New era of trophoblast research: integrating morphological and molecular approaches

Abstract

Many pregnancy complications are the result of dysfunction in the placenta. The pathogenic mechanisms of placenta-mediated pregnancy complications, however, are unclear. Abnormal placental development in these conditions begins in the first trimester, but no symptoms are observed during this period. To elucidate effective preventative treatments, understanding the differentiation and development of human placenta is crucial. This review elucidates the uniqueness of the human placenta in early development from the aspect of structural characteristics and molecular markers. We summarise the morphogenesis of human placenta based on human specimens and then compile molecular markers that have been clarified by immunostaining and RNA-sequencing data across species. Relevant studies were identified using the PubMed database and Google Scholar search engines up to March 2020. All articles were independently screened for eligibility by the authors based on titles and abstracts. In particular, the authors carefully examined literature on human placentation. This review integrates the development of human placentation from morphological approaches in comparison with other species and provides new insights into trophoblast molecular markers. The morphological features of human early placentation are described in Carnegie stages (CS), from CS3 (floating blastocyst) to CS9 (emerging point of tertiary villi). Molecular markers are described for each type of trophoblast involved in human placental development. We summarise the character of human trophoblast cell lines and explain how long-term culture system of human cytotrophoblast, both monolayer and spheroid, established in recent studies allows for the generation of human trophoblast cell lines. Due to differences in developmental features among species, it is desirable to understand early placentation in humans. In addition, reliable molecular markers that reflect normal human trophoblast are needed to advance trophoblast research. In the clinical setting, these markers can be valuable means for morphologically and functionally assessing placenta-mediated pregnancy complications and provide early prediction and management of these diseases.

Introduction

The placenta plays a vital role in the development of the foetus and in the maintenance of pregnancy, including pregnancy-specific hormone production, nutrient transfer, gas exchange and immunotolerance. Abnormalities in the trophoblast, the functional cell type of the placenta, can cause inadequate placentation, leading to preeclampsia, the birth of small-for-gestational age neonates, placental abruption and late pregnancy loss. These placenta-mediated pregnancy complications are estimated to occur in 15% of pregnancies and are associated with the morbidity and mortality of the mother and foetus (Wilcox et al., 1988; Wang et al., 2003; Froen et al., 2004; Kuklina et al., 2009; Damodaram et al., 2011; Larsen et al., 2013; Lees et al., 2013; Rodger et al., 2016; Hiersch et al., 2017; Skeith and Rodger, 2017). However, the pathogenic mechanisms of these diseases remain unclear, and there are no effective treatments for placental dysfunction at the present time.

Placental dysfunction begins in the first trimester of pregnancy, before symptom occurrence (Burton and Jauniaux, 2004; Redman, 2014; Khong et al., 2015). In order to elucidate the pathogenesis of various pregnancy complications, it is important to understand the development of normal placenta. Placental development has been mainly described based on mouse embryogenesis and mouse trophoblast stem cells, but human placenta has unique morphology and trophoblast differentiation. This review elucidates this uniqueness from the aspect of structural development and molecular markers.

Cellular differentiation of the placenta

Cellularly, the placenta is primarily constituted of trophoblasts, but also includes stroma cells, macrophages and foetal endothelial cells (FECs). In this review, we focus on trophoblasts, which are responsible for the main function of the placenta. Trophoblast lineage, which describes the developmental sequence of the placenta, consists of trophectoderm (TE), cytotrophoblast (CT), syncytiotrophoblast (ST) and extravillous trophoblast (EVT; also called intermediate trophoblast) (Fig. 1A) (Lee et al., 2016). These cell types are morphologically and functionally distinct. The TE, which is the outermost layer of the blastocyst (Fig. 2A), is the precursor of all trophoblasts. It contributes largely to implantation. The CT is a uninuclear cell that is mitotically active and the progenitor of ST and EVT. The ST is a multinucleated cell that is generated from the fusion of uninuclear CT and is mitotically inactive. The ST serves as a place for nutrient transport, gas exchange and pregnancy hormone secretion. The microvilli on the surface of ST increase the efficiency of nutrient and gas exchange (Teasdale and Jean-Jacques, 1985). The EVT, also a uninuclear cell, is involved in the remodelling of maternal spiral arteries depending on the blood supply.

Human placental development in Carnegie stages

The differentiation and histological maturation of placental cells occur over a fixed time course. Morphological changes of human embryo development are described using Carnegie stages (CS) (O’Rahilly and Müller, 1987). CS can also be applied to human placental development, especially the earlier stages of human embryo development, such as CS5a-c. Here we review early CS and the foetal period (Fig. 2).

Carnegie stages 1 and 2 (Days 1–3)

The zygote starts from one cell fertilised oocyte. Cleavage begins, and finally the morula is formed. Differentiation into the placental lineage has not started yet.

Carnegie stage 3 (Days 4–6)

CS3 describes the blastocyst stage. The blastocyst is still enclosed in the zone pellucida, where it floats freely and ‘hatches’ from this surrounding extracellular matrix before implantation. At the early blastocyst stage, two cell groups, the inner cell mass (ICM) and outer cell layer, or TE, are visible. At the late blastocyst stage (expanded blastocyst), the ICM segregates into epiblast and primitive endoderm (hypoblast).

Carnegie stage 4 (Days 6–7)

At CS4, the blastocyst is anchored to the endometrium, and interaction between the embryo and maternal tissue begins (Fig. 2A) (Hill, 2020). The TE, which is the source of all trophoblast subtypes, shows a distinct polarisation. The TE plays an important role in embryo implantation and interaction with the maternal endometrium.

Carnegie stage 5 (Days 7–12)

At CS5, implantation is complete, and development of the trophoblast progresses. Based on the original classification of CS, CS5 is further subdivided into stages 5a, 5b and 5c based on trophoblast differentiation.

Carnegie stage 5a (Days 7–8)

At CS5a, the trophoblast exists as a mass of cells and is called solid trophoblast (Fig. 2B). Two distinct layers of cells (bilaminar germ disc), epiblast and primitive endoderm, appear. The amniotic cavity appears between the epiblast and TE and the yolk sac appears between primitive endoderm and TE.

Carnegie stage 5b (Days 9–10)

At CS5b, the trophoblast has two morphologically and functionally distinct layers of cells; the inner layer (i.e. CT) and outer layer (i.e. ST) (Fig. 2C). Lacunae appear in primitive ST, in which small vacuoles fuse with each other, thus leading to CS5b also being called ‘lacunar stage’ or ‘primitive syncytium’ (Fig. 2C) (Hertig et al., 1956; Boyd and Hamilton, 1970; James et al., 2012). Tissue specimens at the time of implantation indicate that CT and ST emerge concurrently (O’Rahilly and Müller, 1987), however, whether TE differentiates into primitive ST through a CT stage is controversial (Boyd and Hamilton, 1970; Knofler and Pollheimer, 2013).

Carnegie stage 5c (Days 11–12)

In CS5c, the primitive ST continues to penetrate into the endometrium, maternal capillaries connect with lacunae in the primitive ST, and maternal bloods flow into the lacunae. The ST forms about three quarters of the total trophoblastic shell, and the CT constitutes the remaining one-quarter (O’Rahilly and Müller, 1987).

Carnegie stage 6 (Days 13–14)

CS6 marks the first stage of chorionic villi development, when the CT grows externally and penetrates into the primitive ST to form cellular columns. The columns are called primary villi and have two cellular layers (Fig. 3A); the outer layer is ST, and the inner layer is CT. The embryonic body at this stage is now about 0.2 mm diameter in size. Furthermore, extra-embryonic cavities coalesce to form the chorionic cavity.

Carnegie stage 7 (Days 15–17)

At the second stage of chorionic villi development, i.e. CS7, extra-embryonic mesoderm proliferates and penetrates into the core of the primary villi. This structure, in which extra-embryonic mesoderm is surrounded by CT and ST, is called the ‘secondary villi’ (Figs 2D, 3B). The embryonic body is now about 0.4 mm in size at the diameter. Laterality is established, gastrulation starts and germ layers are formed from the epiblast.

Carnegie stages 8 and 9 (Days 18–21) and later embryonic stages

Extra-embryonic mesoderm in the secondary villi differentiates, and blood cells and vessels are developed. The vessels of the villi connect with the vessels of the embryonic body, and the maternal-placental (uteroplacental) blood circulation begins. This structure is called the ‘tertiary villi’ and is completed by the end of Week 8 after fertilisation (Fig. 3C). The third stage of chorionic villi development (from Weeks 3 to 8 after fertilisation) is quite long; the embryonic body has three germ layers and organogenesis begins. Furthermore, at the end of Week 8 after fertilisation, the tertiary villi have been completed, and organogenesis has ended.

Cellular differentiation and structural development of the placenta occur during the early embryonic stages, and CS5b-9 is important for these phenomena (O’Rahilly and Müller, 1987). They are not only the periods of organogenesis in the embryonic body, but also the periods of placental differentiation. It is very difficult to obtain samples of embryo and placental specimens, thus, most direct observations of CS depend on old serial sections from a few collections of human embryos.

The foetal embryonic stage

At Week 8 after fertilisation, when the tertiary villi have developed, EVT emerges at the front edge of the villi. In other words, EVT is differentiated from CT cell columns. EVT forms plugs in the maternal spiral arteries to occlude these arteries and block maternal blood from flowing into intervillous spaces, although whether the plugs obstruct these arteries completely is controversial (Kurjak and Kupesic, 1996; Valentin et al., 1996; Weiss et al., 2016; Roberts et al., 2017). The oxygen tension of the intervillous space is at a low concentration at 6–8 weeks after fertilisation, otherwise known as the plugged stage (Rodesch et al., 1992; James et al., 2006). At Week 12 after fertilisation, the EVT plugs dissolve, and the oxygen tension changes from low to normal oxygen concentration (Hustin et al., 1988; Jauniaux et al., 1991; Rodesch et al,. 1992; Coppens et al., 1996; James et al., 2006a). Accordingly, low oxygen tension stimulates CT proliferation, and normal oxygen tension stimulates CT differentiation into ST and EVT (Genbacev et al., 1997; James et al., 2006b; Tuuli et al., 2011). Recently, it was reported that low oxygen tension promotes immature EVT differentiation (Wakeland et al., 2017; Chang et al., 2018; Treissman et al., 2020); therefore, the role of oxygen concentration in EVT differentiation needs to be examined in detail. EVT replaces arterial endothelial and vascular smooth muscles, and this remodelling transforms maternal spiral arteries into low-resistance, high-flow conduits. Consequently, the intervillous space is filled with high amounts of maternal blood (Pijnenborg et al., 2006).

The ST forms the outer layer of the chorionic villi that contacts directly with the maternal vascular space and functions to exchange gases and metabolites between the mother and foetus (Benirschke et al., 2000; Lee et al., 2016). The other main function of ST is to produce most pregnancy-specific proteins and hormones. CT and ST are located on the villi; therefore, these cell types are called villous trophoblasts. In contrast, EVT infiltrates the decidua, myometrium and maternal spinal arteries. The functions of the EVT are to anchor the foetus to the uterine wall, maximise oxygen delivery and establish the maternal–foetal interface (Graham and Lala, 1992; Benirschke et al., 2012b; Lee et al., 2016). The EVT is largely composed of uninuclear cells (Kurman et al., 1984) with various morphologies including round, polyhedral or spindle shape (Kaufmann and Castellucci, 1997; Gersell and Kraus, 2011). In humans, the placenta is mainly constituted of CT, ST and EVT (Fig. 1A).

Human placental development with molecular markers

The anatomical boundary of the placenta has not been defined well, which has motivated a search for new molecular markers. The placenta includes cells that originate from TE and ICM and ultimately from the fertilised oocyte. The TE gives rise to all placental epithelium (CT, ST and EVT), whereas the ICM gives rise to stroma cells (placental connective tissue cell), Hofbauer cells (placental villous macrophage) and FECs. Whether the placenta is also comprised of maternal-derived cells is controversial. We therefore exclude maternal-derived cells, such as decidua, decidual natural killer cells, dendritic cells, decidual macrophages, maternal endothelial cells and epithelial glandular cells, from our definition in this review (Fig. 4A). Details of each trophoblast subtype are described in the sections ‘Cellular differentiation of the placenta’ and ‘Human placental development in Carnegie stages’ above.

