|Home | About | Journals | Submit | Contact Us | Français|
We have systematically examined the developmental potential of neural crest stem cells from the enteric nervous system (gut NCSCs) in vivo to evaluate their potential use in cellular therapy for Hirschsprung disease and to assess differences in the properties of postmigratory NCSCs from different regions of the developing peripheral nervous system (PNS). When transplanted into developing chicks, flow-cytometrically purified gut NCSCs and sciatic nerve NCSCs exhibited intrinsic differences in migratory potential and neurogenic capacity throughout the developing PNS. Most strikingly, gut NCSCs migrated into the developing gut and formed enteric neurons, while sciatic nerve NCSCs failed to migrate into the gut or to make enteric neurons, even when transplanted into the gut wall. Enteric potential is therefore not a general property of NCSCs. Gut NCSCs also formed cholinergic neurons in parasympathetic ganglia, but rarely formed noradrenergic sympathetic neurons or sensory neurons. Supporting the potential for autologous transplants in Hirschsprung disease, we observed that Endothelin receptor B (Ednrb)-deficient gut NCSCs engrafted and formed neurons as efficiently in the Ednrb-deficient hindgut as did wild-type NCSCs. These results demonstrate intrinsic differences in the migratory properties and developmental potentials of regionally distinct NCSCs, indicating that it is critical to match the physiological properties of neural stem cells to the goals of proposed cell therapies.
The characterization of gut NCSCs that give rise to the enteric nervous system during fetal development (Bixby et al., 2002) and then persist in the wall of the gut throughout adult life (Kruger et al., 2002; Bondurand et al., 2003; Molofsky et al., 2003; Molofsky et al., 2005) raises several questions about the developmental potential of these cells. Postmigratory fetal gut NCSCs retain the capacity to migrate through trunk neural crest migration pathways after transplantation and to form neurons and glia in diverse regions of the developing PNS, including in the enteric nervous system (Bixby et al., 2002; Bondurand et al., 2003; Kruger et al., 2003). However, it remains unclear whether these neurons acquire regionally appropriate fates. Neural crest progenitors transplanted heterotopically or heterochronically sometimes form inappropriate types of neurons in their new locations (Schweizer et al., 1983; Le Douarin, 1986; White and Anderson, 1999; White et al., 2001). Central nervous system progenitors also can form inappropriate neurons after heterotopic transplantation (Campbell et al., 1995; Olsson et al., 1997; Na et al., 1998; Takahashi et al., 1998; Desai and McConnell, 2000). These observations raise the question of whether gut NCSCs retain the potential to form appropriate subtypes of neurons throughout the developing PNS.
A second important issue relates to the extent to which gut NCSCs differ from NCSCs isolated from other regions of the developing PNS. Some have argued that neural stem cells might retain the potential to make appropriate types of neurons and glia throughout the nervous system, irrespective of their embryonic origin (Weiss et al., 1996; Panchision et al., 1998; Hitoshi et al., 2002), while others have noted that the patterning differences among regions of the nervous system make that implausible (Anderson, 2001). We have observed cell-intrinsic differences among gut NCSCs and sciatic nerve NCSCs in their responses to the factors that regulate lineage determination in developing peripheral nerve (Bixby et al., 2002). However, a thorough examination of the developmental potential of gut NCSCs in vivo is required to determine whether these cells exhibit global differences relative to other postmigratory NCSCs.
The systematic examination of the developmental potential of gut NCSCs in vivo has gained added importance with the suggestion that these cells might be used to treat Hirschsprung disease (Bondurand et al., 2003; Iwashita et al., 2003; Kruger et al., 2003). Hirschsprung disease, or aganglionic megacolon, is a congenital defect that affects 1 out of 5,000 live births and is characterized by the failure to form enteric nervous system in a variable length of the hindgut (Newgreen and Young, 2002). This potentially fatal condition results in an inability to coordinate peristaltic movements of the bowel and is most commonly caused by mutations that reduce signaling through the Glial Derived Neurotrophic Factor (GDNF) or Endothelin3 (EDN3) signaling pathways (Gariepy, 2001). Both the GDNF receptor Ret and the EDN3 receptor Endothelin receptor B (EDNRB) are expressed by gut NCSCs (Iwashita et al., 2003).
Rodents that are deficient for GDNF or Ret (Schuchardt et al., 1994; Moore et al., 1996; Pichel et al., 1996; Sanchez et al., 1996) or Edn3 or Ednrb (Greenstein Baynash et al., 1994; Hosoda et al., 1994; Gariepy et al., 1996) also exhibit aganglionic megacolon. Signaling through these pathways expands the pool of NCSCs and other neural crest progenitors that initially colonizes the foregut (Barlow et al., 2003; Iwashita et al., 2003; Kruger et al., 2003; Bondurand et al., 2006). These pathways subsequently interact to regulate the caudal migration of NCSCs and other progenitors throughout the length of the gut (Schuchardt et al., 1994; Shin et al., 1999; Young et al., 2001; Barlow et al., 2003; Iwashita et al., 2003; Kruger et al., 2003; Lee et al., 2003). NCSCs never migrate into the aganglionic portion of the gut in animals affected by Ret or Ednrb deficiency, despite the presence of NCSCs in regions of the foregut and midgut that are colonized by neural crest (Iwashita et al., 2003; Kruger et al., 2003). These observations raise the possibility of improving the treatment of Hirschsprung disease by transplanting NCSCs into the aganglionic portion of the gut to generate enteric ganglia by bypassing the migration defects (Natarajan et al., 1999; Bondurand et al., 2003; Iwashita et al., 2003; Kruger et al., 2003).
The prospect of a cellular therapy that could generate enteric ganglia is compelling because while surgery restores considerable gut function, many patients experience ongoing problems (Tsuji et al., 1999) including those due to hypoganglionic segments of gut that remain after surgery. Wild-type NCSCs are able to engraft and form neurons in the aganglionic region of the Ednrb-deficient gut just as efficiently as in wild-type gut, demonstrating that the aganglionic gut environment is permissive for the survival and differentiation of wild-type NCSCs (Kruger et al., 2003). This is also true in the Ret-deficient gut (Natarajan et al., 1999; Bondurand et al., 2003). Nonetheless, it remains to be determined whether the aganglionic environment is also permissive for the engraftment of Ednrb-deficient or Ret-deficient NCSCs (Jacobs-Cohen et al., 1987; Rothman et al., 1996). If the aganglionic gut is not permissive for the engraftment of mutant NCSCs then it would not be possible to autologously transplant NCSCs from the normoganglionic foregut into the aganglionic hindgut of the same patient.
To address these issues we injected prospectively isolated fetal gut NCSCs and sciatic nerve NCSCs into the embryonic chick neural crest migration pathway or directly into the developing chick or rat gut. The transplanted cells were flow-cytometrically isolated from transgenic rats that constitutively expressed human placental alkaline phosphatase (hPAP), allowing us to follow their migration and differentiation regardless of what fates they acquired. Gut and sciatic nerve NCSCs migrated widely throughout the PNS, but these cells differed in migratory properties and fate determination in a number of regions of the developing PNS, particularly in the enteric nervous system. While gut NCSCs formed appropriate types of neurons in parasympathetic and enteric ganglia, they failed to generate noradrenergic neurons in sympathetic ganglia despite engrafting in this location and readily forming noradrenergic neurons in culture. This emphasizes the importance of evaluating the developmental potential of neural stem cells under physiological conditions in vivo as their ability to form progeny in vivo cannot necessarily be predicted based on in vitro experiments. To test whether gut NCSCs might be used in autologous cell therapies for Hirschsprung disease, Ednrb-deficient and wild-type gut NCSCs were expanded in culture and transplanted into the aganglionic guts of Ednrb-deficient recipients. Ednrb-deficient and wild-type cells engrafted and formed enteric neurons with similar efficiency. This demonstrates that the aganglionic Endrb-deficient hindgut is permissive for the engraftment and differentiation of Ednrb-deficient neural crest progenitors.
