INTRODUCTION

Although the mechanisms that guide neurons during migrationare not well understood, the existence of restricted migratorypathways suggests that neuronal migration is under the controlof specific guiding cues (Anderson et al., 1997; Soriano et al.,1997; Wichterle et al., 1997; Pearlman et al., 1998). Inaddition, the heterogeneity of migratory routes followed bymigratory neurons that share their origin in a given germinalmatrix indicates that distinct migratory neurons may responddifferentially to similar extracellular signals.

Netrins are a family of laminin-related secreted proteins thatmay act either as chemoattractants or as chemorepellents fordistinct developing axons (Kennedy et al., 1994; Serafini et al.,1994; Colamarino and Tessier-Lavigne, 1995). Netrin 1 isresponsible for the attraction of commissural axons towards themidline in the spinal cord (Kennedy et al., 1994; Serafini et al.,1996), although these axons also require the interaction of atleast two adhesion molecules, Axonin-1/TAG-1 and Ng-CAM/L1, to reach their targets (Stoeckli and Landmesser,1995). DCC (deleted in colorectal cancer), a neural celladhesion molecule of the Ig superfamily (Ig-CAMs), is acomponent of the receptor complex that mediates the

chemoattractive response to netrin 1 (Keino-Masu et al., 1996;Fazeli et al., 1997).

Loss-of-function mutations of either netrin 1 (Ntn1) ordeleted in colorectal cancer (Dcc) in mice confirm theirinvolvement in the formation of commissural connections(Serafini et al., 1996; Fazeli et al., 1997). A second DCC-related protein, neogenin, also binds netrin 1 but its functionremains unknown (Keino-Masu et al., 1996). In addition, netrin1 also acts as a chemorepulsive cue for the axons of thetrochlear, trigeminal, facial and glossopharyngeal motor nuclei(Colamarino and Tessier-Lavigne, 1995; Varela-Echavarria etal., 1997). Such chemorepulsive action is probably notmediated by DCC, but by a second family of netrin 1 receptors,the unc5-related proteins (Leung-Hagesteijn et al., 1992;Hamelin et al., 1993; Leonardo et al., 1997).

During the development of the nervous system, considerabledistances separate the final destination of migratory neuronsfrom their proliferative matrices. Recent evidence, in bothvertebrates and invertebrates, suggests that some cuesgoverning axonal growth may also be involved in neuronalmigration (Culotti and Kolodkin, 1996; Goodman, 1996; Huand Rutishauser, 1996; Tessier-Lavigne and Goodman 1996).For instance, the lack of pontine nuclei and the severe atrophy

1359Development 127, 1359-1372 (2000)Printed in Great Britain © The Company of Biologists Limited 2000DEV9687

Netrin 1 is a long-range diffusible factor that exertschemoattractive or chemorepulsive effects on developingaxons growing to or away from the neural midline. Herewe used tissue explants to study the action of netrin 1 in themigration of several cerebellar and precerebellar cellprogenitors. We show that netrin 1 exerts a strongchemoattractive effect on migrating neurons from theembryonic lower rhombic lip at E12-E14, which give riseto precerebellar nuclei. Netrin 1 promotes the exit ofpostmitotic migrating neurons from the embryonic lowerrhombic lip and upregulates the expression of TAG-1 inthese neurons. In addition, in the presence of netrin 1, themigrating neurons are not isolated but are associated withthick fascicles of neurites, typical of the neurophilic way ofmigration. In contrast, the embryonic upper rhombic lip,which contains tangentially migrating granule cellprogenitors, did not respond to netrin 1. Finally, in the

postnatal cerebellum, netrin 1 repels both the parallelfibres and migrating granule cells growing out fromexplants taken from the external germinal layer. Thedevelopmental patterns of expression in vivo of netrin 1 andits receptors are consistent with the notion that netrin 1secreted in the midline acts as chemoattractive cue forprecerebellar neurons migrating circumferentially alongthe extramural stream. Similarly, the pattern of expressionin the postnatal cerebellum suggests that netrin 1 couldregulate the tangential migration of postmitoticpremigratory granule cells. Thus, molecular mechanismsconsidered as primarily involved in axonal guidanceappear also to steer neuronal cell migration.

Key words: Netrin 1 receptors, Cerebellum, Pontine nucleus,Neuronal migration, Chemoattraction, Chemorepulsion, Migration

SUMMARY

Netrin 1 acts as an attractive or as a repulsive cue for distinct migrating

neurons during the development of the cerebellar system

S. Alcántara1,2,*, M. Ruiz1, F. De Castro2, E. Soriano1 and C. Sotelo2

1Department of Animal and Plant Cell Biology, Faculty of Biology, University of Barcelona, Barcelona E 08028, Spain2Institut National de la Santé et de la Recherche Médicale U-106, Hôpital de la Salpétrière, 75651 Paris Cédex 13, France*Author for correspondence (e-mail: sole@porthos.bio.ub.es)

Accepted 14 January; published on WWW 7 March 2000

1360

of the inferior olivary complex in both netrin 1 and Dccknockout mice suggest that netrin 1 provides cues not only foraxonal guidance but also for neuronal migration (Serafini et al.,1996; Keino-Masu et al., 1996; Bloch-Gallego et al., 1999).Moreover, mutations in the unc5h3 gene are responsible for themurine “rostral cerebellar malformation”, leading to aberrantmigration of granule and Purkinje cells in the rostral directionand loss of the rostral cerebellar boundary (Ackerman et al.,1997; Leonardo et al., 1997; Przyborski et al., 1998).

The rhombic lip, the germinative neuroepithelium locatedaround the alar recess of the fourth ventricle (His, 1891), givesrise to neurons that follow distinct migratory pathways andsubstrates (Rakic, 1990). In the mouse, the upper rhombic lip(uRL; the rostral aspect of the rhombic lip or germinal trigone;Altman and Bayer, 1997) produces progenitor cells from E13onwards that spread in a rostromedial direction to cover thewhole cerebellar surface by E17 (neurophilic tangentialmigration). These progenitors form the external germinal layer(EGL), which is the origin of the cerebellar granule cells (Gaoand Hatten, 1994; Mathis et al., 1997). During the first twopostnatal weeks in mice, postmitotic neurons from the EGLmigrate inwards along Bergmann glia (gliophilic radialmigration) and differentiate into granule cells under thePurkinje cell layer (Miale and Sidman, 1961; Rakic, 1971).

The lower region of the rhombic lip (lRL) gives rise toprecerebellar neurons in the inferior olive (IO), lateral reticularnucleus (LRN), external cuneate nucleus (ECN), nucleusreticularis tegmenti pontis (NRTP) and basilar pons (BP).Although there is some spatial and temporal overlap, neuronsfor each nucleus have a distinct temporal pattern of birthdatesand, remarkably, distinct migratory routes. Thus, neuronsdestined to IO are generated at E10-E11 and neurons fated tothe LRN and ECN become postmitotic at E11-E12. In contrast,pontine neurons (NRTP and BP) are generated during aprotracted period, extending from E12 to E16 with a late peakat E14-E15 (Taber-Pierce, 1966). Although all precerebellarneurons follow a neurophilic type of migration alongcircumferential routes at the periphery of the brainstem, IOneurons migrate along the submarginal stream (Bourrat andSotelo, 1988; Altman and Bayer, 1997), while LRN andECN neurons migrate through the marginal stream (Bourratand Sotelo, 1991; Altman and Bayer, 1997). NRTP and BPneurons follow a distinct rostral migratory route towardstheir respective pontine nuclei, named the pontobulbar orpontomedullary stream (Harkmark, 1954; Rakic, 1985; Onoand Kawamura, 1990). Finally, while neurons destined to theLRN and ECN cross the midline through the most ventralregion of the floor plate (Bourrat and Sotelo, 1991; Altman andBayer, 1997), olivary and pontine neurons stop ipsilaterally onthe side where they were generated (Taber-Pierce, 1966;Ellerberger et al., 1969; Bourrat and Sotelo, 1988, 1990).

The complexity of migratory routes and locations taken byrhombic lip-originated neurons suggests that multiple andspecific guiding and stopping factors are involved in theirmigration and allocation. To understand the molecular bases ofneuronal cell migration in the rhombic lip, here we study theeffects of netrin 1 in vitro in explants taken from E12-E14rhombic lip and from its postnatal derivative, the EGL. Weshow that secreted netrin 1 exerts an attraction on both axonsand neurons arising from embryonic lRL explants, whereas ithas no effect on uRL explants. In contrast, netrin 1 strongly

repels parallel fibres and granule cells from postnatal EGLexplants. We also show that the patterns of expression of netrin1 and its receptors (Dcc, neogenin, unc5h2 and unc5h3) areconsistent with a role for this chemotactic molecule in themigration of neurons originated from the rhombic lip in vivo.

MATERIALS AND METHODS

AnimalsOF1 embryos and postnatal mice (Iffa Credo, France) were used. Themating day was considered embryonic day 0 (E0) and the day of birthpostnatal day 0 (P0). After ether anaesthesia of the dams, embryoswere dissected out and collected in 0.1 M phosphate-buffered saline(PBS) and 0.6% D-glucose, and used for cultures or fixed with 4%paraformaldehyde (PFA) in 0.1 M phosphate buffer, pH 7.3. Embryosolder than E14 and postnatal animals were transcardially perfusedwith the above fixative. After postfixation, the brains werecryoprotected in 30% sucrose and frozen on dry ice. Coronal sections50 µm thick were collected in a cryoprotectant solution (30% glycerol,30% ethylene glycol, 40% 0.1 M PBS), and stored at −30°C until use.

