Keywords

4.1 Semaphorins Control the Radial Migration of Cortical Neurons

The six-layered mammalian neocortex contains two main types of neurons: (1) pyramidal neurons that primarily project outside the cortex and originate from the ventricular zone of the dorsal telencephalon, or pallium; and (2) interneurons, which make local connections between different layers or across layers and originate from the ganglionic eminence in the ventral telencephalon or subpallium (Rakic 2009; Bartolini et al. 2013). There is a variety of interneurons differing in morphology (dendritic arborization) and neurotransmitter content (Ascoli et al. 2008). Several studies have shown that semaphorins control the migration of both types of cortical neurons (Fig. 4.1). Pyramidal neurons are known to migrate along the processes of radial glia cells, which extend from the ventricular surface to the cortical surface (Metin et al. 2008). There is little dispersion of the pyramidal neurons during their radial migration, which at least partially explains the columnar organization of the cortex (Rakic et al. 2009).

Fig. 4.1
figure 1

Role of secreted semaphorins in cortical neuron migration. (a) In the embryonic forebrain of wild-type mice, cortical interneurons originate from the medial ganglionic eminence (MGE) and migrate tangentially to the cortex avoiding the striatum where Sema3A and Sema3F are expressed (red arrows). Sema3A is also expressed in the upper part of the cortical plate and promotes the radial migration of pyramidal neurons (blue triangles) from the subventricular/ventricular zone (SVZ/VZ) and controls their polarity. (b) Sema3A, neuropilin-1, and Plexin-A2 loss of function perturb the migration of both types of neurons. A large fraction of cortical interneurons invade the striatum. In the cortex, the radial migration of layer II/III neurons from the SVZ/VZ is affected and many neurons differentiate at an ectopic location. Their polarity and morphology are also abnormal

Full size image

Several studies showed that Sema3A is expressed in the superficial layers of the cortical plate (Bagnard et al. 1998; Polleux et al. 1998; Chen et al. 2007) in a decreasing gradient from the surface to the depth of the cortex. Cortical neurons also express the neuropilin-1 receptor during radial migration (Chen et al. 2007). In the embryonic day 16 (E16) mouse cortex, the silencing of neuropilin-1 expression using RNA interference or Sema3A overexpression strongly impaired the ability of layer II/II neurons to reach the cortical plate, although they could enter the intermediate zone and proliferated normally (Chen et al. 2007; Shelly et al. 2011) (Fig. 4.1). The same defects were seen using recombinase-mediated inactivation of neuropilin-1 in cortical neurons. Co-culture experiments mixing cortical slices and sema3A-expressing cells showed that Sema3A has a chemoattractive activity on radially migrating neurons (Chen et al. 2007). Neuropilin-2 loss of function also significantly perturbed radial migration. Similarly, the silencing of Plexin-A4 and Plexin-A2 prevented cortical neurons from leaving the intermediate zone, suggesting that they interact with neuropilin-1 to mediate Sema3A activity. The polarity of the ectopic neuropilin-1-deficient neurons was severely affected, and they adopted a multipolar morphology with multiple neurites rather than a bipolar morphology with a single and unbranched leading process oriented toward the cortical surface (Chen et al. 2007; Shelly et al. 2011). A role for Sema3A in the polarization of newly generated neurons has been also demonstrated for hippocampal neurons, which also migrate radially (Shelly et al. 2011). This step is essential to initiate radial migration (Namba et al. 2014), and therefore Sema3A could act first to polarize neurons and initiate migration and later to attract the neurons toward the surface. Interestingly, similar migration defects were also observed after inactivation of various molecules that are known to act downstream of Sema3A/neuropilin-1 such as the cycling-dependent kinase Cdk5 (Gupta et al. 2003), the Src Kynase Fyn (Sasaki et al. 2002), or Tag-1 (Law et al. 2008; Namba et al. 2014). In addition, Sema3A inhibits the activity of protein kinase A on the phosphorylation of GSK3β and LKB1, two kinases acting on radial migration and neuronal polarity (Asada et al. 2007; Asada and Sanada 2010). However, the exact mechanism of action of Sema3A remains unknown, and the apparently normal cortical layering in the Sema3A knockout (Catalano et al. 1998) suggests that Sema3A acts redundantly with other factors. Sema3A not only controls the migration of pyramidal neurons but also regulates the growth of their apical and basal dendrites and guides their axons in the white matter (Polleux et al. 1998, 2000; Sasaki et al. 2002; Zhou et al. 2013). Surprisingly, much less is known about the function of transmembrane semaphorins in the migration of cortical neurons. A recent study (Azzarelli et al. 2014) revealed that Plexin-B2 is required for radial migration. In the mouse embryo, Plexin-B2 is expressed in the ventricular zone, subventricular zone, and cortical plate. ShRNA-mediated knockdown of Plexin-B2 prevents most neurons from leaving the ventricular zone/subventricular zone (VZ/SVZ) and perturbs the acquisition of cortical neuronal morphology. It was shown that the small GTP-binding protein, Rnd3, which inhibits RhoA signaling through p190RhoGAP (Wennerberg et al. 2003; Pacary et al. 2013), acts downstream of Plexin-B2 during radial migration, confirming previous studies that had shown that Rnds are partners of type-B plexins (Oinuma et al. 2004; Azzarelli et al. 2014). Plexin-B2 inhibits Rnd3, which results in the activation of RhoA activity. The antagonistic activity of Plexin-B2 and Rnd3 allows stabilizing the level of active RhoA during radial migration. Several class 4 semaphorins bind to Plexin-B2, and Sema4D enhances cortical neuron motility in vitro (Hirschberg et al. 2010). However, a role for transmembrane semaphorins in radial migration has not yet been confirmed in vivo. Worthy of note, the organization of the embryonic cortex is severely disturbed in the constitutive Plexin-B2 knockout, suggesting that Plexin-B2 function in cortical development extends beyond radial migration and that it also acts on cell proliferation and neuronal differentiation (Hirschberg et al. 2010).

