ribchester & barry

Experimental Physiology (1994), 79, 465-494Printed in Great Britain

REVIEW ARTICLE

SPATIAL VERSUS CONSUMPTIVE COMPETITIONAT POLYNEURONALLY

INNERVATED NEUROMUSCULAR JUNCTIONS

RICHARD R. RIBCHESTER AND JACQUELINE A. BARRYDepartment of Physiology, University Medical School, Teviot Place, Edinburgh EH8 9AG

(MANUSCRIPT RECEIVED 16 DECEMBER 1993, ACCEPTED 25 FEBRUARY 1994)

CONTENTS PAGEIntroduction 465A definition and some general principles 468Neuromuscular paradigms of neurotrophic competition 471

Innervation of smooth muscle 471Innervation of skeletal muscle 472Indirect evidence for spatial competition in skeletal muscle 473Predictions regarding plasticity of smooth and skeletal muscle innervation 474

Competition versus co-existence of motor nerve terminal boutons 475Role of activity 475Role of selective synapse formation and elimination 478

Molecular basis for neuromuscular synaptic competition 478Diffusible versus fixed molecular resources 478Are there growth inhibitory factors in skeletal muscle? 480

An 'induced-fit' model of neuromuscular synaptic competition 482Concluding discussion 484References 485

INTRODUCTION

A compelling problem in developmental neurobiology during the past few decades has beento resolve how networks of functionally appropriate connections arise during development,and how they are maintained throughout adult life, given that the genetic pre-specificationof every connection would require a degree of encoding that could outstrip an organism'sgenomic capacity. It is more or less taken for granted that the diversity and specificity offunctions in the nervous system arise through some degree of selectivity in the formationand use of neural connections (Purves & Lichtman, 1985), but it has been proposed thatthe full complement of neuronal connections must be determined by epigenetic, trophicdependences of neurones on their target organs (Purves, 1988). According to this' neurotrophic theory', neurones are predisposed to make broadly selective synapticconnections, but the number of neurones innervating a specific target and the number anddiversity of each neurone's connections are regulated by a negative feedback loop: specificnutrients (trophic factors) derived from peripheral targets regulate neuronal survival andgrowth of dendrites, axon collaterals and synaptic terminals; while orthograde patterns ofneural activity limit the production, availability or accessibility of the trophic support (Figs

22-2) at unknown institution on May 27, 2009ep.physoc.orgDownloaded from Exp Physiol (

http://ep.physoc.org/

466 R. R. RIBCHESTER AND J. A. BARRY

A MN

Activity Neurotrophicfactor

Muscle cell/fibre

B C

Smooth muscle Skeletal musclecell fibre

Fig. 1. A, general form of the neurotrophic theory, as it might be applied to the motor innervation of smooth orskeletal muscle. The level of neurotrophic factor produced by the target tissue is up- or down-regulated by thelevel of activity induced in the muscle by its motor nerve supply (MN). Decreased levels of activity give rise toincreased levels of neurotrophic factor, and increased levels of activity reduce the level of neurotrophic support.The level of neurotrophic factor acquired by the neurone influences its survival and the number, disposition andstrength of its connections. Based on Purves (1988). B and C, schematic illustrations of the principal differencesin the distribution of motor innervation to smooth and skeletal muscle, and how this might relate to distributionof neurotrophic resources. In smooth muscle (B), the neurotrophic resource (NGF; stippling of muscle) maybe distributed over the entire area of the target tissue, and the autonomic motor nerve terminals compete forthis on a consumptive basis, leading to a stable steady state in which the resource is partitioned between theboutons of the nerve supply. By contrast, in skeletal muscle (C) the neurotrophic resource may be restricted toa relatively small area, leading to spatial competition between convergent motoneurones and their terminals,emergence of a dominance hierarchy and, ultimately, exclusion of all but one of the convergent terminals.

1 and 2). Thus neurones and their axons and dendrites appear to flourish in proportion tothe capacity of their axon terminals to exploit trophic resources, whose levels they indirectlylimit and control. It is customary to describe this potential interaction between neuronesand their targets in terms of competition for neurotrophic resources.

) at unknown institution on May 27, 2009ep.physoc.orgDownloaded from Exp Physiol (


SPATIAL COMPETITION AT MOTOR ENDPLATES

Fig. 2. Fluorescence micrographs showing the difference in motor innervation of smooth muscle (A and B) andskeletal muscle (C), as visualized with vital staining. A and B, autonomic motor nerve terminals in a preparationof mouse vas deferens stained with the fluorescent carbocyanine dye DiOC2(5). Note the extended, beadeddistribution of fluorescent nerve terminals. C, somatic motor nerve terminals in a reinnervated and paralysedrat lumbrical muscle stained with the fluorescent styryl dye 4-Di-2-Asp. Note the pair of axons converging onthe reinnervated motor endplate on the left (arrow). Calibration bar, 20 /,m. (Panels A and B reproduced withpermission from Lavidis & Bennett, 1992).

Neurotrophic theory has arisen, at least in part, from the notion that the interplaybetween neurones and their synaptic connections during development resembles moreclosely the interactions between organisms in a biological system than the organization ofelements in an electrical circuit (Purves & Lichtman, 1985, p. 363). Competitive, Darwinianprinciples are also the cornerstone of other theories of higher brain function and

467



R. R. RIBCHESTER AND J. A. BARRY

development (Edelman, 1987, 1992). Our aims in this review are therefore threefold.Firstly, we explore the possibility that principles used to describe competition in a differentbiological context - namely the dynamic interactions between members of plant and animalcommunities (Keddy, 1989) - may be applied to the problem of neuronal or synapticcompetition. The intention is to indicate how these principles might be used as basis forfurther experimental investigation of competitive synaptic mechanisms. Secondly, weconsider some potential weaknesses in the way neurotrophic theory is currently applied tothinking about the control of skeletal muscle innervation, particularly in relation to the roleof motoneuronal activity. Thirdly, we consider alternative kinds of molecular mechanismsfor competition at neuromuscular junctions, and we present a speculative model based on'induced fit' between presynaptic and postsynaptic molecules. This model embodies someof the criteria used in population biology to define the nature of competition betweenindividuals vieing for access to limited resources. Finally, we briefly consider the possiblerelevance of our approach to thinking about the control of convergence and divergenceelsewhere in the nervous system during development or nerve regeneration.

