The road to generating transplantable organs: from blastocyst complementation to interspecies chimeras

Over the past half-century, laboratory-made chimeras have been widely used for developmental biology studies. Chimeras generated within the same species (intraspecies chimeras) have proven to be useful experimental models to determine the potency of cultured stem cells, to study gene function during development, and to generate transgenic animal models of human diseases. Chimeras generated between different species (interspecies chimeras), on the other hand, have helped researchers understand evolutionarily conserved and divergent developmental mechanisms. More recently, these studies have generated hope for solving the worldwide shortage of organs for transplant via the approach of interspecies blastocyst complementation (Rashid et al., 2014).

Blastocyst complementation, first developed by Alt and colleagues to study specific gene function in lymphocytes (Chen et al., 1993b), is a technique based on injecting wild-type (WT) pluripotent stem cells (PSCs) into a mutant blastocyst to form a chimera. As the chimera develops, the WT donor cells complement the developmental defect(s) in the mutant embryo. If the gene mutation(s) in the host embryo prevents the development of a particular organ, WT cells from the same or a different species can fill in and generate an organ mostly composed of donor cells.

Since its adaptation for organogenesis by Nakauchi and colleagues (Kobayashi et al., 2010), blastocyst complementation has gained popularity in regenerative medicine (Wu et al., 2016) (Fig. 1). Indeed, the last decade has seen many studies attempting to generate functional organs in vivo using this technique. However, it remains challenging to achieve functional organ complementation between different species owing to developmental incompatibilities, and it will be necessary to study and overcome these xenogeneic barriers before interspecies blastocyst complementation can come into broader use. In this Review, we first discuss the chimera competency of various types of donor cells. We then summarize studies on intra- and inter-species blastocyst complementation, highlighting how they have been used in an attempt to generate various tissue and organs. Finally, we provide a brief account of xenogeneic barriers and conclude with an outlook of the field.

The ability of donor cells to robustly contribute to chimera formation in host embryos is a prerequisite for successful blastocyst complementation. Early embryonic cells, e.g. cells from the inner cell mass (ICM), meet this criterion and, before the derivation of embryonic stem cells (ESCs), were often used as donor cells (Gardner and Johnson, 1973; Mystkowska, 1975; Rossant and Frels, 1980). These early studies lent support to the notion that evolutionary distance is negatively correlated with donor cell chimerism in the host species. For example, when ICM cells from the ryukyu mouse [Mus caroli, which diverged from Mus musculus about 7.4 million years ago (MYA)] were injected into mouse blastocysts, extensive contribution of ryukyu cells to the resultant mice was observed (Rossant and Frels, 1980). In contrast, when ICM cells from the rat (which diverged from mice 20.9 MYA) were injected into mouse blastocysts, they gave rise to limited patches in different germ layers (Gardner and Johnson, 1973). In addition, embryonic cells from the more distant bank vole (Clethrionomys glareol, which diverged from mice 33 MYA) failed to show a detectable chimeric contribution beyond embryonic day (E) 9 in mice (Mystkowska, 1975). The same strategy has also been successfully applied for generating viable chimeras between other mammalian species that are closely related, including sheep and goat (10 MYA) (Fehilly et al., 1984) and two bovine species (Williams et al., 1990). Specifically, Williams et al. showed that, despite the phenotypic differences between the species Bos indicus and Bos taurus, their close genetic relationship (diverged <1 MYA) allowed for very high rates of chimeric calves (63% of all full-term pregnancies) to be born. Overall, although effective, early embryonic cells are limited in quantity and are technically difficult to manipulate, which initially limited the scope and utility of chimera research.

The initial derivation of mouse embryonic stem cells (mESCs) and the subsequent demonstration of their germline chimera and gene-targeting capabilities revolutionized modern biology (Evans and Kaufman, 1981; Martin, 1981; Thomas and Capecchi, 1987). ESCs provide an unlimited supply of cells functionally indistinguishable from epiblast cells within the ICM, and thus constitute ideal donor cells for chimera research. Following the initial success in mice, a great deal of effort has been dedicated to deriving chimera-competent ESCs from other species but this has had limited success. Although a variety of PSCs have been generated from different sources and species, most of them failed to pass the standard pluripotency test, e.g. the ability to form teratomas, let alone chimeras. To date, convincing results for chimera formation and germline transmission are only available for mouse and rat PSCs. This disappointing outcome is largely attributed to our poor understanding of pluripotency in species other than mice and highlights the need for continued research.

Several seminal studies in the 2000s significantly expanded our knowledge of pluripotency and contributed to the greatly expanded collection of PSCs. A milestone discovery was the generation of induced pluripotent stem cells (iPSCs) from fibroblasts (Takahashi and Yamanaka, 2006). Since then, iPSC technology has not only transformed regenerative and personalized medicine, but has also made it possible to obtain PSCs from species in which early embryos are not readily accessible (Stanton et al., 2019). Another important discovery was that the self-renewal of ‘ground state’ mESCs could be enabled by two inhibitors (2i) (Ying et al., 2008). Ground state culture liberated mESC derivation from a few permissive strains and supported de novo derivation of germline-competent rat ESCs (Buehr et al., 2008; Li et al., 2008). Finally, the derivation of epiblast stem cells (EpiSCs) from post-implantation mouse and rat embryos (Brons et al., 2007; Tesar et al., 2007) led to the conceptualization of naïve and primed pluripotency phases in vivo (Nichols and Smith, 2009), and the stabilization of distinct pluripotency states in cultured PSCs (Wu and Izpisua Belmonte, 2015). mESCs mirror ‘naïve’ epiblast cells from a late blastocyst, whereas mouse EpiSCs (mEpiSCs) closely resemble peri-gastrulation epiblast cells and are ‘primed’ for differentiation. Chimera and germline competency are widely regarded as functional features of naïve but not primed pluripotency (Fig. 2), which helps explain the failure of PSCs cultured in primed condition(s) to contribute to blastocyst chimeras (Wu and Izpisua Belmonte, 2015).

