embryonic cell differentiation

7/27/2019 Embryonic Cell Differentiation

1/28

10.1101/gad.1303605Access the most recent version at doi:2005 19: 1129-1155Genes Dev.

Gordon Kellerbiology and medicineEmbryonic stem cell differentiation: emergence of a new era in

References

http://genesdev.cshlp.org/content/19/10/1129.full.html#related-urlsArticle cited in:

http://genesdev.cshlp.org/content/19/10/1129.full.html#ref-list-1This article cites 269 articles, 119 of which can be accessed free at:

serviceEmail alerting

click heretop right corner of the article orReceive free email alerts when new articles cite this article - sign up in the box at the

CollectionsTopic

(105 articles)Cancer and Disease Models(109 articles)Development

Articles on similar topics can be found in the following collections

http://genesdev.cshlp.org/subscriptionsgo to:Genes & DevelopmentTo subscribe to

Cold Spring Harbor Laboratory Press

Cold Spring Harbor Laboratory Presson August 29, 2011 - Published bygenesdev.cshlp.orgDownloaded from
http://genesdev.cshlp.org/lookup/doi/10.1101/gad.1303605http://genesdev.cshlp.org/lookup/doi/10.1101/gad.1303605http://genesdev.cshlp.org/lookup/doi/10.1101/gad.1303605http://genesdev.cshlp.org/content/19/10/1129.full.html#related-urlshttp://genesdev.cshlp.org/content/19/10/1129.full.html#related-urlshttp://genesdev.cshlp.org/content/19/10/1129.full.html#related-urlshttp://genesdev.cshlp.org/content/19/10/1129.full.html#ref-list-1http://genesdev.cshlp.org/content/19/10/1129.full.html#ref-list-1http://genesdev.cshlp.org/content/19/10/1129.full.html#ref-list-1http://genesdev.cshlp.org/cgi/alerts/ctalert?alertType=citedby&addAlert=cited_by&saveAlert=no&cited_by_criteria_resid=genesdev;19/10/1129&return_type=article&return_url=http://genesdev.cshlp.org/content/19/10/1129.full.pdfhttp://genesdev.cshlp.org/cgi/alerts/ctalert?alertType=citedby&addAlert=cited_by&saveAlert=no&cited_by_criteria_resid=genesdev;19/10/1129&return_type=article&return_url=http://genesdev.cshlp.org/content/19/10/1129.full.pdfhttp://genesdev.cshlp.org/cgi/alerts/ctalert?alertType=citedby&addAlert=cited_by&saveAlert=no&cited_by_criteria_resid=genesdev;19/10/1129&return_type=article&return_url=http://genesdev.cshlp.org/content/19/10/1129.full.pdfhttp://genesdev.cshlp.org/cgi/collection/cancer_modelshttp://genesdev.cshlp.org/cgi/collection/cancer_modelshttp://genesdev.cshlp.org/cgi/collection/cancer_modelshttp://genesdev.cshlp.org/cgi/collection/cancer_modelshttp://genesdev.cshlp.org/cgi/collection/cancer_modelshttp://genesdev.cshlp.org/cgi/collection/cancer_modelshttp://genesdev.cshlp.org/subscriptionshttp://genesdev.cshlp.org/subscriptionshttp://genesdev.cshlp.org/subscriptionshttp://genesdev.cshlp.org/subscriptionshttp://genesdev.cshlp.org/subscriptionshttp://www.cshlpress.com/http://www.cshlpress.com/http://www.cshlpress.com/http://genesdev.cshlp.org/http://genesdev.cshlp.org/http://www.cshlpress.com/http://genesdev.cshlp.org/http://genesdev.cshlp.org/subscriptionshttp://genesdev.cshlp.org/cgi/collection/cancer_modelshttp://genesdev.cshlp.org/cgi/collection/developmenthttp://genesdev.cshlp.org/cgi/alerts/ctalert?alertType=citedby&addAlert=cited_by&saveAlert=no&cited_by_criteria_resid=genesdev;19/10/1129&return_type=article&return_url=http://genesdev.cshlp.org/content/19/10/1129.full.pdfhttp://genesdev.cshlp.org/content/19/10/1129.full.html#related-urlshttp://genesdev.cshlp.org/content/19/10/1129.full.html#ref-list-1http://genesdev.cshlp.org/lookup/doi/10.1101/gad.1303605


2/28

REVIEW

Embryonic stem cell differentiation:emergence of a new era in biologyand medicine

Gordon Keller1

Department of Gene and Cell Medicine, Mount Sinai School of Medicine, New York, New York 10029, USA

The discovery of mouse embryonic stem (ES) cells >20years ago represented a major advance in biology andexperimental medicine, as it enabled the routine ma-nipulation of the mouse genome. Along with the capac-ity to induce genetic modifications, ES cells provided thebasis for establishing an in vitro model of early mamma-lian development and represented a putative new sourceof differentiated cell types for cell replacement therapy.While ES cells have been used extensively for creatingmouse mutants for more than a decade, their applicationas a model for developmental biology has been limitedand their use in cell replacement therapy remains a goalfor many in the field. Recent advances in our under-standing of ES cell differentiation, detailed in this re-view, have provided new insights essential for establish-ing ES cell-based developmental models and for the gen-eration of clinically relevant populations for cell therapy.

Embryonic stem (ES) cells are pluripotent cells derivedfrom the inner cell mass of blastocyst-stage embryos(Evans and Kaufman 1981; Martin 1981). Their impor-tance to modern biology and medicine derives from twounique characteristics that distinguish them from allother organ-specific stem cells identified to date. First,they can be maintained and expanded as pure popula-tions of undifferentiated cells for extended periods oftime, possibly indefinitely, in culture. Unlike trans-formed tumor cell lines, ES cells can retain normalkaryotypes following extensive passaging in culture. Sec-ond, they are pluripotent, possessing the capacity to gen-erate every cell type in the body. The pluripotent nature

of mouse ES cells was formally demonstrated by theirability to contribute to all tissues of adult mice, includ-ing the germline, following their injection into host blas-tocysts (Bradley et al. 1984). In addition to their devel-opmental potential in vivo, ES cells display a remarkablecapacity to form differentiated cell types in culture

(Keller 1995; Smith 2001). Studies during the past 20years have led to the development of appropriate cultureconditions and protocols for the generation of a broadspectrum of lineages. The ability to derive multiple lin-eages from ES cells opens exciting new opportunities tomodel embryonic development in vitro for studying theevents regulating the earliest stages of lineage inductionand specification. Comparable studies are difficult in themouse embryo and impossible in the human embryo. Inaddition to providing a model of early development, theES cell differentiation system is viewed by many as anovel and unlimited source of cells and tissues for trans-plantation for the treatment of a broad spectrum of dis-eases. The isolation of human ES cells (hES) in 1998 dra-matically elevated the interest in the cell therapy aspectof ES cells and moved this concept one step closer toreality (Thomson et al. 1998). This review details the

current status of mouse and human ES cell differentia-tion from both the developmental biology and cell re-placement perspectives. The first sections of the reviewhighlight successes to date in the generation and char-acterization of mature populations, while the final sec-tion outlines the challenges for the future with a focuson the identification of progenitor cells representing theearliest stages of embryonic lineage development. Thereader is referred to other recent reviews that provideadditional details for many of the subjects covered here(Kyba and Daley 2003; Nir et al. 2003; Hornstein andBenvenisty 2004; Lang et al. 2004; Pera and Trounson2004; Rippon and Bishop 2004; West and Daley 2004).For the purpose of this review, the term ES will be used

in reference to mouse cells and hES for human cells.

Maintaining undifferentiated ES cells

ES cells were initially established and maintained by co-culture with mouse embryonic feeder cells (Evans andKaufman 1981; Martin 1981). Subsequent studies iden-tified leukemia inhibitory factor (LIF) as one of thefeeder-cell-derived molecules that plays a pivotal role inthe maintenance of these cells (Smith et al. 1988; Wil-liams et al. 1988; Stewart et al. 1992). In the presence ofappropriate batches of fetal calf serum (FCS), recombi-nant LIF can replace the feeder cell function and sup-

[Keywords: ES cells; differentiation; mesoderm; endoderm; ectoderm;embryonic development]1Correspondence.E-MAIL [email protected]; FAX (212) 803-6740.Article and publication are at http://www.genesdev.org/cgi/doi/10.1101/gad.1303605.

GENES & DEVELOPMENT 19:11291155 2005 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/05; www.genesdev.org 1129

http://www.cshlpress.com/http://www.cshlpress.com/http://www.cshlpress.com/http://genesdev.cshlp.org/http://genesdev.cshlp.org/http://www.cshlpress.com/http://genesdev.cshlp.org/


3/28

port the growth of undifferentiated ES cells (Smith et al.1988; Williams et al. 1988). Recently, Ying et al. (2003a)have uncovered a role for BMP4 in ES cell growth anddemonstrated that in the presence of LIF, it can replacethe requirement for serum. With these new develop-ments, it is now possible to grow ES cells with defined

factors in the absence of serum or feeder cells. Molecularanalyses have revealed that LIF functions through thegp130 activation of STAT3 (Niwa et al. 1998; Matsudaet al. 1999), whereas the effect of BMP4 on undifferenti-ated ES cells is mediated by Smad activation and thesubsequent induction of the helixloophelix Id factors.In addition to STAT3 and Id, two other transcriptionfactors, Oct3/4 (Niwa et al. 2000) and nanog (Chamberset al. 2003; Mitsui et al. 2003), have been shown to playpivotal roles in maintaining the undifferentiated state ofES cells. The role of these transcription factors in ES cellrenewal has been recently reviewed (Chambers andSmith 2004) and will not be discussed further here.

