Journal of Neurocytology 9, 647-664 (1980)

Responses to cell contacts between growth cones, neurites and ganglionic non-neuronal cells N O R M A N K. W E S S E L L S 1, P A U L C . L E T O U R N E A U 2, R O B E R T P. N U T T A L L 3, M A R I L Y N L U D U E i q / A - A N D E R S O N 1 a n d J E R E M Y M . G E I D U S C H E K 1

1Department of Biological Sciences, Stanford University, Stanford, California 94305, U.S.A. 2Department of Anatomy, 4-135 Jackson HaIL University of Minnesota, Minneapolis, Minnesota 55455, U.S.A. 3Department of Biology, Emory University, Atlanta, Georgia 30322, U.S.A.

Received 5 February 1979; revised 31 October and 7 December 1979; accepted 7 December 1979

Summary The motility of growth cones of embryonic peripheral neurons is not inhibited by contact with the surfaces of neurites or of non-neuronal cells. Rather, growth cones and microspikes adhere to other cell surfaces and often respond with forward movement and elongation in contact with other ceils, as they do on adhesive surfaces in vitro. Furthermore, non-neuronal ceils do not display contact inhibition when they contact growth cones or neurites. If anything, surface motility and ruffling is stimulated by contact with a neuronal cell surface and some non-neuronal cells prefer to migrate along neurites rather than on the surface of the culture dish.

These observations on the contact behaviour of cells from peripheral nerve ganglia imply that the surfaces of embryonic neurons differ from those of non-neuronal ceils in that the neuronal surfaces do not elicit the typical contact ihhibition response.

Introduction The reaction of cells to contact wi th o ther cells and wi th extracellular surfaces m a y regulate cell m o v e m e n t s and cell associations in embryos . A we l l -known cell interact ion in vitro is contact inhibit ion of f ibroblast locomotion, def ined as the inhibi t ion of con t inued locomotion of a cell in the direct ion which p roduced collision wi th ano the r cell (Abercrombie, 1967; Harris, 1974; Trinkaus, 1976; Heaysman , 1978). The componen t s of contact inhibi t ion have been amply d o c u m e n t e d in reviews and need not be repea ted here (Harris, 1974; Heaysman , 1978). Contact

0300-4864/80/050647-18503.80/0 9 1980 Chapman and Hall Ltd.

648 WESSELLS, LETOURNEAU, NUTTALL, LUDUENA and GEIDUSCHEK

inhibition may occur in vivo as cells leave densely populated areas, for example, when neural crest cells migrate from the dorsal nerve tube (Ebendal, 1977; Tosney, 1978). However, other contact-dependent reactions may affect the social behaviour of embryonic cells, especially when the substratum for migration seems to be the surfaces of other cells, rather than an extracellular matrix (Trinkaus, 1976).

It is, therefore, of interest to examine the contact interactions of cells from nervous tissues, since neurons carry out extensive migration (in the form of neurite extension) and, in doing so, experience frequent cell-cell contacts prior to synaptogenesis. Likewise, neuroblasts and non-neuronal cells of nervous tissues, glia and Schwann cells, may make many contacts with other cells as they migrate (Speidel, 1933; Rakic, 1971; Lopresti et al., 1973). The importance of 'pioneer' fibres in the formation of nerve fascicles and the apparent guidance of Schwann cells and neuroblasts along axons and other cellular processes suggest that cells from nervous tissues may not show the typical contact inhibition phenomenon (Weiss, 1941; Jacobson, 1978).

Previous investigations of contacts between growing neurites have usually involved observations of intact embryonic ganglia embedded in plasma clots (see Dunn, 1971, for literature). The results are often difficult to interpret because the neurites twist about one another, and single neurites and growth cones cannot be resolved from multiple ones. In addition, t h e clots often undergo liquefaction, producing unstable culture conditions.

The results from these studies are conflicting. Nakai & Kawasaki (1959) and Nakai (1960) noted that filopodia of growth cones adhered to other cells or inanimate objects and exerted tension upon them. This phenomenon was proposed to be an important component of fasciculation, whereby a growth cone contacts and joins a bundle of axons. In contrast to this behaviour, Dunn (1971) described filopodial contacts with other axons that resulted in withdrawal of the filopodium and partial retraction of its growth cone. He termed this sequence 'contact inhibition of extension' and invoked it to explain the radial outgrowth of neurites from ganglia explanted in clots (see also, Ebendal, 1976).

In this paper, we report observations on growth cones of neurites cultured at low densities to facilitate both observation and interpretation. In addition, we employ stable planar culture substrata coated with polycations or with collagen, both relatively adhesive materials for neurons (Luduefla, 1973a; Letourneau, 1975a). The events of contact between growth cones, neurites and non-neuronal cells from ciliary and sensory ganglia have been recorded by time-lapse cinematography and compared to the typical events of fibroblastic cell contact inhibition. Our conclusion is that under our conditions, contact inhibition of motility is not a response in encounters between growth cones and neurites or between neurons and non-neuronal cells.

