Little is known about the role of class 3 semaphorins in the development of CNS circuitry. Several class 3 semaphorins, including semaphorin 3F (Sema3F) bind to the receptor neuropilin-2 to confer chemorepulsive responses in vitro. To understand the role of Sema3F in the establishment of neural circuitry in vivo, we have generated sema3F null and sema3F conditional mutant mice. Inspection of the peripheral nervous system in sema3F null mice reveals that Sema3F is essential for the proper organization of specific cranial nerve projections. Analysis of the CNS in sema3F null mice reveals a crucial role for Sema3F in the rostral forebrain, midbrain, and hippocampus in establishing specific Npn-2 (neuropilin-2)-expressing limbic tracts. Furthermore, we identify Sema3F and Npn-2 as the first guidance cue-receptor pair shown to be essential for controlling the development of amygdaloid circuitry. In addition, we provide genetic evidence in vertebrates for a neuronal requirement of a soluble axon guidance cue in CNS axon guidance. Our data reveal a requirement for neuronal Sema3F in the normal development of the anterior commissure in the ventral forebrain and infrapyramidal tract in the hippocampus. Thus, our results show that Sema3F is the principal ligand for Npn-2-mediated axon guidance events in vivo and is a critical determinant of limbic and peripheral nervous system circuitry.
One of the most remarkable aspects of nervous system development is the specification and formation of functionally appropriate synaptic connections. Wiring up roughly 10 billion neurons into meaningful circuits may seem like a hopelessly complicated problem to address, but there are few other problems that are as central to the study of nervous system development. Work from the last three decades has brought increasing insight into the cellular and molecular mechanisms that underlie the patterning of appropriate connections, and progress in this field has come in part from the realization that, from a developmental perspective, the problem of neuronal connectivity can be broken down into three separate developmental events. The first is the specification of neuronal subtypes, a process that leads to the generation of groups of neurons that share certain functional properties. The second is pathfinding, a process that allows axons from functionally related neurons to grow to their appropriate target regions. The third is the formation of appropriate synaptic connections, a process that leads to the formation of functionally appropriate neural circuits. Advances in our understanding of the first two processes have been recently reviewed elsewhere (11Lumsden A Krumlauf R Science. 1996; 274: 1109-1115Crossref PubMed Scopus (924) Google Scholar, 14Tessier-Lavigne M Goodman C.S Science. 1996; 274: 1123-1133Crossref PubMed Scopus (2592) Google Scholar); the focus of this review is a discussion of some recent progress in the field with regard to the problem of specification of synaptic circuits during development. Much of our insight into the mechanisms that underlie the formation of appropriate synaptic connections comes from the studies on the development of the circuit that mediates the stretch reflex. In this simple circuit, the peripheral branch of the sensory axons innervate a limb muscle and fire action potentials upon detecting a rapid mechanical stretch in the muscle. The sensory neurons convey this information to the spinal motor neurons via monosynaptic synapses that the central branch of the sensory axons make onto the dendrites of the motor neurons. The axons of these motor neurons in turn project to the muscle and induce a contractile response. Studies on the development of this circuit indicate that the underlying connections are specified with exquisite precision and require coordinated interactions among the motor nerve, the muscle target, and the sensory nerve. While much is known about the cell biology of the process, the molecular mechanisms that mediate the formation of these precise connections are not well understood. In a current study, the Jessell, Lance-Jones, Anderson, and Saito laboratories (Lin et al., 1998 [this issue of Cell]) provide compelling evidence that the establishment of this functional neural circuitry depends on the coordinate expression of individual ETS-family transcription factors by both motor neurons and corresponding presynaptic sensory neurons. Further, this study establishes an essential role for the peripheral target in directing the expression of similar ETS transcription factors in both classes of neurons. Although the requirement for specific ETS proteins in circuit formation is not yet firmly established, this work provides a logical framework for beginning to understand how specific connections between the motor neurons, the sensory neurons, and the muscle may be established during development. A discussion of the mechanisms that specify reflex circuits requires a brief review of the organization of motor neurons within the spinal cord (see Figure 1). In the chick, the motor neurons that innervate the limb musculature are located at the brachial and lumbosacral levels in the lateral motor columns (LMC) of the spinal cord. Within the LMC, motor neurons that innervate ventral muscles are located medially (LMCm), and those that innervate dorsal muscles are located laterally (LMCl) (8Landmesser L J. Physiol. 1978; 284: 371-389PubMed Google Scholar) (Figure 1C). LMC motor neurons are generated starting at stage 15 in the chick, with the LMCm neurons becoming postmitotic before the LMCl neurons (4Hollyday M Hamburger V Brain Res. 1977; 132: 197-208Crossref PubMed Scopus (222) Google Scholar). The LMCm and LMCl motor neurons can be further subdivided into motor pools that correspond to groups of motor neurons that innervate a specific muscle. Motor neurons of the LMCs can be distinguished from other motor neurons by patterns of expression of LIM homeodomain (LIM-HD) proteins, and this molecular signature provides an important tool for defining the factors that affect motor column determination (15Tsuchida T Ensini M Morton S.B Baldassare M Edlund T Jessell T.M Pfaff S.L Cell. 1994; 79: 957-970Abstract Full Text PDF PubMed Scopus (853) Google Scholar, 13Tanabe Y Jessell T.M Science. 1996; 274: 1115-1123Crossref PubMed Scopus (624) Google Scholar, 2Ensini M Tsuchida T.N Belting H.-G Jessell T.M Development. 1998; 125: 969-982PubMed Google Scholar). LMC motor neurons extend an axon toward their limb muscle targets via the motor nerve, and by stage 23/24 the growth cones of these axons have reached the base of the developing limb. Upon entering the limb mesenchyme, the nerve splits into dorsal and ventral branches, which contain axons from the lateral and medial LMCs, respectively. By stage 35 the motor axons have innervated their appropriate muscle targets. Several perturbation experiments suggest that by the time lateral column motor neurons become postmitotic they are determined with regard to their specificity for appropriate muscle targets. For example, if the limb is rotated before innervation by the motor nerve, the nerve innervates the appropriate muscles although the muscles are now in ectopic locations (6Lance-Jones C Landmesser L J. Physiol. 1980; 302: 581-602Crossref PubMed Scopus (157) Google Scholar, 3Ferns M.J Hollyday M J. Neurosci. 1993; 13: 2463-2476PubMed Google Scholar). Similarly, if a segment of the neural tube is rotated at stage 15 to cause an anterior–posterior reversal, the motor axons are able to grow to their appropriate muscles as long as the rotation does not involve more than 2–3 spinal segments (7Lance-Jones C Landmesser L Proc. Royal Soc. London. 1981; 214: 19-52Crossref PubMed Scopus (183) Google Scholar). Such early specification of motor neurons is not unique to chicks, and elegant cell transplantation experiments in the zebrafish embryo have indicated that individual motor neurons are determined with regard to peripheral muscle innervation even before they extend an axon (1Eisen J.S Science. 1991; 252: 569-572Crossref PubMed Scopus (104) Google Scholar). The ability of emerging motor axons to make appropriate target choices soon after the motor neurons become postmitotic suggests that motor neurons in different motor pools must be molecularly distinct by the time they extend an axon. The demonstration that LIM-HD proteins define subclasses of motor columns prompted a search for a distinct class of transcription factors that might serve to distinguish motor pools from one another. In the current study 10Lin J.H Saito T Anderson D.J Lance-Jones C Jessell T.M Arber S Cell. 1998; 95 (this issue,): 393-407Abstract Full Text Full Text PDF PubMed Scopus (275) Google Scholar report that members of the ETS family of transcription factors (which have previously been characterized as targets of the Ras-MAPK signaling pathway and influence the differentiation of a variety of invertebrate and vertebrate cell types [16Wasylyk B Habman J Gutierrez-Hatmann A Trends Biochem. 1998; 270: 213-216Abstract Full Text Full Text PDF Scopus (434) Google Scholar]) are expressed in discrete populations of motor neurons in the embryonic chick ventral spinal cord. Importantly, retrograde labeling experiments with horseradish peroxidase (HRP) indicate that individual motor pools express a specific complement of ETS and LIM-HD proteins. For example, HRP injections into the adductor (A) muscle (a ventral muscle) retrogradely label a population of motor neurons that expresses the ER81 ETS protein and also the LIM-HD proteins Isl1 and Isl2 (Figure 1C). Therefore, this expression profile defines this ER81+ A motor pool as an LMCm motor pool. Similarly, HRP injections into the iliotrochanterici (ITR) muscle (a dorsal muscle) retrogradely label a motor pool that expresses the PEA3 ETS protein and also the LIM-HD proteins Isl2 and Lim1, defining this PEA3+ ITR motor pool as an LMCl motor pool. A detailed analysis of several lumbosacral motor pools shows that the combinatorial expression of ETS and LIM-HD proteins serves to define uniquely many motor pools. Since previous neural tube reversal experiments have shown that signals from the paraxial mesoderm (PM) help establish the rostrocaudal identity of specific motor pools during a critical period up until stage 15 (Figure 1A), patterns of ETS protein expression should also reflect this influence if they are markers for distinct motor pools. Indeed, Lin et al. report that following neural tube reversal at stage 13 when paraxial mesodermal signals are capable of respecifying motor pools, the distribution of ETS and LIM-HD proteins shows that specific motor pools ultimately are found to reside in their normal rostrocaudal position. Neural tube reversal at stage 15, when this respecification is no longer possible, results in ETS and LIM-HD expression profiles that show a rostrocaudal reversal of motor pool identity. These observations indicate that subsets of motor pools can be uniquely identified molecularly during spinal cord development based on the pattern of ETS and LIM-HD protein expression. Sensory neurons, which reside in the dorsal root ganglia (DRG), can be broadly divided into cutaneous and muscle afferents, which characteristically express receptors for various neural growth factors. The cutaneous afferents convey sensory information from the skin, and predominantly express receptors for NGF (TrkA) or BDNF (TrkB). The muscle afferents, which innervate muscle spindles, express the neurotrophin-3 (NT-3) receptor TrkC. As previously described, the muscle afferents make highly specific connections with motor neurons in individual motor pools during development, suggesting that molecular recognition mechanisms might be involved in the patterning of these connections. Lin et al. provide evidence that ETS proteins are also expressed in the sensory neurons, and that subpopulations of sensory neurons and their motor neuron targets share ETS protein expression patterns. Prior to the onset of the formation of monosynaptic connections between muscle sensory afferents and motor neurons, ∼70% of DRG sensory afferents that express these ETS proteins express both PEA3 and ER81. In addition, at this time some ETS-expressing sensory afferents do not show coexpression of TrkC, which is expressed on more mature muscle sensory afferents. As these sensory neurons begin to form monosynaptic connections with motor neurons, however, these expression patterns change dramatically. Most of these sensory afferents (∼90%) now express PEA3 or ER81 but not both ETS proteins, and all of these sensory neurons express TrkC and not TrkA. To address the relationship between the sensory afferents and motor neurons that express the same ETS proteins, Lin et al. made HRP injections into muscles to label retrogradely sensory and motor neurons that share the same peripheral muscle target. These experiments reveal that many groups of sensory afferents and motor neurons that contact the same muscle show a similar pattern of ETS expression. For example, ∼95% of the sensory afferents that innervate the A muscle express ER81, which is also expressed by the motor neurons that innervate the A muscle (Figure 1C). Similarly, ∼90% of the sensory afferents that innervate the ITR muscle express PEA3, which is also expressed by the motor pools that innervate the ITR muscle. This correlation is not exact, however, since a low but significant number of ETS+ sensory afferents contact ETS− motor neurons, and at least one ETS−/PEA3− motor pool receives ∼50% of its sensory afferent input from PEA3+ sensory afferents. While these exceptions indicate that additional factors must contribute to precise matching of sensory and motor neurons that define an individual