Small-Molecule Induction of Neural Crest-like Cells Derived from Human Neural Progenitors§

Authors

  • Ryo Hotta,

    1. Department of Anatomy & Cell Biology,, The University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author
  • Lana Pepdjonovic,

    1. Centre for Neuroscience, and, The University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author
  • Richard B. Anderson,

    1. Department of Anatomy & Cell Biology,, The University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author
  • Dongcheng Zhang,

    1. Murdoch Children's Research Institute, Royal Children's Hospital, Parkville, Victoria, Australia
    Search for more papers by this author
  • Annette J. Bergner,

    1. Department of Anatomy & Cell Biology,, The University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author
  • Jessie Leung,

    1. Centre for Neuroscience, and, The University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author
  • Alice Pébay,

    1. Centre for Neuroscience, and, The University of Melbourne, Parkville, Victoria, Australia
    2. Department of Pharmacology, The University of Melbourne, Parkville, Victoria, Australia
    Search for more papers by this author
  • Heather M. Young,

    Corresponding author
    1. Department of Anatomy & Cell Biology,, The University of Melbourne, Parkville, Victoria, Australia
    • The University of Melbourne, Department of Anatomy and Cell Biology, Level 5, Medical Building, Parkville, Victoria, Australia 3010
    Search for more papers by this author
    • Telephone: 61 3 8344 5770; Fax: 61 3 9347 5219

  • Donald F. Newgreen,

    Corresponding author
    1. Murdoch Children's Research Institute, Royal Children's Hospital, Parkville, Victoria, Australia
    • Murdoch Children's Research Institute, Department of Embryology, 10th Floor, Royal Children's Hospital, Parkville, Victoria, Australia 3052
    Search for more papers by this author
    • Telephone: 61 3 8341 6276; Fax: 61 3 9348 1391

  • Mirella Dottori

    Corresponding author
    1. Centre for Neuroscience, and, The University of Melbourne, Parkville, Victoria, Australia
    2. Department of Pharmacology, The University of Melbourne, Parkville, Victoria, Australia
    • The University of Melbourne, Centre for Neuroscience and Department of Pharmacology, Level 7, Medical Building, Parkville, Victoria, Australia 3010
    Search for more papers by this author
    • Telephone: 61 3 8344 3988; Fax: 61 3 9349 4432


  • Author contributions: R.H.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript; L.P., R.B.A., D.Z., A.J.B.: conception and design, collection and/or assembly of data, data analysis and interpretation, final approval of manuscript; J.L.: collection and/or assembly of data, data analysis and interpretation, final approval of manuscript; A.P.: conception and design, financial support, data analysis and interpretation, final approval of manuscript; H.M.Y., D.F., M.D.: conception and design, financial support, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript.

  • First published online in STEM CELLS EXPRESS August 26, 2009.

  • §

    Disclosure of potential conflicts of interest is found at the end of this article.

Abstract

Neural crest (NC) cells are stem cells that are specified within the embryonic neuroectodermal epithelium and migrate to stereotyped peripheral sites for differentiation into many cell types. Several neurocristopathies involve a deficit of NC-derived cells, raising the possibility of stem cell therapy. In Hirschsprung's disease the distal bowel lacks an enteric nervous system caused by a failure of colonization by NC-derived cells. We have developed a robust method of producing migrating NC-like cells from human embryonic stem cell–derived neural progenitors using a coculture system of mouse embryonic fibroblasts. Significantly, subsequent exposure to Y27632, a small-molecule inhibitor of the Rho effectors ROCKI/II, dramatically increased the efficiency of differentiation into NC-like cells, identified by marker expression in vitro. NC-like cells derived by this method were able to migrate along NC pathways in avian embryos in ovo and within explants of murine bowel, and to differentiate into cells with neuronal and glial markers. This is the first study to report the use of a small molecule to induce cells with NC characteristics from embryonic stem cells that can migrate and generate neurons and support cells in complex tissue. Furthermore, this study demonstrates that small-molecule regulators of ROCKI/II signaling may be valuable tools for stem cell research aimed at treatment of neurocristopathies. STEM CELLS 2009;27:2896–2905

INTRODUCTION

During embryogenesis, neural crest (NC) cells are specified within the dorsal neural tube and undergo epithelio-mesenchymal transition (EMT). This is mediated by growth factors, including Bone Morphogenic Proteins (BMPs), Fibroblast Growth Factors (FGFs), and Wnts from the surrounding tissues [1–4] These cells then migrate to various regions of the embryo to further differentiate into many cell types, including neural and glial cells of the peripheral nervous system, melanocytes, cardiac smooth muscle, and craniofacial cartilage. The differentiation of NC-derived cells into various lineages depends upon their migratory pathways within the embryo, environmental signals released by surrounding tissues, and intrinsic factors [5, 6].