Explant culturesThe uRL and lRL from E12-E15 embryos were dissected out as asingle piece and cut into 150-300 µm tissue pieces with fine tungstenneedles (Fig. 1A). 250 µm parasagittal sections of postnatal cerebellawere cut with a tissue chopper. After removing the meninges, 150-300 µm tissue pieces from the EGL were obtained (Fig. 1B).

Explants were cocultured at a distance (200-600 µm) withaggregates of EBNA-293 cells stably transfected with a constructencoding netrin 1-c-myc, or with the vector alone (Keino-Masu et al.,1996). Explants and cell aggregates were embedded in rat-tailcollagen as described (Lumsden and Davies, 1986) and cultured inDMEM (Seromed) supplemented with L-glutamine, NaHCO3, D-glucose, and supplements B27 and N2 (all from Gibco LifeTechnologies), for 24-72 hours in a 5% CO2, 95% humidity incubatorat 37°C. To monitor the expression of netrin 1 and the distancecovered by the netrin 1 gradient, some cocultures were fixed andimmunostained with the monoclonal 9E10 anti-c-myc antibody (SantaCruz). Secreted netrin 1-c-myc was detected in a gradient of intensityover a distance of about 300-400 µm away from the EBNA-293aggregates after 2 days in vitro (div), indicating that a long-rangegradient of netrin 1 is formed in our experimental conditions. In someexperiments, explants were incubated with medium conditioned for36 hours with EBNA-293-netrin 1 or control cells diluted 1:1 withfresh medium.

28 experiments were done and up to 2000 explants analyzed.

ImmunohistochemistryExplants were fixed in 4% PFA for 1 hour and processed for thevisualisation of neuronal bodies and neurites. After several rinses inPBS, cultures were incubated with monoclonal antibodies againstneuron-specific class III β-tubulin (1:3000; clone Tuj-1 Babco;Moody et al., 1989), intracellular human DCC (1:500; clone GA7 499,Pharmigen), C-myc (1:200, Santa Cruz) or rabbit polyclonalantibodies against intracellular human DCC (1:1000, Fabre et al.,1999), L1 (1:2000, Persohn and Schachner, 1987), TAG-1 (1:2000,Karagogeos et al., 1991), calbindin D28k (1:10000, Swant) and GFAP(1:1000, DAKO). The incubation with secondary antibodies (1:200;Vector Labs.) was followed by the streptoavidin-peroxidase complex(1:400, Amersham). Enzymatic reaction was developed withdiaminobenzidine and H2O2. Alternatively, FITC-conjugatedsecondary antibodies, or Texas Red-coupled streptoavidin were used,together with fluorescent nuclear markers (bisbenzimide or ethidiumbromide) to visualized both cell bodies and neurites.

Sections from different developmental stages, some of them

S. Alcántara and others

1361Netrin 1 guides neuronal migration

previously hybridised for netrin 1 expression, were immunolabeledwith anti-DCC antibodies.

In situ hybridisationIn situ hybridisation was performed on free-floating sectionsessentially as described (Alcántara et al., 1998). Sections werepermeabilized in 0.2% Triton X-100 (15 minutes), treated with 2%H2O2 (15 minutes), deproteinized with 0.2 N HCl (10 minutes), fixedin 4% PFA (10 minutes) and blocked in 0.2% glycine (5 minutes).Thereafter, sections were prehybridised at 60°C for 3 hours in asolution containing 50% formamide, 10% dextran sulphate, 5×Denhardt’s solution, 0.62 M NaCl, 10 mM EDTA, 20 mM Pipes (pH6.8), 50 mM DTT, 250 mg/ml yeast t-RNA and 250 mg/ml denaturedsalmon sperm DNA. netrin 1, Dcc, unc5h2, unc5h3 and neogeninriboprobes were labelled with digoxigenin-d-UTP (Boehringer-Mannheim) by in vitro transcription. A cDNA fragment of 2.7 kbencoding the 3′ unstranslated region of mouse netrin 1 (Serafini et al.,1996) using T3 polymerase (Ambion), a cDNA fragmentcorresponding to the rat Dcc cytoplasmic region of 0.85 kb, and afragment of 0.72 kb encoding the rat neogenin cytoplasmic region(Keino-Masu et al., 1996) using SP6 polymerase (Promega). cDNAfragments of the rat unc5h2 (1.5 kb) and rat unc5h3 (1.6 kb)(Leonardo et al., 1997) were transcribed using T7 polymerase(Ambion). Labelled antisense cRNA was added to theprehybridisation solution (250-500 ng/ml) and hybridisation wascarried out at 60 ºC overnight. Sections were then washed in 2× SSC(30 minutes, room temperature), digested with 20 mg/ml RNAse A(37°C, 1 hour), washed in 0.5× SSC/50% formamide (4 hours, 55°C)and in 0.1× SSC/0.1% sarkosyl (1 hour, 60°C). After rinsing in Tris-buffered saline (TBS)/0.1% Tween 20 (15 minutes), sections wereblocked in 10% normal goat serum (2 hours) and incubated overnightwith an alkaline phosphatase-conjugated antibody to digoxigenin(Boehringer-Mannheim, 1:2000). After washing, sections weredeveloped with nitroblue tetrazolium (NBT) and 5-bromo-4-chloro-3-indolyl phosphate (BCIP) (Life Technologies), mounted ongelatinised slides and coverslipped with MowiolTM.

ControlsControl hybridisations, including hybridisation with sensedigoxigenin-labelled riboprobes or RNAse A digestion prior tothe hybridisation, prevented alkaline phosphatase staining above

background levels. For immunocytochemical controls, omission ofthe primary antibodies prevented immunostaining.

Analysis and quantificationHybridised and immunostained sections were examined on a ZeissAxiophot microscope. The delimitation of regional and laminarboundaries was performed according to Jacobowitz and Abbott(1998). Rhombic lip or EGL explants cultured alone displayed aradial pattern of neurite outgrowth. Neurite quantification wasobtained from explants immunostained with Tuj-1, whereas thequantification of the neuronal migration was done on bisbenzimide-stained explants. In both instances, explants were photographed(75× final magnification) and the field was divided into fourquadrants as illustrated in Fig. 1C. For each explant, the areascovered by either Tuj-1-positive neurites or by bisbenzimide-stainedcells (from the border of explants to the outer perimeter of theexpanding processes or neurons) were measured in the distal andproximal quadrants using the IMAT image analysis program(Scientific-Technical Services, University of Barcelona). Data werepresented as mean ± s.d. For the statistical analysis of the areas,Anova and LSD tests were performed with the Statgraphic statisticalpackage. To determine the preferential direction of the outgrowth,the ratio of the surface in the proximal and distal quadrants wasobtained for each explant, which gives a ratio of 1 for radialoutgrowth. Ratios differing from the mean ratio obtained forexplants cultured with control cells by more than one s.d. wereassigned therefore to the group attracted or repulsed by netrin 1.

RESULTS

Netrin 1 effects on neurons and neurites derivedfrom the lower rhombic lipThe temporal sequence of proliferation of neuronal precursorsin the lRL allows us to pinpoint the classes of precerebellarneurons affected by netrin 1 in their migration. In ourexperiments, we selected rhombic lip explants taken from E12-E14 embryos. At E12, all IO neurons have left the lRL andmost of the cells in these explants corresponded to LRN andECN neurons. In contrast, the neurons generated at E14 are

Table 1. Quantitative analysis of the neuritic outgrowth (A) and cellular migration (B) in explants from the lower rhombiclip at different ages

Mean area ± s.d. (mm2) Statistical Number of Orientation

Culture condition Proximal quadrant Distal quadrant significance explants + = − % +

(A) Neurite outgrowth (Tuj-1 immunostaining)lRL E12 net 1 2 div 0.302±0.103 0.110±0.072 ** 23 21 2 0 91lRL E12 cont 2 div 0.129±0.058 0.211±0.128 7 1 4 2 29

lRL E14 net 1 1 div 0.072±0.052 0.024±0.018 ** 17 10 7 0 59lRL E14 cont 1 div 0.031±0.001 0.025±0.004 4 1 3 0 25

lRL E14 net1 2 div 0.296±0.178 0.097±0.039 ** 23 23 0 0 100lRL E14 cont 2 div 0.115±0.073 0.131±0.083 7 0 7 0 0

(B) Neuronal cell migration (bisbenzimide nuclear staining)lRL E14 net1 2 div 0.380±0.177 0.056±0.025 ** 6 6 0 0 100lRL E14 cont 2 div 0.056±0.018 0.045±0.014 6 0 6 0 0

div, days in vitro.The area occupied by the bulk of the neuritic processes and/or by the cell bodies in the proximal and distal quadrants in explants cocultured with aggregates of

EBNA-293 control cells or with cells transfected with the netrin 1-c-myc construct was measured and expressed as means ± s.d. Significant differences betweenproximal and distal quadrants were calculated from crude data using the LSD test. **P=0.01.

The preferential orientation of the outgrowth was determined for each explant dividing the area occupied by the neuritic outgrowth or cellular migration in theproximal quadrant into that of the distal quadrant, which gives a ratio of 1 for radial outgrowth. The limits between categories were calculated from the s.d.obtained for the ratios in explants cocultured with control cells. div, days in vitro.

=, radial: ratio = 1±s.d.; +, attraction: ratio >1+s.d.; −, repulsion: ratio <1−s.d.; %+, percentage of the explants attracted by netrin 1.

