The Journal of Neuroscience, February 1995, f5(2): 981-989

Motility and Cytoskeletal Organization of Migrating Cerebellar Granule

Rodolfo J. Rivas and Mary E. Hatten The Rockefeller University, New York, New York 10021-6399

To characterize CNS neuronal precursor migration along as- The cerebellar granule has provided an experimental troglial fibers, we examined the motility of the migratory lead- model for CNS migration, in part becausethe specializedmove- ing process and cytoskeletal-based mechanisms of loco- ment of granule neurons along glial fibers can be reproduced in motion of early postnatal mouse cerebellar granule neurons an in vitro system(Edmondson and Hatten, 1987;Hatten, 1990). in vitro. To visualize the surface motility of the leading pro- During migration, the granule neuron is highly polarized in the cess, granule neurons were labeled with the fluorescent li- direction of cell movement. The rostra1portion of the elongated pophilic dye, PKH-26, and imaged by time lapse fluores- cell body tapers in the direction of migration into a thickened cence microscopy. The motile behavior and cytoskeletal migratory leading processthat extends along and wraps around organization of the migrating neuron had several distinctive the glial fiber (Rakic, 1971; Edmondsonand Hatten, 1987). The features. As the migrating neuron moved along the glial fiber, saltatory movement of the neuronal cell , moving and the leading process rapidly extended, projecting up to 40 pausing in regular intervals (Edmondson and Hatten, 1987) pm, and retracted, withdrawing towards the cell soma. Broad resemblesthe cyclic advance of migrating fibroblasts (Aber- lamellipodia were common along the entire length of the crombie et al., 1970; Abercrombie, 1980). This view is sup- leading process, giving it a ruffled appearance. Within the ported by the findings of Dunn and Heath (Dunn and Heath, cell soma, a cage-like distribution of microtubules encircled 1976) who showed that while fibroblast locomotion is random the and actin filaments formed a subcortical rim on flat substrates,movement is oriented when the cells move underneath the plasma membrane. Disruption of actin fila- on a thin-cylindrical substrate. In this model, the growing tip ments with cytochalasin B inhibited migration, suggesting of the leading processwould advance along the thin glial fiber involvement of actin subunit assembly in neuronal migration. by rapid extension and retraction of the leading process,anal- Both microtubules and actin filaments were heavily concen- ogousto the rapid advance and retreat of the fibroblast leading trated in the leading process; the leading process did not lamella, with a net movement in the direction of migration. show the development of a distinct actin-rich domain at its Two lines of evidence suggesta role for the cytoskeleton in tip. neuronal motility along astroglial fibers. First, by correlated [Key words: neuronalmigration, cytoskeleton, cellpolarity, video and electron microscopy, a system of longitudinally ori- ] ented microtubules extends from a juxtanuclear basalbody into the leadingprocess (Gregory et al., 1988). The microtubule sys- In the developing mammalian brain, neuronal migration estab- tem has been proposed to underlie an oriented flow of mem- lishesthe laminar pattern of cortical regions, ushering young branous elements and vesiclesfrom the soma into the leading neuronsfrom ventricular zones where they are generatedto the process,generating a net forward flow of the neuronalcytoplasm layers where they establish synaptic relationships (Ramon y (Hatten, 1993). Second, cytoskeletal elements are thought to Cajal, 1911; Sidman and Rakic, 1973). Although considerable underlie the two classesof neuron- appositionsseen during progresshas been made towards identifying the primary path- migration-a specializedinterstitial junction along the interface ways for neuronal migration (Rakic, 1990; Gray and Sanes, of the cell soma and glial fiber, and punctae adherentia along 1991; O’Rourke et al., 1992; Hatten, 1993) and the adhesion the length of the leading process.Along the interstitial junction, systemsthat function in directed migration alongthe radial glial the intracellular space contains thin fibers that are contiguous fiber system (Edmondson et al., 1988; Sanes,1989; Fishell and with, or are transmembranousextensions of, submembranous Hatten, 1991; Fishman and Hatten, 1993) the cytoskeletal elements that appear to attach to microtubules (Gregory et al., structures underlying the specialized motility of migrating neu- 1988). Moreover, antibody perturbation studies provide evi- rons during CNS development remain uncharacterized. dencethat antibodies againstthe neuron-glia ligand astrotactin significantly reduce the rate of neuronal migration, induce dis- organization of cytoskeletal components,and causewithdrawal of the leading process(Fishell and Hatten, 1991). In particular, Received Mar. 3, 1994; revised June 30, 1994; accepted July 28, 1994. treatment of granule neurons with anti-astrotactin antibodies We gratefully acknowledge the advice of Drs. Gord Fishell and Renata Fishman results in tanglesof cytoskeletal elementsin the area rostra1to and thank Dr. Fishell and Dr. Sharon Powell for readina the manuscript. Peter Peirce provided expert assistance with photomicrographs. This work was sup- the nucleus(Fishell and Hatten, 1991). This result has suggested ported by NIH Postdoctoral Fellowship NS08834 (R.J.R.) and NIH Grant NS that the neuronal cytoskeleton interacts with neuron-glia re- 15429 (M.E.H.). ceptor systemsto produce movement of the cell soma. Correspondence should be addressed to Mary E. Hatten, The Rockefeller Uni- versity, 1230 York Avenue, New York, New York 10021-6399. To provide a detailed view of the motile behavior of the Copyright 0 1995 Society for Neuroscience 0270-6474/95/150981-09$05.00/O leading process,we labeledgranule neurons,purified from early 982 Rivas and Hatten * Organization of Migrating Cerebellar Granule Neurons postnatal mouse cerebellum, with the fluorescent lipophilic dye, 568 nm illumination from a krypton-argon