Patterning has often taken a back seat to fate determination in cell biology studies. “We've spent buckets of time,” says Cagan, “finding all sorts of signaling molecules that explain cell fate, but not patterning. We can make kidney cells [in vitro], but how do you sort them into a glomerulus? What we really want to know is not what makes a lung cell, but what makes a lung.”
A hint came when others found that mutation of the Ig family member Roughest caused disorganized ommatidia in flies but did not affect cell fate. Differential adhesion had been shown previously to control patterning by minimizing surface free energy—highly adherent cells congregate in the center, excluding cells with lesser adhesion strengths. But the fly eye is much more elaborate; each ommatidium is surrounded and separated by a single layer of cells derived from a pool of interommatidial precursor cells (IPCs) that rearrange to form a hexagon-shaped lattice.
For this lattice, complementary cell types expressing different adhesion proteins were needed. IPCs expressed Roughest, while the ommatidia expressed its homologue, Hibris. Roughest bound more strongly to Hibris than to itself, thus creating competition between IPCs for contact with ommatidia and minimizing interactions between IPCs. The result is that “the IPCs stretch out between the ommatidia,” says Cagan. “The lowest free energy shape they can take is a hexagon.”
Heterophilic interactions led to the formation of lasting junctions between the two cell types. IPCs that lost out in the competition for contact with ommatidia died by apoptosis; Roughest expression level was a determining factor in survival versus death. An IPC that had extra Roughest often inhabited a niche normally occupied by two IPCs. Deletion of Roughest from an IPC, by contrast, increased that cell's chance of dying.
As mammalian homologues of Roughest are strongly expressed in the kidneys, Cagan suspects that they are also a driving force behind glomerulus formation. “The way you can get a complex refined structure,” he says, “is to use a two-piece system and then control the expression of those two pieces.”