reviews
A commondenominatorof growth coneguidanceand collapse? Jochen W a l t e r , T i m o t h y E. A l l s o p p and Friedrich B o n h o e f f e r
Axonal guidance in the retinotectal system and in spinal nerve segmentation is based on repulsion or inhibition. In both systems the membrane glycoprotein responsible for the guiding activity is capable of inducing growth cone collapse. We discuss two models of axonal guidance that correlate a.zonal guidance and growth cone collapse. The models are applicable to axon guidance by membrane-associated or diffusible stimuli, and are not based on preferential adhesion of axons to certain substrata. Axonal growth is commonly accompanied by pronounced alternating forward and backward movements of the growth cone lamellipodia (compare Refs 1-3). This presents one major difficulty in studying the mechanisms of axonal guidance, because these movements are usually much more noticeable than the gradual net forward movement of the growth cone and they are more pronounced than the movements caused by directional guidance towards a target. In recent years in vitro systems have been developed in which growth cones encounter an abrupt (spatial and/or temporal) change in the environment leading to strong and immediate reactions. Studies of these systems, which involve exclusively natural cell membranes as a stimulus, give us new insights into the phenomenon of axonal guidance.
Observations of membrane-induced collapse In 1980, Bray et al. showed that growing sympathetic and retinal axons interact and guide each other in vitro 4. When Kapfhammer and Raper investigated the cell dynamics of this interaction, they discovered that under certain circumstances growth cones of axons collapse when they encounter heterotypic neurites5,6. Similar collapse phenomena have since been observed in other systems7,8 and are reviewed elsewhere9. The time-lapse studies of Kapfhammer and Raper showed that a few minutes after the first contact between the filopodia of a growing retinal axon and a sympathetic axon, the retinal growth cone exhibits an increasing density in phase-contrast microscopy, as it rapidly thickens and shortens its filopodia. LameUipodia are resorbed into the neurite, and within 15 min the neurite has retracted by some tens of microns. Almost always the retraction leaves behind a strand of material that is connected to the contacted neurite and able to withstand the observable tension exerted by the rearward movement. These characteristic morphological changes involved in the collapse of the growth cone are illustrated in Fig. 1. Within minutes a new growth cone emerges from the retracted neurite, and again encounters the heterotypic neurite. "Present address: Department of Anatomy, St George's Hospital Medical School, Cranmer Terrace, Tooting, London SW17 ORE, UK. TINS, Vol. 13, No. 11, 1990
Interestingly, several of these attempts can finally lead to a successful crossing. It is an open question whether this occurs because the growth cone is able to inactivate the membrane-associated component causing the collapse or because it becomes desensitized and no longer detects the collapse-inducing cue, or whether the phenomenon can be explained simply on a statistical basis: every so often a growth cone can cross. Since the growth cone is pulled back by the tension of the axon while contacts between the filopodia and the sympathetic axon still remain, the sympathetic axon appears to exert an inhibitory influence on the growth of the retinal axon. With these experiments it was discovered that axons might be guided by inhibitory effects. Three phenomena are associated with this inhibitory effect - a collapse of growth cone morphology, which is presumably related to changes in the structure of the cytoskeleton, a detachment of the central growth cone domain from the substratum, and an inhibition of motility or paralysis. The causal relationship between these phenomena is not clear. Are such inhibitory stimuli and their effects on growth cones part of the mechanisms underlying the choices that growth cones have been observed to make during embryonic development? We will look at two examples of developmental systems, in which growth cones choose the correct pathway, and which have been investigated in vitro l°,n - spinal nerve segmentation and the retinotectal projection. In both these systems in vitro growth cone guidance and growth cone collapse have recently been described 12-16.
]ochen Walter, TimothyE.AIIsopp" and Friedrich Bonhoefferare at the Max-Planck-lnstitut f~r Entwicklungsbiologie, 5pemannstrasse35, D-7400 T6bingen, FRG.
Guidance and collapse in developmental systems Spinal nerve segmentation is known to be determined early in development by division of the paraxial mesoderm into somites either side of the neural tube and the subsequent division of these somites into anterior and posterior halves. The presumptive vertebral column (sclerotome) is formed by a subdivision of the somite. Later, motor axons from the ventral neural tube and sensory axons from the dorsal root ganglion sprout into the sclerotome. They invade exclusively the anterior half-sclerotome. From transplantation experiments it can be concluded that the pattern of outgrowth is exclusively determined by cues within the somites 17. Posterior sclerotome supports arborized outgrowth less well than anterior sclerotome. These findings are consistent with the hypothesis that the posterior half-sclerotome contains components that might be inhibitory for axonal outgrowth, and so have a strong influence in causing neural segmentation. The molecular components that cause the effect have been identified 13. At the time
© 199o,ElsevierSciencePublishersLtd.(UK) 0166-2236/90/$02.00
447
Fig. 1. Collapseof a growth coneinducedby membraneparticles. Experiment pefformedas describedby Raper and Kapfhammeris. A temporalretinal growth coneis shown 2 rain before, then5 and25 min after the addition of posterior tectalmembranes. Arrow marks referencepoint on the substratum. Scalebar is20#m.
-2
5
/ when segmentation is established, posterior sclerotome contains glycosylated proteins that bind the lectin peanut agglutinin (PNA). These PNA-binding proteins, when isolated and coated onto tissue culture plastic, inhibit axonal outgrowth by 30%. In addition, by using an in vitro assay developed by Raper and Kapfhammer ~8, it has now been shown that these inhibitory membrane proteins, when reconstituted into phospholipid vesicles, induce the collapse of cultured dorsal root ganglion growth cones ~3. The retinotectal system, i.e. the projection of retinal ganglion cells to the optic tectum, has been used for studies of axonal guidance in vivo and in vitro because it is a precise and topographic projection in which the target area of retinal axons is well defined 19. Here we consider nasal and temporal axons, which project to the posterior and to the anterior tectal areas, respectively. Studies in vivo have shown that the tectal surface contains directional cues for the ingrowing retinal axons 2°-22. An in vitro system has been developed to detect and characterize the molecules responsible for the observed guidance phenomenon 15. In this system narrow alternating stripes of anterior and posterior tectal membrane components are used as substratum for axonal growth. It has been shown that in this system temporal retinal axons grow exclusively on anterior stripes because they avoid posterior membranes 23, whereas nasal axons grow preferentially on fractionated posterior membranes, as shown recently by von Boxberg and Schwartz 24. The avoidance of posterior membranes by temporal axons has been found to be due to a glycoprotein with an apparent molecular mass of 33 kDa 25. This protein is anchored in the membrane by a phosphatidylinositol anchor 26. A similar activity can be isolated from fish, chick and mice 27,28. The membranes of posterior tectum express this guiding property only during that developmental period when the retinotectal map is
formed 15,2s. The guiding component of posterior tectum, when incorporated into phospholipid vesicles, induces collapse of temporal retinal growth cones 25. Thus, in both the retinotectal system and the spinal cord segmentation system, molecules have been identified that each express two key properties, a guiding activity and a collapse-inducing activity. Some biochemical properties of these molecules are summarized in Table I. Before we describe models that relate guidance and collapse, we will discuss one more interesting experimental observation which concerns the mechanism of action of the 33 kDa guiding molecule of posterior tectal membranes. Temporal axons that are incapable of crossing the border onto posterior stripes are not inhibited with respect to their growth when they grow continuously on substratum consisting solely of posterior tectal membranes (Fig. 2). Thus the striped outgrowth of axons cannot be explained by simply ascribing growth-permissive properties to anterior membranes and growth-inhibitory properties to posterior membranes. If inhibition of axonal growth is involved at all, it must be postulated that, upon continuous exposure, axons can habituate, i.e. they get used to growing on posterior membranes by some unknown mechanism of adaptation zs. Such mechanisms have been involved in chemotaxis of leucocytes29. We will now discuss two models of axonal guidance which give different explanations of the observations that temporal axons are guided away or deflected from posterior membranes and thus do not grow onto posterior stripes, but that they grow on posterior membranes if no other substratum is encountered.