1362

exclusively destined to pontine nuclei (Taber-Pierce 1966, seealso Discussion).

We first tested whether netrin 1 acts as an attractive orrepulsive cue for LRN and ECN neurons, by coculturing E12lRL explants with aggregates of netrin 1-secreting cells.Explants cultured with control EBNA-293 cell aggregates for2 div showed no preferential orientation (radial) of outgrowingneurites immunostained with the Tuj-1 mAb (Table 1; Fig. 2).Most importantly, the outgrowing neurites were almost devoidof migrating cell bodies. In contrast, when E12 lRL explants

were cultured with netrin 1-secreting cells, 91% of the E12 lRLexplants (Table 1) showed a marked asymmetrical pattern ofneurite outgrowth with far fewer processes in the distal than inthe proximal quadrant (data not shown). This was so even whentissue explants were located at long distances (around 600 µm)from the cell aggregates. Moreover, the neurites in the proximalquadrant formed tight, prominent fascicles, which wereassociated with strands of migrating neurons.

To test the effect of netrin 1 on the migration of pontineprecerebellar neurons, similar in vitro experiments werecarried out using E14 lRL. In cocultures with control cellaggregates, lRL neurites grew symmetrically in 75% of theexplants after 1 div and in 100% of the explants after 2 div(Table 1; Fig. 3A). After both culture times, radial emergingneurites were poorly fasciculated (Fig. 3A), and the few Tuj-1-positive neurons leaving the explants showed long neuriticprocesses, probably corresponding to the leading process ofmigrating pontine cells (Ono and Kawamura, 1990) androughly bipolar morphology (Fig. 3C,E). These neurons didnot migrate for long distances and formed a narrow ring of lessthan 100 µm thick around the explants (Table 1; Figs 3A, 4A).In contrast, 59% (1 div) and 100% (2 div) of the explantscultured with netrin 1-secreting aggregates (Table 1A; Fig. 3B)showed prominent neurite outgrowth directed towards the cellaggregates. Most of these attracted processes had fasciculatedand attained the netrin 1 producing cell masses (Fig. 3B,F).Isolated migrating neurons were practically absent incocultures in which cell nuclei were visualized with DNA-staining dyes. On the contrary, the migrating neurons weretightly apposed, following each other to form thick neuronal

S. Alcántara and others

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

E12 net-1 E12 cont E14 net-1 E14 cont

Are

a of

neu

rite

outg

row

th (

mm

2)

proximal

distal

Fig. 2. Histograms illustrating the area of neuritic processes and cellbodies in the proximal and distal quadrants of E12 and E14 lowerrhombic lip explants cocultured with EBNA-293 control cells or withcells secreting netrin 1 for 48 hours. Data are means ± s.e.m.;**P=0.01, LSD test.

uRLlRL

E12-E14 PostnatalA

EBNA cells

Distal ProximalC

IGL

EGLML

Rostral Caudal

WM

PL

B

4V

Fig. 1. Schematic drawings showing theanatomical location of the different types ofexplants used and the analysis applied. (A) Dorsalview of an E12-E14 embryo showing the locationof the rhombic lip around the IVth ventricle. Theupper region or germinal trigone (uRL) is inpurple and the lower region (lRL) is in yellow.(B) Sagittal section of a postnatal cerebellumshowing the layers used for EGL explants.(C) Scheme illustrating the method used for thequantification of neuritic outgrowth and neuronalmigration. For each explant, the area occupiedrespectively by neuritic processes and cell bodieswas measured in the quadrants (marked in black)proximal and distal to the cell aggregates.

1363Netrin 1 guides neuronal migration

chains, which emerged from the explants, mainlyin their proximal quadrants (Table 1B; Fig.3B,D), but also occasionally in their lateralquadrants (Fig. 4B). In double-stainedcocultures, most of the chains of migrating cellswere intermingled with the fasciculatedprocesses (Fig. 3B,D,F) as occurs in the pontinestream in vivo (Ono and Kawamura, 1990), andspanned the distance between the lRL explantsand the EBNA-293 transfected cell masses (Fig.3B). Thus, precerebellar neurons use aneurophilic type of migration (Rakic, 1990).Moreover, a few neurites also grew in the distalquadrants, but they were not associated withmassive neuronal migration (Figs 3B, 4B), whichsupports the suggestion that, in the cocultures,neuritic outgrowth and neuronal migration arenot a unique process. These results show thatnetrin 1 functions as a chemotactic cue formigrating precerebellar neurons, because itimposes directionality to their migration.

Fig. 3. Netrin 1 is chemoattractive for neurites andcells originating from E14 lRL explants. E14 lRLexplants cocultured 48 hours with aggregates ofEBNA-293 control cells (A,C,E) or cells transfectedwith the netrin 1-c-myc expression vector (B,D,F).Confocal images at a medium plane of the cultureshowing the neuronal processes immunostained withanti-β-tubulin III antibodies, in green, and the cellularnuclei stained with ethidium bromide, in red.(C,E) and (D,F) are respectively higher magnificationsof the boxes shown in A and B, which illustrate theshort distance and the random orientation of migratingneurons confronted with control cells (E) and the longchains of migrating cells oriented to the source ofnetrin 1 (F). c, aggregate of control cells; n, aggregateof netrin 1-secreting cells. Scale bars, 200 µm (A,B)and 40 µm (C-F).

Table 2. Quantitative analysis of the outgrowth in explants from the uRL and EGLMean area ± s.d. (mm2) Statistical Number of Orientation

Culture condition Proximal quadrant Distal quadrant significance explants + = − % −uRL E14 net 1 1 div 0.058±0.017 0.059±0.023 22 5 8 9 41uRL E14 cont 1 div 0.073±0.036 0.062±0.027 5 1 3 1 20

uRL E14 net 1 2 div 0.083±0.052 0.115±0.043 11 1 4 6 55uRL E14 cont 2 div 0.133±0.053 0.149±0.062 5 1 3 1 20

EGL P0 net 1 3 div 0.069±0.025 0.099±0.030 * 18 2 4 12 67EGL P0 cont 3 div 0.112±0.040 0.122±0.045 8 1 6 1 13

EGL P2 net 1 2 div 0.065±0.023 0.103±0.024 ** 22 0 3 19 86EGL P2 cont 2 div 0.087±0.033 0.077±0.030 9 1 7 1 11

EGL P4 net 1 2 div 0.050±0.018 0.095±0.042 ** 32 0 3 29 91EGL P4 cont 2 div 0.173±0.047 0.172±0.042 7 0 6 1 14

EGL P5 net 1 2 div 0.062±0.026 0.131±0.051 ** 31 1 3 27 87EGL P5 cont 2 div 0.119±0.026 0.130±0.037 11 2 7 2 18

The area occupied for the bulk of the neuritic processes and cell bodies in the proximal and distal quadrants of explants cocultured with aggregates of EBNA-293 control cells or with cells transfected with the netrin 1-c-myc construct were measured and expressed as means ± s.d. Significant differences betweenproximal and distal quadrants were calculated from crude data using the LSD test.

*P=0.05; **P=0.01. The preferential orientation of the outgrowth was determined for each explant by dividing the area of the outgrowth in the proximalquadrant into that of the distal quadrant, which gives a ratio of 1 for radial outgrowth. The limits between categories were calculated from the s.d. obtained for theratios in explants cocultured with control cells. div, days in vitro.

=, radial: ratio = 1±s.d.; +, attraction: ratio >1+s.d.; –, repulsion: ratio <1−s.d.; %−, percentage of the cultures repelled by netrin 1.

1364

To substantiate the notion that migrating neurons in E14 lRLexplants are pontine neurons, cultures were immunostainedwith anti-TAG-1 antibodies (Wolfer et al., 1994). As shown inFig. 4C,D, TAG-1 antibodies stained thick strands of migratingneurons with long leading edges, which were directed towardsthe netrin 1 aggregates. Finally, the strong migration ofpostmitotic pontine precerebellar neurons produced by netrin1 was reproduced when single E14 lRL explants were culturedwith medium conditioned for 36 hours with EBNA-293 cells

secreting netrin 1 (Fig. 4E-H). Some of the lRL explants,cultured either with aggregates of netrin 1-secreting cells or inconditioned medium, were immunostained with anti-GFAPantibodies. GFAP-positive cells were missing, suggesting thatthe migration was not gliophilic.

These experiments demonstrate that netrin 1 is not only achemoattractant for neurites emerging from LRN, ECN, NRTPand BP neurons generated in the lRL, but also that itchemoattracts and promotes the massive cell migration of theseneurons.

Netrin 1 effect on neurons and neurites from theupper rhombic lip and external germinal layerTo test the role of netrin 1 in the two modes of migration ofthe cerebellar granule cell precursors and neurons, in vitroexperiments were carried out with either E14 uRL explants orP0-P6 EGL explants. E14 uRL explants cocultured withcontrol cells emitted neurites without preferential orientationafter either 1 or 2 div (Table 2). When cultured with netrin 1-producing cells, the neuritic outgrowth from uRL explants wasless homogeneous (Fig. 5A,B), although no significant changeswere noticed in the area occupied by the emitted processes(Fig. 6; Table 2).