motor unit, the general finding that there is a striking match between ETS expression profiles for certain motor pools and their sensory afferents strongly suggests that the specification of these connections is regulated by the expression of common ETS proteins between these neuronal populations. It will be important to determine whether or not other neurons that innervate specific motor pools, such as spinal inhibitory interneurons, also express ETS proteins. The expression of common ETS proteins in subsets of motor and sensory neurons suggests that the expression of these proteins may be coordinately regulated by a peripheral signal. Given that these neurons innervate common muscle targets, is it possible that both motor pool and muscle sensory afferent specification depends on target-derived signals? To address this issue Lin et al. performed limb ablation experiments between stages 16 and 20, prior to the onset of ETS protein expression and also the onset of axon outgrowth into the limb, but after the time when paraxial mesodermal signals have already defined motor column identity. The early (stage 16/17) hindlimb ablations result in a dramatic loss of PEA3 and ER81 expression in both sensory and motor neurons on the operated side of the spinal cord (not accounted for by cell death) indicating that a peripheral signal is required for ETS protein expression in these neuronal populations. Later (stage 19) limb ablations do not result in a loss of ER81 expression in the A motor pool indicating the motor axons and sensory afferents need to detect the peripheral signal only briefly for persistent ETS expression. It should be noted, however, that these experiments do not prove that a peripheral signal confers motor pool identity, since it is possible that the peripheral signal induces ETS protein expression in previously specified motor neuron pools. Examining ETS protein expression in animals where motor axons are forced to innervate ectopic muscles should help resolve this issue. While the limb ablation experiments suggest that a signal from the periphery is necessary for ETS protein expression in motor neuron pools and in sensory neurons, they do not reveal the specific location and identity of the peripheral signal. It is reasonable to speculate that the peripheral signal that specifies motor pool ETS expression should be produced by the target muscle, but cell biological experiments suggest that although the motor nerve innervates the peripheral muscle mass with great precision, the signal for nerve patterning may not be muscle derived. This possibility is suggested by somite removal and somite reversal experiments in which the motor nerve invades the limb mesenchyme and separates into dorsal and ventral branches although the muscle-derived signals are missing or aberrant (5Lance-Jones C Ciba Found. Symp. 1988; 138: 97-115PubMed Google Scholar, 12Phelan K.A Hollyday M J. Neurosci. 1990; 10: 2699-2716PubMed Google Scholar). Therefore, the signal responsible for the dorsal–ventral patterning of the nerve and muscle-specific innervation may be produced by the connective tissue in the developing limb or the overlying epidermis. It will be instructive to determine if ETS protein expression in individual motor pools is altered in such somite perturbation experiments since this would provide insight into the role of the muscle target in inducing ETS expression in the spinal cord. It will also be important to determine if the same peripheral signal regulates ETS protein expression by both the motor and sensory neurons. While the notion of a single target-derived signal specifying ETS expression in both populations is attractive, certain cell biological observations suggest that motor neurons and sensory neurons may respond to distinct signals. In general the development of sensory projections to the periphery appears to depend upon signals from the motor nerve. During development, the sensory nerve follows the motor nerve to appropriate muscle targets. If the ventral neural tube is removed from a chick embryo at stage 16, most of the motor neurons fail to form and the motor nerve is absent or severely reduced (9Landmesser L Honig M Dev. Biol. 1986; 118: 511-531Crossref PubMed Scopus (88) Google Scholar). In these animals the sensory projections to the muscles are grossly altered suggesting that a signal from the motor neurons may be required for the proper development of peripheral sensory projections. Examining the pattern of ETS protein expression in the sensory ganglia in animals where the motor pool is ablated or the motor