FA10@AikoHIRAKI:FA1014 JSONTXT

Summary The complexity of organogenesis hinders in vitro generation of organs derived from a patient's pluripotent stem cells (PSCs), an ultimate goal of regenerative medicine. Mouse wild-type PSCs injected into Pdx1−/− (pancreatogenesis-disabled) mouse blastocysts developmentally compensated vacancy of the pancreatic “developmental niche,” generating almost entirely PSC-derived pancreas. To examine the potential for xenogenic approaches in blastocyst complementation, we injected mouse or rat PSCs into rat or mouse blastocysts, respectively, generating interspecific chimeras and thus confirming that PSCs can contribute to xenogenic development between mouse and rat. The development of these mouse/rat chimeras was primarily influenced by host blastocyst and/or foster mother, evident by body size and species-specific organogenesis. We further injected rat wild-type PSCs into Pdx1−/− mouse blastocysts, generating normally functioning rat pancreas in Pdx1−/− mice. These data constitute proof of principle for interspecific blastocyst complementation and for generation in vivo of organs derived from donor PSCs using a xenogenic environment. Introduction Current stem cell therapy mainly targets diseases that can be treated by cell replacement, such as Parkinson's disease or diabetes mellitus. One of the ultimate goals of regenerative medicine, however, is to grow organs using the patient's own stem cells and to transplant those organs into the patient. With the development of induced pluripotent stem cell (iPSC) technology, we are now able to obtain patient-derived PSCs (Takahashi et al., 2007, Takahashi and Yamanaka, 2006), although actual developmental potentials remain to be defined, as do risks associated with somatic cell reprogramming. The real challenge is to create a reproductive system for generation of PSC-derived organs. The interactions among cells and tissues during development and organogenesis are so complex that the recapitulation of these interactions to generate organs in vitro is essentially impractical. We have challenged this goal using the biology of blastocyst complementation. Blastocyst complementation was first reported by Chen et al. They demonstrated that deficiency of T and B lymphocyte lineages in Rag2-deficient (Rag2−/−) mice was complemented by injecting normal mouse embryonic stem cells (mESCs) into Rag2−/− mouse-derived blastocysts (Chen et al., 1993). Because Rag2 is an indispensable enzyme for rearrangement of immunoglobulin and T cell receptor genes, the T and B cells generated in the complemented animals were mESC derived; there were no host T or B lymphocytes. We assumed that this complementation was possible because the Rag2−/− host, incapable of generating mature T and B cells, provided a “developmental niche” for ESC-derived T and B cells. We hypothesized that with blastocysts derived from a mutant mouse strain in which the gene necessary to form a particular organ is deficient, the same principle might apply. To test this hypothesis, we used, in this study, blastocyst complementation to generate functional pancreas from donor PSCs. The pancreas, consisting of endocrine and exocrine glands, is formed by early embryonic interactions of mesenchyme and epithelium (Slack, 1995). Pdx1 (pancreatic and duodenal homeobox1) is a Hox-type transcription factor that plays a critical role in pancreatic development and β cell maturation. Homozygous deficiency of Pdx1 in the mouse results in death soon after birth due to pancreatic insufficiency (Offield et al., 1996). Targeted disruption of Pdx1 thus should empty a pancreatic “developmental niche” in embryos derived from Pdx1−/− blastocysts. Therefore, injection of mouse PSCs (mPSCs) into Pdx1−/− blastocysts should result in generation of pancreata almost entirely derived from injected mPSCs. Although our results verified our initial hypothesis, this system cannot be applied to generate human organs. A xenogenic, but not allogenic, blastocyst complementation system must therefore be established. However, little is known about the nature of the xenogenic barrier, how organogenesis by donor PSCs is influenced, or how intrinsic developmental programs can be regulated in xenogenic environments. To address these issues, we attempted to generate interspecific chimeras between mouse and rat using a blastocyst injection technique with mouse and rat PSCs. Generation of interspecific chimeras in livestock animals using preimplantation embryos of each species is described; the “geep,” or chimera between goat and sheep, is famous as an interspecific live chimera (Fehilly et al., 1984). In rodents, however, methods like those producing the geep succeeded only between mouse subspecies, such as Mus musculus and Mus caroli, that are closely related but not capable of interbreeding (Rossant and Frels, 1980). Many groups have sought to generate interspecific chimeras between mouse and rat, successful with chimeric preimplantation embryos in vitro but failing with live chimeric animals (Stern, 1973, Zeilmaker, 1973). Extraembryonic lineage cells like trophectoderm or primitive endoderm, derived from xenogenic embryos, might suffer inhibition of implantation on exposure to host uterus; inhibition of further intrauterine