Neural crest origin of olfactory ensheathing glia

Abstract

Olfactory ensheathing cells (OECs) are a unique class of glial cells with exceptional translational potential because of their ability to support axon regeneration in the central nervous system. Although OECs are similar in many ways to immature and nonmyelinating Schwann cells, and can myelinate large-diameter axons indistinguishably from myelination by Schwann cells, current dogma holds that OECs arise from the olfactory epithelium. Here, using fate-mapping techniques in chicken embryos and genetic lineage tracing in mice, we show that OECs in fact originate from the neural crest and hence share a common developmental heritage with Schwann cells. This explains the similarities between OECs and Schwann cells and overturns the existing dogma on the developmental origin of OECs. Because neural crest stem cells persist in adult tissue, including skin and hair follicles, our results also raise the possibility that patient-derived neural crest stem cells could in the future provide an abundant and accessible source of autologous OECs for cell transplantation therapy for the injured central nervous system.

Olfactory ensheathing cells (OECs) are a unique class of vertebrate glial cells that envelop bundles of olfactory axons, both peripherally in the olfactory nerve and within the olfactory nerve layer (ONL) of the olfactory bulb (1–4). Olfactory receptor neurons (ORNs), whose cell bodies reside in the epithelium of the nasal cavity, are continually replenished: throughout adult life, OECs support axon outgrowth from nascent ORNs and help guide them to their synaptic targets in the olfactory bulb (1–4). Given these properties, OECs are thought to have considerable potential as therapeutic agents for central nervous system repair; indeed, when grafted into spinal cord lesions, OECs intermingle with astrocytes, myelinate demyelinated axons, and promote axon sprouting (1–4). OECs can be cultured from the olfactory mucosa, which is easily accessible within the nasal cavity (1–4). However, this tissue also contains antigenically similar Schwann cells, making the reliable identification of OECs in such cultures a major challenge for OEC-mediated transplantation therapy for human spinal cord repair (4).

OECs are currently thought to arise from the olfactory epithelium (5, 6), which derives from the olfactory placodes (patches of thickened embryonic head ectoderm), whose precursor cells lie next to future olfactory bulb cells in the anterior neural folds (5, 7). All other peripheral glia, i.e., myelinating and nonmyelinating Schwann cells in peripheral nerves, and satellite cells in peripheral ganglia are derived from neural crest cells (NCCs) (8), which emigrate from the dorsal neural tube early in development. OECs are similar in many ways to immature and nonmyelinating Schwann cells (3, 9–11), ensheathing bundles of small-diameter axons without forming myelin (Fig. S1A); furthermore, the OEC transcriptome is closer to that of Schwann cells than astrocytes (12), and OECs will myelinate larger-diameter axons indistinguishably from Schwann cells, both in vitro and after transplantation into demyelinated spinal cord lesions (9–11, 13, 14). [Olfactory axons are much smaller than the threshold axon diameter for myelination (15, 16).] Conversely, NCC-derived cells can adopt OEC-like phenotypes: adult rat Schwann cells transplanted to the olfactory mucosa ensheath olfactory axons in large bundles of closely packed axons without forming myelin, similarly to OECs (17), whereas embryonic rat Schwann cell precursors behave more similarly to OECs than to mature Schwann cells when transplanted into demyelinated spinal cord lesions (18).

The original data interpreted as supporting an olfactory placode origin for OECs were from an anterior neural fold (ANF) fate map (5). When grafted isotopically from quail to chicken embryos, the ANF formed the olfactory bulb, the olfactory placode, and cells on the olfactory nerve that must have migrated from the olfactory placode: these cells were assumed to be glia (5). The other evidence taken to support an olfactory placode origin for OECs was the demonstration that OECs migrate out of explanted embryonic day (E)14/juvenile rat olfactory epithelium after dissecting away the underlying connective tissue layer, the lamina propria (6). However, NCCs are found throughout the frontonasal mass (19), are associated with the olfactory epithelium throughout its development, and form the lamina propria, so these explants could still have contained NCCs. NCCs normally colonize all other developing peripheral nerves, so if OECs are not NCC-derived, the developing olfactory nerve would be an exception. In the extensive literature on NCC migration and development, we found only one hint that NCCs can colonize the olfactory nerve: a study on avian lower jaw development described in passing that hindbrain-level NCCs will heterotopically colonize the olfactory nerve when grafted to the diencephalon or when more rostral host NCC precursors are ablated (20). However, these experimental data do not describe normal development, and the fate of the heterotopic NCC-derived cells on the olfactory nerve (glial or otherwise) was not examined. Here, we use fate-mapping experiments in chicken embryos and genetic lineage-tracing in mice to show that OECs are indeed derived during normal development from NCCs, not from the olfactory placodes.