nerve transected would be useful in determining the contribution of the motor nerve to the molecular specification of sensory neurons. While the motor nerve may be involved specifying the peripheral projection of sensory neurons, it appears that a signal from the peripheral musculature can specify the central projection pattern of the sensory neurons. In experiments in which the ventral muscles are replaced with dorsal muscles during limb development to create a double-dorsal limb, the ventral motor and sensory nerves innervate the ectopic dorsal muscle. In such manipulated animals, the central branch of the ectopically projecting sensory neurons makes synapses onto the LMC neurons that normally innervate the dorsal muscle. This suggests that the central projection of sensory neurons is specified in response to signals from the periphery (17Wenner P Frank E J. Neurosci. 1995; 15: 8191-8198PubMed Google Scholar). According to the model proposed by Lin et al., one would expect the pattern of ETS proteins to be respecified in sensory neurons innervating the ectopic dorsal muscle. Such a respecification would lend support to the idea that ETS protein expression may play an instructive role in specifying connections between the central axons of sensory neurons and their target motor pools. Also, examining the pattern of ETS protein expression in the motor neuron pools would allow one to determine if motor neuron pools can be respecified with regard to ETS expression. If such experiments demonstrate alterations in ETS expression that are diagnostic of neuronal respecification, they would support the notion that peripheral signals specify motor pool and sensory neuron identity and do not simply induce a prespecified differentiation program. While the findings of Lin et al. establish that a peripheral signal is capable of directing patterns of ETS protein expression in motor and sensory neurons, they leave open the identity of such a signal. Identification of these signals and understanding how such signals might regulate coordinated ETS protein expression in motor pools and subsets of sensory neurons should provide important mechanistic insight into the process by which neurons acquire their ETS protein profiles. The experiments described in the study by Lin et al., together with classic cell biological experiments, suggest a model in which a signal from the peripheral target induces the expression of particular ETS proteins in individual motor pools and subsets of sensory neurons. Once ETS protein expression has been induced, the sensory neurons extend their central axons ventrally and make specific connections with motor neurons in appropriate motor pools. Lin et al. suggest that homophilic cell surface interactions might provide for a selective matching of sensory axons and motor neurons that are part of the same circuit. Further, they note that there is emerging evidence implicating ETS proteins in controlling expression of cadherin genes and that at least one member of this family of homophilic cell adhesion molecules (CAMs) is expressed in a motor pool–specific pattern. It will be of interest to explore the possibility that ETS-regulated expression of a CAM by both sensory and motor axons might result in matched central connections.
ABSTRACT Regulation of directed axon guidance and branching during development is essential for the generation of neuronal networks. However, the molecular mechanisms that underlie interstitial axon branching in the mammalian brain remain unresolved. Here, we investigate interstitial axon branching in vivo using an approach for precise labeling of layer 2/3 callosal projection neurons (CPNs), allowing for quantitative analysis of axonal morphology at high acuity and also manipulation of gene expression in well-defined temporal windows. We find that the GSK3β serine/threonine kinase promotes interstitial axon branching in layer 2/3 CPNs by releasing MAP1B-mediated inhibition of axon branching. Further, we find that the tubulin tyrosination cycle is a key downstream component of GSK3β/MAP1B signaling. We propose that MAP1B functions as a brake on axon branching that can be released by GSK3β activation, regulating the tubulin code and thereby playing an integral role in sculpting cortical neuron axon morphology. HIGHLIGHTS - GSK3β activation induces excessive interstitial axon branching in excitatory cortical neurons - MAP1B, acting as a brake, is a downstream effector of GSK3β-mediated axon branching - MAP1B inhibition of axon branching is released by GSK3β phosphorylation - GSK3β/MAP1B regulation of interstitial axon branching is through modification of the tubulin code