multiline laser. A Nomarski PIU-I-26 (Horan and Slezak, 1989; Gao et al., 1992). The labeled image was obtained by transmitting the laser light through an optic fiber cable. Alternatively, PKH-labeled cells were imaged using a Hamamatsu neurons were plated on unlabeled glial fibers and migration was 2400-50/80 I-CCD camera head and controller. Epifluorescent excita- visualized by time lapse fluorescence microscopy. This allowed tion/emission was at 510/550 nm. Three neutral density filters were us to view the motility of the leading process separately from added to reduce the excitation light level by 99%. For time lapse re- the surface activity of the underlying glial fiber. During migra- cordingsusing the I-CCD, oneimage (100 msec exposure) was acquired every 3-5 min. tion the leading process rapidly extended, projecting up to 40 Disruption of cytoskeletal structures of migrating neurons with cyto- pm, and retracted, withdrawing towards the rostra1 portion of chalasin B or nocodazole. To infuse agents that disrupt cytoskeletal the cell soma. Lamellipodial and filopodial extension occurred elements, the cover of the microculture system was modified. Rather along the entire length of the leading process. Treatment of than sealing the microwell with a second #l coverslip, the petri dish migrating neurons with cytochalasin B, a drug that disrupts actin was filled with glial-conditioned medium (14 ml). The culture was sealed with a petri dish cover fitted with a “window” (obtained by drilling a filaments (Forscher and Smith, 1988; Letoumeau et al., 1987), hole (14 mm) in the top and sealing with a # 1 coverslip). Additions of rapidly arrested neuronal migration, suggesting that actin sub- concentrated stocks of nocodazole or cytochalasin were made through unit assembly functions in neuronal locomotion along the glial a second hole (6 mm) in the plastic petri dish cover which was sealed substrate. In contrast to the growth cone, where actin filaments with Parafilm before and after drug additions. For drug addition ex- periments, Nomarski images were acquired with a Zeiss Axiovert mi- and microtubules are separated into distinct peripheral and cen- croscope fitted with a long working distance condenser. tral domains, the leading process did not show the development Immunocytochemical procedures. Tubulin and actin were localized of a distinct actin-rich domain at its tip. by a procedure modified from Wang et al. (1982). In brief, cells were washed once with calcium-free Hanks’ balanced salt solution and once Materials and Methods with cytoskeleton buffer (CB; Hanks’ BSS plus 2 mM MgCl,, 2 mM Purijication of cerebellar granule neurons and astroglial cells. Cell sus- EGTA, 5 mM PIPES, pH 6.9). The cells were then treated for 1 min pensionsenriched in cerebellargranule neurons (95%) and astroglia (5%) with CB plus 0.25% (w/v) glutaraldehyde, 0.5% (w/v) Triton X-100, were purifiedby Percollgradient sedimentation from C57B1/6J mice rinsed in CB, and then postfixed (1% glutaraldehyde in CB, 10 min). on postnataldays 5 or 6 (Hatten, 1985).Cells were plated in serum- After further permeabilizationin 0.5% Triton X-100 in CB (10 min), supplementedmedium (BME plus 10%horse serum, 9 mg/mlglucose, aldehyde autofluorescence was quenched with 1 mg/ml NaBH, in cy- 0.3 mg/ml glutamine,50 U/ml penicillin,and 50 j&ml streptomycin) toskeleton buffer (three 10 min incubations). The cells were washed in asdescribed (Edmondson and Hatten, 1987)on glasscoverslip micro- CB and actin microfilaments were stained with rhodamine-phalloidin culturewells pretreated with laminin(Collaborative Research, Lexing- (Molecular Probes). For tubulin staining, cells were rinsed with antibody ton, MA; 10 j&cm2 in CMF-PBS at 4°C for 12-24 hr), and maintained buffer (PBS plus 1% Triton X-100) and stained sequentially with a at 35.5”C with 100% humidity and 5% CO,. mouse monoclonal antibody to /3-tubulin (Amersham) and fluorescein- Dye labeling of puriJied granule neurons. A purified (>99%) popu- conjugated, goat anti-mouse and rabbit anti-goat, affinity-purified sec- lation of immature granule cells was prepared as described by Gao et ondaryantibodies (Cappel). In someexperiments, granule neurons were al. (Gao et al., 1991) and labeled with the fluorescent lipophilic dye stained for actin filaments and microtubules sequentially with BODIPY PKH-26 (Gao et al., 1992).Labeled granule cells were then addedto phalloidin (MolecularProbes; 558 nm excitationmaximum, 568 nm cultures of unlabeled cerebellar granule cells (95%) and astroglia (5%), emission maximum), mouse monoclonal antibody to P-tubulin (Amer- which had been plated 24 hr earlier to allow extension of the glial fiber sham),and affinity-purified, fluorescein-conjugated, goat anti-mouse an- system in the cultures. tiserum.Fluorescently stained cultures were scanned for neuronshaving Preparation of glass fibers. Glass fibers of a geometry and diameter the elongated, bipolar shape typical of migrating neurons (Edmondson similar to radial glial fibers (0.5-2 pm in width and up to 100 pm in and Hatten, 1987). length)were preparedfrom Whatmanglass fiber filters as described (Fishman and Hatten, 1993). In the present experiments, glass fibers Results were coatedwith polylysine(100 &ml; 2 hr at roomtemperature) and then with laminin (50 rg/ml; 4 hr at 4°C) prior to placement in glass The leading processis highly dynamic coverslip microwells (see above) pretreated with polylysine (100 pg/ml). Previous views of glial-guided neuronal migration in vitro using To provide trophic support for the purified granule cells, astroglial cells VEC-DIC microscopy showed high-resolution details of migra- were plated on a coverslip, which was inverted as a “ceiling” over the microwell. tion of the neuronal soma and the proximal portion of the Migration of dye-labeled granule neurons on native cerebellar glial leadingprocess (Edmondson and Hatten, 1987; Fishell and Hat- fibers. Migration of granule neurons along glial processes was monitored ten, 1991). Although VEC-DIC microscopy allowed visualiza- with time lapse video microscopy. Cultures were washed