Trophoblasts can be identified in accordance with their anatomical location, cellular morphology and expression of particular markers in vivo (Hsi et al., 1991). On the other hand, stage-specific molecular markers are very important for the characterisation because there is no spatial information in vitro. Below we summarise cellular stage-specific molecular markers based on comprehensive immunostaining, microarray and RNA-sequencing analyses (Table I).

Pan-trophoblast

GATA3, TFAP2C and KRT7 are the consensus markers of ‘first trimester trophoblasts’ in vivo and in vitro (Lee et al., 2016). Importantly, Lee et al. confirmed the above consensus markers with post-implantation placenta, which assures GATA3, TFAP2C and KRT7 are expressed in CT, ST and EVT (Table I) (Lee et al., 2016; Liu et al., 2018; Okae et al., 2018; Vento-Tormo et al., 2018). Therefore, CT, ST and EVT can be distinguished using these consensus markers in combination with other markers. Overall, Lee et al. summarised the criteria of trophoblast as (i) the demethylation of ELF5 promoter, (ii) protein expression of GATA3, TFAP2C and KRT7, (iii) negative expression of HLA-A and HLA-B and (iv) high expression of microRNA in the chromosome 19 microRNA cluster (C19MC).

In addition to the consensus markers described above, GATA2 and KRT19 are persistently expressed markers for TE, CT, ST and EVT (Table I) (Shorter et al., 1993; Aghajanova et al., 2012; Assou et al., 2012; Yan et al., 2013; Blakeley et al., 2015; Petropoulos et al., 2016; Liu et al. 2018; Okae et al., 2018; Stirparo et al., 2018). GATA3 and TFAP2C are also consensus markers and expressed in pre-implantation trophoblast (Yan et al., 2013; Blakeley et al., 2015; Nakamura et al., 2016; Petropoulos et al., 2016; Deglincerti et al., 2016a; Stirparo et al., 2018). Therefore, these four genes mark trophoblast in the first trimester (gestational Weeks 1–12).

On the other hand, KRT7 expression is controversial in human TE. KRT7 expression was observed in human TE with immunostaining (Niakan and Eggan, 2013), but single-cell RNA-sequencing (scRNA-seq) data of pre-implantation embryo revealed TE expresses KRT7 at the same low level as epiblast and primitive endoderm (Yan et al., 2013; Blakeley et al., 2015; Petropoulos et al., 2016; Stirparo et al., 2018). Peri-implantation in vitro culture of human embryo demonstrated KRT7 expression in CT just after implantation (Table I) (Shahbazi et al., 2016; Deglincerti et al., 2016a). scRNA-seq data during peri-implantation in vivo have not been published, because specimens for this stage are rare. For these reasons, in this review, we assume KRT7 is expressed just after implantation.

Negative molecular markers are also useful. HLA-A, HLA-B, THY1 (CD90) and VIM are non-trophoblast markers (Shorter et al., 1993; Yan et al., 2013; Blakeley et al., 2015; Petropoulos et al., 2016; Liu et al., 2018; Okae et al., 2018; Stirparo et al., 2018). In addition to these positive and negative markers, evaluating trophoblast cell subtypes in combination with specific molecular markers is recommended.

Trophectoderm

TE is trophoblast during the pre-implantation stage. Cdx2 is specifically expressed in mouse TE and is an important cell lineage determinant between mouse ICM and TE (Niwa et al., 2005). Previous reports also used this gene as a TE marker in human (Table I) (Deglincerti et al., 2016b; Shao et al., 2017; Iwasawa et al., 2019). However, other reports showed that CDX2 is expressed only partially in human TE (Chen et al., 2009; Niakan and Eggan, 2013; Petropoulos et al., 2016; Stirparo et al., 2018), and CDX2 positive cells were contained in first-trimester placenta as residual TE (Hemberger et al., 2010; Horii et al., 2016; Haider et al., 2018; Soncin et al., 2018). To date, there are no specific markers of human TE.

Cytotrophoblast

For the assessment of CT, we can use multiple consensus markers of trophoblast. The identification of CT-specific markers can distinguish CT from other cell types in vitro or in vivo. CT expresses TP63, VGLL1 and ITGA6 (CD49f) specifically (Table I) (Damsky et al., 1992; Shih and Kurman, 2004; Kalhor et al., 2009; Knofler and Pollheimer, 2013; Okae et al., 2018; Soncin et al., 2018). ITGA6 is a cell surface protein that constitutes the principal adhesion receptors for laminin. Stem cell niches have abundant laminin, and ITGA6 plays a role in stemness (cell proliferation and self-renewal) (Krebsbach and Villa-Diaz, 2017). Reflecting the proliferative capacity of CT, a subset of CT expresses Ki67 (MKI67) (Haider et al., 2018; Lee et al., 2018), the positive ratio of which decreases with passing gestational weeks (Arnholdt et al., 1991).

Syncytiotrophoblast

Lee’s consensus markers of trophoblast, such as GATA3, TFAP2C and KRT7, are all expressed in ST (Liu et al., 2018; Okae et al., 2018), although they also are considered CT and EVT markers (Lee et al., 2016). On the other hand, the representative cell surface marker of ST is SDC1 (CD138) (Jokimaa et al., 1998; Okae et al., 2018). A combination of these four markers and ST-specific markers (Table I) may provide a more robust definition of ST in vitro and in vivo. ST cells are mitotically inactive (Ki67 negative) (Turco et al., 2018) and secrete placental specific hormones such as hCG (human chorionic gonadotropin), hPL (human placental lactogen) and PSG (pregnancy-specific glycoprotein) into the maternal systemic circulation. The fusion of CT to make ST is initiated by the upregulation of ERVW-1 (Syncytin-1) (Mi et al., 2000; Yu et al., 2002; Frendo et al., 2004).

Extravillous trophoblast

EVT also expresses the consensus markers GATA3, TFAP2C and KRT7 (Lee et al., 2016; Liu et al., 2018; Okae et al., 2018), and HLA-G is widely accepted as a specific cell surface marker of mature EVT (Table I) (Heap et al., 1988; Tarrade et al., 2001; Shih and Kurman, 2004; Mao et al., 2007; Kalhor et al., 2009; Benirschke et al., 2012c; Knofler and Pollheimer, 2013; Liu et al., 2018; Okae et al., 2018). Although CT and ST express neither Class I (HLA-A, HLA-B, HLA-C, HLA-E, HLA-F and HLA-G) nor Class II (HLA-DR, HLA-DQ and HLA-DP) HLA on their surfaces, EVT expresses HLA-C and HLA-E Class I molecules (Loke, 1989). HLA-C and HLA-G binds maternal killer immunoglobulin-like receptors and leukocyte immunoglobulin-like receptors, respectively (Sharkey et al., 2008; Parham and Moffett, 2013; Vento-Tormo et al., 2018), therefore, these HLA ligands facilitate immune tolerance (Hunt et al., 2005). ITGA5 is a cell surface protein that constitutes the principal adhesion receptors for fibronectin, which is one of the main constituents of the extracellular matrix near the cell column (Humphries et al., 2006). EVT deposits high doses of fibronectin, and the expression of ITGA5 facilitates cell-matrix binding (Earl et al., 1990; Damsky et al., 1992). In addition, EVT invades the decidual layer when it expresses ITGA1, and ITGA1 is essential for the invasion (Damsky et al., 1992; 1994). CD146 (also known as melanoma cell adhesion molecule, MCAM) is another representative cell surface marker of EVT (Table I) (Shih, 1999; Pujades et al., 2002; Wang and Yan, 2013; Lee et al., 2018; Turco et al., 2018), but it is also expressed by neural crest cells, ganglion cells and activated T lymphocytes (Shih, 1999; Pujades et al., 2002; Wang and Yan, 2013). EVT is classified as two types based on its location and immunohistochemical profiles: implantation site EVT and chorionic laeve EVT. The major function of implantation site EVT is to establish the maternal–foetal circulation by remodelling the maternal spiral arteries during early pregnancy, while the function of chorionic laeve EVT is unknown (Shih et al., 2018). MUC4 is expressed in implantation site EVT, but not chorionic laeve EVT (Table I) (Mao et al., 2007). The significance of discriminating implantation site EVT and chorionic laeve EVT is the classification of gestational trophoblastic diseases (GTDs). Immunohistochemistry has revealed that the origin of several GTDs differs with the EVT type. For example, exaggerated placental site and placental site trophoblastic tumour arise from implantation site EVT, and placental site nodule and epithelioid trophoblastic tumour arise from chorionic laeve EVT (Heller, 2018; Shih et al., 2018; Kaur and Sebire, 2019). It should be noted that MUC4 cannot identify GTD subtypes.

Stroma cell

Stroma cells, which are derived from extra-embryonic mesenchyme, are located in the placental villous core (Fig. 4A) (Kaufmann et al., 1977). They provide mechanical support to the villous structures and control intervillous blood flow (Demir et al., 1997). Stroma cells express HLA-A, HLA-B, THY1 (CD90) and VIM, in contrast to trophoblasts, which do not express these molecular markers. The high expression of HLA-A and HLA-B in stroma cells indicates that these cells are derived from ICM, not TE. Furthermore, stroma cells do not express the trophoblast markers GATA2, GATA3, TFAP2C, KRT7 or KRT19 (Liu et al., 2018; Okae et al., 2018).Despite the above knowledge, the development of stroma cells over gestation is unclear at this time.

Hofbauer cell

Hofbauer cells are macrophages in the stroma of chorionic villi (Seval et al., 2007; Schliefsteiner et al., 2017; Reyes and Golos, 2018), which often localise close to FECs and CT (Fig. 4A) (Cervar et al., 1999; Khan et al., 2000; Seval et al., 2007; Loegl et al., 2016; Reyes and Golos, 2018). These macrophages are derived from foetal mesenchymal cells, not trophoblast. Hofbauer cells are characterised by general macrophage markers, such as CD14, FCGR1A (CD64), CD68, CD163 and LYVE1 (Seval et al., 2007; Schliefsteiner et al., 2017; Reyes and Golos, 2018; Vento-Tormo et al., 2018).

Hofbauer cells secrete the pro-angiogenic molecules VEGF and FGF2 (Loegl et al., 2016). They also secrete EGF, which promotes trophoblast proliferation (Maruo et al., 1995; Leach et al., 2004; Barber et al., 2005; Johnstone et al., 2005; Vento-Tormo et al., 2018). In addition, they express sprouty proteins that regulate branching morphogenesis (Anteby et al., 2005). Therefore, Hofbauer cells can play a role in vascular conditioning, the proliferation of CT and the branching morphogenesis of placental villi during development by paracrine signals and cell-to-cell crosstalk in the placental villi. (Cervar et al., 1999; Khan et al., 2000; Seval et al., 2007; Vento-Tormo et al., 2018). Like stromal cells, the development of Hofbauer cells over gestation is unclear at this time.

Foetal endothelial cell

FECs are of mesenchymal origin and lie between circulating foetal blood and the placental stroma (Fig. 4A). In short, they are vascular endothelial cells in the chorionic villi and are characterised by general endothelial markers such as PECAM1 (CD31) and CD34 (Mutema and Stanek, 1999; Vento-Tormo et al., 2018). However, FECs are characterised by no specific marker and can only be distinguished from other cells based on their location in the chorionic villi.

Summary

The majority of molecular markers regarding placental development are intracellular markers, and few cell surface markers have been identified (Table I). Cell surface markers are valuable for trophoblast isolation from placenta and the evaluation of cell lines, because they allow live trophoblasts to be detected with antibodies. Cell surface antibodies also provide qualitative and defined endpoints, and they can be used to elucidate the timing and order of molecular changes. Combinations of cell surface markers can be used to identify the trophoblast state, and thus provide a quality standard for trophoblast experiments.

