First printed in R&D Systems' 2004 Catalog.
During development of the nervous system, axons often navigate considerable distances before finally reaching their appropriate targets.
The highly motile growth cone at the tip of an extending axon is exquisitely sensitive to attractive and repellent cues that act as molecular guideposts for the developing neuron. These cues might be substrate-associated or diffusible, working at short range or over long distances, respectively. Axons often travel in distinct steps, reaching intermediate targets before continuing on to their ultimate destinations. One particularly good model system for studying the factors that regulate axon guidance in this fashion has been the ventral midline of the developing spinal cord.
||Figure 1. Commissural axons expressing Deleted in Colorectal Cancer (DCC) are attracted to the midline by Netrin-1 secreted by cells of the floor plate. A direct association between DCC and the Slit receptor, Roundabout (Robo), could account for a loss in sensitivity to Netrin-1 in contralateral axons.
Commissural neurons of the vertebrate embryo differentiate in the dorsolateral regions of the spinal cord.1 From here they send axons that navigate stereotypical trajectories, traveling ventrally along the lateral margin of the cord before turning at the level of the motor column and taking a more direct route to the midline floor plate. After crossing the midline in the ventral commissure, these axons then turn abruptly and travel in the ventral funiculus, a white matter tract that follows the longitudinal axis of the floor plate. Decussated commissural axons do not recross to the ipsilateral side, or enter inappropriate regions of the dorsal spinal cord. The specific pathways that commissural axons travel are regulated via expression patterns and the integration of signals from several families of guidance cues and their associated receptors. This mini-review focuses on these families with special emphasis given to those molecules and activities regulating the guidance of vertebrate commissural axons at the spinal cord midline.
Netrins make up a small family of secreted, laminin-related, molecules with multifunctional roles in axon guidance, acting as context-dependent chemoattractants or chemorepellents.2 Roles for Netrins and their receptors in axon guidance are conserved across species, ranging from Caenorhabditis elegans to Homo sapiens. To date 4 members of the Netrin family have been identified in rodents and include Netrin-1, -3, -4 (beta-Netrin), and -G1, with counterparts found in humans as well.2 Netrin receptors include members of the Deleted in Colorectal Cancer (DCC) and UNC5 families.2 Several studies have demonstrated the importance of vertebrate DCC in mediating Netrin-1 axon growth promotion and chemoattraction in vitro.3, 4 In vivo, Netrin-1/DCC play important roles in retinal ganglion cell axon pathfinding,5, 6 and the formation of corpus callosum and hippocampal commissures.7, 8, 9 In invertebrates, members of the UNC5 family, in general, mediate chemorepellent responses, while their roles in vertebrates remains less clear.7, 10 In cultured Xenopus neurons, direct interactions between DCC and UNC5 receptors can convert Netrin-1 attraction to repulsion,11 and a recent study shows that mice deficient in UNC5H3 exhibit pathfinding abnormalities in corticospinal tract neurons.12
Role at the Spinal Cord Midline
As they extend from the dorsolateral regions of the spinal cord, commissural axons take an initial ventral trajectory toward the midline suggesting the presence of attractive ventral cues. In vitro, diffusible Netrin-1 is a potent chemoattractant for commissural axons originating from spinal cord explants,13 and in vivo, is secreted by specialized ependymal cells of the midline floor plate where it likely forms a ventral to dorsal gradient.3, 13 Commissural axons in mice deficient in either Netrin-18 or its receptor DCC9 exhibit misprojections, many axons stall, and few reach the ventral commissure, strongly supporting the notion that Netrin-1 is indeed a critical midline attractive guidance cue (Figure 1).
