News & Views | Published:

Developmental biology

Holding it together in the eye

To form tissues, like cells must clump together. The striking resemblance between one cell aggregate in flies and a cluster of soap bubbles points to a crucial role for surface mechanics in biological pattern formation.

How informative are the relatively simple and general rules of physics in explaining the often bewildering complexity and specificity of biology? The large number of proteins and metabolites that work together to produce the specific shapes and functions of different cell types suggests that any apparent similarities between cells and simpler non-biological objects are unlikely to be more than coincidence. However, as Hayashi and Carthew1 report on page 647 of this issue, there is a remarkable analogy between the structures formed from a type of cell in the retina of fruitfly eyes and clusters of soap bubbles. This analogy illuminates the surface forces that are responsible for producing the specific cellular shapes required for biological function in the eye.

In a developing tissue or organism containing dozens or even thousands of different cell types, the formation of uniquely structured subsets of cells is a complex process. It can involve cell motility, cell adhesion, attractive and repulsive signalling, and other factors, all of which are controlled by their own network of biochemical reactions. However, even before very much was known about the molecular biology underlying any of these events, developmental biologists had noted that simple mixtures of cells become sorted out in a manner that is reminiscent of the behaviour of immiscible liquids2. These investigators suggested that similar phenomena, based on minimizing the surface or adhesive energies of cell clusters, might account for cell sorting in vivo. Furthermore, in vitro studies, which manipulated adhesion strength in systems containing small numbers of different cell types, indicated that concepts of surface energies can predict cell sorting in this context3. Now Hayashi and Carthew1 show that such predictions also succeed in vivo.

The insect eye (Fig. 1) is an attractive model for studying development, because it is a relatively small and simple organ, formed by a limited number of cell types. As structures emerge they can be observed by microscopy, and their formation can be manipulated by precise changes in protein structure, produced by mutations in the genes that code for them. A particularly well-defined and beautiful structure in the insect eye is the ommatidium — the single eye within the compound eye — which contains only around 20 cells. The subject of Hayashi and Carthew's study is a set of cone cells, of which there are normally four, but more or less in some mutant flies. These cells lie at the top of the ommatidium, above the light-detecting photoreceptor cells.

Figure 1: View of a fruitfly's eye (and more).


Inset, left, the retina, showing numerous ommatidia and, within each ommatidium, a cluster of four cone cells; right, a cluster of bubbles. The arrow illustrates the presence of tension on the bubble surface. (Inset images taken from ref. 1.)

Hayashi and Carthew first show that the cone cells configure themselves into a cluster that strikingly resembles a cluster of four soap bubbles (Fig. 1, inset). They then focus on a set of proteins called cadherins, which are responsible for the adherence between cells. They observe differences in the types of cadherins that are expressed by different cells in the ommatidium, noting that one cadherin (N-cadherin) is expressed along the interfaces between individual cone cells. They further find that damaging N-cadherin does not prevent the ommatidium from forming, but does disrupt the typical bubble-like structure of the cluster of cone cells.

Hayashi and Carthew also take advantage of a previously characterized mutant fly in which cell–cell contacts are normal, but the number of cone cells varies from three to six rather than being fixed at the usual four. The kinds of clusters that three to six soap bubbles can make are limited and well characterized, and Hayashi and Carthew find examples of each possible type of cluster in the eyes of these mutant flies. This remarkable correlation suggests that, like clusters of soap bubbles, cone-cell aggregates configure themselves in a way that intrinsically minimizes their overall surface energy.

The idea that aggregates of cells and clusters of soap bubbles take on shapes that minimize surface energy does not, of course, imply that the molecular details of cell adhesion bear much resemblance to those in soap bubbles. The proteins (such as N-cadherin) and phospholipids that hold one cell membrane to another, and thereby allow the cells to pack in an energy-minimizing way, are very different from the simple surfaces that partition a collection of bubbles.

Something that Hayashi and Carthew do not emphasize, but which can be clearly inferred from their images, is that cell volumes (as well as bubble volumes) remain relatively constant during adhesion — and this might influence the distribution of cell components. Adhesion leads to a surface tension on the non-adherent parts of the membranes4 that is evident in the rounded appearance of both bubbles and cells away from the zones of contact. According to Laplace's law, this tension is proportional to a hydrostatic pressure difference across the membrane or interface; but osmotic pressures in cells are comparatively high and resist any large variations in cell volume. So membranes are subject to sustained tensions (something that has only recently been documented), which seem to be linked to various cellular processes5. Membrane-channel proteins or cytoskeletal processes that are regulated by tension could therefore be activated locally and feed back to further influence the asymmetric distributions of proteins documented by Hayashi and Carthew.

A surprising aspect of how well the minimization of surface energy accounts for the structures of cone-cell aggregates is that, unlike bubbles, cells are filled not with fluid but with a viscoelastic protein skeleton, both within the cell interior and lining the cell surface. In addition, cells exert forces on each other and on their surrounding extracellular matrix. The internal mechanical properties of retinal cone cells have not been measured directly, but they are unlikely to be those of a simple liquid. And yet these mechanical properties and forces seem — on the long timescales of development — to have little or no influence on the shapes of cell aggregates. This result makes it clear that the laws of physics, which underlie all biological processes, are not always obscured by the dazzling molecular complexity of biology, and that structural similarities between tissues and their simpler inanimate counterparts can be informative about the ways in which biological patterns form.


  1. 1

    Hayashi, T. & Carthew, R. W. Nature 431, 647–652 (2004).

  2. 2

    Steinberg, M. S. & Poole, T. J. Phil. Trans. R. Soc. Lond. B 295, 451–460 (1981).

  3. 3

    Steinberg, M. S. & Takeichi, M. Proc. Natl Acad. Sci. USA 91, 206–209 (1994).

  4. 4

    Hategan, A., Law, R., Kahn, S. & Discher, D. E. Biophys. J. 85, 2746–2759 (2003).

  5. 5

    Raucher, D. & Sheetz, M. P. J. Cell Biol. 148, 127–136 (2000).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

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.