Embryonic and adult stem cell systems in mammals: Ontology and regulation


  • Keiichi Katsumoto,

    1. Department of Stem Cell Biology, Institute of Molecular Embryology and Genetics (IMEG), Kumamoto University, Honjo 2-2-1, Kumamoto 860-0811
    2. The Global COE Cell Fate Regulation Research and Education Unit, Kumamoto University, Honjo 2-2-1, Kumamoto 860-0811, Japan
    Search for more papers by this author
  • Nobuaki Shiraki,

    1. Department of Stem Cell Biology, Institute of Molecular Embryology and Genetics (IMEG), Kumamoto University, Honjo 2-2-1, Kumamoto 860-0811
    Search for more papers by this author
  • Rika Miki,

    1. Department of Stem Cell Biology, Institute of Molecular Embryology and Genetics (IMEG), Kumamoto University, Honjo 2-2-1, Kumamoto 860-0811
    2. The Global COE Cell Fate Regulation Research and Education Unit, Kumamoto University, Honjo 2-2-1, Kumamoto 860-0811, Japan
    Search for more papers by this author
  • Shoen Kume

    Corresponding author
    1. Department of Stem Cell Biology, Institute of Molecular Embryology and Genetics (IMEG), Kumamoto University, Honjo 2-2-1, Kumamoto 860-0811
    2. The Global COE Cell Fate Regulation Research and Education Unit, Kumamoto University, Honjo 2-2-1, Kumamoto 860-0811, Japan
    Search for more papers by this author

*Author to whom all correspondence should be addressed.
Email: skume@kumamoto-u.ac.jp


Stem cells are defined as having the ability to self-renew and to generate differentiated cells. During embryogenesis, cells are initially proliferative and pluripotent and then they gradually become restricted to different cell fates. In the adult, tissue stem cells are normally quiescent, but become proliferative upon injury. Knowledge from developmental biology and insights into the properties of stem cells are keys to further understanding and successful manipulation. Here, we first focus on ES cells, then on embryonic development, and then on tissue stem cells of endodermally derived tissues, particularly the liver and pancreas.


Stem cells studies in mammals have attracted more recent attention than other species, because understanding the properties of stem cells in mammals may provide insights into the biology of development and open the way for possible treatments for regenerative diseases, which could be applicable to humans. The identification and characterization of embryonic stem (ES) cells in mice has contributed much to the genetic engineering and understanding of gene functions. Recent progresses have shown that ES cells could be manipulated to generate diverse cell types in vitro, derived from all three germ layers (Kume 2005b). ES cells provide a useful tool for in vitro dissection of early inductive processes (Shiraki et al. 2009; Shiraki et al. 2005; Shiraki et al. 2008a; Shiraki et al. 2008b). ES cell derived differentiated cells are useful as a cell source for analysis of a certain cell type (Yoshida et al. 2009; Yoshida et al. 2008). Recent establishment of induced pluripotent stem (iPS) cells from somatic cells in mouse and human (Takahashi & Yamanaka 2006; Takahashi et al. 2007) further emphasizes their potential to derive patient-specific ES cell equivalents (Takahashi et al. 2007; Yu et al. 2007; Park et al. 2008). These iPS cells seemingly give fewer ethical concerns than ES cells, which have to be derived from living embryos. However, more research is still required to reveal the nature and the therapeutic usefulness of iPS cells.

Embryonic stem cells derive from preimplantation embryos. During embryonic development, the differentiation potency gradually becomes limited. The progenitor cells of a particular tissue often have limited ability to self-renew and are not easily isolated for detailed studies in vitro. However, knowledge of the progenitor cells will shed light on the origin and nature of the stem cells and will be beneficial for manipulating them. To understand the biology of stem cells, we try to address the following questions. Where do they come from? How do they differentiate into specific cell types? How are the senescence, proliferation and differentiation of stem cells controlled? What are the stem cell niches and the key mobilizing signals? During embryogenesis, cells are initially proliferative and pluripotent and then they gradually become restricted to different cell fates. Knowledge from developmental biology and insights into the properties of stem cells are keys to further understanding and successful manipulation. Here, we first focus on ES cells, then on embryonic development, and then on tissue stem cells of endodermal derived tissues such as the liver and pancreas.

ES cells

Establishment of ES cells

Embryonic stem cells were first isolated in 1981 by several independent groups from the inner cell mass (ICM) of mouse blastocysts at 3.5 days post coitum (dpc) (Evans & Kaufman 1981; Martin 1981). Embryonic carcinoma (EC) cells were the first pluripotent stem cells to be derived from mouse embryonic or fetal tissues, which could participate in entirely normal development following their introduction into the blastocysts. EC cells were found to be able to contribute to most if not all organs and tissue of the resulting offspring. Before the development of ES cell technology, it was reported that EC cells derived from mouse teratocarcinomas, could be cultured in the undifferentiated state in vitro, and a single EC cell could regenerate a tumor containing both EC cells and differentiated cells (Kleinsmith & Pierce 1964). However, the effectiveness of chimerism was often weak and EC-derived tumors were a common feature of these embryos (Rossant & Papaioannou 1984). As one of the applications for ES cells, Gossler et al. (1986) described the ability and advantages of using ES cells to produce transgenic animals (Gossler et al. 1986). Thomas & Capecchi (1987) showed that one could alter the genome of ES cells by homologous recombination (Thomas & Capecchi 1987). Koller later demonstrated that ES cells, modified by gene targeting when reintroduced into blastocysts, could transmit genetic modifications through the germline (Koller et al. 1989). Today, genetic modification of the murine genome by ES cell technology is a powerful approach to understanding the function of mammalian genes in vivo. Moreover, the development of human ES cells led to a great increase of interest in the biology and potential therapeutic application of ES cells (Shamblott et al. 1998; Thomson et al. 1998).