Two models that relate collapse and axonal guidance Let us assume that in the retinotectal system axons are guided by fields of gradients 11. In the chick, the spatial extension of these gradients across the tectal rudiment would TABLE I. Biochemical characterization of guiding proteins be some 100 times larger (several Origin Apparent Peanut GPI-anchor $olubilization buffer millimeters) than the extension of molecular agglutinin growth cones (about 30 I~m). If the mass binding growth cone reads the gradients, Posterior somite 48/55 kDa + ? CHAPS/PBS there must be a mechanism by Posterior optic tectum 33 kDa + + CHAPS/urea/PBS which the growth cone can detect a Abbreviations: GPI, glycosyl phosphatidylinositol; CHAPS, 3-(3-cholamidopropyldimethylammonio)-l- very small concentration difference propanesulphonate; PBS,phosphate-buffered saline. (perhaps of the order of 1%) 448
TINS, VoL 13, No. 11, 1990
between one and the other of its lateral margins, and convert this difference into a guiding force. One conceivable detection and amplification mechanism has been suggested by Gierer n,3°. It is assumed that weak external concentration gradients (1%, Fig. 3) are transduced to the inside of the growth cone, causing small internal concentration differences. This initial gradient is amplified and thus converted to a steep gradient by a process that involves autocatalysis of a locally acting activator and lateral inhibition by a far-reaching inhibitor; the focus of activity thus established within the growth cone is assumed to determine the direction of preferential growth 3°. Discontinuous reactions of the growth cone have been incorporated into the model31 by assuming that, whenever the proximal part of the growth cone becomes activated, indicating that the forward direction is entirely wrong, growth may stop, the direction may change abruptly, or arborization may occur. This model has been analysed mathematically, and has been shown not only to explain guidance in weak external gradients, but also to produce, via computer simulation, axonal trajectories (and terminal arbors) that resemble those found in vivo3L In this model the growth cone would react to small concentration differences of some external guidance molecule, e.g. the 33 kDa component, and it would indeed be a prediction of this model that growth (or growth rate) is not influenced even at high concentrations as long as the component has a homogeneous distribution. The observed collapse of growth cones when they are suddenly confronted with a high concentration of this repulsing guidance molecule was not predicted by the model; however, it is not inconsistent with it. Within the context of the model, collapse may be due to an over-reaction of the amplification because of its high sensitivity to small changes, or it may be a response to the 'forward is wrong' signal, leading to retraction of the growth cone rather than it just stopping if the signal is very strong. These interpretations would be consistent with the observation that growth can proceed on posterior material. Alternatively, the observed collapse phenomenon can be used to design a guidance model that accommodates three experimental observations on temporal retinal axons, namely that (1) posterior material acts as a repellent guiding substance, (2) posterior material induces collapse, and (3) posterior material is not inhibitory to growth. The model has two basic elements - a local paralysis induced by an external inhibitory stimulus and a mechanism for habituation that allows the growth cone to overcome this paralysis. In this model it is assumed that the ability of the axon to grow is controlled by the concentration of some second messenger within the growth cone. Its concentration is influenced by an external guiding molecule: in the stripe assay, for example, contact to posterior membranes at one side of the growth cone is assumed to lead to a local increase of this second messenger and thus to local paralysis of the corresponding part of the growth cone. The unaffected part of the growth cone will continue to advance as normal resulting in the deflection of the growth cone away from the side of contact to posterior membranes. Increase of the concentration of the second messenger all over the growth cone, as for example TINS, Vol. 13, No. 11, 1990
Fig. 2. Temporalretinal axonsgrow equally well on anterior or posterior membranes. Approximately the same number and length of axons is achieved by growth on exclusively(A) anterior or exclusively (B) posterior membranes. However, when both anterior and posterior membranesare available(C), they choose to grow on the anterior membranes. Growth is from left to right in the figure, with lanes labelled as either anterior (A) or posterior (P). Scale baris lO0 #m.
induced in a collapse experiment, will result in general paralysis or general collapse of the growth cone. In order to use local collapse as an explanation for axonal guidance in a gradient field, one has to consider two problems: very small concentration differences between both sides of the growth cone must be sufficient for guidance, and growth must be possible in the presence of a high concentration of collapseinducing activity. One might solve the second problem by invoking an adaptation or 'habituation' mechanism. In contrast to the primary cause of guidance (local paralysis) the habituation must not be local, because local habituation would prevent guidance. Rather, it should be an integrative and general process which regulates the concentration of the second messenger down to a basal level that is optimal for growth. In a homogeneous environment, for example a substratum of pure posterior membranes, the level of second messenger is readjusted to the basal level everywhere within the growth cone. Figure 3 illustrates how a growth cone changes direction in a weak outside gradient of stimuli. A weak outside gradient of a stimulus (Fig. 3D) is converted to an initial inside gradient of the same slope with a second messenger level much above the basal level (Fig. 3E). Since in this gradient, the concentration of second messenger on both sides of the growth cone is approximately the same, it is difficult to imagine how the gradient guides the growth cone. It is assumed that the postulated habituation reduces the level of the 449
exposure
to
a
weak
external
guiding
gradient
• ~°
~
....
D
A
-
~
g
n
a
l
E
,
•
° ° ' "' ,,.
internal signal "• ............
• ........
.
o• ,o
° %1, °
°°~'
,s~•°
•
.
°' •
,. j'...~ ..
; '£o•
• ""-
. ~ °" ' "
, o l t,
•
; .
,4,"°~, I ~~ I , ' ;
•
•
- : . :'.~.~°
•
.°
•°°°•
••
•
°*•
•
basal level
"• ........
"• .......
°°°"°i~ x
X
conversion
of
the
external
signal
to
an
initial
internal
....
B
"
" :'q..
"o
.•*
-'..(;. •
,,
.
.
•
•
•
~ ,
•
°;,,%
.
.% .;. •
.
,,
•
•
..;:..
o
,,,••
°
•• "
• e,•°
oo
.
~*'.~-P~"-'-A.:•
• ,:Ii,, " "s"
E
°•-o
,
•
.
• ,.
signal
t
"*'•
••
:••:C•,~
• •. , .t o ;•II•,~
°o
" . ° . ' ° ° Z "
•-.
•-
•
°
.*
'
•
•
•:;•....:.H.
.o
•
%
• •l,"
•
%
o'~" •
=s~
•
,'°
..o° ...• ".
•
•
° • ° °
oo °'.• t. • •
~ .
,'.."
°~
%
° °•~ • I°°• .,• • • m• i~•,
o,
I. •° .
•
•,'"
•~ . - : . . . . . . . . , : #°•o •
="
°; ,4.
-..
:-
•° • •
.°,
•
° •. °• o, ° o
•
~f~ , . : ~1 • . .•" • •
°•.°
•
•~
°• • ••o° ~
~
-
×
a u t o c a t a l y s i s and lateral inhibition
habituation
• ","
L
• .l.
"~t.""
"
" ......
;oliO,i
.: ".-.i "I~"
°"
:
o~°~°oO ~.. ° •
~~
.-:. ".. ".o•
:
":'.- ~
°*
"
i
•
°•°
. "tQ
: : ~ ~ . . • ~ .. ~ - . . : , . . . . , - .
. ; . . - " r ~ ~ E ~
.,
,. " ~
,
-'..-.~,-
•. .""i ~~; ! " :;':" ' " ""
""'~!-'..i~!~..:" " ' ~ " " ".•"':
."na''n,ern ,s, na, basal
level
/~ x
a u t o c a t a l y s i s
450
model
habituation
model
TINS, Vol. 13, No. 11, 1990
internal signal everywhere by the same amount (Fig. 3I), leading to a gradient that still has the same slope but that reaches from a certain collapse-inducing (or paralysing) activity down to zero. So, while one side of the growth cone is locally collapsed (paralysed), the other side can grow normally. This leads to the axon having a bent trajectory (arrow). One of the major problems with this model is that it requires the reduction of the second messenger level everywhere within the growth cone to be equal and approximately proportional to the total external stimulus and independent of the local concentration of the second messenger. If the second messenger is, for example, an ion, habituation could occur by means of pumps for the second messenger, evenly distributed within the membrane. The habituation level would be reflected in the rate of pumping. An increase in the rate (for example by incorporation of more pumps) should depend on the total amount of second messenger present in the growth cone, but not on its distribution. The model suggests that the levels of habituation in a growth cone vary according to the environment (e.g. growth on pure posterior or pure anterior membranes). This hypothesis could be verified by finding the molecular equivalent of the habituation level. A second prediction of the model is that habituation should be observed in a collapse experiment. Axons should form new growth cones even in the continual presence of the collapse-inducing stimulus. In this model, habituation is a prerequisite for guidance within a gradient field; however, habituation is not required if a growth cone has only to be kept out of a specific area as in the case of posterior somites. It should be emphasized that it is not crucial for the model whether the second messenger primarily influences the cytoskeleton or reduces the adhesiveness of the growth cone surface. Either a local reduction of adhesion or a local interference with the cytoskeleton may lead to local paralysis within the growth cone and thus to guidance. In addition, the model functions whether the external stimulus is membrane-associated or freely diffusible, and irrespective of whether the stimulus is repelling or attracting. Diffusible factors have been shown to guide axons by attraction in a number of different systems 32-34. The two models (autocatalysis with lateral inhibition and collapse with habituation) are not mutually exclusive and have one important feature in common: guidance is not based on preferential adhesion to
different substrata as has been discussed by Letourneau 35. The concepts are rather that external stimuli cause change~ of the cytoskeleton and/or the adhesive properties of the growth cone. In both models the weak outside gradient is converted to an internal concentration gradient. In one model this internal gradient is amplified by the process of autocatalysis and wide-range.lateral inhibition, which leads to an increase in concentration at one edge of the growth cone, and to a decrease at the other edge. In the other model, the internal gradient is not amplified; however, by some kind of regulatory compensation mechanism it is converted into a gradient that has a very low value at one edge of the growth cone. The resultant ratio between concentrations at the two edges of the growth cone becomes rather large - sufficiently large to .act as a guiding force (paralysis at one edge and normal growth at the other). The models resemble each other with respect to two other essential features. They each have one local activity (activator or collapse) and a more widely spread activity (lateral inhibition or habituation). As long as we do not know the second messenger systems involved, it will be difficult to distinguish between the two models. One obvious task for the near future will be to study changes in the intracellular calcium (Ca 2+) concentration during the guidance and collapse phenomena: we know from experiments in recent years that Ca 2+ concentration plays a major role in controlling the motility of the growth cone 8, and that activation of the Ca2+-effector system, involving protein kinase C, can induce the collapse of retinal growth cones 36. We have discussed two reactions of axons growing in vitro - the avoidance reaction in guidance experiments and growth cone collapse. Circumstantial evidence indicates that the observed avoidance reaction is involved in vivo in axonal guidance during the formation of the retinotectal map. However, there is no direct proof, and whether and where growth cone collapse plays a role in vivo is an open question. There is strong evidence, though, that the two in vitro pheomena are related, since they are caused by the same molecule. The two hypotheses described above might explain the relationship between the repulsion of axons and the collapse of growth cones, but may raise more questions than they answer. For instance, does the relationship between growth cone deflection during guidance and growth cone collapse have a counterpart when growth cones are attracted? What could be the analogue to collapse in this situation? It is
Fig. 3. Proposed models that explain the turning of a growth cone as a response to a guiding gradient. The autocatalysis model 29 (AC) proposes that a weak external guiding gradient (A) is converted into an initial weak internal gradient of signal (B). Amplified and steepened by autocatalysis and lateral inhibition the resulting strong final gradient (C) steers the growth cone. The alternative habituation model is exemplified in (D-E) and illustrated in (G-I), assuming a gradient of a repulsive guiding molecule (large dots) ranging from left to right. The concentration of the external signal is proportional to the density of the dots. The density is 10% higher at the left than at the right margin, but otherwise the distribution of the dots is random. The small dots represent the internal signal and their density represents its concentration, indicated in (D), (E) and TINS, Vol. 13, No. 11, 1990
(F). The x-axis represents the lateral extent of the growth cone. Exposure of the growth cone to a weak external guiding gradient (D, G) raises the internal signal from its basal level (D) to a very high value (E). The amplitude of this initial internal signal would be sufficient to paralyse the growth cone. A compensatory mechanism (habituation), introducing a uniform reduction of concentration throughout the growth cone down-regulates the second messenger level The gradient of the final internal signal after habituation (F) is the same as that of the initial signal. The concentration now ranges from a certain paralysing level at the left side of the growth cone to zero at the right side. Therefore the growth cone will turn to the right, and it keeps growing down the gradient in the presence of collapse-inducing material
451
Acknowledgements
hoped that these questions will stimulate ideas for We thank Bernhard new experiments that will further our understanding Mi~llerfor Figs I and2, of axonal guidance mechanisms. and Drs D. Bray, E. Cox, S. E. Fraserand A. Gierer for their helpful comments on the manuscript.