4.2 Semaphorin Function in the Tangential Migration of Cortical Neurons

Most inhibitory interneurons [using gamma-aminobutyric acid (GABA) as a neurotransmitter] are produced from progenitors located in the medial ganglionic eminence that migrate tangentially (between E12 and E16 in the mouse), independently of radial glia fibers, across the basal forebrain and then colonize the cortical wall (de Carlos et al. 1996; Anderson et al. 1997; Tamamaki et al. 1997). In the subpallium, cortical interneurons avoid the striatum where Sema3A and Sema3F are highly expressed (Fig. 4.1). During their migration, GABAergic interneurons express neuropilin-1 or neuropilin-2 receptors and are repelled from the nascent striatum by Sema3A and Sema3F. In the absence of neuropilin-1 or neuropilin-2 signaling, many interneurons enter the striatum (Marin et al. 2001; Tamamaki et al. 2003; Zimmer et al. 2010). Ectopic interneurons are also detected after silencing Plexin-A1 in cortical interneurons (Andrews et al. 2013). Lim kinases (Limk) control actin dynamics through cofilin and were previously seen to be involved in semaphorin signaling (Aizawa et al. 2001; Scott et al. 2009). Limk2 was detected in cortical interneurons, and loss-of-function studies using electroporation of Limk2 siRNA in forebrain slices and medial ganglionic eminence (MGE) suggest that it acts downstream of Sema3A/neuropilin-1/Plexin-A1 during migration (Andrews et al. 2013).

Migrating interneurons also express Roundabout 1 (Robo1), one of the receptors of Slit repellents, but Slits are not required for cortical interneuron migration in the subpallium (Marin et al. 2003). However, in Robo1 knockout mice, some GABAergic interneurons abnormally enter into the striatal anlage, and their density is also increased in the cortex (Hernandez-Miranda et al. 2011), a phenotype not observed in Slit knockouts (Marin et al. 2003; Andrews et al. 2008). It was recently shown that Robo1-deficient interneurons are less responsive to Sema3A/3F repulsion in migration assays and express a lower level of neuropilin-1 and Plexin-A1. This result suggests that there could be crosstalk between the semaphorin/plexin and Slit/Robo pathways in migrating neurons. Accordingly, Robo1 could be co-immunoprecipitated with neuropilins, suggesting that they are in the same receptor complex. Moreover, chondroitin sulfate proteoglycans (CSPGs), which bind to many extracellular matrix components, regulate the spatial distribution of Sema3A in the basal forebrain and potentiate its repulsive activity for migrating cortical interneurons (Zimmer et al. 2010).

Interestingly, neuropilin-2 expression in striatal interneurons is repressed by the transcription factor Nkx2.1 (Nobrega-Pereira et al. 2008), which directly binds to regulatory elements in the neuropilin-2 promoter. In Nkx2.1 knockout, neuropilin-2 is upregulated and there is a reduction in the number of striatal interneurons (Nobrega-Pereira et al. 2008). Therefore, the differential expression of neuropilin-2 by MGE-derived interneurons controls their cortical versus striatal fate. The Dlx1/2 TFs, which control MGE development, are also candidate repressors of neuropilin-2, but the possible influence of this interaction on cortical interneuron migration is unknown (Le et al. 2007).