development also is possible (Rossant et al., 1982, Tarkowski, 1962). Only cells of pre-blastocyst origin can contribute to extraembryonic lineage cells, and mESCs/iPSCs do not thus contribute (Rossant, 2007). Therefore, we rechallenged this old issue with new technology, using iPSCs or ESCs that are not capable of contributing to extraembryonic tissues. Using recently established culture conditions with a combination of signaling inhibitors (Buehr et al., 2008, Li et al., 2008, Ying et al., 2008), we generated rat-ESCs and -iPSCs (rESCs, riPSCs). Our work has now confirmed the existence of interspecific chimeras generated with PSCs, using injection not only of mouse PSCs into rat embryos but also of rat PSCs into mouse embryos. Finally, by combining the principle of blastocyst complementation with the production of interspecific chimeras, we succeeded in generating rat pancreas in Pdx1−/− mice. These sets of experiments provide proof-of-principle data for donor iPSC-derived organ generation in a xenogenic environment. Results Generation of Donor Mouse iPSC-Derived Pancreas in Pdx1-Deficient Neonates Our first goal was to generate pancreas from mPSCs. We performed a blastocyst complementation experiment using Pdx1−/− blastocysts that would provide a niche for pancreatic organogenesis. These mice exhibit pancreatic agenesis due to the absence of Pdx1-specified interactions between prepancreatic epithelium and mesenchyme, a key step in pancreatic development. Homozygote Pdx1-LacZ knockin mice (Pdx1−/−) are born alive but die within 1 week after birth, presumably due to pancreatic insufficiency (Offield et al., 1996). For complementation donor cells, we used GT3.2 mouse iPSCs (miPSCs) that had been generated from tail tip fibroblasts (TTFs) of an adult EGFP-transgenic (TG) C57BL/6 mouse (Okabe et al., 1997) using three factors (Oct3/4, Sox2, Klf4) in retroviral vectors (Figure S1 available online). These GT3.2 miPSCs were injected into blastocysts obtained by an intercross of heterozygote Pdx1-LacZ knockin mice (Pdx1+/−), offspring of C57BL/6(Pdx1+/−) × DBA(Pdx1+/+) or C57BL/6(Pdx1+/−) × BDF1(Pdx+/+) mice. G4.2 mESCs were also injected for comparison. Neonates were assessed macroscopically and histologically for pancreatic development and were genotyped for Pdx1. These neonates were highly chimeric: Due to silencing of EGFP and to contamination by donor cells, accurate determination of the genotypes required analysis at the single-cell level. To identify the genotypes of these neonates accurately, EGFP-negative, c-Kit-positive, and Sca1-positive, lineage marker-negative (EGFP−KSL) bone marrow cells were clone-sorted by fluorescence-activated cell sorting (FACS) to obtain single-cell-derived hematopoietic colonies (Figure S2A ). After culture for 2 weeks, cells collected from each colony were subjected to PCR analysis (Figure S2B). This “colony PCR” method unambiguously determined the host embryo genotype and proved that pancreas had been formed in Pdx1−/− mice by blastocyst complementation (Figure S2C). In all postnatal chimeric mice derived from blastocysts injected with either miPSCs or mESCs, pancreas was present regardless of host genotype (including Pdx1−/−). As expected, the pancreas in Pdx1+/− or Pdx1+/+ chimeric mice was a composite of host-derived cells and EGFP-miPSC- or mESC-derived cells, as in whole-body chimerism (Pdx1+/− + miPSCs or mESCs in Figure 1A ). In contrast, the pancreatic epithelium in Pdx1−/− chimeric mice was almost entirely composed of EGFP-marked miPSC- or mESC-derived cells (Pdx1−/− + miPSCs or mESCs in Figure 1A). The pancreas of these mice was grossly and histologically normal. We examined these pancreatic tissues further for contributions of donor mPSCs to different pancreatic lineages. Both miPSCs and mESCs supplied all pancreatic cell lineages (exocrine and endocrine tissues, including ductal epithelia), but pancreatic stromal elements—vessels, nerves, and fibrocytes—were composites of host- and mPSC-derived cells in all mice (Figure 1B). miPSCs Rescued Pdx1−/− Mice by Blastocyst Complementation We next addressed whether mPSCs can rescue Pdx1−/− lethality via blastocyst complementation. As we predicted based on neonatal analysis (Figure 1), both mESCs and miPSCs contributed to pancreatic organogenesis; Pdx1−/− chimeric mice survived to adulthood (mESCs injection: n = 4, miPSCs injection: n = 15 in Figure 2A ). They even served as Pdx1−/− founders, transmitting their genotype to the next generation. Mating Pdx1−/− founder male mice with Pdx1+/− female mice increased to 50% the proportion of pups of homozygously disrupted genotype (Figure 2A). These data indicate that blastocyst complementation can be used to generate a functional organ and to yield homozygous founder mice even when uncomplemented homozygosity is fatal. The method described here thus can be useful for efficient production of gene-targeted mice that are embryonic lethal without introducing a conditional gene targeting system. As expected, in both Pdx1−/− and Pdx1+/ − mice injected with EGFP-miPSC, miPSC-derived cells contributed to all the nonpancreatic tissues of the body, including lung, heart, liver, muscle, testis, and brain (data not shown). The extent of contribution varied from tissue to tissue and