ES cells in other species

Embryonic stem cells can be established from species of vertebrates other than mouse or human, such as rat, hamster, rabbit, dog, cat, pig, sheep, cow, horse, marmoset, rhesus monkey, chicken, medaka and zebrafish. They can proliferate while maintaining an undifferentiated morphology and expressing at least some ES cell markers. They can differentiate into a variety of cell types in vitro and produce teratomas when transplanted in vivo. ES cells in the hamster and rabbit are very similar to mouse ES cells, but ES cells of most other mammals are not. On the other hand, human ES cells form colonies closely resemble those of ES cells of other primates. In the marmoset, rhesus monkey, and human, ES cells not only form relatively flat colonies, but also exhibit a number of differences from mouse ES cells in the markers they express. Generally, strategies for establishing ES cell lines in other species initially followed the conditions that have been proved successful in the mouse ES cells, such as the use of growth-inactivated feeder cells and either leukemia inhibitory factor (Lif) or a related cytokine. The differences in morphology and growth factor demands between the mouse and human ES cells are thought to due to the differences in their origins: the developmental state of human ES cells is similar to that of the late mouse epiblast (see next section).

In addition to EC of ES cells, a third type of pluripotent stem cell, called embryonic germ (EG) cell, has been isolated from mouse, pig, cow, human and chicken. Primordial germ cells (PGCs) form ES-like colonies when grown on feeder cells in the presence of serum and Lif, basic fibroblast growth factor (bFGF) and stem cell factor (SCF). Like ES cells, EG cells show a full developmental potency, being able to differentiate into all three germ layer-derived cells in vitro and form teratoma in vivo. Overall, ES or EG cells are valuable resources for both basic and applied research, and the evaluation on the similarities or differences among species will clarify the stem cell system.

Self-renewal is the ground state of ES cells

Mouse ES cells are derived and maintained by using a combination of the cytokine Lif to activate the signal transducer and activator of transcription 3 (Stat3), and either serum or bone morphogenetic protein (Bmp) to induce proteins that inhibit differentiation (Ying et al. 2003). Moreover, the elimination of differentiation-inducing signaling mediated by mitogen-activated protein kinase (MAPK) by autocrine fibroblast growth factor 4 (Fgf4) enabled self-renewal of ES cells. Additional inhibition of glycogen synthase kinase 3 beta (Gsk3b) consolidated biosynthetic capacity and suppressed residual differentiation (Ying et al. 2008). ES cells can be expanded efficiently and maintained under EC triple inhibitor or so-called ‘3i conditions’ (Ying et al. 2008). This involves treatment with three chemical inhibitors: SU5402, an Fgf receptor inhibitor; PD184352, a MEK inhibitor and CHIR99021, a Gsk3b inhibitor. The 3i conditions do not have any requirement for growth factors such as Lif or Bmp. Thus, self-renewal turns out to be an intrinsic property of ES cells. Figure 1 shows the schematic drawing of the self-replication of the ES cells. The above finding that ES cell self-renewal is the ground state and the use of the 3i condition would bring us a strong tool in establishing ES cell lines in species that were thought to be difficult.

Figure 1.

 Schematic representation of the self-replication of the pluripotent state. The source of differentiation-inducing signals, Phospho-Erk (pERK), is either inhibited upstream by chemical inhibitors (A), or blocked downstream by leukemia inhibitory factor (Lif) and bone morphogenetic protein (Bmp) signaling (B). Adapted from Ying et al. 2008.

Human ES cells show distinctive characteristics that differ from mouse ES cells. They grow as a monolayer sheet and their growth rate is slower than that of feeder cells. Moreover, human ES cells require Fgf2 and Activin/Nodal signals instead of a Lif signal for self-renewal. Epiblast and the primitive ectoderm of 6.0 dpc mouse embryo are the sources of the other type of pluripotent stem cells, epiblast stem cells (EpiSCs) (Brons et al. 2007; Tesar et al. 2007). Because mouse EpiSCs have similar characteristics to human ES cells, it is likely that the developmental state of human ES cells is similar to that of the late mouse epiblast, rather than the early epiblast. Therefore, in any study of the mechanisms that maintain self renewal, careful comparisons of mouse ES cells, EpiSCs and human ES cells will be required.

