Main

The enteric nervous system (ENS) includes thousands of interconnected ganglia embedded within the wall of the gut4,5, making it the most complex portion of the peripheral nervous system in amniotes. The ENS of jawed vertebrates innervates the entire gastrointestinal tract to regulate muscle contraction, chemosensation, water balance, and gut secretion6. Classic transplantation experiments have demonstrated that the neurons and glia of the gut are largely derived from the ‘vagal’ population of neural crest cells that arise within the post-otic portion of the hindbrain7,8. They subsequently migrate to the foregut and embark upon the longest migration of any embryonic cell type, from foregut to hindgut.

The sea lamprey is a jawless (agnathan) vertebrate and an experimentally tractable representative of the cyclostomes. As the sister group to jawed vertebrates (gnathostomes), lampreys are an important model for identifying traits common throughout vertebrates. Lampreys possess migrating neural crest cells that give rise to many cell types, including melanocytes, cartilage, sensory neurons, and glia, but lack other neural-crest-derived structures that are present in gnathostomes such as jaws and sympathetic chain ganglia9,10. Given that lamprey embryos lack sympathetic ganglia, we sought to examine other components of the autonomic nervous system, with a focus on the ENS. Adult lamprey have a simple ENS that includes a ganglionated plexus of serotonin (5-HT)-producing cells, as well as a smaller number of catecholamine-containing neurons11,12. However, the developmental origin of these enteric neurons is unknown.

As a first step in analysis of lamprey ENS development, we examined the time course of neuronal appearance along the developing embryonic gut by performing in situ hybridization for Phox2b, which is expressed in enteric neurons in many jawed vertebrates5,9. Expression of lamprey Phox2b is detectable along the gut from early stages (Tahara stage 25 (T25)9), and by stage T28 or embryonic day 20 (E20), Phox2b-expressing cells are associated with a depression in the gut that will become the typhlosole, a haematopoietic tissue associated with elements of the ENS (Fig. 1a, b). Using a 5-HT antibody, serotonin-immunoreactive neurons were first observed within the gut wall at ~T28.5 (Fig. 1c, d) in the anterior trunk region. With time, the numbers increased, and 5-HT+ neurons were noted progressively posteriorly, with particularly high cell numbers in the cloacal region, as reported previously13. By T30 (E30), there are >100 neurons along the gut in association with the typhlosole and vagus nerve (Fig. 1e and Extended Data Fig. 1). Interestingly, individual serotonergic cells were also often associated with axon bundles emanating ventrally from the dorsal root ganglia (Fig. 2f and Extended Data Fig. 1a). In addition to 5-HT+ neurons, we also noted 5-HT+ columnar cells that may have represented gut enterochromaffin cells (Extended Data Fig. 1a, b).

Figure 1: Early formation of enteric neurons in the lamprey P. marinus.
figure 1

a, Phox2b expression in dorsolateral cells of a T28 embryo, with anterior to left and dorsal up. Letter ‘b’ marks the site of the cross-section in b. b, Transverse cross-section of a T28 embryo shows Phox2b expression in the depression of the typhlosole (black arrowhead). NT, neural tube; Not., notochord. c, 5-HT and acetylated tubulin immunoreactivity in a slightly older T28.5 embryo. d, 5-HT is detectable in neurons (white arrowheads) adjacent to the vagus nerve (white arrow). e, Serotonergic neurons form small ganglia within the enteric plexus of a T30 embryo. Scale bar, 50 μm.

PowerPoint slide

Full size image
Figure 2: DiI-labelled cells in the neural tube contribute to enteric ganglia.
figure 2

Immunohistochemistry of T30 lamprey transverse cryosections. a, Uninjected control highlighting neuronal projections (white arrowhead) and serotonergic neurons (white arrows) around the gut. Green, acetylated tubulin; blue, DAPI; cyan, 5-HT. b–d, DiI-injected ‘tubefill’ embryos. Red, DiI; green, acetylated tubulin; cyan, 5-HT; blue, DAPI. b, c, DiI (yellow arrows) within the neural tube, dorsal root ganglia, and along axon bundles. d, A DiI-labelled enteric neuron (acetylated tubulin+, 5-HT+; yellow arrowhead) within the gut wall typhlosole. Inset shows DiI only. e, A DiI-labelled neuron (acetylated tubulin+, 5-HT+) in the oesophagus (yellow arrowhead). See cell counts in Table 1. f, 5-HT+ cells along dorsal root (DR) nerve processes but not ventral roots (VR). g, A DiI-labelled GFAP+ cell associated with the lateral line. Scale bar, 20 μm.

PowerPoint slide

Full size image

We next sought to determine the embryological origin of the neurons in the gut by performing lineage labelling with the lipophilic dye DiI14. In chicken, vagal neural crest cells that contribute to the ENS arise from the hindbrain neural tube adjacent to somites 1–7 (ref. 7). After exiting the neural tube, they migrate ventrally, invade the foregut, and undergo a collective cell migration along the rostrocaudal extent of the gut. To test whether lamprey possess a homologous cell population, we performed focal injections of DiI into the dorsal portion of the caudal hindbrain (corresponding to the site of origin of vagal neural crest in gnathostomes) of T20 (E6) embryos. Regardless of the exact injection site, dye-labelled cells spread within the hindbrain and emigrated from the neural tube as a stream directly above the pharynx (Extended Data Fig. 2a–c). From this site, they progressed ventrally, then turned caudally to populate all of the branchial arches (Extended Data Fig. 2g), similar to previously reported migration patterns of cranial neural crest cells15,16. However, despite many focal injections (n = 46) into the caudal hindbrain homologous to ‘vagal’, we failed to find evidence of DiI-labelled cells in the gut. We next looked for expression of the vagal neural crest marker, Ret, in lamprey, since it is required for vagal neural crest development in gnathostomes17,18,19. To this end, we cloned a lamprey Ret homologue and its co-receptor Gfra1 and examined their expression patterns by in situ hybridization. The results showed lamprey Ret and Gfra1 are expressed in many parts of the embryo, including the typhlosole. Ret and Gfra1 expression were first observed in a region anterior to the gut at T26 (Extended Data Fig. 2i, j) in a manner reminiscent of Phox2b expression (Extended Data Fig. 2k). However, we did not observe expression in neural crest cells migrating from caudal hindbrain towards the gut, raising the intriguing possibility that enteric neurons might arise from a different cellular source.

In gnathostomes, migrating trunk neural crest cells fail to invade the immediately underlying gut owing to the presence of repulsive signals such as Slit that block their entry20. Recently, however, a secondary contribution to the mammalian ENS has been uncovered that comes not from vagal neural crest but rather from trunk neural-crest-derived Schwann cell precursors (SCPs) associated with extrinsic nerves that contribute postnatally to calretinin-containing neurons of the gut3. Using Dhh-cre mice to lineage trace immature Schwann cells, it was found that ~20% of colonic neurons in mice were derived from trunk SCPs3. Intriguingly, this population persists in Ret-null mice3. To examine the possibility that a homologous population may exist in lamprey, we asked whether cells emerging from the trunk neural tube might contribute to neurons of the gut. To test this, we performed focal DiI injections into the dorsal trunk neural tube as well as neural tube lumen injections following its cavitation at T22–T23 (~E8), thus labelling the entire neural tube as well as presumptive neural crest cells before their emigration. Both labelling techniques yielded similar results. By T25 (E14), such injections label neural-crest-derived cells in dorsal root ganglia and mesenchyme cells of the fin9. Interestingly, at later stages, we observed DiI-labelled cell bodies closely associated with nerve processes, recognized by acetylated tubulin staining above the gut (Fig. 2a–c). These individual DiI-labelled cells were visible as early as T28 (E18–E20) and persisted until the latest stages examined, T30 (E30–E35), by which time the yolk was reduced, facilitating imaging. In the trunk of lamprey, there are two nerve roots per trunk segment: a dorsal root that contains fibres emanating from neural-crest-derived dorsal root ganglia and a ventral root that contains motor-neuron nerve fibres. Interestingly, DiI-labelled cells were associated with these dorsal roots (Fig. 2c). Similarly, serotonergic presumptive neuroblasts were selectively associated with and appeared to migrate only along these dorsal root bundles (Fig. 2f), consistent with a neural crest origin.

To establish whether the DiI-labelled cells differentiated into enteric neurons, embryos were sectioned and stained with antibodies to serotonin and acetylated tubulin as a mature neuronal marker. The results, based on examination of transverse sections through 25 representative embryos at T30 (E30–E35) in which we quantitated cells that were positive for acetylated tubulin and serotonin, are summarized in Table 1. We noted numerous DiI-labelled serotonergic neurons (DiI/5-HT/acetylated-tubulin-positive cells) in the anterior gut (Fig. 2d), oesophagus (Fig. 2e), typhlosole, and adjacent tissues, as well as other DiI+ neurons that were serotonin-negative (Extended Data Fig. 1c). These results demonstrate that neural crest cells originating from the trunk neural tube can contribute to enteric neuron populations within the gut wall.

Table 1 Number of DiI-labelled neurons (AcTub+) in 25 representative E30–E35 embryos in which DiI had been injected into the neural tube
Full size table

To determine whether markers associated with neural-crest-derived SCPs were present in the lamprey trunk, we stained embryos with an antibody to glial fibrillary acidic protein (GFAP), which labels glial and Schwann cells21. We observed prominent GFAP+ cells in lamprey embryos, including some that were DiI-positive along the lateral line (Fig. 2g), consistent with the possibility that neural crest cells give rise to the non-myelinating Schwann cells of lampreys22.

Finally, as an additional means of testing whether enteric neurons originate from within the trunk neural tube, we performed neural tube ablation (n = 17), a method classically used to demonstrate the neural crest origin of enteric neurons8,23, at approximately T24. To accommodate slight natural variation in the number of ENS neurons between embryos, we expressed counts as the ratio of serotonergic cells in the zone adjacent to the ablation versus an identically sized region spaced approximately one somite length to the anterior. Compared with control non-ablated (n = 6) or sham-ablated (n = 4) embryos (Fig. 3a, c), experimental embryos showed a significant decrease in the ratio of ENS serotonergic cells (Fig. 3b, d). We noted an ~12% decrease of serotonergic neurons in the ablated region compared with sham-ablated embryos and an ~25% decrease relative to stage controls, consistent with a model of ENS neuronal precursors migrating from trunk neural-crest-derived cells. This decrease was particularly surprising given the remarkable regenerative capacity of the lamprey spinal cord24,25.

Figure 3: Neural tube ablation disrupts ENS development.
figure 3

a–c, Whole-mount images of T30 lamprey embryos (red, HuC/D and acetylated tubulin; cyan, 5-HT). Ant., anterior; Post., posterior. a, Control non-ablated (non-abl.) embryo. b, Ablation of the neural tube (neural abl.) at T24 results in a decrease of serotonergic (5-HT+) cells in the adjacent gut relative to an equivalently sized anterior region. c, A sham-ablated (sham-abl.) embryo, in which epidermis and neural tube were cut but not removed, does not show a reduction of cells. d, Ablated embryos show a significantly lower ratio of serotonergic cells (number of cells in surgical zone/number of cells in anterior zone) than sham-ablated controls (t-test; P = 0.0004) and non-ablated batch controls (t-test, P = 0.0176). Error bars, s.e.m. e, Schematic model of neural crest contributions to the ENS in mouse compared with lamprey. V, vagal neural crest (purple); SCP, blue; lamprey caudal hindbrain crest, purple. Scale bar, 100 μm.

PowerPoint slide

Source data

Full size image

Recent evidence suggests that many neural crest derivatives in postnatal mammals, including skin and peripheral ganglia, arise from SCPs that are closely associated with extrinsic innervation to these structures. For example, SCPs along nerve processes differentiate into melanocytes of the skin and parasympathetic ganglia1,2,26. Moreover, SCPs that migrate along trunk spinal nerves contribute to a subpopulation of enteric neurons in mice3. These studies prove the existence of neural-crest-derived cells that contribute to the peripheral nervous system and other derivatives at post-embryonic stages. Our results suggest that these trunk neural-crest-derived cell types may represent an ancient and evolutionarily conserved source of cells that contribute to the ENS. Moreover, our data suggest that agnathans might lack a classic ‘vagal’ neural crest, leading us to speculate that a vagal neural crest population with the ability to form enteric neurons arose27 in stem gnathostomes (Fig. 3e). We cannot rule out contributions from other sources, but focus here on the positive contribution of trunk neural-crest-derived cells to enteric neurons. Although we cannot formally exclude the possibility that the mechanisms for populating the ENS arose independently in agnathans and gnathostomes, we favour the idea that the contribution of SCPs to the ENS might represent a primitive (plesiomorphic) state retained from early vertebrates, perhaps common to all living vertebrates. With the emergence of jawed vertebrates, new traits, including jaws, sympathetic ganglia, and vagal neural-crest-derived enteric ganglia, appeared under the umbrella of embryonic neural crest derivatives.

Methods

No statistical methods were used to predetermine sample size. Adult sea lamprey (P. marinus) were supplied by the US Fish and Wildlife Service and Department of the Interior, and cultured according to previous protocols28 in compliance with California Institute of Technology Institutional Animal Care and Use Committee protocol 1436. Embryo batches with less than 70% survival were excluded from analyses, and all experimental embryos were randomly chosen from appropriately staged cultures. Celltracker CM-DiI (Thermo Scientific) was resuspended as described9, and surgical ablations were performed using forceps and tungsten needles. For surgeries, only embryos with ablations of two to six somite lengths, adjacent to the anterior intestine and without damage to other tissues, were included in analyses. Serotonergic neurons associated with the typhlosole were counted in a 200–300 μm region adjacent to the ablation site, and an equivalently sized non-ablated region beginning approximately one myotome length anterior to the ablation. Measurements were taken in sibling non-ablated embryos and sham-ablated embryos in which ectoderm and neural tissue was cut, but not removed, at similar axial locations. Investigators were not blinded to group allocations. Probes for Phox2, Gfra1, and Ret were cloned from complementary DNA with primers for Ensembl gene models ENSPMAG00000008433, ENSPMAT00000009324, and ENSPMAT00000008763. In situ hybridizations and antibody stainings were performed using previously described protocols29,30. Anti-acetylated tubulin (Sigma clone 6-11b-1, T7541; Mouse IgG2b) and anti-5-HT (Immunostar 20080, Rabbit IgG) were used at 1:500. Anti-GFAP (Dako, Z0334; Rabbit IgG) was used at 1:400. Embryos were processed for cryosectioning according to standard protocols, and were sectioned on a Microm HM550 cryostat. Microscopy was performed on a Zeiss AxioImager.M2 equipped with Apotome.2, and fluorescent z-stack images were processed and presented as maximum intensity projections. Images were cropped, rotated, and resized using Adobe Photoshop CC, and image panels were constructed using Adobe Illustrator CC. Statistics were analysed in Microsoft Excel.

Data availability

The datasets generated during and/or analysed during the current study are available from the corresponding author upon reasonable request.