Selected references 1 Bray, D. and Chapman, K. (1985) J. Neurosci. 5, 3204-3213 2 Burmeister, D. W. and Goldberg, D. J. (1988) J. Neurosci. 8, 3151-3159 3 Smith, S. J. (1988) Science 242,708-715 4 Bray, D., Wood, P. and Bunge, R. P. (1980) Exp. Cell Res. 130, 241-250 5 Kapfhammer, J. P., Grunewald, B. E. and Raper, J. A. (1986) J. Neurosci. 6, 2527-2534 6 Kapfhammer, J. P. and Raper, J. A. (1987) J. Neurosci. 7, 201-212 7 Schwab, M. E. and Caroni, P. (1988) J. Neurosci. 8, 2381-2393 8 Kater, S. B., Mat-tson, M., Cohan, C. and Connor, J. (1988) Trends Neurosci. 11, 31 5-321 9 Patterson, P. H. (1988) Neuron 1,263-267 10 Keynes, R. J. and Stern, C. D. (1985) Trends Neurosci. 8, 220-223 11 Bonhoeffer, F. and Gierer, A. (1984) Trends Neurosci. 7, 378-381 12 Stern, C. D., Sisodiya, S. M. and Keynes, R. J. (1986) J. Embryol. Exp. Morphol. 91,209-226 13 Davies, J. A., Cook, G. M, W., Stern, C. D. and Keynes, R. J. (1990) Neuron 4, 11-20 14 Bonhoeffer, F. and Huf, J. (1982) EMBOJ. 4, 427-431 15 Walter, J., Kern-Veits, B., Huf, J., Stolze, B. and Bonhoeffer, F. (1987) Development 101,685-696 16 Cox, E. C., M/~ller, B. and Bonhoeffer, F. (1990) Neuron 4, 31-37
17 Keynes, R. J. and Stern, C. D. (1984) Nature310, 786-789 18 Raper, J. A. and Kapfhammer, J. P. (1990) Neuron4, 21-29 19 Crossland, W. J., Cowan, W. M. and Rogers, L. A. (1975) Brain Res. 91,1-23 20 Holt, C. E. and Harris, W. A. (1983) Nature301, 150-152 21 Stuermer, C. A. O. (1986) J. EmbryoL Exp. MorphoL 93, 1-28 22 Thanos, S., Bonhoeffer, F. and Rutishauser, U. (1984) Proc. Natl Acad. Sci. USA 81, 1906-1910 23 Walter, J., Henke-Fahle, S. and Bonhoeffer, F. (1987) Development 101,909--913 24 Stahl, B. et al. Cold Spring Harbor Symp. Quant. Biol. (in press) 25 Stahl, B., M/iller, B., Boxberg, Y. V., Cox, E. and Bonhoeffer, F. Neuron (in press) 26 Walter, J., M011er, B. and Bonhoeffer, F. (1990) J. Physiol. (Paris) 84, 104-110 27 Vielmetter, J. and Stuermer, C. A. O. (1989) Neuron 2, 1331-1339 28 Godement, P. and Bonhoeffer, F. (1989) Development 106, 313-320 29 Devreotes, P. N. and Zigmond, S. H. (1988) Ann. Rev. Cell BioL 4, 649-686 30 Gierer, A. (1981 ) Biol. Cybern. 42, 69-78 31 Gierer, A. (1987) Development 101,479-489 32 Lumsden, A. G. S. and Davies, A. M. (1983) Nature 306, 786-788 33 Tessier-Lavigne, M., Placzek, M., Lumsden, A. G. S., Dodd, J. and Jessell, T. M. (1988) Nature 336, 775-778 34 Heffner, C. D., Lumsden, A. G. S. and O'Leary, D. D. M. (1990) Science 247, 217-220 35 Letourneau, P. C. (1984) in Molecular Bases of Neuronal Development (Edelman, G. M., Gall, W. E. and Cowan, W. M., eds), pp. 269-293, John Wiley & Sons 36 Allsopp, T. E. CellBiol. (in press)
Myelin-associatedinhibitors of neuritegrowth and regenerationin the CNS Martin E. Schwab Martin E. Schwabis at the Brain Research Institute, University of ZOrich,August-Fore/Strasse 1, CH-8029 Z~irich, Switzerland.
452
Axons often respond to lesions by spontaneous sprouting which, in the PARS, can be followed by elongation over long distances. In contrast, in the CNS, regenerative axon growth in most fibre systems subsides after 0. 5 1.0 ram. The observation that an identical situation can be found in tissue culture in the presence of trophic factors argued for the existence of inhibitory mechanisms within the CNS tissue. Detailed cell biological and biochemical studies have provided evidence for two membrane proteins localized selectively in oligodendrocytes and CNS myelin and which exert a powerful inhibitory effect on neurite growth. Antibodies raised against these neurite growth inhibitors (NI-35 and NI250) and applied to rats with complete transections of the corticospinal tract (CST) resulted in CST axon regeneration over five to ten mm from the lesion site within two to three weeks. Analogous results were obtained in rats lacking myelin and oligodendrocytes in the spinal cord. During development, the 'fuzzy' appearance of the CST grown in the absence of oligodendrocytes or in the presence of anti-inhibitor antibodies indicates a boundary and guidance function of these inhibitors for late growing CNS tracts.
absence in the path of interrupted nerve fibres of catalytic agents capable of attracting and directing the axonic current to its destination'1. This conclusion was drawn by Ram6n y Cajal in his famous book 'Degeneration and Regeneration of the Nervous System'1, based on a large number of observations on the lesioned CNS, and on transplantation experiments which demonstrated the ability of CNS neurones to regenerate neurites into peripheral nerve transplants. These latter findings of Tello z were contradicted by Le Gros Clark in the forties3 and subsequently forgotten, until Aguayo's group confirmed and greatly extended them by showing that the majority of CNS neurones can regenerate long axons through peripheral nerve bridges4-7. Subsequent functionalreconnection to postsynaptic targets was recently shown in the optic system8. As Schwann cells from peripheral nerve produce a variety of neurotrophic factors and even increase this production after denervation9'1°, the hypothesis of a difference in trophic factor production between the peripheral and central nervous systems being responsible for their different regeneration c~:,~i~;i:~ ,' " -: ::reed plausible. However, identified n :,~ ::: ~ ::.:?,,rs like nerve 'The tendency to restoration (of lesioned nerve fibres) growth factor (NGF), br~l-Gciived neurotrophic is frustrated by two negative conditions: (1) the lack factor (BDNF), ciliary neurotrophic factor (CNTF), of substances able to sustain and invigorate the and fibroblast growth factor (FGF) were found to be indolant and scanty growth of the sprouts; (2) the present in the adult CNS H-~3, and increases in © 1990.ElsevierSciencePublishersLtd,(UK) 0166- 2236•90/$02.00
TINS, VoL 13, NO. I 1, 1990
A commondenominatorof growth coneguidanceand collapse? Jochen W a l t e r , T i m o t h y E. A l l s o p p and Friedrich B o n h o e f f e r
Axonal guidance in the retinotectal system and in spinal nerve segmentation is based on repulsion or inhibition. In both systems the membrane glycoprotein responsible for the guiding activity is capable of inducing growth cone collapse. We discuss two models of axonal guidance that correlate a.zonal guidance and growth cone collapse. The models are applicable to axon guidance by membrane-associated or diffusible stimuli, and are not based on preferential adhesion of axons to certain substrata. Axonal growth is commonly accompanied by pronounced alternating forward and backward movements of the growth cone lamellipodia (compare Refs 1-3). This presents one major difficulty in studying the mechanisms of axonal guidance, because these movements are usually much more noticeable than the gradual net forward movement of the growth cone and they are more pronounced than the movements caused by directional guidance towards a target. In recent years in vitro systems have been developed in which growth cones encounter an abrupt (spatial and/or temporal) change in the environment leading to strong and immediate reactions. Studies of these systems, which involve exclusively natural cell membranes as a stimulus, give us new insights into the phenomenon of axonal guidance.