The derivation of mEpiSCs also helped resolve why there are noticeable differences between mouse and human ESCs (hESCs) despite both being sourced from preimplantation blastocysts. Distinct from dome-shaped mESCs and similar to mEpiSCs, hESCs show flattened colony morphology and depend on ACTIVIN/NODAL and FGF (Vallier, 2005) rather than BMP and LIF (Ying et al., 2003) for self-renewal. It is now commonly accepted that hESCs represent the counterpart of mEpiSCs and reside in the developmentally more advanced primed pluripotency state (Wu and Izpisua Belmonte, 2015). With this realization, researchers have been actively searching for conditions to stabilize human naïve pluripotency. To date, more than a dozen human naïve conditions, with a wide range of media compositions, have been reported (Gafni et al., 2013; Guo et al., 2016; Takashima et al., 2014; Theunissen et al., 2014; Ware et al., 2014; Weinberger et al., 2016; Wu and Izpisua Belmonte, 2015). Owing to the lack of a functional test for human naïve pluripotency, a set of molecular criteria has been proposed (Theunissen et al., 2016). Based on these molecular criteria, hPSCs grown in two conditions, 5iLA (Theunissen et al., 2014) and T2iLGöY (Takashima et al., 2014), most closely resemble pre-implantation human epiblast cells (Guo et al., 2017). Interestingly, several recent reports demonstrated the plasticity of naïve hPSCs toward trophoblast and hypoblast lineages, a feature likely retained in epiblast cells of human blastocysts (Cinkornpumin et al., 2020; Dong et al., 2020; Guo et al., 2021; Linneberg-Agerholm et al., 2019; Io et al., 2021). Strikingly, naïve hPSCs’ embryonic and extra-embryonic dual-competency also enabled the generation of human blastocyst-like structures (Yu et al., 2021a). Although there is controversy around reports of naïve hPSCs contributing to mouse chimeras (Gafni et al., 2013; Bayerl et al., 2021; Hu et al., 2020; Taei et al., 2020; Masaki et al., 2015; Theunissen et al., 2016), a recent study reported limited proliferation of naïve hPSCs within cultured monkey blastocysts (Aksoy et al., 2021). Of note is the observation that monkey naïve PSCs also inefficiently colonize monkey embryos, with the cells differentiating prematurely following injection (Aksoy et al., 2021). These results may reflect differences in naïve pluripotency between rodents and primates and/or suboptimal conditions for human/monkey naïve PSCs. Indeed, differentially methylated regions within imprinted loci have been reported to be lost in 5iLAF and T2iLGöY cultured hPSCs (Pastor et al., 2016; Theunissen et al., 2016).

Chimera competency is not exclusive to naïve pluripotency. A number of recent studies have reported intermediate PSCs showing features of both naïve and primed pluripotency, with some demonstrating chimera and germline competency (Bao et al., 2018; Chang and Li, 2013; Cornacchia et al., 2019; Du et al., 2018; Han et al., 2010; Neagu et al., 2020; Tsukiyama and Ohinata, 2014; Yu et al., 2021b; Cornacchia et al., 2019). Several concepts have emerged to define these intermediate pluripotency states, among which ‘formative’ pluripotency has become most popular (Smith, 2017). Formative pluripotency is postulated to represent an interval between naïve and primed pluripotency during which epiblast cells acquire competence for multi-lineage induction (Smith, 2017). Functionally, formative pluripotency is characterized by dual competence for chimera formation and primordial germ cell (PGC) specification (Fig. 2), which are properties of E5-6 mouse epiblasts (Gardner et al., 1985; Ohinata et al., 2009). Stable PSCs with formative features were most recently reported by the Austin Smith lab and ours (Kinoshita et al., 2020; Yu et al., 2021b). The Smith lab generated mouse and human formative PSCs (designated as ‘FS cells’) using a low concentration of ACTIVIN-A, the canonical WNT pathway inhibitor XAV939 and a pan-retinoic acid receptor inverse agonist (RARi, BMS493) (Kinoshita et al., 2020). In contrast, we used factors activating the FGF, TGF-β and WNT/β-catenin pathways to generate mouse, horse and human cells termed ‘XPSCs’ (Yu et al., 2021b). Both FS cells and XPSCs demonstrated competence for PGC specification, chimera formation, and transcriptome similarity to formative epiblasts. Pera and colleagues also found that formative-like cells exist as a minority population in primed hPSC cultures (Lau et al., 2020). Interestingly, the transcriptomes of hESCs grown in several reported human naïve conditions (Gafni et al., 2013; Hu et al., 2020; Irie et al., 2015; Sperber et al., 2015) resembled those of formative E8-E10 human epiblasts, suggesting that these cells exist in intermediate/formative rather than naïve pluripotency states (Guo et al., 2017; Huang et al., 2014; Nakamura et al., 2016; Pastor et al., 2016). Future studies are needed to determine the relationship between these intermediate PSCs. It should be noted that, despite tremendous progress, there is considerable uncertainty regarding the developmental status of various types of human PSCs.