The regulation of hES cell growth is less well under-

stood and differs from that of the mouse in that LIF andSTAT3 appear to play no role in their self-renewal(Thomson et al. 1998; Reubinoff et al. 2000; Daheronet al. 2004). With current protocols, hES cells can bemaintained on feeder cells in serum-free medium supple-mented with bFGF (Amit et al. 2000). hES cells can alsobe grown in the absence of feeder cells, if cultured onmatrigel- or laminin-coated plates in medium supple-mented with mouse embryonic fibroblast conditionedmedium (MEF CM) (Xu et al. 2001). While not as welldefined as the conditions for the growth of mouse cells,this protocol does provide for relatively easy mainte-nance of hES cell populations. Cells grown in these con-ditions for >100 population doublings retained normal

karyotypes and stem cell characteristics, including theirin vitro and in vivo pluripotent differentiation potential.Recently, Sato et al. (2004) demonstrated that activationof the canonical Wnt pathway could replace the require-ment of MEF CM in the maintenance of undifferentiatedhES cells for short periods of time (57 d). Whether or notWnt signaling has an effect on hES cell self-renewal overlonger periods through multiple passages remains to bedetermined. hES cells do express both Oct4 (Ginis et al.2004) and nanog (Daheron et al. 2004; Richards et al.2004; Sato et al. 2004), suggesting that this aspect of theirregulation may be similar to that observed in mouse EScells. Future studies will no doubt define specific mol-ecules for the maintenance of hES cells and uncover themolecular mechanisms that regulate their self-renewal.

Differentiation of ES cells in culture

When removed from the factors that maintain them asstem cells, ES cells will differentiate and, under appro-priate conditions, generate progeny consisting of deriva-tives of the three embryonic germ layers: mesoderm,endoderm, and ectoderm (Keller 1995; Smith 2001).Wild-type ES cells do not differentiate to trophectodermin culture and, in this respect, reflect the potential oftheir founder embryonic population, the inner cell mass

(Fig. 1). hES cells differ from mouse cells in this respect,as when induced with BMP4, they will give rise to cellsthat display characteristics of the trophoblast lineage (Xuet al. 2002). The reason for this difference is not clear,but may indicate that at least some of the hES cell linesrepresent earlier stages of development than the compa-

rable mouse populations.Three general approaches, outlined in Figure 2, are

used to initiate ES cell differentiation. With the firstmethod, ES cells are allowed to aggregate and form three-dimensional colonies known as embryoid bodies (EBs)(Doetschman et al. 1985; Keller 1995). In the secondmethod, ES cells are cultured directly on stromal cells,and differentiation takes place in contact with thesecells (Nakano et al. 1994). The most commonly usedstromal cell line for such differentiation studies is OP9(Nakano et al. 1994), originally isolated from CSF-1-deficient op/op mice (Yoshida et al. 1990). The third pro-tocol involves differentiating ES cells in a monolayer onextracellular matrix proteins (Nishikawa et al. 1998).

All three approaches to ES cell differentiation are ef-fective and have specific advantages and disadvantages.EBs offer the advantage of providing a three-dimensionalstructure that enhances cellcell interactions that maybe important for certain developmental programs. Thecomplexity of the EBs can also be a disadvantage as thegeneration of cytokines and inducing factors withinthese structures can complicate interpretations of ex-periments in which one is trying to understand the sig-naling pathways involved in lineage commitment. Co-culture with stromal cells provides the beneficial growthpromoting effects of the particular cell line used. How-ever, undefined factors produced by these supportivecells may influence the differentiation of the ES cells to

undesired cell types. An additional problem with thismethod is the difficulty that can be encountered whenattempting to separate the ES-cell-derived cells from thestromal cells. Differentiation in monolayers on knownsubstrates can minimize the influence of neighboring

Figure 1. Scheme of early mouse development depicting therelationship of early cell populations to the primary germ layers.

Keller

1130 GENES & DEVELOPMENT



4/28


5/28

which later contains the developing aorta, gonads, andmesonephros (AGM) (Dieterlen-Lievre 1975; Russel1979; Godin et al. 1995; Medvinsky and Dzierzak 1996;Palis et al. 1999). Detailed analysis of hematopoietic de-velopment in the early embryo strongly suggests that theprograms generated in these two regions are different.

Hematopoietic commitment is detected first in the yolksac, where distinct blood islands appear, shortly follow-ing gastrulation (Moore and Metcalf 1970; Haar and Ack-erman 1971; Palis et al. 1995, 1999). These blood islandsconsist of an inner cluster of maturing erythrocytes sur-rounded by a layer of developing endothelial cells (Haarand Ackerman 1971). The erythroid cells within theseblood islands, known as primitive erythrocytes, are dis-tinct from fetal and adult erythrocytes in that they arelarge, circulate in the bloodstream as nucleated cells formuch of their life span, and contain an embryonic formof hemoglobin (Barker 1968; Brotherton et al. 1979; Rus-sel 1979; Kingsley et al. 2004). Production of primitiveerythrocytes is known as primitive erythropoiesis and is

restricted to the yolk sac during a narrow window ofdevelopment in the mouse embryo (Palis et al. 1999).Development of all other blood cell lineages includingmyeloid, fetal, and adult erythroid and lymphoid is re-ferred to as definitive hematopoiesis. Definitive ery-throid cells enucleate prior to entering the bloodstream,are smaller than those of the primitive lineage, and pro-duce adult forms of hemoglobin.

In addition to primitive erythrocytes, the yolk sac gen-erates a subset of lineages from the definitive hemato-poietic program including the macrophage, definitiveerythroid, and mast cell (Palis et al. 1999). Kinetic analy-sis of the developing yolk sac revealed that these lineagesare produced in a defined temporal pattern with primi-

tive erythroid and macrophage appearing first, followedby definitive erythroid, which, in turn, is followed bymast cells. While the yolk sac does have potential be-yond that of primitive erythropoiesis, it does not appearto be capable of generating lymphocytes or HSCs, whenanalyzed prior to the onset of circulation (Cumano et al.2001). Parallel studies have demonstrated that the hema-topoietic program initiated in the P-Sp includes the gen-eration of the myeloid, lymphoid, and definitive ery-throid lineages as well as the HSCs (Muller et al. 1994;Godin et al. 1995; Cumano et al. 2001). The P-Sp doesnot, however, generate primitive erythrocytes. Thus, thedistinguishing features of the early yolk sac are the gen-eration of the primitive erythroid lineage and a lack oflymphoid and HSC potential, while the P-Sp programcan be defined by the development of the lymphoid lin-eages and HSCs. Collectively, these observations indi-cate that the hematopoietic system initiates with theproduction of a limited number of specialized lineages inthe yolk sac and matures with time into a full multilin-eage system with the switch to the P-Sp. While some-what unusual, this pattern is logical, as the system isresponding to the requirements of the embryo at differ-ent developmental stages. These dramatic changes in thehematopoietic system, in particular the early and tran-sient appearance of the primitive erythroid lineage, pro-

vide a developmental map for monitoring hematopoieticcommitment in the ES cell differentiation cultures.

ES-cell-derived primitive and definitive hematopoiesis

In optimized culture conditions following serum induc-

tion, ES cells will undergo hematopoietic differentiation(Keller 1995). Hematopoietic commitment within thesecultures can be easily monitored by gene expression pat-terns (Schmitt et al. 1991; Keller et al. 1993; Robertson etal. 2000), the appearance of specific cell surface markers(Kabrun et al. 1997; Nishikawa et al. 1998), and the de-velopment of clonable progenitor cells (Schmitt et al.1991; Keller et al. 1993). With these assays, it has beenpossible to demonstrate that development of the hema-topoietic lineages is highly reproducible and efficient.Under appropriate culture conditions, >50% of the cellsin the differentiation cultures will express the hemato-poietic/vascular receptor tyrosine kinase Flk-1 (VEGF re-ceptor 2) (Kabrun et al. 1997) and up to 5% can represent

a clonable hematopoietic progenitor (Keller et al. 1993).Detailed analyses of the early stages of hematopoieticcommitment have shown that both gene expression pat-terns and the kinetics of lineage development within EBsaccurately reflect that found in the yolk sac (Keller et al.1993; Palis et al. 1999; Robertson et al. 2000). Most no-table was the finding that the primitive erythroid lineagedevelops earliest and represents a transient populationthat persists in the EBs for 4 d (Keller et al. 1993). Themacrophage, definitive erythroid, and mast cell lineagesappear following the onset of primitive erythropoiesisand develop in the temporal order found in the yolk sac(Keller et al. 1993). Lymphoid progenitors and HSCs havenot been identified among the progeny of early stage EBs,

suggesting that the initial stages of EB hematopoiesisrepresent the equivalent of yolk sac hematopoiesis.

