Embryo and organ development is driven in large part by the sculpting of sheets of cells called epithelia. Epithelial sheets have a remarkable repertoire of behaviours — they can narrow and elongate in plane, change thickness, and bend out of plane to form structures such as pits and furrows that develop into the body’s tubes. However, our understanding of the 3D shapes of the cells that make up these structures, and of how they change during development, has been dominated by inferences often made from looking only at the most easily imaged surface of an epithelial-cell sheet. Writing in Nature Communications, Gómez-Gálvez et al.1 use a modelling approach to predict how cells are packed to form tubes, and the authors give a name to a type of epithelial-cell shape involved in this packing, which they also find occurs in vivo.
Cells in single-layered epithelia have approximately parallel tops (apices) and bases. The cells are therefore usually thought of as prisms that join together in a honeycomb-like arrangement (Fig. 1a). When either the apices or the bases of cells constrict in all directions (isotropic constriction), wedge-shaped cells called frusta emerge (Fig. 1b), resulting in the formation of a pit2 or furrow3.
During the elongation of fruit-fly embryos, cell–cell junctions at the apices4 or bases5 of epithelial sheets can also contract in just one orientation in the plane of the sheet. This anisotropic constriction can lead to the formation of different contacts between neighbouring cells at the apical end of the cell from contacts at the basal end. Such constriction produces cells that are neither prisms nor frusta — but the shape that they form has had no name.
Gómez-Gálvez et al. analysed 3D cell shapes in epithelia by using computer modelling to study how cells pack into curved epithelia. For epithelia on a sphere, surface curvature is the same in all directions, and inverted frusta are expected. But the authors realized that, when epithelia form tubes and so curve in only one orientation, or form egg shapes that also have differential surface curvature, the cells’ inner and outer surfaces can contact different neighbours. This change in contacts requires the formation of a triangular face on one side of the cell (Fig. 1c) — a 3D shape similar to that seen in the epithelial sheets of elongating fruit-fly embryos. The researchers name this 3D cell shape scutoid, from the Latin word for shield.
So why are scutoids predicted to occur in tubes? It is possible for 3D shapes resembling frusta to form tubes. But this requires the cells’ apical or basal surfaces to be elongated, and Gómez-Gálvez et al. based their modelling on the plausible assumption that apical and basal surfaces of epithelial cells both tend towards isotropic shapes as the most energy-efficient (‘relaxed’) packing solution to differential curvature. This is in contrast to the sheets of fruit-fly embryos, in which the formation of scutoids is a dynamic, temporary process driven by active cell rearrangement5. When the authors modelled cells that had near-isotropic shapes on both the inner and outer surfaces of tubes, they found topological differences between the inner and outer surfaces that required the cells to be scutoid. The authors then confirmed this finding using a mechanical model of the outer surface, in which the lengths of cell–cell interfaces were minimized, encouraging isotropic shapes.
Gómez-Gálvez and colleagues went on to search for scutoids in vivo. They found scutoid-shaped cells in snapshots of various embryonic epithelia, although, interestingly, not at the frequency predicted by their modelling. For example, the larval salivary glands of fruit flies have fewer scutoids than are predicted by the curvature of the gland tube. Furthermore, the authors found scutoids in the spherically curved surface layers of zebrafish embryos where none would be predicted.
There are several possible reasons for these discrepancies. First, scutoid predictions will be altered if they take into account other forces, either generated within the epithelium (for example, caused by anisotropic contractility) or acting on it. Second, in the current model, the authors used 2D modelling of the inner and outer surfaces of cells to infer 3D cell shapes — predictions might differ if modelling was extended to cell shapes that are explicitly 3D. However, a better understanding of the mechanics of the shared interfaces on the sides of cells will be needed before cell shapes more complex than prisms and frusta can be modelled accurately in three dimensions6.
Third, the assumption that apical and basal cell shapes will tend towards isotropy is based on the assumption that the cells’ cortex — a contractile, mesh-like network of proteins that gives cells their structure — acts as a ring at the cell–cell junctions at the ends of cells. But the cortex also spans the rest of the exposed apical and basal surfaces, and the contraction of this medial web seems not to tend to isotropic shapes in the same way3. Finally, scutoid formation might be highly dynamic in some epithelia, or might involve more-convoluted 3D shapes than those considered here, rendering measurement of scutoid frequency difficult.
Prisms, frusta and scutoids found in single-cell layered epithelia are discrete classes of shape, but belong in a continuum of cell shape and 3D arrangement. This continuum has recently been quantified as an additive combination of cell wedging and interleaving2,7. Wedging describes how cell shapes become wider or narrower with depth, whereas interleaving describes changes in how interdigitated a group of cells is with depth, equivalent to a rate of change of cell arrangement from bases to apices. Both wedging and interleaving can be quantified in units of radius of curvature, so are ideal for characterizing curved epithelia. A formal relationship between the degree of interleaving and the frequency of scutoids seems possible, which would make these continuous and discrete measures usefully interchangeable.
With Gómez-Gálvez and colleagues’ characterization of the scutoid, we are beginning to recognize the kinds of 3D shapes and arrangements to look for in epithelia, and to develop the tools for quantifying them. Where else in nature should we expect scutoids? I would not bet against scutoids being found in plants, given the diversity of plant architectures. But because plant development is driven by cell growth and division without rearrangement, any mechanism of scutoid formation is likely to be rather different from that seen in animals. Nest comb construction by bees or wasps might also result in scutoids, particularly on curved surfaces, as in the nests of some paper wasps8. We shall have to wait and see.
Nature 561, 182-183 (2018)