Developmental biology

The plastic face

In vertebrates, the face is formed in part by neural crest cells. It has been assumed that the developmental fate of these cells is inbuilt. New work, however, reveals a role for instructive signals from nearby cells.

Like many developmental processes, the formation of the face in vertebrates depends on the migration of cells from one part of the embryo to another. The main source of cells that contribute to the facial structures is the neural crest1 — a ridge of embryonic tissue generated by the developing brain. Different bones in the head are formed from different populations of neural crest cells, according to the cells' position along the head-to-tail (anterior–posterior) axis of the embryo. So, for example, anterior neural crest cells migrate to form the brain-case and the nose and jaw bones. Neural crest cells from the posterior part of the brain help to make the neck bones. But beyond these general details, little is known about how facial structures are induced to develop at the right place and in the appropriate, species-specific, shape. Such developmental instructions might be intrinsic to the neural crest and fixed before migration. Alternatively, neural crest cells might be instructed by surrounding tissues during or after migration.

An early, landmark study of chick embryos2 favoured the first of these models. But two new papers, published by Trainor et al.3 and Couly et al.4 in Science and Development, respectively, come down firmly in favour of the second. The two groups find that the neural crest cells that contribute to the face are developmentally 'plastic' — they do not themselves carry any information about their fate, but must be instructed by signals from other tissues to generate skeletal elements of appropriate shape and polarity. The authors also identify two sources of these signals.

During early vertebrate development, the neural crest forms from the neural tube — a long structure that will generate the brain and spinal cord, and which runs along the back of the embryo (Fig. 1a). The part of the neural tube that gives rise to the brain is divided into segments — the forebrain, midbrain and hindbrain; the hindbrain is in turn subdivided into smaller segments called rhombomeres. Neural crest cells generated from the forebrain, midbrain and the first two rhombomeres migrate to cover the brain, as well as into the forehead region and first pharyngeal arch, and contribute to the brain-case, nose and jaws. By contrast, cells from the remaining rhombomeres fill the second, third and fourth arches, which form the neck. (Pharyngeal arches are the embryonic structures from which most cranial features develop.)

Figure 1: Shaping the face.

a, Flattened view of the back of a chick embryo. Left, during early development, neural crest cells migrate out of the neural tube and towards the forehead and nasal region (topmost bulge) and the pharyngeal arches. For simplicity, migration is shown only on the left, although it would occur on both sides in the embryo. Right, spatial relationships between the neural crest and the underlying pharyngeal endoderm, which is divided into regions I–IV. Trainor et al.3 show that signals from the neural tube (such as from the isthmus) inform neural crest cells destined for the first pharyngeal arch about their position (for example, through repressing the Hoxa2 gene). However, the cells' fate is not irreversibly set before migration. Then, signals from the underlying endoderm (green stripes) instruct the cells about the size, shape and orientation of the skeletal elements they will form, as Couly et al.4 show. b, Grafting of an extra endodermal region II onto the original yields a fully duplicated lower beak and jaw, highlighting the importance of signals from the endoderm. Modified with permission from ref. 4. r, rhombomere.

Nearly two decades ago, Noden2 made a seminal discovery while investigating facial development. Using chick embryos, he replaced neural crest cells that were destined for the second pharyngeal arch with quail cells destined for the first arch. This resulted in the formation of extraneous jaw- and beak-like structures in the neck region — in other words, the grafted cells migrated to the nearest (second) arch, from which the neck forms, but gave rise to the morphology characteristic of the first arch. At face value, this implies that the fate of neural crest cells is set while they are still in the neural tube, determining which structures they help to form. But Noden pinpointed two potential contradictions. First, the grafted neural crest cells also contributed to normal second-arch-derived structures; and second, the extra skeletal elements invariably included jaw features but never forehead bones.

How can these results be reconciled? Trainor et al.3 provide a solution. These authors carried out similar grafting experiments, and found that extraneous jaw structures were produced only when the isthmus (the boundary between the midbrain and hindbrain) was included with the grafted cells. The isthmus is a source of signalling molecules, particularly fibroblast growth factor-8 (FGF8), and normally represses the Hoxa2 gene — the main determinant of the fate of second-arch cells5,6,7,8 — in the first rhombomere9. Trainor et al. found that when the isthmus was transplanted posteriorly into subsequent rhombomeres, it inhibited Hoxa2 in adjacent second-arch cells, which normally contribute to the neck but now develop as extra jaw bones. Adding FGF8 alone also inhibited Hoxa2 in neural crest cells. But when Trainor et al. moved just the neural crest cells that were destined for the first arch, without the FGF8-secreting isthmus, the cells took on a second-arch fate. In other words, they were not irreversibly committed to their fate before migrating to the arches.