Intriguingly, horse and human XPSCs can contribute to interspecies chimera formation in early mouse and pig embryos, respectively (Wu et al., 2017a; Yu et al., 2021b). In agreement, two neonatal interspecies chimeric piglets were recently obtained using cynomolgus monkey ESCs cultured in a condition similar to that used to generate XPSCs (Fu et al., 2019). These results raise the intriguing possibility that formative/intermediate PSCs are more capable of contributing to interspecies chimeras than other reported PSCs, and it is tempting to exploit the unique properties of these new PSCs for interspecies blastocyst complementation in future studies.

In addition to pluripotent cells, lineage-restricted stem and progenitor cells, e.g. human hematopoietic progenitors, neural crest cells and glial progenitors, have also been used for generating interspecies chimeras by in utero injection (Fujiki et al., 2003; Cohen et al., 2016; Han et al., 2013). Combining animal organotypic cultures with human stem or progenitor cells may provide an alternative means to promote differentiation and/or maturation of cells into functional human tissues ex vivo. As advances in bioengineering technologies enable the growth of larger and more complex structures in vitro from developing tissue, this might present another approach for generating chimeric organs that does not require an embryonic chimera-competent cell.

In summary, forty years have passed since the first discovery of mouse ESCs and, although we have made tremendous progress in generating intra- and interspecies chimera-competent rodent PSCs, much work lies ahead to extend these findings to other species, including humans.

Early blastocyst complementation studies focused on understanding gene function during mouse development (Table 1). Blastocyst complementation was first reported by Chen et al., who created mice deficient in the Rag2 gene necessary for producing functional lymphocytes (Chen et al., 1993b). When blastocysts from Rag2-deficient mice were injected with WT mESCs, the resulting chimeras contained mature B and T lymphocytes exclusively derived from the donor cells. In contrast, when mESCs with a deletion in the IgJ_H locus were injected, donor cells could only generate T but not B lymphocytes. Following this initial success, several other reports used this technique for studying the function of genes [e.g. Rb (also known as Rb1) and Gata2] in the hematopoietic system (Chen et al., 1993a; Tsai et al., 1994). In the following years, blastocyst complementation began to be adopted outside of the hematopoietic system. For example, Liégeois et al. complemented a defect in the aphakia mouse strain (ak), which fails to develop an ocular lens due to a Pitx3 mutation (Liégeois et al., 1996). The authors injected WT mESCs into ak/ak blastocysts and produced chimeras with normal mESC-derived lenses. In contrast, complementation with Rb-deficient mESCs generated an aberrant lens phenotype. Fraidenraich et al. reported that compound Id1/Id3 knockout mouse embryos displayed multiple cardiac defects, but mid-gestation lethality could be rescued by WT mESCs. The rescue seemed to result from non-cell autonomous effects, and in part could be attributed to IGF1 secreted by donor cells (Fraidenraich et al., 2004). Müller et al. generated functional thymic epithelial cells (TECs) from WT mESCs in Foxn1-null mice and revealed a crucial role for VEGFA in TECs (Müller et al., 2005). Bort et al. injected WT embryos with Hex (Hhex) mutant mESCs and showed that mutant mESCs were not able to contribute to liver development (Bort et al., 2006). In 2007, Stanger et al. used blastocyst complementation to study pancreas size control (Stanger et al., 2007). In this study, blastocysts with homozygous mutations in Pdx1, a gene that is crucial for pancreas development (Jonsson et al., 1994; Offield et al., 1996), were complemented by injection of WT mESCs, generating an entire pancreatic epithelium composed of donor cells.

Although Stanger et al. laid the technical foundation, it was not until 2010 that Nakauchi and colleagues developed and popularized the concept of in vivo organogenesis based on blastocyst complementation (Kobayashi et al., 2010). Similar to Stanger et al., Kobayashi et al. performed Pdx1-null complementation with WT mESCs/iPSCs, demonstrating that all exocrine and endocrine tissues were derived from donor mPSCs; however, pancreatic stromal elements (vessels, nerves and fibrocytes) contained both host and donor cells. They further demonstrated that mPSC-derived islets were functional and could correct hyperglycemia in a diabetic mouse model. In a later study, Yamaguchi et al. successfully complemented Pdx1-null rat embryos with WT rat PSCs (rPSCs) (Yamaguchi et al., 2017). Following these successes with the pancreas, researchers sought to use intraspecies blastocyst complementation for generating other organs and tissues.

Usui et al. generated mPSC-derived kidney tissue in Sall1-null mice (Usui et al., 2012). During development, kidneys develop from the interaction between the ureteric bud (UB) and the metanephric mesenchyme (MM) (Takasato and Little, 2015). Sall1 is primarily expressed in the MM, and homozygous Sall1 knockout mice lack kidneys and die shortly after birth (Nishinakamura et al., 2001). WT mESCs/iPSCs were used to complement Sall1-mutant mouse embryos, and the resulting chimeras were born phenotypically normal with well-developed kidneys capable of producing urine. Kidney stromal elements such as vessels and nerves were a mixture of host and donor cells. The collecting duct epithelium, which is derived from UB, was also chimeric. However, no pups with mPSC-derived kidneys survived to adulthood, likely owing to a poor feeding response caused by the anosmic phenotype (olfactory bulb hypogenesis) of Sall1 mutant mice that was not rescued by WT mPSCs. This study underscores the importance of the choice of target gene(s) to disrupt for blastocyst complementation. Genes that are crucial for the development of all aspects of the target organ without significantly affecting other aspects of development are preferable. In this regard, it will be interesting to test the complementation of Pax2/Pax8 double-knockout mouse or rat embryos, in which the intermediate mesoderm (which gives rise to the UB and MM) does not undergo the mesenchymal-epithelial transitions required for nephric duct formation (Bouchard et al., 2002; Kobayashi et al., 2021). Perhaps complementation of this defect at an earlier stage of kidney development will lead to a more completely donor cell-derived kidney.

