The Neural Crest: Cells That Leave Home

A child with Hirschsprung disease has a stretch of colon with no nerve cells in its wall. A person with Waardenburg syndrome may have a white forelock, eyes of two colours and hearing loss. Some patients have both conditions, from a mutation in a single gene.

Why would gut neurons, hair pigment and hearing fail together? Because the gut’s neurons and the pigment cells, called melanocytes, of skin, hair and inner ear share an ancestor: the neural crest. This lesson follows those cells from the edge of the neural plate, along their routes, into ganglia and into the clinic.

Neural crest, the wanderers: a border population that leaves home

In that lesson you saw the ectoderm, the embryo’s outer cell layer, split into neural plate and future skin. The neural crest forms at the border between the two. BMP signaling at an intermediate level helps define that border (see that lesson), and Wnt signals are needed to make crest there (see that lesson). When the plate rolls up into the neural tube, described in that lesson, the border cells end up along the dorsal midline, the top of the tube.

Then they loosen their grip on their neighbours, lose their top-to-bottom polarity, break through the basement membrane and crawl away. That change, from a cell in an epithelium to a free mesenchymal cell, is an epithelial-to-mesenchymal transition, or EMT. Timing differs: chick trunk crest leaves after the tube has closed, while in the mouse head many cells leave while the folds are still open.

What they become depends partly on where along the body they start:

Axial level	Main derivatives
Cranial	Cartilage and bone of the face and much of the skull, connective tissue, some cranial sensory and parasympathetic neurons, glia, melanocytes
Vagal and sacral	Neurons and glia of the gut wall (the enteric nervous system)
Cardiac (part of the vagal level)	Tissue that divides the heart’s outflow into aorta and pulmonary trunk
Trunk	Dorsal root ganglion sensory neurons, sympathetic neurons, adrenal medulla, Schwann cells, melanocytes

In birds and mammals, trunk crest does not normally make skeleton; crest bone and cartilage are a head speciality. Some authors call the crest a “fourth germ layer”. That is a popular label, not a formal term: the crest arises from ectoderm.

Neural crest tools and history: how do you follow a cell that leaves?

To know where a migrating cell came from, you have to mark it before it goes, and the mark has to last.

The question of what crest makes was opened by Julia Platt, working on the mudpuppy Necturus, a salamander with large, yolky embryonic cells. Reading serial sections of embryos fixed at successive stages, she followed cells from the neural crest and thickened head ectoderm into the branchial arches. In 1893 she concluded that ectoderm contributes to the cartilages of the head.

That broke a rule: cartilage was supposed to be mesoderm. Her evidence was also indirect, from fixed sections with no living cell labelled or moved. The conclusion was long doubted. Grafting and removal experiments in amphibians gave it experimental support in later decades, notably the work of Sven Hörstadius and Sven Sellman in the 1940s.

Classic experiment: quail-chick chimeras

The lasting mark came from Nicole Le Douarin. She noticed that quail nuclei, stained for DNA, show a large dense mass of condensed chromatin beside the nucleolus, which chick nuclei lack. Graft a short piece of quail neural tube into a chick embryo at the same level, and the quail cells that emigrate can be recognised in sections long afterwards. The mark neither fades nor dilutes as cells divide.

With Marie-Aimée Teillet and others, Le Douarin used chimeras to map the crest origin of sensory, sympathetic and enteric ganglia, the adrenal medulla and the skeleton of the face. The neurons of the bird gut, for example, come from crest at the vagal level of the neck, with a contribution from the lumbosacral level to the hindgut. The modern equivalent is genetic: a Cre line is a mouse strain that makes the enzyme Cre recombinase in a chosen cell type, and Cre switches on a reporter gene permanently in those cells and all their descendants.

A graft carries many cells, so it shows where a population goes, not whether each cell could make everything the population makes. Two single-cell experiments asked that.

In a dish, Anne Baroffio, Elisabeth Dupin and Le Douarin plated single migrating crest cells from the quail midbrain and anterior hindbrain region onto a feeder layer of 3T3 cells in 1988. Of 243 clones they analysed after 7 to 10 days, some were committed to a single neuronal type, others could form several neuron types, non-neuronal cells and melanocytes, and many had intermediate potential. In the embryo, Marianne Bronner-Fraser and Scott Fraser injected a fluorescent dextran into single premigratory cells in the chick trunk neural tube, also in 1988. Some single clones spread into more than one derivative, and some included cells that stayed in the neural tube.

So some crest cells can make several cell types. Whether the typical premigratory crest cell is broadly multipotent in the living embryo is still debated: later single-cell tracing in chick and mouse has produced both multipotent and restricted clones. Culture shows what a cell can make under one condition; lineage tracing shows what it did make.

Leaving home: an identity first, then a release

Before a crest cell can leave, it has to become a crest cell. In 2004 Daniel Meulemans and Marianne Bronner-Fraser set out a proposed gene-regulatory network: a diagram of which genes switch on which. Border signals (BMP, Wnt, FGF) turn on border genes such as Msx, Pax3/7 and Zic. These turn on crest genes such as Snail2, FoxD3, Sox9 and Sox10, which in turn control the genes for adhesion, motility and later fate. Each arrow was a hypothesis; in chick cranial crest, for example, the Bronner laboratory later showed that Sox9, Ets1 and cMyb act directly on an enhancer of Sox10.