Neurocristopathies are diseases that arise from defects in NC development. One of the best characterized neurocristopathies is Hirschsprung's disease (congenital aganglionosis or megacolon), which is caused by an absence of neurons in the distal gastrointestinal tract, whereas most of the gut has a normal and functional enteric nervous system. Patients with Hirschsprung's disease suffer from severe constipation because enteric neurons are essential for peristalsis. Most enteric neurons arise from NC-derived cells that emigrate from the caudal hindbrain, enter the foregut, and then migrate in an oral-to-anal direction within the gut wall to colonize the entire gastrointestinal tract [7, 8]. Hirschsprung's disease typically is caused by a failure of NC-derived cells to colonize the distal (aganglionic) regions of the gut. Experimental and biomathematical studies have shown that a reduction in the number of pre-enteric or enteric NC-derived cells results in Hirschsprung's disease-like symptoms [9–11]. The genetics and molecular mechanisms underlying Hirschsprung's disease are increasingly well understood [8, 12]. Currently, the only treatment for Hirschsprung's disease is surgical removal of the affected segment of intestine, but patients commonly suffer from ongoing problems following surgery, including soiling and incontinence [13], and these may lead to major psychosocial problems. Hirschsprung's disease seems to be principally a quantitative defect in NC progenitors, and thus this condition could potentially be ameliorated by stem cell therapy after birth.

Cell therapy to generate enteric neurons in the aganglionic bowel is a potential alternative treatment to surgery for Hirschsprung's disease [14–17]. Several sources of enteric neurons have been investigated in in vitro studies and in animal models—these include fetal central nervous system (CNS) stem cells, enteric neural stem/progenitor cells isolated from the embryonic or postnatal gut, and NC cells isolated from peripheral nerves or from neural tube explants [14, 16, 18–27]. Embryonic stem (ES) cells are also a potential source of unlimited quantities of enteric neural stem cells that might be used in cell replacement therapy to treat Hirschsprung's disease and other enteric neuropathies. Mouse ES cells can give rise to “gut-like” structures in vitro that contain neurons [28–30]. However, the neurons in these ES-derived structures have different trophic requirements from enteric neurons. Enteric neuron development is dependent on glial cell line–derived neurotrophic factor (GDNF) and neurotrophin-3 (NT-3), but not brain-derived neurotrophic factor (BDNF) [8, 31]. In contrast, the formation of neural networks in mouse ES–derived gut-like structures requires BDNF, but not GDNF or NT-3 [29]. The ability of mouse ES–derived gut-like structures to produce neurons in host gut following transplantation in vivo, or in gut explants in vitro, has yet to be examined.

In the current study we describe a highly efficient method for deriving NC-like cells from human embryonic stem cells (hESC) using feeder cells and the small-molecule ROCK1/2 inhibitor Y27632. Such hESC-derived NC-like cells migrated along NC pathways and differentiated into neurons when implanted into avian embryos, and when cocultured with embryonic aneural mouse gut. Some of the hESC-derived neurons in gut explants expressed an enteric neuron subtype marker. These results are the first report of the integration of hESC-derived cells into the developing gut. Moreover, the results demonstrate that hESC-derived cells are a promising source of an unlimited number of human NC-like cells for the study of human NC stem cell biology and investigation of cell therapies for some neurocristopathies.

