Navigating growth cones are exposed to multiple signals simultaneously and have to integrate competing cues into a coherent navigational response. Integration of guidance cues is traditionally thought to occur at the level of cytoskeletal dynamics. Drosophila studies indicate that cells exhibit a low level of continuous caspase protease activation, and that axon guidance cues can activate or suppress caspase activity. We base a model for axon guidance on these observations. By analogy with other systems in which caspase signaling has non-apoptotic functions, we propose that caspase signaling can either reinforce repulsion or negate attraction in response to external guidance cues by cleaving cytoskeletal proteins. Over the course of an entire trajectory, incorrectly navigating axons may pass the threshold for apoptosis and be eliminated, whereas axons making correct decisions will survive. These observations would also explain why neurotrophic factors can act as axon guidance cues and why axon guidance systems such as Slit/Robo signaling may act as tumor suppressors in cancer.

Navigating growth cones are exposed to multiple signals simultaneously and have to integrate competing cues into a coherent navigational response. Integration of guidance cues is traditionally thought to occur at the level of cytoskeletal dynamics. Drosophila studies indicate that cells exhibit a low level of continuous caspase protease activation, and that axon guidance cues can activate or suppress caspase activity. We base a model for axon guidance on these observations. By analogy with other systems in which caspase signaling has non-apoptotic functions, we propose that caspase signaling can either reinforce repulsion or negate attraction in response to external guidance cues by cleaving cytoskeletal proteins. Over the course of an entire trajectory, incorrectly navigating axons may pass the threshold for apoptosis and be eliminated, whereas axons making correct decisions will survive. These observations would also explain why neurotrophic factors can act as axon guidance cues and why axon guidance systems such as Slit/Robo signaling may act as tumor suppressors in cancer.

Slit is a secreted protein that has a canonical function of repelling growing axons from the CNS midline. The full-length Slit (Slit-FL) is cleaved into Slit-N and Slit-C fragments, which have potentially distinct functions via different receptors. Here we report that the BMP-1/Tolloid family metalloprotease, Tolkin (Tok), is responsible for Slit proteolysis in vivo and in vitro. In tok mutants lacking Slit cleavage, midline repulsion of axons occurs normally, confirming that Slit-FL is sufficient to repel axons. However, longitudinal axon guidance is highly disrupted in tok mutants and can be rescued by midline expression of Slit-N, suggesting that Slit is the primary substrate for Tok in the embryonic CNS. Transgenic restoration of Slit-N or Slit-C does repel axons in Slit-null animals. Slit-FL and Slit-N are both biologically active cues with distinct axon guidance functions in vivo. Slit signaling is used in diverse biological processes, thus differentiating between Slit-FL and Slit fragments will be essential for evaluating Slit function in broader contexts.