The functioning of the nervous system is dependent on the intricate network of connections between nerve cells, or neurons. These connections form in the embryo when each developing neuron sends out a thin extension, the axon, which migrates through the embryonic environment to its target cells. Our major interest is to identify the molecules that guide axons to their targets in the vertebrate nervous system. We have been particularly interested in the possibility that axons may be guided by long-range cues—diffusible attractants, secreted by target cells, that attract the axons at a distance and diffusible repellents, secreted by nontarget cells, that create exclusion zones for the axons. Our aim is to identify these diffusible factors, to understand their contributions to axon guidance in vivo, and to determine the mechanisms through which they guide axons. We focus on two major systems.
Axon Guidance Within the Spinal Cord
Previous studies in the vertebrate spinal cord provided evidence that so-called commissural axons are guided to one of their intermediate targets, the floor plate of the spinal cord, in part by a diffusible attractant secreted by floor plate cells. Several years ago, we identified the attractant, a protein we named netrin-1. This protein is homologous to proteins with similar functions in attractive axon guidance in the roundworm Caenorhabditis elegans and the fruit fly Drosophila melanogaster. These netrin proteins are bifunctional, capable of attracting some axons and repelling other axons, apparently depending on the receptors made by the responsive neurons. Studies by several laboratories over the past several years in C. elegans, vertebrates, and Drosophila have implicated receptors of the DCC (deleted in colorectal cancer) family in mediating attractive effects of netrins, whereas studies in C. elegans implicated the UNC-5 protein in mediating repulsive actions of the netrin UNC-6 in that species.
In collaboration with Sue Ackermann (Jackson Laboratories), we have identified three vertebrate homologs of UNC-5, called UNC5H1–3, and showed that they are all netrin-binding proteins, consistent with the possibility that they might be repulsive netrin receptors in vertebrates. We and our collaborators in Mu-Ming Poo's laboratory (University of California, Berkeley) obtained support for this hypothesis by showing that expression of vertebrate UNC5 proteins in Xenopus spinal neurons can convert their attractive responses to netrin-1 (which are mediated by DCC) into repulsive responses. We found further that the conversion of attraction to repulsion involves the formation of a complex of UNC-5 proteins and DCC, and direct interaction of the cytoplasmic domains of these receptors. Current efforts are aimed at understanding the guidance roles of these UNC-5 homologs in vivo and at elucidating the signal transduction pathways through which these homologs and DCC mediate their guidance effects.
When commissural axons reach the midline of the nervous system, they cross the midline, then turn to grow alongside it, but never recross it. In Drosophila, studies in the laboratory of Corey Goodman (University of California, Berkeley) have implicated the receptor Roundabout (Robo) and the ligand Slit as a key receptor-ligand pair involved in preventing axonal recrossing of the midline. In collaboration with Goodman's group, we identified two vertebrate homologs of Robo and three homologs of Slit and showed that at least one of these Slit proteins, Slit-2, similarly functions in axonal repulsion in vertebrates. Similar results on repulsive actions of vertebrate Slit proteins have been obtained by Yi Rao's laboratory (Washington University).
Current studies are aimed at elucidating the roles of Slit and Robo proteins in guidance of various neuronal populations in vertebrates, particularly commissural neurons. In addition, we have evidence that various Semaphorin proteins contribute to guiding commissural axons at the midline, through a repulsive mechanism that involves a receptor complex of neuropilin-2 (as a ligand-binding moiety) and a plexin protein. We are attempting to identify the mechanisms through which the neuropilin-plexin complex induces a repulsive response in growth cones.
Finally, we have focused on how commissural axons can change their responsiveness to midline cells so efficiently, such that they recognize the midline as attractive until they reach it, then perceive it as repulsive (which helps them move along their trajectory). We found that one mechanism that contributes to this change in responsiveness is that the Robo receptor, when activated by its ligand Slit, has a dominant inhibitory action on the netrin receptor DCC, preventing DCC from transducing an attractive signal through the formation of a complex of Robo and DCC receptors. This silencing effect helps explain how the growth cones can move on from an environment that was once attractive to them.
Axon Guidance Around the Spinal Cord
To determine whether guidance by diffusible factors is a widespread mechanism and to identify other diffusible guidance molecules, we initiated an analysis of sensory and motor axon guidance, testing whether the targets of these axons can influence the axons' growth at a distance in vitro. This led us to characterize several long-range interactions. First, we found that two intermediate targets of motor axons, the sclerotome and the limb mesenchyme, secrete factors that promote the outgrowth of these axons in vitro at a distance, and we identified the limb mesenchyme–derived factor as hepatocyte growth factor/scatter factor, a ligand for the c-Met receptor tyrosine kinase. Second, we recently identified a chemoattractant activity for trigeminal sensory axons, previously named Max factor, as a combination of two neurotrophins, NT-3 and brain-derived neurotrophic factor.
In addition, we have studied the mechanisms through which sensory axons branch into the gray matter of the spinal cord. We discovered an activity in spinal cord and brain extracts that stimulates the elongation and branching of sensory axons in vitro. Remarkably, purification of the active component from neonatal brain extracts identified an amino-terminal fragment of Slit-2 as the branch-inducing activity. Thus Slit-2 appears to be bifunctional, capable of repelling some axons and of inducing branching of other axons. We are attempting to define more precisely the roles of various Slit proteins in axonal branching through genetic loss-of-function studies in mice and to define the mechanisms through which Slit proteins regulate branching. We have also initiated a collaboration with Cornelia Bargmann (HHMI, University of California, San Francisco) to identify genes involved in controlling axonal branching in C. elegans, with the aim of applying this knowledge to the vertebrate nervous system as well.
We are initiating studies to examine whether molecules involved in guiding axons in the embryo also play a role in regulating axon growth and guidance in the adult nervous system under conditions of injury (such as spinal cord injury) and plasticity. We have also initiated a large-scale genetic screen in mice in collaboration with William Skarnes (University of California, Berkeley) to identify additional genes involved in axon guidance in vertebrates.
Grants from the National Institutes of Health, the Spinal Cord Research Foundation, the International Spinal Research Trust, the American Paralysis Association, and the Human Frontier Science Program provided support for aspects of the work on the netrins, the Semaphorins, and the peripheral projections of sensory axons.