A DEFINITION AND SOME GENERAL PRINCIPLES

The essence of the neurotrophic theory is that neurones are supported by substances whichare limited in supply, and this implies that a neurone may profoundly influence the accessof others to the same neurotrophic factors. In the context of skeletal muscle innervation, acogent illustration of this kind of influence is provided by the process of reinnervation afterpartial denervation. Intact motor axons, which first sprout to reinnervate muscle fibresvacated by degenerating terminals, relinquish some of their additional connections onlywhen regenerating axons return (Guth, 1962; Brown & Ironton, 1978; Thompson, 1978;Ribchester, 1988 a). Thus the number of muscle fibres innervated by each regeneratingmotoneurone (motor unit size) gradually increases, at the expense of the intact, sproutedmotor units (Fig. 3). At a cellular level, the competition is played out on muscle fibres whichreceive convergent, functional innervation by regenerating axons and intact or sproutedmotor axons (Fig. 4).

This example, which we shall return to in sections below, is sufficient to permit a conciseworking definition of synaptic competition: the negative effects which one neurone and itsterminals have upon others by consuming, or controlling access to, a resource that islimited in availability.

This definition of competition is almost identical to that proposed by Keddy (1989) todescribe competitive interactions between organisms in plant and animal communities. Asa definition it generates several important questions. First, what is the precise nature of thenegative influences between neurones? Second, what is the precise nature of the resources?Third, is synaptic competition based on consumption of resources or on control of accessto resources? And finally, what other factors limit the availability of trophic resources andhow is this limitation brought about?

'Resources' are defined in population biology as substances or factors which lead toincreased growth rates as their availability in the environment is increased. 'Consumption'and 'control of access' refer to two different modes of competition. In plant and animalcommunities, organisms either consume a fraction of a resource which is distributed overa large available area; or they vie for access to the whole of a resource which is onlyavailable in a restricted area. In the first case, the competition is referred to as consumptivecompetition; in the second case it is referred to as spatial competition. The availability of

468




A 15 days

B 30 days

C 45 days

5 mN 20 mN

50 ms 500 ms

Fig. 3. Competition between motor axons illustrated by the reinnervation of partially denervated adult skeletalmuscle. Isometric twitch tension recordings on the left, tetanic tension recordings on the right, from musclesstudied 15, 30 and 45 days after crushing the major source of innervation. In each set of three traces, the lowesttrace is the muscle tension response to stimulation of the regenerating nerve, the middle trace is the responseto stimulation of the intact, sprouted nerve, and the top trace is combined nerve stimulation. With time, theregenerating axons exert a progressive negative influence on the sprouted axons, and the motor units suppliedby the regenerating nerve increase in size, at the expense of muscle fibres supplied by the sprouted nerve.

resources may be limited by various means: by competitors exploiting resources andremoving them from the local environment (competition by exploitation), or by competitorsdirectly or indirectly interfering with other competitors seeking access to the resource(competition by interference).Consumptive competition and spatial competition tend to produce different kinds of

stable steady state (Keddy, 1989). Consumptive competition commonly leads to a fairlyuniform spatial distribution of competing individuals in relation to the resources on whichthey rely (resource partitioning). This competitive strategy and steady state is illustrated bymobile organisms; for example, in the distribution of whale populations in relation to theplankton resources on which they feed. Spatial competition, by contrast, almost invariablyleads to establishment of hierarchies, in which one or more participants becomes dominantover others (dominance hierarchies). This strategy and steady state are commonlyexperienced by sessile organisms; for example, in the distribution of barnacles adhering toa rock. The establishment of dominance is interesting because it may be thought of in termsof two positive feedback loops (Fig. 5) which contrive further to strengthen the position of

469




A D

B E

Lt F

C . F

SN LPN SN LPN

J1 10 mV

20 ms

Fig. 4. Examples of intracellular recordings from partially denervated rat lumbrical muscles that weresubsequently paralysed by a tetrodotoxin block applied to the sciatic nerve, during regeneration of injured (SN)motor axons. A and F are mononeuronally innervated muscle fibres; B-E are polyneuronally innervated todifferent extents by sprouted (LPN) and regenerated axons. (From Ribchester, 1993.)

dominant competitors at the expense of others (the subordinates). An essential requirementfor dominance - or, in the limit, exclusion of all but one participant - is some initial formof asymmetry between future dominant or subordinate competitors; otherwise thecompetitors co-exist.From the above standpoint, it seems to us that ifwe can describe and explain competitive

synaptic interactions in terms of negative influences; identify the molecular nature of theresources; determine the extent to which competitive interactions are regulated by theconsumption of, or control of access to, resources; and demonstrate how the resources arelimited; then it would be fair to say that we will have achieved a reasonable understandingof synaptic competition. This understanding might then be translated into rational and

470




Mononeuronalinnervation

Exploitation Interference

symmetriccompetition

Polyneuronalinnervation

Fig. 5. Schematic diagram showing how positive feedback may generate the outcome of spatial competition atneuromuscular junctions, whether the mechanism is by exploitation of a trophic resource or interferencebetween convergent nerve terminals on a muscle fibre. Initial asymmetries are amplified, leading to eliminationof polyneuronal innervation and progressive emergence of mononeuronal innervation.

beneficial ways of manipulating neurotrophic resources and competition. For example,recent reports have indicated that it may be feasible to treat certain neuromusculardisorders by manipulating levels of activity or by administering neurotrophic factors whichmay not normally be present in muscle (Sendtner, Kreutzberg & Thoenen, 1990; Gurney,Yamamoto & Kwon, 1992; Sendtner, Holtmann, Kolbeck, Thoenen & Barde, 1992a;Sendtner, Schmalbruch, Stockli, Carroll, Kreutzberg & Thoenen, 1992b).