Embryonic day 14.5 gut or sciatic nerve NCSCs were obtained from timed pregnant Sprague-Dawley rats (Charles River, Wilmington, MA), Fisher 344 rats expressing the human placental alkaline phosphatase (hPAP) transgene under the control of the ROSA26 promoter (Kisseberth et al., 1999) or Ednrbsl/slhPAP+ rats derived from a cross between Ednrb+/sl (Wistar-Kyoto background) and hPAP+ rats. Fetal guts or nerves were dissected into ice-cold Ca2+-, Mg2+-free HBSS and dissociated as previously described (Morrison et al., 1999; Bixby et al., 2002). Single cell suspensions of E14.5 gut or nerve were stained with antibodies against the neurotrophin receptor p75 (clone 192Ig , EMD Biosciences, San Diego, CA) and α 4 integrin (clone MRα 4–1, Becton Dickinson, San Jose, CA) and resuspended in staining medium (L15 medium containing 1mg/mL BSA (Sigma product A-3912, St. Louis MO), 10 mM HEPES (pH 7.4), penicillin/streptomycin (BioWhittaker, Walkersville, MD), and 25μg/mL deoxyribonuclease type 1 (Sigma, product D-4527, St. Louis, MO)) plus 7AAD (Molecular Probes, Eugene, OR). p75+α 4+ cells were then isolated by flow cytometry on a FACSVantage dual-laser flow cytometer (Becton-Dickinson) as previously reported (Bixby et al., 2002). Cells for transplantation were sorted into a 0.2 ml PCR tube containing staining medium with 15% chick embryo extract and cells for culture were sorted into plates containing standard medium.
Tissue culture plate preparation, culture medium and culture conditions were as described (Bixby et al., 2002). Cells were typically cultured in 6-well plates (Corning, Corning New York) at clonal density so that individual colonies were spatially distinct (fewer than 30 clones per well for 14 day cultures) as previously described (Morrison et al., 1999; Morrison et al., 2000b). Plates were sequentially coated with 150μg/ml poly-d-lysine (Biomedical Technologies, Stoughton MA) and 0.15 mg/ml human fibronectin (Biomedical Technologies) as described (Stemple and Anderson, 1992). The culture medium contained DMEM-low (Gibco, Carlsbad, CA, product 11885–084) with 15% chick embryo extract (prepared as described (Stemple and Anderson, 1992)), 20 ng/ml recombinant human bFGF (R&D Systems, Minneapolis, MN), 1% N2 supplement (Gibco), 2% B27 supplement (Gibco), 50 μM 2-mercaptoethanol, 35ng/ml (110 nM) retinoic acid (Sigma), penicillin/streptomycin (Biowhittaker), and 20 ng/ml IGF1 (R&D Systems). This medium composition is described as ‘standard medium’. Under standard conditions, cells were cultured for 6 days in this medium, then switched to a similar medium (with 1% CEE and 10ng/ml bFGF) that favors differentiation for another 8 days before immunohistochemical analysis of colony composition. To assay for sensory potential, NCSCs were cultured in standard medium supplemented with NGF (50ng/ml). Wnt3a (R&D Systems, 50ng/ml) was also added to some cultures in an attempt to stimulate sensory neurogenesis. All cultures were maintained in gas-tight chambers (Billups-Rothenberg, Del Mar, CA) containing decreased oxygen levels to enhance the survival of NCSCs (Morrison et al., 2000a).
Ednrb+/+ or Ednrbsl/sl cells were cultured for 6 – 12 days in standard medium for expansion prior to injection. Cells to be injected were trypsinized (0.05% Trypsin-EDTA, Gibco, product 25300-054) for 5 min at 37 ° C then quenched in staining medium plus DNase (5 mg/ml), before being washed and resuspended in staining medium.
For transplantation of freshly sorted NCSCs into embryonic chicks, fertile White Leghorn eggs (Bilbie Aviaries, Ann Arbor, MI) were incubated to Hamburger and Hamilton stage 17–18 (Hamburger and Hamilton, 1992) and injected with freshly isolated sciatic nerve or gut NCSCs as previously described (Bixby et al., 2002). Injected embryos were incubated for an additional 72 hr (stage 29) and then fixed for in situ hybridization. To test the potential of NCSCs to engraft and differentiate in the enteric nervous system, guts, from stomach to rectum, were dissected from 5 day old embryonic chicks (Hamburger and Hamilton stage 27) in ice-cold Ca2+ and Mg2+-free HBSS and temporarily (2–4 hr) placed in Opti-MEM plus 10% chick embryo extract and penicillin/streptomycin at 37 ° C in 6% CO2 until injection. NCSCs were injected into the gut wall at the level of the cecum, then the guts were explanted onto the chorioallantoic membrane of 7 to 10 day-old chicks as previously described (Kruger et al., 2003). Injected guts were incubated for an additional 4 to 5 days and then fixed for immunohistochemistry. To determine if cultured Ednrbsl/sl or Ednrb+/+ gut neural crest progenitors could engraft and survive in the aganglionic Ednrbsl/sl colon, cultured E14.5 gut NCSCs were injected into the gut wall of the distal colon from E14.5 rats and incubated on the chorioallantoic membrane of 7 to 10 day old chicks as described (Kruger et al., 2003). Injected guts were incubated for an additional 4 to 5 days and then fixed for immunohistochemistry.
Injected chick embryos were fixed in 4% paraformaldehyde, washed in 0.1M phosphate buffer, cryoprotected in 15% sucrose, embedded in OCT (Tissue-Tek) and 12 μm sections were cut by cryostat (Leica). In situ hybridization methods were previously described (White and Anderson, 1999). Sections were initially hybridized with probes against the pan-neuronal marker stathmin-like 2 (Stmn2, formerly known as Scg10) (Anderson and Axel, 1985) or the glial marker P0 (Lemke et al., 1988) to evaluate the phenotype of engrafted rat cells. To more fully evaluate the identities acquired by the engrafting neurons, neuronal subtypes were assessed using rat-specific riboprobes against sensory, sympathetic or parasympathetic markers. Serial sections flanked by Stmn2+ rat neurons were hybridized with probes against rat-specific VAChT, TH, Ret, or Gata2 to distinguish sympathetic noradrenergic (TH+, Gata2+, VAChT−) from parasympathetic cholinergic (TH−, Gata2+, VAChT+) phenotypes. Additional sections adjacent to Stmn2+ rat neurons in sensory ganglia were hybridized with a probe against the sensory marker Pou4f1 (formerly Brn3a). VAChT, TH and Gata2 probes were amplified from E14.5 rat cDNA using the PCR primers: Gata2 forward = GGACAAGGATGGCGTCAAGTATC; Gata2 reverse = TGCCGATTCTTTTCTTAGGCG; TH forward =TGCCTCTCGTATCCAGCGCC; TH reverse = TTCTTGAAGGAGCGGACTGGCTTC; VAChT forward=ATCCATCGCCTCATGCTAGACC; VAChT reverse = CGGAGTGCAGAGACCAAGTTTG (Schafer et al., 1994). The Ret and Pou4f1 probes were previously described (White et al., 2001). Sections from E14.5 wild-type rat embryos were also hybridized as positive controls to ensure the specificity and sensitivity of the probes. Since injections into the chick embryo were unilateral, the uninjected side of the embryo served as a negative control.
