Integration and Differentiation of Human EmbryonicStem Cells Transplanted to the Chick EmbryoRONALD S. GOLDSTEIN,1* MICHA DRUKKER,2 BENJAMIN E. REUBINOFF,3 AND NISSIM BENVENISTY2
1Gonda Research Center, Faculty of Life Sciences, Bar-Ilan University, Ramat-Gan, Israel2Department of Genetics, The Institute of Life Sciences, The Hebrew University, Jerusalem, Israel3Department of Obstetrics and Gynecology, Goldyn Savad Institute of Gene Therapy, Hadassah University Hospital,Jerusalem, Israel
Key words: human ES cells, xenograft, neural ro-settes, transplantation therapy, hu-man embryogenesis, neuronal differ-entiation
INTRODUCTION
Human embryonic stem (human ES) cells are pluri-potent cell lines with great potential for transplanta-tion therapy and for the study of early human devel-
opmental processes. These cells form differentiatedtumors (teratomas) when injected into immune-defi-cient mice (Thomson et al., 1998; Amit et al., 2000;Reubinoff et al., 2000). We have shown recently thathuman ES cells also form embryoid bodies (EBs) inculture, that differentiate into cells of all three germlayers (Itskovitz-Eldor et al., 2000). Growth factorsapplied to cells from dissociated EBs led to the produc-tion of multiple cell types from several tissues (Schul-diner et al., 2000). More recently, we demonstratedthat neural differentiation of human ES cells can beenhanced in vitro with use of nerve growth factor andretinoic acid (Schuldiner et al. 2001).
Interspecies transplantation has permitted the studyof both differentiation and therapeutic potential ofstem cells. Several systems have been used to study thebehavior of adult or embryonic stem cells in the post-gastrulation embryo. For example, human adult bone-marrow stromal stem cells were injected into sheepembryos, and differentiation of these cells was followedin utero (Liechty et al., 2001). The chick embryo is awell-characterized and accessible experimental systemfor study of inductive interactions and differentiationin development. Several studies have shown that mam-malian cells and tissues transplanted to avian embryoscan respond to local cues and develop into tissues ap-propriate to their location in the host (i.e., Fontaine-Perus et al., 1997; White and Anderson, 1999). Chickembryos also have been used as a host for the study ofthe migration of mouse embryonic stem cells with andwithout genetic manipulation (Beuvais et al., 1999).
In the present study, we decided to examine whetherthe chick embryo could be used as an in vivo system forthe study of human ES cell differentiation. The advan-tage of using the chick embryo rather than teratomasin rodents is the ability to follow human ES cell devel-
Grant sponsor: The Herbert Cohn Chair; Grant sponsor: The AvivFund for Neuroscience Research; Grant sponsor: The Health ScienceCenter at Bar-Ilan University.
Drs. Goldstein and Drukker contributed equally to this work.*Correspondence to: Ronald S. Goldstein, Faculty of Life Sciences,
opment in a normal embryonic environment, as op-posed to the less-natural environment of adult murinetissues. For our initial studies, we chose to transplanthuman ES cells into or in place of paraxial mesoderm,to expose the cells to the large number of potentialinductive influences that are present in the somites. Inaddition, the somites are both a migration pathwayand a destination for neural crest cells, another cellpopulation whose progeny have many potential pheno-types that the human ES cells could be influenced by.
RESULTS AND DISCUSSION
Colonies of human ES cells were microsurgicallygrafted into the trunk region of 1.5- or 2-day-old (E1.5–E2) chick embryos (Fig. 1A,B). One day after surgery,the operation site was always visible. When carboxy-fluorescein diacetate (CFDA) -labeled or green fluores-cent protein (GFP) -expressing cells were implanted,they were clearly visible in the living embryo by usingfluorescence illumination (Fig. 1C,D). The cells re-mained mostly as clumps, although individual cellscould sometimes be observed migrating away from thesite of implantation (not shown). The graft could beobserved by fluorescence microscopy in some cases aslong as 4 or 5 days after surgery, after fixation andremoval of overlying tissues (not shown).
Human ES Transplanted Into SomiticMesoderm Integrates Into Chick Tissues
The somites give rise to multiple tissue types, includ-ing muscle, dermis, and cartilage/bone. In addition,neural crest cells forming peripheral ganglia migratethrough the somites after their epithelial/mesenchy-
mal transformation. Therefore, we initially implantedGFP-expressing human ES colonies into damagedsomites to see whether they would participate in theproduction of somitic or neural crest derivatives. Im-munostaining for GFP demonstrated that some of thehuman ES remained as clumps (Fig. 2A), some cellsformed distinct columnar (Fig. 2B) or cuboidal (Fig. 2D)epithelial structures, whereas others mingled with thechick cells (Fig. 2C,E–G). Some human ES-derivedcells incorporated into the host dorsal root ganglion(DRG; Fig. 2E). Several of these cells had neuronalmorphology, displaying axon-like processes (Fig. 2E,inset). In addition, elongated human ES-derived cellswere observed lining the outside of the vertebral arch,apparently having contributed to the perichondrium(Fig. 2F,G). In some (20%) preparations, structuresresembling neural rosettes of teratomas developedfrom the grafted cells (see below). Although the humanES were implanted into the somite, morphologic differ-entiation suggestive of the normal somitic derivatives(muscle, dermis, and cartilage) was not observed. An-tidesmin and Alcian blue staining confirmed that thesetissues had not formed from the ES up to 5 days aftersurgery (not shown). The lack of position-specific dif-ferentiation by human ES into these tissues could bedue to the much more rapid organogenesis of aviancompared with human embryos.
Hoechst staining revealed that all GFP� cells con-tained larger nuclei that were usually more intenselystained than the surrounding chick cells (Fig. 2C, in-set). Distinction of human from chick nuclei was alsousually possible in preparations stained with Feulgen(Fig. 3A–D) and hematoxylin and eosin (not shown).
Fig. 1. Schematic representation of the surgery performed. The twoor three most recently formed somites of 12–20 somite chick embryoswere crushed (A) or removed (B) and a colony of human embryonic stem(ES) cells maneuvered into the space generated. C,D: Photomicrographsof live embryos that received grafts of carboxyfluorescein diacetate–
stained HES-1 cell colonies (Reubinoff et al., 2000) 24 hr earlier. In C, theembryo was photographed in situ; in D, the embryo was pinned out in adish for photography. Arrows, ES cell colonies; NT, neural tube; Som,somites.
Fig. 2. Differentiation and integration of human embryonic stem (ES)cells transplanted into chick somites. A: A low-magnification view of theregion of a graft of green fluorescent protein (GFP) -expressing humanES cells into a damaged somite 4 days earlier. GFP-expressing cellswere immunostained (red), and nuclei were stained with Hoechst (blue).Two large masses of GFP� cells are present, one (ES) next to the hostdorsal root ganglion (DRG) and the second, more laterally. In betweenthese masses, GFP� cells are interspersed among chick cells. NT,neural tube. B: The area enclosed by the lower box in A shown at highermagnification includes a GFP� pseudostratified columnar epithelium(Epith). Above the epithelia is a mesenchyme composed of cells withlarge nuclei, some of which express lower levels of GFP. C: The area
enclosed by the upper box in A, is shown at higher magnification. Indi-vidual and clumps of GFP� cells mingle with the chick tissue. GFP� cells(arrows) have larger nuclei than the cells of the host (arrowheads). Theinset shows only the Hoechst staining, which allows easy distinction ofhuman and chick nuclei by size. D: Tubules of cuboidal epithelium (Tub)derived from the grafted cells. Arrow � autofluorescent erythrocyte. E: Aphotomicrograph of a section where many GFP� human ES cells werepresent in the DRG (outlined) of the host. Some of these cells hadprocesses suggestive of neurons (inset). NT, neural tube. F,G: HumanES-derived cells that migrated to the vertebral arches (Vert), where theybecame elongated (arrow in G) and mingled with the chick perichondrialcells. Scale bars � 150 �m in A, 40 �m in B–D,G, 80 �m in E,F.
Similarly, grafted mouse cells can also often be distin-guished from those of host chick embryos by nuclearstaining (Fontaine-Perus et al., 1997). Grafted humanES cells, whether transplanted into or replacingsomites, almost never developed a limiting capsule (1 of20 operations). This is in striking contrast to teratomasthat are bounded by a capsule (Walter and Talbot,1996), which prevents ES-derived cells from integrat-ing into host tissues.
Neuronal Differentiation of Human ESReplacing Somitic Mesoderm
When colonies of human ES cells were implantedadjacent to the neural tube and notocord without in-tervening somitic mesoderm, epithelia reminiscent ofneural rosettes were always (7 of 7 embryos analyzed)observed lateroventral to the chick spinal cord (Fig.3A–F). At embryonic days 6–7, these structures con-tained numerous mitotic figures that were localizedprimarily to their lumenal aspect (Fig. 3D). This ar-rangement of a stratified (or pseudostratified) epithe-lium with mitotic figures adlumenal and not basal, ischaracteristic of neural rosettes in human teratomas(Caccamo et al., 1989) and in the early vertebrate neu-ral tube (see below). The neural rosette-like structurescontained nuclei that were much larger than those ofthe host chick cells (Fig. 3A–D).
Compared with human ES cells transplanted intodamaged somites, this series of grafts contained manyfewer individual cells that migrated away from the siteof the surgery. There were also virtually no clumps ofhuman ES cells with indeterminate morphology or co-lumnar epithelia like those observed in the previous setof operations. However, cuboidal tubules (Fig. 3B) sim-ilar to those present in the damaged somite grafts (Fig.2D) were sometimes present. The development of neu-ral rosette-like structures in all preparations wherecolonies were transplanted adjacent to the host neuraltube with no intervening tissues, suggests that thedifferentiation of human ES can be influenced by theirlocal environment in the chick embryo.
Immunostaining for �-3-tubulin showed that humanES-derived rosette-like epithelia indeed contained neu-ral cells with the pattern found in neural rosettes.Immunopositive neuronal somata were observed at thebase but not the lumen of the epithelia (Fig. 3E,F). Inaddition, fine immunopositive processes were seen ex-tending the diameter of the tubules (Fig. 3F). Theseprocesses were similar to those within which nuclei ofneural-precursors in the embryonic neural tube mi-grate while in S-phase (Sauer, 1935).
In operations where human ES cells replacedsomites, we also intentionally damaged the adjacentneural tube, with the hope that this insult would sendsignals to the graft to differentiate neuronally (Mc-Donald et al., 1999). This damage resulted in attach-ment of human ES-derived cells to the neural tube in 2of 10 preparations (Fig. 3G,H). Human axons wereobserved in the nascent white matter of the chick cen-
tral nervous system (CNS), and chick axons coursedamong the human ES cells (Fig. 3H). This finding wasdemonstrated by confocal examination of double immu-nostaining with an antineurofilament antibody thatrecognizes both mammalian and chick neurons, and amammalian-specific antineurofilament antibody.
Immunostaining showed that some human neuronsalso developed from the human ES in structures thatwere not part of rosettes or the host CNS. Ganglion-like clumps of human neurons were observed near theDRG, as shown by double staining specifically for chicknervous tissue with the HNK-1 antibody and generallyfor vertebrate neurons with �-3-tubulin antibodies(Fig. 3I). In addition, axons from the chick CNS wereobserved traversing the human ES-derived neural ro-sette-like structures (Fig. 3J).
In the present study, we have begun to examine thedevelopment of human embryonic stem cells after theirtransplantation in ovo into chick embryos at earlystages of organogenesis. These experiments are not inconflict with the NIH guidelines for research usinghuman ES cells, because we have not “combined” thecells with the chick embryo but rather transplanted thecells next to partially differentiated tissues. These ex-periments are also different from those attempting toproduce true chimeric embryos where stem cells aremixed or injected into intact embryos at the blastula/gastrula stage and could potentially make a major con-tribution to many of the host tissues.
We found that human ES proliferate and differenti-ate in ovo, apparently into several cell types, includingcells with morphological and molecular characteristicsof neurons. Moreover, we found that the position of thegraft influenced the developmental pathways selected,because grafts in direct contact with the embryo’s axialstructures always gave rise to neural rosettes, andgrafts into existing somitic mesoderm produced cellsthat mingled with those of the host and contributed tothe DRG and the perichondrium. It is actually quitestriking and even surprising that human ES cells formrosettes and neurons as rapidly as we observed, be-cause it takes 4 days as embryoid bodies and anotherweek of culture in fibroblast growth factor-2 (11 daystotal) to obtain rosettes in vitro (Zhang et al., 2001),and teratomas take 1–2 months to form in SCID mice(Thomson et al., 1998).
We have also shown that grafted human ES cells notonly differentiate but also migrate away from thetransplanted colony and can integrate into surround-ing tissues such as DRG, neural tube, and the peri-chondrium of the vertebral primordium. In this way, inovo grafting of human ES cells is very different from invivo grafting of these cells into adult immune-compro-mised mice, which results in the production of terato-mas that are isolated from host tissues. Human EScells rapidly differentiate and integrate into host tis-sues in vivo and, thus, may serve as a convenientsystem for studying differentiation and morphogenesisof human cells and tissues, as well as to examine the
potential of human ES cells for transplantation. In ovografts of human ES may aid us to exploit their tremen-dous potential as a source of human cells for transplan-tation therapy in disease and traumatic injury, and inthe study of human embryonic development.
EXPERIMENTAL PROCEDURESCell Culture
Human ES cells were grown as previously described(Schuldiner et al., 2000; Eiges et al., 2001). Briefly, EScells were initially cultured on a mitomycin-C–treatedmouse embryonic fibroblast (MEF) feeder layer (ob-tained from day 13.5 embryos) in 80% KnockOutDMEM medium (Gibco-BRL), supplemented with 20%KnockOut SR—a serum-free formulation (Gibco-BRL),1 mM glutamine (Gibco-BRL), 0.1 mM �-mercaptoetha-nol (Sigma), 1% nonessential amino acids stock (Gibco-BRL), penicillin (50 units/ml), streptomycin (50 �g/ml),and 4 ng/ml basic fibroblast growth factor. Before graft-ing, human ES were cultured for 1–3 days on gelatin-coated plates without MEFs to obtain a pure popula-tion for grafting. In some experiments, cells werevitally stained with Vybrant CFDA (Molecular Probes),according to the manufacturer’s instructions.
Microsurgery
Fertile chicken eggs were incubated from 40–45 hrto obtain embryos of 10–20 somite pairs. A smallamount of India ink was injected sub-blastodermally,and a tear was made in the vitelline membrane abovethe 3–4 most recently formed somites. In some exper-iments, colonies were implanted laterally into somitesmanually damaged with a microscalpel (Fig. 1A). Inother experiments, somites were removed after 5–10min of enzymatic digestion and the space formed wasfilled with a colony of human ES (Fig. 1B). Approxi-mately 100–200 cells were implanted. Eggs were thensealed with cellotape and incubated an additional 1–5
days. Survival of the embryos was between 50 and100% 1 day after surgery, at 5 days 20–50%. Thegrafted cells were found in sections in 50–75% of thesurviving operations. In all, results presented arebased on analysis of 15 grafts recovered from approxi-mately 100 operations.
Histology and Immunocytochemistry
After the second incubation period, embryos wereremoved from the egg, rapidly fixed in buffered para-formaldehyde, and embedded in paraffin. Serial sec-tions were prepared and stained with antibodies byusing microwave antigen retrieval, or Feulgen andFast Green. Antibodies used were rabbit antineurofila-ment (NF) 200 (N4142, Sigma), mammalian-specificanti-NF 160 (clone 2H3, DSHB), rabbit anti-GFP (San-ta-Cruz), mouse anti–HNK-1 (ATCC), and anti–�-3 tu-bulin (TUJ1 [gift from Avihu Klar] or 5B8, Promega).Detection was performed with fluorescein/Texas Red orAlexa 488/594 conjugated secondary antibodies, andimages were captured digitally from an OlympusBX-60 compound or Bio-Rad MRC 600 and 1024 confo-cal microscopes.
ACKNOWLEDGMENTS
The experiments described in this manuscript wereapproved by the ethics committees of both Bar-IlanUniversity and the Hebrew University. Monoclonal an-tibody 2H3 was generated by Tom Jessel and obtainedfrom the Developmental Studies Hybridoma Bank es-tablished under the auspices of the NICHD and main-tained by the University of Iowa, Department of Bio-logical Science, Iowa City, IA, 52242. The excellenttechnical assistance of Drs. A. Chipman and O. Yanukaand of Ms. C. Avivi greatly contributed to this study.We also thank Douglas Melton and members of theBenvenisty and Goldstein laboratories for their in-sightful comments. N.B. received funding from The
Fig. 3. Neural differentiation of human embryonic stem (ES) cells inthe chick observed in sections through chick embryos receiving grafts ofhuman ES cells that replaced epithelial somites. In A–D, sections werestained with Feulgen and counterstained with Fast Green. A: A photomi-crograph of an embryo 2 days after grafting at embryonic day 4, in whichdistinct tubular structures have differentiated near the neural tube (NT)from the ES cells. The epithelium of human cells(asterisk) is much darkerstaining and easily recognized even at this low magnification. DRG,dorsal root ganglion; No, notochord. B: At higher magnification, part ofthe graft is identifiable as an early neural rosette-like epithelium (asterisk)and other cells make tubules of cuboidal epithelium (Tub). C: A low-power photomicrograph from an embryo 5 days after grafting, stainedalso with Alcian Blue to visualize cartilage. A neural-rosette–like structure(Ros) is enclosed within the vertebra (Vert). D: Numerous mitotic figuresare observed adjacent to the lumen, two of which are indicated by arrowsin this higher magnification image. E: A section through a graft stainedwith neuron-specific tubulin (green) and Hoechst nuclear dye (blue) isshown. The small red cells observed are erythrocytes. WM, white matter;GZ, germinal zone. F: Neuronal somata (arrows) at the basal aspect ofthe epithelium and processes running perpendicular to its diameter (ar-rowhead) are clearly defined when the section is viewed at higher mag-nification. The hollow arrow points to a chick spinal nerve. G: A section
through an operation where the grafted human ES cells (GR) fused withthe host neural tube stained with rabbit antibodies to vertebrate neuro-filament 200 (detected with fluorescein isothiocyanate secondary anti-body, green), mouse monoclonal mammalian-specific neurofilament 160(detected with Texas Red secondary antibody, red), and Hoechst (blue).The graft is lateral to the chick’s WM (compare with E), and the ventricularGZ of the neural tube is distorted on the side of the graft (compare withthe contralateral side and E). H: A projection of a Z-series of imagesmade with a confocal microscope in a section adjacent to G. The inte-gration of the human (red) and chick (green) tissues is striking, withhuman axons (bold arrows) coursing through the chick WM, and chickaxons penetrating between the grafted human cells (arrows). I: Humanneurons forming a ganglion-like structure (filled arrow). This section wasstained with antibodies to HNK-1 (green), which is specific for the aviannervous system, and neuron-specific tubulin (red), which recognizes bothhuman and chick neurons. The host DRG and nerve (hollow arrow) arestained yellow, because they are positive for both HNK-1 and neuron-specific tubulin. The nuclei of human neurons are larger even than thelarge neurons of the DRG nearby. J: Chick axons (bottom hollow arrow)traversing a glancing section through a Ros, which contains neuron-specific tubulin single-labeled axons, and neurons (arrows) are observedin this section stained the same as that in I.
Herbert Cohn Chair and R.S.G. received funding fromthe Health Science Center at Bar-Ilan University.
REFERENCES
Amit M, Carpenter MK, Inokuma MS, Ciu C-P, Harris CP, WaknitzMA, Itskovitz-Eldor J, Thomson JA. 2000. Clonally derived hu-man embryonic stem cell lines maintain pluripotency and prolif-erative potential for prolonged periods in culture. Dev Biol 227:271–278.
Beuvais J, Pla P, Bernex F, Dufour S, Salamero J, Fassler R, PanthierJ-J, Thiery JP, Larue L. 1999. A novel model to study the dorsolat-eral migration of melanoblasts. Mech Dev 89:3–14.
Caccamo DV, Herman MM, Franfurter A, Katestos CD, Collins VP,Rubinstein LJ. 1989. An immunohistochemical study of neuropep-tides and neuronal cytoskeletal proteins in the neuroepithelial com-ponent of a spontaneous murine ovarian teratoma. Am J Pathol135:801–813.
Eiges R, Schuldiner M, Drukker M, Yanuka O, Itskovitz-Eldor J,Benvenisty N. 2001. Establishment of human embryonic stem cell-transfected clones carrying a marker for undifferentiated cells.Curr Biol 11:514–518.
Fontaine-Perus JC, Halgand P, Cheraud Y, Rouaud T, Velasco ME,Diaz CC, Rieger F. 1997. Mouse chick chimeras: a developmentalmodel of murine neurogenic cells. Development 124:3025–3036.
Itskovitz-Eldor J, Schuldiner M, Karsenti D, Eden A, Yanuka O, AmitM, Soreq H, Benvenisty N. 2000. Differentiation of human embry-onic stem cells into embryoid bodies comprising the three embryonicgerm layers. Mol Med 6:88–95.
Liechty KW, Mackenzie TC, Shaaban AF, Radu A, Moseley AM,Deans R, Marshak DR, Flake AW. 2001. Human mesenchymal stemcells engraft and demonstrate site-specific differentiation after inutero transplantation in sheep. Nat Med 6:1282–1286.
McDonald JW, Liu XZ, Qu Y, Liu S, Mickey SK, Turetsky D, GottliebDI, Choi DW. 1999. Transplanted embryonic stem cells survive,differentiate and promote recovery in injured rat spinal cord. NatMed 5:1410–1412.
Reubinoff BE, Pera MF, Fong CY, Trounson A, Bongso A. 2000.Embryonic stem cell lines from human blastocysts: somatic differ-entiation in vitro. Nat Biotech 18:399–404.
Sauer FC. 1935. Mitosis in the neural tube. J Comp Neurol 62:377–405.Schuldiner M, Yanuka O, Itskovitz-Eldor J, Melton DA, Benvenisty
N. 2000. Effects of eight growth factors on the differentiation of cellsderived from human embryonic stem cells. Proc Natl Acad Sci U SA 97:11307–11312.
Schuldiner M, Eiges R, Eden A, Yanuka O, Itskovitz-Eldor J, Gold-stein RS, Benvenisty N. 2001. Induced neuronal differentiation ofhuman embryonic stem cells. Brain Res 913:201–205.
Thomson JA, Itskovitz-Eldor J, Shapiro SS, Waknitz MA, SwiergielJJ, Marshall VS, Jones JM. 1998. Embryonic stem cell lines derivedfrom human blastocysts. Science 282:1145–1147.
Walter JB, Talbot IC. 1996. General pathology. 7th ed. New York:Churchill Livingstone. p 472–476.
White PM, Anderson DJ. 1999. In vivo transplantation of mammalianneural crest cells into chick hosts reveals a new autonomic sublin-eage restriction. Development 126:4351–4363.
Zhang S-C, Wernig M, Duncan ID, Brustle O, Thomson JA. 2001. Invitro differentiation of transplantable neural precursors from hu-man embryonic stem cells. Nat Biotech 19:1129–1133.