Contacts between peripheral ganglionic cells 649

Methods

Ciliary ganglia from 6-9-day chicken embryos were dissected and dissociated as in Helfand el aI. (1976). Specifically, ganglia were handled as follows: (1) incubation at 37 ~ C for 20 min in. Ca, Mg-free Hank's solution (CMFH); (2) then in 0.2% bactotrypsin (Difco) in CMFH at 370 C for 20 min; (3) following removal of the trypsin solution, nutrient medium containing 10% foetal calf serum was added and the ganglia were pelleted. The medium was removed and replaced, and ganglia were dissociated with an orally controlled pipette. The cells were repelleted and suspended in fresh heart conditioned medium (HCM; which is essential for survival and neurite growth under these conditions; Helfand et al., 1976). The cells were counted in a haemocytometer, and plated in HCM at 103-104 cells per 35 mm Falcon tissue culture plastic dish precoated with poly-L-ornithine HBR (Letourneau, 1975a).

Alternatively, 8-day chick embryonic dorsal root (sensory) ganglia were dissected, dissociated and cultured on untreated or polyornithine-coated surfaces in the presence of nerve growth factor (Letourneau, 1975a).

Ganglionic explants were prepared from 8-day embryonic dorsal root ganglia, which were trimmed and cut in half, and then 10-15 explants were placed in polyornithine-coated tissue culture dishes with nutrient medium (the medium was a modified F12 containing 10% by volume of foetal calf serum and with 10 ng/ml of purified fl-NGF per ml (gift of E. Shooter). The dishes were placed in a CO 2 incubator and then gently swirled to collect the ganglionic chunks near the centre of each dish, where they soon settled and attached.

Time-lapse cinematography (Zeiss optics, Bolex camera; Sage controls) was performed on cells after 12-36 h of culture. Warm mineral oil was sometimes layered over the culture medium to maintain pH during filming, and an air curtain incubator (Sage) was used 37~ Film speed varied from 4-58 frames/rain. The 16 mm movie film was analysed frame by frame; selected frames were photographed onto Panatomic X 35 mm film to prepare prints for publication (as in Wessells & Nuttall, 1978). The original films for Fig. 3 were underexposed so that the resultant 35 mm copies yielded grainy, low contrast prints. Hence, the original films were projected and tracings made directly for Fig. 3.

Growth cones were steered, using fine glass needles held in a Leitz micromanipulator (Wessells & Nuttall, 1978). Cells were fixed and prepared for scanning electron microscopy as in Helfand et aI. (1976). The gold-coated cells were examined in a field emission microscope (Coates & Welter).

Nomencla ture Both ciliary and dorsal root ganglia contain a mixture of neurons, Schwann (or satellite) cells and fibroblasts. Since the latter two types cannot be easily distinguished (see also Wood, 1976), such cells will be referred to as non-neuronal cells.

Results

N O N - N E U R O N A L - N O N - N E U R O N A L CELL CONTACZ: PRESENCE OF CONTACT INHIBITION

W h e n an active l amel l ipod ium of one n o n - n e u r o n a l cell touches the surface of another , typical ' contac t para lys is ' occurs (see Tr inkaus et al. , 1971, for definit ion and discussion). Ruffles and o ther surface m o v e m e n t s cease in the area of contact (Fig. 1) but not e l s ewhere on the t w o cells. In such cases, the cells have also been seen to

650 WESSELLS, LETOURNEAU, NUTTALL, LUDUEI~A and GEIDUSCHEK

migrate away from the contact site. We cannot easily distinguish fibroblasts from Schwann cells in our cultures and so cannot say with certainty that Schwann-Schwann contacts show contact inhibition. However, we have not observed an encounter between non-neuronal cells in, our cultures which failed to include contact inhibition. Most important in the current context is the conclusion that our culture conditions and polyornithine-treated substrata do not prohibit the expression of typical contact paralysis and inhibition by fibroblast-like cells.

NEURON-NEURON CONTACT

Growth cone--axon contact: lack of contact inhibition When an advancing growth cone touches the side of an axon, motile activity of the growth cone margin continues and no major area of inactivity is seen (equivalent to that in Fig. lf). Usually, the growth cone moves beneath the axon, in a kind of 'underlapping' (Fig. 2) (Boyde et al., 1969; Bell, 1977). This is the case despite apparently large areas of axonal adherence to the polyornithine- or collagen-coated substrata. Adherence is implied by the crooked shape of axons (Luduefia, 1973b; Letourneau, 1975a), interference reflection studies (Letourneau, 1979) and by results of displacing axons upward or sideways using fine needles (Wessells & Nuttall, 1978); in such cases only the locally displaced region moves, whereas the adjacent parts of the same axon keep their original configuration. In other cases, a growth cone that encounters an axon turns aside and moves along the axon; alternatively, the growth cone may immediately divide (Fig. 3) (Wessells & Nuttall, 1978) into two tips, which move in opposite directions along the encountered axon. We have not detected any relationship between specific behaviours and the angle at which growth cone and axon meet; for example, underlapping, bifurcation or movement along an axon all may occur when the angle of encounter is approximately 90 ~ Adhesion of the axon to the substratum may be a factor in the different behaviours when growth cone and axon meet, since interference reflection studies show varying degrees of contact between axons and a polyornithine substratum (Letourneau, 1979).