also varied depending on individual chimera. However, the pancreas in adult Pdx1−/− mice was entirely derived from donor miPSCs (Pdx1−/− in Figure 2B). Detailed analysis of EGFP distributions in adult miPSC-derived pancreas demonstrated that pancreatic islets, exocrine tissues, and duct epithelia were entirely derived from donor miPSCs, as already shown in neonates (Pdx1−/− in Figures 2C and 2D). In expected contrast, pancreas from Pdx1+/− mice was a composite of host and donor derivatives (Pdx1+/− in Figures 2B and 2C). Quantitative analysis of these sections by image J software revealed percentages of EGFP-positive cells to be 27.4% ± 27.8% in the miPSC + Pdx1+/− setting, 95.6% ± 4.6% in the miPSC + Pdx1−/− setting. Of note is that most individual islets were composed of both host-derived and miPSC-derived cells (Pdx1+/− in Figure 2D), indicating, as previously reported (Deltour et al., 1991), nonclonal origin of pancreatic islets. To test whether miPSC-derived pancreata are functional in Pdx1−/− chimeric mice, we performed glucose tolerance testing (GTT). Whereas streptozotocin (STZ)-induced diabetic mice failed to respond to GTT, Pdx1−/− chimeric mice responded well, indicating that Pdx1−/− mice with miPSC-derived pancreas secreted insulin in response to glucose loading and maintained normal serum glucose levels (Figure 2E). Histology and function of exogenously derived pancreas were essentially the same for mice derived from miPSC- and from mESC-complemented blastocysts (Figure S3). These data demonstrate that the vacant “pancreatic niche” provided in Pdx1−/− mice can be occupied, with developmental compensation, by miPSC- or mESC-derived cells after intraspecific blastocyst complementation, generating functionally intact pancreas. Transplantation of miPSC-Derived Islets Corrected Hyperglycemia in Diabetic Mice To assess the functionality of miPSC-derived pancreas, islets from miPSC-derived pancreas were transplanted into mice of the original strain (C57BL/6) in which STZ administration had induced diabetes. As a control, islets from pancreas in Pdx1+/− chimeric mice injected with miPSCs were transplanted. Blastocysts were obtained by an intercross of Pdx1+/− mice (C57BL/6 × DBA2 or C57BL/6 × BDF1 F1 strains) that were semi-allogenic to miPSCs used in this study. EGFP-expressing miPSC-derived islets (Figure 3A ) were isolated conventionally based on a method developed previously (Gotoh et al., 1985) and were transplanted beneath the renal capsule of recipient mice. Nonfasting blood glucose levels were then monitored. As these islets were of donor origin (i.e., C57BL/6 strain), an immunosuppressive regimen was not used. To prevent nonspecific loss of islets due to inflammation, anti-inflammatory cytokine monoclonal antibody (mAb) cocktails were given at transplantation and 2 and 4 days thereafter (arrows, Figure 3E), as described (Satoh et al., 2007). Two months after transplant, EGFP-expressing miPSC-derived islets were still detectable at the graft site (Figures 3B and 3C). Production of insulin by the transplanted islets was confirmed immunohistologically (Figure 3D). The induced-diabetic recipients no longer exhibited hyperglycemia; they maintained normal blood glucose levels and responded normally to GTT (Figures 3E and 3F). This is in contrast to the therapeutic effect conferred by islets, composites of blastocyst- and miPSC-derived cells, obtained from pancreas of Pdx1+/− chimeric mice (C57BL/6 × DBA2 or C57BL/6 × BDF1 F1 origin). This effect lasted only for a short time, presumably due to immune rejection by the host C57BL/6 diabetic mice (Figure 3E). These data strongly indicate that the miPSC-derived pancreas, with islets, formed in an allogenic host is functional and that the “autologous” islets thus formed can be used to treat diabetes, without rejection. Blastocyst complementation thus permits demonstration of proof-of-principle both for pancreas generation from PSCs and for diabetic therapy using donor iPSC-derived syngenic islets. Generation of Interspecific Chimeras between Mouse and Rat Our second goal was to generate interspecific chimeras between mouse and rat. To achieve this goal, we generated not only miPSCs but also riPSCs and rESCs using an established protocol (Hirabayashi et al., 2009, Ying et al., 2008). These mouse and rat PSCs enabled us bidirectionally to generate interspecific chimeras. We injected EGFP-marked GT3.2 miPSCs into rat blastocysts (r-blastocysts) or EGFP-marked riPSCs (Figures S4A and S4B ) into mouse blastocysts (m-blastocysts). Because post-implantation development reportedly is severely hampered after intrauterine transfer of xenogenic blastocysts (Rossant et al., 1982, Tarkowski, 1962), injected r-blastocysts or m-blastocysts were transferred, respectively, into the uteri of pseudopregnant rats or mice. After intrauterine transfer of injected r-blastocysts or m-blastocysts with development to the fetal stage, we evaluated EGFP expression by fluorescence microscopy for each transferred embryo. EGFP-expressing cells were found in the body of each injected conceptus, but never in placenta (Figure 4A ). This finding indicates that injected mouse or rat iPSCs can contribute to xenogenic development, with generation of interspecific chimeras. Next, we tried