How do growing axons in the central nervous system navigate through the dense jungle of cells and processes that they encounter on the way to their targets? On page 540 of this issue, Bonkowsky et al.1 show that the choice of pathways for growth cones (the leading edges of growing axons) in the fruit fly Drosophila melanogaster is regulated by a receptor tyrosine kinase known as Derailed (Drl). These results further indicate that, rather than attracting growth cones to the right pathways, Drl causes them to be repelled from the wrong ones.
The array of axons in the embryonic Drosophila central nervous system has a ladder-like structure (Fig. 1). Anterior and posterior commissural tracts cross the midline in each body segment, and two longitudinal tracts extend the length of the embryo. Each of the roughly 300 neurons within a unit of this structure is thought to extend its axon along a genetically determined pathway. Most interneurons (neurons that synapse with other neurons) extend axons across the midline of the central nervous system, and attractive and repulsive factors that regulate this fundamental crossing decision have been identified2,3. But axon guidance at the midline involves more than just the decision whether or not to cross. Each commissure contains many distinct pathways, and the growth cone of a particular neuron always follows the same one. The unique sequence of navigational decisions made by its growth cone determines the complex and invariant shape of each axon in the central nervous system4.
Although we are far from understanding how complete axon trajectories (such as those in Fig. 1) are determined, the results of Bonkowsky et al.1 are an important step forward. These authors define how axons choose between the two major subdivisions of the crossing pathways — the anterior and posterior commissures. They show that Drl is normally expressed on axons that follow the anterior commissure, but not on those that take the posterior route1,5. Moreover, forced expression of Drl on specific axons that normally take the posterior commissure causes them to choose the anterior tract instead (Fig. 2).
Can Drl redirect any crossing axon into the anterior commissure? To address this question, Bonkowsky et al. simultaneously misexpressed the Commissureless (Comm) protein and Drl on a set of axons that normally never cross the midline (the thoracic Ap axons). Expression of Comm in neurons downregulates a repulsive signal from the midline that is transduced by the Roundabout (Robo) receptor. So, non-crossing axons that express Comm are diverted into pathways that cross the midline6. The authors found that when the thoracic Ap axons expressed only Comm, they crossed in the posterior commissure. But when they expressed Drl as well, they switched to an anterior pathway.
The authors next asked whether Drl regulates commissure choice by making the anterior commissure more attractive, or whether it does so by generating aversion to the posterior commissure. Their data indicate that it repels axons from the posterior commissure. First, when Bonkowsky et al. expressed Drl without Comm on non-crossing Ap axons that grow anteriorly within the longitudinal tract, the axons were not diverted into the anterior commissure. Instead, their growth cones stalled when they reached the edge of the posterior commissure, suggesting that this region had become repulsive to them.
Second, because Drl is a receptor tyrosine kinase, one would expect it to be activated through an interaction with an extracellular ligand. Although such a ligand has not been identified, Bonkowsky et al.1 nonetheless examined its probable distribution by incubating embryos with ‘tagged’ versions of the Drl extracellular domain, then visualizing these tags using immunohistochemistry or immunofluorescence. The tagged Drl fusion proteins bound to a dumbbell-shaped region of the central nervous system that includes the posterior commissure and adjacent domains within the longitudinal tracts (Fig. 2). These results are consistent with the idea that, during normal development, Drl signalling is activated when growth cones from axons destined for the anterior commissure contact the Drl ligand in the vicinity of the posterior commissure. This repels them from the posterior commissure, allowing them to choose anterior pathways instead.
The next steps will be to identify the ligand(s) for Drl and the signalling proteins downstream of Drl in the growth cone. A more global issue raised by these studies is how growth cones integrate the many attractive and repulsive influences that they meet while navigating through the embryonic landscape. For example, in Fig. 1 consider the growth cone of the black Drl-expressing axon in the anterior commissure. It is attracted to the midline by Netrins (and other factors), but is also influenced by repulsion factors. This repulsion is neutralized by Comm — to allow the growth cone to cross the midline — but is later re-activated to prevent the axon from returning to the midline after it has crossed2,3.
Although Drl repels this axon from the posterior commissure, other factors must attract its growth cone to the anterior pathway. (Drl is not essential for an axon to choose the anterior commissure, and most axons still follow their normal pathways1 in loss-of-function Drl mutants.) The tracts in the fly's central nervous system are composed of discrete bundles of axons that are pioneered by growth cones of early neurons. Each later axon then follows a pathway along specific bundles. So, the individual pathways across the midline probably correspond to axon bundles or groups of bundles within the commissures. Finally, then, the growth cone must choose among the axon bundles in the anterior commissure and switch its bundle preference at the midline (Fig. 1). Does a growth cone need to integrate all of these attractive and repulsive signals simultaneously? Or are the signals temporally segregated, so a growth cone responds to only one influence at any time?
The Drl repulsion mechanism may be used at other times and places during fly development. For example, mutations in the Drl (aka linotte) gene affect learning, and adult brain structures implicated in learning cross the midline abnormally in linotte flies7. The Drl protein is also expressed on body-wall muscles, and growing muscle fibres sometimes overshoot their normal attachment points in Drl mutants8. Both of these effects imply that motile cells cannot recognize ‘stop’ signals when Drl is not expressed. Finally, because Drl is homologous to the mammalian RYK tyrosine kinase5, and because other components of midline-crossing pathways are conserved between vertebrates and flies2,3, Drl mechanisms will also probably be involved in development of the vertebrate nervous system.
Bonkowsky, J. L., Yoshikawa, S., O'Keefe, D. D. & Thomas, J. B. Nature 402, 540–544 (1999).
Zinn, K. & Sun, Q. Cell 97, 1–4 (1999).
Van Vactor, D. & Flanagan, J. G. Neuron 22, 649–652 (1999).
Schmid, A., Chiba, A. & Doe, C. Q. Development 126, 4653–4689 (1999).
Callahan, C. A., Muralidhar, M. G., Lundgren, S. E., Scully, A. L. & Thomas, J. B. Nature 376, 171–174 (1995).
Kidd, T., Russell, C., Goodman, C. S. & Tear, G. Neuron 20, 25–33 (1998).
Simon, A. F., Boquet, I., Synguelakis, M. & Preat, T. Mech. Dev. 76, 45–55 (1998).
Callahan, C. A., Bonkowsky, J. L., Scully, A. L. & Thomas, J. B. Development 122, 2761–2767 (1996).
About this article
A two-phase growth strategy in cultured neuronal networks as reflected by the distribution of neurite branching angles
Journal of Neurobiology (2005)