Fig. 1. Contact paralysis between non-neuronal cells from a ciliary ganglion. Ruffles are seen in A and B, as the lamellipodia approach each other. Contact has been established in C (large arrow), and ruffles continue nearby [arrows, (C, D)]. Subsequently, ruffling at this end of the lower cell is restricted to the left portion of the lameUipodium (L in F), while the area of contact (between large arrows in F) and adhesion remains 'paralysed'. Times in rain (m), seconds (s) after A. x 560. Fig. 2. Growth cone of a single parasympathetic motor neuron from a ciliary ganglion approaching an axon. Contact of a microspike in B; 'underlapping' started by C; the region of axon on the upper surface of the growth cone is displaced toward the left in C and D; ruffles on the cone occur in areas of direct contact with the axon [arrow, (E, G)] and the cone continues moving to the right, finally displacing the region of axon on its upper surface in that direction (cf. E, G, H). Times after A. x 420.

652 WESSELLS, L E T O U R N E A U , N U T T A L L , LUDUEI~A and G E I D U S C H E K

Fig. 3. Tracings of a projected film (see Methods) showing the growth cone of a ciliary ganglion neuron contacting an axon. Contact established at time b (arrow); by c and d, the cone commences to move along the axon; in c, d, and e the axon is bent and kinked; some very sharp kinks are seen in the projected film (arrows f, g). Adhesion between the growth cone and axonal surfaces must be sufficiently strong to allow this deformation of the axon to occur. Times: (a-b) 1 min; (b-c) 4 min 52 s; (c-d) 28 s; (d-e) 36 s; (e-f) i min; (f-g) 8 min 40 s; (g-h) 1 rain 20 s.

Growth cone--axon contact: adhesion The absence of contact inhibi t ion of g row th cones might be explained by a failure of g rowth cones to adhere firmly to o ther cell surfaces, a necessary event of contact inhibi t ion of f ibroblasts (Heaysman, 1978). The fol lowing observat ions, however , suggest that adhes ion does occur b e twe e n g rowth cones and o ther cells.

W h e n a g rowth cone under laps an axon, the axon is usually displaced slightly over the u p p e r surface of the cone, backward toward the base of the advancing cone (Fig. 2). Thus , the encoun te red axon is not p u s h e d ahead of the advancing cone. Rather it behaves as wou ld a small particle that is picked up at the front edge of a cone and m o v e d poster ior ly over the u p p e r surface of a g rowth cone (Nuttall, unpub l i shed t ime-lapse films; Bray, 1970). These observat ions imply that a g rowth

Contacts between peripheral ganglionic cells 653

cone and an axon can adhere sufficiently strongly to permit displacement of the taut axon to occur.

When a growth cone passes beneath an axon and continues migrating, a crossover point between the two axons may form (as seen in a later stage of the movie in Fig. 2). Such a crossover site seems to be a point of firm adhesion, since the growth cones and distal axons beyond the crossover may appear to tug and to exert tension upon the crossover point. Thus, such a crossover point is frequently not a site where the two cell surfaces simply slide over each other.

Another observation implies that firm cell-cell adhesion occurs between active growth cones and axons. In Fig. 3, a cone has bifurcated upon contacting an axon and its daughter cones maintain intense surface activity as they move apart along the axon. Note that the overlying axon becomes crinkled and kinked in the region between the two diverging cones. These deformations imply that the axon is subjected to forces acting backwards relative to the two cones and that these forces are sufficient to impose sharp bends upon the axon despite an extensive microtubular and neurofilamentous core in such axons (Yamada et al., 1971; Wessells, unpublished, for ciliary ganglion axons). It seems likely that the force which deforms this overlying axon is the same as that involved in displacement of particulate material on the upper surface of a cone or on the upper surface of a fibroblastic cell (see Harris & Dunn, 1972; Albrecht-Buehler & Goldman, 1976).

Finally, a related interaction has been recorded in a case where a microspike from a growth cone adhered to an adjacent axonal branch of the same neuron. The microspike shortened periodically (17 times in 43 min), each time displacing the axon to which its tip stuck. This corroborates the report of Nakai (1960), who filmed a case in which a microspike 'twitched' 22 times in 50 rain, each time pulling sideways an axon to which it adhered.

In summary, these examples all may be interpreted to demonstrate adherence between neuronal cell surfaces. Despite that, inhibition of motile activity is not seen when growth cones and axons touch.

Growth cone-growth cone contact Inhibition of growth cone motility was never observed when active growth cones met. After contact both cones remain motile, appear to crawl past each other and frequently continue moving along the other axon for some distance (Fig. 4). Careful analysis of our movies fails to reveal regions of quiescent cell surface where growth cones touch. Although there may be very small regions of quiescent surface beyond the resolution of our optical systems, microspikes and ruffles continue to form in the areas of growth cone contact.