Neuropilin-2 was also involved in the migration of another type of forebrain neurons, the “LOT cells.” The lateral olfactory tract (LOT) contains axons projecting from the olfactory bulb to the pyriform cortex and conveys olfactory inputs. The LOT extends caudally along a region of the forebrain at the interface between the pallium (cortex) and subpallium (MGE) (Sato et al. 1998). It was shown that LOT axons follow a path that is delineated by a specific set of cells, called the LOT cells. It was initially thought that the LOT cells were acting as guidepost cells for LOT axons, but these can be misguided even if the LOT cells are properly positioned, suggesting that these are not sufficient for guiding LOT axons (Fouquet et al. 2007). Cell-labeling studies showed that LOT cells are among the first to be born in the dorsal telencephalon and that they migrate tangentially (between E9.5 and E11.5 in the mouse) in a ventral direction to their final position at the surface of the forebrain (Tomioka et al. 2000; Kawasaki et al. 2006). Sema3F (and not other secreted semaphorins) has a repulsive activity on migrating LOT cells, mediated by neuropilin-2 (Ito et al. 2008). In the final phase of LOT cell migration, Sema3F is expressed in the subpallium in a domain that is surrounding the LOT cells. In neuropilin-2- and Sema3F-knockout embryos, LOT cells migrate more deeply in the forebrain rather than being confined to the surface. Together, these data suggest that neuropilin-2/sema3F act at the end of LOT cell migration to confine them to their superficial position. In parallel, other guidance cues such as Netrin-1 are controlling the dorsoventral patterning of the LOT cells (Kawasaki et al. 2006).

As illustrated by these examples, semaphorins have an important function in the control of neuronal migration in the neocortex, suggesting that defects in semaphorin signaling might be linked to various human neurological diseases in which ectopic or mislocalized neurons were detected, such as schizophrenia and autism spectrum disorders. This hypothesis is supported by preliminary studies indicating increase of Sema3A expression in the cerebellum of schizophrenic patients (Eastwood et al. 2003) and the existence of single-nucleotide polymorphisms (SNPs) in a Japanese cohort of schizophrenics (Fujii et al. 2011). Similarly, there is also evidence for SNPs in Plexin-A2 (Mah et al. 2006; Allen et al. 2008) associated with schizophrenia, but this is debated (Fujii et al. 2007).

4.3 Semaphorins and the Migration of GnRH Neurons

Gonadotropin-releasing hormone (GnRH) has a key role in reproduction by controlling the production of follicle-stimulating hormone (FSH) and luteinizing hormone (LH) by the pituitary gland. Hypothalamic neuroendocrine cells secreting GnRH originate outside the central nervous system (CNS) in the nasal placode (Fig. 4.2) and migrate (from about E11 in the mouse) into the median eminence in the central nervous system along olfactory/vomeronasal axons (Cariboni et al. 2007b). This particular type of neuronal migration is called axophilic (Schwanzel-Fukuda and Pfaff 1989). The abnormal migration of GnrH neurons, such as in Kallmann syndrome, causes a depletion of GnRH neurons, leading to hypogonadism and infertility. Many Kallmann patients are also anosmic. So far, ten causative genes have been identified, only accounting for about one third of Kallmann cases: KAL1, which is X-linked and encodes Anosmin-1 (Legouis et al. 1991), fibroblast growth factor 8 (FGF8) and its receptor FGFR1 (Falardeau et al. 2008), HS6ST1 (Tornberg et al. 2011), WDR11 (Kim et al. 2010), CHD7 (Jongmans et al. 2009), Prokineticin-2 (PROK2, and its receptor PROKR2) (Dode et al. 2006; Pitteloud et al. 2007), and more recently SEMA3A (Hanchate et al. 2012). In mouse and human, olfactory and vomeronasal axons express neuropilin-1, neuropilin-2, and Plexin-A1 during GnRH neuron migration (Giger et al. 1996; Murakami et al. 2001; Cloutier et al. 2002; Cariboni et al. 2007a, b; Hanchate et al. 2012). Sema3A and Sema3F are found in the olfactory epithelium and vomeronasal organ (Cloutier et al. 2002). GnRH cells also express neuropilin-1/2, Plexin-A1 mRNA, Sema3A, and Sema3F (Cariboni et al. 2007a).