Transcriptional circuitry for maintaining self-renewal of ES cells

The POU domain class 5, transcription factor 1 (Pou5f1, also known as Oct3/4), Nanog homeobox (Nanog) and sex determining region Y (SRY) box containing gene 2 (Sox2) have been suggested to form a core network that controls the maintenance of pluripotency of mouse ES cells, albeit in distinct pathways (Nichols et al. 1998; Avilion et al. 2003; Chambers et al. 2003; Mitsui et al. 2003). One role of these factors is to counteract molecules that are involved in the differentiation program. In fact, Pou5f1 and Sox2 form a complex that suppresses the expression of caudal type homeo box 2 (Cdx2), which directs differentiation to the primitive endoderm (Niwa et al. 2005; Niwa 2007). Sox2 affects the maintenance of Pou5f1 expression (Masui et al. 2007). Nanog can inhibit the differentiation of primitive endoderm (Mitsui et al. 2003) and mesoderm (Suzuki et al. 2006) by influencing the expression of GATA binding protein 6 (Gata6) and brachyury (T), respectively. Therefore, suppression of differentiation programs is essential for the maintenance of ES cell self-renewal. Niwa et al. (2009) reported that the Lif signaling pathways maintain self-renewal of ES cells by regulating the above core circuitry via Kruppel-like factor 4 (gut) (Klf4) and T-box 3 (Tbx3) (Niwa et al. 2009). Molecules regulating ES cell self-renewal are summarized in Figure 2. Mouse ES cells depend on Lif and Bmp, whereas human ES and EpiSCs rely on Activin and Fgf2 for maintaining their pluripotency. Moreover, Nanog expression is maintained by the Smad2/3 pathway under the control of Activin in EpiSCs (Vallier et al. 2009).

Figure 2.

 Molecular mechanisms regulating embryonic stem (ES) cell self-renewal. Leukemia inhibitory factor (LIF) signaling pathways maintain pluripotency through a core circuit consisting of Sox2, Nanog and Oct3/4. The effectors are also shown. Adapted from Niwa et al., 2007; Niwa et al. 2009.

Cell cycle regulation of ES cells

Proliferation of ES cells is accomplished by self-renewal as in other cell types. However, cell cycle regulation of ES cells is unusual. ES cells transit through the cell cycle much faster than differentiated cells, mainly via a shortened G1 phase. This lasts only 2–4 h in mouse and human ES cells, which is much shorter than mouse embryonic fibroblasts, in which the G1 phase lasts 15–20 h.

Retinoblastoma 1 (Rb1, also known as Rb) and transformation related protein 53 (Trp53, also known as p53) are key senescence-inducing factors. The activity of Rb1 is regulated by phosphorylation. During progression through G1, Rb1 is phosphorylated sequentially by cyclins and cyclin-dependent kinase complexes (Cdks), which induces a partial release of the critical transcription factor E2F and leads to the activation of target genes and the entry into S phase.

To maintain the stemness of ES cells, a distinct cell cycle control is operated. There is a defective Rb pathway by hyperphosphorylation (Savatier et al. 1994), a constitutively high activity of cyclin E/A-associated kinases and a lack of expression of major Cdk inhibitors (Stead et al. 2002), a resistance to the growth inhibitory activity of cyclin-dependent kinase inhibitor 2A (p16INK4a), which is the inhibitor of Cdk4 (Savatier et al. 1996) and a lack of Trp53-dependent cell-cycle arrest and apoptosis (Aladjem et al. 1998). Myeloblastosis oncogne-like 2 (Mybl2, also known as B-Myb) has been implicated in the self-renewal of ES cells (Tanaka et al. 1999) and is essential for normal cell cycle progression and chromosomal stability. Reciprocal activation of Pou5f1 and Mybl2 might be important for the combination of pluripotency-associated transcription and cell cycle regulation (Tarasov et al. 2008). Yamanaka’s group demonstrated that a member of the Ras gene family, ES-cell-expressed Ras (Eras), is involved in self-renewal and teratoma formation by ES cells (Takahashi et al. 2003), probably through the activation of the phosphoinositide 3-kinase (PI3K) pathway (Takahashi et al. 2005). Cartwright et al. (2005) reported that the proto-oncogene myelocytomatosis oncogene (Myc, also known as c-myc) is downstream of the Lif-signal transducer and activator of Stat3 pathway, which is required for ES-cell maintenance (Cartwright et al. 2005). Unlimited proliferation requires the maintenance of telomerase: ES cells have a constitutive telomerase activity and loss of this activity results in limited growth.

Reprograming of somatic cells to pluripotent cells

iPS cells, which are derived from a variety of tissues, has brought into the stem cell field to understand the mechanism of pluripotency. In August 2006, it was demonstrated that pluripotent ES-like stem cells could be generated from both embryonic and adult mouse fibroblasts by retroviral transduction of four genes, Oct4, Sox2, Klf4 and c-Myc (Takahashi & Yamanaka 2006). To date, a large number of studies on iPS derivation have been reported. Besides mouse and human iPS cells, there are reports on the establishment of iPS cells from other mammals: monkey, rat, pig and canine (Liu et al. 2008; Li et al. 2009; Shimada et al. 2009; Wu et al. 2009). When the process of mouse iPS derivation was monitored at different time-points for pluripotency marker gene expression, alkaline phosphatase was first, and then SSEA1, and finally Nanog and Oct4 were found to be switched on in a consecutive temporal order (Brambrink et al. 2008). These data suggest that the reprogramming is a defined and gradual, rather than chaotic event. Moreover, the experiment using cells transduced with inducible lentiviral vectors encoding the above four factors showed that exogenous factors are required for about 10 days, after which cells enter a self-sustaining pluripotent state (Stadtfeld et al. 2008). In August 2009, Hans R. Schöler’s group reported that Oct4 alone is required and sufficient to directly reprogram human neural stem cells to pluripotency (Kim et al. 2009). Like the above reports, iPS research will give helpful information to understand the mechanism governing pluripotency in mammal cells.