After crossing the midline, wild-type axons remain contralateral and never re-enter the floor plate or recross to the ipsilateral side. If Netrin-1 is a floor plate-released chemoattractant, why then do commissural axons leave the floor plate region without re-entering? One possibility is that they somehow lose sensitivity to Netrin-1 on the contralateral side of the midline and, therefore, are no longer attracted by floor plate tissues. Whether this happens in the spinal cord in vivo is unclear, although axons from hindbrain explants do lose responsiveness to Netrin-1 after crossing the midline.14 If spinal commissural axons do lose sensitivity to Netrin-1, the mechanism likely does not involve a downregulation of DCC expression, since the receptor is found on both ipsilateral and contralateral axons.3 However, clues to potential mechanisms may have been provided by studies assessing the differential responses of Xenopus neurons to gradients of Netrin-1 in vitro. Xenopus spinal neurons lose their attraction to Netrin-1 following exposure to Slit, a guidance cue that, like Netrin-1, is secreted by the floor plate.15, 16 The loss in sensitivity results from a direct inhibitory interaction between the Slit receptor, Roundabout (Robo), and DCC.15 The response to a Netrin-1 gradient in vitro also depends on the activities of intracellular cyclic nucleotides. Xenopus neurons are attracted to a gradient of Netrin-1 and repulsed by the same gradient when intracellular cAMP activity is blocked with a competitive cAMP analog or by Protein Kinase A inhibitors.17 There is precedent for altered axonal responses to Netrin-1 in vivo with an apparent association with cAMP levels. In the Xenopus optic pathway, retinal axons change their response to Netrin-1 from attraction to repulsion as they navigate from the retina toward the tectum.18 This altered response correlates with a developmentally regulated decrease in cAMP levels as these neurons age in vitro. Whether spinal commissural axon responses to Netrin-1 are modulated by cAMP in vivo remains to be determined.
Characterized by at least 30 members, the Semaphorins make up the largest family of axon guidance cues yet described.19 Semaphorins are divided into 8 classes, of which, classes 3 through 7 are found in vertebrates. Class 3 Semaphorins are secreted, classes 4 through 6 are transmembrane proteins, and class 7 are membrane associated via glycosylphosphatidylinositol (GPI) linkage. They are classically described as collapsing factors and mediators of axon repulsion in vitro,20, 21 and in vivo, as factors that regulate fasciculation22 and exclude axons from inappropriate regions of the nervous system during development.22, 23, 24 However, they may act as context-dependent chemoattractants as well.25
Multiple receptors or receptor complexes mediate Semaphorin signaling. Neuropilin (Npn)-1 and -2 are transmembrane proteins that bind the widely studied class 3 Semaphorins and are necessary for Semaphorin-mediated signaling in many contexts.21, 26 The short cytoplasmic region of the Npns contains no known signaling sequence, and mutants lacking the cytoplasmic tail still mediate Semaphorin function, suggesting that these receptors comprise only a part of a receptor complex.27 Coreceptors important for mediating Npn/Semaphorin activity include the Plexins and Neural Cell Adhesion Molecule (NCAM)-L1. Plexins -A through -D make up a large family of transmembrane proteins that bind certain Semaphorin classes directly, and can act as coreceptors with Npns in mediating class 3 Semaphorin effects.28, 29, 30 For example, in vitro studies show that receptor complexes including Npns and certain A-type Plexins are necessary for Semaphorin 3A (Sema3A)-mediated axon repulsion.29 In addition to the role of Plexins as coreceptors, recent studies demonstrate that Npn association with NCAM-L1 is necessary for mediating Sema3A chemorepellent effects on mouse cortical and dorsal root ganglion neurons.31, 32
||Figure 2. Semaphorin 3B (Sema3B) is expressed in the floor plate and Sema3F is expressed throughout most of the remaining gray matter excluding the floor plate. Explant assays show that commissural axons gain a postcrossing sensitivity to these Semaphorins, indicating a role in excluding contralateral axons from inappropriate regions of the gray matter and preventing recrossing of the midline. The Semaphorin receptor Neuropilin-2 (Npn-2) is required for proper midline crossing suggesting that it may mediate Semaphorin effects on commissural axon guidance.