Observations of membrane-induced collapse In 1980, Bray et al. showed that growing sympathetic and retinal axons interact and guide each other in vitro 4. When Kapfhammer and Raper investigated the cell dynamics of this interaction, they discovered that under certain circumstances growth cones of axons collapse when they encounter heterotypic neurites5,6. Similar collapse phenomena have since been observed in other systems7,8 and are reviewed elsewhere9. The time-lapse studies of Kapfhammer and Raper showed that a few minutes after the first contact between the filopodia of a growing retinal axon and a sympathetic axon, the retinal growth cone exhibits an increasing density in phase-contrast microscopy, as it rapidly thickens and shortens its filopodia. LameUipodia are resorbed into the neurite, and within 15 min the neurite has retracted by some tens of microns. Almost always the retraction leaves behind a strand of material that is connected to the contacted neurite and able to withstand the observable tension exerted by the rearward movement. These characteristic morphological changes involved in the collapse of the growth cone are illustrated in Fig. 1. Within minutes a new growth cone emerges from the retracted neurite, and again encounters the heterotypic neurite. "Present address: Department of Anatomy, St George's Hospital Medical School, Cranmer Terrace, Tooting, London SW17 ORE, UK. TINS, Vol. 13, No. 11, 1990
Interestingly, several of these attempts can finally lead to a successful crossing. It is an open question whether this occurs because the growth cone is able to inactivate the membrane-associated component causing the collapse or because it becomes desensitized and no longer detects the collapse-inducing cue, or whether the phenomenon can be explained simply on a statistical basis: every so often a growth cone can cross. Since the growth cone is pulled back by the tension of the axon while contacts between the filopodia and the sympathetic axon still remain, the sympathetic axon appears to exert an inhibitory influence on the growth of the retinal axon. With these experiments it was discovered that axons might be guided by inhibitory effects. Three phenomena are associated with this inhibitory effect - a collapse of growth cone morphology, which is presumably related to changes in the structure of the cytoskeleton, a detachment of the central growth cone domain from the substratum, and an inhibition of motility or paralysis. The causal relationship between these phenomena is not clear. Are such inhibitory stimuli and their effects on growth cones part of the mechanisms underlying the choices that growth cones have been observed to make during embryonic development? We will look at two examples of developmental systems, in which growth cones choose the correct pathway, and which have been investigated in vitro l°,n - spinal nerve segmentation and the retinotectal projection. In both these systems in vitro growth cone guidance and growth cone collapse have recently been described 12-16.
]ochen Walter, TimothyE.AIIsopp" and Friedrich Bonhoefferare at the Max-Planck-lnstitut f~r Entwicklungsbiologie, 5pemannstrasse35, D-7400 T6bingen, FRG.
Guidance and collapse in developmental systems Spinal nerve segmentation is known to be determined early in development by division of the paraxial mesoderm into somites either side of the neural tube and the subsequent division of these somites into anterior and posterior halves. The presumptive vertebral column (sclerotome) is formed by a subdivision of the somite. Later, motor axons from the ventral neural tube and sensory axons from the dorsal root ganglion sprout into the sclerotome. They invade exclusively the anterior half-sclerotome. From transplantation experiments it can be concluded that the pattern of outgrowth is exclusively determined by cues within the somites 17. Posterior sclerotome supports arborized outgrowth less well than anterior sclerotome. These findings are consistent with the hypothesis that the posterior half-sclerotome contains components that might be inhibitory for axonal outgrowth, and so have a strong influence in causing neural segmentation. The molecular components that cause the effect have been identified 13. At the time
© 199o,ElsevierSciencePublishersLtd.(UK) 0166-2236/90/$02.00
447
Fig. 1. Collapseof a growth coneinducedby membraneparticles. Experiment pefformedas describedby Raper and Kapfhammeris. A temporalretinal growth coneis shown 2 rain before, then5 and25 min after the addition of posterior tectalmembranes. Arrow marks referencepoint on the substratum. Scalebar is20#m.
-2
5
/ when segmentation is established, posterior sclerotome contains glycosylated proteins that bind the lectin peanut agglutinin (PNA). These PNA-binding proteins, when isolated and coated onto tissue culture plastic, inhibit axonal outgrowth by 30%. In addition, by using an in vitro assay developed by Raper and Kapfhammer ~8, it has now been shown that these inhibitory membrane proteins, when reconstituted into phospholipid vesicles, induce the collapse of cultured dorsal root ganglion growth cones ~3. The retinotectal system, i.e. the projection of retinal ganglion cells to the optic tectum, has been used for studies of axonal guidance in vivo and in vitro because it is a precise and topographic projection in which the target area of retinal axons is well defined 19. Here we consider nasal and temporal axons, which project to the posterior and to the anterior tectal areas, respectively. Studies in vivo have shown that the tectal surface contains directional cues for the ingrowing retinal axons 2°-22. An in vitro system has been developed to detect and characterize the molecules responsible for the observed guidance phenomenon 15. In this system narrow alternating stripes of anterior and posterior tectal membrane components are used as substratum for axonal growth. It has been shown that in this system temporal retinal axons grow exclusively on anterior stripes because they avoid posterior membranes 23, whereas nasal axons grow preferentially on fractionated posterior membranes, as shown recently by von Boxberg and Schwartz 24. The avoidance of posterior membranes by temporal axons has been found to be due to a glycoprotein with an apparent molecular mass of 33 kDa 25. This protein is anchored in the membrane by a phosphatidylinositol anchor 26. A similar activity can be isolated from fish, chick and mice 27,28. The membranes of posterior tectum express this guiding property only during that developmental period when the retinotectal map is
formed 15,2s. The guiding component of posterior tectum, when incorporated into phospholipid vesicles, induces collapse of temporal retinal growth cones 25. Thus, in both the retinotectal system and the spinal cord segmentation system, molecules have been identified that each express two key properties, a guiding activity and a collapse-inducing activity. Some biochemical properties of these molecules are summarized in Table I. Before we describe models that relate guidance and collapse, we will discuss one more interesting experimental observation which concerns the mechanism of action of the 33 kDa guiding molecule of posterior tectal membranes. Temporal axons that are incapable of crossing the border onto posterior stripes are not inhibited with respect to their growth when they grow continuously on substratum consisting solely of posterior tectal membranes (Fig. 2). Thus the striped outgrowth of axons cannot be explained by simply ascribing growth-permissive properties to anterior membranes and growth-inhibitory properties to posterior membranes. If inhibition of axonal growth is involved at all, it must be postulated that, upon continuous exposure, axons can habituate, i.e. they get used to growing on posterior membranes by some unknown mechanism of adaptation zs. Such mechanisms have been involved in chemotaxis of leucocytes29. We will now discuss two models of axonal guidance which give different explanations of the observations that temporal axons are guided away or deflected from posterior membranes and thus do not grow onto posterior stripes, but that they grow on posterior membranes if no other substratum is encountered.
Two models that relate collapse and axonal guidance Let us assume that in the retinotectal system axons are guided by fields of gradients 11. In the chick, the spatial extension of these gradients across the tectal rudiment would TABLE I. Biochemical characterization of guiding proteins be some 100 times larger (several Origin Apparent Peanut GPI-anchor $olubilization buffer millimeters) than the extension of molecular agglutinin growth cones (about 30 I~m). If the mass binding growth cone reads the gradients, Posterior somite 48/55 kDa + ? CHAPS/PBS there must be a mechanism by Posterior optic tectum 33 kDa + + CHAPS/urea/PBS which the growth cone can detect a Abbreviations: GPI, glycosyl phosphatidylinositol; CHAPS, 3-(3-cholamidopropyldimethylammonio)-l- very small concentration difference propanesulphonate; PBS,phosphate-buffered saline. (perhaps of the order of 1%) 448
TINS, VoL 13, No. 11, 1990
between one and the other of its lateral margins, and convert this difference into a guiding force. One conceivable detection and amplification mechanism has been suggested by Gierer n,3°. It is assumed that weak external concentration gradients (1%, Fig. 3) are transduced to the inside of the growth cone, causing small internal concentration differences. This initial gradient is amplified and thus converted to a steep gradient by a process that involves autocatalysis of a locally acting activator and lateral inhibition by a far-reaching inhibitor; the focus of activity thus established within the growth cone is assumed to determine the direction of preferential growth 3°. Discontinuous reactions of the growth cone have been incorporated into the model31 by assuming that, whenever the proximal part of the growth cone becomes activated, indicating that the forward direction is entirely wrong, growth may stop, the direction may change abruptly, or arborization may occur. This model has been analysed mathematically, and has been shown not only to explain guidance in weak external gradients, but also to produce, via computer simulation, axonal trajectories (and terminal arbors) that resemble those found in vivo3L In this model the growth cone would react to small concentration differences of some external guidance molecule, e.g. the 33 kDa component, and it would indeed be a prediction of this model that growth (or growth rate) is not influenced even at high concentrations as long as the component has a homogeneous distribution. The observed collapse of growth cones when they are suddenly confronted with a high concentration of this repulsing guidance molecule was not predicted by the model; however, it is not inconsistent with it. Within the context of the model, collapse may be due to an over-reaction of the amplification because of its high sensitivity to small changes, or it may be a response to the 'forward is wrong' signal, leading to retraction of the growth cone rather than it just stopping if the signal is very strong. These interpretations would be consistent with the observation that growth can proceed on posterior material. Alternatively, the observed collapse phenomenon can be used to design a guidance model that accommodates three experimental observations on temporal retinal axons, namely that (1) posterior material acts as a repellent guiding substance, (2) posterior material induces collapse, and (3) posterior material is not inhibitory to growth. The model has two basic elements - a local paralysis induced by an external inhibitory stimulus and a mechanism for habituation that allows the growth cone to overcome this paralysis. In this model it is assumed that the ability of the axon to grow is controlled by the concentration of some second messenger within the growth cone. Its concentration is influenced by an external guiding molecule: in the stripe assay, for example, contact to posterior membranes at one side of the growth cone is assumed to lead to a local increase of this second messenger and thus to local paralysis of the corresponding part of the growth cone. The unaffected part of the growth cone will continue to advance as normal resulting in the deflection of the growth cone away from the side of contact to posterior membranes. Increase of the concentration of the second messenger all over the growth cone, as for example TINS, Vol. 13, No. 11, 1990
Fig. 2. Temporalretinal axonsgrow equally well on anterior or posterior membranes. Approximately the same number and length of axons is achieved by growth on exclusively(A) anterior or exclusively (B) posterior membranes. However, when both anterior and posterior membranesare available(C), they choose to grow on the anterior membranes. Growth is from left to right in the figure, with lanes labelled as either anterior (A) or posterior (P). Scale baris lO0 #m.
induced in a collapse experiment, will result in general paralysis or general collapse of the growth cone. In order to use local collapse as an explanation for axonal guidance in a gradient field, one has to consider two problems: very small concentration differences between both sides of the growth cone must be sufficient for guidance, and growth must be possible in the presence of a high concentration of collapseinducing activity. One might solve the second problem by invoking an adaptation or 'habituation' mechanism. In contrast to the primary cause of guidance (local paralysis) the habituation must not be local, because local habituation would prevent guidance. Rather, it should be an integrative and general process which regulates the concentration of the second messenger down to a basal level that is optimal for growth. In a homogeneous environment, for example a substratum of pure posterior membranes, the level of second messenger is readjusted to the basal level everywhere within the growth cone. Figure 3 illustrates how a growth cone changes direction in a weak outside gradient of stimuli. A weak outside gradient of a stimulus (Fig. 3D) is converted to an initial inside gradient of the same slope with a second messenger level much above the basal level (Fig. 3E). Since in this gradient, the concentration of second messenger on both sides of the growth cone is approximately the same, it is difficult to imagine how the gradient guides the growth cone. It is assumed that the postulated habituation reduces the level of the 449
exposure
to
a
weak
external
guiding
gradient
• ~°
~
....
D
A
-
~
g
n
a
l
E
,
•
° ° ' "' ,,.
internal signal "• ............
• ........
.
o• ,o
° %1, °
°°~'
,s~•°
•
.
°' •
,. j'...~ ..
; '£o•
• ""-
. ~ °" ' "
, o l t,
•
; .
,4,"°~, I ~~ I , ' ;
•
•
- : . :'.~.~°
•
.°
•°°°•
••
•
°*•
•
basal level
"• ........
"• .......
°°°"°i~ x
X
conversion
of
the
external
signal
to
an
initial
internal
....
B
"
" :'q..
"o
.•*
-'..(;. •
,,
.
.
•
•
•
~ ,
•
°;,,%
.
.% .;. •
.
,,
•
•
..;:..
o
,,,••
°
•• "
• e,•°
oo
.
~*'.~-P~"-'-A.:•
• ,:Ii,, " "s"
E
°•-o
,
•
.
• ,.
signal
t
"*'•
••
:••:C•,~
• •. , .t o ;•II•,~
°o
" . ° . ' ° ° Z "
•-.
•-
•
°
.*
'
•
•
•:;•....:.H.
.o
•
%
• •l,"
•
%
o'~" •
=s~
•
,'°
..o° ...• ".
•
•
° • ° °
oo °'.• t. • •
~ .
,'.."
°~
%
° °•~ • I°°• .,• • • m• i~•,
o,
I. •° .
•
•,'"
•~ . - : . . . . . . . . , : #°•o •
="
°; ,4.
-..
:-
•° • •
.°,
•
° •. °• o, ° o
•
~f~ , . : ~1 • . .•" • •
°•.°
•
•~
°• • ••o° ~
~
-
×
a u t o c a t a l y s i s and lateral inhibition
habituation
• ","
L
• .l.
"~t.""
"
" ......
;oliO,i
.: ".-.i "I~"
°"
:
o~°~°oO ~.. ° •
~~
.-:. ".. ".o•
:
":'.- ~
°*
"
i
•
°•°
. "tQ
: : ~ ~ . . • ~ .. ~ - . . : , . . . . , - .
. ; . . - " r ~ ~ E ~
.,
,. " ~
,
-'..-.~,-
•. .""i ~~; ! " :;':" ' " ""
""'~!-'..i~!~..:" " ' ~ " " ".•"':
."na''n,ern ,s, na, basal
level
/~ x
a u t o c a t a l y s i s
450
model
habituation
model
TINS, Vol. 13, No. 11, 1990
internal signal everywhere by the same amount (Fig. 3I), leading to a gradient that still has the same slope but that reaches from a certain collapse-inducing (or paralysing) activity down to zero. So, while one side of the growth cone is locally collapsed (paralysed), the other side can grow normally. This leads to the axon having a bent trajectory (arrow). One of the major problems with this model is that it requires the reduction of the second messenger level everywhere within the growth cone to be equal and approximately proportional to the total external stimulus and independent of the local concentration of the second messenger. If the second messenger is, for example, an ion, habituation could occur by means of pumps for the second messenger, evenly distributed within the membrane. The habituation level would be reflected in the rate of pumping. An increase in the rate (for example by incorporation of more pumps) should depend on the total amount of second messenger present in the growth cone, but not on its distribution. The model suggests that the levels of habituation in a growth cone vary according to the environment (e.g. growth on pure posterior or pure anterior membranes). This hypothesis could be verified by finding the molecular equivalent of the habituation level. A second prediction of the model is that habituation should be observed in a collapse experiment. Axons should form new growth cones even in the continual presence of the collapse-inducing stimulus. In this model, habituation is a prerequisite for guidance within a gradient field; however, habituation is not required if a growth cone has only to be kept out of a specific area as in the case of posterior somites. It should be emphasized that it is not crucial for the model whether the second messenger primarily influences the cytoskeleton or reduces the adhesiveness of the growth cone surface. Either a local reduction of adhesion or a local interference with the cytoskeleton may lead to local paralysis within the growth cone and thus to guidance. In addition, the model functions whether the external stimulus is membrane-associated or freely diffusible, and irrespective of whether the stimulus is repelling or attracting. Diffusible factors have been shown to guide axons by attraction in a number of different systems 32-34. The two models (autocatalysis with lateral inhibition and collapse with habituation) are not mutually exclusive and have one important feature in common: guidance is not based on preferential adhesion to
different substrata as has been discussed by Letourneau 35. The concepts are rather that external stimuli cause change~ of the cytoskeleton and/or the adhesive properties of the growth cone. In both models the weak outside gradient is converted to an internal concentration gradient. In one model this internal gradient is amplified by the process of autocatalysis and wide-range.lateral inhibition, which leads to an increase in concentration at one edge of the growth cone, and to a decrease at the other edge. In the other model, the internal gradient is not amplified; however, by some kind of regulatory compensation mechanism it is converted into a gradient that has a very low value at one edge of the growth cone. The resultant ratio between concentrations at the two edges of the growth cone becomes rather large - sufficiently large to .act as a guiding force (paralysis at one edge and normal growth at the other). The models resemble each other with respect to two other essential features. They each have one local activity (activator or collapse) and a more widely spread activity (lateral inhibition or habituation). As long as we do not know the second messenger systems involved, it will be difficult to distinguish between the two models. One obvious task for the near future will be to study changes in the intracellular calcium (Ca 2+) concentration during the guidance and collapse phenomena: we know from experiments in recent years that Ca 2+ concentration plays a major role in controlling the motility of the growth cone 8, and that activation of the Ca2+-effector system, involving protein kinase C, can induce the collapse of retinal growth cones 36. We have discussed two reactions of axons growing in vitro - the avoidance reaction in guidance experiments and growth cone collapse. Circumstantial evidence indicates that the observed avoidance reaction is involved in vivo in axonal guidance during the formation of the retinotectal map. However, there is no direct proof, and whether and where growth cone collapse plays a role in vivo is an open question. There is strong evidence, though, that the two in vitro pheomena are related, since they are caused by the same molecule. The two hypotheses described above might explain the relationship between the repulsion of axons and the collapse of growth cones, but may raise more questions than they answer. For instance, does the relationship between growth cone deflection during guidance and growth cone collapse have a counterpart when growth cones are attracted? What could be the analogue to collapse in this situation? It is
Fig. 3. Proposed models that explain the turning of a growth cone as a response to a guiding gradient. The autocatalysis model 29 (AC) proposes that a weak external guiding gradient (A) is converted into an initial weak internal gradient of signal (B). Amplified and steepened by autocatalysis and lateral inhibition the resulting strong final gradient (C) steers the growth cone. The alternative habituation model is exemplified in (D-E) and illustrated in (G-I), assuming a gradient of a repulsive guiding molecule (large dots) ranging from left to right. The concentration of the external signal is proportional to the density of the dots. The density is 10% higher at the left than at the right margin, but otherwise the distribution of the dots is random. The small dots represent the internal signal and their density represents its concentration, indicated in (D), (E) and TINS, Vol. 13, No. 11, 1990
(F). The x-axis represents the lateral extent of the growth cone. Exposure of the growth cone to a weak external guiding gradient (D, G) raises the internal signal from its basal level (D) to a very high value (E). The amplitude of this initial internal signal would be sufficient to paralyse the growth cone. A compensatory mechanism (habituation), introducing a uniform reduction of concentration throughout the growth cone down-regulates the second messenger level The gradient of the final internal signal after habituation (F) is the same as that of the initial signal. The concentration now ranges from a certain paralysing level at the left side of the growth cone to zero at the right side. Therefore the growth cone will turn to the right, and it keeps growing down the gradient in the presence of collapse-inducing material
451
Acknowledgements
hoped that these questions will stimulate ideas for We thank Bernhard new experiments that will further our understanding Mi~llerfor Figs I and2, of axonal guidance mechanisms. and Drs D. Bray, E. Cox, S. E. Fraserand A. Gierer for their helpful comments on the manuscript.