Results

The Olfactory Placode Does Not Form OECs.

We repeated the ANF fate-mapping experiments interpreted as showing an olfactory placode origin for OECs (5) by unilaterally grafting the ANF from 3 to 6 somite-stage (ss) transgenic ^GFPchicken embryos [which ubiquitously express cytoplasmic GFP (21)] into wild-type hosts (Fig. S2 A and B). Using ^GFPchick donors, instead of quail, enabled us to examine cell morphology as well as expression of molecular markers (quail cells are identified by a peri-nuclear antigen). Only ANF-grafted embryos in which the olfactory epithelium was labeled and the olfactory nerve developed normally on the grafted side were analyzed further (n = 10).

Embryos fixed at E4.5 or E6.5 (n = 7) were sectioned and immunostained for GFP and neuronal β-III tubulin (Fig. 1 A–B2 and E–H2) or the neuron-specific RNA-binding protein HuC/D (Fig. 1 C–D2). Many ANF-derived cells were present on the olfactory nerve (Fig. 1 A–D2), as found at E5 in the original study (5). However, all these cells seemed to express neuronal β-III tubulin or HuC/D; i.e., they were neurons (Fig. 1 A–D2). In three ANF-grafted embryos surviving to E10.25 in which the olfactory epithelium was labeled on the grafted side (Fig. 1 E1–G1), immunostaining on sections showed bundles of olfactory axons ensheathed by processes expressing the OEC marker p75^NTR (the low-affinity neurotrophin receptor), both in the lamina propria adjacent to the olfactory epithelium (Fig. 1 F and G) and more proximally along the main body of the nerve (Fig. 1H). However, almost none of the p75^NTR-positive OEC processes were GFP-positive; i.e., they did not originate from the graft (Fig. 1 G–H2). In all three embryos, occasional GFP-positive cells were seen in the frontonasal mesenchyme and in the cartilage of the nasal septum (Fig. 1 E1 and H), so the graft also contributed a few NCCs. Given this, the occasional p75^NTR-positive/GFP-positive cells seen in the olfactory nerve in these embryos (Fig. 1H2) are most likely NCC-derived (compare Fig. 1 H–H2 with Fig. 3 C–D2). Our results do not support the hypothesis that OECs are derived from the olfactory placodes.

Fig. 1.

The olfactory placode does not form OECs. Unilateral isotopic grafts of ^GFPchick ANF at the 3–5 ss label the olfactory epithelium (and prospective olfactory bulb) on the operated side at E4.5 (A and C) and E10.25 (E and F, which show examples from two different embryos). At E4.5, all of the olfactory placode-derived cells on the olfactory nerve seem to be neurons, expressing the neuronal markers neuronal β-III tubulin (B–B2) and/or HuC/D (D–D2). At E10.25, p75^NTR-positive OEC processes (red) ensheath bundles of olfactory axons (blue), both in the lamina propria beneath the olfactory epithelium (G and G1) and more proximally along the nerve (H–H2). Virtually none of the p75^NTR-positive OEC processes (red) are graft-derived (G1, H1, and H2). Arrowheads in G and G1 indicate examples of p75^NTR-positive/GFP-negative OEC processes ensheathing bundles of GFP-positive olfactory axons. Occasional p75^NTR-positive/GFP-positive processes (arrow in H–H2) are most likely derived from the few NCCs contributed by the graft (scattered GFP-positive cells are seen in the nasal septum and frontonasal mesenchyme: arrowheads in H–H2). nβ3-tub, neuronal β-III tubulin; OB, prospective olfactory bulb; OE, olfactory epithelium; ON, olfactory nerve; ns, nasal septum.

Fig. 3.

Chick neural crest cells form OECs along the length of the olfactory nerve from the lamina propria to the olfactory bulb. Isotopic grafts of ^GFPchick midbrain-level neural fold at the 5 ss label migratory NCCs. At E10.25, p75^NTR-positive OECs are associated with p75^NTR-positive olfactory axons/neurons in the lamina propria (A and B). These p75^NTR-positive OECs, like the rest of the lamina propria and the cribriform plate, are ^GFPchick NCC-derived (arrowheads highlight examples) (A1, A2, B1, and B2). Similarly, the p75^NTR-positive OEC processes ensheathing olfactory axon bundles in the olfactory nerve (C and D) are from ^GFPchick NCC-derived cells (C1, C2, D1, and D2), whose cell bodies can be seen as brighter patches of immunofluorescence between the axon bundles (the p75^NTR immunoreactivity is confined to their cell processes). P0-positive OEC processes (E) are also from ^GFPchick NCC-derived cells (E1 and E2). (F) Low-power composite view showing that ^GFPchick NCC-derived cells not only surround the olfactory bulb but also are found throughout the ONL. [The scattered ^GFPchick NCC-derived cells in the olfactory bulb are smooth muscle actin-positive pericytes associated with the forebrain vasculature (also see Fig. S3).] (G–G2) Higher-power view showing ^GFPchick NCC-derived cells throughout the ONL, with cells in the outer layer expressing both p75^NTR (blue) and P0 (red), and cells in the inner layer expressing P0 but not p75^NTR. (H–H2) In a different E10.25 embryo, the olfactory bulb is surrounded by ^GFPchick NCC-derived cells, and almost all of the P0-positive OECs inside the olfactory bulb are ^GFPchick NCC-derived (arrowheads highlight examples). Arrows highlight some P0-positive OECs that are not ^GFPchick graft-derived; these are likely to be derived from host NCCs. CP, cribriform plate; LP, lamina propria; nβ3-tub, neuronal β-III tubulin; OE, olfactory epithelium; ON, olfactory nerve; ONL, olfactory nerve layer.

Neural Crest Cells Form OECs in Avian Embryos.

To test whether NCCs contribute to OECs, we isotopically grafted 4–7 ss midbrain-level NCC precursors (i.e., neural folds) unilaterally or bilaterally from either transgenic ^GFPchick or quail donors into wild-type chick hosts (Fig. S2 C–F). Embryos were fixed and sectioned at stages ranging from E4.5 (when the olfactory nerve first reaches the telencephalon) to E10.25 (when OECs ensheath axon bundles along the olfactory nerve). After such grafts (n = 35), the olfactory epithelium was unlabeled and graft-derived NCCs filled the frontonasal mass (e.g., Fig. 2 A1, C1, and E). At E4.5, many nonneuronal graft-derived NCCs were closely associated with the olfactory nerve (Fig. 2 A–B2; n = 12/13). At E5.5–E6.5, graft-derived NCCs were found in abundance throughout the olfactory nerve (Fig. 2 C–F2; n = 16/17). They expressed the transcription factor gene Sox10 (Fig. 2 C and D; n = 9/9), which is expressed in migrating NCCs and required for the differentiation of NCC-derived glia (23). Sox10 also directly regulates expression of the “myelin” P0 gene (24), a marker for chick and rat OECs (22–26) as well as for migrating NCCs and their glial derivatives (26, 27). Many of the ^GFPchick NCCs and their processes on the olfactory nerve at E6.5 expressed the OEC marker P0 (Fig. 2 F–F2). In embryos surviving to E8.5–E10.25, graft-derived NCCs expressing OEC markers (p75^NTR and/or P0) were associated with olfactory axons in the lamina propria (Fig. 3 A–B2), ensheathed bundles of axons in the main body of the olfactory nerve (Fig. 3 C–E2), and were found in abundance in the ONL of the olfactory bulb (Fig. 3 F–H2) (n = 5/5). We note that, consistent with previous reports from both rat and mouse (28, 29), OECs in the innermost layer of the embryonic chick ONL did not express p75^NTR (Fig. 3 G–G2). [We also saw ^GFPchick NCCs scattered throughout the forebrain (Fig. 3F): these cells express smooth muscle actin (Fig. S3), so are likely to be pericytes in forebrain blood vessels, which, like forebrain meninges, are derived from midbrain-level NCCs (30).]

Fig. 2.

Avian neural crest cells form OECs on the olfactory nerve. Isotopic grafts of ^GFPchick (A–B2 and E–F2) or quail (C–D) midbrain-level neural fold at the 5-ss label migratory NCCs that colonize the olfactory nerve as well as the surrounding frontonasal mesenchyme. At E4.5 (A–B2), nonneuronal NCCs are associated with the olfactory nerve. (Faint green staining in the apical olfactory epithelium is background.) At E5.5 (C–D), many Sox10-positive cells are seen on the olfactory nerve on the grafted side of the embryo. Quail NCC-derived cells (green nuclei) on the olfactory nerve express Sox10 (D). At E6.5 (E–F2), a significant proportion of olfactory nerve-associated ^GFPchick NCCs and their processes express the OEC marker P0 (arrowheads, F–F2) (22). nβ3-tub, neuronal β-III tubulin; OB, olfactory bulb; OE, olfactory epithelium; ON, olfactory nerve.

These results show that avian NCCs form OECs, identified by virtue of their location (i.e., along the length of the olfactory nerve, from the lamina propria to the ONL), their expression of the OEC markers p75^NTR and/or P0, and their morphology (i.e., their processes ensheath bundles of olfactory axons). Furthermore, Sox10 is a previously unrecognized marker for OECs at all stages examined (Fig. S4).

Other NCC Populations Are Competent to Form OECs.

To determine whether competence to form OECs is restricted to midbrain-level NCCs, we heterotopically grafted ^GFPchick neural folds (before NCC emigration) from prospective hindbrain or spinal cord in place of rostral midbrain neural folds in 4–9 ss wild-type hosts (Fig. S2 G and H). Embryos were fixed at either E6.5 or E9.5–10.25 (n = 19). ^GFPChick NCCs, at least some of which expressed p75^NTR, were found on the olfactory nerve after heterotopically grafting either hindbrain rhombomeres 4–6 (n = 7/9) or prospective spinal cord (n = 7/10) (Fig. 4 A–C2). In embryos surviving to E9.5–E10.25, ^GFPchick NCC processes ensheathed bundles of axons in the olfactory nerve and were found abundantly in the ONL, after heterotopic grafts of rhombomeres 4–6 (n = 2/3) or of prospective spinal cord (n = 3/4) (Fig. 4 C–E2). We conclude that NCCs from both the hindbrain and the spinal cord can form OECs.

Fig. 4.

Neural crest cells from the hindbrain can form OECs. (A–B2) At E6.5, ^GFPchick NCC-derived cells, at least some of which express p75^NTR (arrowheads highlight examples), are present on the olfactory nerve after heterotopic grafts of rhombomeres 4–6 to the rostral midbrain. (C–C2) At E9.5–10.25, ^GFPchick hindbrain NCC-derived cells, at least some of which express p75^NTR (arrowheads highlight examples), ensheath bundles of axons in the olfactory nerve. (D–E2) At E9.5–10.25, p75^NTR-positive ^GFPchick hindbrain NCC-derived cells are abundantly found in the ONL. nβ3-tub, neuronal β-III tubulin; OB, olfactory bulb; OE, olfactory epithelium; ON, olfactory nerve; ONL, olfactory nerve layer.

Neural Crest Cells Form OECs in Mouse Embryos.

Embryos from crosses between Wnt1^Cre (31) and R26R reporter mice, in which NCCs are permanently lineage-labeled genetically with lacZ (32) or YFP (33), have been used extensively to fate-map mouse NCCs. Wnt1 is exclusively expressed in the developing central nervous system and early migrating NCCs, with a rostral limit in the diencephalon (31, 32). We confirmed that Wnt1 is not expressed by any cells of the developing olfactory system, either at E11.5 (when the olfactory nerve has reached the telencephalon) or at E14.5 (when the definitive ONL can be identified), although Wnt1 expression can be detected in the central nervous system at both stages (Fig. S5). We immunostained frontal sections through the olfactory region of E13.5 and E14.0 Wnt1^Cre;R26R^YFP embryos (33) for YFP, neuronal β-III tubulin, and p75^NTR (Fig. 5). Only faint background YFP immunostaining was seen in Wnt1^Cre-negative littermates (compare Fig. 5 A1 and A2 with Fig. 5 B1 and B2; also see Fig. S6 M–P1). In contrast, Wnt1^Cre;R26R^YFP embryos showed robust YFP immunostaining (i.e., NCC-derived cells) throughout the frontonasal mass, as expected (Fig. 5 B1 and B2 and Fig. S6). As in our ^GFPchick NCC grafts, p75^NTR-positive OECs associated with olfactory axons/neurons in the lamina propria were NCC-derived (Fig. 5 C–C2; compare with Fig. 3 A–B2). YFP-positive NCC-derived cells, some of which coexpressed p75^NTR, were also found throughout the ONL (Fig. 5 B–B2 and D–D2; compare with Fig. 3 F–G2). The lack of p75^NTR expression in many OECs in the ONL is consistent with previous reports (28, 29). p75^NTR-positive NCC-derived cells outside the ONL are part of the forebrain meninges (28, 30, 34). We also saw YFP-positive NCC-derived cells scattered throughout the forebrain in association with blood vessels (Fig. S6 K and L), presumably NCC-derived pericytes in forebrain blood vessels (30). We confirmed by in situ hybridization on sections of E17.5 wild-type embryos that, as in chick embryos, Sox10 is a marker for mouse OECs (Fig. 5 E–F1). At E16.5–E17.5, lacZ-positive NCCs in Wnt1^Cre;R26R^lacZ embryos made a significant contribution to the ONL (Fig. S7 A–B2) and to cells associated with the olfactory nerve (Fig. S7 C–D2), as well as to the lamina propria, cribriform plate, nasal septum, and forebrain meninges (Fig. S7 A–C2). These genetic lineage-labeling results are consistent with the NCC fate-mapping data from our avian neural fold grafts. Overall, we conclude that OECs are derived from NCCs, not from the olfactory placodes.

Fig. 5.

Genetic lineage-tracing using Wnt1^Cre;R26R^YFP embryos shows that mouse neural crest cells form OECs. (A–A2) No above-background YFP immunoreactivity is seen in the olfactory region of an E13.5 Wnt1^Cre-negative embryo from a cross between Wnt1^Cre and R26R^YFP reporter mice (compare A1 and A2 with B1 and B2). (B–B2) In a Wnt1^Cre;R26R^YFP littermate, YFP-positive NCC-derived cells are found throughout the frontonasal mass, as well as in the ONL. The olfactory epithelia and vomeronasal organ epithelia show no above-background YFP immunoreactivity (compare B1 and B2 with A1 and A2). (C–C2) p75^NTR-positive OECs associated with olfactory axons/neurons in the lamina propria are NCC-derived (YFP-positive), as is the rest of the lamina propria. Arrowheads highlight examples. (D–D2) YFP-positive NCC-derived cells, some of which express p75^NTR, are found throughout the ONL. Arrowheads highlight examples of p75^NTR-positive OECs in the ONL. (E–F2) In situ hybridization on sections of a wild-type E17.5 embryo shows that Sox10 is a marker for mouse OECs. [Sox10 expression in the olfactory epithelium likely represents developing Bowman's glands (Fig. S8).] LP, lamina propria; nβ3-tub, neuronal β-III tubulin; OB, olfactory bulb; OE, olfactory epithelium; ON, olfactory nerve; ONL, olfactory nerve layer; vno, vomeronasal organ.