S. Alcántara and others

Fig. 4. The cells and neurites attracted by netrin 1 at E14 are TAG-1immunoreactive. E14 lRL explants cocultured for 48 hours withaggregates of EBNA-293 control cells (A,C), or with cells secretingnetrin 1 (B,D). Explants stained with bisbenzimide (A,B) illustratethe distribution of migrating cells, and the remarkable effect of netrin1 on neuronal migration (B). Similar explants immunostained withTAG-1 antibodies (C,D) show that, in the absence of netrin 1, fewsymmetrically arranged neurites and cell bodies exit the explants;whereas, they are much more numerous and oriented towards thesource of netrin 1 when cocultured with cells secreting netrin 1.Comparison of C and D indicates that netrin 1 upregulates TAG-1immunoreactivity (E-H). E14 lRL explants cultured for 72 hourseither in conditioned medium from EBNA-293 control cells (E,G) orfrom cells transfected with netrin 1-c-myc construct, immunostainedwith anti-TAG-1 antibodies (F,H). In the presence of netrin 1, TAG-1immunoreactivity increases and cells migrate out of the explantforming long cellular strands. (G,H) Higher magnifications of theboxes shown in E,F, illustrating the chains of migrating cells. c,control; n, netrin 1. Scale bars, 200 µm (A-F) and 75 µm (G-H).

Fig. 5. Netrin 1 effects on embryonic granule cell progenitors andneonatal granule cells. Explants from E14 uRL (A,B), P0 EGL (C,D)were cocultured for 36-72 hours with EBNA-293 control cells (A,C) orwith cells transfected with the netrin 1-c-myc construct (B,D), andimmunostained with anti-β-tubulin III antibodies. Note that netrin 1exerts inhibitory effects on the neuritic outgrowth from P0 EGLexplants (D). Explants from P5 EGL cultured for 24 hours withconditioned medium from parental EBNA-293 cells (E) or from cellstransfected with netrin 1-c-myc construct (F) labelled with anti-β-tubulin III antibodies. Note that immunoreactive fibers are reduced inthe presence of netrin 1. Perimeters of aggregates of control cells (c)and netrin 1-secreting cells (n), outlined with black circles, white plotsoutline perimeters of uRL and EGL explants. Scale bars, 200 µm.

1365Netrin 1 guides neuronal migration

To dissect EGL explants from the parasagittal slices ofpostnatal cerebella, we used the easiest cleavage plane, whichpasses just below the multicellular (P0-P2) or monocellular(P4-P6) Purkinje cell layer. Thus, these EGL explants initiallycontain the subpial basal and glial laminae, the entire EGL, thenascent molecular layer and most of the Purkinje cells(Fig.1B). To favour the radial outgrowth of granule cellprocesses and neurons, explants were placed either on theirupper subpial or their deeper Purkinje cell surface. Theoutgrowth from such EGL explants cultured in a tridimensionalcollagen gel matrix was symmetrical and radial, which allowedus to monitor the effects of netrin 1.

To identify the classes of cerebellar neurons and processesthat grew from EGL explants, cultures were immunostainedwith Tuj-1 mAb or with antibodies to different markers: TAG-1 or L1 to visualise granule cell bodies and processes,calbindin for Purkinje cells and GFAP for Bergmann glia(Wassef et al., 1985; Buttiglione et al., 1996; Soriano et al.,1997). When P0 EGL explants were cocultured for 1 to 3 dayswith netrin 1-expressing cells, 67% of the explants showed Tuj-1-positive emerging neurites that grew preferentially towardsthe distal quadrant (Table 2; Fig. 5C,D). Older EGL explants(P2, P4 or P5) showed a similar, although more dramatic,repulsive effect (Figs 6, 7A-D). For instance, P4 and P5 EGLexplants showed a preferential outgrowth in the distal quadrantin about 90% of cases (Table 2). The repulsive effect of netrin1 on neurite extension was manifested in preferential neuriticoutgrowth in the distal quadrant rather than in veering awayfrom the source of netrin 1, and was easily mimicked whenEGL explants were cultured with conditioned medium fromnetrin 1-secreting cells, which reduced neurite outgrowth (Fig.5E,F).

The use of distinct cellular markers showed that calbindin-positive Purkinje cells did not survive in these conditions anddegenerated during the second day in culture (not shown).GFAP-immunostaining revealed positive cell bodies withinthe explants, most probably belonging to Bergmann glialcells, but no GFAP-positive processes grew out of theexplants (data not shown). In contrast, TAG-1 (Fig. 7E) andL1 (not shown) antibodies, which in the developingcerebellum mark the most immature and mature parallelfibres, respectively (Buttiglione et al., 1996), labelled themajority of processes arising from EGL explants. Thisindicates that most, if not all, the neurites repelled by netrin

1 in our assay correspond to the parallel fibres. Finally,together with the parallel fibers, TAG-1- or L1-positive cellbodies were observed leaving the tissue explants (Fig. 7E). Inthe presence of control EBNA-293 cells, few labelled cellbodies were present near to the explants, in the radial-emerging bundles of parallel fibres. In contrast, in the

0.00

0.05

0.10

0.15

0.20

E14net-1

E14cont

P0net-1

P0cont

P2net-1

P2cont

P4net-1

P4cont

P5net-1

P5cont

Are

a of

neu

rite

outg

row

th (

mm

2)

proximal

distal

Fig. 6. Histograms illustrating thearea of neuritic outgrowth in theproximal and distal quadrants ofE14 upper rhombic lip andpostnatal EGL explantscocultured for 48-72 hours withEBNA-293 control cells or withcells transfected with the netrin 1-c-myc construct. Data arepresented as means ± s.e.m.;*P=0.05; **P=0.01, LSD test.

Fig. 7. Netrin 1 repels neuritic outgrowth from postnatal mice EGLexplants. Explants from P2 (A,B), P4 (C,D) and P6 (E) cerebellawere cocultured for 48 hours with EBNA-293 control cells (A,C) orwith cells transfected with netrin 1-c-myc expression vector (B,D,E)and immunostained with antibodies against β-tubulin III (A-D) orTAG-1 (E). Note the strong chemorepulsion exerted by netrin 1 inB,D. (E) Photomicrograph showing TAG-1-positive granule cells(arrows) migrating along fibres in the quadrant distal to the source ofnetrin 1. Perimeters of aggregates of control cells (c) and netrin 1-secreting cells (n), outlined with white circles. Scale bars, 250 µm(A-D), 25 µm (E).

1366

presence of netrin 1-expressing cells, the immunopositive cellbodies were more abundant and exclusively found in thequadrant distal to the netrin 1 cell aggregate (Fig. 7E). TAG-1- or L1-immunoreactive cell bodies were very small (6-8µm) and therefore correspond to migrating granule cells.

Taken together, the above results show thatnetrin 1 has no effect on E14 uRL explants,indicating that this factor does not play a role inthe initiation of the tangential migration ofgranule cell progenitors originating from thegerminative trigone. In contrast, netrin 1 exertsa strong repulsive effect on the outgrowth ofparallel fibres (the axons of granule cells) and onthe intraneuritic translocation of granule cellnuclei in these axons. The netrin 1-generatedrepulsion is already evident at birth (P0) but ismore dramatic at later postnatal stages.

Developmental expression of netrin 1,Dcc, neogenin, unc5h2 and unc5h3genesTo examine whether in vivo netrin 1 mediatesthe attractive and repulsive effects observed invitro, we studied by in situ hybridisationand immunohistochemistry the patterns ofexpression of netrin 1 and its receptors in thelower brainstem and cerebellum of embryonicand postnatal mice. At E12, netrin 1 was highlyexpressed in the epithelial cells at the origin ofthe floor plate. The expression spreadmediolaterally (with a decreasing gradient)through the ventricular basal plate, up to thesulcus limitans. Dcc mRNA and protein werenot expressed in the ventricular zone, but in thesubventricular zone of both the basal and alarplates. DCC protein labelled fibres in themarginal and submarginal migratory streams. Atthis stage of development, the hybridisationsignals for unc5h2 and unc5h3 were almostabsent (not shown; Bloch-Gallego et al., 1999).

At E14, the expression of netrin 1 was similarto E12 (Fig. 8A). Dcc expression remained highin the subventricular zone and along thepontomedullary migratory stream, up to thepons. Less intense expression was detected inthe nucleus of the solitary tract and the vestibularregion (Fig. 8C,D). neogenin mRNA was highlyexpressed in cells of the ventricular zone in boththe basal and alar plates and in two dorsalmostlines, underlying the lower rhombic lips (Fig.8B). Thus, Dcc and neogenin genes show non-overlapping patterns of expression. Unc5h2signals were high and occupied most of theventricular alar plates, including the lRL,whereas unc5h3 signals were practically absent(Fig. 8E,F). The expression of unc5h2 wascomplementary to that of netrin 1 and partiallyoverlapping with neogenin mRNA distribution(Fig. 8A,B,E). At pontine levels, netrin 1 wasexpressed in the midline (Fig. 8G) and Dccsignals were intense in the pontine migratory

stream (Fig. 8H). Unc5h2 and unc5h3 were faintly expressedin the pons at E14, and increased with age (Fig. 8I).

At E14, the expression of netrin 1 and its receptors was weakin the developing cerebellar plate. netrin 1 was virtually absent,with the exception of a faint staining in deep nuclear regions

S. Alcántara and others

Fig. 8. Embryonic expression of netrin 1 and its receptors in the lower brainstem andpons. (A-F) Coronal sections through the brainstem showing the lRL and the originof the pontine migratory stream at E14. (A) netrin 1 mRNA (NET-1) is stronglyexpressed in the floor plate and in the ventricular zone of the basal plate.(B) Neogenin mRNA (NEOG) is expressed in the ventricular zone of the basal plateand less intensely in the ventricular zone of the alar plate including the lRL. (C) DccmRNA and (D) protein (DCC-P) are expressed in early postmitotic pontine neuronsin the subventricular zone of the lRL and more intensely in cells migrating along thepontine migratory stream. (E) Unc5h2 mRNA is expressed in the ventricular zone ofthe alar plate including the lRL. Note that the strongest expression is adjacent to thearea of netrin 1 expression and the small patches of unc5h2 delimitating the pontinemigratory stream. (F) unc5h3 mRNA is not expressed at this age. (G-I) Coronalsections at the level of the pons at E14 (G,H) and E17 (I). (G) netrin 1 is stronglyexpressed in midline cells. (H) DCC protein is present in the pontine migratorystream at the level of the pons. (I) at E17 unc5h2 is expressed in pontine neurons.AP, alar plate; BP, basal plate; FP, floor plate, lRL, lower rhombic lip; MFN, motorfacial nucleus; PN, pontine nuclei; PMS, pontine migratory stream; SL, sulcuslimitans; SVZ, subventricular zone; VL, nucleus of the ventral lemniscus; VZ,ventricular zone; IV, IVth ventricle. Scale bars 250 µm.

1367Netrin 1 guides neuronal migration

(Fig. 9A). Dcc mRNA and protein werefound in the cerebellar primaryneuroepithelium, the germinal trigoneand the EGL. DCC protein present infibre bundles might correspond to thevestibular fibres, which are the firstextracerebellar projections to arrive(Fig. 9C,D). neogenin colocalized itsexpression with Dcc, mainly at themiddle region of the ventricular zone. Inaddition, neogenin was also expressedin deep nuclear neurons (Fig. 9B). Asdevelopment proceeded, the cerebellarexpression of netrin 1 and its receptorschanged. Thus, at P5 netrin 1expression was prominent in the EGLand in both the stellate and basket cells(the inhibitory interneurons of themolecular layer), but almost absent inthe IGL (Fig. 10A). neogenin mRNAwas expressed in granule cells in boththe EGL and IGL (Fig. 10B). DccmRNA was found throughout the EGL,whereas DCC protein was only found inthe parallel fibres at the interfacebetween the deep EGL and the nascentmolecular layer. DCC-positive fibreswere also observed in the granule celllayer (IGL) and in the white matter (Fig.10C,D). Unc5h2 and unc5h3 mRNAswere confined to granule cell precursors(EGL) and postmigratory granule cells(IGL) (Fig. 10E,F).

DISCUSSION

Netrin 1 is a guiding cue for themigration of neurons fated to thepontine nuclei In the present study, we have addressedthe role of netrin 1 in the migrationof precerebellar neurons by culturing

Fig. 9. Expression of netrin 1 and its receptors in thecerebellar anlage at E14. (A) netrin 1 mRNA (NET-1)is almost absent in the cerebellar anlage. (B)Neogenin mRNA (NEOG) is strongly expressed inthe ventricular zone with a mediolateral gradient ofexpression and also in the EGL and deep nuclei of thecerebellum. (C) Dcc mRNA and (D) protein (DCC-P), as neogenin, are localized in the ventricular zoneand EGL. Note that several fibres tracts, probablycerebellar afferents and the trochlear nerve (arrow),are strongly immunoreactive for DCC. DN, deepnuclei; EGL, external granule layer; lRL, lowerrhombic lip; VZ, ventricular zone; URL, upperrhombic lip; IV, IVth ventricle. Scale bars, 250 µm.

Fig. 10. Expression of netrin 1 and its receptors in the P5 cerebellum. (A) netrin 1 mRNA(NET-1) is expressed in the EGL and in the molecular layer interneurons. (B) neogeninmRNA (NEOG) is exclusively expressed in granule cells in the EGL and IGL. (C) DccmRNA and (D) protein (DCC-P) are expressed in premigratory granule cells in the EGL andin the white matter. Arrows in D indicate parallel fibres. (E) unc5h2 and (F) unc5h3 areexpressed in granule cells in both the EGL and IGL. EGL, external germinal layer; IGL,internal granule layer; ML, molecular layer; PC, Purkinje cell layer; WM, white matter. Scalebars, 250 µm (A-C, E-F), 50 µm (D).

1368

explants of the inferior lateral recess of the IVth ventricle atE12-E14 (lRL explants). Although we cannot rule out thepossibility that some olivary neurons may be present in ourexplants, several arguments indicate that we have monitoredthe migration of neurons fated to the pontine precerebellarnuclei. First, although a few olivary neurons are late-generatedat E12 in the lRL, this stage contributes mainly to thegeneration of precerebellar neurons fated to the externalcuneatus (EC), the lateral reticular (LR) and the raphe nuclei(Taber-Pierce, 1966; Bourrat and Sotelo, 1988; Altman andBayer, 1997). This is especially so at E14, when the lRL onlygenerates neurons destined to the pontine nuclei (PN) (Taber-Pierce, 1966; Altman and Bayer, 1997). Second, during theformation of the precerebellar system, the surface glycoproteinTAG-1/Axonin 1 is strongly expressed along the extramuralmigratory streams (which contains neurons destined to the EC,LR and PN nuclei), but not along the intramural migratorystream, the pathway followed by migrating olivary neurons(Wolfer et al., 1994), which agrees with the present datashowing that the neurons that respond to netrin 1 are stronglyimmunoreactive for TAG-1. We thus conclude that the lRLexplants selected in our experiments correspond to theprecerebellar neuroepithelium that gives rise to the LR and ECnuclei at E12, and to the basilar pons at E14 (Taber-Pierce,1966).

Analysis of the precerebellar nuclei in mice deficient inexpression of netrin 1 has revealed that basilar pontine neuronsare missing (Serafini et al., 1996) and that less than 14% ofinferior olivary neurons reach their ventromedial position(Bloch-Gallego et al., 1999), although it is not known how sucha phenotype arises. The present data show that netrin 1 acts asa long-range chemoattractant factor for pontine nucleimigrating neurons, which represents the first direct evidencefor netrin 1 influencing the direction of migrating neurons. Ourdata support the notion that the absence of pontine nuclei innetrin 1− hypomorphic mice directly reflects abnormalneuronal migration from the lRL caused by the lack of netrin1. Since Dcc knockout mice share similar defects in pontinestructures (Fazeli et al., 1997), it is likely that this receptor isinvolved in such chemoattraction. In favour of this conclusionare the observations on inferior olivary neurons in netrin 1-deficient mice. The visualisation of these neurons by their Brn-3b expression (Bloch-Gallego et al., 1999) suggested that theatrophy of the inferior olive is mainly the result of impairedmigration, because Brn-3b-expressing cells were distributedalong the submarginal migratory stream. Thus, the presentresults, together with those in inferior olivary neurons (Bloch-Gallego et al., 1999), allow us to conclude that netrin 1 is anessential cue for the circumferential migration of precerebellarneurons.

However, other mechanisms involving netrin 1 maycontribute to the mutant phenotype. For instance, DCC mayinduce apoptosis in the absence of ligand-binding, but blockscell death when engaged to netrin 1 (Mehlen et al., 1998). Asthe early postmitotic cells in the lRL express DCC, the netrin1/DCC interaction may regulate the survival of early pontinenuclei cells (Cho and Fearon, 1995). In addition to exerting astrong chemoattractive effect, our explants incubated withnetrin 1-conditioned media show that this factor dramaticallypromotes the exit of migrating neurons from theneuroepithelium, even in the absence of well-defined gradients.

Thus, although our findings show that netrin 1 acts as achemoattractive cue for the migration of pontine nuclei alongthe anterior extramural stream, as occurs for the navigation ofventrally growing axons (Kennedy et al., 1994; Shirasaki et al.,1995, 1996), we cannot rule out additional mechanisms ofaction of netrin 1 that may contribute to the generation of thephenotype in netrin 1 mutant mice.

The present observations also show that netrin 1 induces theformation of thick strands and fascicles of migrating neuronsand neurites, in which these cells attach closely to each other,which is reminiscent of the neurophilic migration that takesplace in the extramural migratory streams in vivo (Rakic, 1985;Ono and Kawamura, 1989, 1990) and in other migratorypathways such as the rostral migratory stream to the olfactorybulb (Jankovski and Sotelo, 1996; Lois et al., 1996; Wichterleet al., 1997). Previous studies have reported the presence ofTAG-1 in developing axons that respond to netrin 1 (Placzeket al., 1990; Serafini et al., 1994; Wolfer et al., 1994;Colamarino and Tessier-Lavigne, 1995; Guthrie and Pini,1995; Tamada et al., 1995; Shirasaki et al., 1996) and theparticipation of this molecule in the transduction of floor-plate-derived signals has been hypothesised (Shirasaki et al., 1996).The upregulation of TAG-1 protein in migrating cells by netrin1 reported here represents the first direct link between TAG-1expression and a guidance protein secreted by the floor plate.TAG-1 forms homophilic trans-interactions and heterophilicinteractions with other IgCAMS, such as F3 or L1, whichresults in changes in adhesivity and fasciculation (Lemmon etal., 1989; Felsenfeld et al., 1994; Buttiglione et al., 1998).TAG-1 may contribute to the formation of neuritic fasciclesand, particularly, of the strands of migrating pontine neurons,similar to the participation of PSA-NCAM to the formation ofneuronal chains in the rostral migratory stream of the olfactorybulb (Hu et al., 1996).

Correlation of netrin 1 effects with the pattern ofexpression in vivoThe pattern of expression of netrin 1 and their receptors in theearly development of the pontine nuclei is summarised in Fig.11. Some precerebellar neurons originating in the lRL migratealong the marginal stream to cross the midline and allocate inthe LRN and ECN contralateral to their site of origin. Incontrast, neurons migrating within the submarginal (inferiorolivary neurons) and pontomedullary streams (basilar pontineneurons) do not cross the midline, and populate nucleiipsilateral to their side of origin (Taber-Pierce, 1966; Bourratand Sotelo, 1990). The present study shows that pontine nucleicell progenitors express unc5h2 and neogenin receptor genesbut not Dcc, whereas, postmitotic pontine neurons along theanterior extramural migratory pathway express the oppositecombination of receptors, i.e., high Dcc and no neogenin,unc5h2 or unc5h3. Finally, as these neurons end theircircumferential migration near the midline, they re-express lowlevels of unc5h2 and unc5h3 genes. Conversely, netrin 1 isexpressed in the midline area near the target region (pontinenuclei), which suggests that a gradient of netrin 1 produced inthis area may be responsible for the unidirectional migrationof neurons towards the precerebellar nuclei. Also, the patternof receptor expression in vivo is consistent with current viewsthat DCC is a component of the receptor complex mediatingchemoattractive responses by netrin 1 along the migratory

S. Alcántara and others

1369Netrin 1 guides neuronal migration

pathway. Finally, the re-expression of the unc5h2 and unc5h3receptor genes when migrating pontine neurons approach thetarget region suggests that these receptors may prevent theseneurons from crossing the midline, thereby allowing netrin 1to act as a stop signal, as proposed for olivary neurons (Bloch-Gallego et al., 1999).

However, netrin 1 is also expressed in the floor plate regionclose to the precerebellar neuroepithelium, so additionalmechanisms may prevent young postmitotic neurons frommigrating towards this source of netrin 1. Possible mechanismsinclude the sequestration of netrin 1 by association with theextracellular matrix or by neogenin receptors (Kennedy et al.,1994; Serafini et al., 1994; Keino-Masu et al., 1996), which arehighly expressed near the site of netrin 1 expression, or thepresence of a non-permissive substrate for neuronal migrationin this brainstem area.

Age-dependent chemorepulsion of granule cellprogenitors and axons mediated by netrin 1During embryonic development, granule cell progenitors aregenerated in the uRL and migrate tangentially over thecerebellar anlage to form the EGL, where they re-enter themitotic cycle. The present data with E14-uRL explants showthat embryonic granule cell progenitors do not respond tonetrin 1, although they express the Dcc and neogenin receptorgenes. This is consistent with observations in the cerebellumof newborn netrin 1 defective mice showing an almost intactEGL layer (Bloch-Gallego et al., 1999). Since netrin 1 is notexpressed in the cerebellar anlage at early embryonic stages,Dcc expression in granule cell progenitors derived from theuRL may reflect a netrin 1-independent action for this receptor.It has been suggested that granule cell progenitors migratealong preexisting axonal tracts (Hynes et al., 1986). At

embryonic stages, the dorsal surface of the cerebellar anlage iscrossed by DCC-containing axons that overlap with and run inparallel with migrating granule cell progenitors (Fig. 10D). Asoccurs with other IgCAMS such as L1 or NCAM (reviewed inWalsh and Doherty, 1997), homophilic trans DCC interactionsbetween axons and neurons may contribute to the tangentialmigration of embryonic granule cell precursors. In the unc5h3null mutant embryos, granule cell progenitors extend morerostrally than in wild-type mice, indicating that unc5h3receptors transduce a chemorepulsive signal for tangentiallymigrating granule cell progenitors (Przyborski et al., 1998).The results in the unc5h3 mutant in vivo and our negativefinding with netrin 1 in vitro, suggest that unc5h3 receptorsmay bind additional ligands, as recently postulated inCaenorhabditis elegans (Colavita et al., 1998).

Recent publications (Wu et al., 1999; Hu, 1999) have shownthat tangentially migrating neuronal progenitors in the rostralmigratory stream and some neurons in the cerebral cortexmight be guided by long-range repulsive factors, the Slitproteins. These results, together with those reported here forprecerebellar neurons, strongly suggest that chemotacticmechanisms exert a guidance effect on tangentially migratingneurons. Although we have not observed chemotropic effectsof netrin 1 on embryonic granule cell progenitors, it will beinteresting to determine whether other diffusible factors, suchas Slit proteins, are guiding these progenitors.

Starting at birth, postmitotic granule cells descend to thedeep EGL and extend their axons, the parallel fibres (Ramóny Cajal, 1911; Miale and Sidman, 1961). Granule cell somatathen become oriented perpendicular to the axons and migrateradially through the molecular layer, attached to the Bergmannglia, to form the IGL (Rakic, 1971, 1981, 1985; Hatten andLiem, 1981). The present data show that netrin 1 exerts a strong

High Netrin-1

High DCC

High unc5H3

Low Netrin-1

High unc5H2 Low unc5H2

Low unc5H3

Low DCC

High Neogenin Low Neogenin

lRL

BP

AP

FP

SVZ

SLVL

RP

PNPMS

IC

R

VZ

C

B

E14 E17

AC B

PMS

SVZVZ

ICP

STT

PMS

Fig. 11. Schematic drawings illustrating the pontinemigratory stream in relation to the space-temporaldistribution of netrin 1 and its receptors. (A) Dorsolateralview of an E14-E17 brain showing the localization of thepontine migratory stream in green and the plane of thesections showed in B and C. The arrow in A indicate thecaudorostral direction of the pontine migration. (B) Schemeof a coronal section at E14 illustrating the distribution ofnetrin 1 and its receptors at the origin of the pontinemigratory stream (PMS). Note that the PMS only expressDcc and that areas of low unc5h2 and neogenin expressionsurround the migratory pathway. (C) Scheme of a coronalsection at the level of the midbrain and pons at E17showing the destination of the neurons migrating along thePMS in the pontine nuclei. Pontine neurons start to expressunc5h2 and unc5h3 receptors and downregulate Dccexpression at the end of their migration. AP, alar plate; BP,basal plate; FP, floor plate; IC, inferior colliculus; ICP,inferior cerebellar peduncle; lRL, lower rhombic lip; PMS,pontine migratory stream; PN, pontine nuclei; SL, sulcuslimitans; STT, spinal trigeminal tract; SVZ, subventricularzone; R, red nucleus; RP, raphe nucleus; VL, ventrallemniscus; VZ, ventricular zone.

1370

chemorepulsive effect on axons growing from postnatal EGLexplants, which were characterised as parallel fibres on thebasis of their immunoreactivity for L1 and TAG-1 proteins(Persohn and Schachner, 1987; Yamamoto et al., 1990). Sucha chemorepulsive effect is age-dependent, and thus moredramatic at P5 than at P0, which again correlates with theperiod of parallel fibre formation in vivo (Altman and Bayer,1997). In the postnatal cerebellum, netrin 1 is expressed in theEGL and interneurons of the molecular layer. We suggest thatnetrin 1 contributes to the initiation of the correct parallel fibreextension, which could be a prerequisite for the ordered exit ofpostmitotic premigratory granule cells from the EGL(Engelkamp et al., 1999). The coexpression of Dcc, unc5h2and unc5h3 receptor genes in granule cell precursors as wellas in postmitotic granule cells is consistent with the view thatco-participation of DCC and unc5 receptors is required fornetrin 1 to exert a chemorepulsive response (Hong et al., 1999).

In addition to axons, our data show migrating neuronsleaving the explants in the quadrants distal to the source ofnetrin 1, which are identified as granule cells because of L1and TAG-1 immunostaining. We believe that this observationreflects an action of netrin 1 on migrating neurons for thefollowing reasons. (1) Granule cell perikarya outside theexplants are exclusively observed in the distal quadrants, butnot in the proximal quadrants where granule cell axons alsogrow, which rules out the possibility that migrating cells movepassively through the parallel fibres. (2) The rostral cerebellarmalformation, which results from the disruption of the unc5h3gene, leads to abnormal migration and positioning of granulecells, which implies a possible role for netrin 1 signallingthrough this receptor in cerebellar migration. (3) unc5h2 andunc5h3 receptors are highly expressed in premigratory granulecell bodies. However, Bergmann fibres did not leave our EGLexplants and the granule cells migrate along fascicles of TAG-1-positive fibres belonging to parallel fibres. We thus interpretour results as showing that the non-gliophilic migrationobserved in vitro mimics the tangential migration of granulecells within the deeper EGL, just before their inwards radialmigration (Ryder and Cepko, 1994), and that this tangentialmigration is under the control of netrin 1 and its receptors.

Netrin 1, a factor involved in axonal growth andneuronal migrationNetrin 1 secreted by the floor-plate acts as a chemoattractantfactor for ventrally directed axonal navigation (Kennedy et al.,1994; Keino-Masu et al., 1996; Shirasaki et al., 1995, 1996)and as a chemorepellent cue for dorsally oriented axons(Colamarino and Tessier-Lavigne, 1995). The present datashow that netrin 1 is involved in the growth and targeting ofintrinsic cerebellar circuits, which are independent of themidline, by providing chemorepulsive signals to developingparallel fibres. Thus, together with reports on corticofugalfibres (Métin et al., 1997; Richards et al., 1997) and ongoingstudies on the hippocampus (J. Barallobre, personalcommunication), the present data indicate that netrin 1 is alsoinvolved in the guidance of some ipsilateral projections and intarget-layer selection.