to quantitate contribution of mouse or rat iPSCs to these interspecific chimeras. Chimerism in interspecific embryos appeared to vary individual-to-individual and organ-to-organ. Since quantitation of PSC-derived cells was difficult in organs, we analyzed embryonic fibroblasts and hematopoietic cells. FACS analysis of embryonic fibroblasts revealed that donor-derived EGFP+ cell percentages of about 28.0% and 26.5%, respectively, were detected in mouse and rat interspecific chimeras (representative FACS data shown in Figure 4B). We also examined chimerism in hematopoietic cells by staining cells from livers of interspecific chimera fetuses with antibodies specific for mouse and rat CD45 antigens. Cells that expressed mouse or rat CD45 represented distinct populations in interspecific chimeras, with only cells derived from injected iPSCs expressing EGFP (Figure 4B). Whereas a high proportion (28.3%) of mouse blood cells was detected in r-blastocyst-derived chimeric fetal liver, rat blood cells were only rarely present (less than 3.3%) in m-blastocyst-derived chimeric fetal liver (Figure 4B). This tendency was specific to interspecific chimeras and not observed in intraspecific chimeras (Figure 4E). It is not clear why the difference in contribution of iPSCs to hematopoietic cells between mouse and rat interspecific chimeras was so marked. To further confirm interspecific chimerism, genomic DNA extracted from FACS-sorted cells expressing CD45 was PCR-amplified using primers common to the mouse and rat Oct3/4 loci (Figure 4C). PCR products of different lengths, indicating origin in each species, were clearly present (Figure 4D). These results strongly indicated that the animals harboring these cells were mouse/rat interspecific chimeras. To investigate the influence of iPSC contribution to xenogenic development at the fetal stage, we assessed embryonic development rate of interspecific chimeras, and the extent of chimerism, by FACS analysis using established embryonic fibroblasts. Both embryonic development rate and degree of chimerism were lower in interspecific chimeras than in intraspecific chimeras (Figures 4F and 4G). In addition, high contributions by xenogenic cells appeared to be associated with morphological abnormalities and embryonic lethality (data not shown). To exclude the possibility that these abnormalities were caused by donor iPSCs, we also attempted to generate interspecific chimeras using mouse or rat ESCs. DsRed-marked EB3DR mESCs could also generate interspecific rat chimeras (top panels in Figure S5A ), with embryonic development rate and degree of chimerism similar to those generated by miPSC injection (Figure S5B). The Venus-marked WIv3i-1 and -5 rESC lines, with high contribution to rat embryo development and germline competency (Hirabayashi et al., 2009), could also generate interspecific chimeras after injection into m-blastocyst (middle and bottom panels in Figure S5A), but embryonic development rate and degree of chimerism were lower than reported for intraspecific chimeras (Figure S5C). These results suggest that generation of interspecific chimeras between mouse and rat is less efficient than generation of intraspecific chimeras. Interspecific Chimeras Were Live-born; Some Grew into Adulthood To investigate the developmental potential of generated chimeras and to assess the functionality of the cells, tissues, or organs derived from injected cells, we analyzed interspecific chimeras at neonatal and adult stages. Mouse- or rat-iPSC-injected interspecific chimeras survived after birth and expressed EGFP ubiquitously (as did intraspecific chimeras; Figure 5A , r-blastocyst + miPSCs: n = 5, m-blastocyst + riPSCs: n = 10). As these chimeras developed into adulthood, chimerism could be judged by coat color because miPSCs (C57BL/6, black coat) were injected into r-blastocyst (Wistar, white coat) or riPSCs (Wistar) were injected into m-blastocyst (BDF11 × C57BL/6, black coat) (Figure 5C, r-blastocyst + miPSCs: n = 8, m-blastocyst + riPSCs: n = 4). The full-term development rate of interspecific chimeras, either with miPSCs into r-blastocyst or with riPSCs into m-blastocyst, was ∼20%. In both settings it was lower than that for intraspecific chimeras (∼50%). Determination of Body Size in Interspecific Chimeras Adult rats typically are ten times bigger than adult mice, whereas newborn rats are three times bigger than newborn mice. Because mouse and rat gestations are of similar length (19 and 21 days, respectively), organogenesis requires more cell proliferation and differentiation during rat development than during mouse development. What determines the size of interspecific chimeras is an intriguing biological question. Interestingly, body size and weight of interspecific chimeras born from rat foster mothers were similar to those of normal newborn rats, whereas for those born from mouse foster mothers they were similar to those of newborn mice (Figures 5A and 5C). However, high contributions by xenogenic cells may also affect interspecific chimera size. One chimera obtained after injection of miPSCs into r-blastocysts showed body weight and size equivalent to those of a newborn mouse, although its birth mother was a rat (Figure 5B). This particular chimera's donor miPSC-derived