Discussion

Our avian fate-mapping results show that the olfactory placodes do not form OECs: the Schwann cells on the olfactory nerve described in the original quail-chick ANF fate-mapping experiments (5) were in fact olfactory placode-derived neurons, which migrate into the forebrain (35, 36). (The existence of migratory olfactory placode-derived neurons was not reported until several years after the ANF fate-mapping experiments, so at the time, it was a reasonable assumption that cells associated with the developing olfactory nerve were glial cells.) In our fate-mapping experiments, we used transgenic ^GFPchick as well as quail donors (21), allowing us to use both cell morphology and molecular markers (p75^NTR, P0) to identify OECs. After grafting the ANF from ^GFPchick donors to unlabeled host embryos, we saw only a very few ^GFPchick OECs at E10.25: these most likely reflected the presence of a small number of graft-derived NCCs, as ^GFPchick NCCs were occasionally seen in cartilage (e.g., the nasal septum), as well as in frontonasal mesenchyme. In contrast, our midbrain grafts showed that midbrain-derived NCCs abundantly form OECs, identified by their location (along the olfactory nerve from the lamina propria to the ONL), their expression of the OEC markers p75^NTR and/or P0, and their morphology (i.e., their processes ensheathed bundles of olfactory axons). Heterotopic grafts further showed that hindbrain- and spinal cord-derived NCCs can form OECs. We also identified Sox10 as a molecular marker for OECs: this is consistent both with the maintenance of Sox10 expression in all NCC-derived glia (23) and with the fact that Sox10 directly regulates the “myelin” P0 gene (24), an OEC marker in chick and rat (22, 25, 27). Our analysis of Wnt1^Cre;R26R^lacZ and Wnt1^Cre;R26R^YFP embryos, in which NCCs are permanently labeled genetically (32, 33), yielded results entirely consistent with our fate-mapping experiments in avian embryos. Hence, OECs are derived from the neural crest, not from the olfactory placodes.

This finding explains the similarities between OECs and Schwann cells (3, 9–11) and raises interesting questions about the control of NCC differentiation into OECs versus Schwann cells. It is also potentially of high clinical significance. Human midbrain/hindbrain NCC-derived stem cells [so-called skin-derived precursors, or SKPs (37)] can be isolated and expanded from scalp or facial skin or from the dermal papillae of isolated scalp/beard hair follicles (38–40): the latter can be directed in culture to form cells with at least some characteristics of Schwann cells (40). These important proof-of-principle experiments raise the exciting prospect (once the signals that promote adoption of an OEC fate by NCCs have been identified) of producing large, pure populations of patient-specific OECs for transplant-mediated spinal cord repair by expanding NCC-derived stem cells from the patient's scalp/hair follicles in culture and inducing them to form OECs. Overall, therefore, the finding that OECs are NCC-derived has important implications not only for our basic understanding of OEC biology and the formation of different glial cell types from NCCs, but also potentially for the future production of autologous OECs for spinal cord repair.

Materials and Methods

Fertilized wild-type chick (Gallus gallus domesticus) and quail (Coturnix coturnix japonica) eggs were obtained from commercial sources. Fertilized transgenic ^GFPchick eggs (21) were obtained from Helen Sang and Adrian Sherman at the Roslin Institute, Edinburgh, Scotland (funded by the BBSRC). Grafting procedures were performed as previously described (5, 41). Embryos were fixed in 4% paraformaldehyde or modified Carnoy's (60% ethanol, 11.1% formaldehyde, 10% glacial acetic acid) and embedded for cryo- or wax sectioning. Sections (5–10 μm) were processed for in situ hybridization and immunohistochemistry as previously described (42). Sox10 probes for chick and mouse were gifts of M. Bronner-Fraser (Caltech, Pasadena, CA) and J. Briscoe (National Institute of Medical Research, London), respectively. Primary antibodies used were the following: anti-GFP (rabbit; Invitrogen), FITC-conjugated anti-GFP (goat; Abcam), anti-neuronal β-III tubulin (TuJ1, mouse IgG2a; Invitrogen), anti-P0 [1E8, mouse IgG1, Developmental Studies Hybridoma Bank (DSHB), developed under the auspices of the National Institute of Child Health and Human Development and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA 52242], anti-chick p75^NTR and anti-mouse p75^NTR (both rabbit; gifts of L. Reichardt, University of California at San Francisco), anti-quail (QCPN, mouse IgG1; gift of M. Bronner-Fraser, Caltech, Pasadena, CA; also available from the DSHB), and anti-smooth muscle actin (mouse IgG2a; Sigma). Alexa⁴⁸⁸- and Alexa⁵⁶⁸-conjugated secondary antibodies and Alexa³⁵⁰-conjugated NeutrAvidin were obtained from Molecular Probes/Invitrogen and biotinylated goat anti-mouse secondary antibodies were obtained from Southern Biotech. Sections were analyzed using epifluorescence or confocal microscopy. LacZ staining of mouse embryos was performed according to standard protocols (43).