The faithful recapitulation of this yolk sac develop-mental program provides strong evidence that regulationof hematopoietic commitment in the ES/EB model issimilar, if not identical, to that of the early embryo. Sup-port for this interpretation has been provided by genetargeting studies that helped define the role of specifictranscription factors including Scl/tal-1 (Begley et al.1989), Runx1 (Wang and Speck 1992; Ogawa et al. 1993),and GATA-1 (Orkin 1992) in the establishment of thehematopoietic system. Each of these factors functions atspecific stages of blood cell differentiation as demon-strated by the observations that Scl/tal-1 is required forthe development of all hematopoietic (primitive ery-throid and definitive) lineages (Robb et al. 1995; Shiv-dasani et al. 1995), Runx1 for the definitive lineages butnot primitive erythropoiesis (Okuda et al. 1996; Wang etal. 1996), and GATA-1 for late-stage primitive and de-finitive erythroid maturation (Pevny et al. 1991; Weiss etal. 1994). All of these defects have been accurately rep-licated in the ES cell differentiation model (Weiss et al.1994; Porcher et al. 1996; Wang et al. 1996; Lacaud et al.2002). In addition to further validating the ES cell systemas a model of hematopoietic development, the ability toanalyze mutations in EBs provides a powerful model for

Keller




6/28

structure/function studies as well as for the identifica-tion of molecular targets of the gene of interest.

Lymphoid and HSC development from ES cells

While the early stages of EB differentiation do not give

rise to lymphoid progeny, it has been possible to generatecells of both the T- and B-cell lineages following ex-tended periods of time in culture. B-cell potential wasdemonstrated following the coculture of ES cells withOP9 stromal cells in medium containing lymphoid cy-tokines (Nakano et al. 1994; Cho et al. 1999). More re-cently Schmitt et al. (2004) demonstrated that expres-sion of the Notch receptor ligand, delta-like 1, in theOP9 cells facilitates the differentiation of the developinglymphoid cells to a T-cell fate. These findings are en-couraging as they indicate that populations similar tothat of the P-Sp may be generated from differentiating EScells and that signaling pathways, such as Notch, knownto play a role in B-cell and T-cell fate in vivo may func-

tion in a similar manner in culture.Two recent reports have provided evidence that cells

with HSC properties can be generated from ES cells inculture. In the first, Kyba et al. (2002) transplanted re-cipient animals with ES-cell-derived hematopoietic cellsthat had been induced with forced expression of theHoxB4 gene. Donor cells were clearly evident in thetransplanted animals up to 12 wk following repopulationin primary recipients and as long as 20 wk in secondaryrecipients. The majority of donor cells, however, ap-peared to be myeloid as the levels of lymphoid engraft-ment were very low. These patterns of repopulation dif-fer from those generated by fetal liver or adult bone mar-row HSCs that typically show extensive myeloid and

lymphoid repopulation (Jordan et al. 1990; Kondo et al.2003). Given these patterns of engraftment, it is unclearif the repopulation originates from the equivalent of aP-Sp multipotential HSC that is not sufficiently matureto display multilineage potential or from a yolk sac-likeprogenitor, whose ability to survive in vivo has been pro-longed by the HoxB4 gene.

In the second study, CD45+c-kit+ cells isolated fromEBs cultured for 710 d in the presence of c-kit ligand,IL-3, and IL-6 were transplanted into irradiated recipientanimals (Burt et al. 2004). Even when transplanted intoallogeneic recipients, these cells generated extensive he-matopoietic chimerism and contributed to both the my-eloid and lymphoid lineages. These findings are some-what surprising, given that many different hematopoi-etic populations from ES cell differentiation cultureshave been transplanted into different types of recipientswith little evidence of repopulation. One possible reasonfor the success in this study is that the particular batchof FCS used may have contained factors that promote thedevelopment of HSCs. Establishment of serum-free con-ditions for the generation of these cells would enableother investigators to reproduce these findings. One in-teresting observation in this study was that the level ofES-cell-derived hematopoietic contribution was signifi-cantly higher in recipients in which the cells were trans-

planted directly into the femur rather than intravenouslyinto the circulation. This observation suggests that re-populating cells generated in the ES cell differentiationcultures may not be fully differentiated and lack criticaladhesion molecules that enable them to home to thebone marrow. A lack of homing potential may account

for some of the failures in detecting HSCs in previousstudies.

While these findings indicate that it is possible to gen-erate cells that can persist and function in recipient ani-mals, additional studies will be required to determine ifthese cells are comparable to HSCs found in the fetalliver and adult bone marrow. Given that the P-Sp is con-sidered to be the site of HSC development in the earlyembryo, one important approach will be to establish con-ditions to generate populations comparable to the P-Spin the ES cell differentiation cultures. Access to suchcells should ultimately enable the routine generation ofHSCs from ES cells.

Establishment of the hematopoietic system:identification of the hemangioblast

One of the outstanding strengths of the ES cell differen-tiation model is that it provides access to early develop-mental stages that are difficult to access in the embryo.This unique advantage has been fully exploited to inves-tigate the earliest stages of hematopoietic commitment(Choi et al. 1998; Nishikawa et al. 1998) and to test along-standing hypothesis that the hematopoietic and en-dothelial lineages develop from a common progenitor, acell known as the hemangioblast. The hemangioblasthypothesis was put forward many years ago, based on theobservation that the blood cell and endothelial lineages

develop in close proximity at the same time in the yolksac blood islands (Sabin 1920; Murray 1932; Haar andAckerman 1971). Circumstantial evidence supportingthis hypothesis was provided by studies demonstratingthat immature hematopoietic and vascular cells sharethe expression of a large number of genes (Orkin 1992;Watt et al. 1995; Young et al. 1995; Fong et al. 1996;Takakura et al. 1998) and that specific genes are essentialfor the proper development of both lineages (Dicksonet al. 1995; Robb et al. 1995; Shalaby et al. 1995; Shiv-dasani et al. 1995). Formal proof that a progenitor withproperties of the hemangioblast does exist was providedby studies using the ES differentiation model (Choi et al.1998; Nishikawa et al. 1998). Analysis of early-stage,carefully timed EBs led to the identification of a progeni-tor known as the blast colony-forming cell (BL-CFC) thatgives rise to blast colonies consisting of hematopoieticand vascular progenitors in methylcellulose cultures inthe presence of vascular endothelial growth factor(VEGF) (Kennedy et al. 1997; Choi et al. 1998). The he-matopoietic component of these colonies consisted ofprimitive erythroid progenitors and the subset of defini-tive hematopoietic lineages that is found in the yolk sac,while the vascular potential included both the endothe-lial and vascular smooth muscle lineages (VSM)(Kennedy et al. 1997; Choi et al. 1998; Ema et al. 2003).

Lineage development from ES cells

GENES & DEVELOPMENT 1133



7/28

Initial studies identified the BL-CFC as a transient pro-genitor that develops early in EBs prior to the onset ofprimitive erythropoiesis. Subsequent experiments haveshown that it expresses the receptor tyrosine kinaseFlk-1 (Faloon et al. 2000), the transcription factor Runx1(Lacaud et al. 2002), and the mesoderm gene brachyury

(Fehling et al. 2003), suggesting that this progenitor rep-resents a subpopulation of mesoderm undergoing com-mitment to the hematopoietic and vascular lineages (Fig.3). The BL-CFC does not, however, express markers as-sociated with hematopoietic and vascular development,including CD31, VE-Cadherin (VE-cad), CD34, or c-kit(Fehling et al. 2003; M. Kennedy and G. Keller, unpubl.),a finding consistent with the interpretation that this pro-genitor represents the earliest stage of hematopoieticcommitment.

The BL-CFC is not an artifact of the ES cell model, asa similar progenitor has recently been identified in theearly mouse embryo (Huber et al. 2004). This progenitorarises in the posterior primitive streak of the embryo,

coexpresses Flk-1 and brachyury, and displays the samedevelopmental potential as the EB-derived BL-CFC(Fig. 3). Given these characteristics, this embryo-derivedprogenitor can be considered to be the illusive heman-gioblast. The isolation and characterization of the em-bryo hemangioblast were made possible through the useof strategies developed for the identification of theBL-CFC in the ES/EB system, a clear demonstration thatthis in vitro model can provide important insights intoearly embryonic development.

The ES differentiation system has also been instru-mental in characterizing the earliest stages of hemato-poietic development, immediately following the appear-ance of the hemangioblast. Analyses of these ES-cell-

derived hematopoietic populations at different timepoints have revealed dynamic changes in the expressionof cell surface proteins that likely reflect changes in thelineage composition of the system as well as maturationof the cells within a specific lineage (Kabrun et al. 1997;Nishikawa et al. 1998; Mikkola et al. 2003). Of particular

interest is the observation that these early hemato-poietic cells express markers that are not found on fetalliver and adult hematopoietic populations. Conversely,certain markers associated with fetal and adult hemato-poietic cells are absent from embryonic hematopoieticcells. For instance, the earliest hematopoietic popula-

tions express the endothelial markers Flk-1 (Kabrunet al. 1997; Nishikawa et al. 1998), and VE-cad (Nishi-kawa et al. 1998) and the IIb (CD41) component of theplatelet glycoprotein receptor IIb3 (Mitjavila-Garciaet al. 2002; Ferkowicz et al. 2003; Mikkola et al. 2003).These markers are expressed prior to the onset of CD45,a hematopoietic-specific marker present on most fetalliver and adult bone marrow cells. In the fetal liver andadult, Flk-1 (Millauer et al. 1993; Yamaguchi et al. 1993)and VE-cad (Matsuyoshi et al. 1997) are restricted to theendothelial lineage, while CD41 is expressed exclusivelyin the megakaryocyte lineage (Phillips et al. 1988). Theappearance of endothelial markers prior to CD45 hasbeen interpreted by some as evidence that hematopoietic

cells develop from a specialized population of endothe-lial cells, known as hemogenic endothelium. An equallyplausible explanation is that the earliest embryonic he-matopoietic progenitors that give rise to the later hema-topoietic populations express Flk-1 and VE-cad.