Neither does there appear to be stimulation of surface motility at the sides or bases of two growth cones that meet, unlike the situation when fibroblasts meet, where lamellipodial extension elsewhere at the cell margins leads the cells away from the site of contact. Other observations of growth cones turning and experiments with cut

654 WESSELLS, L E T O U R N E A U , N U T T A L L , L U D U E N A and G E I D U S C H E K

axons reveal that surface motil i ty can begin from quiescent regions of a g rowth cone or along the sides of an axon and redirect neuri te g ro w th or p roduce a branch (Bray et al., 1978; Wessells & Nuttall, 1978; Wessells et al., 1978). Such de novo activity is not elicited w h e n g rowth cone meets g r o w th cone.

Contact between two parts of the same cell The above observat ions were made on g rowth cones undergo ing normal, undirected locomot ion and ones that were s teered (Wessells & Nuttall, 1978) to encounte r o ther cell surfaces. A fine needle ma y be used to induce branching of a g ro w th cone and the daugh te r cones may subsequent ly be s teered so as to encounte r each o ther directly or to encoun te r the side of the o ther axonal branch. In this case, behaviour is exactly as descr ibed above, g rowth cones move un d e r or along axons belonging to their o w n cell or g rowth cones touch bu t move past each o ther wi thou t loss of activity. Interestingly, no hint of fusion b e t w e e n the cell surfaces is seen even though two parts of the same cell are meet ing and even t h o u g h m e m b r a n e fusion has been hypo thes i zed to be par t of microspike regression wi thin a single g rowth cone (Spooner et al., 1974).

NEURON-NON-NEURONAL CELL CONTACT Growth cone-non-neuronal cell contact G r o w t h cones m a y migrate on the uppe r surface of non -neu rona l cells (Fig. 7). Such cones usual ly appear more bu lbous than do g rowth cones on plastic, collagen or polycat ion- t rea ted surfaces (Bray & Bunge, 1973; Ludueha , 1973b; Letourneau, 1975a; Hel fand et al., 1976). Howe v e r , microspikes and veils cont inue to form from the r o u n d e d cones and actual axonal e longat ion m ay occur on the up p e r surface of a non -neu rona l cell. Fig. 9 shows a rare case in which a g rowth cone has m o v e d to the front edge of a locomotory non-neurona l cell; in the films, ruffles cont inue to form on the non -neu rona l cell while, directly above, g rowth cone surface m o v em en t s also go on. Nei ther cell seems to inhibit the other ' s motile activity.

Unde r some condit ions, a high propor t ion of the sensory neurons that g row axons

Fig. 4. Two growth cones meeting (ciliary ganglion neurons). Contact is established at approximately time B; ruffling continues as the cone on the upper cell moves down toward the axon on the lower cell (ruffles near the tip of the advancing cone indicated with arrows in C, D, E, F); by time G the upper cone has underlapped the lower axon, and some of its veils extend beyond (arrows G, H); the cone then bifurcates and moves in both directions along the axon [limits of cone shown by arrows (I)]. Times after A. x 420. Fig. 5. An encounter between a non-neuronal cell and the axon of a ciliary motor neuron. One branch of the axonal outgrowth of a neuron (MN in inset) passes over the margin of a non-neuronal cell (cell body at N). A large vertical ruffle (R) of the cell margin region which contacts an axon (A) is typical and similar to those seen in the frames from the movie shown in Fig. 6. x 4140; inset, x 400.

656 WESSELLS, LETOURNEAU, NUTTALL, LUDUENA and GEIDUSCHEK

are situated on the upper surface of ganglionic non-neuronal cells or heart fibroblasts (Ludueha, 1973a; Letourneau, 1975a). Such axons may be profusely branched upon the upper surface of the non-neuronal cell (Fig. 7). While observing living cells with phase contrast microscopy, we have seen cases in which the growth cones of such branched axons remain near the edge of the upper surface on the non-neuronal cell and, though extraordinarily active, fail to move onto the plastic or collagenous substratum. The behaviour appears to be similar to that seen when a sensory or motor neuron growth cone moves along edges of palladium-polycation patterns: veils and microspikes contact both substrata but the growth cones remain upon the more adhesive substratum (as defined by other procedures; Letourneau, 1975a, b; Helfand et al., 1976). Thus, the observations of growth cone activity on top of non-neuronal cells could be interpreted to mean that adhesion between those cell types is greater than adhesion between the growth cone and the artificial substratum (for example, plastic, collagen; Letourneau, 1975b). This behaviour resembles that displayed by fibroblasts which do not cross a boundary onto a nonadhesive surface like agar (Abercrombie, 1967); this has been called contact inhibition, type 2 by Heaysman (1978).

Non-neuronal cell-neuron contact Non-neuronal cells in our cultures show no hint of paralysis of surface movements when an axon or an active growth cone is encountered. On the contrary, extraordinarily large ruffles often extend upward around the sides of the encountered growth cone or axon (Figs. 5, 6). The non-neuronal cell retains contact with the substratum and appears to migrate with its upper surface in a hyperactive state where the axon is encountered. We have never observed such large ruffles in the absence of contact with neurons.