twice with tion of the leading processin real time, the small diameter of Leibovitz’s L-15 medium (GIBCO) containing Redu-Ser II (Upstate the process(less than 1 km) and the motility of the surface of Biotechnology), 10 mM HEPES, 0.3 mg/ml glutamine, 50 U/ml peni- cillin, 50 &ml streptomycin, and 9 mg/ml glucose. An 18 mm coverslip the underlying glial fiber made it difficult to image leading pro- was mounted over the microwell (Edmondson and Hatten, 1987), and cessmotility. To provide a detailed view of the movementsof the sealed chamber was placed on the stage of a Zeiss Axiovert micro- the leading process,in isolation from the surfaceactivity of the scope. The specimen, objective, and condenser were heated to 33-35°C underlying glial fiber, freshly purified granule neuronswere la- by a Zeiss TRZ 3700 heating stage. Images of migration were acquired with three optical systems. For video-enhanced contrast beled with the fluorescent lipophilic membrane dye, PKH-26 differential-interference contrast (VEC-DIC) microscopy, cells were (Horan and Slezak, 1989; Gao et al., 1991, 1992; Gao and mounted on a Zeiss Axiovert-35 inverted microscope equipped with Hatten, 1993) and were added to neuron-glial migration cultures differential interference contrast optics (20 x /0.5, 40 x /0.75, 63 x / 1.4, that had been plated earlier to allow extension of the glial fiber or 100 xi 1.3 NA objectives). Images were acquired with a Hamamatsu system. PKH-labeled neurons attached to glial fibers and ap- C 1965 video camera. A Uniblitz shutter and T 132 shutter driver were controlled by the Image-l system (Universal Imaging). Images were pearedindistinguishable from unlabeled neuronsby differential- recorded with a Panasonic optical memory disk recorder every 3 min interference contrast microscopy, adopting the elongatedbipolar in the case of low magnification assays, and every 3-5 min in the case shapecharacteristic of migrating neurons(Edmondson and Hat- of high magnification assays. ten, 1987; Gregory et al., 1988; Fishell and Hatten, 1991). Using Images ofgranule cells labeled with PKH-26 were acquired with either confocal microscopy or with a light-intensified charged-coupled device PKH-26 we were able to image the entire extent of the leading (I-CCD). Confocal microscopy was performed using a Zeiss Axiovert processmoving through a complex culture environment of glial microscope fitted with a Bio-Rad MRC 600 confocal scan head with processesand neurites. The Journal of Neuroscience, February 1995, U(2) 983

Figure 1. PKH-26 labeling of migrating granule neurons. Freshly purified granule neurons were labeled with the fluorescent lipophilic membrane dye, PKH-26, and added to neuron-glial migration cultures. The labeled neurons attached to extended glial processes in the cultures and were examined by time lapse fluorescencemicroscopy with an I-CCD camera.Highly motile lamellipodia(large arrow in u-e) werepresent along the lengthof the leadingprocess. A juxtanucleararea of fluorescentlylabeled organelles (long, thin nrrows in u-e), which apparentlytook up the dye from the plasmamembrane by endocytosis,was located just rostra1to the caudallyoriented nucleus. Fluorescently labeled vesicles (short arrows in a. c, e) could be seenstreaming down the leadingprocess during migration.A short processwas observed trailing from the migratingcell (open arrowhead). Time points (min): u, 0; b, 10; c, 33; d, 67; e, 139. Scale bar, 5 urn.

A novel finding revealed with PKH-26 labeling was that mi- area of densestaining of membranousorganelles just rostra1to grating neurons in vitro had highly elongated leading processes the nucleus (Fig. 2b). This juxtanuclear area has been reported (1O-40 pm in length) extending along the underlying glial fiber in ultrastructural studiesto contain microtubule-associatedor- (seeFigs. 1, 5u-c). During migration, the leading processex- ganellessuch as ER, mitochondria, the Golgi apparatus, and tended away from, and then retracted towards the cell soma in basal bodies (Gregory et al., 1988). an accordion-like fashion over a period of minutes or hours (see The nucleusremained in the posterior portion of the cell soma Figs. 1, 5u-c). The average length of the leading process, as during neuronal migration (seeFigs. 1, 2, 4, 5). The rostra1end determined from fluorescencemicrographs, was 30.3 f 6.3 pm of the nucleus frequently showed an indentation, or notch, in (mean + SEM); extensions or retractions of the leadingprocess the direction of migration. This notch was seenin both time during migration of the cell soma averaged 8.9 f: 2.1 pm. In lapseobservations of PKH-labeled neurons(not shown) and by someinstances, the leadingprocess retracted into the cell soma examination of fixed neurons stained with DiOC,(3) (Fig. 26), and then reextendedin the direction ofcell migration (not shown). confirming previous video and ultrastructural studies (Ed- Altematively,‘as the neuron reversedthe direction of migration, mondson and Hatten, 1987; Gregory et al., 1988). Thus, the the leading process withdrew into the soma and reextended migrating granule neuron is highly polarized in the direction of along the glial fiber in the direction opposite to the previous migration, positioning the nucleusin the posterior aspectof the direction of cell migration. PKH-26 labelingrevealed numerous cell and localizing membranousorganelles in a juxtanuclear area broad lamellipodial extensions along the entire length of the just rostra1to nucleus. leading process(see Figs. 1, 5u-c). Rapid extension and retrac- tion of lamellipodia was common, with occasionalfine filopodia Organization of microtubules in migrating neurons evident. The numerouslamellipodial extensionsalong the lead- Evidence for a role for microtubules in neuronal migration has ing processgave it a ruffled appearance(see Figs. 1,5u-c). Thus, come from electron microscopic studiesrevealing an association during migration the granule neuron extends and retracts a of submembranousmicrotubules with the interstitial junction, thickened, highly dynamic leading processwith lamellipodial the site of adhesionbetween