Models for human trophoblast research

Common in vitro models of human trophoblast include primary cultures, placental villous explants, choriocarcinoma cell lines and immortalised cell lines derived from placenta. Additionally, BMP-treated human embryonic and induced pluripotent stem cells (ESCs and iPSCs, respectively) have been established, and recent studies have reported the long-term culture system of human CT, both monolayer and spheroid. The characterisation of human trophoblast cell lines is summarised in Table II.

Conventional human trophoblast cell lines

As conventional human trophoblast cell lines, primary trophoblast cell culture, placental villous explant, trophoblast tumour-derived cell lines (choriocarcinoma cell lines) and immortalised trophoblast cell lines have all been used in placental research because of their expected similarities with in vivo trophoblast (Table I). The first two models are culture systems and cannot be expanded for a long time. Choriocarcinoma cell lines, such as BeWo, JEG-3 and JAR, have been around for five decades (Pattillo and Gey, 1968; Pattillo et al., 1968a,b, 1971a,b; Kohler and Bridson, 1971), but they have various chromosomal aberrations (Poaty et al., 2012). Owing to widespread transfection methods, immortalised cells, such as HTR-8/SVneo, Swan 71 and TCL-1, have also been established (Graham et al., 1993; Lewis et al., 1996; Straszewski-Chavez et al., 2009). Choriocarcinoma and immortalised cell lines overcome the problem of long-term passage, but their genome-wide DNA methylation patterns are clearly different from primary trophoblast (Novakovic et al., 2011). The poor differentiation potency and transcriptome discrepancy are important limitations for human placental research (Bilban et al., 2010; Apps et al., 2011; Ji et al., 2013).

Pluripotent stem cell-derived trophoblasts

Human ESCs and iPSCs differentiate into trophoblast lineage in a species-dependent manner (Xu et al., 2002; Sudheer et al., 2012; Amita et al., 2013; Li et al., 2013; Yang et al., 2015; Horii et al., 2016; Lee et al., 2016; Yabe et al., 2016; Horii et al., 2019). The conventional way to induce trophoblast lineage is to expose these pluripotent stem cells (PSCs) to bone morphogenetic protein-4 (BMP4) (Xu et al., 2002), and the induced trophoblasts were recently shown to express the distinctive placental signature (Roberts et al., 2018). Several efficient differentiation protocols for trophoblast have been published: (i) BAP treatment (BMP4 plus the activin A signalling inhibitor A83-01 and the FGF2 signalling inhibitor PD173074) (Amita et al., 2013; Yang et al., 2015; Yabe et al., 2016), and (ii) BMP4 plus a Wnt/beta-catenin inhibitor (Horii et al., 2019). The PSCs can thus terminally differentiate into hCGβ^positive ST-like cells and HLA-G^positive EVT-like cells. However, whether they differentiate into trophoblast remains controversial. First, BMP-treated PSCs do not agree with the criteria of first-trimester trophoblast, that is, they express HLA-A and HLA-B and have a low expression of C19MC microRNA (Lee et al., 2016). In addition, these models are contradictory in developmental biology, as conventional human PSCs correspond to the post-implantation stage, which is different from mouse ESCs (Nakamura et al., 2016), and the lineage segregation between ICM and TE takes place at the blastocyst stage (De Paepe et al., 2013; Petropoulos et al., 2016; Rossant and Tam, 2018; Leng et al., 2019). To solve these problems, a new type of stem cell has been established; expanded potential stem cells retain the ability to differentiate into any embryonic lineage as well as trophoblasts (Gao et al., 2019), although this model has not been fully evaluated.

Trophoblast stem cells and trophoblast organoids

New culture systems for human trophoblast stem cells (monolayer, 2D) and trophoblast organoids (spheroid, 3D) allow for long-term expanding trophoblast, thus providing the next generation of human trophoblast cell lines and human placental embryology (Table II) (Haider et al., 2018; Okae et al., 2018; Turco et al., 2018). Since both trophoblast stem cells and organoids meet the first-trimester trophoblast criteria, these culture systems are considered improvements to previous trophoblast cell lines (Haider et al., 2018; Okae et al., 2018; Turco et al., 2018). Trophoblast stem cells derived from human blastocyst and first-trimester placenta behave as long-term expanding CT and can differentiate into ST and EVT with high purity (Okae et al., 2018). ST specification from human trophoblast stem cells is induced by forskolin (cAMP activator), whereas EVT specification is induced by Neuregulin-1, ALK4/ALK5/ALK6 inhibitor and Matrigel (Okae et al., 2018). Trophoblast organoids derived from first-trimester placenta behave as long-term expanding 3D villous trophoblasts, consisting of CT and ST (Haider et al., 2018; Turco et al., 2018). In these organoids, ST are surrounded with CT without stroma cells (Haider et al., 2018; Turco et al., 2018). This reversed polarity is characteristic of ordinary organoid culture systems (Sato et al., 2009, 2011; Bartfeld and Clevers, 2015; Bartfeld et al., 2015). Trophoblast organoids can differentiate into HLA-G^positive EVT cells under Okae’s EVT differentiation medium (Turco et al., 2018) or regulation of Wnt signalling (Haider et al., 2018). Although the proliferative capacity of CT decreases prominently after 10 weeks of gestation (Mayhew, 2014), trophoblast stem cells and trophoblast organoids from second trimester or term placenta are strongly needed to model placenta-mediated pregnancy complications, because complications do not appear until these later stages.

Placental classification by animal taxonomical approaches

Species with placenta-like structures

Animals can be grouped as oviparous species, which lay eggs, and viviparous species, which have placentas and include most mammals. One major characteristic of oviparity is the constant volume of the egg during development. To excrete nitrogen, the egg produces uric acid, which maintains the constant egg volume (Baggott, 2009). In contrast, viviparous species produce urine, which is excreted through the placenta. The placenta itself is classified into yolk-sac placenta and chorio-allantoic placenta. Yolk-sac placenta is developed from the yolk sac, in which the allantoic membrane does not adhere to the chorion, and material exchange occurs between the mother and the foetus through a part of the yolk sac. Fish, reptiles and most marsupials have yolk-sac placenta (Freyer and Renfree, 2009). Chorio-allantoic placenta includes the human placenta. Here, the allantoic membrane and serosa coalesce to form the chorio-allantoic membrane, develop blood vessels in the mesodermal part of the chorio-allantoic membrane and adhere closely to the maternal endometrium. Placentalia (Eutheria) have chorio-allantoic placenta as too does the marsupial bandicoot. The subclade Euarchontoglires in Placentalia consists of Glires and Euarchonta; Glires includes Rodentia, and Euarchonta includes Primates.