In 72-120-h-old cultures of ciliarj ganglion cells, many highly elongated non-neuronal cells are seen to move to and fro along axons (Fig. 8; as seen by Nakai, 1956; Nakai & Kawasaki, 1959); ruffling may occur at one end, then the other, or both simultaneously. It is not clear in these cases whether the axon is the primary substratum for the non-neuronal cell, or whether both axon and dish are substrata. Because of our culture methods, Schwann cells and fibroblasts are not easily distinguished in this behavior towards axons, although Wood (1976) has reported a specific adhesive preference of Schwann cells for axons. Finally, it is emphasized that these examples of an absence of contact inhibition be tween non-neuronal cells and neurons occur in the same dish in which non-neuronal-non-neuronal cell contact inhibition occurs (Fig. 1).

Ganglion-ganglion interaction Our inability to observe contact inhibition by neurites in cell culture led us to examine the question in explants of whole sensory ganglia cultured on polyornithine-treated substrata. Usually when ganglia are cultured close to each

Con tac t s b e t w e e n pe r iphe ra l gangl ion ic ceils 657

Fig. 6. A sensory neuron 's growth cone contacting a non-neuronal cell. Contact has been made at A by the long microspike; the non-neuronal cell extends an intensely ruffling extension along the microspike (B-H); the limits of the ruffles are indicated with arrows in F-I; by times I-K, the most advanced ruffles of the non-neuronal cell are already beyond the edge of the photographs to the left; arrows in H - K indicate ruffles and veils that continue to form in areas of non-neuronal cell-neuron contact. Times after A. x 700.

Contacts between peripheral ganglionic cells 659

other in a semisolid plasma clot (Dunn, 1971), the region between the ganglia has a lower density of neurites than the rest of the periphery of those same ganglia. Dunn (1971) proposed that contact inhibition between neurites extending in opposite directions from the apposed ganglia could account for the low density, since growth cones might turn aside upon encountering ones coming in the other direction.

In our cultures, many axons extended from the ganglia but few non-neuronal cells left the explants (possibly because fibroblastic cell locomotion is inhibited by high adhesion to the substratum; Gaff & Boone, 1972). The density of axons between adjacent ganglia was just as high as elsewhere around the ganglia (Figs. 10, 11). There was no evidence of growth cones being turned aside by encounters with growth cones extending from other ganglia.

D i s c u s s i o n

Our observations of the reaction of growth cones to contact with each other, with neurites and with non-neuronal cells reveal differences from typical fibroblastic behaviour: (1) growth cones are not paralysed by contact with other cells and do not withdraw from cell contacts; instead they extend beneath or along encountered neurites and on top of non-neuronal cells; and (2) non-neuronal cells do not show contact inhibition when encountering neurites, rather, they are stimulated to extend their cell margins and migrate along neurites. It is unclear why growth cones do not show contact inhibition because the mechanism of fibroblastic contact inhibition itself is unclear. However, there are differences between neurite growth cones and fibroblasts in their adhesion to other cells and in the need to break adhesions during locomotion which may bear strongly upon why these two cell types react differently to cell contact. Some of these behaviours of growth cones resemble those of

Fig. 7. Sensory axons on the upper surface of a highly spread non-neuronal cell (G). Much branching has apparently taken place upon the upper surface of this spread cell (see Luduefla, 1973a). One growth cone (A) has moved off onto the substratum, while another (B) remains near the edge of the underlying cell, though some of its microspikes extend off onto the substratum, x 1725. Fig. 8. A non-neuronal cell (B) in contact with the axon (A) of a ciliary ganglion neuron. Ruffles (arrows) appear at both ends of this cell, and give the impression of wrapping around the axon when the movie is projected, x 540. Fig. 9. A non-neuronal cell (G) with an axon (A) and growth cone (C) of a ciliary ganglion neuron above. In the projected film, the right end of the non-neuronal cell is seen to ruffle actively (arrow) while, immediately above, microspikes and veils form and move about from the growth cone. x 540. Figs. 10 and 11. Sensory ganglia (G) cultured near each other on a polyornithine substratum for 18 h. Many neurites extend outward and appear to bridge the gap between adjacent ganglia. No intervening region of low neurite density is seen. x 125.

660 WESSELLS, LETOURNEAU, NUTTALL, LUDUEI~A and GEIDUSCHEK

pigmented retinal epithelial cells which spread on choroid fibroblasts and are not contact inhibited by them (Parkinson & Edwards, 1978).

A striking feature of our observations is that growth cones apparently adhere to all the cell surfaces they encounter: the sides of neurites, the upper surface of other growth cones, the underside of neurites which are underlapped and the upper surface of non-neuronal cells. Measurements of growth cone-substra tum adhesion suggest that adhesion to the upper surface of non-neuronal cells is stronger than growth cone adhesion to many in vitro substrata, though comparable to adhesion to a polyornithine-treated surface (Letourneau, 1975a). Neurites, too, make adhesions to other neurites; crossover points, which remain stable in spite of lateral displacement of the adjacent free portions of those neurites. Neurites also adhere to the upper surface of non-neuronal cells, where they have the same crooked shapes as neurites on a highly adhesive surface. These instances of adhesivity by neuronal cell surfaces contrast with the behaviour of fibroblasts, which do not adhere to the upper surface of other cells (Harris, 1973; Di Pasquale & Bell, 1974, 1975).