The release needs adhesion to change. Premigratory chick crest makes Cadherin-6B, a cadherin that holds cells together; a subset of migrating crest cells switches to Cadherin-7. In 1994 a group including Angela Nieto showed that blocking the transcription factor Slug, now called Snail2, reduced crest emigration in chick embryos. In 2007 Lisa Taneyhill, Edward Coles and Bronner-Fraser linked the two. They lowered Snail2 with a morpholino, a synthetic molecule that blocks the making of one protein. Cadherin-6B RNA rose within about half an hour. Snail2 bound clusters of E boxes, short DNA motifs, in the Cadherin-6B regulatory region, and the authors concluded that it represses the gene directly. In a companion study, Coles, Taneyhill and Bronner-Fraser found that lowering Cadherin-6B made crest cells emigrate early, so Cadherin-6B acts as a brake on departure. Snail2 probably promotes departure at least partly by lifting that brake, and EMT involves more than one adhesion gene.

Neural crest migration patterns: repulsion, collisions and a gathering

In the chick trunk, crest cells take two main routes. The ventromedial route runs through the somites, the blocks of mesoderm beside the neural tube, and gives dorsal root ganglia, sympathetic ganglia and adrenal medulla. The dorsolateral route, taken later, runs under the skin and gives mostly melanocytes. On the ventral route the cells pass only through the rostral (front) half of each sclerotome, the part of the somite that will form vertebrae. That is why dorsal root ganglia are segmented.

The caudal half keeps them out. Crest cells carry EphB receptors, and caudal-half sclerotome cells carry ephrin-B, a membrane-bound ligand. In 1997 two groups, Catherine Krull and colleagues working with chick found that trunk crest plated on alternating stripes avoids ephrin-B1 stripes, and Hai Wang and David Anderson showed that clustered ephrin-B ligands repel crest cells and motor axons. Ephrin-B can also signal back into its own cell (see that lesson), but here the forward signal into the crest cell repels.

Design problem

Trunk crest cells in culture avoid stripes coated with ephrin-B. A sceptic says the stripe is simply a poor surface to stick to, and the cells drift onto the better one. How would you show that the avoidance is active repulsion through Eph receptors?

One way to do it

Add soluble ephrin-B to the medium, or a reagent that blocks Eph receptors. Soluble ligand engages the receptors everywhere, so the cell can no longer read the stripe as different. If cells now cross the stripes and spread evenly, the avoidance needed Eph-ephrin signaling. That is what the experiment showed: soluble ephrin-B in the medium blocked the avoidance. An adhesion problem would not be relieved by adding ligand, because the stripe surface itself is unchanged. As a control, add a soluble protein that does not bind Eph receptors; avoidance should persist.

Crest cells also steer each other. In contact inhibition of locomotion, first described by Michael Abercrombie in cultured fibroblasts in the 1950s, two cells that touch stop moving towards each other and head off in new directions. In 2008 Roberto Mayor’s laboratory showed that frog and zebrafish crest cells do this. At the contact, components of planar cell polarity signaling, a Wnt branch that acts without β-catenin, activate the GTPase RhoA, which collapses protrusions there, while Rac drives new protrusions at the free edge (see that lesson). Cells at the edge of the group move outward. Add a mutual attraction that keeps the group together, and chemotaxis towards the chemokine Sdf1, which in Xenopus a cluster follows better than a single cell, and you get collective cell migration.

Then the travellers have to stop and gather. This is “The Coalescence Problem”: no central planner tells cells where to form a ganglion. The “Three Ingredients for Clustering” are adhesion molecules such as N-cadherin and integrins, which stabilise a cluster once cells arrive; chemoattractants such as SDF1, which bring them there; and boundaries such as Eph-ephrin and semaphorin signals, which divide the embryo into zones. The three act together, not in sequence. Ask yourself: “If you remove N-cadherin from neural crest cells, what happens to dorsal root ganglia formation?” Commit to an answer. If adhesion stabilises rather than initiates, what should you see?

Neurocristopathies: one failure at each step

Robert Bolande named disorders of neural crest cells neurocristopathies in 1974. For each, ask which step failed.

Step	What fails	Example
Specification at the border	Border gene PAX3	Waardenburg syndrome type 1: pigment changes and deafness
Survival of early crest	Ribosome biogenesis gene TCOF1; in mice, cell death reduces crest	Treacher Collins syndrome: small jaw and cheekbones
EMT	Few human disorders trace to this step alone; the clearest evidence is from chick
Migration and colonisation	RET, EDNRB, EDN3, SOX10: crest fails to colonise the distal gut	Hirschsprung disease
Differentiation	Melanocytes arrive but cannot make melanin (tyrosinase)	Albinism

Albinism sits at the very end of the chain: the cells are present and only their product is missing. Piebaldism, with patches that lack melanocytes altogether, is the migration or survival version. Cleft palate is harder to place: the palatal shelves are built largely of crest-derived mesenchyme that must grow, lift and fuse, and many genes and environmental factors can make that fail. Genes such as SOX10 and EDNRB, needed by both enteric and melanocyte lineages, explain the patient in the opening.