Ontogeny of the endodermally derived liver and pancreas: Identification of progenitors and signals regulating their differentiation


Progenitors The genes for albumin (Alb), transthyretin (Ttr) and α-fetoprotein (Afp) are among the earliest markers expressed during mammalian hepatic differentiation. These genes start to be expressed at the 7-somite stage in mice (8.25 dpc) (Gualdi et al. 1996; Jung et al. 1999) and Alb/Afp double-positive hepatoblasts are bipotential progenitors that differentiate into hepatocytes and bile duct cells (Germain et al. 1988; Clotman et al. 2005). Fate mapping studies showed that the pre-liver region exists in two separate lateral domains of the ventral endoderm and in the medial ventral endoderm at an early somite stage. These liver progenitor cells translocate to the anterior medial ventral region and are brought together during development (Tremblay & Zaret 2005).

Fgf and Bmp Fgf, secreted from the adjacent cardiac mesoderm, is reported to induce hepatic gene expression in the ventral foregut endoderm (Jung et al. 1999; Rossi et al. 2001; Chen et al. 2003; Zhang et al. 2004; Serls et al. 2005; Shin et al. 2007). The prehepatic endoderm receives a low concentration of Fgf from the adjacent developing cardiac mesoderm, which permits liver differentiation through MAPK (Calmont et al. 2006). Later, the septum transversum develops between the liver primordium and cardiac mesoderm. This acts as a barrier against direct signaling to the liver primordium from the high concentration of Fgf secreted from the cardiac mesoderm. This higher concentration of Fgf induces lung development (Serls et al. 2005). The septum transversum secretes the Bmp 2 and Bmp 4, which are also important for generating liver progenitor cells.

Retinoic acid signaling Retinoic acid (RA) is well known to have a significant role in the anterior-posterior axis formation during development. In the zebrafish but not in the Xenopus and mouse system, inhibition of RA signaling blocks the endodermal cells from differentiating into a hepatic or pancreatic fate but not the initial endoderm differentiation (Stafford & Prince 2002). Ectopic RA signaling causes the expansion of early hepatic and pancreatic marker gene expressions in the endoderm. Relevant RA from the paraxial mesoderm is required in the endoderm (Stafford et al. 2006).

Wnt signaling It is reported that Wnt signaling has a critical role at the initiation of liver development. In the zebrafish Wnt2bb mutants, initial liver bud generation is inhibited, although liver rudiments do form in the ductal regions of pancreas at a later developmental stage (Ober et al. 2006). On the other hand, Wnt antagonist secreted frizzled-related sequence protein 5 (Sfrp5) is expressed in the foregut endoderm in both the mouse and Xenopus (Pilcher & Krieg 2002; Finley et al. 2003). Ectopic Wnt signaling in the foregut endoderm results in reducing liver-related genes and inhibiting liver bud development. However, overexpression of Wnt antagonist results in an expansion of liver bud domain (McLin et al. 2007). This discrepancy can not be explained in detail and more research is required to resolve this problem. One possibility is that an appropriate Wnt signaling level is required for liver development, so that the balance between Wnt signaling and the antagonist is very important to develop the liver, which is similar for RA signaling within the pancreas region (Kinkel et al. 2009).

Transcription factors Liver primordium is differentiated from the ventral endoderm (Zaret & Grompe 2008). Foxa1 and Foxa2 have crucial roles in liver development. In Foxa1 and Foxa2 double-knockout mice, no liver bud is generated and Afp cannot be detected in the foregut endoderm. Furthermore, in the endoderm culture experiments of Foxa1 and Foxa2 double-deficient mice, Ttr and Alb can not be induced, regardless of the presence of Fgf2. (Lee et al. 2005). Therefore, Foxa1 and Foxa2 expression establish the competence in the foregut endoderm to react to the hepatic-inducing signals. An early hepatoblast morphogenic event, from a cuboidal to a columnar shape generating pseudostratified epithelium, is controlled by the homeobox transcription factor gene Hex (Bort et al. 2006). The basal lamina is broken down and the cells move into the surrounding stroma and proliferate. Hepatoblast progenitor cells in the stroma receive signals from the endothelial cells to undergo maturation (Matsumoto et al. 2001). Moreover, prospero-related homeobox 1 (Prox1) is involved in the delamination of hepatoblasts, and Tbx3 has a role in mediating proliferation and is required for hepatoblast migration (Sosa-Pineda et al. 2000; Ludtke et al. 2009). The liver progenitor cells differentiate into ductal cells when Notch signaling is activated. Onecut1 (also known as OC1 or HNF6) is the regulator of biliary development and the hepatic nuclear factor 4, alpha (Hnf4a) is expressed in parenchymal cells (Zong et al. 2009). The cell lineages and molecules involved in liver development are summarized in Figure 3.

Figure 3.

 Molecules involved in pancreatic and hepatic development. BMP, bone morphogenetic protein; FGF, fibroblast growth factor; RA, retinoic acid; VEGF, vascular endothelial growth factor.