Fig. 4.2
figure 2

Control of GnRH neuron migration by secreted semaphorins. (a) In the wild-type embryo, GnRH neurons (red) migrate along vomeronasal axons (blue) and olfactory axons (black) to the basal forebrain and hypothalamus. Sema3A and Sema7A are expressed along their migration pathway. (b) In neuropilin and Sema3A knockouts, the growth of olfactory axons and vomeronasal axons is perturbed, which results in the misrouting of GnRN neurons. Abbreviations: VNO vomeronasal organ, OE olfactory epithelium, Hypo hypothalamus, MOB main olfactory bulb, AOB accessory olfactory bulb

Full size image

Many studies suggest that semaphorin/neuropilin are required for guiding vomeronasal/olfactory axons to the CNS. In the Sema3A knockout, olfactory axon projections to the olfactory bulb are disorganized (Schwarting et al. 2000) and most GnRH neurons accumulate ectopically close to the olfactory bulb (OB) (Cariboni et al. 2011b), which likely explains hypogonadism in adult Sema3A −/− (Cariboni et al. 2011b; Hanchate et al. 2012). In Neuropilin-1 Sema/Sema knock-in mice, a large fraction of olfactory axons fails to enter the OB, and vomeronasal axons are also misrouted (Cariboni et al. 2011b). In these mice, about half the GnRH neurons still migrate along these axons and, as a result, settle in the cortex and thalamus instead of the hypothalamus (Fig. 4.2). Puberty is delayed in Neuropilin-1 Sema/Sema knock-in mice, and fertility is severely affected; this is also true for Neuropilin-2 knockout mice, which have reduced gonadal size (Cariboni et al. 2007a). Their phenotypic analysis revealed that the number of GnRH neurons in the hypothalamus is significantly reduced (Cariboni et al. 2007a) and that many migrate into the dorsal telencephalon or accumulate at the nasal septum. In Sema3F knockouts, vomeronasal axons defasciculate at the surface of the CNS (Cloutier et al. 2002, 2004), but GnRH migrate normally (Cariboni et al. 2011b). Surprisingly, in this system, Sema3A seems to signal through neuropilin-1 and neuropilin-2 as a Neuropilin-1 Sema/Sema; Neuropilin-2 double-mutant phenocopy of Sema3A knockouts (Cariboni et al. 2011b).

Interestingly, exon sequencing in Kallmann patients identified families with heterozygous nonsense and missense mutations in the SEMA3A gene (Hanchate et al. 2012; Young et al. 2012). These mutations either prevent Sema3A from being produced or result in the secretion of an inactive Sema3 protein (Hanchate et al. 2012; Young et al. 2012). It was also proposed that Sema3A heterozygous mutations alone are not sufficient to induce Kallmann syndrome but that it acts synergistically with other proteins or pathways also mutated in the patients (Hanchate et al. 2012). Together, these observations suggest that the inability of GnRH to enter the brain when semaphorin/neuropilin signaling is altered is probably secondary to a failure of olfactory axons to penetrate into the olfactory bulb.

However, there is also evidence for a cell-autonomous action of neuropilin-1 in GnRH neurons. In chemotactic assays, Sema3A and Sema3F have a repulsive activity on the migration of the GnRH cell line (GN11) whereas vascular endothelial growth factor (VEGF) has an attractive activity (Cariboni et al. 2007a). VEGF also controls their survival (Cariboni et al. 2011a). Interestingly, both activities appear to be mediated by neuropilin-1, at least in vitro.

The phenotypic analysis of Neuropilin-1 null embryos (Cariboni et al. 2007a) revealed a severe decrease of the number of GnRH neurons in the forebrain, which could be explained by a combination of increased cell death and abnormal migration. This reduction was also observed following conditional deletion of neuropilin-1 in neuronal precursors (Cariboni et al. 2011a). However, it was also reported that the conditional inactivation of neuropilin-1 in GnRH neurons (using a GnRH:cre line) does not perturb their migration (Hanchate et al. 2012), which would argue in favor of the non-cell-autonomous model of action of neuropilin-1 in GnRH neurons.