More importantly, the present results show that netrin 1exerts a direct chemoattractive influence on precerebellarmigrating neurons along the circumferential extramural stream,and a chemorepulsive effect on granule cell migration. These

data, together with our previous study in the olivary nucleus(Bloch-Gallego et al., 1999) and the analysis of un5h3 mutants(Przyborski et al., 1998), show a novel role for netrin 1 inneuronal migration, which, as for axonal growth, may be eitherchemoattractant or chemorepellent. Thus, together with othermolecules with a role in neuronal migration and axonal growth,such as reelin or semaphorin III (D’Arcangelo et al., 1995; DelRio et al., 1997; Eickholt et al., 1999), netrin 1 is likely to havea dual role in the construction of neural networks and in theguidance of migrating neurons towards their target nuclei andlayers.

This work was supported by Grants SAF98-106 (ComisionInterministerial de Ciencia y Tecnología, Spain), by the HumanScience Frontier Program and by Marató TV3 Foundation (Spain) toE. S., and by the Institut National de la Santé et de la RechercheMédicale financial support to C. S.; S. A. is a recipient of MECpostdoctoral fellowship (Spain). F. de C. has EC Fellowship ERB-4001-GT-970077. We are indebted to Dr D. Karagogeos for thegenerous gift of anti-TAG-1 antibodies, Dr M. Fabre and Dr F. X. Realfor anti-DCC antibodies, Dr M. Schachner for anti-L1 antibodies, D.Le Cren for photographic work and R. Rycroft for editorial assistance.We are specially gratefull to Dr M. Tessier-Lavigne for providing uscDNA probes and cell lines.

REFERENCES

Ackerman, S., Kozak, L. P., Przyborski, S. A., Rund, L. A., Boyer, B. B.and Knowles, B. B. (1997). The mouse rostral cerebellar malformationgene encodes an UNC-5-like protein. Nature 386, 838-842.

Alcántara, S., Ruiz, M., D’Arcangelo, G., Ezan, F., de Lecea, L., Curran,T., Sotelo, C. and Soriano, E. (1998). Regional and cellular patterns ofreelin mRNA expression in the forebrain of the developing and adult mouse.J. Neurosci. 18, 7779-7799.

Altman, J. and Bayer, S. A. (1997). Development of Cerebellar System inRelation to its Evolution, Structure and Functions. Boca Raton, FL: CRCPress.

Anderson, S. A., Eisenstat, D. D., Shi, L. and Rubenstein, J. L. R. (1997).Interneuron migration from basal forebrain to neocortex: dependence onDLX genes. Science 278, 474-476.

Bloch-Gallego, E., Ezan, F., Tessier-Lavigne, M. and Sotelo, C. (1999).Floor plate and netrin-1 are involved in the migration and survival of inferiorolivary neurons. J. Neurosci. 19, 4407-4420.

Bourrat, F. and Sotelo, C. (1988). Migratory pathways and neuriticdifferentiation of inferior olivary neurons in the rat embryo: Axonal tracingstudy using the in vitro slab technique. Dev. Brain Res. 39, 19-37.

Bourrat, F. and Sotelo, C. (1990). Early development of the precerebellarsystem: Migratory routes, Selective aggregation and neuritic differentiationof the inferior olive and lateral reticular nucleus neurons. An overview. Arch.Ital. Biol. 128, 151-170.

Bourrat, F. and Sotelo, C. (1991). Relationships between neuronal birthdatesand cytoarchitecture in the rat inferior olivary complex. J. Comp. Neurol.313, 509-521.

Buttiglione, M., Revest, J. M., Rougon, G. and Faivre-Sarrailh, C. (1996).F3 neuronal adhesion molecule controls outgrowth and fasciculation ofcerebellar granule cell neurites: a cell type specific mediated by the Ig-likedomains. Mol. Cell Neurosci. 8, 53-69.

Buttiglione, M., Revest, J. M., Pavlov, O., Karagogeos, D., Furley, A.,Rougon, G. and Faivre-Sarrailh, C. (1998). A functional interactionbetween the neuronal adhesion molecules TAG-1 and F3 modulates neuriteoutgrowth and fasciculation of cerebellar granule cells. J. Neurosci. 17,6853-6870.

Cho, K. R. and Fearon, E. R. (1995). DCC: linking tumor suppressor genesand altered cell surface interactions in cancer? Curr. Opin. Genet. Dev. 5,72-78.

Colavita, A., Krishna, S., Zheng, H., Padget, R. W. and Culotti, J. G.(1998). Pioneer axon guidance by UNC-129. a C. elegans TGF-β. Science281, 706-709.

Colamarino, S. A. and Tessier-Lavigne, M. (1995). The axonal

S. Alcántara and others

1371Netrin 1 guides neuronal migration

chemoattractant netrin-1 is also a chemorepellent for trochlear motor axons.Cell 81, 621-629.

Culotti, J. G. and Kolodkin, A. L. (1996). Functions of netrins andsemaphorins in axon guidance. Curr. Opin. Neurobiol. 6, 81-88.

D’Arcangelo, G., Miao, G. G., Chen, S.-C., Soares, H. D., Morgan, J. I.and Curran, T. (1995). A protein related to extracellular matrix proteinsdeleted in the mouse mutant reeler. Nature 374, 719-723.

Del Río, J. A., Heimrich, B., Borrell, V., Förster, E., Drakew, A., Alcántara,S., Nakajima, K., Miyata, T., Ogawa, M., Mikoshiba, K., Derer, P.,Frotscher, M. and Soriano, E. (1997). An essential role for Cajal-Retziuscells and reelin in the layer-specific targeting of hippocampal afferents.Nature 385, 70-74.

Eickholt, B. J., Mackenzie, S. L., Graham A., Walsh, F. S. and Doherty, P.(1999). Evidence for collapsin-1 functioning in the control of neural crestmigration in both trunk and hindbrain regions. Development 126, 2181-2189.

Ellerberger, C., Hanaway, J. and Netsky, M. G. (1969). Embryogenesis ofthe inferior olivary nucleus in the rat: A radioautographic study andreevaluation of the rhombic lip. J. Comp. Neurol. 137, 71-88.

Engelkamp, D., Rashbas, P., Seawright, A. and van Heyningen, V. (1999).Role of Pax6 in development of the cerebellar system. Development 126,3585-3596.

Fabre M., Martin, M., Ulloa, F. and Real, F. X. (1999). In vitro analysis ofthe role of DCC in mucus-secreting intestinal differentiation. Int. J. Cancer81, 799-807.

Fazeli, A., Dickinson, S. L., Hermiston. M. L., Tighe, R. V., Steen, R. G.,Small, R. G., Stoeckli, E. T., Keino-Masu, K., Masu M., Rayburn, H.,Simons, J., Bronson, R. T., Gordon, J. I., Tessier-Lavigne, M. andWeinberg, R. A. (1997). Phenotype of mice lacking functional deleted incolorectal cancer (DCC) gene. Nature 386, 796-804.

Felsenfeld, D. P., Hynes M. A., Skoler, K. M., Furley, A. J. and Jessell, T. M.(1994). TAG-1 can mediate homophilic binding but neurite outgrowth onTAG-1 requires a L1-like molecule and β1 integrins. Neuron 12, 257-267.

Gao, W.-Q. and Hatten, M. E. (1994). Immortalizing oncogenes subvert theestablishment of granule cell identity in developing cerebellum.Development 120, 1059-1070.

Goodman, C. S. (1996). Mechanisms and molecules that control growth coneguidance. Annu. Rev. Neurosci 19, 341-377.

Guthrie, S. and Pini, A. (1995). Chemorepulsion of developing motor axonsby the floor plate. Neuron 14, 1117-1130.

Hamelin, M., Zhou, Y., Su, M. W. and Culotti, J. G. (1993). Expression ofthe UNC-5 guidance receptor in the touch neurons of C. elegans steers theiraxons dorsally. Nature 364, 327-330.

Harkmark, W. (1954). Cell migrations from the rhombic lip to the inferiorolive, the nucleus raphe and the pons. A morphological and experimentalinvestigation on chick embryos. J. Comp. Neurol. 100, 115-205.

Hatten, M. E. and Liem, R. K. H. (1981). Astroglia provide a template forthe positioning of developing cerebellar neurons in vitro. J. Cell Biol. 90,622-630.

His, W. (1891). Die Entwicklung des menschlichen Rautenhirns vom Ende desersten bis zum Beginn des dritten Monats. I. Verlängertes Mark. Abhandlungder königlicher sächsischen Gesellschaft der Wissenschaften,Mathematischephysikalische Klasse 29, 1-74.

Hong, K., Hinck, L., Nishiyama, M., Poo, M., Tessier-Lavigne, M. andStein, E. (1999). A ligand-gated association between cytoplasmic domainsof UNC5 and DCC family receptors converts Netrin-induced growth coneattraction to repulsion. Cell 97, 927-942.

Hu, H. (1999). Chemorepulsion of neuronal migration by Slit2 in thedeveloping mammalian forebrain. Neuron 23, 703-711.

Hu, H. and Rutishauser, U. (1996). A septum-derived chemorepulsive factorfor migrating olfactory interneuron precursors. Neuron 16, 933-940.

Hu, H., Tomasiewics, H., Magnuson, T. and Rutishauser, U. (1996). Therole of polysialic acid in migration of olfactory bulb interneuron precursorsin the subventricular zone. Neuron 16, 735-743.

Hynes, R. O., Patel, R. and Miller, R. H. (1986). Migration of neuroblastsalong preexisting axonal tracts during prenatal cerebellar development. J.Neurosci. 6, 867-876.