With a more detailed understanding of the earlieststages of hematopoiesis, it has been possible to use theES cell system to begin to investigate the regulation ofhematopoietic commitment in a manner that could notbe done in the embryo. Findings from such studies havedemonstrated that the transcription factors Scl/tal-1 andRunx1 function early in development, specifically at thestage of hematopoietic commitment of the BL-CFC(Faloon et al. 2000; Robertson et al. 2000; Lacaud et al.

2002; DSouza et al. 2005). With respect to induction andgrowth regulation, different groups have shown that thedevelopment of the BL-CFC and hematopoietic re-stricted progenitors is positively regulated, directly orindirectly by bFGF (Faloon et al. 2000), VEGF (Na-kayama et al. 2000; Park et al. 2004), and Ephrin/Eph(Z. Wang et al. 2004) signaling together with serum-de-rived factors. Studies conducted in serum-free conditionsrevealed that BMP4 together with VEGF can support he-matopoietic differentiation of ES cells (Nakayama et al.2000; Park et al. 2004). These factors appear to act atspecific developmental stages, with BMP4 functioning toinduce Flk-1+ cells and VEGF playing a role in the gen-eration of Scl/tal-1-expressing hematopoietic and vascu-lar progenitors within this Flk-1+ population (Park et al.2004). Molecular analysis revealed that the effects ofBMP4 and Flk-1 are mediated through the activation ofthe SMAD1/5 and MAP kinase pathways, respectively(Park et al. 2004). The findings from these studies aresignificant as they demonstrate that hematopoiesis canbe induced from ES cells with defined factors that arethought to function in a similar fashion in the early em-bryo. What is not resolved is the role of BMP4 as it couldbe functioning to induce mesoderm, to specify meso-derm to the hematopoietic program, or both. Furtherstudies using approaches that enable one to monitor the

Figure 3. A comparison of BL-CFC development in EBs to he-mangioblast development in the early mouse embryo. The EBand embryo are derived from an ES cell line in which the GFPcDNA has been targeted to the brachyury locus (Fehling et al.2003; Huber et al. 2004). The presence of GFP in the EB and theprimitive streak of the embryo is indicative of brachyury ex-pression.

Keller




8/28

earliest stages of germ layer induction will be required todefine the precise role of such factors.

In addition to probing early stages of the hematopoi-etic system, the ES differentiation model offers the po-tential to generate large numbers of cells from specifichematopoietic lineages for both molecular and bio-

chemical analyses as well as for transplantation forshort-term lineage replacement therapy. To date, meth-ods have been established for selectively expanding mul-tipotential cell populations (Pinto do et al. 1998), neu-trophils (Lieber et al. 2004), megakaryocytes (Eto et al.2002), mast cells (Tsai et al. 2000), eosinophils (Hamagu-chi-Tsuru et al. 2004), dendritic cells (Fairchild et al.2003), and definitive erythroid cells (Carotta et al. 2004)from ES cells in culture.

Hematopoietic development from hES cells

Several studies have documented hematopoietic devel-

opment in hES cell cultures. Differentiation wasachieved by serum induction of cells either through co-culture with mouse bone marrow stromal cells (Kauf-man et al. 2001) or the generation of EBs (Chadwick et al.2003; Cerdan et al. 2004). While differentiation wasserum-induced, hematopoietic development was aug-mented in the EBs by the addition to the cultures ofBMP4, VEGF, and a mixture of hematopoietic cytokines(Cerdan et al. 2004). Kinetic analysis revealed the devel-opment of definitive erythroid and myeloid progenitorsfollowing 2 wk of differentiation, a pattern that suggeststhat hematopoietic induction, under the conditionsused, is slower than observed in mouse ES cell differen-tiation cultures. Distinct primitive and definitive ery-

throid populations have not yet been identified in thehuman cultures, although changes in patterns of hemo-globin expression within the ES-cell-derived erythroidlineages have been documented (Cerdan et al. 2004).These changes suggest that at least some aspects of glo-bin switching are taking place, reflecting the changesfound during normal fetal development (Stamatoyan-nopoulos and Grosveld 2001). The hematopoietic poten-tial of hES cells has been recently extended to includethe lymphoid lineages, with the observation that cellsexpressing B-cell markers develop from CD34+ cells fol-lowing extensive culture on stromal cells (Vodyanik etal. 2005). Analysis of the earliest stages of hematopoieticdevelopment in the human system identified a CD45

Flk-1+ VE-cad+ CD31+ population at day 10 of differen-tiation that generated CD45+ hematopoietic cells follow-ing further culture (L. Wang et al. 2004). These findingssuggest that human hematopoietic development withinthe EBs parallels that of the mouse in that the earliesthematopoietic progenitors express endothelial markersprior to their maturation to CD45+ cells. Together, thefindings from this limited number of studies indicatethat it is possible to generate hematopoietic cells fromhES cells in culture and that the sequence of develop-mental events may reflect the onset of hematopoiesis inthe early embryo.

Vascular development

The early appearance of the BL-CFC in ES cell differen-tiation cultures not only defines the onset of hemato-poietic commitment, but also represents the earlieststages of vascular development. This pattern of vascular

commitment was anticipated, as endothelial cells can bedetected early in the yolk sac blood islands of the embryo(Haar and Ackerman 1971). As with the hematopoieticlineages, vascular development appears to be quite effi-cient in serum-stimulated differentiation cultures (Wanget al. 1992; Bautch et al. 1996; Vittet et al. 1996; Kabrunet al. 1997; Hirashima et al. 1999). Kinetic analysis ofendothelial development in intact EBs revealed a sequen-tial up-regulation of the following markers associatedwith the development and maturation of the lineage;flk-1, CD31, tie2, tie1, and VE-cad (Vittet et al. 1996).This pattern is similar to that observed in the early em-bryo, suggesting that development of the endotheliallineage in vitro recapitulates its development in vivo.

A similar pattern of differentiation was observed whenFlk-1+ progenitors were isolated from the differentiationcultures and recultured on type IV collagen-coateddishes or on OP9 stromal cells (Hirashima et al. 1999).Further analysis of this ES-cell-derived Flk-1 populationrevealed that it also displayed the capacity to generatecells of the VSM lineage. Clonal analysis demonstratedthat both the endothelial and VSM lineages develop froma common progenitor, a cell that can be considered to bea vascular progenitor (Yamashita et al. 2000). A compa-rable progenitor was also identified in the yolk sac of theembryo. These findings are important as they define anew branch point within the vascular system and, indoing so, further highlight the power of the ES cell sys-

tem in studying early lineage commitment.Studies on the regulation of vascular commitment

have revealed that factors essential for the developmentand maturation of the endothelial lineage in the earlyembryo, including VEGF (Carmeliet et al. 1996; Ferraraet al. 1996) and the receptors Flt1 (Fong et al. 1995) andFlk-1 (Shalaby et al. 1995), are also required for the es-tablishment of the lineage in ES cell differentiation cul-tures (Bautch et al. 2000; Kearney et al. 2002; Zippo et al.2004). The regulation of endothelial and VSM develop-ment in ES cell cultures is also controlled, in part, by thetranscription factor Scl/tal-1. In the absence of Scl/tal-1,ES cells favor the VSM pathway, while when expressed,they differentiate to both the endothelial and VSM lin-eages (Ema et al. 2003; DSouza et al. 2005).

The functional capacity of the ES-cell-derived vascu-lar progenitors has been evaluated both in culture andin animal models following transplantation. ES-cell-derived vascular cells are able to organize into vessel-likestructures in EBs (Doetschman et al. 1985), in explantcultures (Bautch et al. 1996), or when cultured on colla-gen I (Yamashita et al. 2000). When transplanted intotumor-bearing mice, ES-cell-derived progenitors incorpo-rated into the newly formed vessel in the tumors, indi-cating that they can function as vascular cells and par-ticipate in neovascularization in vivo (Marchetti et al.





9/28

2002; Yurugi-Kobayashi et al. 2003). These observationsare significant as they demonstrate that the ES cell dif-ferentiation system can be used as a model for investi-gating the mechanisms regulating angiogenesis in tumorformation.

Endothelial differentiation has also been demonstrated

in hES cell differentiation cultures (Levenberg et al.2002). The cells that develop in these cultures expressmarkers associated with the endothelial lineage, formtube-like structures in matrigel in vitro, and generatecapillary structures when embedded in sponges andtransplanted into SCID mice. Taken together, the find-ings from studies on vascular development of ES cellsdemonstrate that the lineages develop efficiently andthat the cells that are generated display properties of nor-mal endothelial and VSM cells.

Cardiac development

The development of the cardiac lineage in ES cell differ-

entiation cultures is easily detected by the appearance ofareas of contracting cells that display characteristics ofcardiomyocytes. Development of the cardiomyocyte lin-eage has been analyzed in most detail in cultures in-duced with serum (Hescheler et al. 1997; Boheler et al.2002). As observed with the hematopoietic and vascularsystems, development of the cardiomyocyte lineageprogresses through distinct stages that are similar to de-velopment of the lineage in vivo. An ordered pattern ofexpression of cardiac genes is observed in the differen-tiation cultures, with expression of the transcription fac-tors gata-4 and nkx2.5 that are required for lineage de-velopment preceding the expression of genes such asatrial natriuretic protein (ANP), myosin light chain

(MLC)-2v, -myosin heavy chain (-MHC), -myosinheavy chain (-MHC), and connexin 43 that are indica-tive of distinct maturation stages within the developingorgan in vivo (Hescheler et al. 1997; Boheler et al. 2002).Maturation of the lineage in the cultures is associatedwith changes in cell size and shape, progressing fromsmall, round cells to elongated cells with well-developedmyofibrils and sarcomeres (Boheler et al. 2002). Electro-physiologicial measurements of cells from differenttimes in culture suggest that the cardiomyocyte popula-tion undergoes a change from early-stage cells with pace-maker-like activity to more terminally differentiatedatrial- and ventricular-like cells (Maltsev et al. 1993;Hescheler et al. 1997; Banach et al. 2003). These changescorrelate with the observed changes in cellular morphol-ogy, characteristic of each cell type.

While these studies clearly demonstrate the develop-ment of the cardiomyocyte lineage from differentiatingES cells, they are carried out in heterogeneous culturesin which these cells represent a minority of the entirepopulation (Klug et al. 1996). As there are relatively fewantibodies available for the isolation of cardiac progeni-tors, investigators have genetically engineered ES cells toenable specific selection of cells representing differentstages of development within the lineage. ES cells havebeen generated to express either drug-resistance or fluo-

rescent genes under the control of promoters that driveexpression at specific stages of cardiac development. Inthe first of these approaches, Klug et al. (1996) expressedthe neomycin-resistance gene under the control of the-cardiac MHC promoter. With G418 selection at appro-priate stages of development, populations highly en-

riched (>99%) for cardiomyocytes were isolated. Whenapplied to large-scale cultures, this strategy enabled thegeneration of large numbers of cardiomyocytes (Zandstraet al. 2003). Other strategies involve expressing the greenfluorescent protein (GFP) from cardiac specific promot-ers including Nkx2.5 (Hidaka et al. 2003), cardiac -actin(Kolossov et al. 1998), and myosin light chain-2v (Mulleret al. 2000). Expression from myosin light chain-2v wasdesigned to specifically select for ventricular cells fromthe ES cell differentiation cultures (Muller et al. 2000).Cells selected on this basis displayed electrophysiologi-cal properties of ventricular cardiomyocytes, indicatingthat the strategy was successful.

Several different studies have begun to investigate the

mechanisms regulating the development of the cardiaclineage in ES cell differentiation cultures. Parisi et al.(2003) demonstrated that the EGF-CFC factor Cripto,known to be essential for cardiomyocyte development invivo (Ding et al. 1998), plays a pivotal role in differentia-tion of ES cells to the cardiac lineage. Cripto/ ES cellsdisplay a deficiency in generating cardiomyocytes in cul-ture that could be restored by the addition of solubleCripto to the differentiation cultures. Of interest was theobservation that the factor had to be supplied within thefirst 48 h of differentiation, suggesting that its primaryfunction may be on the induction of mesoderm, which,in turn, differentiates to the cardiac lineage. Notch sig-naling also plays a role in cardiac development from ES

cells (Schroeder et al. 2003). However, in this case, inhi-bition of the pathway appears to be important for cardiacdifferentiation, as ES cells lacking the recombination sig-nal sequence-binding protein Jk, a downstream signalingmolecule of all Notch receptors, generate more cardiaccells than wild-type ES cells. Other factors, includingBMP2 and FGF2 (Kawai et al. 2004) as well as nitric oxide(Kanno et al. 2004) and ascorbic acid (Takahashi et al.2003), have been shown to promote or improve cardio-myocyte differentiation in ES cell cultures. As observedwith Cripto, the effects of BMP-2 and FGF were mostdramatic when the factors were added early in the dif-ferentiation cultures, again suggesting that some of theeffects may be mediated at the level of mesoderm induc-tion. A role for FGF signaling in ES-cell-derived cardiacdevelopment is further supported by the observation thatgfr1/ ES cells show a marked defect in their ability todifferentiate to cardiomyocytes (DellEra et al. 2003).Factors produced by visceral endoderm also appear toplay a role in cardiomyocyte differentiation as cocultureof ES cells with a visceral endoderm-like cell line, END-2, significantly enhanced cardiac development (Mum-mery et al. 2002). The nature of the END-2-derived mol-ecules responsible for cardiac development remains to bedetermined.

While these studies have identified several factors that

Keller




10/28

regulate cardiac development from ES cells, the precisestage at which they act and their relationship to eachother remain to be elucidated. To characterize themechanisms regulating cardiac development in the EScell differentiation cultures in more detail, it will be nec-essary to use approaches that enable one to monitor and

isolate the earliest stages of development within this lin-eage.

Transplantation of ES-cell-derived cardiomyocytes

One of the most promising applications of cardiomyo-cyte differentiation of ES cells is to provide a source oftransplantable cells for the treatment of cardiovasculardisease (Kehat and Gepstein 2003; Nir et al. 2003). Kluget al. (1996) first demonstrated that ES-cell-derived car-diomyocytes selected for -cardiac MHC expressioncould incorporate and survive in the hearts of dystrophicmice, following direct transplantation into the organ.More recent studies have demonstrated that mouse ES-

cell-derived cells can survive for up to 32 wk followingtransplantation into the hearts of rats with myocardialinfarction (Min et al. 2002, 2003). For these studies, areasof contracting cardiac cells were dissected from hetero-geneous differentiation cultures and used for transplan-tation. Analysis of the animals indicated improved car-diac function in those transplanted with the ES-cell-de-rived cells compared to controls that received cell-freemedia. Histological analysis demonstrated the presenceof differentiated donor-derived cardiac cells. To improvethe survival of the ES-cell-derived cardiomyocytes, Yanget al. (2002), engineered the expression of VEGF in thecardiomyocytes prior to transplantation into the heartsof mice that had suffered myocardial infarction. Mice

that received the VEGF-expressing cells displayed en-hanced neovascularization and improved cardiac func-tion compared with animals transplanted with wild-typecells.

Although the findings from these studies are encour-aging, it is unclear to what extent the improvement isdue to the myocyte function of the cells rather than toindirect effects such as induced vascular development atthe site of injection. Future studies need to include theinjection of noncardiac cells to control for the cardiaclineage specificity of the observed improvement. An ad-ditional issue relates to the differentiation status oftransplanted cells. In the studies described above, rela-tively mature contracting cardiomyocytes were trans-

planted. It is conceivable that populations consisting ofimmature progenitor cells may be more effective in re-pair of the damaged heart, as such cells should displaygreater proliferative potential than the contracting cellsand thus generate a larger graft. Identification and char-acterization of early-stage cardiac progenitors in ES celldifferentiation cultures are important goals for the fu-ture.

Cardiac development from hES cells

Cardiomyocyte differentiation has also been demon-strated in hES cell differentiation cultures (Nir et al.

2003). As observed with the mouse system, the cells dif-ferentiate over time in culture, following a maturationprocess similar to that reported in vivo (Kehat et al. 2001;Nir et al. 2003; Snir et al. 2003). Although the hES-cell-derived cardiomyocytes do not undergo maturation tothe stage of adult cardiomyocytes, cells with electrical

properties of nodal, atrial, and ventricular cells havebeen identified (He et al. 2003). The induction events forcardiac development in the hES cell cultures have notbeen defined in any detail, and in most studies the popu-lations are generated following serum stimulation. Asobserved with mouse ES cells, coculture of hES cellswith END-2 endoderm cells enhanced the developmentof cardiac cells in these differentiation cultures (Mum-mery et al. 2003).

Recently, hES-cell-derived cardiomyocytes have beenused in xenogeneic transplantation as biologic pace-makers for the treatment of bradycardia (Kehat et al.2004). To demonstrate this potential, clusters of con-tracting cardiomyocytes isolated from the differentiation

cultures were transplanted into the left ventricle of pigsthat had their atrioventricular node ablated. The recipi-ent hearts had spontaneous rhythms that appeared tooriginate from the transplanted cells as assessed by high-resolution electroanatomical mapping. Though promis-ing, these results also raise the concern that transplantedcells could serve as a nidus for arrhythmia.

Other mesoderm-derived lineages

While the hematopoietic, vascular, and cardiac lineageshave been studied in most detail, the ES cell system doesoffer the potential to develop many other mesoderm lin-

eages. To date, cell populations representing the skeletalmuscle (Rohwedel et al. 1994), the osteogenic (Butteryet al. 2001; zur Nieden et al. 2003), the chrondrogenic(Kramer et al. 2000), and adipogenic (Dani et al. 1997)lineages have been generated from ES cells differen-tiated in culture. Where analyzed in detail, the develop-ment of the lineage in vitro recapitulated its develop-ment in vivo, indicating that these populations pro-gressed through normal differentiation programs. Aswith the other lineages discussed in this review, accessto these cell types from ES cells offer unprecedented op-portunities to establish model systems to study lineagecommitment and to provide a source of cells for trans-plantation.