Role at the Spinal Cord Midline
After crossing the midline, commissural axons make an orthogonal turn along the longitudinal axis into white matter tracts of the ventral funiculus. Once on the contralateral side, these axons do not recross the midline or enter inappropriate regions of dorsal gray matter.1 In addition to the possibility that commissural axons may lose sensitivity to midline attractants like Netrin-1 in order to remain on the contralateral side, it is also plausible that they could gain a postcrossing sensitivity to midline repellent cues. In the rat embryo, Sema3B is expressed in the spinal cord floor plate region, while Sema3F is found throughout most of the spinal cord gray matter apart from the floor plate.16 Because decussated commissural axons are normally excluded from these regions during development, these expression patterns could suggest a role for Sema3B and Sema3F in directing axons into the appropriate fiber tracts (Figure 2). Indeed, it has been demonstrated using an in vitro explant assay, that commissural axons gain a postcrossing sensitivity to floor plate-secreted repellent factors, a response that is mimicked by transfected COS cells expressing secreted Sema3B or Sema3F.16 Consistent with a role for Semaphorins in expelling crossing commissural axons from the midline, axons from Npn-2-deficient mice stall and/or exhibit disorganized projections upon reaching the floor plate.16
Morphogens are best known for their effects on cell fate determination during development. Local concentration gradients convey positional information used during organization of the major axes, limb development, and patterning of the nervous system. Recent studies suggest that in addition to these roles, morphogens, including members of the Wnt family, Bone Morphogenetic Protein (BMP) family, and Sonic Hedgehog (Shh), might also function in axon guidance. For instance, knockout of the Wnt receptor, Frizzled-3, results in severe defects in several major fiber tracts in the vertebrate forebrain33 and Wnt-3 slows axon outgrowth and may mediate terminal branching of vertebrate muscle afferents.34 Although the role of Wnts in axon guidance at the spinal cord midline is unknown, in the Drosophila ventral nerve cord, Wnt-5-binding to the receptor tyrosine kinase Derailed is required for targeting axons to the appropriate midline commissure.35 Other evidence for morphogen involvement in axon guidance includes defects in retinal ganglion cell projections in mice deficient in the BMP receptor, BMPR-IB,36 and Shh has been shown to inhibit the outgrowth of neurons from retinal explants in vitro.37
||Figure 3. Sonic Hedgehog (Shh) secreted by the floor plate acts as a chemoattractant for commissural axons. Smoothened (Smo), the signaling component of the Shh receptor, is important for this activity. Bone Morphogenic Proteins (BMPs) -6 and -7, and BMP family member Growth Differentiation Factor 7 (GDF-7) are expressed in the roof plate. BMP-6 acts as a dorsal repellent for commissural axons, an activity that is enhanced through the formation of BMP-6/GDF-7 heterodimers. The BMP receptor responsible for the activity is at present unknown.
Role at the Spinal Cord Midline
It is well accepted that Netrin-1 is an important attractive factor secreted by the floor plate.13 However, commissural axons do still exhibit some attraction to Netrin-1-deficient floor plate tissues, and a small percentage of axons reach the floor plate in Netrin-1-deficient mice.8 These results suggest that Netrin-1 is likely not the sole attractive cue at the midline. Like Netrin-1, Shh is secreted by cells of the vertebrate floor plate and is also a chemoattractant for commissural axons in vitro, suggesting that it may be the factor regulating Netrin-1-independent midline attraction (Figure 3).38 Transcription factors of the Gli family (Gli1 through Gli3) are important players in Shh-mediated signal transduction pathways in the developing vertebrate nervous system.39 The