure 1G. This variation is likely due to stochastic developmental rates amplified across 5 days of development. Use of animals and protocols were approved by the University of Massachusetts IACUC.

Tissue Collection, Embedding, and Preparation for Immunofluorescence

Whole uteri were dissected between 103 hp-hCG and 139 hp-hCG as indicated. Freshly dissected uteri were cut away from the mesometrium and nutated in 4% paraformaldehyde for 2 hr at room temp or overnight at 4°C. The uteri were then dehydrated in a series of methanol washes of 1 hr each in 25%, 50%, 75%, methanol in phosphate buffered saline/0.01% Tween-20 (PBT), followed by two 1-hr washes in 100% methanol, and storage at −20°C. Before paraffin embedding the uteri were soaked in xylenes twice for 1 hr at room temperature and then molten paraffin (65°C) for 1 hr and then overnight. Uteri were cut into 5–7 pieces and incubated at 65°C for 1 additional hour in paraffin, embedded in plastic molds and cooled to room temperature. The 7.5-micrometer sections were cut with a microtome and floated in water, collected on Superfrost Plus slides (Fisher 12-550-15) and then dried overnight at 42°C. Sections were deparaffinized with three 10-min xylene washes and rehydrated with three 5-min washes in 100% ethanol, followed by 1 minute wash in 90%, 80%, 70% ethanol/water and finally rinsed in distilled water.


Antigen retrieval was performed by boiling for 5 min in 0.01 M Tris Base pH 10.0 with 0.05% Tween-20. After heating, slides were allowed to cool to room temperature then were washed three times in PBT for 5 min, and blocked with 0.5% milk in PBT for 2 hr at room temperature in a humidified chamber. Slides were incubated with primary antibodies in 0.05% milk/PBT overnight at 4°C in a humidified chamber at the concentrations listed below. The following day, three 15-min PBT washes preceded a 1-hr secondary antibody incubation in 0.05% milk/PBT in a humid chamber at room temperature. Slides were washed in PBT for 15 min twice and then in PBS for 15 min. Nuclei were counter stained with DAPI (Roche or Molecular Probes) in PBS (1:10,000) for 2 min and then rinsed once with PBS. Slides were sealed with ProLong Gold Antifade Reagent (Invitrogen) and coverglass. For sequential analysis, slides were incubated at 37°C for 30 min in PBS to remove the coverglass, washed in PBS 3 times for 10 min and then underwent antigen retrieval and immunofluorescence as described above. The following primary antibodies and concentrations were used: CDX2 (BioGenex AM392-5M 1:200); E-Cadherin (BD Biosciences 610181 1:250); Laminin (Sigma Aldrich L9393 1:250); NANOG (CosmoBio RCAB0002P-F 1:200); OCT3/4 (Santa Cruz sc5279 1:200); SOX2 (Santa Cruz sc17320 1:200); SOX17 (R&D Systems NL1924R 1:500); GATA-6 (R&D Systems AF1700 1:100); YY1 (Santa Cruz sc1703 1:100); ZO1 (Invitrogen 339100 1:200). Secondary antibodies were used at a concentration of 1:500 and include the following: 488 donkey anti-mouse (Invitrogen A-21202), 488 donkey anti-rabbit (Invitrogen A-21206), 546 donkey anti-mouse (Invitrogen A-10036), 546 donkey anti-rabbit (Invitrogen A-10040), 546 goat anti-rabbit (Invitrogen A-11035), 647 donkey anti-goat (Invitrogen A-21447), 647 donkey anti-rabbit (Invitrogen A-31573).


Digital images of whole-mount embryos were captured with a Nikon SMZ-1500 stereomicroscope equipped with a Spot Idea Digital Camera and Spot software (v4.6). Digital images of sectioned embryos were taken with a Nikon Eclipse TE2000-S inverted fluorescence microscope and QImaging Retiga Exi Fast 1394 camera fitted with a color-slider for use with brightfield images and Nikon NIS-Elements BR software or a Nikon Eclipse Ti inverted microscope with an Andor DR-228C camera and Nikon NIS-Elements AR software.

Cell Quantification

Cells were manually counted in digital images using Elements BR or AR software. ICM cells were identified by the absence of CDX2 expression, as such these include both the differentiating inner cell mass and primitive endoderm. Analysis of images from successive immunofluorescence experiments was used to assess combinatorial gene expression (Fig. 2A). For cells that spanned several sections, the image with the section that had the largest nuclear diameter was counted and that cell was excluded from counts of adjacent images.


Adobe Illustrator was used to generate schematic diagrams.