cell contribution was extremely high (Figure 5B, inlet). However, chimeras with high contributions of xenogenic iPSC-derived cells generally were not identified, suggesting an association with embryonic lethality. In most chimeras, the size of newborns seemed to conform with that in the species from which the blastocyst originated. A correlation between body weight and contribution of donor iPSC-derived cells as estimated from hair color and peripheral blood cells is shown in Figure 6. The origins of placenta and uterus may have a key role in size determination. Injected xenogenic PSCs never developed into placenta, which was always of host blastocyst origin (Figure 4A, Figure S5A). Therefore it is difficult to determine whether it is placenta or uterine environment that influences the size of embryos. Distribution of Donor iPSC-Derived Cells in the Xenogenic Environment We determined the distribution of mouse- or rat-iPSC-derived EGFP-positive cells in neonatal interspecific chimeras. With miPSC injection into r-blastocyst (Figure S6A ) and with riPSC injection into m-blastocyst, almost all organs contained EGFP-positive cells (Figure S6B). In sections immunostained with anti-EGFP antibody, representative images demonstrated that EGFP-positive cells become various types of tissues (Figure 5D). Of particular note was a pancreatic islet generated by injection of miPSCs into r-blastocyst. Cells in this islet marked immunohistochemically for insulin; the islet was a composite of EGFP-positive and -negative cells, indicating that the islet consisted of host rat cells and exogenous donor mouse cells (Figure 5E). Islet polyclonality also was seen with intraspecific chimeras (Figure 2E). The expression of other functional molecules in xenogenic cells (i.e., hepatocytes or cholangiocytes in the liver or leukocytes in the peripheral blood) was also detected (Figure 5E and Figure S6C). It is known that rats do not have gallbladders, whereas mice do. To date, when mPSCs were injected into r-blastocysts, the interspecific chimeras produced (generally the size of rats) have not had gallbladders (n = 8). In contrast, interspecific chimeras generated from m-blastocysts complemented with rPSCs (more like mice in size) have had gallbladders (n = 4). To see the xenogenic contribution to germ cells, testis and developing gonad of the interspecific chimeras were examined. Cells were not found that coexpressed EGFP or Venus and the germ cell marker mouse vasa homolog (MVH) (Figure 5F). As the injected miPSCs or rESCs were both confirmed as germline-competent pluripotent stem cells in intraspecific settings (data not shown), germ cell development may be impaired or more limited in the xenogenic environment. Generation of Rat Pancreas in Mouse via Interspecific Blastocyst Complementation Our last goal was to generate xenogenic rat pancreas in Pdx1−/− mice by interspecific blastocyst complementation. For efficient production of Pdx1−/− mice, embryos were generated by intercross of Pdx1−/− founder male mice (in which pancreas arose principally from exogenous miPSCs) with Pdx1+/− female mice. With this founder system, half the mice born were Pdx1−/− (in contrast to intercross of Pdx1+/− heterozygotes in which only 25% of offspring were Pdx1−/−). For donor riPSCs, we used the riPSC#3 line (Figure S4A). This line was selected among 11 established riPSC clones for embryonic developmental rate and degree of chimerism after injection into mouse embryos (data not shown). After injection, 139 embryos were transferred into uteri of pseudopregnant mice and 34 mice were born. They were analyzed at neonatal and adult stages. The contribution of EGFP-marked riPSC-derived cells in the pancreas of neonatal Pdx1+/− interspecific chimeras was small relative to that of host-derived cells, as seen in whole-body chimerism (Pdx1+/− + riPSCs in Figure 7A bottom panel; n = 5). In contrast, the pancreatic epithelia in Pdx1−/− interspecific chimeras were entirely composed of EGFP-marked riPSC-derived cells (Pdx1−/− + riPSCs in Figure 7A top panel; n = 10). Each genotype was confirmed by PCR using genomic DNA extracted from FACS-sorted mouse CD45 (mCD45)-positive splenocytes (Figure S7A ). Neonates with entirely EGFP-positive pancreata thus were clearly identified as of Pdx1−/− genotype (Figure S7B). The existence of interspecific chimeras between mouse and rat was also confirmed by FACS patterning, which demonstrated distinct populations of mCD45- and rat CD45 (rCD45)-positive cells, with only rCD45-positive cells expressing EGFP after riPSC injection (Figure S7A). FACS-sorted mCD45- and rCD45-positive cells were also confirmed as, respectively, mouse or rat in origin by genomic PCR testing using Oct3/4 locus primers (Figure S7C), which clearly identified cell origin. On immunostaining, riPSC-derived pancreas expressed EGFP almost universally (Figure 7B) and also expressed α-amylase (an exocrine tissue marker) and insulin, glucagon, and somatostatin (endocrine tissue markers; Figure 7B). As in wild-type experiments, successful maturation into adulthood (8 weeks) was uncommon in Pdx1−/− mice complemented with riPSCs; however, adult mice with riPSC-derived pancreas (Figure 7C; n = 2) had intact pancreas expressing EGFP (Figures 7E and 7F). Quantitative