Structural classification of the placenta

Chorio-allantoic placenta is grossly classified as the following four types (Fig. 5A). (i) Diffuse placenta: the placenta is scattered throughout the surface of the chorion (horses and pigs). (ii) Cotyledonary placenta: the placenta has multiple small placentae isolated on the amniotic membrane in a mottled pattern (cows and sheep) (Furukawa et al., 2014). (iii) Zonary placenta: the placenta is shaped like a band around the centre of the amniotic membrane (dogs and cats). (iv) Discoidal placenta: the placenta forms a discoid region in the uterus (humans, macaques, mice and brown bears).

The placenta is responsible for nutrient and gas exchange between the mother and foetus at the surface where the chorion and allantoic membrane are facing. Although the structure on the foetal side typically has an epithelial cell layer (trophoblast cell layer), connective tissue and capillary vessels, the structure on the maternal side varies across species. Therefore, the placenta is also classified histologically into the four types below (Fig. 5B) (Wagner et al., 2014). (i) Epitheliochorial placenta: the maternal side is composed of capillaries, connective tissue and an endometrial epithelium, and the endometrium is in contact with the maternal–foetus boundary (horses and pigs). (ii) Syndesmochorial placenta: there is no epithelium on the maternal side, and the connective tissue is in contact with the maternal–foetus boundary (cows and sheep). (iii) Endotheliochorial placenta: there is no connective tissue on the maternal side, and the endothelial cells of the capillary are in contact with the maternal–foetus boundary (dogs and cats). (iv) Haemochorial placenta: there are no capillary vessels on the maternal side, thus, maternal blood is in direct contact with the foetus (humans, macaques and mice).

Notably, epitheliochorial placenta has three layers (capillaries, connective tissue and an endometrial epithelium), and haemochorial placenta has no cell layer between the maternal blood and trophoblast. These histological differences suggest that maternal damage during delivery is relatively large in the haemochorial placenta (Abrams and Rutherford, 2011). In contrast, the extent of maternal damage has not been established for epitheliochorial and syndesmochorial placenta because they have endometrial epithelium or endometrial stroma cells next to the decidua membrane.

Differences in placentation between humans and other species

Research on placental development with other species

Our understanding of trophoblast has been refined with long-time research using different animal models, but mainly mouse and rat. Studies with mouse trophoblast stem cells (Tanaka et al., 1998), which were established 20 years ago, model mouse trophoblast development in vivo with remarkable accuracy (Perez-Garcia et al., 2018). An analysis of mutant mouse trophoblast stem cells and conditional knockouts suggests that early miscarriage (embryonic Days 9.5–14.5) is usually associated with severe placental malformations (Perez-Garcia et al., 2018). Although mouse and human placentation are not exactly the same, mouse trophoblast stem cells provide a useful platform for placental research, because experiments in vivo are easy and the gestational period is short.

Other popular animal model is macaques (a subfamily of old world monkey) and marmoset (a subfamily of new world monkey). In general, these monkeys have served as valuable models for human diseases and treatments due to their close similarities to humans in terms of genetic and physiological features (Chan, 2013). Monkeys have also been studied for many years in placental research (Myers, 1972; Hearn et al., 1988; Enders, 1995, 2000; Carter and Pijnenborg, 2011; Carter et al., 2015). Past studies were mainly based on morphological approaches, although scRNA–seq approaches have begun recently (Nakamura et al., 2016). In addition, the in vitro culture of macaque embryos has been established (Ma et al., 2019; Niu et al., 2019). Great apes (gorillas, chimpanzees, bonobo and orangutans) are closely related to humans and have very similar placenta structures (Carter and Pijnenborg, 2011; Pijnenborg et al., 2011). However, many governments ban or severely restrict research on great apes, because great apes are cognitively similar to humans (Knight, 2008).

As useful as these models are, placenta formation between humans and other animal species differs in many ways. Below we summarise some of these differences.

Gross and histological differences of placenta

As described above, humans, other primates and mice all have discoidal placenta, but the shapes are different. Humans, great apes and mice have a single placental disc, but macaques and marmosets have bidiscoidal placenta (Table III) (Myers, 1972). Some researchers have observed that a secondary attachment generates bidiscoidal placenta (Carter et al., 2015), but the true cause is a matter of debate. In terms of histomorphology, primates and mice have the same haemochorial placenta among the four types of histological classification in Fig. 5B. Implantation begins at the blastocyst stage in human (Day 7) (O’Rahilly and Müller, 1987) and mouse (Day 4) (Theiler, 1972), but at the embryonic period in cow (Day 20) (King et al., 1980; Wathes and Wooding, 1980) and horse (Day 35) (Bowen and Burghardt, 2000; Allen and Stewart, 2001). Since mouse placenta shares more similarities to human placenta compared with other species in early placental development except those whose experimental use is greatly restricted, like great apes, toxicity studies with mouse have been used to evaluate teratogenicity and developmental toxicity (Pijnenborg et al., 1981).