Adhesion to a substratum is required for all tissue cells to be able to migrate on that surface. For neurites, there is always at least one area of adhesion beneath the growth cone and, if the neurite has no other firm adhesive contacts behind the growth cone, it forms a taut line be tween the growth cone and the cell soma, as if it were under tension (Luduefia, 1973b; Letourneau, 1975a). It is believed that the adhesive contacts formed at the leading edge of a growth cone by microspikes and lamellar protrusions (veils; Wessells et al., 1976; Bunge, 1977) from the growth cone act to stabilize these transient extensions of the cell margin and promote the net assembly of precursor subunits into structures of the growing neurite (Bray, 1973a; Letourneau, 1975a, 1979). Thus, one might predict that growth cones will use other cell surfaces as substrata for extension when adhesion to those cell surfaces is sufficient to stabilize the growth cones, microspikes and veils. Of course, this implies that such contact and stabilization should not also evoke contact paralysis or contact inhibition.

Another feature of neurite growth of importance to cell contact behaviour is the proposal that the surface of the neurite (and perhaps the internal cytoskeleton, as well) becomes stable upon its assembly in the growth cone and need not move relative to the substratum or cell soma (Bray, 1970, 1973b). This means that neurite elongation may not require release of posterior adhesions for net elongation to occur (Luduefia & Wessells, 1973). Fibroblasts, on the other hand, must rupture posterior adhesions for net translocation and are slowed down on highly adhesive surfaces (Gail & Boone, 1972). So, a growth cone may touch a neurite or non-neuronal cell, may adhere to that surface as it moves over it or underlaps it, and may continue on, adding new membrane and axonal structures beyond the point of adhesion. More importantly, the growth cone may grow along a previously extended pioneer axon by the same mechanism: as new axon is spun out behind the advancing cone, that

Contacts be tween peripheral ganglionic cells 661

axon need not move, but can persist in a stable, side-to-side adhesion with the pioneer axon. Only the growth cone moves according to this scheme.

Dunn's films of microspike and growth cone retraction following contact with neurites lead to the proposition of 'contact inhibition of extension', analogous to the retraction of cell margins evident during fibroblast contact inhibition (Abercrombie & Dunn, 1975). In contrast, we have observed growth cones moving towards neurites they encounter, and as reported by others (Nakai, 1956; Nakajima, 1965), we saw single microspikes shortening and pulling on neurites to which they had attached. In such cases, the growth cone drew close to the neurite and grew along it in a type of fasciculation. We believe that both Dunn's contact inhibition of extention and our microspike pulling on other neurites involve exertion of the same intracellular (possibly contractile) forces. Under our culture conditions traction is provided to the growth cone by its strong adhesion to the polyornithine-coated substratum and maybe strong adhesion of the extended microspike to the encountered neurite, as well, so that exertion of contractile forces in a microspike does not cause its retraction, but rather draws the neurites together. Thus, the total adhesive environment of the growth cone must be considered in interpreting the outcome of a particular cell contact.

The stimulation of cell surface activity in non-neuronal cells by contact with neuronal surfaces is quite unlike contact inhibition. Instead, contact with a neurite seems to elicit a haptotactic response, such as when a filopodium, extended by a fibroblast, contacts an adhesive planar surface and follows with expansion of the cell margin towards the adhesive site (Albrecht-Buehler, 1976). It is not known whether strong adhesion or other signals keep Schwann cells migrating along neurites. The ultrastructure of these contacts between non-neuronal cells and neurites should be examined for junctions or cytoskeletal specializations and compared to the microfilament bundles found at sites of fibroblast-fibroblast contact (Heaysman & Pegrum, 1973). The Schwann cells and perineural fibroblasts could be separated from each other in the non-neuronal cell population and recornbined with isolated neurons, to analyse more specifically the interacting of neurites with non-neuronal cells.

In conclusion, we find that neurons differ from fibroblasts in major aspects of their reactions to cell contact. Growth cones adhere to many cell surfaces, including the upper surface of non-neuronal cells and subsequently move over these cells. We see no contact paralysis of growth cones to the limit of our microcinematography. If paralysis does occur, it must be extremely local, since a growth cone can protrude its cell margin and elongate beyond any contact site without needing to disrupt it. Neither are non-neuronal cells contact inhibited when they touch growth cones or axons. This suggests that there are distinct and important differences in the cell surface properties of neurons and non-neuronal cells. Whether these differences are in surface molecules, surface-cytoskeleton relationships or other properties of the

662 WESSELLS, L E T O U R N E A U , N U T T A L L , LUDUEI~A and G E I D U S C H E K

locomotory system mus t be def ined by fur ther work. It will also be interest ing to compare the cell surface and cytoskeletal proper t ies of neurons wi th those of mal ignant cells which fail to s h ow typical contact inhibi t ion (S tephenson & S tephenson , 1978).

Acknowledgements

We thank J, T. Wrenn and Belen S. Palmer for aid and criticisms dur ing these studies, and Eric Shooter for the gift of NGF. This work was suppor ted by grants HD 04708 (NKW) and NS 13501 (MLA) f rom the National Insti tutes of Heal th and PCM 77-21035 (PCL) from the National Science Foundat ion.

References ABERCROMBIE, M. (1967) Contact inhibition: The phenomenon and its biological implications.

National Cancer Institute Monograph 26, 249-73. ABERCROMBIE, M. A. & DUNN, G. A. (1975) Adhesions of fibroblasts to substratum during

contact inhibition observed by interference reflection microscopy. Experimental Cell Research 92, 57-62.

ALBRECHT-BUEHLER, G. (1976) Filopodia of spreading 3T3 ceils. Journal of Cell Biology 69, 275-86.

ALBRECHT-BUEHLER, G. & GOLDMAN, R. D. (1976) Microspike mediate particle transport towards the cell body during early spreading of 3T3 cells. Experimental Ceil Research 97, 329-39.

BELL, P. B. (1977) Locomotory behavior, contact inhibition, and pattern formation of 3T3 and polyoma virus-transformed 3T3 ceils in culture. Journal of Cell Biology 74, 963-82.

BOYDE, A., GRAINGER, F. & JAMES, D. W. (1969) S c a n n i n g e lec t ron microscopic o b s e r v a t i o n s of chick embryo fibroblasts in vitro, with particular reference to the movement of cells under others. Zeitschrifl ffir Zellforschung und mikroskopische Anatomie 94, 46-55.

BRAY, D. (1970) Surface movements during the growth of single explanted neurons. Proceedings of the National Academy of Sciences (U.S.A.) 65, 905-10.

BRAY, D. (1973a) Model for membrane movements in the neural growth cone. Nature 244, 93-6.

BRAY, D. (1973b) Branching patterns of individual sympathetic neurons in culture. Journal of Cell Biology 56, 702-12.

BRAY, D. & BUNGE, M. B. (1973) The growth cone in neurite extension. In Ciba Foundation Symposium, Vol. 14, Locomotion of Tissue Cells, pp. 195-209. Amsterdam: Associated Scientific Publishers.

BRAY, D., THOMAS, C. & SHAW, G. (1978) Growth cone formation in cultures of sensory neurons. Proceedings of the National Academy of Sciences (U.S.A.) 75, 5226=9.

BUNGE, M. B. (1977) Initial endocytos!s of peroxidase or ferritin by growth cones of cultured nerve cells. Journal of Neurocytology 6, 407-39.

DI PASQUALE, A. & BELL, P. B., JR (1974) The upper cell surface: its inability to support active cell movement in culture. Journal of Cell Biology 62, 198-214.

DI PASQUALE, A. & BELL, P. B., JR (1975) C o m m e n t s on r e p o r t e d o b s e r v a t i o n s of ceils spreading on the upper surfaces of other cells in culture. Journal of Cell Biology 66, 216-8.

DUNN, G. A. (1971)'Mutual contact inhibition of extension of chick sensory nerve fibers in vitro. Journal of Comparative Neurology 143, 491-508.

Contacts be tween peripheral ganglionic cells 563

E BE N DAL, T. (1976) The relative roles of contact inhibition and contact guidance in orientation of axons extending on aligned collagen fibrils in vitro. Experimental Cell Research 98, 159-69.

EBENDAL, T. (1977) Extracellular matrix fibrils and cell contacts in the chick embryo. Cell and Tissue Research 175, 439-58.

GAIL, M. H. & B O O N E , C. W. (1972) Cell-substrate adhesivity. Experimental Cell Research 70, 33-40.

HARRIS, A. (1973) Behaviour of cultured cells on substrata of variable adhesiveness. Experimental Cell Research 77, 285-97.

HARRIS, A. (1974) Contact inhibition of cell locomotion. In Cell Communication (edited by c o x , R. P.), pp. 147-185. New York: Wiley.

HARRIS, A. & DUNN, G. A. (1972) Centripetal transport of attached particles on both surfaces of moving fibroblasts. Experimental Cell Research 73, 519-22.

H E A Y S M A N, J. E. M. (1978) Contact inhibition of locomotion: a reappraisal. International Review of Cytology 55, 49-66.

H E A Y S M A N , J. E. M. & PEGRUM, S. M. (1973) Early contact between fibroblasts. An ultrastructural study. Experimental Cell Research 78, 71-8.

H E L F A N D , S. L., SMITH, G. A. & WESSELLS, N. K. (1976) Survival and development in culture of dissociated parasympathetic neurons from ciliary ganglia. Developmental Biology 50, 541-7.

JACOBSON, M. (1978)Developmental Neurobiology. 2nd edn. New York: Holt, Rhinehart, Winston.

LETOURNEAU, P. C. (1975a) Possible roles for cell-to-substratum adhesion in neuronal morphogenesis. Developmental Biology 44, 77-91.