analysis of the sections by image J software revealed percentages of EGFP-positive cells to be 81.9% ± 3.4%. Additionally, on GTT in adulthood, insulin was secreted in response to glucose loading and normal serum glucose levels were maintained (Figure 7D). These results indicate that generation of a xenogenic iPSC-derived organ is possible via interspecific blastocyst complementation. Discussion We report three innovative observations, using proof-of-principle approaches. (1) If an empty developmental niche for an organ is provided (as with the Pdx1−/− mouse and the pancreatic niche), PSC-derived cellular progeny can occupy that niche and developmentally compensate for the missing contents of the niche, forming an organ almost entirely composed of cells derived from donor PSCs. (2) Generation of interspecific chimeras between mouse and rat is possible with injection of mouse or rat PSCs into embryos from the other species; injected PSC-derived cells are distributed throughout the body and appear to function normally. (3) The combination of (1) and (2) successfully generates rat pancreas in mouse with injection of riPSCs into Pdx1−/− mouse embryos, a technique that we term “interspecific blastocyst complementation.” We demonstrated that the Pdx1−/− mice derived from blastocysts complemented with mPSCs were born with functional pancreas almost entirely derived from donor PSCs and grew into adulthood without showing any evidence of pancreatic insufficiency. Several groups have used the same technique to study the development of thymic epithelium (Muller et al., 2005), to compensate for cardiac defects (Fraidenraich et al., 2004), or to determine if yolk sac hematopoiesis and germ cell development are of clonal or nonclonal origin (Ueno et al., 2009, Ueno and Weissman, 2006). Although Stanger et al. tested development of pancreas and liver in embryos to define organ size determinants (Stanger et al., 2007), no study has exploited this technique to produce donor-derived functional organs and rescued a lethal phenotype to adulthood. Direct in vitro differentiation of insulin-producing cells from PSCs has been a major focus of stem cell therapy, as recently demonstrated by a sophisticated protocol to generate pancreatic endoderm efficiently via stepwise endodermal differentiation (Kroon et al., 2008). However, the in vitro generation of insulin-producing cells still needs further improvement in differentiation efficiency, in insulin production levels, or in speed of insulin response to glucose changes. In addition, the risk of tumor development due to contamination with undifferentiated PSCs must be rigorously assessed before clinical use. Compared with those generated in vitro, insulin-producing cells obtained from pancreas that is formed in vivo by blastocyst complementation must have gone through near-normal differentiation processes with proper epigenetic changes. The tissues obtained, such as insulin-producing cells, thus are presumed to be fully functional and the risk of teratoma development due to contamination of undifferentiated PSCs to be negligible. Nonetheless, the oncogenicity of iPSC-derived cells due to reactivation of introduced genes (Miura et al., 2009, Nakagawa et al., 2008) or to genome abnormality due to long-term culture remains to be assessed. To establish iPSCs without genomic integration of a retroviral sequence should further reduce the risk associated with the use of iPSCs (Okita et al., 2008). We assumed that aggregation of early embryos would lead to the presence in trophectoderm of xenogenic cells that are reported to be harmful to embryonic development after uterine implantation. We therefore injected mESCs/iPSCs, but not blastomere cells, into r-blastocysts. As predicted, EGFP-positive PSC-derived cells were not detected in placentas (Figure 4A and Figure S5A), and we succeeded in generating interspecific chimeras. To confirm this further, we then attempted to generate interspecific chimeras by injecting rESCs/iPSCs into m-blastocysts. After injection into 8-cell/morula stage mouse embryos, the riPSCs were eventually enclosed within the inner cell mass of the m-blastocyst and were never detected in the m-blastocyst trophectoderm (Figures S4C and S4D). Our study, consistent with others, indicates that the presence of xenogenic cells among extraembryonic lineage cells, with exposure of xenogenic cells to the uterine environment, is inhibitory to implantation and/or to further intrauterine development of interspecific chimeras. It is of prime interest that body size and weight of interspecific chimeras conformed with those of the species of the foster mother (Figures 5A–5C). What xenogenic components contribute to the phenotypic determination of interspecific chimeras? It seems that placenta and/or uterine environment are responsible for size determination of adult interspecific chimeras both as embryos and as adults. Given that placenta must be of the same origin as the foster mother for successful generation of interspecific chimeras, it is not clear which is primarily responsible for this determination. Degree of chimerism may also influence the phenotype. As observed in intraspecific chimeras, contribution and distribution of xenogenic cells in interspecific chimeras vary