neural tube of mice lacking Gli2 function fails to develop a functional midline floor plate and, therefore, lacks Shh expression.40 In Gli2-/- mice, most axons do reach the floor plate, but many exhibit wandering projections into the motor column on the ipsilateral side and disorganization in the ventral commissure. When the Gli2 deficiency is combined with a knockout for Netrin-1, commissural axons fail to enter the ventral spinal cord, and consequently, the ventral commissure does not form.38 The mechanism underlying Shh effects is likely mediated by the prototypical Shh receptor complex consisting of a binding component termed Patched (Ptc) and an associated signaling subunit termed Smoothened (Smo). Pharmacological inhibition of Smo with cyclopamine blocks Shh-mediated attraction in vitro, and targeted disruption of the Smo gene in commissural axons results in a phenotype similar to that of Gli2-/- mice. These results suggest that Shh activity mediated via Ptc/Smo receptor complex works in concert with Netrin-1/DCC for proper targeting of commissural axons to the ventral midline in vivo.38
Although commissural axons require attractive molecules like Netrin-1 and Shh in order to reach the ventral midline, even in the absence of these cues, axons do still acquire at least an initial ventral trajectory.38 This suggests that repellent factors known to exist in roof plate tissues might also play a role in deflecting axons from the dorsal regions of the spinal cord.41 Members of the BMP family, including BMP-6, BMP-7, and Growth Differentiation Factor 7 (GDF-7), are expressed in the vertebrate roof plate and are potential candidates as dorsal repellent cues (Figure 3).41 Secreted BMP-7 alone, and in synergy with GDF-7 through the formation of BMP-7/GDF-7 heterodimers, exhibit the ability to repel commissural axons in vitro.41, 42 In vivo, a subset of commissural axons at early stages of outgrowth in mice deficient in BMP-7 and/or GDF-7 exhibit abnormal medial or dorsal projections and occasionally misproject across the midline roof plate.42 The receptors mediating roof plate repulsion via the BMP family remain to be identified.
Slit proteins make up a family of multifunctional guidance cues with putative roles in regulating neuronal migration,43 axonal and dendritic branching,44, 45, 46 and axon guidance.47, 48 These large glycoproteins are conserved across species with 3 family members (Slit1 through 3) identified in developing and adult mammalian nervous systems.47, 49 Slits are susceptible to proteolytic cleavage resulting in fragments (Slit-N and Slit-C) that can have variable functions.44, 47, 50 Knockout studies demonstrate that Slits1 and 2 play critical roles in the formation of several mammalian fiber tracts including corticofugal, thalamocortical, callosal, optic, and lateral olfactory tracts.51, 52, 53 The receptor Robo is well known for mediating the repellent effects of Slit at the midline of the Drosophila ventral nerve cord. Ipsilaterally projecting axons and decussated commissural axons that express Robo are prevented from inappropriate crossing of the midline via interaction with Slit.48 Genetic mutations in either Robo or Slit lead to aberrant crossing and recrossing, or failure of these axons to leave the midline.48 In Drosophila, commissural axons acquire a postcrossing sensitivity to Slit resulting from increased surface expression of Robo. This occurs via a mechanism that includes inactivation of Comm, an intracellular sorting protein that normally targets Robo for an endosomal degradation pathway.54
||Figure 4. Slits1 through 3 are expressed in the floor plate, while Slit2 is also found in the roof plate and motor column. Slit receptors Robo1 and 2 are found in regions consistent with expression by commissural axons. Explant assays show that commissural axons gain sensitivity to Slit2 only after crossing the midline, suggesting a role in the exclusion of contralateral axons from inappropriate regions of the spinal cord.