Transmembrane semaphorins have also been involved in the migration of GnRH neurons. The GPI-linked Sema7A is expressed along the migratory pathway followed by GnRH neurons to the olfactory bulb. GnRH neurons express two of its receptors, integrin-β1 and Plexin-C1, and their migration is altered in Sema7A knockout mice, which have fertility defects (Messina et al. 2011). GnRH neurons and olfactory axons also express Plexin-B1 during their migration, and its ligand Sema4D promotes, at least in vitro, the migration of GnRH cells by coupling B1 to methylethyltryptamine (N-methyl, N-ethyltryptamine, MET), a tyrosine kinase receptor for hepatocyte growth factor (HGF) (Giacobini et al. 2008). The migration of GnRH neurons is perturbed in Plexin-B1 knockouts although the olfactory projections look normal. However, GnRH neurons are not affected in Sema4D knockouts, suggesting that other class 4 semaphorins might act redundantly. Interestingly, at later stages, semaphorin signaling (such as 4D/B1) regulates cellular rearrangements that accompany the cyclic production of the pituitary hormones (see Messina and Giacobini 2013, for review).

4.4 When Semaphorins Restrain Neuronal Migration at CNS Boundaries

In the hindbrain and spinal cord, several studies showed that semaphorins act at boundaries between CNS compartments to restrict cell migration.

Boundary cap (BC) cells are neural crest cell derivatives localized at the motor axon exit points at the border between the CNS and PNS in vertebrate embryos (Vermeren et al. 2003). Following genetic or surgical ablation of BC cells, motor axons exit the brain at their normal position, but some motor neuron cell bodies migrate out of the CNS inside the motor nerves, which suggests that BC cells constitute a repulsive barrier for motor neurons (Fig. 4.3). It was later found that in mouse and chick embryos BC cells express Sema6A, Sema3B, and Sema3G (Bron et al. 2007; Mauti et al. 2007). Motor neurons express neuropilin-2 (a Sema3B receptor) and Plexin-A2. In neuropilin-2 knockouts, a significant fraction of the motor neurons enter the nerve roots (Bron et al. 2007), but this is not seen when sema3B is inactivated, suggesting that BCs express another neuropilin-2 ligand. Motor neuron emigration in the peripheral nervous system (PNS) was also observed following silencing of Sema6A expression in BC cells and of its receptor plexin-A2 in motor neurons (Bron et al. 2007).

Fig. 4.3
figure 3

Semaphorins and boundary cap cells. (a) Schematic representation of the embryonic spinal cord. Motor neurons (green) are localized in the ventral part of the spinal cord and send their axons outside the cord. (b) Region outlined in (a). In the wild type, motor neurons (MN) express Plexin-A2 and their cell bodies are prevented from leaving the spinal cord by semaphorins, expressed by boundary cap cells (BCC). (c) After Sema6A or Plexin-A2 loss of function, some MN exit the spinal cord and migrate inside motor nerves

Full size image

During early development, the hindbrain is transiently segmented into cellular units called rhombomeres (Lumsden and Keynes 1989; Kiecker and Lumsden 2005). Each rhombomere expresses specific combinations of transcription factors and contain progenitors that generate various neuronal populations. There is little cell dispersion and mixing between neighboring rhombomeres, and specialized cells, with distinct molecular properties and that do not produce neurons, delineate rhombomere boundaries (Lumsden and Krumlauf 1996). In zebrafish embryos, a group of neurons expressing fibroblast growth factor 20a (fbfg20a) occupy the center of each rhombomere, where it represses neuronal differentiation (Gonzalez-Quevedo et al. 2010). Fgf20a neurons were found to express neuropilin-2 whereas boundary cells express its ligands, Sema3fb and Sema3gb (Terriente et al. 2012). The knockdown of either neuropilin-2 or Sema3fb/Sema3gb leads to the dispersion of Fgf20 neurons inside the rhombomere and increased neurogenesis; this suggests that Sema3fb/Sema3gb, secreted by boundary cells, repels fgf20 cells and confines them to the center of each rhombomere, thereby influencing neurogenesis.

Neuropilin-1 also influences neuronal migration between rhombomeres. In the mouse, neurons that will form the facial motor nucleus are generated in rhombomere 4 but migrate caudally to settle in rhombomere 6 (Studer et al. 1996). In neuropilin-1-null mice, the tangential migration of facial motor neurons is compromised and most fail to reach rhombomere 6 (Schwarz et al. 2004). In this case, it was found that VEGF, and not class 3 semaphorins, is the neuropilin-1 ligand.

4.5 Semaphorins and Plexins Control the Migration of Granule Cells in the Cerebellum and Olfactory Bulb

In the developing cerebellum of most vertebrates, granule cells (GCs) derive from precursors localized in the external granular layer (EGL), which covers the surface cerebellar cortex (Chedotal 2010). In mice, GCs are generated during the first 3 postnatal weeks. They first migrate tangentially in the EGL and then radially across the molecular layer to reach the internal granule cell layer (IGL) (Ramon and Cajal 1911; Altman 1972) (Fig. 4.4).