The pancreas is composed of endocrine, exocrine and ductular cells. The exocrine cells make up about 95–99% of the total pancreas. The endocrine cells, which compose the islets of Langerhans scattered in the exocrine tissue, make up about 1–5% of the total pancreas in the mouse (Bonal & Herrera 2008). The islets of Langerhans are composed of α, β, δ, ε and pancreatic peptide cell. These produce hormones such as glucagon, insulin, somatostatin, pancreatic polypeptide and ghrelin. Each islet is made up from a central core of β cells surrounded by the other endocrine cells. Insulin downregulates blood glucose level, whereas glucagon upregulates it. Ghrelin and pancreatic polypeptide are orexigenic hormones, and somatostatin controls the secretion of insulin, glucagon and pancreatic polypeptide (Bonal & Herrera 2008).

Pancreas progenitors The pancreas starts to differentiate from the foregut/midgut junction of the endoderm at the 6–10 somite stage (E 8.5) in the mouse. The earliest pancreatic marker gene, Pancreatic and duodenal homeobox1 (Pdx1) (Kume 2005a; Offield et al. 1996), is expressed in the dorsal and ventral pancreatic bud and a portion of the stomach and duodenal endoderm. The dorsal pancreas is differentiated from the dorsal endoderm, which expresses Sox17, Foxa2 and OC1, and the ventral pancreas is generated from the ventral endoderm, which expresses Sox17, Foxa1, Foxa2, Gata4, Gata6 and Hnf1b (Zaret 2008; Zaret & Grompe 2008). The dorsal pancreatic progenitor cells first appear near Hensen’s node immediately after the completion of gastrulation, whereas the ventral pancreas progenitor cells begin to appear at the somite border at the 4-somite level near the vitelline vein at the 17 somite stage in the chick embryo (Katsumoto et al. 2009; Matsuura et al. 2009). The Pdx1 expression is detected in the pre-dorsal pancreas region at the 10 somite stage and in the pre-ventral pancreas region at the 12 somite stage by whole mount in situ hybridization analysis in the chick embryo (Katsumoto et al. 2009; Matsuura et al. 2009). Hnf1b is required for the initial ventral pancreatic bud generation but not for dorsal pancreas bud generation (Haumaitre et al. 2005). In the absence of Gata4, the development of ventral pancreas and the liver was perturbed (Watt et al. 2007). Lineage tracing experiments have shown that all pancreatic cells are derived from the Pdx1-expressing precursor cells (Gu et al. 2003). Pdx1-deficient mice do form a pancreatic bud, but development is arrested at an early somite stage. Although Pdx1 plays a key role in pancreatic differentiation (Kume 2005a; Offield et al. 1996), Pdx1 is not a determination factor because an ectopic pancreas is not generated when Pdx1 is overexpressed in the pre-stomach or pre-intestinal regions in the chick embryo (Grapin-Botton et al. 2001). In contrast, the ventral pancreas arises from the ventral foregut endoderm, where bipotential progenitor cells can differentiate into either liver or pancreas at around E8.5 (8–10 somite stage). Fgf signals from the cardiac mesoderm induce liver differentiation, while suppression of the Fgf signaling in the ventral foregut endoderm permits development of the ventral pancreas. These observations suggest that the default state of the ventral foregut endoderm is for pancreatic development (Deutsch et al. 2001).

RA signaling It was reported that in the zebrafish, aldehyde dehydrogenase family 1, subfamily A2 (Aldh1a2 or Raldh2) mutant neckless (nls) and treated with retinoic acid receptor (RAR) inverse agonist BMS493 showed a lack in both the endocrine and exocrine pancreatic cell types and markers of pancreatic progenitors (Stafford & Prince 2002). Furthermore, the late gastrula embryos treated with RA expanded the pancreatic field to the anterior region, but not to the posterior region. This critical role of RA is conserved in at least the Xenopus, quail and mouse (Chen et al. 2004; Stafford et al. 2004; Martin et al. 2005; Molotkov et al. 2005). It was shown that in the Raldh2 mutant mice, Pdx1 and Prox1 expression were inhibited and the pancreas could not be generated in the dorsal endoderm region. However, the ventral pancreas development is not perturbed. Interestingly, in zebrafish embryo, RA is required for both pancreas and liver development. However, in the mouse embryo, RA is required for only the dorsal pancreas development. RA treatment is an absolutely necessary component for the differentiation of human or mouse stem cells to pancreatic cell fate (D’Amour et al. 2006; Jiang et al. 2007; Shim et al. 2007; Shiraki et al. 2008b). It was also reported that RA have a critical role in anterior-posterior regionalization within the endoderm in the chick embryo (Bayha et al. 2009) and Cyp26 enzymes, which are retinoic acid catabolic enzymes, play a critical role in defining the anterior limit of the pancreatic region (Kinkel et al. 2009). On the other hand, posterior limit of the pancreatic region is controlled by transcription factor caudal type homeo box 4 (Cdx4) (Kinkel et al. 2008).