NEUROMUSCULAR PARADIGMS OF NEUROTROPHIC COMPETITION

Innervation of smooth muscleAn established paradigm for neurotrophic theory is provided by the innervation of

smooth muscle targets by autonomic ganglion neurones, especially those arising insympathetic ganglia. It is well known that the innervation of smooth muscles is adistributed one (e.g. Lavidis & Bennett, 1992; Fig. 2). There is good evidence that smoothmuscle regulates sympathetic ganglion cell numbers and the number and disposition oftheir peripheral (axonal) and intraganglionic (dendritic) branches (Olson & Malmfors,1970; Hume & Purves, 1988; Voyvodic, 1989). This regulation is evidently mediated, atleast in some cases, by a family of structurally related, diffusible growth factors, including'neurotrophins' such as nerve growth factor (NGF), brain-derived neurotrophic factor(BDNF), neurotrophins 3 and 4/5 (NT3, NT4/5), and structurally dissimilar moleculessuch as ciliary neurotrophic factor (CNTF), leukaemia inhibitory factor (LIF) andfibroblast growth factor (FGF; see Levi-Montalcini, 1987; Korsching, 1993, for reviews).The survival and growth of a sympathetic neurone probably depends on the capacity of itsnerve terminals to bind these neurotrophin resources (mediated by specific, tyrosine-kinase-associated neurotrophin receptors), followed by uptake by endocytosis and retrogradetransport (Yankner & Shooter, 1983; Berkemeier, Winslow, Kaplan, Nikolics Goeddel &Rosenthal, 1991; Lindsay, Alderson, Friedman, Hyman, Furth, Maisonpierre, Squinto &Yancopoulos, 1991; Thoenen, 1991; Dechant, Biffo, Okazawa, Kolbeck, Pottgeisser &Barde, 1993a; Dechant, Rodriguez-Tebar, Kolbeck & Barde, 1993b; Rodriguez-Tebar,Dechant, Gotz & Barde, 1993). Fractions of the total amount of neurotrophin resource

471




appear to be partitioned between and consumed by the terminal boutons of the autonomicnerve supply.The smooth muscle paradigm provides a compelling illustration of neurotrophic theory

in its most conventional form (Fig. 1A and B); and it seems to represent a form aconsumptive competition based on exploitation of an essential resource. Levels ofneurotrophin molecules appear to be regulated by negative feedback: the production of theneurotrophic molecules is down-regulated in the target muscle and consumption is up-regulated by the autonomic innervation. This regulation appears to be activity dependent,although the evidence for this remains indirect (Shelton & Reichardt, 1984; Heumann,Korsching, Scott & Thoenen, 1984; Heumann & Thoenen, 1986; Snider, 1988).Innervation of skeletal muscleA quite different form of neurotrophic interaction and different competitive pressures

appear to prevail at skeletal neuromuscular junctions, where interactions involving morethan one motoneurone occur within a much more restricted area of the muscle fibre surface(Figs 1 C and 2 C). The pattern of innervation of skeletal muscle, and the manner in whichit normally emerges, are what one might expect if spatial competition rather thanconsumptive competition were driving the interaction between motor nerve terminals andmuscle fibres.The events leading to the mature pattern of innervation ofmotor endplates are quite well

documented (reviewed by Jansen & Fladby, 1990). Motoneurones grow selectively intospecific skeletal muscles in early development (Lance-Jones & Landmesser, 1980;Landmesser, 1984; Tanaka & Landmesser, 1986). Normally about half these motoneuronessubsequently die (Lance-Jones, 1982; Oppenheim, 1991). Motoneurone death is largely aprenatal phenomenon (Hardman & Brown, 1985; Oppenheim, 1986), and neurotrophingrowth factors produced by muscle or in peripheral nerve have recently been implicated init (Oppenheim, Qin-Wei, Prevette & Yan, 1992; Yan, Elliott & Snider, 1992; Sendtneret al. 1992 a). Motoneurones which survive cell death provide a transient, focal polyneuronalinnervation of muscle fibres (Boeke, 1925; Redfern, 1970; Bennett & Pettigrew, 1974).Significantly, the terminals converge on a common synaptic site, the motor endplate, whichconstitutes about 0 1 % or less of the total area of exposed muscle fibre membrane (Bekoff& Betz, 1977; Bixby, 1981). The surviving neurones typically withdraw between one-halfand nine-tenths of their initial complement of peripheral branches during postnataldevelopment, a process commonly referred to as synapse elimination (Brown, Jansen &Van Essen, 1976; Korneliusen & Jansen, 1976; Riley, 1977; Betz, Caldwell & Ribchester,1979; Fladby, 1987). Recent data strongly suggest that the loss of functional input isprotracted and occurs by stepwise elimination of individual terminal boutons, rather thanby a sudden disconnection of axon collaterals or retraction of the entire terminal. Thesurviving terminal progressively adds boutons to the endplate over the same period (Balice-Gordon, Chua, Nelson & Lichtman, 1993 a; Balice-Gordon, Garriga & Lichtman, 1993 b).

Synapse elimination during development ultimately yields a familiar, stereotyped patternof adult muscle fibre innervation (at least in mammalian skeletal muscle and many of thosein birds, reptiles and amphibia) of a single motor nerve terminal attached to each musclefibre. Similar processes occur during reinnervation of completely or partially denervatedadult skeletal muscle: like immature muscle fibres, reinnervated muscle fibres in adultmuscles experience a transient polyneuronal innervation, which in most cases issubsequently eliminated (Boeke, 1916; McArdle, 1975; Gorio, Carmignoto, Finesso,Polato & Nunzi, 1983; Taxt, 1983 a, b). From the viewpoint of either development or

472




reinnervation, the key (related) questions are: why do terminals of different motoneuronesnot normally co-exist at a motor endplate and how is it that only one terminal normallypersists ?

Indirect evidence for spatial competition in skeletal muscleEvidence that neuromuscular synapse elimination results from some form of spatial

competition can be deduced from the kinds of experiment which population biologistswould call 'removal' experiments and 'additive' experiments. A removal experimentinvolves depleting an existing population and then searching for evidence of competitive'release' in the remainder; a result of alleviation of pressure on resources. In the case ofmuscle, a removal experiment may be instigated by performing a partial denervation. In anadditive experiment, the ratio of putative colnpetitors to the level of resource is artificiallyincreased, and the investigator searches for a resulting decrease in some index ofperformance of the competitors: the predicted result of increased pressure on resources.Additive experiments can be implemented by performing an implant, forcing additionalaxons to innervate a muscle.Motor unit size is preserved or may even continue to decline in neonatal muscle after

partial denervation at birth: there is no evidence for an expansion of the motor units(Brown et al. 1976; Thompson & Jansen, 1977; Betz, Caldwell & Ribchester, 1980a;Fladby & Jansen, 1987). In adults, however, the response to partial denervation is quitedifferent. Rampant sprouting and occupancy of denervated motor endplates by intactaxons occurs (Brown, Holland & Hopkins, 1981a). In rodents, the transition of theresponse occurs around 3 weeks after birth, when most of the polyneuronal innervation hasbeen eliminated (Brown, Hopkins & Keynes, 1982).When excess motor axons are implanted into the endplate region of an adult muscle,

elimination of some - actually rather few- of the intact motor terminals occurs (Bixby &Van Essen, 1979). An alternative approach, successfully carried out in frogs, has been toforce motor axons from ipsilateral and contralateral sides of the spinal cord to grow intoonly one hindlimb (Lamb, 1980). This also results in hyperinnervation of muscles, by twiceas many motoneurones as normal. Individual motor units are evidently smaller in thesemuscles, as predicted from an increase in pressure on resources, but the interpretation iscomplicated because the number of muscle fibres forming during development depends onthe number of motor axons innervating the muscle, and nerve implants induce formationof more muscle fibres than normal (Betz et al. 1980 a; Ross, Duxson & Harris, 1987; Sheard& Lamb, 1991).The reinnervation of partially denervated muscle may be regarded as a combination of

a removal and an additive experiment. When regenerating axons return to a partiallydenervated muscle, they recover innervation of fewer fibres than they innervated initially,evidently because their endplates become occupied by sprouts from intact axons, asdescribed above. It appears that these opportunistic connections are not readily winkledout by the regenerating axons (Fig. 3), and the longer the interim between nerve injury andregeneration, and the fewer the number of regenerating axons, the less competitive theyappear to become (Brown & Ironton, 1978; Thompson, 1978; Luff & Torkko, 1990;Ribchester, 1993). The success of the regenerating axons in reinnervating partiallydenervated muscle is dramatically improved by simply cutting the intact axons immediatelybefore the regenerating ones return, causing the intact terminals and their sprouts todegenerate (Ribchester & Taxt, 1984). Although substantial improvement in the amount ofsynapse formation by the regenerating axons can also be obtained merely by paralysing the

473




muscle during reinnervation - rendering the muscle functionally denervated - the increaseis not as great as when sprouts from intact axons are improved by complete, structuraldenervation (Ribchester & Taxt, 1984; Ribchester, 1993).The absence of sprouting in neonatal muscles following partial denervation and the

difference between the effects of paralysis and surgical denervation on sprouting in adultmuscles are findings that are compatible with a mechanism of spatial competition atendplates based on interference (the presence of either a functional or non-functionalterminal preventing or inhibiting synapse formation by an ingrowing nerve) but not readilyexplained by a mechanism based on consumptive exploitation of a trophic resource whoselevels are controlled solely by activity. These findings together suggest that the gradualelimination of polyneuronal innervation during normal development or regeneration maybe represented as the emergence of a dominance hierarchy (Fig. 5), based on some form ofinitial asymmetry (not necessarily a size or geometric asymmetry) in the capacity of theconvergent axons to respond to the resources concentrated at the developing motorendplate and leading, via positive feedback, to the exclusion of all the subordinate motorterminals.The proposition that spatial competition regulates the focal innervation of skeletal

muscle fibres and that consumptive competition regulates the distributed innervation ofsmooth muscle is compromised by a few possible exceptions, however. For example, theextensive axon collateral and motor nerve terminal sprouting which occurs followingpartial denervation or paralysis of adult muscle (Betz, Caldwell & Ribchester 1980b;Holland & Brown, 1980; Brown, Holland & Hopkins, 1981 b; Brown, Hopkins & Keynes,1982), the distributed innervation of tonic muscle fibres in some muscles (Gordon, Perry,Tufferey & Vrbova, 1974; Lichtman, Wilkinson & Rich, 1985), and the distributed motorinnervation of intrafusal muscle fibres in muscle spindles (Barker, 1974; Banks, 1981) allprovide a superficial resemblance to the distributed pattern of innervation of smoothmuscle. A further complication is that supernumerary motor nerve terminals in skeletalmuscle are not always eliminated. For instance, following treatment of neonatal male ratswith testosterone, muscle fibres in the bulbocavernosus and levator ani muscles acquire apolyneuronal innervation which persists into adulthood (Jordan, Letinsky & Arnold, 1989;Lubischer, Jordan & Arnold, 1992). Also, following reinnervation of partially denervatedmuscles in normal adults, a small proportion of fibres appear to acquire a stablepolyneuronal innervation (Ribchester, 1988 a; Werle & Herrera, 1991; Barry & Ribchester,1994). These exceptions suggest that perhaps different factors regulate the growth andbranching of axons in skeletal muscle, independently of the factors which motivate thecompetition between convergent terminals at neuromuscular junctions. Overall, however,the differences in the patterns of innervation of smooth and skeletal muscle suggest to usthat, normally, consumptive competition is the principal basis for the control ofinnervation in the case of smooth muscle, whereas convergent terminals in skeletal musclemainly experience a form of spatial competition.