Fig. 4.4
figure 4

Control of cerebellar granule cell migration by transmembrane semaphorins. (a) In newborn wild-type mice, granule cell progenitors (in purple) proliferate exclusively in the upper external granule cell layer (uEGL) and express Plexin-B2. Postmitotic granule cells start differentiating and migrate tangentially in the lower EGL (lEGL). During this phase, they express Sema6A and Plexin-A2. Next, they migrate radially through the molecular layer (ML) and the Purkinje cell layer (PCL) to the internal granule cell layer (IGL). Sema6A is downregulated at the onset of radial migration. At this stage Purkinje cells (in blue) are still distributed in multiple layers. In adult mice, Purkinje cells are aligned in a monolayer, above differentiated granule cell bodies all localized in the inner granular layer (IGL). (b) In Plexin-B2 mutants, some ectopic granule cells are found at the top of the molecular layer and the cerebellar cortex is fragmented. Islands of Purkinje cells (arrow) are embedded in differentiated granule cells. (c) In Sema6A and Plexin-A2 knockouts, about half the granule cells fail to migrate radially and are found in the molecular layer

Full size image

Previous studies showed that during their tangential migration GCs express the transmembrane semaphorin Sema6A (Kerjan et al. 2005), which is downregulated at the onset of radial migration. Migrating GCs also express Plexin-A2, one of the receptors for Sema6A (Renaud et al. 2008). Phenotypic analysis of Sema6A knockout and PlexinA2 knockout mice revealed that migration from the EGL of about half of the GC population is aborted and that their cell body remains trapped inside the molecular layer (Kerjan et al. 2005; Renaud et al. 2008). Time-lapse studies of GC migration in postnatal EGL explants showed that, in both Sema6A and PlexinA2 knockouts, the migration defects are probably related to an anomaly of nuclear translocation (Renaud et al. 2008). In Sema6A knockouts, the centrosome appears unable to detach from the nucleus, whereas in Plexin-A2 knockouts, the distance separating the centrosome from the nucleus becomes significantly increased. Therefore, in both cases, an abnormal coupling of the nucleus and centrosome might block migration. However, this could just illustrate a more global disorganization of the migration machinery, in particular of the actin dynamics. In the current working model, Sema6A–Plexin-A2 interaction is thought to have a role in the control of the switch from a tangential to radial mode of migration (Chédotal 2010) as this is correlated with the downregulation of Sema6A expression. The analysis of GC migration in mouse chimeras combining the wild type and either Plexin-A2 or Sema6A knockouts suggests that Sema6A is a ligand for Plexin-A2 in migrating GCs (Renaud et al. 2008). In vitro assays have also shown that GC migration is impaired on Sema6A-expressing cells (Janssen et al. 2010), suggesting that Sema6A could repel GCs away from the EGL. However, it might promote cell motility rather than simply inhibiting tangential migration. In vitro data suggest that the guanine nucleotide exchange factor (GEF) trio might act downstream of Plexin-A2/Sema6A (Peng et al. 2010). Rac1 might also be involved, because the conditional inactivation of Rac1 in migrating GCs somehow phenocopies the defects observed in Sema6A and Plexin-A2 knockouts (Tahirovic et al. 2010). As Sema6A and Plexin-A2 are also coexpressed in GCs, this suggests that they could also signal in cis in tangentially migrating GCs, as described for Sema6A and Plexin-A4 in sensory axons (Haklai-Topper et al. 2010).

During their proliferation in the upper part of the EGL, GC precursors express Plexin-B2, which is downregulated in postmitotic GCs entering the lower EGL (Fig. 4.4). Therefore, there is a correlation between Plexin-B2 extinction and the initiation of tangential migration and cell-cycle exit (Friedel et al. 2007). In Plexin-B2 knockout mice, GC differentiation is initiated and the GCs migrate away from the cerebellar surface along radial glia. However, mutant GCs do not stop their proliferation and keep dividing outside the upper EGL, within the molecular layer, and even in the IGL, which profoundly disorganizes the structure of the adult cerebellar cortex (Deng et al. 2007; Friedel et al. 2007). The Plexin-B2 mechanism of action and its physiological ligand(s) in the developing cerebellum are still unknown, although genetic and in vitro evidence suggests that Sema4C and Sema4G are involved (Maier et al. 2011). Both semaphorins are expressed in migrating GCs; they bind to Plexin-B2, and in Sema4C/sema4G double knockouts a subset of GCs does not migrate properly. However, the cerebellar defects are much more severe in Plexin-B2 knockouts, supporting the existence of other ligands.