organ-to-organ. There is a negative correlation between contribution of donor (mouse) PSC-derived cells and body weight (Figure 6). Although we could obtain interspecific chimeras consistently (Figure 4 and Figure 5), embryonic lethality was high and postnatal development was poor. Some interactions between mouse- and rat-derived cells indispensable for organism survival may not work across species, resulting in death during embryonic development or in retarded postnatal development. Another example is the formation of gallbladders in interspecific chimeras. Those derived from m-blastocysts have gallbladders but those from r-blastocysts do not. It is conceivable that the temporo-spatial development of donor PSC-derived cells is regulated by the xenogenic host microenvironment, which governs morphogenesis and organogenesis. However, the data also indicate that the intrinsic developmental program imprinted in PSCs may create competition between host blastocyst-derived and donor PSC-derived cells to form chimeric organs. The balance between host and donor cells at certain critical points during embryonic development thus may be important. Xenogenic developmental systems may be useful to elucidate these and other developmental questions and to help in generation of chimeras between species evolutionarily more distant from one another than are mouse and rat. Most importantly, we succeeded in generating functional rat pancreas in Pdx1−/− mice via interspecific blastocyst complementation. In all interspecific neonates derived from Pdx1−/− blastocysts injected with riPSCs, pancreas was present. Although full maturation into adulthood was not common, once the mice matured into adulthood, the generated riPSC-derived pancreas was morphologically and histologically normal and was not associated with any sign of diabetes or other abnormalities; GTT results strongly indicated normal function. Generation of functional cells (sperm, hepatocytes) in xenogenic environments has been reported (Mercer et al., 2001, Shinohara et al., 2006). In addition, hematopoietic xeno-chimeras have been used commonly as a method to study functionality of hematopoietic stem cells (Kamel-Reid and Dick, 1988, McCune et al., 1988). No study, however, has demonstrated generation in a xenogenic environment of a PSC-derived functional organ that can rescue embryonic lethality to adulthood. The organ generation system described may be applied to treat organ failure in humans if pigs or other large animals are used. There are, however, several issues that need to be addressed to bring this principle into the clinic. For example, though we were able to generate interspecific chimeras between mouse and rat, their embryonic lethality is high and maturation into adulthood is uncommon. The nature of this xenogenic barrier is not clear, but it is evident that the evolutionary distance accounts for this, as we do not see these problems in intraspecific chimeras. Livestock animals such as pigs or sheep may be too distant evolutionarily for successful complementation. In addition, as described in the allogenic system, vessels, nerves, and some interstitial elements that are not under the influence of Pdx1 expression were composites of host- and miPSC- or mESC-derived cells. Although we showed in this study that islets prepared from miPSC-derived pancreas generated in allogenic hosts indeed were successfully transplanted into “autologous” mice with STZ-induced diabetes without rejection, whether the same principle applies to transplantation of islets obtained from pancreas generated in xenogenic animals remains to be seen. Transplantation of islets from rat pancreas generated in Pdx1−/− mice into diabetic rats should answer this question. However, due to the size difference between mouse and rat, and to high embryonic lethality and poor postnatal maturation of interspecific chimeras, it is not possible to obtain sufficient numbers of islets to treat diabetic rats. Generation of mouse pancreas in Pdx1−/− rats will be necessary to do such experiments. Production of organ-deficient livestock animals and generation of chimeras is another issue, but nuclear transfer technology available for livestock animals may permit establishment of organ-deficient pig lines, for example (Lai et al., 2002). The successful generation of pig chimeras using blastocyst injection has also been reported (Nagashima et al., 2004). The major difficulty seems to be that primate and rodent PSCs differ (Nichols and Smith, 2009); limits of primate PSCs in contributing to embryo development have been suspected. Poor contribution of human ESCs to embryo development after injection into mouse blastocysts or chick embryo has been demonstrated (Goldstein et al., 2002, James et al., 2006). Generation of interspecific chimera technology may prove a tool useful in assessment of pluripotent stem cell potential, thereby addressing this issue. Another issue of concern is the fact that PSC-derived cells are found not only in pancreas but in all organs and tissues, including brain and gonads. Therefore, without proper control of the differentiation potential of PSCs, generation of human organs in livestock animals will face an ethical issue. There are several approaches to address this. One is use of