MATERIALS AND METHODS

Cell Culture

HES3 (WiCell, www.wicell.org), HES4 (WiCell), and ENVY (ES Cell International, www.escellinternational.com) human embryonic cell lines were cultured in 20% serum-containing medium on mitomycin C–treated mouse embryonic fibroblasts (mEFs; 6 × 104 cells per cm2) and passaged weekly by mechanical dissociation. For neural induction, 500 ng/ml recombinant noggin (R&D Systems Inc., Minneapolis, http://www.rndsystems.com) was added to hESC medium at the time of colony transfer onto mEFs and cells were cultured for 14 days without passage. After noggin treatment, colonies were mechanically transferred to individual wells of a nonadherent 96-well plate to allow neurosphere formation. Neurospheres were cultured in suspension in neural basal media (NBM) supplemented with epidermal growth factor (EGF) and basic fibroblast growth factor (bFGF) (20 ng/ml each; R&D Systems). These steps are detailed by Dottori and Pera [32]. Subsequent differentiation to NC was carried out by plating day 4 old neurospheres onto a feeder layer (6 × 104 cells per cm2) of mitomycin-treated mEFs or PA6 cells in NBM supplemented with bFGF and EGF (20 ng/ml each) for 24 hours. After 24 hours, the medium was replaced with fresh medium with or without Y27632 (25 μM; Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com). In some experiments, Y27632 was added to the culture medium at the time of plating. To examine the effects of inhibition of bone morphogenetic protein (BMP), canonical Wnt, and/or FGF2 signaling, neurospheres were cultured on mEFs as described above in the presence of noggin (500 ng/ml; R&D Systems), dickkopf1 (Dkk1, 100 ng/ml; Peprotech, Rocky Hill, NJ, http://www.peprotech. com), and/or SU5409 (20 μM; Merck & Co., Whitehouse Station, NY, http://www.merck.com) for 24 hours. After 24 hours, Y27632 (25 μM) was added to each culture in fresh media combined with its corresponding inhibitor.

Immunostaining

Cultures, quail embryos, and gut explants were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer overnight. Quail embryos were then cryosectioned. The antibodies used were rabbit anti-SoxE [33] (1:2,500), goat anti-GFP (1:400; Rockland Immunochemicals, Inc., Gilbertsville, PA, http://www.rockland-inc.com), sheep anti-NOS [34] (1:1,000), mouse anti-Hu (1:100; Invitrogen, Carlsbad, CA, http://www.invitrogen.com), human anti-Hu [35], rabbit anti-p75 (1:500; Promega, Madison, WI, http://www.promega.com), mouse anti-S100b (clone SH-B1; Sigma-Aldrich), and mouse anti-HNK1 (hybridoma; MCRI, Royal Children's Hospital, Parkville, VIC, Australia, http://www.mcri.edu.au).

Analysis of Cell Migration

Neurospheres plated on mEF, PA6, or fibronectin substrates were photographed using an Olympus CKX41 inverted microscope and Nikon Coolpix 4500 digital camera system. Cell migration of each plated neurosphere was assessed as being “positive” if at least 50 isolated cells were observed in each image. For measurement of the extent of migration, ENVY-derived neurospheres plated on mEF were photographed using an Olympus IX81 inverted fluorescent microscope. For each image, the total area covered by GFP+ cells was measured in pixels using the Image J software program. Images and measurements were taken at 24 hours after plating and again at 24- and 48-hour treatment with Y27632 or media. All data are expressed as mean ± SEM. Movies were made on an Olympus IX70 microscope with the Image-Pro Plus 4.5 program.

In Ovo Transplantation

hESC-derived neurospheres were cultured on mEFs in medium with and without 25 μM Y27632 for 24 and 48 hours and then mechanically harvested into small fragments up to the size of a quail somite (200-μm diameter or less) at E2. With use of techniques identical to chick-quail grafting [36], the fragments were implanted into E2 (HH13) quail embryos, in the segmental plate next to the host neural tube just caudal to the last somite. Incubation was then continued until E5 (HH26-28).

Coculture of Neurospheres with Explants of Embryonic Aneural Gut

Explants of postcaecal hindgut were suspended across the “V” cutout in filter paper supports. Neurospheres were cut in half, and each half was apposed to the rostral end of explants of hindgut and cultured for 8 days as described previously [37, 38].

RESULTS

Derivation of NC-like Cells from Human Embryonic Stem Cells

Neural progenitors can be efficiently derived from hESC using the BMP 2/4/7 inhibitor, noggin [32, 39]. Briefly, hESC colonies were treated with noggin for 14 days to induce differentiation toward neuroectoderm. Noggin-treated hESC were then harvested and cultured in suspension in NBM supplemented with bFGF and EGF to generate neurospheres. Neurospheres derived by this protocol have been characterized, showing they can differentiate to neural and glial cell types in vitro and in vivo [39–41]. We have previously reported that neural progenitors derived from noggin-treated hESC, when cultured as neurospheres, show a downregulation of Pax6 and an upregulation of the dorsal markers Pax3 and Pax7 [41]. This suggests that their differentiation may be biased toward dorsal neural tube lineages, including NC cells. Since neurospheres consist of heterogeneous neural progenitor cell types, we sought to modify the neural differentiation protocol to enrich for NC-like progenitors.

During embryogenesis, the NC is specified in the dorsal regions of the embryonic neural tube by factors released from surrounding tissues, including Wnts, FGF, and BMP proteins [2]. Previous studies have identified that mEFs also secrete FGF and Wnt proteins and thus may be able to induce NC cells [42, 43]. To examine this possibility, we cultured hESC-derived neurospheres on a monolayer of mitomycin-treated mEFs. After 24 hours of culture, only 17.4 ± 10.8% of the neurospheres showed migrating cells, with 90.3 ± 6.5% of such cells immunopositive for the NC marker p75 (Fig. 1A–1C). In contrast, about 70% of neurospheres cultured on fibronectin showed migrating cells but very few were p75+ (5.3 ± 5.3% p75+ migrating cells, p < .001; Fig. 1A). HNK1 and Sox8, Sox9, and Sox10 are also upregulated during NC EMT. Consistent with a NC phenotype, migrating cells from the neurosphere-mEF cocultures were also immunopositive for HNK1 and SoxE (the antibody recognizes Sox8, Sox9, and Sox10 proteins; Fig. 1H–1K, supporting information Fig. 1). Thus, as assessed by fundamental NC properties of cell migration and NC marker expression, mEFs promote the differentiation of hESC-derived neural progenitors toward migrating NC-like cells.

Figure 1.

Induction of migrating NC from hESC-derived neural progenitors. (A): Quantification of p75+ cells within the population cells that migrate from neurospheres plated on mouse embryonic fibroblasts (mEFs) or fibronectin. At 24 hours after plating, cultures were treated with 25 μM Y27632 or media only. Cells were fixed and immunostained the following day. Cell counts were performed on 100-700 migrating cells per experiment (n = 3 experiments). Data were analyzed using one way ANOVA with Tukey's multiple comparison test. ***, p < .001. (B, C): Neurospheres cultured on mEFs show migrating p75+ cells (B) and DAPI counterstain (C). Scale bar = 200 μm. (D): Quantification of neurospheres that display cell migration after 24 hours when cultured on mEFs or fibronectin, with or without 25 μM Y27632. n denotes number of experiments, and each experiment had between 3 and 7 neurospheres per condition. Data were analyzed using one way ANOVA with Tukey's multiple comparison test. ***, p < .001. (E, F): Neurospheres derived from ENVY-hESC cell line were cocultured on mEFs for 24 hours and then treated with medium only (E) or medium with Y27632 (F). Representative images show extent of cell migration in each condition at 24 hours after treatment. Scale bar = 500 μm. (G): Graph showing extent of migration across 2 days in cocultured neurospheres with or without 25 μM Y27632. Y scale shows the ratio value of total area (pixels) covered by GFP+ cells at 24 hours after treatment (24h) or at 48 hours after treatment (48h), divided by total area (pixels) of GFP+ cells at 24 hours after plating. Data were analyzed using repeated measures one way ANOVA with Tukey's multiple comparison test. n denotes the number of samples. ***, p < .001; *, p < .05. (H–K): Representative pictures showing migrating cells from cocultured neurospheres (derived from ENVY-hESC) treated with 25 μM Y27632, coexpressed SoxE (H), p75 (I), and GFP (J). (K) Triple immunostaining showing coexpression of SoxE (red), p75 (pink), and GFP (green). Scale bar = 200 μm. Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; Fibro, fibronection.

Proportion of Migrating NC-like Derived from hESC Is Significantly Increased by Y27632 Treatment

Previously, NC properties including EMT have been generated from avian neural epithelium in vitro by drugs that disrupt actin bundles and dissociate N-cadherin junctions [44]. Similar effects are seen on neural epithelium generated from mouse ESC [45]. More recently, the small-molecule ROCK1/2 inhibitor Y27632 was shown to be a potent inducer of EMT in avian neural epithelium explants [46]. We tested the effect of Y27632 on neurospheres that had been exposed to mEFs for 24 hours.

After Y27632 treatment (25 μM) for 24 hours, there was a fivefold increase in the proportion of neurospheres that displayed cell migration (mEF, 17.4 ± 10.8%; mEF + Y27632, 87.0 ± 5.7% of spheres displaying migration; p < .001; Fig. 1D–1F). In addition, with use of the GFP-expressing ENVY ES cell line, the area of GFP+ cell migration on the substrate (compared to the area of the neurosphere itself) was significantly greater in neurospheres cultured on mEFs with 25 μM Y27632 compared to that of controls on mEFs, both at 24 hours (p < .05) and 48 hours after treatment (mEF + Y27632, 3.35 ± 0.20-fold increase in area of GFP+ cells after 48 hours, p < .001; mEF, 2.63 ± 0.17-fold increase; Fig. 1E–1G). Significantly, nearly all (97.7 ± 0.7%) of the migrating cells were immunopositive for p75 in neurospheres cocultured with mEFs and also treated with Y27632 (Fig. 1A, 1H–1K, supporting information Fig. 2). In addition, the Y27632-treated cells showed a more dynamic multipolar form of motility (supporting information Fig. 3, supporting information Movie 1, supporting information Movie 2). In contrast, when Y27632 was added to neurospheres plated on fibronectin, the proportion that displayed cell migration was not significantly different from controls (fibronectin + Y27632, 75.0 ± 15.8% of neurospheres with cell migration, compared to fibronectin alone, 66.7 ± 12.2%) (Fig. 1D). In other points of difference from mEF cocultures, regardless of Y27632, few migrating cells derived from neurospheres plated on fibronectin were p75+ (fibronectin + Y27632, 8.0 ± 6.6% of migrating cells, fibronectin alone, 5.3 ± 5.3%; p > .05) and most migrating cells expressed the glial marker S100β (Figs. 1A, 2). Taken together, these data suggest that after exposure to mEFs, Y27632 significantly enhances the yield of migrating NC-like cells from hESC-derived neural progenitors.

Figure 2.

Neurospheres plated on fibronectin with or without Y27632 treatment. Neurospheres plated on fibronectin for 24 hours and then treated with medium only (A) or medium treated with Y27632 (B). Migrating cells from neurospheres treated with Y27632 are immunopositive for the glial marker, S100β (C), and very few cells, if any, are p75+ (E). (D, F): DAPI counterstains of (C) and (E), respectively. Scale bar = 200 μm.

Recent studies have shown induction of NC cell progenitors from hESC-derived neural progenitors using PA6 and MS-5 stromal cell lines [47, 48]. Y27632 improved the generation of NC-like cells equally when added to neurospheres cultured on mEFs or on the stromal cell line PA6 (Fig. 3). We tested the time of Y27632 treatment relative to mEF exposure for efficiency of NC induction by adding Y27632 at the time of plating neurospheres onto mEFs and after 24-hour exposure to mEFs. We found a significant enhancement of cell migration in the latter condition (p = .0002; Fig. 3). These data suggest that neural progenitors require priming by factors secreted by mEFs or PA6 cells to become specified as NC-like and that inducing migration alone does not specify NC fate. Nevertheless, once the neural progenitors are primed, addition of Y27632 significantly augments the number of NC-like cells arising in culture.

Figure 3.

Neurospheres cultured on mEFs versus PA6, with or without Y27632 treatment. Quantification of neurospheres that display cell migration after 24 hours when cultured on mEFs (A) or PA6 (B) with or without Y27632 at different time points. The different time points of conditions were as follows: no treatment; Y27632 added at time of plating neurospheres onto mEFs or PA6 (+pl); Y27632 added at time of plating neurospheres and again at 24 hours after (+pl + 24); or only at 24 hours after plating (+24). (mEF, 17.4 ± 10.8%; mEF + pl, 11.0 ± 11%; mEF + pl + 24, 66.7 ± 33.3%; mEF + 24, 86.6 ± 5.8%; PA6, 11.0 ± 7.0%; PA6 + pl, 41.7 ± 22.0%; PA6 + pl + 24, 91.7 ± 8.3%; PA6 + 24, 79.3 ± 11.3%). Data were analyzed using ANOVA with Tukey's multiple comparison test. Data are expressed as the means ± SEM. Error bars indicate SEM. n denotes number of experiments, and each experiment had between 3 and 10 replicates per condition. Abbreviations: ***, p < .001; **, p < .01; mEFs, mouse embryonic fibroblasts.

NC cells in vivo are induced by FGFs, Wnts, and BMP-4 [1, 2]. To investigate which factors may be secreted by mEFs, neurospheres were cocultured on mEFs in the presence of the BMP antagonist, noggin, the canonical Wnt antagonist, Dkk1, or the small-molecule inhibitor of FGF2 signaling, SU5402. There was no significant difference in the proportion of neurospheres that displayed migrating p75+ cells in cultures treated with each individual inhibitor or in any combination (supporting information Fig. 4). This suggests that additional or sequential factors may be required for NC specification.

hESC-Derived NC-like Cells Migrate and Differentiate in Quail Embryos

The ability of hESC-derived NC-like cells to behave like NC cells was assayed first by allowing them access to NC migration and differentiation sites by in ovo implantation into quail embryos. Neurospheres derived from hESC were cultured on mEFs for 24 hours, and then in medium with and without 25 μM Y27632 for 24 and 48 hours. Neurosphere fragments up to the size of a somite were implanted into E2 (HH13) quail embryos. The site of implantation, the segmental plate next to the host neural tube (supporting information Fig. 5), ensured that the GFP+ hESC-derived graft cells had access to NC migratory pathways. After development to E5 (HH26-28), neurospheres fragments not treated with Y27632 (n = 9) remained as a coherent group in the form of a rosette with a lumen and laminin+ basement membrane, similar to an early neural tube in the host paraxial mesenchyme (Fig. 4B, supporting information Fig. 6A). Neural tube markers such as NCAM and N-cadherin were expressed by implanted neurospheres, and some cells expressed markers of neuronal (Hu) and glial (S100β) differentiation. A few GFP+ cells were observed outside, but close to, some transplanted neurospheres.

Figure 4.

Transplantation of hESC-derived progenitors into quail embryos. Transverse sections through quail embryos 2 days after implantation into paraxial tissue of GFP+ve hESC-derived neurospheres that were initially cultured on mEFs (mouse embryonic fibroblasts) in medium with (A), (C), (D), and (E) and without (B) 25 μM Y27632 for 24 hours. (A): Grafts of neurosphere fragments treated with Y27632 showed a distribution of GFP+ cells within regions of endogenous neural crest-derived cells, including the dorsal root ganglia, ventral root (solid arrow), and primary sympathetic ganglia (open arrow). Scale bar = 100 μm. (B): Neurospheres not treated with Y27632 remained aggregated within the somite and consisted of rosette-like structures. Scale bar = 100 μm. (C, D): GFP+ cells in dorsal root ganglia of quail embryos that received Y27632-treated implants showed neuronal differentiation. The sections were immunostained using antibodies to the pan-neuronal marker, Hu (red). GFP+ cell in the dorsal part of the ganglion projects a process (C, arrows) to the neural tube. Blood cells (asterisks) show nonspecific fluorescence. Scale bar = 20 μm. (D): A GFP+ cell (arrow) in the ventral part of the ganglion projects a process toward the ventral root. Scale bar = 20 μm. (D, D): Inset of (D), showing that the GFP+ cell with an axon (D, arrow) was also Hu+ (D, arrow). (E, E): GFP+ cells located along the ventral nerve roots were S100β+ (arrows). Scale bar = 20 μm. Abbreviations: noto, notochord; DRG, dorsal root ganglia.

Grafts of neurosphere fragments treated with Y27632 (n = 20) formed less coherent cell masses of irregular shape and incomplete laminin+ve basement membrane (supporting information Fig. 6B), and extending from these were numerous widely distributed GFP+ cells (Fig. 4A). GFP+ cells that were immunoreactive for the early NC and glial marker SoxE were found in regions of endogenous NC migration, near and within the nerve roots of the host (18/20 implants), including the ventral nerve roots (supporting information Fig. 6C) and at the primary sympathetic chain (Fig. 4A). GFP+/SoxE− cells expressing the pan-neuronal marker, Hu, identifying them as graft-derived neurons, occurred in the dorsal root ganglia of 13/20 Y27632 pretreated implants, but were relatively sparse (GFP+/Hu+ cells were present in about every second section) (Fig. 4C, 4D, 4D′, 4D″). GFP+/Hu+ cells within ganglia were usually present individually, not in clumps. Apparently unbranched axons projected from these GFP+/Hu+ cells into the dorsal horn of the neural tube or to the periphery via the ventral root (Fig. 4C, 4D). Most of the scattered GFP+/SoxE+ve cells in the ventral roots were immunopositive for the glial marker, S100b (Fig. 4E′, 4E″). Taken together, these results show that the hESC-derived NC-like cells behaved similarly to NC cells in vivo in that they distributed along NC migratory pathways and differentiated into neural and glial cells within the embryonic peripheral nervous system.

hESC-Derived NC-like Cells Can Give Rise to Enteric Neurons in Explants of Embryonic Mouse Gut

The potential of hESC-derived NC-like cells to colonize explants of embryonic mouse gut and differentiate into enteric neurons was assessed. Cocultures were established between segments of hindgut from E11.5 mice and hESC-derived neurospheres that had been cultured on mEFs for 24 hours and then exposed to Y27632 for an additional 6, 24, or 48 hours. In E11.5 mice, NC cells had not yet colonized the hindgut, and thus the recipient guts were aneural [38]. In cocultures between aneural gut and hESC-derived cells exposed to mEFs only, or to mEFs and then to Y27632, GFP+ cells were present in approximately 80% of recipient guts (Fig. 5A, 5D). Exposure to Y27632 did not change the distance migrated by GFP+ hESC-derived cells along recipient gut explants (Fig. 5E). However, in only 5% of cocultures between hESC-derived cells exposed to mEFs only, or to mEFs followed by Y27632 for 6 hours, did any of the GFP+ cells within the recipient gut express SoxE (Fig. 5G) or the pan-neuronal marker Hu (Fig. 5F). In contrast, after exposure to mEFs and then Y27632 for 24 or 48 hours prior to coculturing, GFP+ hESC-derived cells in approximately 50% of the recipient guts expressed SoxE and/or Hu (Fig. 5B–5B″′, 5F, 5G). GFP+ cells that did not express SoxE or Hu were also observed in the recipient guts. In the enteric nervous system, there are different types of enteric neurons, and approximately 30% express neuronal nitric oxide synthase (nNOS) [49]. A subpopulation of hESC-derived Hu+ cells in the explants showed nNOS immunostaining (Fig. 5C, 5C′); thus, hESC-derived cells can also give rise to a major enteric neuron subtype. The results of these gut explant cultures are consistent with the in vitro data described above (Fig. 1), showing that NC-like induction from human neural progenitors is significantly improved when initially primed on mEF substrate followed by 24 hours or longer treatment with Y27632.

Figure 5.

Transplantation of hESC-derived progenitors into mouse gut tissue. HES-derived neural crest (NC) cells grown on mEFs-colonized explants of aneural embryonic mouse gut (A). Exposure to Y27632 had no effect on the percentage of recipient guts colonized (D) or the distance migrated by hES-derived NC cells (E). However, exposure to Y27632 for 24 or 48 hours significantly increased the proportion of explants with Hu+ (BB″′, F) and SoxE+ (BB″′, G) cells. A subpopulation of the Hu+ cells (asterisks) expressed the enteric neuron subtype marker, NOS (C,C), were analyzed using a χ2 test (D, F, G) or a one-way ANOVA with Tukey's multiple comparison test (E). Scale bars: (A) = 100 μm; (B) = 50 μm; (C) = 10 μm. Abbreviations: *, p = .05; **, p = .01; ***, p = .001; mEFs, mouse embryonic fibroblasts; N, number of experiments; n, number of preparations; NS, neurosphere.

DISCUSSION

Previous studies have shown that hESC can give rise to neurospheres resembling neuroepithelium, and hence to some NC-like cells and NC derivatives spontaneously, and this is improved by exposure to stromal cell lines [48, 50, 51]. Our study shows that mEFs also show this NC promoting capacity to a degree equal to that of stromal cells. The mEF-induced NC-like cells occurred in limited numbers compared to neuroepithelial-like cells, but they express NC markers including p75, SoxE, and HNK1, and could migrate and differentiate in vitro, in embryonic mouse gut and in avian embryos. The combination of these characteristics are important for identifying these cells as being NC-like cells since hESC-derived neurospheres consist of a heterogeneous population of neural progenitors and expression of one NC marker, such as p75, is not sufficient for unequivocal identification. Significantly, our data demonstrates that subsequent exposure to Y27632, a small-molecule inhibitor of the Rho effectors ROCKI/II, dramatically increases the number of migrating hESC-derived NC-like cells in vitro. Inhibition of Rho/ROCK signaling has been implicated in the EMT of NC cells from avian neural epithelium [46].

We found that hESC-derived neurospheres exposed to mEFs and then transplanted into the NC migratory pathways of avian embryos resulted mostly in coherent neural tube-like structures. In contrast, exposure to mEFs + Y27632 resulted in extensive migration of NC-like (SoxE+ve) cells in association with host NC cells. The hESC-derived cells also assembled at sites appropriate for NC cells and showed neuronal (Hu+, axons) and glial (S100b++) differentiation. Cells from hESC-derived neurospheres exposed to mEFs migrated into explants of embryonic gut, but further Y27632 exposure was required to elicit expression of the NC marker SoxE, the generic neuronal marker Hu, or the enteric neuron subtype marker, nNOS. Although we showed hESC-derived cells in NC-appropriate sites and development of markers consistent with NC differentiation, the generic markers mean that further studies will be required to ascertain whether hESC-derived neurons have the same functional properties as dorsal root ganglion and enteric neurons. Overall, our data show that exposure of hESC-derived neural progenitors to both mEFs and Y27632 is efficient in inducing NC-like cells capable of context-appropriate migration, assembly, and differentiation.

The efficient induction of migrating NC-like cells from hESC-derived neural progenitors by Y27632 treatment was effective only when Y27632 was added 24 hours after exposure to mEFs. Also, Y27632 treatment of hESC-derived neural progenitors cultured on fibronectin substrate did not promote the expression of NC markers. This suggests that the signaling pathways manipulated by Y27632 treatment are not sufficient to induce the NC phenotype in otherwise uncommitted neural progenitors. Thus, a priming event is required to initially specify hESC-derived neural progenitors to differentiate toward NC, and subsequent exposure to Y27632 efficiently promotes NC EMT and migration. This matches with the normal development of the NC from neural epithelium via molecularly complex instructions to specify the NC lineage followed by triggering of the morphogenetic step of EMT [1–4]. The mEF feeder layer appears to have the appropriate signals to bias the differentiation of hESC-derived neural progenitors toward NC. This suggests that signals expressed by feeder layers that serve to maintain embryonic stem cell growth may also play instructive roles in differentiating early progenitors. Candidate factors that are known to be secreted by mEF and may contribute toward NC specification are members of the Wnt and FGF family [42, 43]. Blocking canonical Wnt, BMP, and/or FGF2 signaling did not significantly decrease the neural crest–inducing signals mediated by mEF cocultures. Further studies will be required to decipher what the NC-inducing signals are, but it is likely that NC induction requires several factors, with sequence and concentration being of importance, to optimally induce NC specification.

For the purpose of therapies, the advantage of using tissue-derived NC cells is that they can be generated autologously and thus not be immunorejected once transplanted back into the patient. Some studies state that NC-like cells isolated from adult non-neural tissues may be restricted in their differentiation potential and therefore be incapable of some lines of differentiation [52]. However, other reports have suggested differently [53]. Conversely, NC-like cells isolated from embryonic stem cells are not patient-specific and hence are subject to rejection, but are more likely to retain their full differentiation and functional potential. This caveat of patient specificity may now be addressed with the recent advancements of induced pluripotent stem (iPS) cell technology [54]. It is becoming more feasible that patient-specific iPS cell lines may be generated and perhaps therapeutically used. It is also very likely that the technologies and methods developed with hESC may be transferred to iPS cells and their neural derivatives, such as the NC-induction system described in this study. Our data, together with the advancements in stem cell technologies, show great promise of the potential therapeutic use of NC cells.

CONCLUSION

We describe a simple method for efficiently generating NC-like cells from hESCs. Like CNS-derived stem cells [24] and NC-derived stem/progenitor cells derived from the gut [20, 25, 26], hESC-derived cells are a potential source of NC-like cells to generate enteric neurons or other NC derivatives in diseases such as Hirschsprung's disease. At the same time, they are a resource for understanding the molecular and cell biological requirements for any type of human NC stem cell therapy. The data in this study provide a foundation for the production of an unlimited number of enteric neurons and other NC derivatives from patient-derived iPS cells.

Acknowledgements

We thank Craig Smith, Murdoch Children's Research Institute, Melbourne, for SoxE antibody, and Miles Epstein, University of Wisconsin, for the human anti-Hu. This study was supported by The University of Melbourne, The National Health and Medical Research Council of Australia, and Friedreich Ataxia Research Association Australia.

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

The authors indicate no potential conflicts of interest.

Ancillary