Interestingly, Plexin-B2 is also expressed in the postnatal and adult brain in the precursors of olfactory bulb (OB) granule cells (Saha et al. 2012). These interneurons are generated throughout life from neural stem cells located in the subventricular zone lining the cerebral ventricles. These cells produce neuroblasts that migrate tangentially to the OB along the rostral migratory stream (RMS) (Lois and Alvarez-Buylla 1994; Sanai et al. 2011). These cells differentiate and switch to a radial mode of migration on entering the OB (Bovetti et al. 2007; Snapyan et al. 2009). As in the cerebellum, the downregulation of Plexin-B2 expression is correlated with the final cell division and the initiation of the radial migration of postmitotic granule cells. However, in this case, Plexin-B2 promotes the proliferation of SVZ neuroblasts in addition to acting on GC migration (Saha et al. 2012). It also acts on periglomerular cells, the second type of OB interneurons generated by SVZ neuroblasts. RMS neuroblasts migrate more rapidly in Plexin-B2 knockout than in wild-type mice. Similarly, macrophages and dendritic cells purified from Plexin-B2 knockout mice migrate more rapidly (Roney et al. 2011), suggesting that in various cell types, including cerebellar cortical and OB neurons, Plexin-B2 could provide a molecular tuning mechanism allowing migrating cells to control migration/proliferation choices at specific locations and time points. Again, the Plexin-B2 ligands involved in this process remain to be characterized.

4.6 Semaphorins Control the Migration of Oligodendrocyte Precursors During Myelination and Remyelination

Oligodendrocytes are the myelinating cells in the central nervous system of jawed vertebrates (Zalc et al. 2008). These cells extend processes that wrap around the axon and form the myelin sheath, interrupted at the level of the nodes of Ranvier where voltage-gated sodium channels accumulate, which allows the rapid saltatory conduction of action potentials (Sherman and Brophy 2005 for a review).

Oligodendrocytes are generated across the CNS from precursors (or OPCs, oligodendrocyte precursors) localized in multiple neuroepithelial foci of the ventricular zone (Pringle and Richardson 1993; Vallstedt et al. 2005). Starting at embryonic ages, OPCs migrate from the ventricular zone to colonize the CNS. In rodents, OPCs start to produce postmitotic oligodendrocytes after birth, but myelination continues for several weeks. Moreover, some OPCs are still present in the adult CNS and produce new oligodendrocytes throughout life. In demyelinating diseases, such as multiple sclerosis (MS), the myelin sheath is destroyed and oligodendrocytes die, leading to myelin-poor areas called plaques. OPCs also exist in the adult CNS and can proliferate to generate new oligodendrocytes in response to injury (Franklin and Ffrench-Constant 2008). These adult OPCs appear to migrate into demyelinated regions, and their recruitment is required for myelin repair. The migration and proliferation of OPCs during development is influenced by trophic/growth factors such as platelet-derived growth factor (PDGF) and fibroblast growth factor 2 (FGF-2). However, mounting evidence indicates that axon guidance molecules, including secreted semaphorins, also pattern their migration.

OPCs invade the optic nerve (ON) from the preoptic area in the brain, around 2 weeks of gestation, and reach the optic nerve head before birth (Spassky et al. 2002). In mammals, OPCs do not enter the retina. In mouse and rat embryos, ON nerve OPCs express neuropilin-1 and neuropilin-2 (Sugimoto et al. 2001; Spassky et al. 2002) as well as postnatal rat OPCs (Cohen et al. 2003). Neuropilin-1 is still expressed in mature oligodendrocytes (Ricard et al. 2001), unlike neuropilin-2 (Xiang et al. 2012). In three-dimensional (3D) collagen gel cultures, OPCs migrating from ON explants are repelled by cells secreting Sema3A, and this repulsive activity is blocked by anti-neuropilin-1 antibodies (Sugimoto et al. 2001; Spassky et al. 2002). By contrast, OPCs are attracted by cells secreting Sema3F via a receptor complex comprising neuropilin-2 (Spassky et al. 2002) and Plexin-A3 (Xiang et al. 2012). These chemotropic activities appear specific, as migrating OPCs do not respond to gradients of soluble Sema3C or Sema3E. However, postnatal OPCs avoid substrate-bound Sema3C and Sema3B (Cohen et al. 2003). Sema3A is enriched in cells that surround the ON and might prevent OPCs from leaving the ON. Sema3F, which is expressed by retinal ganglion cells, might attract OPCs toward their axons. Moreover, oligodendrocytes themselves express several secreted and transmembrane semaphorins (Cohen et al. 2003; Moreau-Fauvarque et al. 2003; Bernard et al. 2012), which might also act influence cell–cell interaction during oligodendrocyte migration. For instance, Sema4F is a transmembrane semaphorin (Encinas et al. 1999) expressed by OPCs in the mouse CNS (Armendariz et al. 2012). In vitro migration assays suggest that Sema4F inhibits the migration of ON OPCs, possibly in a paracrine/autocrine manner. Of note, direct in vivo support for these models is still lacking as neither the distribution of oligodendrocytes nor myelination seems to be perturbed in Sema3A, Sema3F, or neuropilin-1/2 knockouts (Kitsukawa et al. 1997; Taniguchi et al. 1997; Chen et al. 2000; Giger et al. 2000). A recent study suggests that the distribution of OPCs is uneven in the neocortex of newborn Plexin-A4 knockout mice (Okada and Tomooka 2012). Plexin-A4 forms, with neuropilin-1, a receptor complex for Sema3A, but it is also a receptor for Sema6A. Sema6A is expressed during myelination by postnatal oligodendrocytes and could therefore influence the migration of OPCs. Accordingly, 3T3 cells expressing Sema6A have a repulsive activity for cells of an OPC line, which is mediated by Plexin-A4 (Okada et al. 2007; Okada and Tomooka 2013). However, these cells also express Sema6A and are repelled by Plexin-A4, suggesting that Sema6A might act as a receptor in OPCs. Although the analysis of Sema6A knockout revealed a myelination delay of CNS white matter tracts, including the ON, the recruitment of oligodendrocytes to these axonal tracts does not seem to be altered (Bernard et al. 2012), which would support a ligand function of Sema6A in OPC migration, if any.

Recent data suggest that secreted semaphorins might also influence the remyelination of demyelinating lesions by acting on the migration of adult OPCs. It has been proposed that impaired remyelination in chronic MS lesions is linked to a poor recruitment of OPCs to the plaques. Adult OPcs still express neuropilins and Plexin-A1/-A3 receptors, suggesting that these and OPCs could still be attracted or repelled by Sema3F and Sema3A, respectively (Piaton et al. 2011). It was shown that the expression of Sema3A and Sema3A transcripts is enriched and upregulated in neurons and glia (mostly astrocytes) at active inflammatory MS lesions and in demyelinating lesions induced by lysophosphatidylcholine (LPC) in the rat spinal cord (Williams et al. 2007). Therefore, abnormal semaphorin expression might perturb the ability of OPCs and new oligodendrocytes to disseminate inside the lesion, opening new therapeutic perspectives for MS treatment. A first step toward the in vivo validation of this strategy has been obtained in LPC mouse models in which lentiviral vectors expressing Sema3A or Sema3F were injected into the demyelinated lesion. Sema3F overexpression increased significantly the number of OPCs at the lesion whereas Sema3A decreased it (Piaton et al. 2011). In addition, OPC recruitment after LPC lesions was impaired in neuropilin-1 Sema/Sema knock-in mice, which are unable to bind Sema3A but still bind vascular endothelial growth factor (VEGF) (Gu et al. 2003). Although another study has confirmed the inhibitory action of Sema3A on remyelination in a model of ethidium bromide-induced demyelination (Syed et al. 2011), it was suggested that this was primarily through an action on OPC differentiation rather than on their migration. In conclusion, the physiological function of semaphorins in myelination and remyelination is still an open question. Another transmembrane semaphorin, Sema5A, is expressed by cells of the oligodendrocyte lineage (Goldberg et al. 2004; Hilario et al. 2009), as well as its receptor Plexin-B3 (Artigiani et al. 2004). However, no myelin defects were detected in Plexin-B3 knockout mice (Worzfeld et al. 2009). By contrast, Sema5A was found to inhibit the motility of glioma cells through Plexin-B3 and the inactivation of Rac1. Sema5A also promotes the glial differentiation of glioma cells, suggesting that a downregulation of Sema5A expression, as observed in human astrocytomas, might influence the dissemination of tumor cells within the CNS (Li and Lee 2010; Li et al. 2012).