Nevertheless, mouse and human placenta have significant structural differences. The haemochorial placenta is classified into villous type (humans, apes and old world monkeys) and labyrinth type (mice, rats and new world monkeys) (Fig. 5, Table III). Human villous haemochorial placenta has two layers of trophoblasts that separate maternal blood from foetal blood (Fig. 5A, Table III). The bichorial trophoblast layers consist of a single syncytial layer and a single cytotrophoblastic layer (Rossant and Cross, 2001; Soncin et al., 2015). The cytotrophoblastic layer becomes discontinuous and covers 20% of chorionic villi in term placenta, whereas the syncytial layer does not change (Benirschke et al., 2012a). On the other hand, mouse labyrinth haemochorial placenta has three layers of trophoblasts between the maternal and foetal blood (Table III) (Enders and Blankenship, 1999). The mouse placenta is called labyrinth because of its maze-like appearance in cross-section (Fig. 5B). The trichorial trophoblast layers consist of two syncytial layers and a single mononuclear layer of unknown function. The labyrinth zone of mouse placenta does not have a proliferative cytotrophoblastic layer (Simmons, 2013). These species differences may be related to the much shorter duration of mouse pregnancies than human. From the perspective of drug and nutrient diffusion, the ability for molecules to cross the placenta is strongly influenced by the number or the thickness of cell layers between the maternal and foetal blood (Mihaly and Morgan, 1983; Schroder, 1995; Pere, 2003; Furukawa et al., 2014). It is important to consider the diversity of the histological structures for drug metabolism and placental transfer, although active and facilitated transport also have effects.

Another difference between mouse and human is the supplied blood volume. In the villous haemochorial placenta (human), maternal spiral arteries open directly into the intervillous spaces. The chorionic villi protrude into the intervillous space and are bathed directly in maternal blood (Furukawa et al., 2014). Interestingly, maternal blood flow regularly spouts into the intervillous space, as if washing the foetal villi (Dancis and Schneider, 1986; Burton et al., 2009). In contrast, maternal blood is exposed to a large amount of foetal blood indirectly through the trophoblast layers in labyrinth haemochorial placenta (mouse).

In addition, several articles have reported that placental drug transfer is determined largely by placental blood flow (Mihaly and Morgan, 1983; Schroder, 1995; Furukawa et al., 2014). Circulation between the mother and foetus also affects the transfer ratio of nutrients ([concentration in foetus]/[concentration in maternal plasma]) (Tarui et al., 2018). In placental circulation, the flow rate and the flow direction affect the transfer ratio (Faber et al., 1992; Faber, 1995; Perazzolo et al., 2017). Mouse placenta has countercurrent blood flow (opposite directions) through the maternal and foetal vessels (Faber et al., 1992; Adamson et al., 2002). On the other hand, human chorionic villi are exposed to the maternal pool of blood in the intervillous space (Costa et al., 1992; Acharya et al., 2016; Plitman Mayo, 2018), resulting in a lower transfer ratio than the countercurrent system (Faber et al., 1992). Although monkeys and humans share more similar placental development compared with mice, it is difficult to conduct large-scale toxicity tests with monkeys. At the same time, when using mouse models, the structural differences between species must be considered.

Morphology during implantation

The first morphological difference between primates (human and monkey) and rodents (mouse and rat) is the direction of implantation. The polar TE overlying ICM is the site of primate implantation (Herzog, 1909), whereas rodent implantation is initiated in the opposing mural TE (Table III) (Wimsatt, 1975). Despite the difference in the direction of implantation, the main component of the placenta originates from polar TE in both groups. The distinct behaviour of polar and mural TE has been proposed to result from differential signal transduction from the epiblast (Gardner and Johnson, 1972; Gardner et al., 1973; Papaioannou, 1982).

The blastocysts of macaque and mouse remain in the uterine cavity; in other words, the blastocysts embed partially within the uterine cavity. This type of implantation is called superficial implantation. On the other hand, the implantation type of human is interstitial (Table III) (Chapman et al., 2013). That is, the human blastocyst embeds entirely within the endometrial connective tissue and there is a massive endometrial reaction. Thereafter, EVT invades through the decidua to the inner one-third of the myometrium (interstitial invasion) (Ferretti et al., 2007).

The primitive syncytium forms lacunae filled with maternal blood (Hertig et al., 1956; Enders, 1989). The lacunae coalesce with each other gradually and become one large intervillous space. The primitive syncytium is observed only in primates and guinea pig (Enders and Schlafke, 1969; Schlafke and Enders, 1975; James et al., 2012; Soares et al., 2018). In contrast, the ectoplacental cone is a specific structure in mice and rats (Table III). The ectoplacental cone is a cellular mass formed by intensive proliferation of polar TE (Amoroso, 1952; Snell and Stevens, 1966; Barlow and Sherman, 1972; Gardner et al., 1973; Enders and Blankenship, 1999). It is located in the maternal endometrium and contains progenitors for a number of trophoblast subtypes (Simmons et al., 2007; El-Hashash et al., 2010; Mould et al., 2012).

Predominantly expressed genes of TE

A comprehensive analysis of the expressed genes in mouse, monkey and human embryos is valuable for understanding trophoblast specification. These three animals share certain similarities in that their trophoblasts express GATA2, GATA3 and TFAP2C (Nakamura et al., 2016). The main signalling pathway that governs TE specification in the mouse is HIPPO signalling (Nishioka et al., 2008, 2009), but in human this is unknown at this time. In addition, unlike in the mouse, the expressions of EOMES, ETS2 and ELF5, three transcription factors for TE specification, were undetectable in the human pre-implantation embryo (Table III) (Petropoulos et al., 2016; Kubaczka et al., 2017; Roberts et al., 2018). In particular, EOMES is not expressed in any human trophoblast subtype. In contrast, Dab2, a primitive endoderm marker in mouse (Morrisey et al., 2000; Yang et al., 2002; Moore et al., 2013), is expressed in human TE (Blakeley et al., 2015; Petropoulos et al., 2016; Stirparo et al., 2018). Thus, some master regulators in mouse trophoblast lineage are not expressed in human TE, and the expression timing of other regulators differs in the two animals. Consequently, it is unclear whether the mechanisms of trophoblast differentiation and maintenance are similar among mouse, monkey and human. In mice, about 70 knockout lines have been shown to cause lethality with placenta and placental dysmorphologies (Perez-Garcia et al., 2018), but these genes have not been examined in humans. Knockout experiments in human trophoblast cell lines would therefore be useful. Gene expression has been investigated in cymolgous monkey in vivo, showing different kinetic gene expression before and after implantation (Nakamura et al., 2016).

Cell types of trophoblast

Spongiotrophoblasts in mouse are the putative mouse counterpart of human CT in CT cell columns. Both mouse trophoblast giant cells (TGCs) and glycogen cells (glycogen trophoblasts), which are derived from spongiotrophoblasts, are analogous to human EVT (Georgiades et al., 2002; Cross, 2005; El-Hashash et al., 2010; Soncin et al., 2015) (Fig. 1B). TGCs have their name because of their ‘giant’ nuclei, which is caused by DNA synthesis without nuclear division. Human EVT do not have giant nuclei (Soncin et al., 2015). Although mouse TGCs and human EVT commonly function as an anchor for the placenta to the uterus, some functional features of these cells are different. TGCs are the main endocrine cells producing hormones in mouse placenta, whereas in humans those cells are ST, and human EVT secrete a small amount of hormones (e.g. human placental lactogen) (Tarrade et al., 2001). Another difference between human and mouse trophoblast is the invasive capacity. Both TGCs and glycogen cells have little invasive capacity (Simmons et al., 2007). Human EVT, on the other hand, contributes largely to vascular remodelling of the spiral arteries, which then dilate (Soncin et al., 2015). It has been reported that all trophoblast cell types among monkeys, apes and humans are similar morphologically and functionally (Myers, 1972; Enders, 1995, 2000; Carter et al., 2015).

Peri-implantation stage research with experimental animals

Ethics, including the use of experimental animals, will continue to limit placental research and prevent the use of in vivo human specimens from reaching the implantation phase. Therefore, placental research using other animals, such as macaques, for insights of early trophoblast development is required. Recently, the specification of primordial germ cells was identified by immunostaining and in situ hybridisation using macaque embryos during the peri-implantation stage (Sasaki et al., 2016). Other studies revealed the development trajectories of epiblast, primitive endoderm, TE and primordial germ cells (Ma et al., 2019; Niu et al., 2019). In total, our understanding of early human placental development will depend on a combination of (i) human cells in vitro, (ii) non-human primates in vivo and in vitro and (iii) mice in vivo and in vitro.

Future perspectives

In the past several years, new models of in vitro trophoblast, such as trophoblast stem cells and trophoblast organoids, have been developed. Trophoblast organoids still lack blood vessels, stroma cells and Hofbauer cells compared to in vivo chorionic villi, but in the future, innovative bioengineering systems will be used to create chorionic villi-like models or blood-placental barrier-like models (Takebe and Wells, 2019). In addition, crosstalk between the endometrium and trophoblasts will be elucidated using endometrial organoids (Boretto et al., 2017; Turco et al., 2017; Ezashi et al., 2019).

Owing to the advent of high throughput technology, scRNA-seq is enabling single-cell transcriptome analysis at high resolution (Wagner et al., 2018), including placental development. Quantifying the transcriptome of individual cells can lead to the discovery new cell types, reveal cellular heterogeneity (Islam et al., 2014) of the developmental trajectory (Cheng et al., 2019), identify gene regulatory mechanisms and dissect cell–cell interactions. In the past several years, scRNA-seq has been performed with trophoblasts at the pre-implantation stage (Yan et al., 2013; Blakeley et al., 2015; Petropoulos et al., 2016) and peri-implantation stage with in vitro culture systems to study the human embryo (Zhou et al., 2019; ,Xiang et al., 2020) and post-implantation stage (Liu et al., 2018; Vento-Tormo et al., 2018). These data identified new maternal–fetal interactions and epigenetic programmes. In this way, scRNA-seq can be a powerful tool for the study of early embryogenesis and reproductive medicine.

Other new methods for placental research, such as single-cell proteomics (Su et al., 2017; Lundberg and Borner, 2019; Marx, 2019) and single-cell epigenetics (Clark et al., 2016; Kelsey et al., 2017; Zhou et al., 2019), are also progressing. Differences in gene expression among individual cells can be controlled by epigenetic diversity. Single-cell epigenome methods provide high resolution profiling of DNA modifications, such as transcription factor binding, histone modifications and DNA accessibility (Kelsey et al., 2017). These new techniques can reveal trophoblast cell-fate decisions, identity and function in normal placental development (Kelsey et al., 2017). Additionally, single-cell western blotting can identify cell-to-cell variations in protein-mediated cell functions (Hughes et al., 2014).

Conclusion

Cell biological research of the human placenta began five decades ago with the use of choriocarcinoma and immortalised cell lines (King et al., 2000; Shiverick et al., 2001). However, these cell lines have transcriptome discrepancy and poor differentiation potency compared to in vivo CT (Bilban et al., 2010), thus demanding other cells lines. To evaluate cell lines, new criteria for first-trimester trophoblast were made (Lee et al., 2016). In addition, more molecular markers, epigenetic features and morphological appearance of trophoblast subtypes are also required when selecting new cell lines (Gamage et al., 2018a,b). In parallel, it is important to recognise species differences in various features, such as placental structure, early embryonic shape, implantation type, trophoblast type and gene expression. This review deals with basic knowledge of various aspects of placental development. Compared to other organs, the placenta exhibits great biodiversity among species, but also common features across species. Although there are ethical and technical constraints, continued efforts to elucidate human placental development in normal and abnormal pregnancies are encouraged.

Acknowledgements

The authors thank Dr. Mark Anthony Hill (Anatomy, School of Medical Sciences, Faculty of Medicine, UNSW, Australia) for permission to reuse images of Carnegie stages. The authors thank Dr. Yasuhiro Takashima (Department of Life Science Frontiers, Center for iPS Cell Research & Application, Kyoto University, Japan) for helpful discussions. The authors also thank Dr. Peter Karagiannis for English language editing and are grateful to Dr. Misaki Ouchida for the science illustrations.

Authors’ roles

S.I. and S.Y. contributed to the conceptual planning of the manuscript and drafted the manuscript. S.I. designed the figures and tables. Y.C., K.K., E.K., M.M. and S.Y. edited the manuscript.

Funding

This work was supported by ISHIZUE 2019 of Kyoto University Research Development Program (to S.Y.), Grant-in-Aid for JSPS Research Fellow (Grant numbers: 19J13503), the Smoking Research Foundation (to S.I.) and Fujiwara Memorial Foundation (to S.I.).

Conflict of interest

The authors declare no conflicts of interest.

References