Activin and Fgf2 Activin and Fgf2 (also known as bFgf) are thought to be factors secreted from the notochord that direct dorsal pancreatic morphogenesis at around Hamburger–Hamilton stage 12 (16 somite stage) in the chick embryo and maintain the expression of early pancreatic genes through the repression of sonic hedgehog (Shh) expression (Hebrok et al. 1998). Because the notochord cannot induce the posterior non-pancreatic endoderm to express pancreatic marker genes, notochord signals are considered to be permissive rather than instructive (Kim et al. 1997). Mice deficient for activin receptor IIA (Acvr2a, also known as ActRIIa) and activin receptor IIB (Acvr2b, also known as ActRIIB) express Shh in the prospective dorsal pancreas region and pancreatic development is disrupted (Kim et al. 2000).

Vascular endothelial growth factor A Vascular endothelial growth factor A (Vegfa, also known as Vegf) is well known as a maturation signal for the dorsal pancreas and potentiates insulin expression within the pancreatic endoderm (Lammert et al. 2001). The first signs of morphological change (budding from the endoderm) occur at the 22–25 somite stage (E9.5) in the mouse. Mice deficient in the Vegfa receptor gene, kinase insert domain protein receptor (Kdr, also known as Flk1), exhibit disruption in the formation of blood vessels and the dorsal mesenchyme (Yoshitomi & Zaret 2004). In Kdr-deficient mice, in which mice aorta endothelial cells are immature, Pdx1 expression is normal at the 20–25 somite stage but pancreas specific transcription factor, 1a (Ptf1a) expression is reduced in the dorsal but not in the ventral pancreas. Hence, mature aortal endothelial cells induce the expression of Ptf1a in the dorsal pancreas, which is required for the maintenance of Pdx1 expression. An overlapping expression of Pdx1 and Ptf1a defines the dorsal pancreas region in the Xenopus embryo (Afelik et al. 2006).

Islet 1, N-cadherin and FGF10 The dorsal mesenchyme, which surrounds the dorsal pancreas, has a very significant role in the development of the dorsal pancreas. Mice deficient in ISL1 transcription factor, LIM/homeodomain (Isl1) or N-cadherin display a lack of dorsal mesenchyme and abnormality in the dorsal pancreas (Esni et al. 2001). Fgf10 is known to be secreted from the dorsal mesenchyme and promotes the accumulation of Pdx1-positive pancreatic progenitor cells (Jacquemin et al. 2006).

Notch signaling pathways One of the Notch signaling target genes, hairy and enhancer of split family (Hes), represses the expression of Neurog3, an endocrine progenitor marker gene and cyclin-dependent kinase inhibitor 1C (P57) (Cdkn1c), regulates the cell cycle (Lee et al. 2001; Georgia et al. 2006). All pancreatic endocrine cells are generated from Neurog3-positive cells (Gu et al. 2002, 2003). Neurog3 null mice completely lack pancreatic endocrine cells and show defects in acinar morphogenesis. In Neurog3 null mice, neurogenic differentiation 1 (Neurod1 also known as NeuroD and BETA2) expression also disappeared. In the Neurod1-deficient mice, islets and acinar cells were abnormal. Therefore, it would appear that the defects in acinar cells of Neurog3 null mice are a consequence of the loss of Neurod1 expression (Naya et al. 1997; Gradwohl et al. 2000). Overexpression of Neurog3 or of the intracellular form of Notch 3 (a repressor of Notch signaling) (Notch 3-ICD) leads to premature endocrine cell differentiation at the expense of the exocrine lineage (Apelqvist et al. 1999). On the other hand, overexpression of Pdx1-Notch 1-ICD leads to reduced numbers of Neurog3-positive and endocrine cells, and diminished acinar differentiation (Hald et al. 2003; Murtaugh et al. 2003). Mice deficient in recombination signal binding protein for immunoglobulin kappa J region (Rbpj), delta-like 1 (Drosophila) (Dll1), or Hes1 exhibit increased numbers of Neurog3-positive cells, immature endocrine differentiation and a ‘plastic’ pancreas (Apelqvist et al. 1999; Jensen et al. 2000; Fujikura et al. 2006). These results suggest that Notch signaling in the early phase of pancreatic development has an important role for maintaining the undifferentiated state of Neurog3-positive progenitor cells.

Other intrinsic molecules It has been suggested that the numbers of Pdx1-positive progenitor cells between E8.5 and E12.5 determines the size of the pancreas (Stanger et al. 2007). Furthermore, in explant experiments, Pdx1 deficient epithelium cannot mature in the presence of wild type mesenchyme. Thus Pdx1 might provide competency to the pancreatic epithelium to respond to growth factors from the mesenchyme (Ahlgren et al. 1996). From E9.5, epithelial budding starts branching morphogenesis. Multipotent progenitor cells, which can became endocrine cells, exocrine cells or ductal cells exist at the tips of the branches. These cells are marked by Pdx1, Ptf1a, carboxypeptidase A1 (Cpa1) and c-myc (Zhou et al. 2007).

Many genes have been shown to be involved in endocrine cell fate determination through gene knockout or transgenic mouse studies. Genes implicated in specification of the four endocrine cell types are shown in Figure 3 (Kume 2005a; Kaestner et al. 1999; Shih et al. 1999; Lee et al. 2002; Jorgensen et al. 2007; Bonal & Herrera 2008). The cell lineages and molecules involved in pancreatic development are summarized in Figure 3.