Endoderm derivatives

The generation of endoderm derivatives, in particularpancreatic -cells and hepatocytes, has become the focusof many investigators in the field of ES cell biology. Theinterest in the efficient and reproducible development ofthese cell types derives from their clinical potential forthe treatment of Type I diabetes and liver disease, respec-tively. Despite the interest in these lineages, progress ingenerating endoderm-derived cell types has been slow.The lack of progress in this area can be attributed to





11/28

several different factors. First, several genes used asmarkers of definitive endoderm (Foxa2, Gata4, andSox17) (Arceci et al. 1993; Monaghan et al. 1993; Sasakiand Hogan 1993; Laverriere et al. 1994; Kanai-Azumaet al. 2002), early liver (-fetoprotein and albumin)(Dziadek and Adamson 1978; Meehan et al. 1984; Sellem

et al. 1984), and early pancreas (Pdx1 and insulin)(McGrath and Palis 1997) development are also ex-pressed by visceral endoderm. Visceral endoderm, apopulation of extraembyonic endoderm, is an extra-embryonic tissue that functions in a regulatory capacitybut does not contribute directly to the formation of anyadult organs (Fig. 1; Gardner and Rossant 1979; Thomasand Beddington 1996). Given the overlapping expressionpatterns, it can be difficult to distinguish definitive andextraembryonic endoderm in the ES cell differentiationcultures.

While ES cells do not contribute extensively to vis-ceral endoderm in vivo following injection into blasto-cysts (Beddington and Robertson 1989), they do display

some capacity to generate this population in culture(Doetschman et al. 1985; Soudais et al. 1995). The tran-scription factors nanog, GATA-4, and GATA-6 all appearto play some role in the regulation of visceral endodermdevelopment from ES cells. ES cells lacking nanog (Mit-sui et al. 2003) or those engineered to overexpressGATA-4 or GATA-6 (Fujikura et al. 2002) differentiate tovisceral endoderm in the presence of LIF. Recently,Hamazaki et al. (2004) demonstrated that aggregation ofES cells in the presence of LIF is sufficient to down-regulate nanog expression and induce visceral endodermdevelopment. With hES cell cultures, BMP2 has beenshown to induce cells with visceral endoderm character-istics (Pera et al. 2004). The fact that ES cells can gener-

ate visceral endoderm in culture makes it imperativethat one establish appropriate strategies for identifyingdefinitive endoderm and its derivatives. Approaches toaddress this issue could include the identification ofgenes that are differentially expressed between the endo-derm populations and the establishment of conditionsthat specifically support the development of definitiveendoderm.

A second problem encountered in endoderm differen-tiation from ES cells is the lack of specific inducers ofthis lineage. In most studies to date, endoderm develop-ment has been analyzed in ES cell cultures differentiatedin the presence of FCS. While FCS appears to efficientlyinduce mesoderm populations, particularly the hemato-poietic, vascular, and cardiac lineages, findings from arecent study demonstrate that it is not the optimal in-ducer of endoderm (Kubo et al. 2004).

In spite of these obstacles, several studies have pro-vided evidence for the generation of endoderm-derivedcell types including populations that display character-istics of pancreatic islets (Colman 2004; Stoffel et al.2004), hepatocytes (Hamazaki et al. 2001; Jones et al.2002; Yamada et al. 2002a), thyrocytes (Lin et al. 2003),lung (Ali et al. 2002), and intestinal cells (Yamada et al.2002b). This review focuses on pancreatic and hepato-cyte development.

Pancreatic development from ES cells

Two general protocols have been used for the generationof pancreatic islet-like cells from ES cells in culture. Thefirst, a five-step protocol, described by Lumelsky et al.(2001), is an adaptation of a protocol for neural cell dif-

ferentiation and involves the transfer of serum-inducedEBs to serum-free medium followed by treatment withFGF and factors that promote maturation of endocrinecells. Clusters resembling pancreatic islets developed inthese cultures and cells within the clusters were shownto contain insulin, glucagon, or somatostatin by immu-nohistochemistry. While the insulin content of the cellswas low, they were able to secrete insulin in response toglucose. However, when tested for their ability to func-tion in vivo following transplantation into strepto-zotocin (STZ)-induced diabetic mice, these cells failed tocorrect the hyperglycemia of these animals. The inabil-ity to cure these animals could be due to the fact thatthe cells were too immature or that they were not islet

cells.Several other groups have modified this approach and

reported improved development of -like cells in the cul-tures. The modifications included constitutive expres-sion of the transcription factor Pax4 (Blyszczuk et al.2003) during differentiation, inducible expression ofPdx1 (Miyazaki et al. 2004) during differentiation, or ex-posure of the cells to the inhibitor of phosphoinositide3-kinase (PI3K), LY294002 (Hori et al. 2002), at the finalmaturation step. Pdx1 and Pax4 are essential for -celldevelopment in vivo (Murtaugh and Melton 2003).Transplantation of the Pax4-induced and LY294002-treated cells into STZ-treated recipients did improvethe hyperglycemia in these animals, suggesting that ES-

derived cells were producing insulin. Of significance wasthe observation that removal of the LY294002-treatedgraft resulted in reversion to a diabetic state, indicatingthat the observed improvement was due to the im-planted cells. While these findings are encouraging, sev-eral issues remain to be resolved regarding the types ofcells generated and the amount and type of insulin pro-duced in these cultures.

Rodents contain two nonallelic insulin genes, insulinI and insulin II (Melloul et al. 2002). While insulin I isexpressed predominantly in -cells, insulin II is found inthe yolk sac and developing brain in addition to the pan-creas (Deltour et al. 1993; Devaskar et al. 1993; Giddingset al. 1994). Given that the ES cell cultures describedabove do contain significant numbers of neuronal cells,it is possible that some of the insulin detected is derivedfrom these cells, as demonstrated by the recent study ofSipione et al. (2004). A second issue relates to the obser-vation that ES-cell-derived cells can absorb insulin fromthe tissue culture medium and then release it whenstressed or undergoing apoptosis (Rajagopal et al. 2003).It is unclear to what extent neuronal-derived insulin orthe release of absorbed insulin is involved in the abovestudies as the cultures contain mixed populations ofcells. Resolution of these issues and the development oflarge numbers of functional insulin-producing cells will

Keller




12/28

require improved protocols for endoderm induction and-cell specification and maturation.

Some of the concerns relating to -cell develop-ment in ES cell cultures were addressed in a recent studyof Ku et al. (2004). In this work, modification of thecomponents of the serum-free culture and maturation of

the differentiating cells in the presence of activin B,exendin-4, and nicotinamide, resulted in the develop-ment of cell populations that expressed insulin I, fol-lowing the appearance of cells that expressed variousgenes indicative of endoderm differentiation. Immuno-staining revealed that cells within the cultures alsocontained c-peptide, the cleavage by-product of pro-insulin, indicating that at least some of the insulin de-tected is produced by the cells. The modification tothe maturation cultures significantly increased the lev-els of insulin I mRNA as well as the frequency of cellsthat express insulin (to 2%3% of the total) in the popu-lation. These findings are encouraging and offer hopethat further modifications together with selection ap-

proaches will yield populations highly enriched for-cells.

A second approach for the generation of islet-likestructures relies on their development in heterogeneouspopulations derived from ES cells following serum in-duction. Such approaches have yielded, at a low fre-quency, cells that display many characteristics of -cells,including the presence of c-peptide and insulin in dis-crete granules within the cells (Kahan et al. 2003). Toenrich for -cells from heterogeneous serum-inducedcultures, Soria et al. (2000) developed a selection strategybased on the insulin promoter driving expression of theselectable gene -geo. Addition of G418 to the culturesresulted in the isolation of an insulin-secreting clone

that had insulin contents similar to that of normal islets.When injected into STZ diabetic mice, these cells cor-rected the hyperglycemia of these animals for up to 12wk. Beyond this time, however, a significant number ofanimals reverted back to a hyperglycemic stage, indicat-ing that the cells were unable to provide long-term -cellfunction.

As a strategy to select for early pancreatic cells as theydevelop in ES cell differentiation cultures, Micallef et al.(2005) targeted GFP to the pdx1 locus. Pdx1 is expressedin the earliest stages of pancreatic development as theorgan rudiments are specified from the gut endoderm(Murtaugh and Melton 2003). Treatment of the develop-ing EBs with retinoic acid in this study led to the devel-opment of a small GFP-pdx1+ population by day 8 ofdifferentiation. Analysis of the isolated GFP-pdx1+ cellsrevealed that they expressed endoderm-specific genesbut not genes associated with pancreatic maturation,suggesting that they represent an early stage of endo-derm development. Alternatively, some of the cellswithin the population could be visceral endoderm, a tis-sue that also expresses pdx1 (McGrath and Palis 1997).Further studies will be required to demonstrate that thispopulation does represent the earliest stages of pancre-atic differentiation. The generation of the pdx1 select-able marker is a good strategy that will enable the quan-

titation and isolation of pancreatic lineage progenitorsfrom cultures induced under optimal conditions for de-finitive endoderm development.

Insulin-expressing cells have also been detected in hEScell differentiation cultures (Assady et al. 2001). How-ever, as with many of the studies with mouse cells, the

frequency of these cells in the hES cell cultures is cur-rently too low to allow detailed characterization andfunctional analysis.

Hepatocyte development from ES cells

Several reports have documented the generation of cellswith hepatocyte characteristics in ES cell differentiationcultures. Hamazaki et al. (2001) developed a multistepprotocol that included the addition of specific growthfactors at various stages of differentiation to promote thegrowth and differentiation of hepatocyte cells within thecultures. Genes indicative of hepatocyte development

and maturation were expressed in these cultures with akinetic pattern similar to that found in vivo. Hepatocyte-like cells generated with this protocol were subsequentlyshown to contain albumin protein and to produce urea(Chinzei et al. 2002). Cells from these cultures were alsotransplanted into recipient mice pretreated with 2-ace-tylaminofluorene to prevent proliferation of host hepa-tocytes. Four weeks following transplantation, low num-bers of albumin-producing donor cells were detected inthe livers of the recipient animals, suggesting that thesecells might be able to function in vivo (Chinzei et al.2002).

To be able to specifically monitor hepatocyte develop-ment and isolate these cells from the differentiation cul-

tures, several studies have incorporated approaches todistinguish definitive endoderm and early liver popula-tions from visceral endoderm. Jones et al. (2002) used anES cell line carrying a lacZ gene trap vector inserted intoa gene known as Gtar. Gtar is expressed in definitiveendoderm committed to a hepatic fate and the develop-ing liver but is not found in visceral endoderm. Follow-ing serum induction, LacZ-positive cells that co-expressed -fetoprotein (afp) and albumin (alb) were de-tected, indicating development of the hepatocyte lineagein the culture. In a recent study, Asahina et al. (2004)identified cytochrome P450 7A1 (Cyp7a1) as a liver-spe-cific gene that is not expressed in visceral endoderm.Cyp7a1 expression was detected at low levels in serum-stimulated EBs at 2 wk of differentiation. As the Cyp7a1-positive cells were not isolated or characterized in thisstudy, their developmental potential remains to be de-termined. Yamada et al. (2002a) used specific uptake ofthe organic anion indocyanine green (ICG) as a marker ofhepatocyte development in ES differentiation cultures.ICG-positive cells isolated from the cultures expressedhepatocyte markers, displayed ultrastructural character-istics similar to those of hepatocytes and gave rise to lownumbers of albumin-positive cells following transplan-tation into recipient animals. Taken together, the find-ings from these various studies strongly suggest that





13/28

cells with hepatocyte characteristics are generated inmouse ES cell differentiation cultures.

Cells with hepatocyte characteristics have also beenidentified in hES cell cultures (Rambhatla et al. 2003). Togenerate these cells, the cultures were treated with so-dium butyrate, a procedure that killed significant num-

bers of the differentiating population. The cells that sur-vived this treatment gave rise to a population that dis-played many features of hepatocytes. Although theapproach is promising and results in the development ofhepatocyte-like cells, the physiological relevance of thesodium butyrate treatment is unclear.

The findings from these different studies indicate thatcells with characteristics of pancreatic -cells and hepa-tocytes can be generated in ES cell differentiation cul-tures. A problem with many of the current approaches isthe very low frequency of differentiated cells identifiedand the cellular heterogeneity within the cultures. Im-provements in generating endoderm derivatives will re-quire the development of more efficient protocols for

endoderm induction and for pancreatic and hepato-cyte specification combined with technologies for select-ing the appropriate cell types from the differentiationcultures. Only when large numbers of highly enrichedprogenitors are accessible can methods for their matura-tion be defined and their functional capacity be rigor-ously tested in animal models of diabetes and liver fail-ure.

We have recently investigated the potential of ES cellsto differentiate to endoderm derivatives and developedtwo different protocols that promote the generation ofthese cell types (Kubo et al. 2004). The first is a restrictedexposure of the EBs to serum followed by a period ofserum-free culture, and the second is induction with ac-

tivin A (activin) in the absence of serum. Endoderm de-velopment was quantified based on the proportion ofcells that expressed Foxa2, a transcription factor foundin the earliest stages of definitive endoderm develop-ment (Monaghan et al. 1993; Sasaki and Hogan 1993).All of the Foxa2+ cells that developed in these cul-tures also expressed the primitive streak marker brachy-ury, a gene that is not expressed in visceral endo-derm. This observation strongly suggests that theFoxa2+ cells represented definitive endoderm. Based onthe number of Foxa2+ cells, the activin protocol wasfound to be the most efficient as >50% of the total popu-lation in these cultures expressed this protein. Tissuespecification was detected in the activin-induced cul-tures as demonstrated by the presence of cells that ex-pressed Afp and Alb, indicative of hepatocyte differen-tiation, Pdx1 for early pancreas specification, and surfac-tant protein C (Sp-c) for early lung development (Wert etal. 1993). When transplanted under the kidney capsule ofrecipient mice, the activin-induced cells generated lung-like structures that expressed Sp-c and gut structuresthat expressed intestinal fatty acid-binding protein(IFABP). These findings demonstrate that it is possible toefficiently generate endoderm in ES differentiation cul-tures with the use of specific inducers in the absence ofserum.

Ectoderm derivatives

Ectoderm differentiation of mouse ES cells is well estab-lished, as numerous studies have documented and char-acterized neuroectoderm commitment and neural differ-entiation. Given extensive efforts in this field over the

past decade, several different protocols have evolved topromote neuroectoderm differentiation. The various ap-proaches include (1) treatment of serum-stimulated EBswith retinoic acid (Bain et al. 1995), (2) sequential cultureof EBs in serum followed by serum-free medium (Okabeet al. 1996), (3) differentiation of ES cells as a monolayerin serum-free medium (Tropepe et al. 2001; Ying et al.2003b), and (4) differentiation of ES cells directly on stro-mal cells in the absence of serum (Kawasaki et al. 2000;Barberi et al. 2003). As with the mesoderm and endo-derm lineages, development of the ectoderm lineages inthe ES differentiation cultures appears to recapitulatetheir development in the early embryo (Barberi et al.2003). Commitment to neuroectoderm appears to be

rapid and efficient as the majority of the cells that de-velop in these cultures display characteristics of neuralcells (Okabe et al. 1996; Kawasaki et al. 2000; Barberi etal. 2003). By targeting the -geo selection marker gene toSox2, a gene expressed in neuroepithelium, Li et al.(1998) were able to generate highly enriched neural popu-lations (>90% of the population) using the retinoic acid-induction protocol followed by selection with G418.

Each of the three major neural cell types of the centralnervous systemneurons, astrocytes, and oligodendro-cytescan be generated, and relatively pure populationsof each can be isolated when cultured under appropriateconditions (Okabe et al. 1996; Barberi et al. 2003). Inaddition to the generation of these different neural popu-

lations, conditions have been established for the devel-opment of different subtypes of neurons. The protocolsfor differentiation to specific types of neurons have in-cluded the sequential combination of regulators that areknown to play a role in the establishment of these lin-eages in the early embryo. For instance, midbrain dopa-minergic neurons have been generated in the EB systemby overexpression in the cells of the transcription factornuclear-receptor-related factor1 (Nurr1), and the addi-tion to the cultures of SHH and FGF8 (Kim et al. 2002).Nurr1, SHH, and FGF8 are required for the developmentof this class of neurons in the early embryo (Ye et al.1998; Simon et al. 2003). More recent studies have dem-onstrated the development of cholinergic, serotonergic,and GABAergic neurons in addition to dopaminergicneurons, when differentiated on MS5 stromal cells in thepresence of different combinations of cytokines (Barberiet al. 2003). Using the coculture approach together withthe appropriate signaling molecules and selection steps,Wichterle et al. (2002) successfully generated cells thatdisplay many of the characteristics of motor neurons. Aswith many other populations discussed in this review,the generation of these cells from ES cells recapitulatedthe pathway of motor neuron development in vivo.

The ES cell model is being used to investigate the ear-liest stages of neural development. When cultured at low

Keller




14/28

density in serum-free medium in the presence of LIF, EScells generate a population that has been called primitiveneural stem cells (Tropepe et al. 2001). These cells havebeen characterized by their ability to generate neuro-sphere-like colonies composed of cells that express theneural precursor cell marker, nestin (Lendahl et al. 1990).

When cultured on a matrigel substrate in the presence oflow amounts of serum, cells within these colonies gen-erated neurons, astrocytes, and oligodendrocytes. Meso-derm and definitive endoderm derivatives were not de-tected in these cultures. In contrast to the restricteddevelopmental pattern observed in culture, the cellsfrom these colonies were able to contribute to most tis-sues of the embryo following morula reaggregation andtransplantation in vivo. These observations would posi-tion the primitive neural stem cell at a developmentalstage between the ES cell and a neural-restricted progeni-tor.

The ability to generate different types of neurons fromES cells has dramatically raised the interest in repair of

nervous system disorders by cell replacement therapy.Early transplantation studies demonstrated the feasibil-ity of engrafting ES-cell-derived neural cells into recipi-ent animals. Brustle et al. (1997) transplanted such cellsinto ventricles of fetal rats and demonstrated the incor-poration of donor-derived neurons, astrocytes, and oligo-dendrocytes into the brains of the recipient animals, sev-eral weeks later. As a demonstration of cell-based repair,this group showed that ES-cell-derived oligodendrocytescould form myelin sheaths around host neurons whentransplanted into a myelin-deficient rat model of mul-tiple sclerosis (Brustle et al. 1999). Other studies demon-strated that ES-cell-derived neural cells could lead to par-tial functional recovery of spinal cord injury in rats,

when injected directly into the spinal cord near the in-jury site (McDonald et al. 1999). Although the basis forrecovery in this study was not determined, remyelin-ation of the damaged nerves may have played some role,as a later study demonstrated that ES-cell-derived neuralcells transplanted into spinal cords of rats with chemi-cally induced demyelination preferentially differentiateinto oligoendrocytes and myelinated host axons (Liuet al. 2000).