the neuronal cell somaand the glial and filopodial extensions all along its length. process(Gregory et al., 1988). To view the organization of mi- crotubules in the cell somaand the leading processof migrating Polarity of the migrating neuron cells, we immunostained migrating neuronswith a monoclonal As,the neuron moved along the glial guide, fluorescently labeled antibody against /3-tubulin. By immunostaining, microtubules internal structures were visible, owing to an apparent uptake of formed a “cage-like” web of filaments circumscribing the nu- PKH-26 from the surface membrane. Dense fluorescencewas cleus in the caudal aspect of the cell (Fig. 3~). Rostra1to the seenin the cell somajust rostra1to the nucleusin the direction nucleus, a densearray of longitudinally oriented microtubules of migration; this staining probably representedendosomes, ly- extended into the leading process (Fig. 3~; cf. Gregory et al., sosomes,or the Golgi (seeFigs. 1, 5u-c). This view was con- 1988). firmed by staining for internal membranes with the dicarbo- Although optical sectioning of immunolabeled microtubules cyanine dye, DiOC,(3) (Terasaki, 1989), which also showed an by confocal microscopy delineated the cage of microtubules 984 Rivas and Hatten l Organization of Migrating Cerebellar Granule Neurons

Figure 2. DiOC,(3) stainingof intracellularmembranes in migrating granuleneurons. The dicarbocyaninedye DiOC,(3) wasused to stain mitochondria,endoplasmic reticulum, and other internalmembranes in glutaraldehyde-fixedneuron-glial migration cultures, which were then examinedby confocalmicroscopy. a, Nomarskimicrograph of granule neuronmigrating on a glial fiber. The directionof migrationis towards the bottomof the panel.b, Fluorescenceconfocal image corresponding to the cell in a. A brightjuxtanuclear area of fluorescentlylabeled in- ternalmembranes (singlelongarrow) wasobserved rostra1 to the nucleus in the directionof migration.A distinctivenotch in the nucleus(double arrows in b) pointedin the directionof migration.Fluorescently labeled vesicles(short arrow) could alsobe seenin the leadingprocess. Scale Figure 3. Distributionof microtubulesand actin filamentsin migrating bar, 4 pm. granuleneurons. Neuron-glial migration cultures were fixed andstained by indirectimmunofluorescence with P-tubulinantibody (a) or, alter- natively, with rhodamine-labeledphalloidin to stainfilamentous actin surrounding the nucleus, the organization of microtubules in (b). Cultureswere examined by confocalmicroscopy. The directionof z-planesnear the interstitial junction and in regions of the lead- migrationis towards the bottomofthe figure.a, Examinationofneurons ing processthat apposed the glial fiber was difficult to assess. with migrationprofiles revealed a cagelikenetwork of microtubulesin To view neuronal microtubules in isolation from glial micro- the neuronalcell body (double arrows) and longitudinallyarrayed mi- crotubulesextending into the leadingprocess (single large arrow). The tubules, we immunostained granule cells migrating along lam- microtubulesin the underlyingglial fiber are indicatedby the open inin-coated glassfibers (Fishman and Hatten, 1993). By video arrowhead.b, Filamentousactin wasparticularly concentrated in the microscopy, the dynamics and general mode of movement of leadingprocess (/urge arrow) and in a subcorticalrim in the cell body granule cell migration along glassfibers closely resemblesneu- (doublearrows). Actin filamentstaining in the underlyingglial fiber is rons migratingalong ghal processes(Fishman and Hatten, 1993). indicatedby the open arrowhead. Scalebar, 4 pm. Immunocytochemical localization of microtubules confirmed the perinuclear, cage-like distribution of microtubules seenin actin filaments in migrating granule neurons, we visualized fil- cells migrating on native astroglial fibers (Fig. 4b,d). Within the amentousactin by rhodamine-phalloidin staining. Within the leading processes,a densearray of longitudinally oriented mi- cell soma, actin staining was concentrated in a thin, subcortical crotubules could be seenspiraling around the shaft of the glass rim underneath the plasma membrane (Fig. 3b). Filamentous fiber (Fig. 4c,d). actin was also seenin lamellipodial and filopodial extensions of the leading process(Fig. 3b). To discriminate actin networks Organization of actin$laments in migrating neurons in the leading processfrom those in the underlying glial fiber, The regulation of assemblyof actin subunitsplays a critical role we imaged the actin filaments of granule cells migrating on glass in generating fibroblast locomotion (Abercrombie et al., 1970; fibers. The organization of actin filaments in granule cells un- Bray and White, 1988; Stossel,1993) and in guiding the direc- dergoingmigration alongglass fibers closelyresembled that seen tionality of the growth cone (Letoumeau and Ressler, 1983; for cells migrating on native glial fibers. By phalloidin staining, Marsh and Letoumeau, 1984; Letoumeau et al., 1987; Forscher a thin rim of cortical actin was seenunder the cell soma of the and Smith, 1988; Mitchison and Kirschner, 1988; Smith, 1988; migrating cell, with heavy staining observed along the length of Bridgman and Dailey, 1989). To examine the organization of the migratory process(see Fig. 86). The Journal of Neuroscience, February 1995. 75(2) 985

Figure 4. Distributionof microtubulesin granuleneurons migrating on laminin-coatedglass fibers. Granule neurons migrating on laminin-coated glassfibers were fixed, stainedby indirectimmunofluorescence with a monoclonalantibody to @-tubulin,and examinedby confocalmicroscopy. a, Nomarskimicrograph of migratinggranule neuron. The directionof migrationis towardsthe bottom of the panel.!&, Confocalfluorescence imagescorresponding to samecell as in a. Opticalsections started 1.5 pm (b) from the top of the cell body and weretaken towards (d) the glass covcrslip;optical sectionsare 1.0 pm apart. Note the cage-likenetwork of microtubulesin the cell body (singlearrow in b and d). Spiraling microtubulescould be seencoursing down the leadingprocess (double arrow in c and d). Scalebar, 4 pm.