Adult endodermal stem cells

The main role of an adult stem cell is the maintenance and repair of its own tissue type. Adult stem cells have been defined in high cell turnover organs, including the skin and intestine, and low turnover organs such as the kidney and brain (Potten & Morris 1988; Galli et al. 2003; Oliver et al. 2004). Among the endodermally derived tissues, the liver and pancreas have a low cell turnover rate (Finegood et al. 1995; Michalopoulos & DeFrances 1997). The presence of liver adult stem cells and pancreatic adult stem cells has been discussed in several publications (Vessey & de la Hall 2001; Fausto & Campbell 2003; Fausto 2004; Bonner-Weir & Weir 2005; Guo & Hebrok 2009). On the other hand, the intestines exhibit a high turnover rate (Barker et al. 2008). Here we will focus on adult stem cells of the liver and pancreas, in the normal situation and in regeneration upon injury.

Adult liver stem cells

The adult liver plays an essential role in homeostasis, metabolism and detoxification throughout adult life (Michalopoulos & DeFrances 1997). These functions are carried out by hepatocytes, which constitute 80% of the liver cells (Koniaris et al. 2003). The liver shows a high regenerative potential after injury, which differs from a low turnover in the normal state (Koniaris et al. 2003). Thus, the liver-to-body mass ratio is maintained constant. The existence of liver stem cells is not clearly defined. Rapid regeneration of liver tissue is associated with replication of mature hepatocytes and oval cells and/or bone marrow cells are involved in regeneration under conditions when hepatocyte proliferation is suppressed (Fausto & Campbell 2003; Koniaris et al. 2003). Figure 4 shows the schematic drawing of hypothesized cell sources that are involved in liver regeneration.

Figure 4.

 A schematic representation of the hypothesized cell sources involved in liver regeneration.

Replication of hepatocytes Partial hepatectomy (PH) in the adult rat has been used to study hepatic regeneration (Higgins & Anderson 1931). After 70% PH, normal liver tissue size recovered completely. Likewise, the human liver regenerates after PH and reaches its original size by 3–6 months (Court et al. 2002). DNA synthesis in hepatocyte peaks at 24 h after PH and cell division is terminated by 2–3 days (Michalopoulos & DeFrances 1997). Treatment with carbon tetrachloride (CCl4) induces necrosis of hepatocytes and does not reduce liver mass (Reynolds 1963; LeSage et al. 1999). After CCl4 treatment, the surviving hepatocytes replicate rapidly and maintain normal liver functions (Greenbaum et al. 1995; Czaja 1998). These results indicate that hepatocytes can proliferate and contribute to regeneration of the liver.

Oval cells Several methods are available to induce oval cells. Effective methods involve inhibiting proliferation of hepatocyte by exposure to 2-acetylaminofluorene (2-AAF), then subjecting the animals to PH to induce oval cell proliferation. Other approaches that do not include PH, such as treatment with a choline-deficient, ethionine-supplemented diet induces oval cells in mouse liver (Farber 1956; Tatematsu et al. 1984; Spelman et al. 1986; Lemire et al. 1991; Steinberg et al. 1991; Factor et al. 1994; Petersen et al. 1998; Akhurst et al. 2001; Paku et al. 2001). Oval cells have an ovoid nucleus and a high nucleus-to-cytoplasmic ratio (Farber 1956). Oval cells express cholangiocyte markers of keratin 19 (Krt19, also known as cytokeratin 19) and gamma-glutamyltransferase 1 (Ggt1), a hepatoblast/immature hepatocyte marker of Afp and a hepatocyte marker of Alb (Fausto & Campbell 2003). They also express hematopoietic stem cell markers such as kit oncogene (Kit, also known as c-KIT), CD34 antigen (Cd34) and thymus cell antigen 1, theta (Thy1) (Fujio et al. 1994; Omori et al. 1997; Petersen et al. 1998; Matsusaka et al. 1999). Oval cells are thought to give rise to ductal epithelial cells in the canals of Hering because these cells express both Krt19 and Afp in the adult liver (Factor et al. 1994).

Bone marrowPetersen et al. (1999) demonstrated that donor-derived bone marrow could repopulate the liver in rats treated with 2-AAF (Petersen et al. 1999). Wang et al. (2003) reported that bone-marrow-derived hepatocytes could repopulate the liver of mice with fumarylacetoacetate hydrolase deficiency, which represents a model of tyrosinemia type 1 with marked liver disease. Furthermore, karyotyping demonstrated that the repopulated hepatocytes arose from cell fusion between host hepatocytes and donor-derived bone marrow cells (Wang et al. 2003).

Adult pancreatic stem cells

In the normal pancreas, few β cells replicate during the adult life and it is a slow cell turnover organ (Finegood et al. 1995). The β cell mass is maintained during postnatal life in response to physiological damage and stresses such as disease, aging and pregnancy. Physiological stimuli can induce the proliferation of β cells by 10-fold compared with the normal pancreas in rats (Parsons et al. 1992). Likewise, a 30-fold increase in β cells was observed in mice resistant to insulin (Bruning et al. 1997). Pancreatic regeneration models have shown that pancreatic duct ligation induces robust β cell hyperplasia. Moreover, pancreatitis induces exocrine damage; pancreatectomy induces regeneration of all pancreatic cell types and streptozotocin (STZ) treatment induces regeneration of β cells (Risbud & Bhonde 2002). Here, previous reports suggested that β cells could be regenerated from progenitors within the islet, ducts or acini, or by replication of the β cells themselves are summarized. Figure 5 shows a schematic drawing of the cell sources for pancreas regeneration.

