Pluripotent stem cells for the study of early human embryology

Abstract Forty years have passed since the first pluripotent stem cells (PSCs), mouse embryonic stem cells (ESCs), were established. Since then, several PSCs have been reported, including human ESCs in 1998, mouse epiblast stem cells (EpiSCs) in 2007, induced PSCs (iPSCs) in 2006 and 2007, and naïve human PSCs in 2014. Naïve PSCs are thought to correspond to pre‐implantation epiblast cells, whereas conventional (or primed) human PSCs correspond to post‐implantation epiblast cells. Thus, naïve and primed PSCs are classified by their developmental stages and have stage‐specific characteristics, despite sharing the common feature of pluripotency. In this review, we discuss the current status of PSCs and their use to model human peri‐implantation development.


| INTRODUC TI ON
The establishments of mouse and human embryonic stem cells (ESCs) were separated by more than 15 years (Evans & Kaufman, 1981;Martin, 1981;Thomson et al., 1998). This lag was primarily due to the fact that the culture condition for mouse ESCs did not work for human ESCs. Specifically, leukemia inhibitory factor (LIF), a key factor for the self-renewal of mouse ESCs (Smith et al., 1988;Williams et al., 1988), does not provide a sufficient signal to maintain human ESCs. Instead, FGF and TGFβ/ACTIVIN signaling were found essential to maintain human pluripotency (Ludwig et al., 2006;Vallier et al., 2005). Since the founding of ESCs, other major pluripotent stem cells (PSCs) reported included induced pluripotent stem (iPS) cells (Takahashi & Yamanaka, 2006) and mouse epiblast stem cells (EpiSCs) (Brons et al., 2007;Tesar et al., 2007). Notably, like human PSCs, mouse EpiSCs depend on FGF and ACTIVIN. They also correspond to post-implantation epiblast. Other similarities between mouse EpiSCs and human PSCs include a flat shape morphology, metabolism, signaling pathways, and epigenetic patterns (Table 1).
They also share the same pluripotency, primed pluripotency, whereas mouse ESCs exhibit naïve pluripotency. Naïve human PSCs were finally reported in 2014 (Takashima et al., 2014;Theunissen et al., 2014). Much of the knowledge gained from mouse ESCs was used to establish culture systems that maintain naïve human PSCs. In particular, the establishment of two-inhibitor (2i) medium , which contains a MEK inhibitor, PD0325901 (PD03), and a GSK3 inhibitor, CHIR99021 (CHIR), enabled the establishment of ESCs from non-obese diabetic (NOD) mice ) and other rodents, such as rats (Buehr et al., 2008;Li et al., 2008), which cannot be established using the traditional serum + LIF medium. Notably, 2i medium was fundamental for establishing the original naïve human PSCs. Since naïve human PSCs have the characteristics of the pre-implantation epiblast (Nakamura et al., 2016;Stirparo et al., 2018;Takashima et al., 2014;Theunissen et al., 2014), they are viewed as models of human peri-implantation development in vitro.
Ideally, human embryos would be used to study human periimplantation development, but access to them is limited due to ethical reasons. Furthermore, analysis of embryonic development at the post-implantation stage is forbidden, as this stage occurs in utero.
Overall, very few reports offer histological data (Hertig & Rock, 1949;O'Rahilly & Muller, 1987); instead, mouse models are commonly used. However, single-cell RNA sequencing (scRNA-seq) data have shown that the pre-implantation embryos of humans and mice exhibit species-specific gene expression patterns (Blakeley et al., 2015;Boroviak et al., 2018). To understand the post-implantation stages of development in humans, several technological approaches have been attempted. In this review, we examine the methods available for analyzing both the pre-and post-implantation stages of development in humans. In particular, we describe long-term ex vivo embryo cultures for mice, non-human primates, and humans and the construction of ESC-derived spheroids using aggregates of PSCs with or without extraembryonic lineage cells. We also discuss naïve human PSCs and their potential to analyze human embryonic development.

| E S TAB LIS HMENT OF NAÏVE HUMAN PSC s
Research has revealed that there are several pluripotent states.
The first PSCs reported, mouse ESCs, have naïve pluripotency and resemble pre-implantation epiblast. On the other hand, mouse EpiSCs have primed pluripotency and represent post-implantation epiblast. Consistently, these cells were derived from pre-and postimplantation embryonic cells, respectively ( Figure 1a)  Mouse EpiSCs can be reset to naïve PSCs by the overexpression of a single transcription factor, such as Klf2, Klf4,Nr5a1,Nr5a2,Nanog,or Esrrb (Figure 1a) (Festuccia et al., 2012;Guo & Smith, 2010;Guo et al., 2009;Hall et al., 2009;Silva et al., 2009), followed by culture in 2i medium . These findings suggested that primed human PSCs could also be reset to naïve pluripotency. The 2i medium is a serum-free medium containing PD03 to inhibit FGF signaling and CHIR to activate Wnt signaling and inhibit GSK3 function . Thus, 2i medium inhibits differentiation cues and induces pluripotent genes. These findings suggest that multiple signaling pathways are involved in maintaining naïve pluripotency.
In addition, t2iLGö and 5iLA media can be used to establish naïve ESCs directly from human pre-implantation blastocysts Theunissen et al., 2014). These two media commonly contain a MEK inhibitor (PD03) and Wnt signal activator (CHIR or IM-12). Interestingly, the removal of the Wnt signal activator did not cause a significant reduction in OCT4 distal enhancer activity or KLF4 expression compared to removal of the MEK inhibitor (Theunissen et al., 2014). In addition, lower concentrations of CHIR in t2iLGö medium induced a more homogenous colony morphology than did lower concentrations in the conventional t2iLGö medium (Guo et al., 2017). Together, these findings indicated that the activation of Wnt signaling may not be required for establishing naïve human PSCs. More recently, PXGL medium, which replaces CHIR in t2iLGö medium with a tankyrase inhibitor, XAV939 (XAV), to suppress Wnt signaling, was found to maintain naïve pluripotency ( Overexpression of transgenes, such as NANOG and KLF2, is another way to robustly induce naïve PSCs (Table 2) (Takashima et al., 2014).
Overexpressing KLF4 only in t2iLGö medium can also reset human primed PSCs to naïve PSCs (Liu et al., 2017). Alternatively, chemicalonly induction methods for resetting are available (Table 2). Austin Smith's group reset primed human PSCs by culturing them in PDLIF/ HDACi (PD03, LIF, and histone deacetylase inhibitor (HDACi)) for 3 days and then in PXGL (Guo et al., 2017). The addition of HDACi enables the induction of naive human PSCs without any forced expression of transcription factors. However, using the 5iLA medium is the simplest method because primed human PSCs can be reset to naïve PSCs in 5iLA without any medium change (Theunissen et al., 2014). However, karyotypic abnormalities are more likely to occur in 5iLA medium than in t2iLGö or PXGL medium (Guo et al., 2017;Liu et al., 2017). Such karyotypic abnormalities can be reduced by a lower PD03 concentration and alternative MEK inhibitors (Di Stefano et al., 2018).

| IN VITRO CULTURE S OF MOUS E AND PRIMATE EMB RYOS TOWARD G A S TRUL ATI ON
Pre-implantation embryos can be studied relatively easily as far as they can be cultured in vitro. However, the analysis of embryonic development after implantation into the uterus is difficult even in animal models. Therefore, research on in vitro cultures beyond the implantation stage is being conducted. To reproduce implantation in vitro, maternal tissues or alternative materials and culture media for maintaining the embryos beyond implantation are required. Several groups have succeeded in performing 2D embryo cultures until the egg-cylinder stage in vitro using medium supplemented with serum from calf, rat, or human placental cord blood (Hsu, 1972;Masaki et al., 2015;Morris et al., 2012;Wu et al., 1981). Mouse embryos can grow on a non-coated plastic plate or on a plate coated with a matrix, such as collagen, fibronectin, or inactivated feeder cells.
In 2014, a new protocol for in vitro embryo culture was developed, which allowed the observation of mouse embryo development from E3.5 to post-implantation epiblast with an egg-cylinder-like structure (Table 3) (Bedzhov et al., 2014). This protocol is based on two media, IVC1 and IVC2, which could be used to culture embryos from humans and cynomolgus monkeys as well as mice (Table 3

| RECON S TITUTI ON OF THE EMB RYO -LIKE S TRUC TURE US ING PLURIP OTENT S TEM CELL S
Although in vitro embryo cultures provide an advanced method to study peri-implantation development, they require a large number of human and non-human primate embryos. Compared to mice, it is difficult to prepare a sufficient number of high-quality embryos from humans and primates. Moreover, to analyze genetic functions in early embryogenesis, gene knockout is often required. Although genome editing techniques, such as CRISPR, can be applied directly to fertilized primate eggs to induce gene knockouts (Fogarty et al., 2017;Sato et al., 2016), a large number of embryos are required to obtain accurately genome-edited embryos in addition to the issue of mosaicism. Therefore, unless genome editing improves in terms of accuracy, efficiency, and homogeneity, other methods are needed.
Mouse ESCs have the ability to generate mice composed only of ESC-derived somatic cells using tetraploid embryos (Nagy et al., 1990;Wang et al., 1997). This ability suggests that it may be possible to form an embryo-like structure from ESCs in vitro. Such a structure must be capable of differentiating into the three germ

| CON CLUS IONS
Early developmental studies on humans are limited because of ethical issues. In particular, embryos beyond the implantation stage are not readily accessible. As alternatives, non-human primate embryos and reconstituted embryos using PSCs are serving as accurate representations of human embryo. In this review, we summarized the current status of mouse and human PSCs, with particular focus on pre-implantation naïve PSCs to post-implantation primed PSCs. The formative state is midway between naïve and primed PSCs and bestows the capacity to differentiate into the three germ layers and germ cells. The latest development regarding PSC research for the study of the embryo is the preparation of embryo-like structures.
This approach may enable the best ethically sound study of human post-implantation embryos. Ultimately, all these efforts are being made to understand the mechanisms of early embryonic development, which defines the growth of the entire body.

F I G U R E 4
Comparison of natural embryos, blastoids, ETS-embryos, and ETX-embryos. Diagram of mouse development from zygote to post-implantation stages (above) and synthetic embryos corresponding to each of those stages (below).

ACK N OWLED G M ENTS
The authors thank all Takashima Lab members for helpful discussions and assistance. We also thank Peter Karagiannis for reading the manuscript. This work was supported by MEXT KAKENHI (grant number 16H02465, 20H05762 to Y. T) and AMED (grant number 20bm0104001h0008, 20bm0704035h0002 to Y. T).