Mori et al. disabled lung development using conditional Ctnnb1 or Fgfr2 knockout, which rendered mouse embryos unable to specify or expand early respiratory endoderm progenitors, respectively. They complemented these mutant embryos with WT mPSCs, and the resulting chimeras survived into adulthood and had lungs functionally indistinguishable from those of WT littermates (Mori et al., 2019). Donor contribution to the lung tissue was found to be consistently high in the case of epithelial cells, but variable for endothelial and mesenchymal cells. Another study that reported the complementation of lung tissue by mESCs disabled lung formation in host embryos by mutating Fgf10, which binds to Fgfr2 during lung progenitor cell expansion (Kitahara et al., 2020). In addition to regulating epithelial proliferation and lung branching morphogenesis, FGF10 plays essential roles in multiple mesenchymal lineages during lung development. Consequently, not only the epithelial cells, but also the interstitial cells of the lung predominantly originated from donor mESCs. These results suggest that Fgf10 deficiency empties a more complete organ developmental niche. In another study, Wen et al. injected WT mESCs into blastocysts mutant for Nkx2.1 (Nkx2-1), which is crucial for both lung and thyroid development (Wen et al., 2020). Although the resulting chimeras contained thyroid and lung tissues mostly derived from the donor cells, they died shortly after birth due to the presence of tracheoesophageal fusion.

Emx1 is expressed in progenitors of the dorsal telencephalon that eventually develop into the cerebral cortex and olfactory bulbs (Simeone et al., 1992), but mutating Emx1 alone is not enough to prevent forebrain development (Yoshida et al., 1997). Chang et al. induced targeted ablation of the dorsal telencephalon via inducible expression of diphtheria toxin A (DTA) under the control of the Emx1 promoter to open a niche for donor cells to proliferate and create an entire donor mESC-derived forebrain (Chang et al., 2018). This approach highlights how cell ablation can help create an emptied developmental niche in tissues for which key lineage differentiation gene(s) have not yet been identified.

Miura et al. used a triple-targeting CRISPR strategy to knock out Nanos3 in mouse zygotes, then performed complementation with WT mESCs (Miura et al., 2020), which resulted in the generation of fertile chimeras with spermatozoa fully derived from donor cells. Kobayashi et al. used rat blastocysts mutant for Prdm14, which is crucial for germline development in both mouse and rat (Yamaji et al., 2008; Kobayashi et al., 2020), and complemented the germline defect with WT rPSCs. This led to the generation of fertile chimeric rats, the F1 offspring of which displayed the coat color of the original donor rPSCs (Kobayashi et al., 2021).

One issue with most blastocyst complementation studies is that, although the parenchyma can be complemented by donor cells, the endothelium and stroma are still populated with a mixture of host and donor cells. The presence of host vascular endothelial cells is particularly problematic, as these cells can elicit hyperacute rejection following transplantation. To address this limitation, a recent study emptied the endothelial niche in mice by mutating the Flk1 gene (Hamanaka et al., 2018). Flk1 (Kdr, Vegfr2), is a VEGF receptor involved in early embryonic vasculogenesis, and is necessary for the development of both endothelial and hematopoietic lineages (Shalaby et al., 1995, 1997). Flk1-mutant mouse embryos do not survive past E9.5, but WT mESCs were able to fully rescue the embryonic lethality phenotype (Hamanaka et al., 2018). The rescued chimeras could survive to adulthood with all vascular endothelial cells and hematopoietic cells derived from the donor cells. Similarly, Wang et al. complemented Flk1-null mouse embryos with WT mESCs, and showed that all the CD31+ endothelial cells in the blood vessels were derived from the donor cells (Wang et al., 2020). Most recently, Das et al. targeted the gene Etv2, a master regulator of hematoendothelial development (Das et al., 2020). Etv2-null mouse embryos die early in gestation and completely lack hematoendothelial lineages (Lee et al., 2008). By using WT mESCs, Das et al. were able to rescue the Etv2-null mouse phenotype and generate hematoendothelial lineages derived exclusively from the donor cells (Das et al., 2020).

Even though the focus of blastocyst complementation has shifted to interspecies organogenesis, intraspecies blastocyst complementation remains a useful part of the developmental biology toolbox and serves as a necessary control for interspecies studies. Notwithstanding the progress made, however, there are still several limitations that need to be overcome in order to create a more robust blastocyst complementation platform for human organogenesis. Goals to work toward include the full complementation of the target organ including supporting tissues, as well as the limitation of donor cell contribution to undesired lineages. As mentioned above, the organs resulting from blastocyst complementation are primarily composed of donor cells, but some supporting tissues such as stroma and endothelium still contain host cells. It seems likely that simultaneous knockout of multiple genes will be needed to enable the generation of a completely donor-derived organ. In addition to reducing the contribution of host cells to the target organ, the successful application of blastocyst complementation depends on the reciprocal ability to limit donor cell contribution to organs outside of the target. This would address one of the primary ethical concerns about interspecies organogenesis and chimera formation with human cells, which is the possibility of unwanted contributions to the brain and/or germline. There have been some efforts to help address these concerns. For example, Hashimoto et al. created mouse chimeras in which donor mPSCs harbored null mutations in both the Otx2 and Prdm14 genes that rendered them capable of normal development in most organs, but unable to contribute to neural and germ cell lineages (Hashimoto et al., 2019). In addition, Kobayashi et al. discovered that forced expression of the Mixl1 gene in donor mPSCs limited their contribution to only endodermal lineages, which avoided contribution to neurons and germ cells (Kobayashi et al., 2015).

Blastocyst complementation has also been applied to large livestock species, including pigs and cattle. Owing to the lack of chimera-competent PSCs from these species, WT blastomeres from cleavage-stage embryos are used as donors instead. As a first proof of concept, Matsunari et al. produced pancreatogenesis-disabled pig fetuses by introducing a PDX1-HES1 (hairy and enhancer of split-1 gene driven by PDX1 promoter) transgene into in vitro-matured pig oocytes (Matsunari et al., 2013). Primary cultures of fibroblasts were established from these fetuses and used as nuclear donors for somatic cell nuclear transfer (SCNT). The chimeric pigs generated by morula injection of WT blastomeres showed contribution of donor cells throughout the entire body, developed fully functional pancreata and grew into fertile adults (Matsunari et al., 2013). Following this success, Zhang et al. used WT blastomeres to complement an eye defect caused by mutation of the MITF gene. MITF-null pigs were completely anopthalmic, but complemented embryos exhibited normal eye development (Zhang et al., 2018). Almost all the retinal pigment epithelial cells and corneal epithelial cells of the chimeric piglets were derived from donor cells. More recently, Das et al. complemented ETV2-null pig embryos with WT blastomeres (Das et al., 2020). ETV2-null pig embryos lack hematoendothelial lineages. Following morula injection, WT blastomeres rescued the lethality of ETV2-null embryos and generated all endothelial and hematopoietic lineages in full-term piglets. In a tour de force study, Matsunari et al. systematically evaluated blastocyst complementation for several organs and tissues, including the pancreas, vasculature, liver and kidney, using allogenic WT blastomeres (Matsunari et al., 2020). For the pancreas, they generated PDX1-null pig embryos that have an apancreatic phenotype similar to that of Pdx1-null mice and rats (Kang et al., 2017; Wu et al., 2017b). They also tested complementation of both the pancreas and the vasculature in PDX1/KDR double-knockout embryos. Remarkably, the single full-term chimera had a well-developed pancreas, endothelial tissue and hematopoietic cells derived from the donor cells. They then turned their attention to the kidney, attempting a similar SALL1 knockout strategy as had been performed previously in mice and rats. SALL1-null pig fetuses showed an anephric phenotype but seemingly died earlier (mid-gestation) than Sall1-null mice and rats (neonatal). Complementation of SALL1-null embryos with WT blastomeres resulted in a fetus exhibiting normally developed kidneys. Finally, the authors successfully complemented the ahepatogenic phenotype in hematopoietically expressed homeobox (HHEX)-deficient embryos and generated full-term fetuses with normal-sized livers composed of cells mostly derived from donor blastomeres. Most recently, Maeng et al. produced pig embryos lacking native skeletal muscle by simultaneously inactivating the MYF5, MYOD and MYF6 genes (Maeng et al., 2021). They then used blastomeres to complement the lack of skeletal muscle phenotype in MYF5/MYOD/MYF6-null pigs and generated viable chimeras showing normal histology, morphology and function.

Blastocyst complementation has also been performed in cattle. Ideta et al. generated null mutations in NANOS3 in Wagyu cattle fibroblasts, then performed SCNT to create NANOS3-null embryos. NANOS3-null cattle exhibited a complete loss of germ cells, similar to the phenotype observed in mice (Tsuda, 2003; Ideta et al., 2016). Injection of WT blastomeres into NANOS3-null embryos resulted in the production of chimeric livestock complemented with allogeneic germ cells in their ovaries, although these chimeras were not analyzed beyond the fetal stage.

Despite the progress made, only limited success of blastocyst complementation has been realized in the interspecies context. In retrospect, the success of intraspecies blastocyst complementation seems obvious. In fact, in a technique known as tetraploid complementation, an entire adult organism can be solely derived from donor mESCs (Nagy et al., 1990). However, tetraploid complementation and many forms of organ blastocyst complementation do not work across species, even between closely related rats and mice (Yamaguchi et al., 2018). Therefore, more study is needed to examine the barriers preventing blastocyst complementation from working effectively between species. Below we summarize the current state of interspecies blastocyst complementation research and discuss its levels of success in varying organs (summarized in Table 2).

In a milestone study, Nakauchi and colleagues provided the first proof-of-concept of interspecies blastocyst complementation by generating a mouse-sized pancreas primarily derived from rPSCs in a Pdx1-null mouse (Kobayashi et al., 2010). Importantly, Pdx1-null mice complemented with rPSCs grew into adulthood, maintained normal serum glucose levels and displayed normal insulin secretion in response to glucose loading. Later, the Nakauchi lab also performed the reverse experiment in which mPSCs were used to complement Pdx1-null rat embryos, thereby generating a rat-sized pancreas mostly composed of mouse cells (Yamaguchi et al., 2017). They then isolated and transplanted islets from these pancreata into mice with streptozotocin-induced diabetes. Remarkably, xeno-derived mouse islets normalized and maintained host blood glucose levels for over 1 year without immunosuppression. In a separate study, Wu et al. developed a one-step approach combining CRISPR-Cas9 zygote editing with blastocyst complementation, and independently demonstrated that functional rat pancreatic tissues could be generated in Pdx1-null mice (Wu et al., 2017a). These pioneer rodent studies demonstrated the efficacy of tissues generated via interspecies blastocyst complementation and gave an indication of their therapeutic potential. Despite a number of trials in other organs, however, rat-mouse or mouse-rat pancreas complementation remains the only successful example of generating a fully functional organ cross species to date. With regard to other organs and tissues, there have been varying degrees of success, as discussed below.

Isotani et al. complemented blastocysts from nude mice that lacked thymi with WT rESCs and obtained a thymus predominantly constituted of rat cells. Interestingly, the size of the rESC-derived thymus was much smaller than that of a WT mouse, although it appeared to be functional as CD4⁺ and CD8⁺ single-positive T cells could be detected in these chimeras (Isotani et al., 2011). They also transplanted the xeno-derived rat thymus under the kidney capsule of a nude rat lacking a thymus, showing that rat CD4⁺ and CD8⁺ cells appeared after 8 weeks, further demonstrating functionality. Despite the success, however, detailed characterization and more extensive functional analysis of the rESC-derived thymus is lacking. It also remains unresolved whether the small size of the thymus generated is due to interspecies developmental incompatibility or low levels of rat cell contribution. In this regard, it will be interesting to perform the reverse experiment and generate a mouse thymus in a nude rat for comparison.

Usui et al. found that, unlike mPSCs, rPSCs could not complement the kidney agenesis defect of Sall1-null mice (Usui et al., 2012). This could be attributed to the poor contribution of rPSCs to mouse MM (Goto et al., 2019; Yamaguchi et al., 2018). In contrast, mPSCs could efficiently contribute to rat kidneys, thereby enabling the generation of histologically normal neonatal kidneys in anephric Sall1-null rats, with mouse cells occupying all MM derivatives (Goto et al., 2019). The other compartments in the generated kidneys, however, were comprised of a mixture of mouse and rat cells. Importantly, proper ureter and bladder connections were noted in the mPSC-derived kidneys, suggesting urine secretion ability. Unfortunately, extensive functional tests were not performed as these chimeras died shortly after birth due to their inability to suckle milk. Strategies such as targeting other genes essential for kidney development, conditional knockout or conditional ablation via DTA may be considered to generate fully functional mouse kidneys in rats.

An interesting observation was that the size of the xeno-derived mouse kidneys was smaller than that of rat kidneys and similar to mouse kidneys, which is different from what was observed for the pancreas. In the case of the pancreas, size is controlled by the host species, which is likely as a result of non-cell autonomous control of progenitor cell numbers in the developing pancreatic bud (Stanger et al., 2007; Kobayashi et al., 2010; Wu et al., 2017a; Yamaguchi et al., 2017). For the kidney, by contrast, the resulting size might be determined by donor cell-intrinsic factor(s). Intriguingly, the total number of glomeruli in Sall1-null chimeric rats was similar to that found in mice, whereas the size of glomeruli was more similar to that of control rats. These findings suggest that both cell autonomous and non-cell autonomous mechanisms are at play during interspecies nephrogenesis.

Wu et al. demonstrated that rat eye tissues can be generated in Pax6-null mice (Wu et al., 2017a). Pax6 is a master regulator for eye morphogenesis, and Pax6-null mice lack eye development in addition to other phenotypes (Gehring and Ikeo, 1999). Although restored development of eyes enriched with donor rat cells was obtained in Pax6-null neonatal chimeras, detailed characterization was lacking. The authors also used rPSCs to rescue the severe heart malformation and growth retardation phenotypes caused by Nkx2.5 (Nkx2-5) deficiency in mouse embryos (Wu et al., 2017a). However, even though complementation appeared successful at E10.5, full-term rescue was not achieved.

Isotani et al. demonstrated that it was possible to derive functional rat sperm from rat-mouse chimeras (Isotani et al., 2016). Via the formation of interspecies chimeras, Honda et al. generated gametes using iPSCs from the endangered Ryukyu spiny rat (Tokudaia osimensis) in mice (Honda et al., 2017). Most recently, Kobayashi et al. succeeded in performing germline interspecies blastocyst complementation, which involved the generation of mouse gametes in Prdm14-null rats (Kobayashi et al., 2021). After several intraspecies trials demonstrating proof of concept, the authors injected mESCs carrying a Blimp1 (Prdm1)-GFP reporter into Prdm14-null rat blastocysts. At E9, both mouse and rat PGCs were found in the posterior epiblast, but the rat PGCs died out shortly thereafter owing to the Prdm14 mutation. By E15, the rat gonads were exclusively colonized by mouse PGCs, and in adult chimeras two out of three testes contained mouse-derived germ cells undergoing normal spermatogenesis. The xeno-derived mouse sperm had impaired motility but were able to fertilize mouse oocytes in vitro and produced normal offspring. Germline blastocyst complementation may therefore provide a robust system for the efficient generation of animals with desired traits and for obtaining germ cells from endangered and extinct species.

Although mPSCs could fully complement Flk1-null mouse embryos, only partial rescue has been observed with rPSCs (Hamanaka et al., 2018; Wang et al., 2020). rPSC-derived vasculature and hematopoietic cells could be observed up to E11.5 in Flk1-null mouse embryos, but successful complementation was not obtained beyond E13.5. Similarly, Garry and colleagues demonstrated partial complementation of ETV2-null pig embryos with hPSCs overexpressing the anti-apoptotic factor BCL2 (Das et al., 2020). Remarkably, in E17 ETV2-null embryos, essentially all TIE2+ cells were derived from hPSCs. Most recently, Garry and colleagues succeeded in enriching TP53-null hPSCs in the skeletal muscles of E20 and E27 MYF5/MYOD/MYF6 triple-knockout pig embryos (Maeng et al., 2021). Although in-depth analysis is lacking and complementation at a later stage was not examined, these two studies represent the first reported attempts at human-pig blastocyst complementation.