So it seems that signals from the neural tube inform neural crest cells about their original position. But where does information about the shape and polarity of facial bones come from? Couly et al.4 provide evidence in favour of a tissue — the endoderm — that lies beneath the neural crest and lines the pharyngeal arches.

Using chick embryos again, these authors first excised all the neural crest cells that contribute to the face. They then showed that any short fragment of the region of the neural crest that does not express Hox genes can generate the entire facial skeleton. The findings again show that these cells are not endowed with intrinsic developmental information before migration. Couly et al. further discovered that distinct regions of the endoderm (regions I–IV; Fig. 1a) instruct Hox-negative neural crest cells to produce specific face and jaw bones. The removal of single regions of the endoderm prevented the formation of distinct subsets of facial neural-crest-derived elements. However, this might simply reflect the removal of a general signal needed for skeleton formation. More compellingly, individual stripes of endoderm grafted near their normal counterparts led to spectacular duplications of facial cartilage and adjacent bones (Fig. 1b), with orientations that depended on the orientation of the graft.

These results3,4 suggest that distinct populations of neural crest cells respond to regional cues from the neural tube (such as FGF8 from the isthmus). Such signals must also be maintained in the arch, possibly to ensure that the cells segregate and migrate correctly. But the cells remain uncommitted until further instructed by the pharyngeal endoderm, each region of which contains information that guides the size, shape and orientation of parts of the facial skeleton. It will be important to identify the factors that control regionalization of the endoderm, and the exact signals that instruct the neural crest. During vertebrate evolution, the distribution or timing of such factors or signals may have been changed so as to mould different facial morphologies.

Some past results may need revisiting in light of the new work. For example, grafting rhombomeres 1 and 2 plus the isthmus more posteriorly in chick embryos2,3 — leading to the inhibition of Hoxa2 — and knocking out the Hoxa2 gene in mice5,8 resulted in jaw structures in the neck region. Given Couly et al.'s discovery4 that the pharyngeal endoderm instructs neural crest development, this might mean that the endoderm of the second arch provides a signal like that from the first arch. But in that case it seems odd that posterior grafts of only the dorsal parts of rhombomeres 1 and 2, or of rhombomeres 1 and 2 without the isthmus, result in a largely normal second-arch morphology3,10 even in the continued absence of Hoxa2 in the neural crest cells10.

A related puzzle is why Hoxa2-expressing neural crest cells, induced to migrate into the first arch, are unable to generate any skeletal elements at all6,10. A clue might come from an experiment in which jaw-to-neck transformations could be produced only when Hoxa2 was artificially expressed in both the neural crest and the corresponding (first) arch6,7. Perhaps, to form the neck bones, Hoxa2 is required not only in neural crest cells but also in other arch components. Tissue-specific knockouts of Hoxa2 in mice, and further grafting experiments in chicks, may lead to the answers.


  1. 1

    Le Douarin, N. M. The Neural Crest (Cambridge Univ. Press, 1983).

  2. 2

    Noden, D. M. Dev. Biol. 96, 144–165 (1983).

    CAS  Article  Google Scholar 

  3. 3

    Trainor, P. A., Ariza-McNaughton, L. & Krumlauf, R. Science 295, 1288–1291 (2002).

    ADS  CAS  Article  Google Scholar 

  4. 4

    Couly, G., Creuzet, S., Bennaceur, S., Vincent, C. & Le Douarin, N. M. Development 129, 1061–1073 (2002).

    CAS  PubMed  Google Scholar 

  5. 5

    Gendron-Maguire, M., Mallo, M., Zhang, M. & Gridley, T. Cell 75, 1317–1331 (1993).

    CAS  Article  Google Scholar 

  6. 6

    Grammatopoulos, G. A., Bell, E., Toole, L., Lumsden, A. & Tucker, A. S. Development 127, 5355–5365 (2000).

    CAS  PubMed  Google Scholar 

  7. 7

    Pasqualetti, M., Ori, M., Nardi, I. & Rijli, F. M. Development 127, 5367–5378 (2000).

    CAS  PubMed  Google Scholar 

  8. 8

    Rijli, F. M. et al. Cell 75, 1333–1349 (1993).

    CAS  Article  Google Scholar 

  9. 9

    Irving, C. & Mason, I. Development 127, 177–186 (2000).

    CAS  PubMed  Google Scholar 

  10. 10

    Couly, G., Grapin-Botton, A., Coltey, P., Ruhin, B. & Le Douarin, N. M. Development 125, 3445–3459 (1998).

    CAS  PubMed  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Filippo M. Rijli.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Pasqualetti, M., Rijli, F. The plastic face. Nature 416, 493–494 (2002).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Sign up for the Nature Briefing newsletter for a daily update on COVID-19 science.
Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing