Diffusible chemorepellents play a major role in guiding developing axons toward their correct targets by preventing them from entering or steering them away from certain regions. Genetic studies in Drosophila have revealed a novel repulsive guidance system that prevents inappropriate axons from crossing the CNS midline; this repulsive system is mediated by the Roundabout (Robo) receptor and its secreted ligand Slit. In rodents, Robo and Slit are expressed in the spinal cord and Slit can repel spinal motor axons in vitro. Here, these findings are extended into higher brain centers by showing that Robo1 and Robo2, as well as Slit1 and Slit2, are often expressed in complementary patterns in the developing forebrain. Human Slit2 can repel olfactory and hippocampal axons and collapse their growth cones (Nguyen Ba-Charvet, 1999).

To determine the potential role of Robo and Slit family members in mediating axon pathfinding in the forebrain, the expression patterns of Robo1, Robo2, Slit1, and Slit2 mRNAs in the developing telencephalon were examined at the time olfactory and hippocampal projections are forming, near the end of the second week of gestation. In the rat olfactory system (the expression pattern in the mouse is identical), fibers begin to leave the olfactory bulb (OB) by E14, and by E15 the lateral olfactory tract is clearly formed. OB axons project ipsilaterally, never cross the midline, and avoid the septal area. In vitro, the septum has been shown to repel olfactory tract axons. As early as E12, Slit2, but not Slit1, is highly expressed in the midline of the telencephalon, including the region of the presumptive septum. By E14 and continuing until at least E18, both Slit1 and Slit2 are expressed by the septum. At these stages, mitral cells and tufted cells of the OB express high levels of Robo2 mRNA, while Robo1 is almost undetectable. During this time period, Slit1 is also found in a subset of mitral cells, but Slit2 is completely absent from the OB. Thus, their expression patterns in the olfactory system are complementary (Nguyen Ba-Charvet, 1999).

Tracing studies in mouse hippocampus have shown that hippocampal afferents invade their target territories in a highly specific fashion. Such stereotyped, directed growth suggests the involvement of long-range and short-range guidance cues, and previous studies have focused on the role of Semaphorins in shaping this trajectory. To explore potential contributions of Slit and Robo proteins in patterning hippocampal connections, their expression patterns in this region were examined. The earliest entorhinal axons leave the entorhinal cortex by E14 and reach the hippocampus by E15, where they first project onto neurons in the CA1 and CA3 subfields and later in the dentate gyrus (DG). In rodents, most of the development of the DG occurs postnatally; however, a DG primordium can clearly be observed by E16 in the mouse (Nguyen Ba-Charvet, 1999).

In the rat, Robo and Slit mRNAs are highly expressed in the developing hippocampal formation. From E12-E13, Slit2 can be detected in the so-called 'cortical hem', a neuroepithelial structure that forms a boundary between the hippocampus, the most medial part of the cerebral cortex, and the telencephalic choroid plexus. This expression persists until at least E15, when it can also be observed in the entorhinal cortex. At E18, Slit2 mRNA is found in the DG, in a portion of the CA3 subfield, and in the entorhinal cortex but is not present in the neocortex. Slit2 is also detectable in the choroid plexuses and in cells lining the telencephalic ventricles. Neither Slit1, Robo1, nor Robo2 can be observed in the hippocampal formation until E14, and by E18 Slit1 is found at the level of the hippocampal plate, in the presumptive pyramidal cells of the Ammon horn, and in most of the other hippocampal-related structures, with the exception of the DG. In contrast to Slit2, Slit1 expression is strong in the neocortex, where it is largely restricted to the cortical plate. By P3, Slit1 is expressed almost exclusively in CA3 and in the subiculum, while Slit2 can still be detected throughout the hippocampus proper and presubiculum. At E20, both Robo1 and Robo2 are expressed in the hippocampus in relatively similar structures, with Robo1 at a higher level, but only Robo1 is found in the dentate granule cell layer. Interestingly, Robo2 is also expressed by Cajal-Retzius cells, which have been shown to have a role in guiding the entorhinal cortex projections. At this stage, both Robo genes are expressed in the entorhinal cortex and neocortex, but they remain confined to distinct and complementary layers: Robo1 mRNA is found in neurons of the cortical plate and marginal zone, while Robo2 is localized to the intermediate zone. By P3, the expression patterns of Robo1 and Robo2 have changed, such that Robo1 expression has now disappeared from the subiculum and Robo2 expression has increased in the subiculum. In addition, at P3 both genes are also expressed in the Ammon horn pyramidal cell layer and in the hilus. Moreover, Robo1 expression is maintained at a low level in the DG granule cell layer, and Robo2 expression persists in Cajal-Retzius cells. Thus, although there is some overlap, Slit and Robo expression patterns in the developing hippocampus are to a large extent complementary (Nguyen Ba-Charvet, 1999).

When cultured in a three-dimensional collagen gel matrix, E14 rat OB axons can be repelled by a diffusible factor released by the septum. At that stage, mitral cells in the OB express Robo2, while its ligand Slit2 and putative ligand Slit1 are found in the septum and telencephalic midline. Since it is known that COS cells secreting hSlit2 can repel spinal motor axons, a test was performed to see whether Slit2 can also act as a repulsive molecule in the olfactory system by culturing E14-E15 OB explants with aggregates of hSlit2-expressing cells. When cultured either directly adjacent to or at a distance (up to 200 µm) from control COS cells, axons extend from all OB explants in a radial pattern. In contrast, axons from 98% of OB explants cultured at a distance from COS cells secreting hSlit2 are repelled. In addition, in the distal quadrant, both the number of OB neurite bundles and the area covered by OB neurites are significantly larger than in the proximal quadrant, facing hSlit2-expressing cells. As early as 18-24 hr in culture, an asymmetric pattern of outgrowth can be observed, with far fewer axons in the quadrant proximal to the cell aggregate. This repulsive effect lasts until at least 36-48 hr in culture. At these later time points, only a few axons can be observed in the proximal quadrant, but most axons are clearly directed away from the hSlit2-expressing cells. Explants were not cultured for more than 48 hr, because transfected COS cells start to die after this point. These results indicate that hSlit2 is a chemorepellent for OB axons (Nguyen Ba-Charvet, 1999).

It was next examined whether Slit proteins are chemorepellents for axons of the hippocampal formation. During development, hippocampal axons never invade the adjacent entorhinal cortex, whereas axons from the entorhinal cortex project massively to the DG. Axons from the DG, CA3, and CA1 subfields of the embryonic hippocampus can be repelled by the entorhinal cortex. Cells secreting either Semaphorin III or IV can mimic this activity. Since Slit2 mRNA and Robo1 and Robo2 mRNAs are expressed at high levels in the entorhinal cortex and in the DG, respectively, attempts were made to determine whether Slit2 could repel DG axons. DG axons grow symmetrically in 82% of the cases when confronted with control COS cells. In contrast, in 93% of the explants cultured with COS cells expressing hSlit2, axons preferentially grow away from the cell aggregates. Repulsion could be observed after 24 hr and is maintained for one more day in vitro. In the distal quadrant, both the number of DG neurite bundles and the area covered by DG neurites are significantly larger than in the proximal quadrant, facing hSlit2-expressing cells. This demonstrates that Slit2 is also a chemorepellent molecule for DG axons. It will be interesting to study Slit1 function using similar assays, but Slit1 expression constructs are not available yet (Nguyen Ba-Charvet, 1999).

The afferent axonal projections of the rodent hippocampus are very precisely organized, with subsets of axons arborizing in specific layers and synapsing on specific portions of the dendritic trees of their target neurons. These afferent axons project directly to their proper target layers, suggesting that their final targeting involves layer-specific positional cues. In rodents, the DG consists of three layers: the molecular layer, the granule cell layer, and the hilus or polymorphic layer. The principal cells of the DG are the granule cells, but other neuronal subtypes, such as pyramidal basket cells, stellate cells, and mossy cells, can be found scattered thought the molecular layer and hilus. The major afferents to the DG arise from the entorhinal cortex and innervate the superficial two-thirds of the molecular layer via the perforant pathway, beginning at E19 in the mouse. The inner third of the molecular layer is innervated postnatally (around P2) by commissural/associational axons originating from the mossy cells from the hilus. Interestingly, both Robo1 and Robo2, as well as their candidate ligands, Slit1 and Slit2, are expressed in somewhat complementary patterns in the embryonic hippocampus when these hippocampal connections are being formed. In addition, the entorhinal cortex secretes a diffusible repellent for axons of the DG, and hSlit2-expressing cells can mimic this activity. Granule cell axons, the mossy fibers, project to the CA3 subfield of the hippocampus proper, where they are restricted to the proximal segment of the pyramidal cell stem dendrites, while entorhinal axons occupy their distal segment. Therefore, Slit2 produced by entorhinal axons could block the invasion of the upper dendritic segment by mossy fibers. Robust expression of Slit2 mRNA was also found in CA3 pyramidal neurons. A heterogeneous distribution of Slit2 on the surface of pyramidal cell dendrites could also participate in the dendritic segregation of mossy fiber afferents. In contrast with other hippocampal projection neurons in CA1 and CA3, mossy fibers never cross the telencephalic midline, which expresses high levels of Slit2 mRNA from very early stages. These results suggests that Slit2 may have a similar function in pushing noncommissural axons away from the midline of the forebrain, as has been proposed for Robo and Slit at the Drosophila ventral midline (Nguyen Ba-Charvet, 1999 and references).

Semaphorins function as chemorepellents when presented chronically to growth cones but induce growth cone collapse when presented acutely. Since hSlit2 can repel axons of a variety of classes of neurons, it was next examined whether hSlit2, like Semaphorins, could induce growth cone collapse. Membranes from hSlit2-expressing cells were used. Explants from OB or DG were cultured on a poly-L-lysine/laminin substrate. After a day in culture, individual axons extending from these explants could be observed. Membrane extracts prepared from hSlit2-expressing COS cells or control COS cells were applied to the OB and DG cultures for 1-2 hr. Cultures were fixed and collapsed growth cones, identified by a lack of lamellipodia and filopodia, were quantified in blinded experiments. Four independent experiments gave similar results. Application of hSlit2 membrane extracts to OB growth cones results in a marked and significant increase in the percentages of axons with collapsed growth cones, as compared with addition of control membranes. hSlit2 membranes can also induce collapse of 58% of DG growth cones, although in this case the difference between hSlit2 and control membranes is less significant than for OB growth cones. Western blots of membrane extracts reveal that the hSlit2-containing membrane extracts consist primarily of the full-length and N-terminal cleavage fragments, with most, if not all, of the C-terminal cleavage fragment presumably being lost during treatment. These results suggest that the collapse-inducing activity is contained in the full-length or N-terminal fragments (Nguyen Ba-Charvet, 1999).

The olfactory bulb plays a central role in olfactory information processing through its connections with both peripheral and cortical structures. Axons projecting from the olfactory bulb to the telencephalon are guided by a repulsive activity in the septum. The molecular nature of the repellent is not known. The isolation is reported of vertebrate homologs of the Drosophila slit gene. Slit protein is shown to bind to the transmembrane protein Roundabout (Robo). In chick embryos, Slit expression is detectable first in Hensen's node at stage 4. From stage 5 to stage 8, it is expressed in Hensen's node, the notochord, the prechordal plate, and the paraxial mesoderm. At stage 6, Slit expression is in the ventral midline of the neural tube, including the floor plate. Slit expression in the roof plate proceeds spatiotemporally in a rostral to caudal order: Slit is expressed in the roof plate at the prechordal level in a stage 10 embryo, but not yet at the spinal cord level in the same embryo. Slit expression extends to the most rostral ends of both the ventral and dorsal midlines of the neural tube. Slit expression in the motoneuron columns can be detected in stage 17 embryos. Slit disappears from the mesodermal midline so that, by stage 21, strong expression at the spinal cord level can be detected in the floor plate, the motor neurons, and the roof plate, but not in the notochord. Slit expression is not only in the neural tube, but also in the somites (the dermomyotome), the amacrine cells in the retina, and the limb bud. These results suggest that Slit may function in many regions of the embryo. Slit is expressed in the septum whereas Robo is expressed in the olfactory bulb. Functionally, Slit acts as a chemorepellent for olfactory bulb axons. These results establish a ligand-receptor relationship between two molecules important for neural development; suggest a role for Slit in olfactory bulb axon guidance, and reveal the existence of a new family of axon guidance molecules (Li, 1999).

In addition to the olfactory bulb axons, another system in which Slit functioning is related to the midline is the projection of commissural axons in the spinal cord. The expression of Slit in the floor plate and its binding to Robo support a role for Slit as a repellent at the floor plate for commissural axons. One function of Slit in the spinal cord would be to prevent commissural axons that have already crossed the floor plate from recrossing it. The expression level of Robo protein determines the response of commissural axons to Slit, so that commissural axons would respond to Slit after, but not before, the axons have crossed the floor plate. The expression of Slit in motoneurons suggests another possible function: this is to turn the circumferentially growing commissural axons into the longitudinal direction. Although longitudinal turning is a well-known phenomenon, its molecular mechanism is not clear. It is tempting to speculate that the presence of Slit in the floor plate and in the motor column would work together to force commissural axons that have crossed the midline to turn longitudinally. Thus, the expression pattern and repellent activity of Slit suggest that it is possible for a single molecule to play two important roles: to prevent commissural axons from recrossing the floor plate and to turn these axons longitudinally. The mechanism for rostral-caudal guidance of the longitudinal axons remains unknown at the present. It can not be readily explained by Slit functioning alone because no rostrocaudal difference of Slit mRNA distribution in the floor plate and the motor neurons has been detected by in situ hybridization in vertebrate embryos, suggesting possible involvement of other molecules for rostrocaudal guidance. Expression of Slit outside the spinal cord, such as the retina and the limb bud, suggests that Slit functioning is not limited to the midline of the central nervous system (Li, 1999).

Olfactory sensory neurons (OSNs) expressing a given odorant receptor project their axons to specific glomeruli, creating a topographic odor map in the olfactory bulb (OB). The mechanisms underlying axonal pathfinding of OSNs to their precise targets are not fully understood. This study demonstrates that Robo2/Slit signaling functions to guide nascent olfactory axons to the OB primordium in zebrafish. robo2 is transiently expressed in the olfactory placode during the initial phase of olfactory axon pathfinding. In the robo2 mutant, astray (ast), early growing olfactory axons misroute ventromedially or posteriorly, and often penetrate into the diencephalon without reaching the OB primordium. Four zebrafish Slit homologs are expressed in regions adjacent to the olfactory axon trajectory, consistent with their role as repulsive ligands for Robo2. Masking of endogenous Slit gradients by ubiquitous misexpression of Slit2 in transgenic fish causes posterior pathfinding errors that resemble the ast phenotype. The spatial arrangement of glomeruli in OB is perturbed in ast adults, suggesting an essential role for the initial olfactory axon scaffold in determining a topographic glomerular map. These data provide functional evidence for Robo2/Slit signaling in the establishment of olfactory neural circuitry in zebrafish (Miyashaka, 2005).

The floor plate is known to be a source of repellent signals for cranial motor axons, preventing them from crossing the midline of the hindbrain. However, it is unknown which molecules mediate this effect in vivo. Slit and Robo proteins are candidate motor axon guidance molecules, since Robo proteins are expressed by cranial motoneurons, and Slit proteins are expressed by the tissues that delimit motor axon trajectories, i.e. the floor plate and the rhombic lip. In vitro evidence is presented showing that Slit1 and Slit2 proteins are selective inhibitors and repellents for dorsally projecting, but not for ventrally projecting, cranial motor axons. Analysis of mice deficient in Slit and Robo function shows that cranial motor axons aberrantly enter the midline, while ectopic expression of Slit1 in chick embryos leads to specific motor axon projection errors. Expression of dominant-negative Robo receptors within cranial motoneurons in chick embryos strikingly perturbs their projections, causing some motor axons to enter the midline, and preventing dorsally projecting motor axons from exiting the hindbrain. These data suggest that Slit proteins play a key role in guiding dorsally projecting cranial motoneurons and in facilitating their neural tube exit (Hammond, 2005).

The mammalian Slit protein guides neuronal and leukocyte migration through the transmembrane receptor Roundabout. The intracellular domain of Robo interacts with a novel family of Rho GTPase activating proteins (GAPs). Two of the Slit-Robo GAPs (srGAPs) are expressed in regions responsive to Slit. Slit increases srGAP1-Robo1 interaction and inactivates Cdc42. A dominant negative srGAP1 blocks Slit inactivation of Cdc42 and Slit repulsion of migratory cells from the anterior subventricular zone (SVZa) of the forebrain. A constitutively active Cdc42 blocks the repulsive effect of Slit. These results have demonstrated important roles for GAPs and Cdc42 in neuronal migration. A signal transduction pathway is proposed from the extracellular guidance cue to intracellular actin polymerization (Wong, 2001).

The yeast two-hybrid system was used to search for proteins interacting with the C-terminal region from amino acid residues (aa) 1455 to 1657 of rat Robo1. 20 positive clones were isolated from a mouse brain cDNA library, eight of which encode a novel family of rhoGAP proteins named here as slit-robo (sr) GAPs 1, 2, and 3, corresponding to the human KIAA 1304, KIAA0456, and KIAA0411, respectively. The srGAPs contain an FCH domain, a rhoGAP domain, and an SH3 domain. The FCH domain located from aa 11 to 110 in srGAP1 is similar to the FCH domains of srGAP2, srGAP3, C1 rhoGAP, cdc15, and Fer. The centrally located rhoGAP domain (aa 483 to 657) in srGAP1 is highly similar to rhoGAP domains of srGAP2, srGAP3, and C1 rhoGAP. The SH3 domain (aa 712 to 767) in srGAP1 is similar to the SH3 domains in srGAP2, srGAP3, C1 rhoGAP, and CSK. The overall primary structures of the srGAP proteins are highly conserved (Wong, 2001).

A pathway mediating Slit-Robo signaling is proposed. In this pathway, the leucine-rich regions of each Slit protein interact with the extracellular immunoglobulin domains of the Robo receptor. The extracellular interaction of Slit and Robo increases the interaction of the SH3 domain in an srGAP with the CC3 motif in Robo, resulting in localized activation of the srGAPs. srGAPs increase the intrinsic GTPase activity of Cdc42, which converts the GTP-bound form of Cdc42 into its GDP-bound form, therefore inactivating Cdc42. Inactivation of Cdc42 leads to a reduction in the activation of the Neuronal Wiskott-Aldrich Syndrome protein (N-WASP), thus decreasing the level of active Arp2/3 complex. Because active Arp2/3 promotes actin polymerization, the reduction of active Cdc42 eventually decreases actin polymerization. The repulsive effect of Slit can therefore be explained by the relative amounts of actin polymerization on the sides of the cell proximal and distal to the Slit source, with the proximal side having relatively less actin polymerization than the distal side. Of course, this simple model needs to (and can be) adapted to the growth cones of projecting axons (Wong, 2001).

Kidney development occurs in a stereotypic position along the body axis. It begins when a single ureteric bud emerges from the nephric duct in response to GDNF secreted by the adjacent nephrogenic mesenchyme. Posterior restriction of Gdnf expression is considered critical for correct positioning of ureteric bud development. Mouse mutants lacking either SLIT2 or its receptor ROBO2, molecules known primarily for their function in axon guidance and cell migration, develop supernumerary ureteric buds that remain inappropriately connected to the nephric duct, and the SLIT2/ROBO2 signal is transduced in the nephrogenic mesenchyme. Furthermore, Gdnf expression is inappropriately maintained in anterior nephrogenic mesenchyme in these mutants. Thus these data identify an intercellular signaling system that restricts, directly or indirectly, the extent of the Gdnf expression domain, thereby precisely positioning the site of kidney induction (Grieshammer, 2004).

Much of the current literature on the function of SLIT/ROBO signaling is focused on its chemorepulsive role in cell motility. Therefore it is appealing to consider a model in which SLIT2/ROBO2 signaling eliminates Gdnf expression by providing a chemorepulsive signal to Gdnf-expressing cells present in anterior nephrogenic mesenchyme, causing them to move posteriorly and thereby accumulate in the condensing metanephric mesenchyme at the stage when the UB begins to form. To explore this possibility, intermediate mesoderm was explanted from normal embryos, labeled nephrogenic mesenchymal cells in the region anterior to the prospective site of UB formation with a lipophilic dye, and whether they had moved posteriorly toward the vicinity of the nascent UB was assessed after 24 hr of culture. However, significant movement of cells in the nephrogenic mesenchyme was never observed (Grieshammer, 2004).

A second possible mechanism by which SLIT2/ROBO2 signaling might function to eliminate Gdnf-positive cells from the region anterior to the site of UB formation is by inducing cell death. If that were the case, one would expect that in the normal embryo, dying cells would be present in the nephrogenic mesenchyme immediately anterior to the Gdnf-positive domain as it is becoming restricted to the prospective site of UB formation. However, few or no dying cells were detected in this region (Grieshammer, 2004).

Furthermore, it is thought unlikely that loss of anterior Gdnf expression is achieved by removal of Gdnf-positive cells via an effect on cell migration or survival, because Robo2-expressing cells, which are presumably SLIT2 responsive, are present anterior to the Gdnf-positive domain in normal E10.5 embryos. If Robo2-positive cells exposed to SLIT2 had migrated posteriorly or had died in order to eliminate Gdnf expression, one would not expect to find Robo2-expressing cells anterior to the Gdnf-positive domain. Instead, it seems plausible that these Robo2-positive cells have responded to SLIT2 by turning off Gdnf expression (Grieshammer, 2004).

Therefore a third possible mechanism involving effects on Gdnf expression is favored to explain how SLIT2/ROBO2 signaling restricts the size of the Gdnf-positive domain. ROBO2 activation may downregulate Gdnf expression by negatively affecting transcriptional activators or positively affecting transcriptional repressors of Gdnf. To explore this issue, the expression of Eya1, Pax2, and Foxc1, presumed transcriptional regulators of Gdnf, was examined in Slit2 mutant embryos: no evidence was found that loss of Slit2 function affects their expression (Grieshammer, 2004).

One unexpected finding was that in normal embryos Eya1 and Pax2 expression is detected not only where Gdnf is expressed, but also in the region anterior to the Gdnf-positive domain. Thus, downregulation of Gdnf does not appear to be caused by a loss of Eya1 or Pax2 expression, but instead may occur via an inhibitory effect on EYA1 or PAX2 translation or activity. These findings raise the possibility that SLIT2/ROBO2 signaling controls such posttranscriptional effects on EYA1 and/or PAX2. Consistent with this possibility, it has been suggested that SLIT and ROBO are required to promote an asymmetric cell division during Drosophila neural development via a negative effect on Nubbin and Mitimere protein levels (Mehta, 2001). Additional studies will be needed to define the mechanism by which SLIT2/ROBO2 signaling interacts with the GDNF and perhaps other signaling systems to produce the single UB that invariantly forms at a specific time and place in the normal embryo (Grieshammer, 2004).

Vertebrate cranial sensory ganglia have a dual origin from the neural crest and ectodermal placodes. In the largest of these, the trigeminal ganglion, Slit1-Robo2 signaling is essential for proper ganglion assembly. This study demonstrates a crucial role for the cell adhesion molecule N-cadherin and its interaction with Slit1-Robo2 during gangliogenesis in vivo. A common feature of chick trigeminal and epibranchial ganglia is the expression of N-cadherin and Robo2 on placodal neurons and Slit1 on neural crest cells. Interestingly, N-cadherin localizes to intercellular adherens junctions between placodal neurons during ganglion assembly. Depletion of N-cadherin causes loss of proper ganglion coalescence, similar to that observed after loss of Robo2, suggesting that the two pathways might intersect. Consistent with this possibility, blocking or augmenting Slit-Robo signaling modulates N-cadherin protein expression on the placodal cell surface concomitant with alteration in placodal adhesion. Lack of an apparent change in total N-cadherin mRNA or protein levels suggests post-translational regulation. Co-expression of N-cadherin with dominant-negative Robo abrogates the Robo2 loss-of-function phenotype of dispersed ganglia, whereas loss of N-cadherin reverses the aberrant aggregation induced by increased Slit-Robo expression. This study suggests a novel mechanism whereby N-cadherin acts in concert with Slit-Robo signaling in mediating the placodal cell adhesion required for proper gangliogenesis (Shiau, 2009).