Despite the fact that rats and mice are relatively close in evolutionary distance, various malformations are often observed in chimeras generated between them, and their frequencies are positively correlated with levels of donor cell contribution. Moreover, high levels of chimerism often lead to early embryonic lethality (Yamaguchi et al., 2018). These results demonstrate early developmental incompatibilities between the two rodent species. In addition, chimeric levels vary among different organs in rat-mouse chimeras, suggesting xenogeneic incompatibilities also exist at late developmental stages. Researchers have encountered far greater difficulty in creating chimeras between evolutionarily more distant species. For example, hPSCs were found to contribute inefficiently to chimera formation in mouse and pig embryos (Masaki et al., 2015; Theunissen et al., 2016; Wu et al., 2017a), suggesting that the elimination of human cells occurs early on during development. Such loss of chimerism before the onset of organogenesis could preclude successful complementation. Overall, these results indicate that greater xenogeneic barriers exist between evolutionarily distant versus closely related species during early development. It will therefore be imperative to study and overcome these barriers, several of which we discuss below (and summarize in Fig. 3).

Differences in developmental time between donor and host species may constitute a barrier for interspecies chimerism. Several studies have shown that the species-specific pace of development can be maintained by PSCs ex utero (Barry et al., 2017; Chu et al., 2019; Diaz-Cuadros et al., 2020; Matsuda et al., 2020b). Barry et al. demonstrated that PSCs from different species pass through the cell cycle at different rates and differentiate at varying speeds (Barry et al., 2017). When applying the same neural differentiation protocol, it took a much longer time for primed hESCs to generate desired neuronal cell types than mEpiSCs. Interestingly, a human-specific neural differentiation rate was also maintained in teratomas generated from hESCs in immunodeficient mice, demonstrating the inability of host factors to accelerate the developmental clock of donor human cells (Barry et al., 2017). These results suggest that developmental time involves a meaningful degree of cell autonomy, which is in part controlled by species-specific biochemical reaction speeds (Matsuda et al., 2020a; Rayon et al., 2020). Incompatibilities in developmental tempo may thus prevent donor cells from responding to proliferation and/or differentiation cues in a synchronized manner with host cells; it is not a leap to speculate that hPSCs might find it difficult to follow the fast pace of the mouse developmental timeline during chimera formation.

However, emerging evidence from interspecies chimera experiments suggests that, despite intrinsic timing differences among species, xenogeneic donor cells are able to follow the developmental time of the host species in some contexts. By developing new naïve hPSC conditions, several recent reports claimed improved human chimerism in mice, even at the E17.5 stage (Bayerl et al., 2021; Hu et al., 2020; Taei et al., 2020). Although validation from independent labs is needed, these findings raise the tantalizing possibility that hPSCs can be induced to accelerate their developmental rate to match that of mice. In agreement, recent work from our lab showed that XPSCs from horses, which have a gestation period much longer than that of mice (∼11-12 months versus ∼20 days), could contribute to chimera formation in E7.75 and E9.5 mouse embryos (Yu et al., 2021b). Most recently, Brown et al. provided additional support by reporting that co-differentiation with mPSCs accelerated the differentiation speed of hPSCs (Brown et al., 2021). Thus, likely due to their inherent plasticity, PSCs may be more flexible than originally thought in terms of their differentiation pace. It also appears that non-cell autonomous mechanism(s) likely exist to control developmental timing in both donor and host cells during embryogenesis, which warrants future studies on understanding how time is enacted in development.

Another point to consider is matching the developmental stage of donor PSCs with host embryos. Although mESCs robustly contribute to blastocyst chimeras, they fail to generate chimeras following injection into post-implantation epiblasts (Fig. 2). In contrast, although mEpiSCs undergo apoptosis following blastocyst injection, they efficiently chimerize post-implantation epiblasts (Huang et al., 2012). Interestingly, grafting primed hPSCs into mouse E6-E7 epiblasts followed by ex utero embryo culture also resulted in robust human-mouse chimera formation (Wu et al., 2015; Mascetti and Pedersen, 2016). Similar results were obtained after injecting primed hPSCs into primitive streak-stage chick embryos (Akhlaghpour et al., 2021). These studies demonstrate that stage matching donor PSCs with host embryos enables robust intraspecies chimera formation and should also be considered when generating interspecies chimeras.

Primed mEpiSCs are normally inefficient in contributing to chimeras following blastocyst injection. Intriguingly, overexpressing anti-apoptotic genes Bcl2, Bcl-xl (Bcl2l1) or CrmA enabled mEpiSCs to contribute to blastocyst chimeras (Masaki et al., 2016). Similarly, Bcl2-expressing rat EpiSCs were able to generate adult rat-mouse chimeras. These results demonstrate that blocking donor cell apoptosis can override a heterochronic barrier within the same species and across different species, thereby constituting a promising strategy to improve chimerism between evolutionarily distant species. Indeed, recent studies demonstrated that BCL2L1- or BCL2-overexpressing primed hPSCs showed improved chimerism in E6.5-E10.5 mouse embryos (Wang et al., 2018), and that BCL2-overexpressing and TP53-null PSCs resulted in enhanced chimeric levels of hPSCs in E17-E27 pig embryos (Das et al., 2020; Maeng et al., 2021). In addition, Huang et al. found that forced expression of BMI1 suppressed dissociation-induced apoptosis in primed hPSCs, which resulted in enhanced human cell chimerism in E10.5 mouse embryos (Huang et al., 2018).

Donor cell death may originate from cell competition, which describes an evolutionarily conserved cell selection process to eliminate viable but ‘less fit’ neighboring cells, thereby ensuring normal development and tissue homeostasis (Amoyel and Bach, 2014). Almost 45 years since its discovery in Drosophilamelanogaster (Morata and Ripoll, 1975), cell competition is now recognized in a wide range of developmental processes and diseases in different tissues and organisms (Baker, 2020). Cell competition may constitute a barrier to interspecies chimerism, with xenogeneic donor cells being seen as unfit cells that are actively removed from development. To study interspecies cell competition during early development, we developed an in vitro system based on co-culturing PSCs from different species (Zheng et al., 2021). We chose conditions that supported the long-term maintenance of PSCs from several species (Guo et al., 2017; Gafni et al., 2013; Theunissen et al., 2014; Wu et al., 2015; Yang et al., 2017) and, through this approach, uncovered surprising competitive interactions between PSCs from evolutionarily distant (e.g. human and mouse) but not closely related (e.g. mouse and rat) species. Interestingly, interspecies PSC competition occurred in primed but not naïve pluripotency. Mechanistically, the NF-κB signaling pathway and MYD88, a canonical adaptor protein downstream of the mammalian toll-like receptor (TLR) and interleukin 1 (IL1) receptor families, triggered apoptosis in human cells during co-culture with mEpiSCs. Knockout of P65 (RELA), a core component of the NF-κB pathway, or of MYD88 in human cells, could overcome primed human-mouse PSC competition, thereby improving human cell chimeric efficiency. NF-κB transcription factors are pivotal regulators of both innate and adaptive immune responses so, as the adaptive immune system is not developed in early embryos, this suggests that donor cell elimination is likely activated by the ancient and conserved innate immune recognition system mediated by TLRs.

Based on the differential adhesion hypothesis proposed by Malcom Steinberg in 1963, cells that adhere well to each other will freely intermingle, but cells expressing either incompatible or different levels of cell adhesion molecules (CAMs) will segregate away from each other (Steinberg, 1963). Differential cell adhesion is key to tissue separation and boundary formation during development and may constitute another xenogeneic barrier (Townes and Holtfreter, 1955; Winklbauer and Parent, 2017). During chimera formation, mismatch in CAMs may prevent donor cells from adhering to the corresponding tissue(s) of the host embryo, thereby resulting in a low degree of chimerism. For example, cell adhesion incompatibility has been noted between ICM cells and neural stem cells (NSCs) (Karpowicz et al., 2007) or mEpiSCs (Ohtsuka et al., 2012). In both cases, overexpressing E-cadherin in donor cells overcame this incompatibility and resulted in improved chimerism.

Another potential barrier is incompatibility between ligands and receptors due to genetic diversification, which often involves ligand and receptor co-evolution to improve binding affinity and/or specificity. The case of the prolactin receptor in mammals shows how episodes of rapid evolution in certain animal species have modified receptor and ligand genes (Markov et al., 2008). As prolactin has to bind its receptor to fulfill its function, it was anticipated that the gene encoding the prolactin receptor should be subjected to selective pressure in the same mammals, and this has been shown to be the case. Similar examples of G protein-coupled receptors (GPCRs) such as the receptors for PRXamides or secretins (PACAP and VIP) and their ligands also support the ligand-receptor coevolution theory (Markov et al., 2008). As a result of this rapid co-evolution, ligands from one species may not recognize the receptors of another, and this is a barrier that needs to be studied further. It is impossible to examine and optimize all cross-species ligand-receptor interactions in a developing embryo, but identifying key signaling pathways blocked by ligand/receptor incompatibility is an important step toward overcoming this barrier. Incompatible ligand/receptor interactions in human and mouse are well documented in cell culture. Genetically replacing or modifying key receptors might restore interspecies signaling, however dose responses should also be considered.

Overall, a number of general xenogeneic barriers appear to prevent robust interspecies chimerism. There may also be additional species-specific barriers. For example, day 8-12 pig conceptuses undergo substantial and rapid elongation, growing from 2-6 mm in diameter on day 10 to ∼100-150 mm in length on day 12 (Geisert et al., 2017). This rapid growth may limit the survival, proliferation and differentiation of donor hPSCs inside a developing pig embryo (Liu et al., 2021). Efforts in understanding and overcoming such xenogeneic barriers will be crucial to enable successful and efficient blastocyst complementation with hPSCs.

In the 27 years that have passed since it was first developed, blastocyst complementation has transformed from a niche developmental biology technique into a promising strategy to solve the worldwide shortage of donor organs. However, to realize the dream of generating human organs in animals, several further advances are likely needed. These include: (1) gaining a deeper and more systemic understanding of organogenesis across different species; (2) generating or engineering more chimera-competent donor PSCs; (3) creating precise and large-scale genome and/or epigenome editing tools to promote donor-host cell cooperation during embryogenesis; and (4) developing innovative strategies to enable the generation of a complete organ, including all parenchymal, stromal and endothelial components. These scientific endeavors should go hand in hand with efforts to assuage ethical concerns and increase societal awareness. First, we need to ensure there are robust and reliable methods to prevent human cell contribution to the host central nervous system and the germline. Guidelines and regulations also need to be set to regulate the use of interspecies blastocyst complementation without overly constraining its use for legitimate medical efforts. Finally, the general public needs to be educated about the potentials and limitations of this technique. Although it seems there is still a long way to go before interspecies blastocyst complementation becomes a viable source of human organs, we are in for an exciting journey of new scientific discoveries.

We apologize to authors whose work has not been included in this Review due to space limitations. We thank all members from the Jun Wu laboratory for their experimental and conceptual contributions, which led to some of the ideas presented in this review.