Predictions regarding plasticity of smooth and skeletal muscle innervationOur proposals yield some immediate predictions about the scope and extent of on-going,

systematic and enduring changes in the structure of neuromuscular connections (i.e. theirplasticity) in smooth and skeletal muscle. For example, if the innervation of smooth muscleis regulated by a form of consumptive competition, then, like mobile organisms, their axonterminals might be expected to be continually shifting in their distribution over the surfacesof smooth muscle cells, in response to local changes in the concentration of neurotrophic

474




resources. By contrast, if the mononeuronal innervation of skeletal muscle arises by spatialcompetition, then, like sessile organisms, we might expect the motor nerve terminals whichexclusively innervate single extrafusal muscle fibres to be structurally rather inert. Fromsimilar reasoning we would predict that the motor innervation of muscle spindles and tonicextrafusal muscle fibres should be more labile than that of extrafusal twitch muscle fibres,and that during nerve sprouting after partial denervation there should be continuousremodelling of nerve terminals, with making and breaking of neuromuscular contacts ason-going processes.

There is some supporting evidence for these proposals. Repeated visualization of nerveterminals in mature skeletal muscle in vivo indeed shows that once the basic architecture ofa nerve terminal is established it remains quite invariant in form (Lichtman, Magrassi &Purves, 1987; Balice-Gordon & Lichtman, 1990, 1993). There is also evidence, albeitindirect, in support of the prediction that terminals belonging to sprouted axons should bemore labile, based on the distribution of sizes of innervated and denervated muscle fibresin partially denervated muscles (Ridge & Rowlerson, 1990). Either direct or indirectevidence in support of on-going remodelling of terminals in muscle spindles or in smoothmuscle is lacking, however. Interestingly, repeated observation of dendrites and cell bodiesof the sympathetic neurones themselves, which also receive a distributed innervation,suggest that their synaptic inputs are quite dynamic in form (Purves, Voyvodic, Magrassi& Yawo, 1987). It would seem to be worthwhile to apply similar repeated-visualizationtechniques to study the dynamic properties of motor terminals in smooth muscle.

COMPETITION VERSUS CO-EXISTENCE OF MOTOR NERVE TERMINAL BOUTONS

We now consider in more detail some of the difficulties in applying neurotrophic theoryin its conventional form to the problem of competitive elimination at neuromuscularjunctions in skeletal muscle. In particular, we address the proposed importance of neuralactivity in determining the rate and outcome of elimination, and the possible importanceof selective, recognition mechanisms that might influence the competition. The former is alinchpin of neurotrophic theory; the latter is normally excluded from it.

After synapse elimination is completed in early postnatal development, each survivingmotor nerve terminal is seen typically to arborize into a number of twigs and boutons. AsPurves (1988) has pointed out, this actually poses an interesting paradox. How is it thatindividual boutons which make up the single surviving presynaptic terminal are allowed toco-exist, while similar boutons belonging to different axons are eliminated? (Co-existenceof organisms and species undergoing spatial competition for a common resource is also anintriguing problem to population biologists; Keddy, 1989.) Put another way, whatproperty of a synapse might normally protect intraterminal synaptic boutons fromcompetitive elimination, while simultaneously promoting interterminal competition?

Role of activityActivity and the processes which hinge upon it are natural candidates for explaining the

synaptic bouton paradox. When an action potential occurs in a motor axon collateral, itspreads almost simultaneously into all the boutons of the nerve terminal, producingsynchronous and uniform exocytosis, release of transmitter and recycling of synapticvesicles (Betz, Mao & Bewick, 1992; Ribchester, Mao & Betz, 1994). It is possible toimagine a number of mechanisms by which terminal boutons that are synchronously activebehave as equals in competition, while boutons that are not simultaneously active will not

475




A B

100 100

80 80

-~60 600

40 40

20 20

0 0OGroup 1 2 3 4 5 Group 1 2 3 4 5

LPN------- Dual-- SN LPN ------- Dual--- SN

Fig. 6. Nerve dominance histograms showing the proportion of mononeuronally innervated fibres (Groups 1 and5) and polyneuronally innervated (Groups 2-4) muscle fibre after reinnervation of partially denervated ratlumbrical muscles, with (A) or without (B) paralysis during nerve regeneration. The SN was crushed and thedistribution of innervation was studied about 30 days later. Levels of both polyneuronal innervation andmononeuronal innervation by the regenerated axons are increased by paralysis during nerve regeneration,indicating an activity-independent element to synaptic competition. (From Ribchester, 1993.)

have equivalent status with respect to access to neurotrophic resources (for example, seePurves, 1988, p. 157). Such proposals imply that if simultaneous activation of boutons weresomehow to inhibit competition, then terminals belonging to different, convergent axonsthat are compelled to express identical patterns of activity would exert no mutual negativeinfluence, and would therefore co-exist rather than compete. In other words, a predictionis that if individual muscle fibres are innervated by convergent axons expressing identicalactivity, they should remain polyneuronally innervated. The available experimentalevidence does not entirely support this prediction.