Selected references 1 Bray, D. and Chapman, K. (1985) J. Neurosci. 5, 3204-3213 2 Burmeister, D. W. and Goldberg, D. J. (1988) J. Neurosci. 8, 3151-3159 3 Smith, S. J. (1988) Science 242,708-715 4 Bray, D., Wood, P. and Bunge, R. P. (1980) Exp. Cell Res. 130, 241-250 5 Kapfhammer, J. P., Grunewald, B. E. and Raper, J. A. (1986) J. Neurosci. 6, 2527-2534 6 Kapfhammer, J. P. and Raper, J. A. (1987) J. Neurosci. 7, 201-212 7 Schwab, M. E. and Caroni, P. (1988) J. Neurosci. 8, 2381-2393 8 Kater, S. B., Mat-tson, M., Cohan, C. and Connor, J. (1988) Trends Neurosci. 11, 31 5-321 9 Patterson, P. H. (1988) Neuron 1,263-267 10 Keynes, R. J. and Stern, C. D. (1985) Trends Neurosci. 8, 220-223 11 Bonhoeffer, F. and Gierer, A. (1984) Trends Neurosci. 7, 378-381 12 Stern, C. D., Sisodiya, S. M. and Keynes, R. J. (1986) J. Embryol. Exp. Morphol. 91,209-226 13 Davies, J. A., Cook, G. M, W., Stern, C. D. and Keynes, R. J. (1990) Neuron 4, 11-20 14 Bonhoeffer, F. and Huf, J. (1982) EMBOJ. 4, 427-431 15 Walter, J., Kern-Veits, B., Huf, J., Stolze, B. and Bonhoeffer, F. (1987) Development 101,685-696 16 Cox, E. C., M/~ller, B. and Bonhoeffer, F. (1990) Neuron 4, 31-37
17 Keynes, R. J. and Stern, C. D. (1984) Nature310, 786-789 18 Raper, J. A. and Kapfhammer, J. P. (1990) Neuron4, 21-29 19 Crossland, W. J., Cowan, W. M. and Rogers, L. A. (1975) Brain Res. 91,1-23 20 Holt, C. E. and Harris, W. A. (1983) Nature301, 150-152 21 Stuermer, C. A. O. (1986) J. EmbryoL Exp. MorphoL 93, 1-28 22 Thanos, S., Bonhoeffer, F. and Rutishauser, U. (1984) Proc. Natl Acad. Sci. USA 81, 1906-1910 23 Walter, J., Henke-Fahle, S. and Bonhoeffer, F. (1987) Development 101,909--913 24 Stahl, B. et al. Cold Spring Harbor Symp. Quant. Biol. (in press) 25 Stahl, B., M/iller, B., Boxberg, Y. V., Cox, E. and Bonhoeffer, F. Neuron (in press) 26 Walter, J., M011er, B. and Bonhoeffer, F. (1990) J. Physiol. (Paris) 84, 104-110 27 Vielmetter, J. and Stuermer, C. A. O. (1989) Neuron 2, 1331-1339 28 Godement, P. and Bonhoeffer, F. (1989) Development 106, 313-320 29 Devreotes, P. N. and Zigmond, S. H. (1988) Ann. Rev. Cell BioL 4, 649-686 30 Gierer, A. (1981 ) Biol. Cybern. 42, 69-78 31 Gierer, A. (1987) Development 101,479-489 32 Lumsden, A. G. S. and Davies, A. M. (1983) Nature 306, 786-788 33 Tessier-Lavigne, M., Placzek, M., Lumsden, A. G. S., Dodd, J. and Jessell, T. M. (1988) Nature 336, 775-778 34 Heffner, C. D., Lumsden, A. G. S. and O'Leary, D. D. M. (1990) Science 247, 217-220 35 Letourneau, P. C. (1984) in Molecular Bases of Neuronal Development (Edelman, G. M., Gall, W. E. and Cowan, W. M., eds), pp. 269-293, John Wiley & Sons 36 Allsopp, T. E. CellBiol. (in press)
Myelin-associatedinhibitors of neuritegrowth and regenerationin the CNS Martin E. Schwab Martin E. Schwabis at the Brain Research Institute, University of ZOrich,August-Fore/Strasse 1, CH-8029 Z~irich, Switzerland.
452
Axons often respond to lesions by spontaneous sprouting which, in the PARS, can be followed by elongation over long distances. In contrast, in the CNS, regenerative axon growth in most fibre systems subsides after 0. 5 1.0 ram. The observation that an identical situation can be found in tissue culture in the presence of trophic factors argued for the existence of inhibitory mechanisms within the CNS tissue. Detailed cell biological and biochemical studies have provided evidence for two membrane proteins localized selectively in oligodendrocytes and CNS myelin and which exert a powerful inhibitory effect on neurite growth. Antibodies raised against these neurite growth inhibitors (NI-35 and NI250) and applied to rats with complete transections of the corticospinal tract (CST) resulted in CST axon regeneration over five to ten mm from the lesion site within two to three weeks. Analogous results were obtained in rats lacking myelin and oligodendrocytes in the spinal cord. During development, the 'fuzzy' appearance of the CST grown in the absence of oligodendrocytes or in the presence of anti-inhibitor antibodies indicates a boundary and guidance function of these inhibitors for late growing CNS tracts.
absence in the path of interrupted nerve fibres of catalytic agents capable of attracting and directing the axonic current to its destination'1. This conclusion was drawn by Ram6n y Cajal in his famous book 'Degeneration and Regeneration of the Nervous System'1, based on a large number of observations on the lesioned CNS, and on transplantation experiments which demonstrated the ability of CNS neurones to regenerate neurites into peripheral nerve transplants. These latter findings of Tello z were contradicted by Le Gros Clark in the forties3 and subsequently forgotten, until Aguayo's group confirmed and greatly extended them by showing that the majority of CNS neurones can regenerate long axons through peripheral nerve bridges4-7. Subsequent functionalreconnection to postsynaptic targets was recently shown in the optic system8. As Schwann cells from peripheral nerve produce a variety of neurotrophic factors and even increase this production after denervation9'1°, the hypothesis of a difference in trophic factor production between the peripheral and central nervous systems being responsible for their different regeneration c~:,~i~;i:~ ,' " -: ::reed plausible. However, identified n :,~ ::: ~ ::.:?,,rs like nerve 'The tendency to restoration (of lesioned nerve fibres) growth factor (NGF), br~l-Gciived neurotrophic is frustrated by two negative conditions: (1) the lack factor (BDNF), ciliary neurotrophic factor (CNTF), of substances able to sustain and invigorate the and fibroblast growth factor (FGF) were found to be indolant and scanty growth of the sprouts; (2) the present in the adult CNS H-~3, and increases in © 1990.ElsevierSciencePublishersLtd,(UK) 0166- 2236•90/$02.00
TINS, VoL 13, NO. I 1, 1990
