Report

Parasympathetic ganglia derive from Schwann cell precursors

See allHide authors and affiliations

Science  04 Jul 2014:
Vol. 345, Issue 6192, pp. 87-90
DOI: 10.1126/science.1253286

Exploiting nervous paths already traveled

The parasympathetic nervous system helps regulate the functions of many tissues and organs, including the salivary glands and the esophagus. To do so, it needs to reach throughout the body, connecting central systems to peripheral ones. Dyachuk et al. and Espinosa-Medina et al. explored how these connections are established in mice (see the Perspective by Kalcheim and Rohrer). Progenitor cells that travel along with the developing nerves can give rise to both myelinforming Schwann cells and to parasympathetic neurons. That means the interacting nerves do not have to find each other. Instead, the beginnings of the connections are laid down as the nervous system develops.

Science, this issue p. 82, p. 87; see also p. 32

Abstract

Neural crest cells migrate extensively and give rise to most of the peripheral nervous system, including sympathetic, parasympathetic, enteric, and dorsal root ganglia. We studied how parasympathetic ganglia form close to visceral organs and what their precursors are. We find that many cranial nerve-associated crest cells coexpress the pan-autonomic determinant Paired-like homeodomain 2b (Phox2b) together with markers of Schwann cell precursors. Some give rise to Schwann cells after down-regulation of PHOX2b. Others form parasympathetic ganglia after being guided to the site of ganglion formation by the nerves that carry preganglionic fibers, a parsimonious way of wiring the pathway. Thus, cranial Schwann cell precursors are the source of parasympathetic neurons during normal development.

Parasympathetic ganglia form one of the three divisions of the peripheral autonomic nervous system. They are located deep in cephalic, thoracic, and abdominal tissues, close to or embedded in their target organs, on which they exert actions opposite to sympathetic ganglia. The systems together are essential for cardiovascular, respiratory, and digestive functions (1). Activity of parasympathetic neurons is modulated by preganglionic, or visceromotor, neurons, most of which are located in the hindbrain and project in cranial nerves (2). Like all other peripheral autonomic neurons, parasympathetic neurons derive from the neural crest (3). Migration pathways of sympathetic and enteric precursors toward the dorsal aorta or in the walls of the gut are well charted (4). We studied how parasympathetic precursors are dispatched in the embryo to arrive at sites far away from their origin. Genetic deletion of two branches of the facial nerve (nVII) eliminated their associated parasympathetic ganglia, suggesting a role for cranial nerves in parasympathetic gangliogenesis (5). We thus sought to analyze the role of cranial nerves in the development of other parasympathetic ganglia: the otic ganglion, which innervates the parotid gland, and the cardiac ganglia, which innervate the heart (6).

The proneural gene Neurog2 is expressed during the differentiation of viscerosensory neurons in the geniculate (gVII), petrosal (gIX), and nodose (gX) ganglia that project in the facial (nVII), glossopharyngeal (nIX), and vagus (nX) nerves, respectively (7). In Neurog2CreERT2/CreERT2 embryos [hereafter Neurog2 knockout (KO) (8)], cranial ganglia and nerves were affected to various extents: gVII was atrophic, and nVII reduced to its main branch (5) (Fig. 1, A and C and B and D); gIX was absent and so was nIX and its branch, Jacobson’s nerve (7) (Fig. 1, A and C and fig. S1); gX was merely atrophic (fig. S1), and nX was preserved (Fig. 1, A and C). Subsequently, the otic ganglion, which normally forms at the end of Jacobson’s nerve, was absent, whereas the cardiac ganglia, which normally lie on cardiac branches of nX, were present (Fig. 1, B and D). Neurog2 is never expressed in the otic or any other parasympathetic ganglion (fig. S2). Thus, deletion of the otic ganglion in Neurog2 KO is non–cell-autonomous and most likely results from the deletion of Jacobson’s nerve. To further test whether the cardiac ganglia depend on nX, we examined embryos lacking both Neurog2 and its paralog Neurog1. In Neurog1/Neurog2 double KO, gX was absent (fig. S3), but nX still formed (Fig. 1, A and E) (made solely of motor fibers) and so did the cardiac ganglia (Fig. 1, B and F). We thus attempted to destroy nX by deleting motor in addition to sensory fibers. Having observed that Cre-dependent diphteria toxin chain A action (9) was too slow for the required time frame, we adapted a technique originally developed in Caenorhabditis elegans (10), conditionally expressing a toxic allele of the sodium channel ASIC2a (11) under the promoter of Paired-like homeodomain 2a (Phox2a) (fig. S4). In Pgk::Cre; Phox2aASIC2a embryos (where Cre is expressed in the germ line), precursors of viscerosensory neurons should be killed in the epibranchial placodes and visceromotor neurons upon exit of the cell cycle, according to the expression schedule of Phox2a (12). As predicted, both types of neurons were absent in embryonic day 11.5 (E11.5) Pgk::Cre; Phox2aASIC2a pups (fig. S5). The main branch of nVII persisted (Fig. 1, A and G), most likely formed by the projections of somatosensory neurons that are located in the proximal part of gVII and do not express Phox2 genes (13); nIX and Jacobson’s nerve were missing (Fig. 1, A and G) and so was the otic ganglion (Fig. 1, B and H). As to nX, it was reduced to a vestigial ramus (Fig. 1, A and G), and cardiac ganglia failed to form (Fig. 1, B and H). The agenesis of parasympathetic ganglia in Pgk::Cre; Phox2aASIC2a embryos was non–cell-autonomous: When we recombined the Phox2aASIC2a locus selectively in premigratory neural crest with a Wnt1::Cre allele (14)—thus, in sympathetic and parasympathetic precursors—sympathetic ganglia were absent at E13.5 (fig. S6), but parasympathetic ganglia were unaffected (Fig. 1, B and H). They degenerated only one day later (fig. S6), in line with the fact that they start expressing Phox2a at E12.5, that is, 2 days later than sympathetic ganglia (15). Therefore, the absence of cardiac ganglia in Pgk::Cre; Phox2aStopASIC2a at E13.5 is most likely secondary to the atrophy of nX. In summary, the Neurog2 KO and Pgk::Cre; Phox2aASIC2a mutations each delete a different set of cranial nerves cell-autonomously, and the corresponding parasympathetic ganglia fail to form. We conclude that parasympathetic ganglia formation depends on the cranial nerves that innervate them.

Fig. 1 Dependency of parasympathetic ganglia on cranial nerves.

(A, C, E, and G) Views of whole-mount embryos at E11.5 stained for neurofilament and centered on cranial nerves VII, VIII, and IX (top) or IX, X, and XI (bottom) in wild-type (WT) embryos (A) and the indicated genotypes (C, E, and G). Dorsal is left, rostral up. (B, D, F, H) Immunohistochemistry for PHOX2b combined with in situ hybridization for peripherin at E13.5 on parasagittal sections through the head (top) or transverse sections through the thorax (bottom) in WT embryos (B) and the indicated genotypes (D, F, and H). In Neurog1 KO, only the somatosensory neurons of the vestibulo-acoustic (VIII) and trigeminal (T) ganglia are missing, with no incidence on parasympathetic ganglia. C, cardiac ganglia; CT, chorda tympani; ENS, enteric nervous system; G, gVII (geniculate ganglion); G*, atrophic gVII (corresponding to its somatic part); JN, Jacobson’s nerve; N, gX (nodose ganglion); O, otic ganglion; P, gIX (petrosal ganglion); yellow asterisk, vestigial nX.

This dependency suggests that the nerves form before the ganglia and then recruit the ganglionic cells. To gather evidence for this, we monitored the formation of Jacobson’s nerve and the behavior of neural crest cells in its vicinity by combined detection of βIII-tubulin (TUBB3), SOX10 [an early marker for neural derivatives of the neural crest (16)], and PHOX2b [a general determinant of autonomic ganglia (17)]. Between E10 and E11 (Fig. 2, A to C), in short sequence, Jacobson’s nerve emerged from the distal pole of gIX and projected toward its endpoint, close to gVII, accompanied by SOX10+ cells with elongated nuclei, which lagged behind the most advanced fibers (Fig. 2, A and B) and whose numbers progressively increased. The pan-autonomic determinant Phox2b was switched on along the nerve in a salt-and-pepper manner between E10.5 and E11 (Fig. 2, B and C) and was expressed by all SOX10+ nerve-associated cells at E11.5 (Fig. 2D). Between E12.5 and E13.5, the otic ganglion anlage, first visible as a buildup of cells around the distal end of the nerve (Fig. 2E), condensed and became demarcated from the nerve, which was concomitantly depleted of cells (Fig. 2, F and G). At the same time, Phox2b and Sox10 expression domains became mutually exclusive, PHOX2b weak or absent in most nerve-associated cells, and SOX10 weak in most PHOX2b+ ganglion cells. Nerve-associated and ganglion cells at this stage had most likely switched off Phox2b or Sox10, respectively, because all nerve-associated cells equally expressed both genes a day earlier (and see below). The observed sequence of events suggests that the SOX10+ cells, which then become SOX10+/PHOX2b+ double-positive, migrate along the nerve and reach in that way the site of ganglion formation, although a directed proliferation could also play a role. Accordingly, in Neurog2 KO, where Jacobson’s nerve is absent (Fig. 1), the stream of cells was absent (fig. S1). This scenario is not unique to the otic ganglion, because a similar sequence was evident for the sphenopalatine and cardiac ganglia (fig. S7).

Fig. 2 Stages in the formation of the otic ganglion.

(A to G) Detection of TUBB3, SOX10, and PHOX2b on Jacobson’s nerve (JN on the three-dimensional rendering of a whole-mount neurofilament stain of cranial nerves V, VII, IX, and X at E10.5 in the top left) and the otic ganglion that forms at its distal end. (A and B) The most distal SOX10+ cell (arrowheads) lags behind the most advanced fibers (asterisks). (E to G) A cluster of PHOX2b+/SOX10weak cells forms a third of the way along JN, presumably corresponding to a previously undescribed parasympathetic ganglion. GSPN, greater superficial petrosal nerve; other abbreviations as in Fig. 1. Scale bars indicate 50 μm in (A) and (B) and 100 μm for (C) to (G).

Schwann cell precursors, which later give rise to myelinating and nonmyelinating Schwann cells, occupy embryonic nerves from E11 (18). We therefore explored whether the nerve-associated SOX10+/PHOX2b+ cells might be Schwann cell precursors. Like Schwann cell precursors and like otic ganglion cells (fig. S8), PHOX2b+ Jacobson’s nerve–associated cells were derived from the neural crest, as evidenced by Wnt1::Cre lineage tracing (Fig. 3). They expressed markers of neural crest (FOXD3, SOX2, and p75) and Schwann cell precursors (ErbB3, Cadherin19, and the myelin protein PLP) (Fig. 3). In contrast, none of them detectably expressed neuronal markers, such as NeuN and TUBB3, or the proneural factors Neurog2 or Ascl1 (fig. S9). Ascl1 (but not Neurog2) was up-regulated in ganglion cells (fig. S9) in keeping with its later roles in parasympathetic ganglia (19). Thus, cranial nerve–associated parasympathetic precursors have all the hallmarks of Schwann cell precursors.

Fig. 3 Molecular signature of parasympathetic precursors.

Labeling of Jacobson’s nerve and associated cells at E11.5 with the indicated antibodies and probes. (Top left) Intersectional lineage tracing for a history of Phox2b expression and a neural crest origin in a Phox2b::FLPo; Wnt1::Cre; RC::Fela embryo. Green fluorescent protein (GFP) and a nuclear-localized lacZ (nlslacZ) are expressed from the RC::Fela allele (21), respectively, after the action of FLPo recombinase under the control of Phox2b (20) or after the combined actions of Phox2b::FLPo and Cre under the promoter of Wnt1 (14). The fibers that belong to viscerosensory and visceromotor neurons (PHOX2b+ but not derived from the neural crest) are GFP+ (red), whereas the cells along the nerve (PHOX2b+ and derived from the neural crest) are nlslacZ+ (green). (Bottom left) Anlage of the otic ganglion at E11.5. All other bottom images are enlargements of a boxed area or a single cell from the top images.

Because cranial nerve–associated cells have features of both autonomic and Schwann cell precursors, we tested whether they actually produced Schwann cells. In Phox2b::Cre; ROSAtdTomato embryos (13), as early as E13.5, Jacobson’s nerve was coated with SOX10+/PHOX2b cells that expressed tdTomato (Fig. 4A), confirming that PHOX2b+ nerve-associated cells subsequently switch it off. At E16.5, several cranial nerves contained numerous tdTomato+ cells intermingled with the fibers (Fig. 4B and fig. S10), most of them PHOX2b, an occasional one still expressing the gene (fig. S10). A different set of recombinase and reporter [Phox2b::FLPo (20) and RC::Fela (21)] (fig. S10) produced similar results. Many cranial nerve–associated PHOX2b tdTomato+ cells expressed the Schwann cell markers Oct6 at E16.5 and S100 at P7 (Fig. 4, C and D) (18), signs of their differentiation into Schwann cells. Thus, cranial Schwann cells derive in part from a contingent of nerve-associated cells with a history of Phox2b expression. Such cells, albeit sparser, were also associated with limb nerves (Fig. 4E and fig. S11), and, from E13.5 on, we found (in 9 out of 10 limbs) a small ganglion made of PHOX2b+ neurons on the median nerve of the forelimb (Fig. 4, E and F), showing that the dual Schwann cell and neuron fate, although mostly associated with cranial nerves, is not unique to them. The persistence in the adult and physiological meaning, if any, of this new ganglion and the nature of its preganglionic innervation—known to control the survival of parasympathetic neurons (22)—remain to be determined.

Fig. 4 Schwann cell fate of nerve-associated Phox2b+ cells.

Sections through various nerves revealing tdTomato (tdT) fluorescence triggered by a Phox2b::Cre and stained by DAPI (4′,6-diamidino-2-phenylindole) and antibodies as indicated. (A) SOX10+ cells associated with Jacobson’s nerve (yellow arrowhead) are tdT+, hence have expressed Phox2b but no longer do, whereas cells in the otic ganglion (O) have maintained PHOX2b expression; bottom images are enlargements of the cell marked in the top image. (B) Segment of nX immunostained for neurofilament (NF) at E16.5. (C) Section through a lingual nerve at E16.5. (D) Section of the greater superficial petrosal nerve at P7. Bottom images are the S100 signal in the cells marked in the top image. (E) Section through the median nerve of the upper limb at E13.5 showing cells with a history of Phox2b expression along the nerve and a cluster of Phox2b+ cells (inset, close up of anti-PHOX2b immunofluorescence) that are neuronal, based on the detection on serial sections of peripherin and PHOX2b or Tubb3 and PHOX2b (F). Scale bars are 30 μm for (A), 5 μm for (B), 10 μm for (C) and (D), and 60 μm for (E).

Thus, parasympathetic ganglion neurons are derivatives of multifated Schwann cell precursors, known to give rise in vivo to melanocytes (23) and endoneurial fibroblasts (24) in addition to Schwann cells. The neuronal potential of these precursors, reported in vitro or after back transplantation for sciatic nerve–derived cells (25, 26), is achieved during the normal development of cranial nerves. Accordingly, the numerous locations of parasympathetic ganglia throughout the head and trunk are specified by the pattern of cranial nerve outgrowths, a parsimonious way of hooking up pre- and postganglionic partners in development and, conceivably, of synchronizing their appearance in evolution.

Supplementary Materials

www.sciencemag.org/content/345/6192/87/suppl/DC1

Materials and Methods

Figs. S1 to S11

References (2733)

References and Notes

  1. Acknowledgments: We thank M. Lazdunski for the ASIC2aG430F clone; C. Birchmeier, T. Müller, and M. Wegner for antibodies; M.-C. Tiveron for the original discovery of the limb ganglion; V. Michel for preliminary experiments; the imaging facilities of IBENS and Institut de la Vision, B. Mathieu, and D. Godefroy for technical assistance; and C. Goridis for helpful discussions. This work was supported by the Agence Nationale de la Recherche (ANR-12-BSV4-0007-01) and Investissements d’Avenir (ANR-10-LABX-54 MEMO LIFE and ANR-11-IDEX-0001-02 PSL* Research University) (to J.-F.B.); Fédération pour la Recherche sur le Cerveau (to Institut de la Vision); and NIH grants R01 DK067826, R21 DA023643, and NIH P01 HD036379 (to S.D.). G.G.C. was funded by the Fondazione CARIPLO, Milan, Italy, and I.E.M. received a Ph.D. studentship.
View Abstract

Navigate This Article