committed stem or progenitor cells in place of PSCs. If they are introduced into an appropriate microenvironment at an appropriate developmental time point, to restrict differentiation toward a particular organ may be possible. An alternative is to use genetically modified PSCs whose differentiation potential is restricted to certain tissues or organs. In conclusion, the approach described here will be of use not only for better understanding of the mechanism of organogenesis but also as an initial step toward the ultimate regenerative medicine of the future. Experimental Procedures Animals C57BL/6NCrSlc, BDF1, DBA/2CrSlc, ICR mice, and Wistar rats were purchased from SLC Japan (Shizuoka, Japan). Pdx1-LacZ heterozygous mice (Offield et al., 1996), kindly provided by Dr. Y. Kawaguchi (Kyoto University) and Dr. C.V. Wright (Vanderbilt University), were crossed with C57BL/6, DBA2, or BDF1 strain mice. C57BL/6 mice were given STZ (Sigma, St. Louis, MO, USA) to induce diabetes. Mice with nonfasting blood glucose levels > 400 mg/dL 1 week after STZ administration (200 mg/kg) were regarded as hyperglycemic and thus as diabetic mice. All experiments were performed in accordance with the animal care and use committee guidelines of the Institute of Medical Science, University of Tokyo. Culture of ESCs/iPSCs Undifferentiated mESCs were maintained on gelatin-coated dishes without feeder cells in Glasgow's modified Eagle's medium (Sigma) supplemented with 10% fetal bovine serum (FBS; Nichirei Bioscience, Tokyo, Japan), 0.1 mM 2-mercaptoethanol (Invitrogen, San Diego, CA, USA), 0.1 mM nonessential amino acids (Invitrogen), 1 mM sodium pyruvate (Invitrogen), 1% L-glutamine penicillin streptomycin (Sigma), and 1000 U/ml of mouse leukemia inhibitory factor (LIF; Millipore, Bedford, MA, USA). The G4.2 mESCs and EB3DR mESCs, kindly provided by Dr. H. Niwa (Center for Developmental Biology, RIKEN), were derived from EB3 mESCs (Niwa et al., 2000) and carried the CAG promoter-driven EGFP or DsRed gene. Undifferentiated miPSCs were maintained on mitomycin-c treated mouse embryonic fibroblasts (MEFs) in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 15% knockout serum replacement (Invitrogen), 0.1 mM 2-mercaptoethanol, 0.1 mM nonessential amino acids, 1 mM HEPES buffer solution (Invitrogen), 1% L-glutamine penicillin streptomycin, and 1000 U/ml of mouse LIF. The GT3.2 miPSCs were generated from TTFs of a male EGFP Tg mouse (kindly provided by Dr. M. Okabe, Osaka University) by introducing three factors (Klf4, Sox2, Oct3/4) in retroviral vectors (Okabe et al., 2009). The GT3.2 miPSCs ubiquitously express EGFP under the control of the CAG promoter. Undifferentiated rESCs/iPSCs were maintained on mitomycin-c treated MEFs in N2B27 medium (Ying et al., 2003) containing 1 μM MEK inhibitor PD0325901 (Axon Groeningen, The Netherlands), 3 μM GSK3 inhibitor CHIR99021 (Axon), with or without FGF receptor inhibitor SU5402 (Calbiochem, La Jolla, CA, USA), and 1000 U/ml of rat LIF (Millipore). The riPSCs were generated from Wistar rat embryonic fibroblast by introducing three mouse factors (Oct3/4, Klf4, Sox2) in lentiviral vectors as a doxycycline-inducible expression unit using N2B27 medium containing the three inhibitors described above. Expression of representative ESC marker genes has been confirmed and teratoma formation by these riPSCs after injection into immunodeficient mice has also been confirmed (our unpublished data). The riPSCs ubiquitously express EGFP under the control of the Ubiquitin-C promoter. The WIv3i-1 or WIv3i-5 rESCs were generated from Venus Tg rat blastocyst (Hirabayashi et al., 2009). These rESCs ubiquitously express Venus under the control of the CAG promoter. Embryo Culture and Manipulation Preparation of wild-type or Pdx1 heterozygous intercrossing embryos was carried out according to published protocols (Nagy et al., 2003). In brief, mouse 8-cell/morula stage embryos were collected in Medium 2 (Millipore) from oviduct and uterus of mice 2.5 days postcoitum (dpc). These embryos were transferred into potassium simplex optimized medium with amino acids (Millipore) and were cultured for 24 hr for blastocyst injection. Rat blastocysts were collected in a bicarbonate-buffered medium composed of Roswell Park Memorial Institute medium (RPMI) 1640, Eagle's solution, and Ham's F12 containing 18% FBS medium (Ogawa et al., 1971) from oviduct and uterus of rats 4.5 dpc. These embryos were transferred into modified rat 1-cell embryo culture medium (Oh et al., 1998) containing 80 mM NaCl (Wako Pure Chemical Industries, Osaka, Japan) and 0.1% polyvinyl alcohol (Sigma) and were cultured for about 1 hr until injection. For micromanipulation, ESCs or iPSCs were trypsinized and suspended in ESC or iPSC culture medium. A piezo-driven micromanipulator (Prime Tech, Tokyo, Japan) was used to drill zona pellucida and trophectoderm under the microscope and 10–15 ESCs or iPSCs were introduced into blastocyst cavities near the inner cell mass. After blastocyst injection, embryos underwent follow-up culture for 1–2 hr. Mouse blastocysts then were transferred into the uteri of pseudopregnant recipient ICR female mice (2.5 dpc) and rat blastocysts were transferred into the uteri of pseudopregnant recipient Wistar female rats (3.5 dpc).