Jacobowitz, D. M. and Abbott, L. C. (1998). Chemoarchitectonic Atlas ofthe Developing Mouse Brain. Boca Raton, FL: CRC Press.

Jankovski, A. and Sotelo, C. (1996). Subventricular zone-olfactory bulbmigratory pathway in the adult mouse: cellular composition and specificityas determined by heterochronic and heterotopic transplantation. J. Comp.Neurol. 371, 376-396.

Karagogeos, D., Morton, S. B., Casano, F., Dodd, J. and Jessell, T. M.

(1991). Developmental expression of the axonal glycoprotein TAG-1:Differential regulation by central and peripheral neurons in vitro.Development 112, 51-67.

Keino-Masu K., Masu, M., Hinck, L., Leonardo, E. D., Chan, S. S. Y.,Culotti, J. G. and Tessier-Lavigne, M. (1996). Deleted in colorectal cancer(DCC) encodes a Netrin receptor. Cell 87, 175-185.

Kennedy, T. E., Serafini, T., de la Torre, J. and Tessier-Lavigne, M. (1994).Netrins are diffussible chemotropic factors for commissural axons in theembryonic spinal cord. Cell 78, 425-435.

Lemmon, V., Farr, K. L. and Lagenaur, C. (1989). L1-mediated axonoutgrowth occurs via a homophilic binding mechanism. Neuron 2, 1597-1603.

Leonardo, E. D., Hinck, L., Masu, M., Keino-Masu, K., Ackerman, S. L.and Tessier-Lavigne, M. (1997). Vertebrate homologues of C. elegansUNC-5 are candidate Netrin receptors. Nature 386, 833-837.

Leung-Hagesteijn, C., Spence, A. M., Stern, B. D., Zhou, Y., Su, M.-W.,Edgecock, E. M. and Culotti, J. G. (1992). UNC-5, a transmembraneprotein with immunoglobulin and thrombospondin type I domains, guidescell and pioneer axon migrations in C. elegans. Cell 71, 289-299

Lois, C., García-Verdugo, J. M. and Alvarez-Buylla, A. (1996). Chainmigration of neuronal precursors. Science 271, 978-981.

Lumsden, A. G. S. and Davies, A. M. (1986). Chemotropic effect of specifictarget epithelium in the development of the mammalian central nervoussystem. Nature 323, 538-539.

Mathis, L., Bonnerot, C., Puelles, L. and Nicolas, J.-F. (1997). Retrospectiveclonal analysis of the cerebellum using genetic lac Z/lacZ mouse mosaics.Development 124, 4089-4104.

Mehlen, P., Rabizadeh, S., Snipas, S. J., Ass-Munt, N., Salvesen, G. S. andBredesen, D. E. (1998). The DCC gene product induces apoptosis by amechanism requiring receptor proteolysis. Nature 395, 801-804.

Métin, C., Deléglise, D., Serafini, T., Kennedy, T. E. and Tessier-Lavigne,M. (1997). A role for netrin-1 in the guidance of cortical efferents.Development 124, 5063-5074.

Miale, I. and Sidman,R. L. (1961). An autoradiographic analysis ofhistogenesis in the mouse cerebellum. Exp. Neurol. 4, 277-296.

Moody, S. A., Quigg, M. S. and Frankfurter, A. (1989). Development of theperipheral trigeminal system in the chick revealed by an isotype-specificanti-beta-tubulin monoclonal antibody. J. Comp. Neurol. 279, 567-580.

Ono, K. and Kawamura, K. (1989). Migration of immature neurons alongtangentially oriented fibers in the subpial part of the fetal mouse medullaoblongata. Exp. Brain Res. 78, 290-300.

Ono, K. and Kawamura, K. (1990). Mode of neuronal migration of thepontine stream in fetal mice. Anat. Embryol. 182, 11-19.

Pearlman, A. L., Faust, P. L., Hatten, M. E. and Brunstrom, J. E. (1998).New directions for neuronal migration. Curr. Opin. Neurobiol. 8, 45-54.

Persohn, E., and Schachner, M. (1987). Immunoelectron microscopiclocalization of the neural adhesion molecules L1 and N-CAM duringpostnatal development of the rat cerebellum. J. Cell. Biol. 105, 569-576.

Placzek, M., Tessier-Lavigne, M., Jessell, T. M. and Dodd, J. (1990).Orientation of commisural axons in vitro in response to a floor plate-derivedchemoattractant. Development 110, 19-30.

Przyborski, S. A., Knowles, B. B. and Ackerman, S. L. (1998). Embryonicphenotype of Unc5h3 mutant mice suggest chemorepulsion during theformation of the rostral cerebellar boundary. Development 125, 41-50.

Rakic, P. (1971). Neuron-glia relationship during cell migration in developingcerebellar cortex: A Golgi and electron microscopic study in Macacusrhesus. J. Comp. Neurol. 141, 283-312

Rakic, P. (1981). Neuron-glial interaction during brain development. TrendsNeurosci. 4, 184-187.

Rakic, P. (1985). Contact regulation of neuronal migration. In The Cell inContact. (eds. G. E. Edelman and J. P.Thiery), pp 67-90. New York: J. Wiley.

Rakic, P. (1990). Principles of neural migration. Experientia 46, 882-891.Ramón y Cajal, S. (1911). Histologie du Système Nerveux de l’Homme et des

Vértébres. Maloine, Paris. Richards, L. J., Koester, S. E., Tuttle, R. and O’Leary, D. D. M. (1997).

Directed growth of early cortical axons is influenced by a chemoattractantreleased from an intermediate target. J. Neurosci. 17, 2445-2458.

Ryder, E. F. and Cepko, C. L. (1994). Migration patterns of clonally relatedgranule cells and their progenitors in the developing chick cerebellum.Neuron 12, 1011-1029.

Serafini, T., Kennedy, T. E., Galko, M. J., Mirzayan, C., Jessell, T. M. andTessier-Lavigne, M. (1994). The Netrins define a family of axon outgrowth-promoting proteins homologous to C. elegans UNC-6. Cell 78, 409-424.

Serafini, T., Colamarino, S. A., Leonardo, E. D., Wang, E., Beddington,

1372

R., Skarnes, W. C. and Tessier-Lavigne, M. (1996). netrin-1 is requiredfor commissural axon guidance in the developing vertebrate nervous system.Cell 87, 1001-1014.

Shirasaki, R., Tamada, A., Katsumata, R. and Murakami, F. (1995).Guidance of cerebellofugal axons in the rat embryo: Directed growth towardthe floor plate and subsequent elongation along the longitudinal axis. Neuron14, 961-972.

Shirasaki, R., Mirzayan, C., Tessier-Lavigne, M. and Murakami, F. (1996).Guidance of circumferentially growing axons by Netrin-dependent and -independent floor plate chemotropism in the vertebrate brain. Neuron 17,1079-1088.

Soriano, E., Alvarado-Mallart, R. M., Dumesnil, N., Del Río, J. A. andSotelo, C. (1997). Cajal-Retzius cells regulate the radial glia phenotype inthe adult and developing cerebellum and alter granule cell migration. Neuron18, 563-577.

Stoeckli, E. T. and Landmesser, L. T (1995). Axonin-1, Nr.CAM and Ng-CAM play different roles in the in vivo guidance of chick commissuralneurons. Neuron 14, 1165-1179.

Taber-Pierce, E. (1966). Histogenesis of the nuclei griseum pontis, corporispontobulbaris and reticularis tegmenti pontis (Bechterew) in the mouse. J.Comp. Neurol. 129, 219-240.

Tamada, A., Shirasaki, R. and Murakami, F. (1995). Floor platechemoattracts crossed axons and chemorepels uncrossed axons in thevertebrate brain. Neuron 14, 1083-1093.

Tessier-Lavigne, M. and Goodman, C. S. (1996). The molecular biology ofaxon guidance. Science 274, 1123-1132.

Varela-Echevarría, A., Tucker, A., Püschel, A. W. and Guthrie, S. (1997).Motor axon subpopulations respond differentially to the chemorepellentsnetrin-1 and semaphorin D. Neuron 18, 193-207.

Walsh, F. S. and Doherty, P. (1997). Neural cell adhesion molecules of theimmunoglobulin superfamily: Role in axon growth and guidance. Ann. Rev.Cell Dev. Biol. 13, 425-456.

Wassef, M., Zanetta, J. P., Brehier, A. and Sotelo, C. (1985). Transientbiochemical compartmentalization of Purkinje cells during early cerebellardevelopment. Dev. Biol. 111, 129-137.

Wichterle, H., García-Verdugo, J. and Alvarez-Buylla, A. (1997). Directevidence for homotypic, glia-independent neuronal migration. Neuron 18,779-791.

Wolfer, D. P., Henehan-Beatty, A., Stoeckli, E. T., Sondererger, P. andLipp, H. P. (1994). Distribution of TAG-1/Axonin-1 in fibre tracts andmigratory streams of the developing mouse nervous system. J. Comp.Neurol. 345, 1-32.

Wu, W., Wong, K., Chen, J-h., Jiang Z-h, Dupuis, S., Wu, J. and Rao, Y.(1999). Directional guidance of neuronal migration in the olfactory systemby the protein Slit. Nature 400, 331-336.

Yamamoto, M., Hassinger, L. and Crandall, J. E. (1990). Ultrastructurallocalization of stage-specific neurite associated proteins in the developingrat cerebral and cerebellar cortices. J. Neurocytol. 19, 619-627.

S. Alcántara and others