Role at the Spinal Cord Midline
In the rat E13 embryo, Slits1 through 3 are expressed in the floor plate, while Slit2 is also found in the roof plate and motor column.47 In addition, Robo1 and 2 are localized to regions of the rat spinal cord consistent with expression on commissural axons.47, 48 As mentioned, in Drosophila, Slit/Robo activity is upregulated in decussated axons and is responsible for preventing aberrant crossing of the midline.48 Although much less clear in the vertebrate, embryonic rat commissural axons from spinal cord explants are repelled by COS cells expressing secreted Slit2 only after crossing the midline.16 This activity is similar to what has been reported for Sema3B and 3F using this same model and suggests that, like Semaphorin signaling, Slit signaling may underlie at least part of the mechanism that prevents recrossing of the midline or axon entry into inappropriate gray matter (Figure 4).16
Ephrins and their tyrosine kinase receptors, Ephs, are divided into two classes: Ephrin-As, which are anchored to the membrane via GPI linkage and preferentially bind EphA receptors, and Ephrin-Bs, which are transmembrane proteins that preferentially interact with receptors of the EphB subtype. The primary exception is the EphA4 receptor, which has been shown to interact with members of both class A and class B Ephrins.55, 56, 57 Ephrins are promiscuous, often exhibiting the ability to bind several Eph receptor subtypes within the same class.58 Ephrin/Ephs, including the class A Ephrins associated via membrane GPI linkage only, have the unusual capacity of bidirectional signaling, involving activation of transduction pathways in both ligand and receptor-expressing cells.59 Both classes have been implicated as regulators of axon guidance. For instance, topographic mapping of the anterior-posterior tectal/superior collicular axis is dependent upon expression patterns of EphA/Ephrin-A,60, 61 while EphB/Ephrin-B signaling is important for mapping along the lateral-medial axis.62, 63 In addition, EphA4 and Ephrin-3B are required for the development of neural networks regulating alternating left-right movement during locomotion.64, 65 EphA4 and Ephrin-3B-deficient mice exhibit defects in corticospinal tract development64, 65 and deficiencies in the formation of central pattern generators, local spinal circuits necessary for coordinated locomotion.56
||Figure 5. EphA2 and EphB expression is upregulated on contralateral commissural axons. Class B Ephrins are expressed in the floor plate and a broad region of the dorsal spinal cord. A subset of decussated commissural axons take a more dorsal trajectory before turning at the border of class B Ephrin expression, indicating a role in excluding these axons from dorsal regions of the spinal cord.
Role at the Spinal Cord Midline
Several lines of evidence suggest that Ephrins play a role in commissural axon guidance. For instance, in vitro, Ephrins B1 through B3 induce collapse of commissural growth cones suggesting that they may act as chemorepellents for these neurons.66 In addition, it has been shown that EphA and EphB receptors are expressed by commissural axons in vivo.66, 67 Interestingly, both EphA2 and EphB receptor expression appears to be upregulated on commissural neurons after midline crossing.66, 67 In the case of EphA2, increased expression is due to the induction of local protein synthesis occurring in the distal segments of the axons.67 These results implicate Ephrins/Ephs in the guidance of commissural axons along the contralateral side of the spinal cord. Class B Ephrins are expressed in the vertebrate floor plate region and in a broad domain of the dorsal spinal cord.66, 68 When commissural axons cross the midline they make an abrupt orthogonal turn into the ventral funiculus.1 A subset of these axons then take a more dorsal trajectory before again turning in the longitudinal axis along a border of class B Ephrin expression, that in the E13 mouse embryo, encompasses roughly the dorsal half of the spinal cord.68 Blockade of EphB/Ephrin-B activity in spinal cord explants results in axon misprojections into normally restricted areas of the dorsal spinal cord.68 These studies support a role for Ephrins/Ephs in repelling decussated commissural axons from inappropriate regions of the dorsal gray matter and directing their projections along the appropriate white matter tracts on the contralateral side of the developing spinal cord (Figure 5).
