Much recent research into axon guidance has focused on the roles of the Roundabout (Robo) receptors and their ligand, Slit, in the developing Drosophila nervous system. Just over a year ago, it was reported that in addition to preventing growth cones from re-crossing the midline, these molecules also control how far axons are repelled away from the midline after crossing. Several years ago, vertebrate homologues of Slit and Robo were identified, and in vitro studies implicated these in the control of axonal repulsion and branching, but their precise functions in vivo have not been described. A series of studies published in Neuron now provide some interesting new insights.

Hutson and Chien investigated the role of the Robo2 homologue Astray in axon guidance at the zebrafish optic chiasm. They showed that Astray is required both to prevent crossing errors and to correct those that do occur. Even in the background of the bel mutation, in which axons turn back sharply before reaching the chiasm and project ipsilaterally, axons are misrouted if Astray is inactivated. So, unlike the Drosophila Robos, Astray acts on both sides of the midline.

In a separate study, Plump et al. showed that in Slit1 ; Slit2 double-knockout mice, an ectopic optic chiasm forms anterior to the true chiasm. Slit1 and Slit2 are expressed in complementary domains that line the route taken by retinal ganglion cell (RGC) axons in the ventral diencephalon, so they might channel the RGC axons to the correct crossing point by creating a corridor with repulsive 'walls'. If growth cones stray too close to the Slit expression domains, they are repelled back into their tract. Intriguingly, the ectopic chiasm always developed in the same position in the Slit1;Slit2 mutants, indicating that other cues also restrict RGC axon movement.

Another study by Bagri et al. in the same laboratory revealed that Slit proteins are also important for axon guidance in other forebrain projections, including the corticofugal, cortico-cortical (callosal) and thalamocortical tracts. The Slit1 single mutant mouse had no obvious axon guidance defects, but in the Slit2 mutant, some axons in these tracts took an abnormally ventral course. In Slit1;Slit2 mutants, the phenotype was even more severe, with some growth cones veering towards the midline. This indicates that one function of the Slit proteins is to maintain a dorsal trajectory for certain forebrain projections. The effect was not confined to projections that originate in the forebrain — 5-hydroxytryptaminergic projections from the raphe nuclei and dopaminergic projections from the substantia nigra/ventral tegmental area were also misrouted in the Slit2 and Slit1;Slit2 mutants. There is circumstantial evidence that the Slits act through Robo receptors in the forebrain, as Robo1 and Robo2 are expressed in the cortical plate and dorsal thalamus at the time that projections are emerging from these regions.

These results indicate that the functions of Slits and Robos in axonal repulsion have been conserved during evolution. One key difference between flies and vertebrates is that the vertebrate axons examined in these studies seem to rely on Slit and Robo throughout their journey, whereas in Drosophila, the Slit/Robo interactions only come into play once the growth cones have crossed the midline. Previous in vitro studies showed that spinal commissural axons in vertebrates, like their Drosophila counterparts, change their sensitivity to Slits on crossing the midline. Triple mutants of Slit1, -2 and -3, all of which are expressed at the spinal cord midline, will be required to determine whether this holds true in vivo.