Abstract From the initial stages of axon outgrowth to the formation of a functioning synapse, neuronal growth cones continuously integrate and respond to multiple guidance cues. To investigate the role of semaphorins in the establishment of appropriate axon trajectories, we have characterized a novel secreted semaphorin in grasshopper, gSema 2a. Sema 2a is expressed in a gradient in the developing limb bud epithelium during Ti pioneer axon outgrowth. We demonstrate that Sema 2a acts as chemorepulsive guidance molecule critical for axon fasciculation and for determining both the initial direction and subsequent pathfinding events of the Ti axon projection. Interestingly, simultaneous perturbation of both secreted Sema 2a and transmembrane Sema I results in a broader range and increased incidence of abnormal Ti pioneer axon phenotypes, indicating that different semaphorin family members can provide functionally distinct guidance information to the same growth cone in vivo.
Bone morphogenetic protein (BMP) signaling has emerged as an important regulator of sensory neuron development. Using a three-generation forward genetic screen in mice we have identified Megf8 as a novel modifier of BMP4 signaling in trigeminal ganglion (TG) neurons. Loss of Megf8 disrupts axon guidance in the peripheral nervous system and leads to defects in development of the limb, heart, and left-right patterning, defects that resemble those observed in Bmp4 loss-of-function mice. Bmp4 is expressed in a pattern that defines the permissive field for the peripheral projections of TG axons and mice lacking BMP signaling in sensory neurons exhibit TG axon defects that resemble those observed in Megf8 (-/-) embryos. Furthermore, TG axon growth is robustly inhibited by BMP4 and this inhibition is dependent on Megf8. Thus, our data suggest that Megf8 is involved in mediating BMP4 signaling and guidance of developing TG axons. DOI:http://dx.doi.org/10.7554/eLife.01160.001.
The past several years have seen a revolution in our understanding of how axons find their targets during neural development. These new insights are due in part to a melding of detailed cellular and genetic characterizations of axon guidance events with the molecular description of several families of phylogenetically conserved cues capable of mediating these steering decisions (reviewed by 5Mueller B.K. Annu. Rev. Neurosci. 1999; 22: 351-388Crossref PubMed Scopus (378) Google Scholar). Neurons use these cues, which include both attractants and repellents acting over short and long distances, to direct their axons to appropriate intermediate or final targets. Refinement of these projections is imparted by the ability of individual axons to respond to multiple guidance cues presented at different points along their trajectories. Further complexity in these early guidance events is provided by the bifunctionality of many of these guidance cues: attracting certain populations of axons and repelling others, or even attracting or repelling the same axon depending on the state of certain intracellular signaling molecules. Though it is likely that many families of guidance cues and their receptors remain to be discovered, the tools are now in hand to begin dissecting the molecular basis of attractive and repulsive guidance mechanisms. Two studies published in the June 25 issue of Cell, one from the Goodman laboratory (1Bashaw G.J. Goodman C.S. Cell. 1999; 97: 917-926Abstract Full Text Full Text PDF PubMed Scopus (167) Google Scholar) and the other a collaborative effort between the Tessier-Lavigne and Poo laboratories (3Hong K. Hinck L. Nishiyama M. Poo M.-M. Tessier-Lavigne M. Stein E. Cell. 1999; 97: 927-941Abstract Full Text Full Text PDF PubMed Scopus (573) Google Scholar), provide insight into the logic of how growth cones interpret guidance cues as being attractive or repulsive. Using complementary in vivo and cell culture approaches, and focusing on distinct but overlapping guidance receptor families, both groups demonstrate a key role for the cytoplasmic domains of guidance receptors in mediating attraction and repulsion. In addition3Hong K. Hinck L. Nishiyama M. Poo M.-M. Tessier-Lavigne M. Stein E. Cell. 1999; 97: 927-941Abstract Full Text Full Text PDF PubMed Scopus (573) Google Scholar provide evidence for a novel molecular mechanism whereby a heteromultimeric receptor complex can dictate whether the steering response to a single cue, netrin-1 (Net-1), is attractive or repulsive. Perhaps there is no better place to investigate attractive and repulsive guidance mechanisms than at the CNS midline (2Flanagan J.G. Van Vactor D. Cell. 1998; 92: 429-432Abstract Full Text Full Text PDF PubMed Scopus (31) Google Scholar). In both vertebrates and invertebrates, axons from specific populations of neurons are attracted toward the midline by long-range chemoattractants belonging to the Netrin family. Netrin attractive functions are mediated by receptors belonging to the DCC family, a branch of the immunoglobulin (Ig) superfamily that includes DCC in vertebrates and Frazzled (Fra) in Drosophila. Upon arriving at the midline, contralaterally projecting axons undergo a conversion; they cross the midline, lose responsiveness to midline-derived Netrin cues, and do not recross the midline. Recent analyses reveal that Slit proteins, also expressed on the midline, can act as repellents and are likely responsible for moving these contralaterally projecting axons across the midline and keeping them from recrossing (6Zinn K. Sun Q. Cell. 1999; 97: 1-4Abstract Full Text Full Text PDF PubMed Scopus (32) Google Scholar). Slit receptors are members of the Roundabout (Robo) family, also a branch of the Ig superfamily, but they differ from DCC proteins in the number of extracellular Ig and fibronectin domains and share no similarity over their large cytoplasmic domains. Using the well-characterized Drosophila midline1Bashaw G.J. Goodman C.S. Cell. 1999; 97: 917-926Abstract Full Text Full Text PDF PubMed Scopus (167) Google Scholar recognized a great opportunity to test in vivo two important and related questions: are Netrin and Slit receptors—Fra and Robo, respectively—modular such that their cytoplasmic domains are responsible for the nature of the growth cone response to a guidance cue; and are attractive and repulsive intracellular signaling components generally present in diverse cell types? The answer to both questions is yes. Chimeric Fra or Robo receptors, containing either the extracellular domain of Fra and the intracellular domain of Robo (Fra–Robo) or the extracellular domain of Robo and the intracellular domain of Fra (Robo–Fra), were expressed on all neurons in Drosophila embryos. This resulted in repulsive guidance responses to Netrins (Fra–Robo) and attractive responses to Slit (Robo–Fra) (see figure, panel A). For example, ectopic Fra–Robo directs axons that would normally cross the midline away from it, leading to a commissureless phenotype. In addition, Fra–Robo also directs motor axons away from the Netrin-expressing muscles that they would normally innervate. These effects are not likely to be due to dominant-negative effects of Fra–Robo. The chimeric Robo–Fra receptor produces complementary phenotypes; Slit can now function as a midline attractant. Remarkably, muscle precursors that normally migrate away from the midline in response to the Slit repellent can interpret Slit as an attractant when they ectopically express Robo–Fra. These results indicate that the attractive or repulsive nature of a particular guidance cue resides in the cytoplasmic domain of the receptor and that this response can be independent of binding a specific class of ligand. Further, a variety of neuronal and nonneuronal cells are shown to be capable of novel attractive and repulsive responses, demonstrating that the downstream signaling components necessary for Netrin and Slit guidance responses are present in different cell types. But what about guidance cues that are bifunctional—how can the same cue be both an attractant and a repellent? 3Hong K. Hinck L. Nishiyama M. Poo M.-M. Tessier-Lavigne M. Stein E. Cell. 1999; 97: 927-941Abstract Full Text Full Text PDF PubMed Scopus (573) Google Scholar directly address this issue. Genetic evidence in C. elegans and direct evidence in vertebrates have motivated the search for understanding the molecular basis of Netrin bifunctionality (reviewed by 5Mueller B.K. Annu. Rev. Neurosci. 1999; 22: 351-388Crossref PubMed Scopus (378) Google Scholar). Altering intracellular cyclic nucleotide levels in vertebrate neurons in vitro can convert an attractive Netrin response to a repulsive one in a DCC-dependent fashion. Though this shows that a single receptor can mediate both Netrin attraction and Netrin repulsion, work in C. elegans and in vertebrates demonstrates that UNC5 receptors, yet another branch of the Ig superfamily containing members distinct from both DCC and Robo proteins, are involved in Netrin-mediated repulsive guidance events and are also Netrin binding proteins. However, a simple model whereby DCC and UNC5 receptors independently signal Netrin attraction and repulsion is challenged by observations that UNC5 and UNC40 (a DCC family member) are both required for many Netrin-mediated repulsive guidance events in C. elegans. First3Hong K. Hinck L. Nishiyama M. Poo M.-M. Tessier-Lavigne M. Stein E. Cell. 1999; 97: 927-941Abstract Full Text Full Text PDF PubMed Scopus (573) Google Scholar directly show, by introducing the vertebrate UNC5H2 receptor into Xenopus spinal neurons grown in culture, that an UNC5 receptor is required cell-autonomously to elicit a repulsive response to Net-1. These spinal neurons express endogenous DCC, and previous work showed that antibody neutralization of DCC abolishes an attractive response to Net-1 (4Ming G.-L. Song H.-J. Berninger S. Holt C.E. Tessier-Lavigne M. Poo M.-M. Neuron. 1997; 19: 1225-1235Abstract Full Text Full Text PDF PubMed Scopus (495) Google Scholar). UNC5H2-dependent repulsion is also abolished by DCC neutralization, directly demonstrating a requirement for both UNC5 and DCC for Net-1 repulsion and strongly suggesting that the role of UNC5 is to convert an attractive response to a repulsive one. To understand how UNC5 effects this conversion, a series of chimeric and altered receptors were constructed and introduced into spinal neurons. Interestingly, Net-1 functions as a repulsive cue in the presence of DCC as long as a membrane-associated cytoplasmic domain of UNC5H2 is also present. A chimeric UNC5H2 consisting of a DCC ectodomain and an UNC5H2 cytoplasmic domain, a similar construct with a TrkA ectodomain, or simply the UNC5H2 cytoplasmic domain targeted to the inner plasma membrane by a myristoylation sequence are all capable of converting Net-1 attraction to repulsion in the presence of endogenous DCC (see figure, panel B). These data show that Net-1 need not bind the ectodomain of UNC5 to produce repulsion, and they suggest that the cytoplasmic domains of UNC5 and DCC form a receptor complex. This point is shown directly by an extensive series of coimmunoprecipitation experiments which demonstrate that this association is ligand dependent. Therefore, Net-1 binding to its receptor serves to overcome an inhibition of the association between DCC and UNC5 cytoplasmic domains to activate a molecular switch that signals repulsion. Since an UNC5 ectodomain is not required for this switch to occur, Net-1 binding is likely to induce an intramolecular conformational change in DCC that allows the UNC5 and DCC cytoplasmic domains to associate. Additional experiments show that Net-1 can mediate repulsion by binding to the ectodomains of either UNC5 or DCC, so long as both DCC and UNC5 cytoplasmic domains are present. Finally, extensive yeast two-hybrid analysis and in vitro competition experiments define conserved regions of both the DCC and the UNC5 cytoplasmic domains that appear to mediate this association in the absence of additional factors. Though this work defines certain minimal requirements for converting Netrin attraction to repulsion, it raises several important questions. What roles do UNC5 ectodomains play in Netrin responses, and do all Netrins within a species interact similarly with DCC and UNC5? Regulation of Netrin-mediated guidance may also result from modulatory events that serve to spatially segregate UNC5 from DCC proteins in those portions of an axon’s trajectory where attraction occurs. Future structure–function analyses of DCC and UNC5 extracellular domain associations with DCC, UNC5, Netrin, and possibly other families of proteins will begin to shed light on these issues. In addition, both the 1Bashaw G.J. Goodman C.S. Cell. 1999; 97: 917-926Abstract Full Text Full Text PDF PubMed Scopus (167) Google Scholar and 3Hong K. Hinck L. Nishiyama M. Poo M.-M. Tessier-Lavigne M. Stein E. Cell. 1999; 97: 927-941Abstract Full Text Full Text PDF PubMed Scopus (573) Google Scholar studies raise crucial questions about the nature of repulsive and attractive guidance mechanisms. Might the conversion of the sign of the response to other guidance cues also employ heteromultimeric receptor switches, such that making Semaphorins or Slits attractive is dependent upon the addition or loss of unidentified receptor components? And, finally, though alteration of cyclic nucleotide levels in the growth cone can convert attraction to repulsion and vice versa, does this mean that the signaling outputs for diverse guidance cues and their equally diverse receptors converge on only one or two common signaling pathways? Continued inwardly directed experimental reflection will undoubtedly address these questions.