Figure 5.

 A schematic drawing of the cell sources for pancreatic regeneration. Adapted from Bonner-Weir & Weir 2005. Genes used for genetic tracing are shown in parentheses.

Progenitor cells within islets Xu et al. (2008) demonstrated that the number of proliferating pancreatic cells increased 10-fold in a duct ligation model (Xu et al. 2008). The proliferating cells expressed Neurog3, which give rise to all endocrine cell types, including β cells, thereby suggesting the existence of stem cells within islets. However, Neurog3-positive stem cells have not been identified in the normal adult pancreas. On the other hand, Thyssen et al. (2006) observed the appearance of proliferating cells expressing insulin or glucagon in the ducts, after a loss of β cells following STZ-treatment to neonatal rats (Thyssen et al. 2006). Some glucagon-positive cells expressed Pdx1, which is a pancreatic progenitor marker. It was reported that genes normally associated with undifferentiated pancreatic progenitor cells are reactivated during regeneration in a caerulein treated regeneration model (Jensen et al. 2005).This therefore suggests that switching on immature markers is an indicator of the dedifferentiation state of cells turned into an immature cell type. However, whether the immature α cells are involved in the regeneration of β cells remains unknown.

Progenitor cells in ductsBonner-Weir et al. (1983) reported that, following 90% pancreatectomy in the rat, a replication of pre-existing endocrine or exocrine cells and an increase in the number of ductal structures occurred in focal regions (Bonner-Weir et al. 1983). In the regenerating area, all duct cells expressed Pdx1 transiently. Glucagon-positive cells and insulin-positive cells were observed (Bonner-Weir et al. 1993; Sharma et al. 1999), thereby suggesting that ductal cells might dedifferentiate and transdifferentiate into endocrine cells. On the other hand, carbonic anhydrase 2 (Car2, also known as CAII) is expressed in the pancreatic duct in adult, but not in the embryonic tubular epithelium. Using a mouse line bearing a Car2 promoter driving the expression of Cre recombinase (Car2-CreERTM), in a duct ligation regeneration model, both endocrine and exocrine cells were observed as the progenies of Car2-positive mature duct cells (Inada et al. 2008).

Progenitor cells in acini Analysis of Cre/loxP lineage tracing system using elastase or amylase promoters has demonstrated transdifferentiation of acinar cells into insulin secreting cells in vitro (Minami et al. 2005). The same group showed that acinar cells gave rise to insulin-expressing cells in a model of type 1 diabetes (Okuno et al. 2007). Baeyens et al. (2009) have indicated that Notch signaling is involved in the acinar to β cell conversion (Baeyens et al. 2009). However, another group demonstrated that mature acinar cells did not contribute to β cells, in a duct ligation and pancreatitis model, which suggested that elastase-expressing cells do not give rise to endocrine or duct cells in normal conditions in vivo (Desai et al. 2007).

Beta cell replication It has been shown that β cells are themselves the main source of new β cells, throughout adulthood as well as during regeneration after pancreatectomy (Dor et al. 2004; Teta et al. 2007). The mature β cells retain proliferative capacity and expansion of adult β cells mass occurs by simple replication. Using a transgenic mouse system for the specific and conditional ablation of β cells, Nir et al. (2007) reported that the surviving β cells were induced to proliferate during regeneration (Nir et al. 2007).

Pancreatic cancer stem cells p16INK4a is a negative regulator of the cell cycle and is associated with tumor suppression (Sharpless 2005; Kim & Sharpless 2006). Moreover, p16INK4a is inactivated in pancreatic intraepithelial neoplasia (PanIN); p16INK4a deficiency induced proliferation in the islets and the expression of p16INK4a increased with age (Krishnamurthy et al. 2006; Guerra et al. 2007; Ramsey et al. 2007). Bmi1 polycomb ring finger oncogene (Bmi1), which is one of the members of the polycomb repressive complex 1 (PRC1 complex), binds to the p16INK4a locus and regulates p16INK4a expression negatively (Molofsky et al. 2003; Leung et al. 2004). During β cell regeneration, the binding of Bmi1 to the p16INK4a locus led to a decrease in p16INK4a expression (Dhawan et al. 2009).


There has been dramatic progress in regenerative medicine in recent years. There has been increasing interest in the therapeutic application of stem cells, albeit numerous obstacles remain to be solved. Knowledge from studies of specific gene-deficient mice and transgenic approaches has led to better understanding of the molecular mechanisms in embryonic development. Identification of tissue-specific adult stem cells and revealing the mechanisms of how their differentiation and self-renewal are controlled will provide breakthroughs for the successful manipulation of stem cells (Kume 2005a,b). Here we have reviewed the mechanism of self-renewal of ES cells, the establishment and the developmental regulation of embryonic precursor cells and recent understanding of the adult stem cells of endodermal tissues. In the next few decades, we should be able to manipulate tissue stem cells in our bodies and be able to apply them in clinical settings.