One of the greatest expectations of clinical utility forES-cell-based therapy for neurodegenerative disease isthe replacement of dopamine neurons in Parkinsonspatients. Using a rat model for Parkinsons disease,Kim et al. (2002) demonstrated that ES-cell-derived do-pamine neurons survived, developed functional syn-apses, and displayed electrophysiolgic properties charac-teristic of midbrain neurons following transplantationinto these animals. In addition, the animals showedsome recovery, suggesting that the transplanted cellswere functional. The animals that received cells over-expressing Nurr1 showed the greatest improvement inthis study. Populations not overexpressing exogenousNurr1 generated with the stromal cell coculture protocolhave also been transplanted into Parkinsonian mice (Bar-beri et al. 2003). Donor dopamine neurons were detected2 mo following transplantation, and grafted animals

showed alleviation of the behavioral deficits displayedby the control animals. Taken together, these studiesestablish that it is possible to generate specific subsets ofneural cells from mouse ES cells and that these cells canbe transplanted into models of human diseases. Addi-tional studies will be required to determine the extent to

which these cells can function over extended periods oftime.

In addition to the neural lineages, ES cells can also giverise to epithelial cells that will undergo differentiation topopulations that express markers of keratinocytes (Ba-gutti et al. 1996). The temporal sequence of expression ofdifferent keratins associated with development of thelineage in culture is very similar to patterns found in themouse embryo. When maintained for 3 wk in organ cul-tures, the ES-cell-derived keratinocytes were able toform structures that resemble embryonic mouse skin,indicating that they possess some capacity to generate afunctional tissue (Coraux et al. 2003). Differentiation ofthe epidermal cells appears to be controlled, in part, by

BMP4 (Kawasaki et al. 2000). When added to culturesundergoing neuroectoderm differentiation, BMP4 pro-moted keratinocyte development and inhibited neuraldifferentiation.

Neural development from hES cells

Ectoderm derivatives have also been generated from hEScells and as with the mouse, most studies have focusedon neuroectoderm and neural cells (Hornstein and Ben-venisty 2004). The three CNS cell types can be derivedfrom hES cells using several different protocols (Reu-binoff et al. 2001; Zhang et al. 2001). When transplanted

into newborn mice, hES-cell-derived neural cells gener-ated neurons and glia that could be detected in differentregions of the recipient brain 4 wk later (Zhang et al.2001). Morphological analysis indicated that the donorcells had matured and were indistinguishable from thesurrounding host cells. As observed with the mousecells, coculture of hES cells with stromal cells led to thedevelopment of neural populations with midbrain dopa-mine characteristics (Perrier et al. 2004; Zeng et al.2004). Cell populations displaying varying degrees of do-paminergic neuron differentiation have been trans-planted into Parkinsonian rats (Ben-Hur et al. 2004; Zenget al. 2004). Low numbers of human cells could be de-tected in the animals between 5 and 12 wk post-trans-plantation, and improvements in the behavior of some ofthe transplanted animals were noted in one of the stud-ies (Ben-Hur et al. 2004). However, given that the graftthat these animals received consisted of a mixture ofneural and nonneural cells and that very few dopamin-ergic cells were detected in the recipients, the signifi-cance of these improvements is unclear. Future studiesin which highly enriched populations of appropriatestaged dopaminergic neurons are transplanted into a va-riety of different models will be required to determinethe potential of this approach as a treatment of Parkin-sons disease.





15/28

Germ cells

In addition to somatic tissues, several studies have docu-mented the development of germ cells from differenti-ated ES cells. In the first study, ES cells were induced todifferentiate in two-dimensional cultures in the presence

of serum without additional factors (Hubner et al. 2003).To identify the developing germ cells in the differentia-tion cultures, the ES cells carried a modified Oct4 pro-moter that drove GFP expression specifically in thegermline. GFP expression was detected as early as day 4of differentiation, and by day 8 up to 40% of the popu-lation was positive. By day 26 of culture, oocyte-likecells enclosed in a structure resembling a zona pellucidawere released into the culture from the differentiatingcell mass. Following several additional weeks of culture,structures similar to blastocysts were observed, likelyoriginating from the parthenogenic activation of the oo-cytes.

Male germ cell development has also been demon-

strated in ES cell differentiation cultures. With one ap-proach, developing germ cells were identified using anES cell line in which GFP was targeted to the mouse vasahomolog (Mvh), a gene specifically expressed ingermline cells (Toyooka et al. 2003). While GFP+ cellswere detected in EBs stimulated with serum alone, thenumber was significantly enhanced when the EBs werecoaggregated with a BMP4-expressing cell line. This ob-servation is encouraging as BMP4 is a known regulator ofgerm cell development in the mouse embryo (Lawsonet al. 1999). To determine the potential of these putativegerm cells, GFP+ cells were isolated from the EBs, co-cultured with gonadal cells, and then transplanted undera host testis capsule, where they were shown to partici-

pate in spermatogenesis. In the second study, germ cellsgenerated in serum-stimulated EBs were isolated on thebasis of SSEA1 expression and cultured in the presence ofretinoic acid, LIF, bFGF, and c-kit ligand (Geijsen et al.2004). Cells differentiated under these conditions dis-played imprint erasure, a distinguishing characteristic ofgerm cells. If maintained in culture for 2 wk, these se-rum-stimulated EBs generated a small population ofspermatid-like cells that could be isolated using an an-tibody recognizing the sperm acrosome. This populationwas enriched for haploid cells that were able to completefertilization, following injection into oocytes.

Germ cell development has also been analyzed in hEScell differentiation cultures (Clark et al. 2004). In thisstudy, gene expression profiles suggested the develop-ment of both male and female germ cells in serum-stimulated EBs. However, as the cells were not isolatedand further characterized, the efficiency of differentia-tion and the extent of lineage maturation are unknown.

The findings from these different studies clearly dem-onstrate that it is possible to generate germ cells frommouse ES cells and, in doing so, open new and excitingavenues for researchers to study the development of thislineage. Before the model can be widely used for suchstudies, however, it will be important to define the regu-lators that induce germ cell development, preferably in

the absence of serum. With appropriate conditions, itshould be possible to reproducibly generate large num-bers of such cells.

The studies reviewed in the previous sections indicatethat it is possible to generate derivatives of the threeprimary germ layers as well as germ cells from ES cells

differentiated in culture. They also highlight the factthat the efficiency in forming different cell types variesconsiderably as neuroectoderm and certain mesodermderivatives including the hematopoietic, vascular, andcardiac lineages are reasonably easy to generate, whereasthe development of definitive endoderm and its deriva-tives such as mature hepatocytes and pancreatic -cellsare considerably more challenging. Given the complex-ity of lineage development in the early embryo, it is re-markable that it is possible to generate any mature popu-lations in a reproducible manner from differentiated EScells. While the successes to date highlight the impor-tance and potential of this model system, significant im-provements in lineage development, in particular endo-

derm and its derivative populations, will ultimately de-pend on gaining a better understanding of the eventsregulating the induction of the primary germ layers. Toachieve this, it will be important to recapitulate the de-velopmental events of the early embryo in the ES cellsystem, to use serum-free conditions and defined mol-ecules for differentiation, and to develop reporter sys-tems that enable one to quantify lineage development.

Development of the embryo: generation of the earliestcell populations

Early in mouse embryo development, the inner cell massproliferates rapidly and differentiates to generate a popu-

lation of pluripotent cells known as primitive ectoderm.Shortly after implantation, the innermost cells of theprimitive ectoderm cell mass undergo apoptosis andform a cavity through a process known as cavitation(Coucouvanis and Martin 1995). The surviving primitiveectoderm cells that surround the cavity differentiate toform a pseudostratified columnar epithelium, giving riseto a structure known as the epiblast (Fig. 4). Althoughthey retain pluripotentiality, primitive ectoderm/epi-blast cells can be distinguished from the inner cell masson a morphological basis, by the fact that they have up-regulated fgf5 expression (Haub and Goldfarb 1991; He-bert et al. 1991) and down-regulated rex1 expression(Rogers et al. 1991), and by the fact that they can nolonger contribute to chimera formation following blas-tocyst injection (Rossant 1977; Beddington 1983). Theepiblast responds to extrinsic signals and gives rise to theprimary germ layers (Gardner and Rossant 1979) as wellas the primordial germ cells (Ginsburg et al. 1990).

The primary germ layers in the embryo are formedduring the process of gastrulation, which begins at ap-proximately embryonic day 6.5 (E6.5) in the mouse (Tamand Behringer 1997). At the onset of gastrulation, theepiblast cells in the region that will define the posteriorpart of the embryo thicken and form a transient struc-ture known as the primitive streak (Fig. 4). During gas-

Keller




16/28

trulation, epiblast cells traverse the primitive streak andundergo an epithelial to mesenchymal transition, givingrise to mesoderm and definitive endoderm. Fate-map-ping studies have demonstrated that the development ofsubpopulations of mesoderm and endoderm is not ran-dom but rather appears to be controlled, both temporallyand spatially (Kinder et al. 1999). For instance, the firstmesoderm cells to develop are derived from epiblast cellsthat move through the most posterior region of thestreak. This mesoderm colonizes the developing yolk sacand gives rise to the hematopoietic and vascular cells of

the blood islands. Epiblast cells that traverse the streakslightly later and in a more anterior position will formcardiac mesoderm, head mesenchyme, and paraxial me-soderm, whereas epiblast cells that move through themost anterior region of the primitive streak will generateendoderm and axial mesoderm, which gives rise to thenotochord (Tam and Behringer 1997). Cells in the mostanterior region of the epiblast that do not move throughthe primitive streak will form ectoderm. While themechanisms controlling epiblast migration and meso-derm and endoderm induction are not fully understoodin the mouse, expression analysis and gene targetingstudies have shown that members of the TGF familyincluding BMP4 (Hogan 1996) and nodal (Conlon et al.

1994; Schier and Shen 2

embryonic cell differentiation

Documents