It is generally agreed that paralysis inhibits synapse elimination; thus when the terminalsof the same and different axons are synchronously inactive they show a tendency to co-exist(Benoit & Changeux, 1975; Thompson, Kuffler & Jansen, 1979; Brown et al. 1982;Duxson, 1982; Caldwell & Ridge, 1983; Taxt, 1983b; Callaway & Van Essen, 1989). Thisfinding is broadly in accord with the standard form of neurotrophic theory (Fig. 1).However, recent experiments indicate that some synapse elimination can occur in theabsence of propagated neural activity. Competition between completely inactive axons wasstudied by blocking activity in the entire motor nerve supply during reinnervation ofpartially denervated muscle (Ribchester, 1993). As expected, this procedure increased boththe overall amount of synapse formation and the amount of polyneuronal innervation inthe muscles. But the number of mononeuronally innervated fibres, exclusively innervated by

476




regenerating, inactive axons was also increased (Fig. 6). Thus it is apparently possible forsome inactive terminals to be displaced during competition with other inactive terminals.The results of synchronizing neuromuscular activity using chronic electrical stimulation

also pose some problems. Chronic electrical stimulation of immature muscles, contrary tothe above predictions, does not cause convergent terminals on polyneuronally innervatedmuscle fibres to co-exist; rather, synapse elimination is accelerated (O'Brien, Ostberg &Vrbovai, 1978; Thompson, 1983). It could be argued that chronic muscle stimulation merelysuperimposes some synchronous activity on the natural, asynchronous activation of motorunits, and the net effect is to reduce the amount of trophic resource available; thus, inaccordance with neurotrophic theory, increasing the competitive pressure and henceincreasing the rate of synapse elimination. The outcome of the competition could still bedetermined by the endogenous asynchronous activity of the convergent axons. Thisexplanation cannot be refuted with available data. It could be tested, however, perhaps byelectrically stimulating motor nerve terminals and muscle fibres directly, while remotelyblocking the conduction of natural activity in the peripheral nerves which supply themuscles.The question of a physiological role for differences in the activity in determining the

outcome of neuromuscular synaptic competition has been specifically examined in anumber of studies. The first studies involved selectively blocking activity in motor axonsinnervating polyneuronally innervated muscle fibres in reinnervated adult muscles, andthese studies suggested that active terminals are preferentially stabilized at the expense ofinactive ones (Ribchester & Taxt, 1983, 1984). However, the effects of blocking activity insome axons but not others in neonatal animals have been interpreted differently;apparently the blocked motor axons acquire larger motor units (Callaway, Soha & VanEssen, 1987; but see Ribchester, 1988 b).

Experiments with selective stimulation of motor units in neonatal muscle have also beencarried out. When some axons but not others are electrically stimulated, the more activeterminals appear to have a competitive advantage and their motor units remain expanded,at the expense of the less active axons (Ridge & Betz, 1984). An acute, heterosynapticdepression has also been reported during stimulation oftwo axons convergently innervatingsingle muscle fibres (Betz, Chua & Ridge, 1989). Selective stimulation of motoneuronesmaking convergent synapses on myotubes in culture has also been reported to stabilize thestimulated synapses and repress the non-stimulated connections (Magchielse & Meeter,1986; Lo & Poo, 1991; Dan & Poo, 1992). However, it has also been reported that whilestimulation of neuromuscular preparations in tissue culture accelerates synapse elimination,there is no effect of selective versus non-selective stimulation on the terminals which survive(Nelson et al. 1993). As yet, there are no reports of the effects of selective stimulation onelimination of polyneuronal innervation in reinnervated adult muscle.One alternative to presuming a central role for activity is the idea that competitive

success might be based on the capacity of the motoneurone cell body to support only alimited number of terminals, and that trophic resources gathered by one collateral orbouton are redistributed or shared among others (e.g. Smallheiser & Crain, 1984).However, such ideas are not very compelling. For example, inactive regenerating axons areadditionally disadvantaged in competition with intact axons, even when these havesprouted to their maximum extent (Ribchester, 1988a). Also motoneurones are clearlyquite capable of supporting more terminals than they actually do, whether competingterminals are present or not, as shown by the effects of muscle paralysis in delaying orinhibiting elimination of polyneuronal innervation in both neonates and reinnervated adultmuscle (Thompson et al. 1979; Taxt, 1983b; Barry & Ribchester, 1994).

477




Role of selective synapse formation and eliminationAn alternative or additional explanation of the one-to-one relationship between a motor

nerve terminal and a skeletal muscle fibre is that this arises through some form of selectiverecognition process, occurring either during synapse formation or synapse elimination. Itis very clear that skeletal muscle is not a homogeneous tissue from the viewpoint ofdeveloping or regenerating motoneurones. For instance, motoneurones are evidentlycapable of making crude distinction between muscles on the basis of their position orsegmental origin (Bennett & Lavidis, 1984; Wigston & Sanes, 1985; Donahue & English,1987, 1989; Hardman & Brown, 1987; Balice-Gordon & Thompson, 1988; Bennett & Ho,1988; Laskowski & Sanes, 1988; Laskowski & High, 1989). Other data suggest thatmatching of motor terminals to muscle fibres of a pre-specified, complementary type isimportant in competition, both during postnatal development and following reinnervationin adults. This is due partly to selectivity during synapse formation, and partly to selectivesynapse elimination, mismatched terminals being preferentially withdrawn in competitionwith functionally matched motoneurone and muscle fibre types (Thompson, Sutton &Riley, 1984; Gordon & Van Essen, 1985; Jones, Ridge & Rowlerson, 1987 a, b; Lichtman& Wilkinson, 1987; Soha, Yo & Van Essen, 1987; Bennett, Davies & Everett, 1989; Fladby& Jansen, 1990; Gates & Ridge, 1992).