Injected guts were fixed overnight in 4% paraformaldehyde at 4 ° C, washed in 0.1M phosphate buffer, cryoprotected overnight in 15% sucrose, embedded in OCT (Tissue-Tek) and 12–14 μm sections were cut using a cryostat (Leica). Sections were processed either chromogenically with the alkaline phosphatase substrate NBT/BCIP (Roche, products 1 087 479 and 1 383 221) or immunohistochemically with antibodies against hPAP (Sigma, A-2951), the neuronal marker β-III tubulin (Promega, G712A, Madison, WI), the glial marker GFAP (Chemicon, AB5804, Temecula, CA) or a neuronal subtype marker; anti-neuropeptide Y (NPY; Peninsula Laboratories, IHC 7180, San Carlos, CA), anti-vasoactive intestinal peptide (VIP; Peninsula Laboratories, IHC 7161), anti-neuronal nitric oxide synthase (nNOS; Chemicon, AB5380), and anti-calbindin (CalB; Chemicon, AB1778). Some engrafted sections were also stained with an antibody against activated caspase-3 (BD Pharmingen, 559565) or microphthalmia (MiTF, MS-771-P, clone C5, NeoMarkers, Fremont, CA). Cultured cells stained for neuronal subtype antibodies were fixed in fresh 4% paraformaldehyde for 20 min at 4 ° C, washed and blocked as described (Shah et al., 1994). Cultures were stained with antibodies against, GFAP, smooth muscle actin (A-2547) and one of the above neuronal subtype antibodies.
The ability of NCSCs to form melanocytes was tested in a hair morphogenesis assay (Lichti et al., 1993; Hutchin et al., 2005; Zheng et al., 2005). This assay recapitulates follicular morphogenesis by combining keratinocytes, melanocyte progenitors, and dermal cells under the skin of immunocompromised adult NOD/scid mice. Primary dermal cells and keratinocyte preparations, which contain melanoblasts, were isolated from the skins of newborn CD-1 (albino) or C57BL (pigmented) mice following established protocols (Dlugosz et al., 1995; Hutchin et al., 2005) and grown in culture for 3 days prior to their use in the assay. Gut and sciatic nerve NCSCs were freshly isolated from 2 litters of E14.5 Norway brown rats (Charles River) as previously reported (Bixby et al., 2002). Approximately 7 x 106 CD-1 keratinocytes and an equal number of CD-1 fibroblasts were injected alone or in combination with 5 x 104 keratinocytes from C57BL mice (a positive control that contains progenitors which give rise to pigmented melanocytes), or 4.5 x 104 sciatic nerve NCSCs or 2 x 104 gut NCSCs. Three weeks after injection, the mice were sacrificed and tissue from the injection sites was fixed in 4% formalin, embedded in paraffin, sectioned, and stained with hematoxylin and eosin. Sections were analyzed for melanin in follicles or elsewhere in the grafted patches of cells.
To systematically examine the potential of regionally distinct postmigratory NCSCs to engraft, migrate, and differentiate throughout the PNS we compared the in ovo engraftment of flow-cytometrically isolated p75+α 4+ NCSCs obtained from the E14.5 rat gut and sciatic nerve. These p75+α 4+ cells accounted for only 1.9±0.5% of cells in the gut and 12.0±6.7% of cells in sciatic nerve respectively, and were highly enriched for NCSC activity. Clonal analysis of these cells indicated that 74±11% of the colonies formed by p75+α 4+ gut cells (60±12% of cells survived to form colonies) and 80±9% of the colonies formed by p75+α 4+ sciatic nerve cells (35±15% of cells survived to form colonies) were large multilineage colonies containing neurons, glia, and myofibroblasts, characteristic of NCSCs (Stemple and Anderson, 1992; Shah et al., 1996; Morrison et al., 1999; Bixby et al., 2002; Kruger et al., 2002).
Gut and sciatic nerve p75+α 4+ NCSCs were isolated from transgenic rats that expressed hPAP under the control of the ubiquitously expressed ROSA26 promoter (Kisseberth et al., 1999). The NCSCs were injected into 1–2 hindlimb bud somites of Hamburger and Hamilton stage 17–18 chick embryos (Morrison et al., 1999; White and Anderson, 1999; White et al., 2001; Bixby et al., 2002). Three days after injection, embryos were fixed, sectioned, and reacted chromogenically to identify hPAP expressing rat cells. hPAP+ cells were found throughout sacral neural crest derivatives in all 5 of the chicks injected with gut NCSCs (Table 1). Transplanted hPAP+ gut NCSCs engrafted in the DRG (3 of 5 chicks; 2.3 ± 1.2 hPAP+ sections/engrafted chick), sympathetic chain (5 of 5 chicks; 6.2 ± 3.6 hPAP+ sections/engrafted chick), peripheral nerve (5 of 5 chicks; 12.4 ± 3.1 hPAP+ sections/engrafted chick), Remak’s ganglion (4 of 5 chicks; 17.5 ± 10.7 hPAP+ sections/engrafted chick), pelvic plexus (1 of 5 chicks; 10 hPAP+ sections/engrafted chick), gut (1 of 5 chicks; 4 hPAP+ sections/engrafted chick), ventral root (2 of 5 chicks; 4.0 ± 0.0 hPAP+ sections/engrafted chick), and dermis (2 of 5 chicks; 4.5 ± 3.5 hPAP+ sections/engrafted chick). Note that we did not try to count the number of hPAP+ cells per engrafted section when hPAP activity was detected chromogenically, because it was difficult to distinguish individual hPAP+ cells or to accurately count them.
Sciatic nerve NCSCs also migrated widely and engrafted throughout the PNS in all 11 chicks injected (Table 1). hPAP+ sciatic nerve NCSCs engrafted in the DRG (7 of 11 chicks; 4.6 ± 2.5 hPAP+ sections/engrafted chick), sympathetic chain (11 of 11 chicks; 7.1 ± 4.0 hPAP+ sections/engrafted chick), peripheral nerve (11 of 11 chicks; 15.3 ± 4.5 hPAP+ sections/engrafted chick), Remak’s ganglion (10 of 11 chicks; 19.2 ± 9.3 hPAP+ sections/engrafted chick), pelvic plexus (1 of 11 chicks; 1.0 hPAP+ section/engrafted chick), ventral root (2 of 11 chicks; 4.0 ± 4.3 hPAP+ sections/engrafted chick) and dermis (11 of 11 chicks; 10.2 ± 8.6 hPAP+ sections/engrafted chick). Sciatic nerve NCSCs did not engraft in the gut (0/11 chicks).
In addition to engrafting in known neural crest-derivatives, some transplanted cells also engrafted in ectopic locations. hPAP+ cells derived from gut NCSCs were found in the ventral column of the neural tube (3 of 5 chicks; 3.0 ± 3.5 hPAP+ sections/engrafted chick; Table 1) and in the dorsal mesenchyme between the notochord and Remak’s ganglion (2 of 4 chicks; not shown). Additionally, hPAP+ cells were found in the developing urogenital region (e.g. the mesonephros, 3 of 5 chicks; not shown), consistent with other groups that have found neural crest cells in the mesonephros after the transplantation of vagal neural crest cells into the thoracic or sacral neural tube (Le Douarin and Teillet, 1974; Burns et al., 2002). We did not detect endogenous chick neurons in any of these locations using a chick-specific neuronal marker, Stmn2. hPAP+ cells derived from sciatic nerve NCSCs were also found in the urogenital region (7 of 11 chicks; not shown) and the dorsal mesenchyme (9 of 11 chicks; not shown) but no sciatic nerve-derived cells migrated into the ventral neural tube. Since NCSCs were transplanted both heterotopically and heterochronically, their ectopic migration might reflect responses to signals that these cells would normally not encounter during their physiological migration.
The above data suggest that differences exist in the migratory potential of gut and sciatic nerve NCSCs. Sciatic nerve NCSCs engrafted more efficiently than gut NCSCs in some locations (such as dermis) while engrafting less efficiently in other locations (such as the gut). To confirm these migratory differences and to determine the fate of the engrafted cells a more extensive analysis of engraftment was performed.
The experiments with hPAP-expressing NCSCs demonstrated that gut and sciatic nerve NCSCs can migrate widely throughout the PNS but the fate of these cells was not determined. To examine the fate of these cells and to test whether there is a difference among NCSCs in their ability to migrate into the gut, stage 17–18 chicks were injected with p75+α 4+ gut or sciatic nerve NCSCs and allowed to develop for 3 days before being fixed and sectioned. The progeny of NCSCs were then identified by in-situ hybridization using rat-specific probes against Stmn2 to identify neurons (Anderson and Axel, 1985) and P0 to identify glia (Lemke et al., 1988). Gut and sciatic nerve NCSCs gave rise to neurons and glia in multiple locations (Table 1). Gut NCSCs gave rise to neurons and glia in the DRG (5 of 37 chicks were positive for Stmn2+ neurons and 1 of 37 chicks had P0+ glia), sympathetic chain (32 of 37 had neurons and 3 of 37 had glia), peripheral nerve (34 of 37 had neurons and 16 of 37 had glia), Remak’s ganglion (35 of 37 had neurons and 2 of 37 had glia), pelvic plexus (20 of 37 had neurons and 7 of 37 had glia), gut (20 of 37 had neurons and 9 of 37 had glia) and ventral root (1 of 37 had neurons and 14 of 37 had glia). In all but one of these locations (the ventral root), the number of chicks with engrafted neurons was greater than the number chicks with engrafted glia. Additionally, there were more sections that contained neurons than sections that contained glia and more neurons per section than glia per section in most of these locations.
Nonetheless, there were some locations in which gut NCSCs gave rise primarily to glia. In the ventral nerve root gut NCSCs formed predominantly glia (14 of 37 chicks had P0+ glia, with 4.0 ± 3.0 glia/positive section) while forming only a single neuron in the ventral root of a single chick (Table 1). In the ventral column of the neural tube, a region not normally populated by neural crest cells, gut NCSCs gave rise exclusively to glia (30 out of 37 chicks; 4.1 ± 2.3 P0+ cells/positive section). Thus, gut NCSCs were capable of efficiently and exclusively forming glia in some environments. Since E14.5 gut NCSCs do not exhibit a gliogenic response to known factors in culture (Bixby et al., 2002), these results suggest that they are either responding to yet unidentified gliogenic factors in these locations, or there are powerful selective factors in these environments that promote the survival of rare glia at the expense of other cells that arise from gut NCSCs. This demonstrates that gut NCSCs are capable of generating different types of cells in different locations, despite their tendency to undergo neurogenesis in most locations.
Sciatic nerve NCSCs also gave rise to neurons and glia in vivo but in contrast to gut NCSCs they did so in fewer locations and they preferentially formed glia. When sciatic nerve NCSCs engrafted in the peripheral nerve, 11 of 11 injected chicks had P0+ glia while only 4 of 11 had Stmn2+ neurons (Table 1) and significantly more sections contained glia than neurons. These results are consistent with our previous finding that sciatic nerve NCSCs are primarily gliogenic in developing peripheral nerves while gut NCSCs are primarily neurogenic (Bixby et al., 2002). However, sciatic nerve NCSCs were not always gliogenic. As observed previously (White et al., 2001), sciatic nerve NCSCs were exclusively neurogenic in Remak’s ganglion (7 of 11 chicks had Stmn2+ neurons with an average of 1.6±0.7 neurons per engrafted section) and the pelvic plexus (1 of 11 chicks had Stmn2+ neurons). Nerve NCSCs thus acquire different fates in different locations but are primarily gliogenic in some locations where gut NCSCS are primarily neurogenic. This further emphasizes the intrinsic differences in lineage determination between these spatially distinct NCSCs (Bixby et al., 2002), but extends this observation by demonstrating that gut NCSCs are also more neurogenic throughout much of the developing PNS. In both peripheral nerve and sympathetic chain, gut NCSCs formed neurons in significantly (p<0.001) more sections per engrafted chick, and significantly (p<0.001) more neurons per engrafted section as compared to sciatic nerve NCSCs (Table 1).
Sciatic nerve NCSCs never formed neurons or glia in the guts of injected chicks (0/11 injected with hPAP+ cells and 0/11 tested for neuronal and glial differentiation for a total of 0/22 chicks; Table 1). In contrast, gut NCSCs migrated into the guts of most transplanted chicks and formed neurons (20/37) or glia (9/37) or hPAP+ cells (1/5) (Table 1). Overall 22 of 42 chicks had gut NCSC-derived cells that migrated into the gut. These results demonstrate an intrinsic difference among gut and sciatic nerve NCSCs in terms of their ability to migrate into the gut.
To further compare the ability of sciatic nerve and gut NCSCs to migrate into the gut, we examined transplanted chicks 40 hours (HH stage 28) after injection of hPAP+ sciatic nerve or gut NCSCs. If sciatic nerve NCSCs were migrating into the gut but undergoing cell death prior to the time point at which chicks were analyzed in the above experiments, then we expected to detect sciatic nerve-derived cells in the gut at earlier time points after transplantation. Both of the chicks transplanted with gut NCSCs and all 3 of the chicks transplanted with sciatic nerve NCSCs had hPAP+ cells in Remak’s ganglion, immediately dorsal to the gut. Nonetheless, while progeny of gut NCSCs were detected in the guts of 1 out of 2 transplanted chicks none of 3 chicks transplanted with sciatic nerve NCSCs had hPAP+ cells in the gut (data not shown). To test whether the progeny of sciatic nerve NCSCs might simply undergo cell death after migrating into the gut we stained for activated caspase-3. No activated caspase-3+ cells were found in the guts or in Remak’s ganglion, though activated caspase-3+ cells were observed in other locations. These data demonstrate that sciatic nerve NCSCs efficiently migrate into Remak’s ganglion after sacral transplantation but unlike gut NCSCs are unable to migrate further into the gut.
Gut NCSCs give rise to neurons in parasympathetic ganglia ((Bixby et al., 2002); Table 1). However, we wondered whether these neurons acquired an appropriate cholinergic fate. To test this in vivo we examined the expression of neuronal subtype markers in the progeny of transplanted cells that had migrated into parasympathetic ganglia. Every sixth section cut from injected chicks was probed with the rat-specific pan-neuronal marker Stmn2 to identify sets of consecutive sections that were likely to contain rat neurons. Upon identifying sets of 6 consecutive sections in which the first and sixth sections contained rat neurons in the same ganglion, we hybridized the intervening four sections with rat-specific probes characteristic of cholinergic or noradrenergic neurons: tyrosine hydroxylase (TH), vesicular acetylcholine transferase (VAChT), and GATA binding protein 2 (Gata2). These markers distinguish cholinergic neurons normally found in parasympathetic ganglia (TH−, Gata2+, VAChT+) from noradrenergic neurons normally found in sympathetic ganglia (TH+, Gata2+, VAChT−), and cholinergic neurons normally found in the enteric nervous system (TH−, Gata2−, VAChT+) (Groves et al., 1995; White and Anderson, 1999; White et al., 2001). We also examined the general autonomic neuronal marker Ret. The fraction of cells that expressed each marker was estimated as a percentage of the number of Stmn2+ neurons in flanking sections.
Most of the progeny of gut NCSCs that differentiated in the parasympathetic pelvic plexus expressed markers consistent with a developmentally appropriate cholinergic fate. In the 9 chicks analyzed for neuronal subtype markers, most of the clusters of engrafted cells contained cells that expressed Gata2 (8 of 9 chicks had Gata2+ cells; the Gata2+ cells were equivalent to 64 ±47% of the number of neurons in adjacent sections), VAChT (8 of 9 chicks; 52±33%), and/or Ret (8 of 9 chicks; 69±49%) (Fig. 1B–D). In Remak’s ganglion, another parasympathetic ganglion, engrafted cells also acquired a cholinergic phenotype. In the 15 engrafted chicks examined for cholinergic differentiation in Remak’s ganglion, VAChT (12 of 15 chicks; 30±37%), Gata2 (10 of 15 chicks; 18±18) and/or Ret (14 of 15 chicks; 60±37) positive cells were observed (Fig. 1E), though the percentage of cells that acquired a cholinergic phenotype appeared lower than in the pelvic plexus. TH-positive cells were never observed in either parasympathetic ganglion. The expression of Gata2 in these parasympathetic ganglia is important given that Gata2 is not normally expressed in the gut (Groves et al., 1995; White and Anderson, 1999). Gata2 expression thus demonstrates that gut NCSCs are capable of efficiently adopting a parasympathetic cholinergic phenotype rather than simply making enteric neurons in parasympathetic ganglia. Sciatic nerve NCSCs were previously observed to form cholinergic neurons in parasympathetic ganglia (White et al., 2001).
In culture, virtually every E14.5 gut NCSC is capable of forming large numbers of TH+DβH+ noradrenergic neurons in clonal assays (Kruger et al., 2002). However, gut NCSCs rarely gave rise to sympathoadrenal neurons (TH+, Gata2+, VAChT−) in vivo (Fig. 2). All 7 chicks examined for neuronal subtype markers in sympathetic ganglion had Stmn2+ neurons in the sympathetic chain, averaging 5.1±2.7 sections with Stmn2+ neurons per chick and 5.7±3.8 rat neurons per engrafted section. However, only two chicks contained any cells that expressed noradrenergic markers. One chick had 2 TH+ cells while a different chick had 2 Gata2+ cells. No chicks were found with TH+ and Gata2+ cells in adjacent sections. In contrast, the most frequently expressed marker was VAChT. In 5 of 7 chicks VAChT-positive cells (equivalent to 38 ± 57% of the number of Stmn2+ cells in adjacent sections) were found in the sympathetic chain. A subset of cells in developing sympathetic ganglia express a cholinergic phenotype (Ernsberger et al., 1997); however, few of the engrafted cells in sympathetic ganglia expressed the autonomic marker Gata2 (Fig. 2G). Thus gut NCSCs do not fully acquire an appropriate fate in sympathetic ganglia, or rarely do so. This is consistent with our prior observation that sciatic nerve NCSCs readily formed noradrenergic neurons in culture when stimulated by high levels of BMPs (Morrison et al., 2000a), but rarely formed noradrenergic neurons in vivo, where physiological BMP concentrations are presumably much more limiting (White et al., 2001).
Gut NCSCs also migrated into DRG (Table 1). To assay whether gut NCSCs adopted a sensory neuronal fate, DRG sections adjacent to those containing Stmn2+ cells were hybridized with a rat-specific Pou4f1 (Brn3a) probe, a marker of DRG sensory neurons (Fedtsova and Turner, 1995; Willert et al., 2003; Lee et al., 2004). Of the 5 chicks examined, none contained Pou4f1+ cells (data not shown). To further evaluate whether gut or sciatic nerve NCSCs had the potential to adopt a DRG sensory fate in vitro, gut and sciatic nerve NCSCs were cultured in conditions permissive for sensory neuron survival (see methods). Under these conditions, gut and sciatic nerve NCSCs rarely formed Pou4f1+ neurons, with 1 out of 182 gut NCSC colonies containing a single Pou4f1+ neuron and 1 out of 89 sciatic nerve NCSC colonies containing 3 Pou4f1+ neurons (data not shown). The addition of soluble Wnt3a (Wnt pathway activation promotes sensory neurogenesis by early migrating NCSCs (Lee et al., 2004)) had no effect on the ability of E14.5 NCSCs to form DRG sensory neurons. The limited ability of NCSCs to express Pou4f1 in vitro or in vivo suggests these cells lack significant DRG sensory potential.
Our use of a ubiquitously expressed genetic marker of transplanted cells (hPAP) in concert with rat markers of differentiation revealed that NCSCs engraft in some locations without differentiating into recognizable neuronal or glial fates. Both gut and sciatic nerve-derived cells migrated into the dermis, but P0+ glia and Stmn2+ neurons were never observed in the dermis (Table 1; Suppl. Fig. S1). Sciatic nerve NCSCs also failed to differentiate to either P0+ glia or Stmn2+ neurons (0 of 11 chicks) in DRG despite the presence of numerous hPAP+ cells in this location in 7 out of 11 transplanted chicks (Table 1). In control sections through rat DRG, neurons expressed Stmn2 and glia expressed P0 (data not shown). These data suggest that either the transplanted cells failed to differentiate in some locations, or that they acquired alternative non-neural fates.
To test whether the cells that engrafted in the dermis acquired a melanocyte fate, sections from 3 gut-injected and 3 sciatic nerve-injected chicks were stained with an antibody against Microphthalmia (MiTF), a marker of melanoblasts (Opdecamp et al., 1997). This antibody identified melanoblasts in positive control sections through the E14.5 rat eye and dermis. Sections from chick dermis that were flanked by hPAP+ rat cells were stained for MiTF or double labeled for MiTF and hPAP. However, no MiTF+ cells were found in any of these sections (data not shown). This suggests that the dermal cells had not adopted a melanocyte fate. Since the injected NCSCs were from albino (Sprague-Dawley) rats it was not informative to assay for melanin in this experiment.
To independently test whether gut or sciatic nerve NCSCs retained the ability to form melanocytes, NCSCs from Brown Norway rats were mixed with primary epidermal keratinocytes and dermal fibroblasts isolated from newborn CD-1 (albino) mice and injected into 8 week old NOD/scid mice in a hair follicle induction assay (Lichti et al., 1993; Hutchin et al., 2005; Zheng et al., 2005). While control experiments using primary epidermal cells and melanoblasts from pigmented C57BL mice led to the production of melanin, including pigmented hair shafts, no melanin or pigmented shafts were found in mice injected with gut or sciatic nerve NCSCs (Suppl. Fig. S2). These results suggest that gut and sciatic nerve NCSCs lack the ability to form melanocytes, consistent with the idea that melanocytic potential diverges from other neural crest fates around the time of neural crest migration (Duff et al., 1991; Erickson and Goins, 1995). Nonetheless, it also remains possible that mouse and chick dermis are not permissive for melanocyte differentiation by rat cells.
After the injection of gut NCSCs into the sacral somites of chick embryos, small numbers of Stmn2+ neurons were found in the guts of most of the chicks analyzed (Table 1). However, the relatively small numbers of neurons observed in the gut made it difficult to test for the expression of enteric neuronal markers. In culture, nearly 100% of fetal gut NCSCs and sciatic nerve NCSCs give rise to colonies containing neurons that express neuronal Nitric Oxide Synthase (nNOS), Vasoactive Intestinal Peptide (VIP) and Neuropeptide Y (NPY) (Kruger et al., 2002)(data not shown). These three neuropeptides are expressed in subsets of early developing neurons within the enteric nervous system (Pham et al., 1991; Young et al., 1998) and other PNS neurons such as DRG sensory neurons (Qian et al., 1996) and sympathetic chain neurons (Tyrrell and Landis, 1994; Ernsberger et al., 1997).
To more precisely test the ability of gut and sciatic NCSCs to acquire appropriate enteric fates these NCSCs were directly injected into the gut. Guts from 5 day-old chick embryos were dissected and freshly isolated hPAP+ NCSCs were injected into the cecum. Guts were then placed onto the chorioallantoic membrane of a 10 day old chick embryo to develop, in ovo, for 4–5 days (Kruger et al., 2003).The percentage of injected guts containing hPAP+ cells was similar irrespective of whether gut or sciatic nerve NCSCs were injected (Fig. 3C). However, guts injected with gut NCSCs had nearly 3 times as many hPAP+ sections/gut compared with guts injected with sciatic nerve NCSCs (Fig. 3C). All of the guts that engrafted with gut NCSCs also showed significantly (p<0.05) higher levels of engraftment, with an average of 74.4±68.7 cells per section while guts injected with sciatic nerve NCSCs had only 1.7±1.0 hPAP+ cells per section (Fig. 3). Sciatic nerve-derived cells were thus not able to engraft in the gut as extensively as gut NCSCs.
To determine the phenotype of the engrafted cells, sections were triple labeled with antibodies against hPAP, the neuronal marker β III-tubulin and either another marker found in enteric neurons (NPY, nNOS, VIP, Calbindin) or the glial marker GFAP (Fig. 4). Gut NCSC-derived hPAP+ cells almost always co-expressed β III-tubulin but never expressed GFAP, indicating that these cells acquired neuronal fates after engrafting in the gut in this assay (Fig. 4M). These gut NCSC-derived neurons frequently co-expressed the enteric neuronal markers nNOS (E-H) and NPY (A-D). A smaller proportion of these neurons co-expressed VIP (I-L) and none of the neurons expressed calbindin (Fig. 4M). Importantly, none of the cells that engrafted in the gut expressed Gata2 (Fig. 4O), indicating that these cells were differentiating to developmentally appropriate enteric neuronal phenotypes. The failure to express calbindin may reflect developmental timing as adult gut NCSCs make much larger numbers of calbindin+ neurons in culture than fetal gut NCSCs (data not shown).
In contrast to the appropriate differentiation of gut NCSCs, sciatic nerve NCSC-derived cells failed to express any neuronal or glial markers in tissue sections (data not shown). This indicates that sciatic nerve NCSCs lack the ability to respond to differentiation cues in the gut. Sciatic nerve and gut NCSCs thus exhibit intrinsic differences in their ability to migrate into the gut, the ability to form significant numbers of progeny after direct transplantation into the gut, and the ability to differentiate within the gut. The observation of such extensive differences among NCSCs from different regions of the developing PNS emphasizes the importance of using physiologically appropriate progenitors for cell therapy. For therapies contemplated in the enteric nervous system, gut NCSCs are likely to be more effective than stem cells isolated from other regions of the nervous system.
Hirschsprung disease is caused by an inability of neural crest progenitors to colonize a variable length of the hindgut in the absence of GDNF/Ret or Edn3/EDNRB signaling (Schuchardt et al., 1994; Gershon, 1999; Natarajan et al., 2002; Newgreen and Young, 2002; Iwashita et al., 2003; Kruger et al., 2003; Bondurand et al., 2006). This suggests that Hirschsprung disease might be more effectively treated by transplanting NCSCs into aganglionic or hypoganglionic segments of gut, in addition to conventional surgical therapies (Natarajan et al., 1999; Bondurand et al., 2003; Iwashita et al., 2003; Kruger et al., 2003).
Wild-type NCSCs can be transplanted into the aganglionic region of Ednrb-deficient guts and they engraft and undergo neurogenesis as efficiently as in the wild-type hindgut (Kruger et al., 2003). While this provides proof-of-principle that the aganglionic environment is permissive for the engraftment of transplanted NCSCs, the transplantation of wild-type NCSCs into an Ednrb-deficient gut would be allogeneic, raising the issues of donor identification, potential immune rejection, and immunosuppression. If the primary defect in the absence of EDNRB signaling represents a migration defect (Kruger et al., 2003), then it should also be possible to bypass the migration defect by transplanting Ednrb-deficient NCSCs into the aganglionic portion of the Ednrb-deficient hindgut. If Ednrb-deficient NCSCs are able to engraft and undergo neurogenesis, this might make it possible to perform autotransplants in which NCSCs are isolated from the foregut of a Hirschsprung patient, expanded in culture, and then grafted back into the hindgut of the same patient. Alternatively, if there are other environmental defects in the Ednrb-deficient hindgut that are non-permissive for the survival or neuronal differentiation of Ednrb-deficient NCSCs then this approach will not be feasible.
To test this, p75+α 4+ gut NCSCs isolated from Ednrbsl/sl rat fetuses or Ednrb+/+ littermates were cultured for 6–12 days in standard medium. After this expansion period individual wells of cells were dissociated and injected into E14.5 Ednrbsl/sl guts at a point midway between the cecum and rectum. The injected guts were then grown as explants on the chorioallantoic membrane of 7–10-day-old chicken embryos for 4–5 days (Kruger et al., 2003). Five of the seven mutant guts injected with Ednrbsl/sl cells survived on the chorioallantoic membrane and were engrafted by hPAP+ cells in the gut wall (Fig. 5A). Four of 6 mutant guts injected with wild-type cells survived and were positive for hPAP+ cells in the gut wall (Fig. 5B). The degree of engraftment was similar irrespective of whether Ednrbsl/sl cells (13.4±5.5 hPAP+ sections/gut and 13.7±10.8 hPAP+ cells/section) or Ednrb+/+ cells (10.8±7.3 hPAP+ sections/gut and 3.8±3.2 hPAP+ cells/section) were injected. In both cases, the hPAP+ cells included similar proportions of β-III tubulin, nNOS, and/or NPY and VIP expressing cells (Fig. 5C–O). Ednrb+/+ and Ednrbsl/sl cells engrafted over a similar distance in the aganglionic gut (Fig. 5P). Thus, cultured Ednrbsl/sl and Ednrb+/+ gut neural crest progenitors were similarly capable of engrafting in the aganglionic region of the Ednrbsl/sl gut and undergoing neurogenesis.
Even when the cultured cells were injected into the more proximal aganglionic region of the distal colon, at a level where endogenous neurons were present, the transplanted cells survived irrespective of genotype and formed neurons that expressed β III-tubulin, nNOS, NPY, and/or VIP (data not shown). These results demonstrate that the aganglionic and hypoganglionic regions of the Ednrbsl/sl gut are permissive for the survival and neuronal differentiation of Ednrbsl/sl neural crest progenitors expanded in culture.
We have systematically examined the developmental potential of prospectively identified E14.5 gut NCSCs in vivo. These cells were able to migrate widely throughout the developing PNS after transplantation into the sacral neural crest migration pathway of chick embryos (Table 1). They formed neurons and glia in DRG, sympathetic chain, peripheral nerve, Remak’s ganglion, pelvic plexus, and gut (Table 1). They formed primarily glia in the ventral nerve root and the ventral column of the neural tube (Table 1). In contrast, sciatic nerve NCSCs exhibited striking differences in migration and fate determination. They were more likely to migrate into the dermis and less likely to migrate into the ventral column of the neural tube (Table 1). They were also unable to migrate into the gut or to form neurons or glia after direct transplantation into the gut (Table 1). Importantly, sciatic nerve NCSCs do not express Ret, in contrast to gut NCSCs which express Ret uniformly (Iwashita et al., 2003). Given that Ret regulates the migration of gut NCSCs and plays a critical role in enteric nervous system development (Schuchardt et al., 1994; Taraviras et al., 1999; Natarajan et al., 2002; Iwashita et al., 2003), the lack of Ret expression by sciatic nerve NCSCs may contribute to their inability to engraft or differentiate in the gut.
Cell intrinsic differences in the migratory properties of mixed populations of neural crest progenitors have previously been observed (Bronner-Fraser et al., 1980; Erickson et al., 1992; Reedy et al., 1998), including between vagal neural crest cells (which give rise to most of the enteric nervous system) and other trunk or sacral neural crest cells (Le Douarin and Teillet, 1974; Burns et al., 2002; De Bellard et al., 2003). However, the extent to which these migratory differences also apply to regionally distinct populations of postmigratory NCSCs has been less certain. Our data demonstrate that regionally distinct NCSC populations exhibit pervasive cell-intrinsic differences in terms of migratory potential and lineage determination.
Vagal neural crest cells give rise to most of the enteric nervous system while sacral neural crest cells contribute to a lesser extent and only in the hindgut. Consistent with this, p75+α 4+ gut NCSCs express Hox genes consistent with a vagal origin ((Iwashita et al., 2003), data not shown). Burns et al (2002) previously demonstrated that when vagal neural crest cells were transplanted into the sacral region they migrated into the gut by embryonic day 4 (HH stage 23–24). Our finding that fetal gut NCSCs migrated into the gut by day 5.5 (HH 28) is consistent with their vagal origin and the timing observed by Burns et al (2002).
Gut NCSCs generated a variety of enteric neuronal phenotypes in the gut (Fig. 4) and appropriately generated autonomic cholinergic neurons in parasympathetic ganglia (Fig. 1). Importantly, the cholinergic neurons in parasympathetic ganglia expressed the autonomic marker Gata-2 while the enteric neurons did not (Fig.1, Fig. 4). This demonstrates that these cells acquired appropriate fates rather than just generating the same progeny in both locations. However, gut NCSCs did not generate the full complement of neurons normally found in parasympathetic ganglia as no TH+ noradrenergic neurons were found in Remak’s ganglion or in the pelvic plexus. Rare TH+ neurons are normally present in the day 6 embryonic chick pelvic plexus and eventually appear in Remak’s ganglion as well (Teillet, 1978). Perhaps with a longer incubation, TH+ cells would have arisen from the progeny of gut NCSCs in parasympathetic ganglia. Alternatively, gut NCSCs may not be able to make noradrenergic neurons in parasympathetic ganglia. Gut NCSCs also failed to generate neurons that fully differentiated to a noradrenergic phenotype in the sympathetic chain (Fig. 2). These results contrast with the ability of gut NCSCs to readily form noradrenergic neurons in culture (Kruger et al., 2002).
Gut NCSCs never convincingly differentiated into DRG sensory neurons in vitro or in vivo, despite engrafting in sensory ganglia (Table 1). Self-renewing and multipotent postmigratory gut NCSCs thus retain the ability to form appropriate types of neurons and glia in multiple regions of the PNS, including regions that they would not normally engraft under physiological circumstances, but they do not retain the potential to make all types of PNS cells.
A prior study that compared the engraftment of enteric neural crest precursors and early migrating neural crest cells in ovo also concluded that the enteric cells were able to form parasympathetic but not sympathetic or DRG sensory neurons, despite engrafting in each of these locations (White and Anderson, 1999). However, the enteric cells used in the prior study were selected as Ret+ and B2- and likely represented a more heterogeneous population of progenitors that included many restricted progenitors. Our study shows that this sublineage restriction also applies to a highly enriched population of gut NCSCs, and our studies extend the results of White and Anderson (1999) by showing that gut NCSCs can give rise to appropriate neurons after transplantation into the enteric nervous system.
Gut and sciatic nerve NCSCs also did not give rise to melanocyte lineage cells in our experiments. Although the progeny of both gut and sciatic nerve NCSCs were found in the dermis of 40% and 100% of engrafted chicks, respectively (Table 1), these cells never expressed the early melanoblast transcription factor MiTF (data not shown) nor did they produce melanin in a hair follicle morphogenesis assay (Suppl. Fig. S2). Based on reports that the dorsolateral pathway is accessible only to cells that are fated to make melanocytes (Erickson and Goins, 1995) it was surprising to find progeny of transplanted NCSCs in the dermis despite their failure to form melanocytes. Fetal avian sciatic nerve (and DRG) or adult mouse nerve can give rise to pigmented cells in culture when treated with EDN3 and 12-O-tetradecanoylphorbol-13-acetate (TPA), or after injury (Ciment, 1990; Rizvi et al., 2002; Dupin et al., 2003). Our results suggest that these melanogenic cells were either not NCSCs or that melanogenesis in culture or after injury does not necessarily reflect melanogenic potential during normal development.
Our results demonstrating the ability of gut NCSCs to form enteric neurons are important in light of the current interest in cell therapies for Hirschsprung disease (Natarajan et al., 1999; Bondurand et al., 2003; Iwashita et al., 2003; Kruger et al., 2003; Burns et al., 2004). While we had demonstrated previously that wild-type gut NCSCs could engraft and undergo neurogenesis in the aganglionic region of the Ednrbsl/sl gut (Kruger et al., 2003), it remained uncertain whether the aganglionic environment is permissive for the engraftment or differentiation of Ednrbsl/sl NCSCs. In this study we show that even Ednrbsl/sl NCSCs were able to engraft and undergo neurogenesis as efficiently as wild-type cells in the aganglionic region of the Ednrbsl/sl gut. These results demonstrate proof-of-principle that it is possible to extract Ednrbsl/sl NCSCs from the gut, expand these cells in culture, and then transplant them into the aganglionic region of the Ednrbsl/sl gut. This raises the possibility of performing autotransplants of culture expanded NCSCs from the normoganglionic foregut to the aganglionic/hypoganglionic hindgut of patients with Hirschsprung disease. This would eliminate the challenges of finding suitable donors and the need for immunosuppression to avoid rejection of allogeneic cells.
Despite these encouraging results impediments remain. Cultured gut NCSCs tended to give lower levels of engraftment as compared to freshly isolated, uncultured gut NCSCs (Suppl Fig. 3). NCSCs undergo considerable changes within days of being added to culture (Bixby et al., 2002). Indeed, Ret expression by NCSCs declines in culture (Suppl. Fig. 4). All uncultured gut NCSCs express Ret (Iwashita et al., 2003) and the vast majority of neurosphere-forming cells among uncultured neural crest progenitors are Ret+ (Suppl. Fig. 4). However, after 10 days in culture, neurospheres formed by Ret+ cells contained both Ret+ and Ret- cells that were equally likely to form neurospheres upon subcloning into secondary cultures (Suppl. Fig. 4). This loss of Ret expression on a subset of multipotent cells in culture may explain the observation that cultured neural crest progenitors are predominantly Ret- (Bondurand et al., 2003) despite the fact that most neural crest progenitors (Young et al., 2003) and NCSCs (Iwashita et al., 2003) in the developing gut are Ret+. Since Ret signaling regulates the migration of NCSCs as well as the proliferation and differentiation of downstream progenitors (Heuckeroth et al., 1998; Natarajan et al., 2002; Barlow et al., 2003; Iwashita et al., 2003), these results might at least partially explain the reduced engraftment levels from cultured NCSCs.
Our results argue against the idea that Edn3 signaling is necessary for the maintenance of NCSCs from the rat enteric nervous system. A recent study (Bondurand et al., 2006) suggested that in the absence of Edn3 multipotent mouse neural crest progenitors could not be maintained in culture, and prematurely differentiated while migrating through the gut. In contrast, we have found that rat NCSCs can be maintained in culture in chemically defined medium lacking Edn3 (Iwashita et al., 2003; Kruger et al., 2003), and addition of Edn3 to culture does not inhibit differentiation or promote NCSC proliferation, or self-renewal (Kruger et al., 2003). We did observe a 2.5-fold reduction in the frequency of NCSCs in the E12.5 Ednrbsl/sl rat gut, reflecting an early role for Edn3 in the expansion of the pool of gut NCSCs. However, we did not detect any further reduction in NCSCs after E12.5, when the migratory defect became apparent in the rat gut, nor did we detect decreased NCSC proliferation in vivo or premature differentiation at the neural crest migration front after E12.5 in Ednrbsl/sl rats (Kruger et al., 2003).
Our ability in the current study to expand Ednrbsl/sl gut NCSCs in culture prior to engrafting these cells in the Ednrbsl/sl gut further demonstrates that rat gut NCSCs can be maintained in culture in the absence of EDNRB signaling. These observations raise the possibility that rat and mouse NCSCs respond differently to Edn3. One possibility is that Endothelin’s role in promoting the proliferation of undifferentiated neural crest progenitors may persist later into development in mice as compared to rats, and the failure of NCSCs to colonize the hindgut in Ednrbsl/sl rats may not reflect a role for Edn3 in the maintenance of migrating NCSCs in the intestine. Indeed, normal frequencies of NCSCs persist in the foregut and midgut of Ednrbsl/sl rats in late fetal development and postnatally, well after neural crest cells have failed to migrate into the hindgut (Kruger et al., 2003). The failure of NCSCs to migrate into the Ednrbsl/sl hindgut, despite their maintenance in normal numbers within the midgut, suggests that in rats the aganglionosis caused by Ednrb-deficiency is primarily caused by effects on migration.
This study demonstrates that gut NCSCs intrinsically differ from sciatic nerve NCSCs in terms of migratory potential, fate determination and the ability to form enteric cell types. The finding that regional differences in neural stem cell populations strongly affect their capacity to migrate and differentiate may have broad implications for the potential use of stem cells for tissue repair. Although many stem cell populations may be able to generate desired cell types in culture, the use of physiologically appropriate stem cells may be crucial to achieve appropriate migration, differentiation, and integration of desired cell types in vivo. Our data suggest that the physiological progenitors of the enteric nervous system, gut NCSCs, are likely to be more effective in cell therapies for Hirschsprung disease than neural stem cells from other regions of the nervous system. Our observation that Ednrbsl/sl gut NCSCs can be expanded in culture and engrafted in the aganglionic region of the Ednrbsl/sl gut supports this idea. Nonetheless, additional work will be required to test whether transplanted cells can engraft and differentiate robustly enough to restore the function of aganglionic segments of gut.
Supplementary Figure 1: NCSC-derived hPAP+ cells migrate into the dermis. E14.5 hPAP+ gut and sciatic nerve NCSCs were injected into the hindlimb bud of stage 17–18 chick embryos. A representative image of the hindlimb area of a stage 29 embryo with sciatic nerve NCSC-derived hPAP+ cells in the nerve (arrow) and in the dermis (arrow heads). hPAP+ cells were found in the dermis of chicks injected with gut (2 of 5 injected chicks) and sciatic nerve (11 of 11 injected chicks) NCSCs. Sciatic nerve NCSCs yielded more engrafted sections per chick in the dermis (10.2 ± 8.6 versus 4.5 ± 3.5 in chicks injected with gut NCSCs; p < 0.05; Table 1).
Supplementary Figure 2: NCSC-derived cells did not produce melanin in a hair morphogenesis assay. To test whether postmigratory NCSCs retained the ability to form melanocytes, gut or sciatic nerve NCSCs from E14.5 Brown Norway rats were mixed with primary epidermal preparations (containing epithelial cells and melanocyte progenitors) and dermal cells isolated from newborn CD-1 (albino) mice and injected into 8 week old nude mice in a hair follicle induction assay (Lichti et al., 1993; Hutchin et al., 2005; Zheng et al., 2005). Three weeks after injection, the animals were sacrificed and tissue from the injection sites was fixed in 4% formalin, embedded in paraffin, sectioned and stained with hematoxylin and eosin (red). Sections from positive controls injected with keratinocytes from C57BL mice (A) consistently contained both melanin-containing hair shafts (arrow head) and melanin-negative follicles (arrow). Sections from mice injected with gut or sciatic nerve NCSCs (B) did not contain melanin (arrows). Three control mice, one mouse injected with gut NCSCs, and one mouse injected with sciatic nerve NCSCs were examined in this experiment (approximately 50 sections per mouse).
Supplementary Figure 3: Uncultured fetal gut NCSCs tend to engraft more efficiently than cultured NCSC. Freshly isolated E14.5 fetal gut NCSCs or E14.5 fetal gut NCSCs that had been cultured for 6 to 12 days were injected into the hindgut of wild-type or Ednrbsl/sl rats. Injected guts were explanted onto the chorioallantoic membrane of 7–10 day old chicks for 4–5 days and then fixed for immunohistochemistry and sectioned. Initially, every tenth section from each gut was processed chromogenically with the alkaline phosphatase substrate NBT/BCIP. Sections adjacent to those that were positive for hPAP+ cells were stained with an antibody against hPAP and the nuclear marker DAPI. Cells double positive for hPAP and DAPI were counted on each engrafted section. The number of engrafted rat cells per section was 2–10 fold greater when uncultured cells were transplanted as compared to cultured cells (this paper, blue bars). We previously saw similarly high levels of engraftment from uncultured cells (Kruger et al., 2003) (gray bars). Note that the high variability between sections meant that these differences were not statistically significant.
Supplementary Figure 4: Ret expression by NCSCs declines in culture. Uncultured Ret+ cells sorted from E14.5 rat guts were approximately 30-fold more likely to form neurospheres in culture than Ret− cells from the same guts (A; p < 0.01). In other experiments, uncultured neural crest progenitors were isolated by flow-cytometry from E14.5 rat guts based on p75 staining then subdivided into Ret+ and Ret− fractions (B). The Ret+ subset was significantly more enriched for cells with the ability to form neurospheres than the Ret− subset (p < 0.01). Indeed, it was possible that the low level of neurosphere-forming activity in the Ret− fraction in these experiments reflected contamination by Ret+ cells. Alternatively, the small population of Ret− progenitors may represent trunk-derived crest cells which are not dependent on Ret expression and which migrate as far as the proximal stomach (Durbec et al., 1996). After 10 days in culture, individual neurospheres were dissociated, stained with Ret and resorted into secondary cultures. Ret+ and Ret− cells were now equally capable of forming neurospheres in secondary cultures (C), even when isolated from primary neurospheres that arose from Ret+ progenitors.
This work was supported by the National Institutes of Health (RO1 NS40750), the Searle Scholars Program, and the Howard Hughes Medical Institute. Thanks to David Adams, Martin White, Ann Marie Deslaurier, and the University of Michigan Flow-Cytometry Core Facility. Flow-cytometry was supported in part by the UM-Comprehensive Cancer Center NIH CA46592, and the UM-Multipurpose Arthritis Center NIH AR20557. Thanks to Chris Edwards and Bruce Donohoe for assistance with confocal microscopy and Eric Turner for providing an antibody against Pou4f1. Thanks To Evan Michael for technical assistance in the hair morphogenesis assay and to Cathy Krull for discussions related to the timing of neural crest migration in chicks.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.