The activities of adhesion molecules play critical roles in many aspects of neurodevelopment ranging from axon growth promotion and pathfinding, to synapse formation, plasticity, and learning and memory.69, 70, 71 The adhesion molecule F-spondin is a secreted extracellular matrix protein and a member of the Thrombospondin Type-1 Repeat (TSR) superfamily.72 In addition to effects on commissural neurons, F-spondin regulates cell adhesion and axon growth promotion in cultured hippocampal and dorsal root ganglion neurons, while acting as a contact repellent for embryonic motor neurons.72, 73, 74, 75
Several cell adhesion molecules (CAMs) of the immunoglobulin (Ig) superfamily have also been implicated as players regulating axon guidance at the midline including mammalian NCAM-L1, NrCAM, and TAG-1 (Mammalian ortholog of chick Axonin-1). As implied by membership in the Ig superfamily, these CAMs all have extracellular regions containing six Ig domains, and are further characterized by the presence of four (NCAM-L1, NrCAM) or five (TAG-1, Axonin-1) Fibronectin type III (FNIII) extracellular domains.69 Although these CAMs contain no intrinsic kinase activity, heterophilic and homophilic interactions can lead to the induction of intracellular signaling cascades. For instance, crosslinking of NCAM-L1 leads to Src-dependent endocytosis and activation of PI3-kinase, Rac1, and MAP Kinase signaling cascades in neuroblastoma cells.76 NCAM-L1 has also been shown to directly interact with, and activate, FGF receptors to stimulate neurite outgrowth.77
||Figure 6. F-spondin is expressed in the floor plate region and is required for proper turning of contralateral commissural axons along the longitudinal axis of the spinal cord. NCAM-L1 and TAG-1 are primarily found associated with contralateral and ipsilateral commissural axons, respectively. Proper crossing of the floor plate requires that Axonin-1, the chick ortholog of TAG-1, interacts with NrCAM expressed in the floor plate. NrCAM is also expressed by commissural axons and is upregulated in the ventral commissure.
Role at the Spinal Cord Midline
F-spondin is expressed in the floor plate region of the developing embryo, and when used as a cell culture substrate, can promote the outgrowth of commissural axons.72, 74 This activity can be mimicked by the protein's TSR domain alone.74 Using function-blocking antibodies in vivo against the TSR domain, chick commissural axons still reach the contralateral side of the midline, but exhibit defects in the stereotypical turn into the ventral funiculus in the longitudinal axis. Instead of crossing the midline, turning, and maintaining a longitudinal trajectory along the floor plate, many axons defasciculate and travel inappropriately into more lateral regions of the spinal cord. These results, and the ability of F-spondin to enhance commissural axon outgrowth in vitro, suggest a potential role as a midline attractant responsible for directing axons along the correct longitudinal trajectory on the contralateral side of the floor plate (Figure 6).74
Although their precise function in spinal cord development remains unclear,78, 79 intriguing expression patterns of the Ig CAMs, NCAM-L1 and TAG-1, make them of interest as potential guidance cues for commissural axons (Figure 6). In rodents, NCAM-L1 appears to be upregulated only after reaching the ventral spinal cord, or during and after crossing the midline.1, 66, 80, 81 In contrast, rodent TAG-1 has been shown to be downregulated upon midline crossing.1, 81 The chick ortholog of rodent TAG-1, Axonin-1, is expressed by commissural axons and interacts with NrCAM expressed by cells of the floor plate. Exposure to soluble Axonin-1, or function blocking antibodies to Axonin-1 or NrCAM, inhibits commissural axon entry into floor plate explants.82 In addition, repeated injection of function-blocking antibodies in ovo results in up to 50% of commissural axons failing to cross to the contralateral spinal cord.82 In addition to the floor plate, NrCAM is also expressed on commissural axons and is upregulated upon entry into the ventral commissure.81 The function of axonal NrCAM upregulation in the ventral commissure is unknown.
The study of commissural neurons of the spinal cord has provided a wealth of information regarding several families of molecules critical for guiding these axons as they extend toward and across the midline. Strong evidence suggests that floor plate-secreted molecules including Netrin-1 and Shh are important for attracting axons to the midline, while roof plate secreted members of the BMP family may serve to deflect growing axons from dorsal regions of the cord. Evidence also seems to suggest that members of the Semaphorin, Slit, and Ephrin families act as repellents that guide commissural axons to the appropriate fiber tracts on the contralateral side of the spinal cord and prevent inappropriate recrossing of the midline. Lastly, adhesion molecules, including the extracellular matrix protein F-spondin and CAMs of the Ig superfamily, including Axonin-1 and NrCAM are important for regulating entry of commissural axons into the ventral commissure and coordinated longitudinal turning into the ventral funiculus. However, many important questions still remain.
Commissural axons require an initial attraction to the midline, but upon crossing must either lose that attraction and/or change to a repellent response in order to remain on the contralateral side. In the vertebrate hindbrain decussated axons are no longer attracted to Netrin-1.14 Whether the underlying mechanism involves Slit-induced inhibitory interaction between Robo and DCC, as is the case with cultured Xenopus spinal neurons, remains to be determined. Evidence using spinal cord explant assays suggests that after midline crossing, commissural neurons gain sensitivity to the repellents Slit2, and Sema3B and 3F, an activity that could play an important role in preventing inappropriate recrossing of the floor plate. Little is known regarding the mechanism that may underlie the upregulation of Semaphorin activity and/or whether a Comm-regulated sorting mechanism is in place for Robo as occurs in Drosophila. Certain members of the Eph family including EphB and EphA2 are upregulated on the distal segments of decussated commissural axons. In the case of EphA2, upregulation appears to require local protein synthesis, but the underlying trigger that may be present as axons cross the midline remains to be identified. Finally, several reports suggest that the CAMs, NCAM-L1, and TAG-1 appear to be localized to the distal and proximal segments of commissural axons, respectively. This expression pattern suggests that these molecules could play a role in commissural axon guidance. Both NCAM-L1 and TAG-1 can act as axonal growth promoting substrates in vitro, but their function at the spinal midline in vivo remains unclear.79, 83, 84 Future studies will undoubtedly address questions such as these, while at the same time raising many more. The answers will ultimately provide an increased understanding of the complex processes that underlie nervous system development.
- Dodd, J. et al. (1988) Neuron 1:105.
- Manitt, C. & T.E. Kennedy (2002) Prog. Brain Res. 137:425.
- Keino-Masu, K. et al. (1996) Cell 87:175.
- de la Torre, J.R. et al. (1997) Neuron 19:1211.
- Deiner, M.S. et al. (1997) Neuron 19:575.
- Deiner, M.S. & D.W. Sretavan (1999) J. Neurosci. 19:9900.
- Keleman, K. & B.J. Dickson (2001) Neuron 32:605.
- Serafini, T. et al. (1996) Cell 87:1001.
- Fazeli, A. et al. (1997) Nature 386:796.
- Hedgecock, E.M. et al. (1990) Neuron 4:61.
- Hong, K. et al. (1999) Cell 97:927.
- Finger, J.H. et al. (2002) J. Neurosci. 22:10346.
- Kennedy, T.E. et al. (1994) Cell 78:425.
- Shirasaki, R. et al. (1998) Science 279:105.
- Stein, E. & M. Tessier-Lavigne (2001) Science 291:1928.
- Zou, Y. et al. (2000) Cell 102:363.
- Ming, G.-l. et al. (1997) Neuron 19:1225.
- Shewan, D. et al. (2002) Nat. Neurosci. 10:955.
- Fiore, R. & A.W. Puschel (2003) Front. Biosci. 8:s484.
- Luo, Y. et al. (1993) Cell 75:217.
- Chedotal, A. et al. (1998) Development 125:4313.
- Taniguchi, M. et al. (1997) Neuron 19:519.
- Kitsukawa, T. et al. (1997) Neuron 19:995.
- Schwarting, G.A. et al. (2000) J. Neurosci. 20:7691.
- Wong, J.T. et al. (1999) Nat. Neurosci. 2:798.
- Bagri, A. & M. Tessier-Lavigne (2002) Adv. Exp. Med. Biol. 515:13.
- Nakamura, F. et al. (1998) Neuron 21:1093.
- Winberg, M.L. et al. (1998) Cell 95:903.
- Tamagnone, L. et al. (1999) Cell 99:71.
- Rohm, B. et al. (2000) Mech. Dev. 93:95.
- Castellani, V. et al. (2000) Neuron 27:237.
- Castellani, V. et al. (2002) EMBO J. 21:6348.
- Wang, Y. et al. (2002) J. Neurosci. 22:8563.
- Krylova, O. et al. (2002) Neuron 35:1043.
- Yoshikawa, S. et al. (2003) Nature 422:583.
- Liu, J. et al. (2003) Dev. Biol. 256:34.
- Trousse, F. et al. (2001) Development 128:3927.
- Charron, F. et al. (2003) Cell 113:11.
- Ruiz i Altaba et al. (1998) Development 125:2203.
- Ding, Q. et al. (1998) Development 125:2533.
- Augsburger, A. et al. (1999) Neuron 24:127.
- Butler, S.J. & J. Dodd (2003) Neuron 38:389.
- Wu, W. et al. (1999) Nature 400:331.
- Wang, K.H. et al. (1999) Cell 96:771.
- Ozdinler, P.H. & R.S. Erzurumlu (2002) J. Neurosci. 22:4540.
- Whitford, K.L. et al. (2002) Neuron 33:47.
- Brose, K. et al. (1999) Cell 96:795.
- Kidd, T. et al. (1999) Cell 96:785.
- Marillat, V. et al. (2002) J. Comp. Neurol. 442:130.
- Nguyen-Ba-Charvet, K.T. et al. (2001) J. Neurosci. 21:4281.
- Bagri, A. et al. (2002) Neuron 33:233.
- Plump, A.S. et al. (2002) Neuron 33:219.
- Nguyen-Ba-Charvet, K.T. et al. (2002) J. Neurosci. 22:5473.
- Keleman, K. et al. (2002) Cell 110:415.
- Kullander, K. et al. (2001) Genes Dev. 15:877.
- Kullander, K. et al. (2003) Science 299:1899.
- Brors, D. et al. (2003) J. Comp. Neurol. 462:90.
- Himanen, J.-P. & D.B. Nikolov (2003) Int. J. Biochem. Cell Biol. 35:130.
- Holmberg, J. & J. Frisen (2002) Trends Neurosci. 25:239.
- Feldheim, D.A. et al. (2000) Neuron 25:563.
- Brown, A. et al. (2000) Cell 102:77.
- Mann, F. et al. (2002) Neuron 35:461.
- Hindges, R. et al. (2002) Neuron 35:475.
- Dottori, M. et al. (1998) Proc. Natl. Acad. Sci. USA 95:13248.
- Yokoyama, N. et al. (2001) Neuron 29:85.
- Imondi, R. et al. (2000) Development 127:1397.
- Brittis, P.A. et al. (2002) Cell 110:223.
- Imondi, R. & Z. Kaprielian (2001) Development 128:4859.
- Walsh, F.S. & P. Doherty (1997) Annu. Rev. Cell Dev. Biol. 13:425.
- Skaper, S.D. et al. (2001) Prog. Neurobiol. 65:593.
- Panicker, A.K. et al. (2003) Front. Biosci. 8:d900.
- Klar, A. et al. (1992) Cell 69:95.
- Feinstein, Y. et al. (1999) Development 126:3637.
- Burstyn-Cohen, T. et al. (1999) Neuron 23:233.
- Tzarfati-Majar, V. et al. (2001) Proc. Natl. Acad. Sci. USA 98:4722.
- Schmid, R.-S. et al. (2000) J. Neurosci. 20:4177.
- Williams, E.J. et al. (1994) Neuron 13:583.
- Cohen , N.R. et al. (1997) Curr. Biol. 8:26.
- Fukamauchi, F. et al. (2001) Biochem. Biophys. Res. Commun. 281:220.
- Tran, T.S. & P.E. Phelps (2000) Dev. Neurosci. 22:228.
- Matise, M.P. et al. (1999) Development 126:3649.
- Stoekli, E.T. et al. (1997) Neuron 18:209.
- Williams, E.J. et al. (1992) J. Cell Biol. 119:833.
- Felsenfeld, D.P. et al. (1994) Neuron 12:67