cells were treated with nocodazole, a drug that disrupts micro- Disruption of the organization of actin jilaments in migrating tubules (Bamburg et al., 1986; Joshi et al., 1986). In contrast to granule cells resultswith cytochalasin B, treatment of neuronal-glial cultures To examine the contribution of actin filamentsto the movement with either 10 or 20 I.LM nocodazole had more general effects, of neuronsalong the glial fiber, migrating cells were treated with inducing a collapse of both glial processesand net&es. The cytochalasin B, an agent that disrupts actin filaments (Marsh neuronal cell bodies were pulled towards the glial cell bodies as and Letourneau, 1984; Letourneau et al., 1987; Forscher and the glial processesretracted, apparently due to the rapid collapse Smith, 1988). A computer-assistedVEC-DIC microscopy assay of microtubules within the glial process(not shown). This made was used to track the behavior of migrating granule neurons it difficult to determine neuronal migration ratesafter drug treat- before and after drug treatment (Fishell and Hatten, 1991). A ment and quantitation was not performed. marked reduction in the movement of migrating neuronsalong glial processesoccurred after treatment with cytochalasin B. Granule cell extension: comparison with cell migration Migration rates in untreated cells averaged 24.1 f 8.6 rcmlhr; To compare the basic mechanismsunderlying the extension of after a 2 hr treatment with cytochalasin B, migration rates av- the granule cell neurite with the locomotion of the cell soma eraged3.3 + 1.7 pm/hr (10 PM cytochalasin B) or 6.9 -t 3.4 along the glial fiber, we first examined the motility of granule pm/hr (20 PM cytochalasin B). By VEC-DIC microscopy, the cell neurites. In contrast to the leading process,the shaft of the cell soma of cytochalasin-treated neurons assumeda rounded granule cell neurite was smooth (Fig. 6a-c), ending in a growth profile characteristic ofstationary neurons(Gregory et al., 1988). cone with numerous filopodia and lamellipodia (Fig. 6a). In- Notably, treated cells remained attached to the glial processes, deed, the extension of filopodia and lamellipodia in the growing suggestingthat disruption of actin filaments was not sufficient neurite was generally confined to the growth cone (cf. Harrison, to disrupt neuron-glial adhesion. 19 10; Ramon y Cajal, 1911; Weiss, 1941; Bray and Hollenbeck, The effect of cytochalasin B treatment on the motility of the 1988; Bray and White, 1988). By time lapsemicroscopy, as the leadingprocess was examined in PKH-26 labeledneurons using growth cone advanced, the neuritic shaft remained stable and time-lapse confocal microscopy. Treatment with cytochalasin did not undergocyclic extension and retraction towardsthe cell B appeared to freeze the broad lamellipodia enwrapping the soma (not shown). Thus, a key feature, evident in time lapse glial fiber within a period of a few minutes (Fig. 5d,e). Over a recordings of PKH-labeled granule cell net&es, was growth period of 20-30 min, the leading processappeared to collapse cone locomotion resulting in net growth of the axon. Granule into a flattened expanse of membrane, and motility of the cell cell neurites were observed to project up to several hundred soma halted (Fig. Sd,e). Although lamellipodial and filopodial micrometers along the culture surfaceover an observation pe- activity was rapidly reduced, movement of intracellular vesicles riod of several days in vitro. was seenin PKH-26 labeled cells after the addition of cyto- Localization of @-tubulin and actin filaments within granule chalasinB, suggestingthat organelleand vesicle transport in the neuron growth cones revealed a densemicrotubular system in cell somawas not affected, at least in the short term (not shown). the shaft of the granule cell neurite, with two cytoplasmic do- Theseresults suggestthat actin filaments are necessaryboth for mains evident within the growth cone region (cf. Forscher and the motility of the leading processand for the movement of the Smith, 1988; Mitchison and Kirschner, 1988; Bridgman and neuronal cell soma. Dailey, 1989; Rivas et al., 1992). Whereas the central domain To assessthe role of microtubules in neuronal migration, the of the growth cone was rich in microtubules (Fig. 7b,c), actin 988 Rivas and Hatten * Organization of Migrating Cereballar Granule Neurons

Figure 5. Effectof cytochalasinB on granuleneuron migration. PIG-I-26 labeled granule neurons were observed in neuron-glialmigration cultures by timelapse confocal microscopy before (u-c) and after addition(d, e) of 10 PM cytochalasinB. u-c, In untreatedcells, PKH-26 wasparticularly concentratedin the juxtanucleararea (single small arrows) rostra1to the nucleus.Large, highly motile lamellipodia(large arrow) werepresent in the leadingprocess. A short processtrailed the migratingcell (open arrowhead in b and c). d and e, Addition of cytochalasinB to the culture mediumwithin minutescaused a lossof motility of lamellipodiain the leadingprocess (large arrow). The time point in d is 9 min after the addition of cytochalasinB; e, 45 min after the additionof the drug. The juxtanucleararea (smd arrows) of fluorescentlylabeled organelles was no longer polarizedin the directionof migrationafter the additionof cytochalasinB, althoughmovement of vesiclesand fluorescentlylabeled membranous tubulescould still be observed.Time points(hours, minutes): a, 00:35; b, 3:42;c, 4:Ol; d, 4:16;e, 4:52. Scalebar, 6 pm. filament labelingwas intensein the periphery of the growth cone that the leading processundergoes an accordion-like extension (Fig. 7~2,~).In contrast to growth cones, both microtubule and and retraction during neuronal migration along the glial guide. actin filament staining was evident along the length of the lead- While it is clear that there is overlap in some aspectsof the ing process(Fig. 8), with double-labeling experiments demon- motility and cytoskeletal organization of the growth cone and strating that a distinct actin-rich domain was not evident at the leading process,the leading processhas several distinctive fea- tip of the leading process(Fig. 8b,c). tures. Perhaps the most dramatic is the rapid extension and retraction of the leading processduring cell locomotion. This Discussion differed from granule cell axon extension becausethe motility In the present experiments, visualization of cerebellar granule of the growth cone results, over time, in net lengthening of the neurons with the fluorescent lipophilic dye PHK-26 revealed axon. Indeed, granule cell extend over distancesof several

Figure 6. Extensionof granulecell neurites.A singleneurite from a PKH-26 labeledgranule neuron was observed by time lapsefluorescence confocalmicroscopy extending on the laminin-coatedglass substrate of a neuron-glialmigration culture at three differenttime points(u-c). A highly motilegrowth cone (long arrow in u-c) is seenat the endingof a long,thin neurite(large arrow in u-c). Motile activity of lamellipodiaand filopodiawas confined to the growthcone. Time points(min): a, 0; b, 21; c, 54. Scalebar, 4 pm. The Journal of Neuroscience, February 1995, 15(2) 987

Fgure 7. Distributton of microtubules and actin filaments in granule cell neurites. Granule neurons extending neurites on a polylysine substrate were fixed and double stained for /3-tubulin using indirect tmmunofluorescence and for actin filaments using BODIPY phalloidin. The same cell is shown in (u-c). Well-spread growth cones were observed on polylysine, with a heavy concentration ofactin filaments in the peripheral lamellipodium (double arrows in a). Longitudinally oriented microtubules were observed in the neurites; these microtubules ended in the central domain of the growth cone (arrow in b). c, Digital merge of a and b showing distinct actin-rich peripheral (double arrows) and microtubule-rich central domains. Redfluorescence indicates actin, greenfluorescence indicates tubulin. Scale bar, 7 pm. hundred micrometers both in vitro and in vivo (Ramon y Cajal, for movement. Second, the cage-like web of microtubules seen 19 11; Gao et al., 199 1, 1992; Gao and Hatten, 1993). One clear in the cell soma may function to maintain caudal positioning difference in the cytoskeletal organization of the leading process of the nucleus as the rostra1 aspect of the cell extends. Finally, is the absence of a distinct actin-rich domain at its tip. We the array of longitudinal microtubules in the leading process propose that these differences in cytoskeletal organization and may serve to orient the directionality of movement by providing motility reflect initial steps in the development of granule cell a substrate for the directed flow of membranous organelles seen polarity. in migrating cells (Edmondson and Hatten, 1987). Cytoskeletal organization of the migratory process Comparison of cell migration with axon extension A network of subcortical actin is thought to provide mechanical In the present study, several differences in the motility and support to the surface of the cell, enabling the cell to undergo cytoskeletal organization of granule cell neurites and the mi- shape changes associated with locomotion. Since receptors for gratory leading process were evident. First, while the surface of several classes of cell adhesion molecules bind actin filaments the extending neuritic shaft was relatively quiescent (cf. Bray et within the subcortical network, this network and the overlying al., 1978; Joshi et al., 1986) with filopodial and lamellipodial surface membrane are often considered to be a single functional extensions confined primarily to the growth cone region, the unit (Bray and White, 1988; Luna, 199 1; Luna and Hitt, 1992). entire length of the migratory process was “ruffled” with mem- In the present experiments, the subcortical actin filaments ob- branous extensions. Moreover, the leading process underwent served in the neuronal soma may function in the inchworm- rapid extension and retraction as the cell moved along the glial like extension and contraction of the soma characteristic of CNS guide; in contrast, forward movement of the growth cone led migration along astroglial fibers (Hatten, 1990). The dense con- over time to net extension of the neurite. The cyclic extension centration of actin filaments seen in the leading process of cells and retraction of the leading process during cell locomotion undergoing migration is likely to function in the extension and suggests that the microtubules of the leading process may be retraction of lamellipodia and filopodia along the leading pro- less stable than that of extending neurites (cf. Mitchison and cess. The arrest of cell locomotion observed after cytochalasin Kirschner, 1988; Bulinski and Gundersen, 199 1). Finally, B treatment suggests that the assembly of actin subunits pro- whereas the growing tip of the granule cell neurite has a central vides an important mechanism for cell migration. region rich in microtubules and a peripheral region rich in actin In addition to actin-based mechanisms, it is likely that the filaments, the leading process did not have a distinct actin-rich microtubules of migrating neurons are specialized for locomo- domain. tion of the cell soma. Several roles for microtubules in CNS cell migration are suggested by the current experiments. First, the Development of cell polarity in the migrating neuron accordion-like extension and retraction of the leading process Migrating neurons are highly polarized in the direction of mi- along the glial guide suggests that the microtubules of the leading gration, with the nucleus positioned in the caudal aspect of the process are highly unstable, raising the possibility that assembly soma and microtubule-associated organelles, such as the ER, and disassembly of microtubules provides a net forward force Golgi, and centrioles concentrated just rostra1 to the nucleus 988 Rivas and Hatten * Organization of Migrating Cerebellar Granule Neurons

Granule cell axonal extension in vivo is segregated both spa- tially and temporally from cell body migration and subsequent formation (Ramon y Cajal, 1911; Gao and Hatten, 1993). This spatiotemporal segregation is reflected both by dif- ferential gene expression at the various stages of granule cell development (Kuhar et al., 1993) and by differential protein distributions in the membrane domains of the developing neu- ron. A number of adhesive glycoproteins of the Ig superfamily (reviewed in Faivre-Sarrailh and Rougon, 1993) including Ll (Pershon and Schachner, 1987) and the glycosylphosphatidyli- nositol-linked molecules F3/Fll (Faivre-Sarrailh et al., 1992) and TAG-l (Furley et al., 1990; Yamamoto et al., 1986, 1990) are localized only to the parallel fiber axons, not to the migratory or postmigratory granule cell bodies, leading processes, or den- drites. TAG-l is also localized exclusively to the long, thin parallel fiber axon-like neurites we have observed in vitro (S. K. Powell, R. J. Rivas, E. Rodriguez-Boulan, and M. E. Hatten, unpublished observations). The distinctive features of cytoskeletal organization and mo- tility in the leading process are likely to subserve different pro- tein expression patterns and functions from those of the growth cone. Thus, the development of distinct differences between the leading process and growth cone reflect the overall program of granule cell development, where changing patterns of gene ex- pression mark the spatially and temporally discrete steps of axon extension, migration, and dendrite formation (Kuhar et al., 1993). In vivo, as visualized by implantation of PKH-labeled granule neurons back into cerebellar cortex, this spatiotemporal segre- gation is a striking feature of granule cell identity (Gao and Hatten, 1993, 1994; Rivas and Hatten, unpublished observa- tions), with implanted neurons developing leading process and axonal morphologies remarkably similar to those we have ob- served in the present experiments. Thus, the in vitro systems offer the opportunity to dissect the molecular events controlling targeting of proteins to distinct cellular locations in the granule neuron and the intracellular signaling events that control this specialized form of cytoskeletal organization. References Abercrombie M (1980) The crawling movement of metazoan cells. Proc R Sot Land [Biol] 207:129-147. Abercrombie M, Heaysman JEM, Pegrum SM (1970) The locomotion of fibroblasts in culture I. Movements of the leading edge. Exp Cell Res 59:393-398. Figure 8. Distribution of microtubules and actin filaments in granule Bamburg JR, Bray D, Chapman K (1986) Assembly of microtubules neurons migrating on laminin-coated glass fibers. Granule neurons mi- at the tip of growing axons. Nature 321:788-790. grating on polylysine laminin-coated glass fibers were fixed and double Bray D, Hollenbeck PJ (1988) Growth cone motility and guidance. stained for fl-tubulin by indirect immunofluorescence and for actin fil- Annu Rev Cell Biol 4:43-6 1. aments with BODIPY phalloidin. The same cell is shown in a-c. The Brav D. White JG (1988) Cortical flow in animal cells. Science 239: micrograph was taken from the same culture shown in Figure 7; the 883-888. . ’ glass coverslip substrate is polylysine. a, Nomarski image of granule Bray D, Thomas C, Shaw G (1978) Growth cone formation in cultures neuron with migration profile on glass fiber. b, Actin filament staining. of sensorv neurons. Proc Nat1 Acad Sci USA 75:5226-5229. c, Tubulin staining. Actin filament and microtubule distributions ap- Bridgman PC, Dailey ME (1989) The organization of myosin and peared to overlap extensively in the leading process (arrows in b and a&n in rapid frozen growth cones. J Cell Biol 108:95-109. c). Scale bar, 7 pm. Bulinski JC. Gundersen GC (1991) Stabilization and post-transla- tional modification of microtubules during cellular morphogenesis. Bioessays 13:285-293. Craig AM, Jareb M, Banker G (1992) Neuronal polarity. Curr Opinion (Gregory et al., 1988). The presence of ribosomes and Golgi Neurobiol 2:602-606. elements in the leading process (Gregory et al., 1988) gives it a Dunn GA, Heath JP (1976) A new hypothesis of contact guidance in dendritic, rather than axonal character (cf. Bridgman and Dail- tissue cells. Exp Cell Res 101: 1-14. ey, 1989; Craig et al., 1992; Rodriguez-Boulan and Powell, 1992). Edmondson JC, Hatten ME (1987) Glial-guided granule neuron mi- Observations of the leading process of migrating neurons led gration in vitro: a high-resolution time-lapse video microscopic study. J Neurosci 7: 1928-l 934. Ramon y Cajal to suggest that the leading process eventually Edmondson JC, Liem RKH, Kuster JE, Hatten ME (1988) Astrotactin, transforms into the seen in postmigratory neurons a novel cell surface antigen that mediates neuron-glia interactions in (Ramon y Cajal, 19 11). cerebellar microcultures. J Cell Biol 106:505-517. The Journal of Neuroscience, February 1995, 15(2) 999

Faivre-Sarrailh C, Rougon G (1993) Are the glypiated adhesion mol- Letoumeau PC, Shattuck TA, Ressler AH (1987) “Pull” and “push” ecules preferentially targeted to the axonal compartment? Mol Neu- in neurite elongation: observations on the effects of different concen- robiol 7:49-60. trations of cytochalasin B and taxol. Cell Motil Cytoskel 8: 193-209. Faivre-Sarrailh C, Gennarini G, Goridis C, Rougon G (1992) F3/Fll Luna EJ (199 1) Molecular links between the cytoskeleton and mem- cell surface molecule expression in the developing mouse cerebellum branes, Curr Opinion Cell Biol 3: 120-I 26. is polarized at synaptic sites and within granule cells. J Neurosci 12: Luna EJ, Hitt AL (1992) Cytoskeleton-plasma membrane interactions. 257-267. Science 258:955-964. Fishell G, Hatten ME (199 1) Astrotactin provides a receptor system Marsh L, Letoumeau PC (1984) Growth of neurites without filopodial for CNS neuronal migration. Development 113:755-765. or lamellipodial activity in the presence of cytochalasin B. J Cell Biol Fishman R, Hatten ME (1993) Multiple receptor systems promote 9912041-2047. CNS neuronal migration. J Neurosci 13:3485-3495. Mitchison T, Kirschner M (1988) Cytoskeletal dynamics and nerve Forscher P, Smith S (1988) Actions of cytochalasins on the organi- growth. Neuron 1:761-772. zation of actin filaments and microtubules in a neuronal growth cone. O’Rourke NA, Dailey ME, Smith SJ, McConnell SK (1992) Diverse J Cell Biol 107:1505-1516. migratory pathways in the developing . Science 258: Furley AJ, Morton SB, Manalo D, Karagogeos D, Dodd J, Jesse11TM 299-302. (1990) The axonal glycoprotein TAG- 1 is an immunoglobulin su- Persohn E, Schachner M (1987) Immunoelectron microscopic local- perfamily member with neurite outgrowth-promoting activity. Cell ization of the neural cell adhesion molecules Ll and N-CAM during 61:1.57-170. nostnatal develonment of the cerebellum. J Cell Biol 105:569-576. Gao WQ, Hatten ME (1993) Neuronal differentiation rescued by im- Rakic P (197 1) Neuron-glia relationship during granule cell migration plantation of weaver granule cell precursors into wild-type cerebellar in developing cerebellar cortex. A golgi and electronmicroscopic study cortex. Science 260~367-369. in Mucacus rhesus. J Comp Neurol 14 1:283-3 12. Gao WQ, Heintz N, Hatten ME (199 1) neu- Rakic P (1990) Principles of neural cell migration. Experientia 46: rogenesis is regulated by cell-cell interactions in vitro. Neuron 6:705- 882-891. 715. Ramon y Cajal S (19 11) Histologie du systeme nerveux de l’homme Gao WQ, Liu XL, Hatten ME (1992) The weaver gene encodes a non- et des vertebres. Reprint (Madrid: Conseio Superior de Investiga- autonomous signal for CNS neuronal differentiation. Cell 68:841- ciones Cientificas, 1955). Maloine: Paris. - 854. Rivas RJ. Burmeister DW. Goldberg DJ (1992) Ranid effects of lam- Gasser UE, Hatten ME (1990) CNS neurons migrate on astroglial inin on the growth cone: Neuron Is: 107-l 15. ’ - fibers from heterotypic brain regions in vitro. Proc Nat1 Acad Sci USA Rodriguez-Boulan E, Powell SK (1992) Polarity of epithelial and neu- 87145434547. ronal cells. Annu Rev Cell Biol 8:395-427. Gray GE, Sanes JR (1991) Migratory paths and phenotypic choices Sanes J (1989) Extracellular matrix molecules that influence neural of clonally related cells in the avian optic tectum. Neuron 6:2 1 l-225. development. Annu Rev Neurosci 12:49 l-5 16. Gregory WA, Edmondson JC, Hatten ME, Mason CA (1988) Cytology Sidman RL, Rakic P (1973) Neuronal migration with special reference and neuron-glial apposition of migrating cerebellar granule cells in to developing : a review. Brain Res 62: l-35. vitro. J Neurosci 8: 1728-I 738. Smith SJ (1988) Neuronal cytomechanics: the actin-based motility of Harrison RG (19 10) The outgrowth of nerve fibers as a mode of growth cones. Science 242:708-715. orotoulasmic movement. J Exn Zoo1 9:797-846. Stossel TP (1993) On the crawling of animal cells. Science 260: 1086- H&ten ME (1985) Neuronal regulation of astroglial morphology and 1094. proliferation in vitro. J Cell Biol 100:384-396. Terasaki M (1989) Fluorescent labeling of endoplasmic reticulum. Hatten ME (1990) Riding the alial monorail: a common mechanism Methods Cell Biol 29: 125-135. for glial-guided migration in drfferent regions of the developing brain. Wang K, Feramisco JR, Ash JF (1982) Fluorescent localization of Trends Neurosci 13: 179-187. contractile proteins in tissue culture cells. Methods Enzymol85:5 14- Hatten ME (1993) The role of migration in central 562. neuronal development. Curr Opinion Neurobiol 3:3844. Weiss P (194 1) Nerve patterns: the mechanics ofnerve growth. Growth Horan PK, Slezak SE (1989) Stable cell membrane labeling. Nature [Suppl] 5: 163-203. 340:167-168. Yamamoto M, Boyer AM, Crandall JE, Edwards M, Tanaka H (1986) Joshi HC, Baas P, Chu DT, Heidemann SR (1986) The cytoskeleton Distribution of stage specific neurite-associated proteins in the de- of neurites after microtubule depolymerization. Exp Cell Res 163: veloping murine nervous system recognized by a monoclonal anti- 233-245. body. J Neurosci 613576-3594. Kuhar SG, Feng L, Vidan S, Ross ME, Hatten ME, Heintz N (1993) Yamamoto M, Hassinger L, Crandall JE (1990) Ultrastructural lo- Changing patterns of gene expression define four stages of cerebellar calization of stage-specific neurite-associated proteins in the devel- granule neuron differentiation. Development 117:97-l 04. oping rat cerebral and cerebellar cortices. J Neurocytol 19:6 19-627. Letoumeau PC, Ressler AH (1983) Differences in the organization of actin in the growth cones compared with the neurites of cultured neurons from chick embryos. J Cell Biol 97:963-973.