LETOURNEAU, P. C. (1975b) Cell-to-substratum adhesion and guidance of axonal elongation. Developmental Biology 44, 92-101.

LETOURNEAU, P. C. (1979) Cell-substratum adhesion of neurite growth cones, and its role in neurite elongation. Experimental Cell Research 124, 127-38.

LOPRESTI , V., M A C A G N O , E. R. & L E V I N T H A L , C. (1973) Structure and development of neural connections in isogenous organisms: Cellular interactions in the development of the optic lamina of Daphnia. Proceedings of the National Academy of Sciences (U. S.A.) 70, 433-7.

LUDUENA, M. A. (1973a) Nerve cell differentiation in vitro. Developmental Biology 33, 268-4. LUDUEI'/qA, M. A. (1973b) The growth of spinal ganglion neurons in serum-free medium.

Developmental Biology 33, 470-6. LUDUEI'~IA, M. A. & WESSELLS, N. K. (1973) Cell locomotion, nerve elongation and

microfilaments. Developmental Biology 30, 427-40. NAKAI, J. (1956) Dissociated dorsal root ganglia in tissue culture. American Journal of Anatomy

99, 81-130. NAKAI, J. (1960) Studies on the mechanism determining the course of nerve fibers in tissue

culture. II. The mechanism of fasciculation. Zeitschrift fidr Zellforschung und mikroskopische Anatomie 52, 427-49.

NAKAI, J. & KAWASAKI, Y. (1959) Studies on the mechanism determining the course of nerve fibers in tissue culture. I. The reactions of the growth cone to various obstructions. Zeitschrift fiir Zellforschung und mikroskopische Anatomie 51, 108-22.

NAKAJIMA, S. (1965) Selectivity in fasciculation of nerve fibers in vitro. Journal of Comparative Neurology 125, 193-204.

P A R K I N S O N , E. K. & EDWARDS, S. G. (1978) Non-reciprocal contact inhibition of locomotion of chick embryonic choroid fibroblasts by pigmented retina epithelial cells. Journal of Cell Science 33, 103-20.

RAKIC, P. (1971) Neuron-glia relationship during granule cell migration in developing cerebellar cortex, a Golgi and EM study in Macacus rhesus. Journal of Comparative Neurology 141, 283-312.

664 W E S S E L L S , L E T O U R N E A U , N U T T A L L , L U D U E I ~ A a n d G E I D U S C H E K

SPEIDEL, C. C. (1933) Studies of living nerves. II. Activities of ameboid growth cones, sheath cells, and myel in segments, as revealed by pro longed observat ion of individual nerve fibers in frog tadpoles. American Journal of Anatomy 52, 1-79.

SPOONER, B. S., LUDUEI~A, M. A. & WESSELLS, N. K. (1974) Membrane fusion in the growth cone-microspike region of embryonic nerve cells undergoing axon elongation in cell culture. Tissue and Cell 6, 399-409.

STEPHENSON, E. M. & STEPHENSON, N. G. (1978) Invasive locomotory behaviour be tween mal ignant human melanoma cells and normal fibroblasts filmed in vitro. Journal of Cell Science 32, 389-412.

TOSNEY, K. W. (1978) The early migrat ion of neural crest cells in the t runk region of the avian embryo: An electron microscopic study. Developmental Biology 62, 317-33.

TRINKAUS, J. P. (1976) O n the mechanism of metazoan cell movements , In The Cell Surface in Animal Embryogenesis and Development (edited by POSTE, G. and NICOLSON, G. L.), pp. 225-329. Amste rdam: North-Hol land.

TRINKAUS, J. P., BETCHAKU, T. & KRULIKOWSKI, L. S. (1971) Local inhibit ion of ruffling dur ing contact inhibit ion of cell movement . Experimental Cell Research 64, 291-300.

WEISS, P. (1941) Nerve Patterns: the Mechanics of Nerve Growth. 3rd Growth Symposium, Vol. 5, pp. 163-203.

WESSELLS, N. K., JOHNSON, S. R. & NUTTALL, R. P. (1978) Axon initiation and growth cone regenerat ion in cul tured motor neurons. Experimental Cell Research 117, 335-45.

WESSELLS, N. K. & NUTTALL, R. P. (1978) Normal branching, induced branching and steering of cul tured parasympathet ic motor neurons in vitro. Experimental Cell Research 115, 111-22.

WESSELLS, N. K., NUTTALL, R. P., WRENN, J. T. & JOHNSON, S. (1976) Differential labeling of the cell surface of single ciliary ganglion neurons in vitro. Proceedings of the National Academy of Sciences (U.S.A.) 73, 4100-4.

WOOD, P, M. (1976) Separat ion of functional Schwann cells and neurons from normal per ipheral nerve tissue. Brain Research 115, 361-75.

YAMADA, K. M., SPOONER, B. S. & WESSELLS, N. K. (1971) Ultrastructure and function of growth cones and axons of cul tured nerve cells. Journal of Cell Biology 49, 614-35.