In sum, to date most experiments suggest that active motor terminals have an advantageover inactive ones, but differences in propagated activity in motor axons alone appear tobe insufficient to ensure persistence of specific terminals at polyneuronally innervatedjunctions. Thus explanations other than activity appear to be required to account for bothcompetition between terminals and co-existence of boutons belonging to the same motornerve terminal. Evidence that some form of selective recognition process plays a role insynaptic competition is now quite compelling. A possible molecular basis for activity-dependent adjustment of selectivity of neuromuscular connections is presented in thepenultimate section of this review.

MOLECULAR BASIS FOR NEUROMUSCULAR SYNAPTIC COMPETITION

Our conclusion thus far is that criteria used to categorize competition in an ecologicalcontext may have utility in a neurobiological context, at least in relation to understandingthe control of muscle innervation. Notions of 'spatial competition' or 'consumptivecompetition' appear to provide a useful analogy, because they may potentially influence theway we might formulate more detailed hypotheses about the cellular and molecularmechanisms underlying competition in smooth and skeletal muscle.

Diffusible versus fixed molecular resourcesAccording to Keddy (1989), the existence of competition in ecosystems is best

demonstrated by showing a dependence of the interacting participants on the level ofessential resources. In skeletal muscle, unlike smooth muscle, a critical problem is that wedo not yet have any clear idea of the molecular identity of the trophic resources for whichmotoneurones and their terminals appear to compete. The search for products of skeletalmuscle which function as specific motoneurone growth factors continues unabated(Dohrmann, Edgar & Thoenen, 1987; Henderson, 1988; Oppenheim, Haverkamp,Prevette, McManaman & Appel, 1988; Arakawa, Sendtner & Thoenen, 1990; Henderson,Camu, Mettling, Gouin et al. 1993). Interestingly, it has recently been shown that

478




expression of mRNA encoding the neurotrophin BDNF increases in skeletal musclefollowing denervation (Funakoshi, Frisen, Timmusk, Zachrisson, Verge & Persson, 1993;Koliatsos, Clatterbuck, Winslow, Cayouette & Price, 1993). BDNF and NT3 have alsobeen shown to increase the efficacy of synapses in tissue culture (Lohof, Ip & Poo, 1993).The mechanism of the synaptic strengthening is unclear. One possibility is that retrogradelytransported second messengers are involved (Harish & Poo, 1992). Another is that sinceneurotrophic factors are normally taken up by receptor-mediated endocytosis, neuro-trophins might stimulate activity-dependent synaptic vesicle exocytosis and recycling.Such effects could provide a cellular basis for the kind of positive feedback required toproduce exclusive innervation of a motor endplate. Techniques have recently beendescribed for quantitative study and analysis of the control of exocytosis, endocytosis andsynaptic vesicle recycling at neuromuscular junctions and synaptic boutons (Betz &Bewick, 1993; Ryan, Reuter, Wendland, Schweizer, Tsien & Smith, 1993; Ribchester et al.1994), so it should be possible to test this idea.To date, however, while there is suggestive evidence that neurotrophins may play a role

in promoting motoneurone survival and possibly in the control of nerve branching, thereis no clear indication that neurotrophins have anything to do with competitive eliminationof terminals at endplates. Indeed, if the mode of competition at polyneuronally innervatedneuromuscular junctions is predominantly one of spatial competition- for the reasonsoutlined in sections above - it would seem inherently more likely that competitive synapseelimination should involve motor nerve terminals vieing for access to fixed molecules intheir local environment.What sorts of molecule might fixed neurotrophic resources comprise, and where might

they be located? The neural cell adhesion molecule (N-CAM) is a plausible candidate.Tight adhesions form between motoneurone growth cones and muscle fibre membraneswithin a few minutes of contact (Buchanan, Sun & Poo, 1989). Neural cell adhesionmolecule has been implicated in the control of nerve growth and branching in muscle, bothduring development and following partial denervation in adults (Booth, Kemplay &Brown, 1990; Landmesser, Dahm, Tang & Rutishauser, 1990; Tang & Landmesser, 1993),and the distribution of N-CAM changes in skeletal muscle after denervation, from arestricted distribution to the endplate region in innervated muscle to a more diffusedistribution in denervated muscle (Sanes, Schachner & Covault, 1986). Activity alsoregulates the expression of N-CAM (Couvault & Sanes, 1986). There are presently thoughtto be about 200 molecular forms of N-CAM, based on alternative splicing and/or variableglycosylation. Some of these isoforms promote and others inhibit neurite growth (Small &Akeson, 1990; Saffell, Walsh & Doherty, 1991; Walsh, Furness, Moore, Ashton &Doherty, 1992; Zorn & Krieg, 1992). Different isoforms of N-CAM are expressed inneonatal muscles, compared with mature or ageing muscles (Andersson, Olsen,Zhernosekov, Gaardsvoll, Krog, Linnemann & Bock, 1993).Another possibility is that acetylcholine receptors (AChR) or molecules tightly associated

with them may provide a substrate for spatial competition, and there is some evidence insupport of this. It is well known that junctional AChR become locked into the endplateregion during early postnatal development (Slater, 1982). Forms of AChR differing insubunit composition or metabolic half-life are selectively expressed at endplates underdifferent conditions (e.g. Witzemann, Bremner & Sakmann, 1991; Andreose, Xu, L0mo,Salpeter & Fumagalli, 1993). Data from J. W. Lichtman's laboratory support a hypothesis,originally proposed by Stent (1973), that local activation of junctional ACh receptors feedsinto an intracellular mechanism which eliminates inactive receptors (Lichtman & Balice-

479




Gordon, 1990). Repeated visualization of nerve terminals in living animals, either duringpostnatal development or during nerve regeneration in adults, appears to show thatacetylcholine receptors in the postsynaptic membrane are eliminated prior to retraction ofdisadvantaged terminals (Rich & Lichtman, 1989; Balice-Gordon & Lichtman, 1991,1993). Remaining terminal boutons evidently fail to occupy the synaptic space vacated bythe retreating terminals. Interestingly, when a patch of receptors at an intact endplate isinactivated by focal binding of a-bungarotoxin, the patch of receptors is removed, theoverlaying terminal boutons are withdrawn and the remaining boutons fail to occupy thevacated postsynaptic space (Balice-Gordon & Lichtman, 1991). Complementary approachesalso indicate that junctional ACh receptors are inherently or potentially more labilein their spatial distribution in mature muscle than presumed hitherto. Junctional receptorsspread from motor endplates in muscle fibres stripped of their basal lamina and cultured(Lupa & Caldwell, 1991). Interestingly, the junctional AChR reaggregate at sites wherenewly synthesized receptors and myonuclei are co-localized (Anderson & Ribchester,1993). Junctional ACh receptors also become redistributed beneath nerve terminal sproutsin muscles poisoned with botulinum toxin (Balice-Gordon et al. 1993b).

However, to date there are no experiments which definitively show that either N-CAMor AChR are the molecules which subserve spatial competition at endplates. We know thatgenes expressed by myonuclei localized to motor endplates are regulated in a different wayfrom those expressed by extra-junctional myonuclei (Merlie & Sanes, 1985; Sanes,Johnson, Katzbauer, Mudd, Hanley & Martinou, 1991; Witzemann et al. 1991), and, notsurprisingly, a number of molecules, in addition to N-CAM or acetylcholine receptors, areknown to be selectively expressed at endplates. For example, numerous sodium channelisoforms are expressed in muscle fibre membranes and these are concentrated atneuromuscular junctions (Caldwell & Milton, 1988; Schaller, Krzemien, McKenna &Caldwell, 1992; Lupa, Krzemien, Schaller & Caldwell, 1993). Acetylcholinesterase(Massoulie & Bon, 1983), cytotactin (Sanes et al. 1986), s-laminin (Hunter, Shah, Merlie &Sanes, 1989) and acetylgalactosaminyl transferase activity (Scott, Balsamomo, Sanes &Lilien, 1990) are also localized to the neuromuscular junction. It would be interesting toknow whether any of these components is selectively up-regulated or down-regulatedduring synapse elimination, or whether experimentally induced manipulation of thesepotential resources influences the rate or the outcome of elimination.

Are there growth inhibitory factors in skeletal muscle?Neurotrophic theory is normally discussed in terms of competition between nerve

endings for growth-promoting factors. Yet inhibition of growth is arguably as importanta part of synapse formation and stabilization (Watson, 1976; Patterson, 1988). Perhaps thecompetition is so intense because the perijunctional or extra-junctional membrane orextracellular matrix provide more than just a non-permissive environment for growth.The capacity of nerve terminals to form synapses ectopically, outside the normal

endplate region is germane to this issue. Ectopic synapses will form when a foreign nerveis implanted into skeletal muscle, but in adults this only occurs when the muscle isdenervated (Frank, Jansen, L0mo & Westgaard, 1975), and in neonatal muscle ectopicsynapse formation normally occurs only when the nerve implant is positioned more thanabout 1 mm from the native endplates. Thus distance between endplates seems to be aregulated variable (Brown et al. 1976; Kuffler, Thompson & Jansen, 1977, 1980; Grinnell,Letinsky & Rheuben, 1979; Haimann, Mallart, Ferre & Zilber-Gachelin, 1981; Werle &Herrera, 1988). Paralysis reduces the allowable distance between endplates in such muscles

480




(Jansen, Lomo, Nicolaysen & Westgaard, 1973; Srihari & Vrbovai 1978; Ding, Jansen,Laing & Tonnesen, 1983). It is interesting that ectopic synapses do not readily form indenervated adult fast-twitch muscle; instead the implanted axons grow along the musclesurface and synapse preferentially on the motor endplates originally innervated by thenative nerve (Taxt, 1983 a).

There is no direct evidence that endogenous neurite-growth inhibitors have anything todo with these findings, indeed they do not allow us to distinguish this interpretation fromone based on the effects of a passive, non-permissive environment. However, taking theseobservations together with data from other systems brings into sharper focus the possibilitythat active, growth inhibitory mechanisms operate in muscle.

First, neurite growth inhibitors are expressed on axons and on neuroglia in the centralnervous system, and such molecules may be responsible for the failure of nerve regenerationin the CNS (Caroni & Schwab, 1988a, b; Schnell & Schwab, 1990). A specific role forastrocytes in providing a 'stop growth' or inhibitory signal in the CNS has been proposed(Luizzi & Lasek, 1987). Neuroglial cells, specifically terminal Schwann cells, also overliemotor nerve terminals at all neuromuscular junctions, and there is evidence that whenSchwann cells are added to neurones and muscle fibres in co-culture this either inhibitssynapse formation or promotes synapse elimination (Chapron & Koenig, 1989).

Secondly, during early development, growing sensory and motor axons are selectivelysteered into the anterior half of the somites (Keynes & Stern, 1984). This appears to be dueto the production of a specific, peanut-lectin binding molecule by cells in the posterior halfof the somite (Davies, Cook, Stern & Keynes, 1990). Peanut-lectin binding activity is alsofound at vertebrate motor endplates (Ko, 1987), but it is not known whether this hasneurite growth inhibiting activity.

Finally, there is some evidence that a balance between the activity of proteases secretedby growth cones and protease inhibitors in the extracellular environment can determinewhether neurites will extend or withdraw (Monard, 1988; Fawcett & Housden, 1990). Agroup of hypotheses for an explicit role for proteases in synapse elimination have beenproposed, based on the reported effects of ions and protease inhibitors on the time courseof synapse elimination (O'Brien et al. 1978; O'Brien, Ostberg & Vrbova, 1984; Connold,Evers & Vrbova, 1986; Vrbova, Lowrie & Evers, 1988; