The olfactory systems of insects and vertebrates are remarkably similar anatomically, so the fruit fly Drosophila melanogaster is a valuable model for investigating the underlying organizational principles. As reported in Development, Jefferis and colleagues have used the Drosophila system to examine how a spatial map is established in an olfactory relay centre in the brain.

In insects and vertebrates, axons from olfactory receptor neurons (ORNs) that express the same olfactory receptor converge on structures known as glomeruli. In Drosophila, these structures reside in the antennal lobe. The ORN axons synapse with the dendrites of projection neurons (PNs), which relay information to higher olfactory centres in the brain. The PN dendrites segregate, according to birth order and lineage, into different glomeruli.

How is the glomerular map established during development? Previous findings favoured the idea that the antennal lobe is initially patterned by ORN axons, which then act as targets for the incoming PN dendrites. However, by labelling clones of PNs and tracking the early development of their dendrites, Jefferis et al. showed that the PN dendrites in fact segregate into rudimentary patterns resembling future glomeruli before the ORN axons arrive at the antennal lobe.

What is the source of the signals that drive the segregation of PN dendrites? One candidate that the authors considered was the larval olfactory system, which shows a similar glomerular organization to the adult system. However, contact between the adult and larval antennal lobes is minimal during the stage when PN dendrites are invading the adult lobe, so it seems unlikely that residual larval neurons impose a pattern on these dendrites.

The glomeruli also contain the processes of glia and interneurons, and Jefferis et al. investigated whether these processes could provide positional information. No glial processes were detectable in the antennal lobe while the prototypic glomerular map was emerging, although glia could still exert a patterning influence outside the lobe. The involvement of interneurons is uncertain, as there are no reliable reagents to detect them. The authors suggest that segregation is at least partly driven by interactions among the PN dendrites themselves; dendrites with the same molecular signature might be induced to cluster through mutual attraction.

So, these findings indicate that PN dendrites establish the initial glomerular map in the antennal lobe. Although this represents a significant shift from previous models that emphasized the role of ORNs, it is not incompatible with the idea that ORN axons are also prepatterned. Also, the map is probably refined through interactions between ORN axons and PN dendrites. The next challenge will be to identify the signals that pattern these two sets of neuronal projections.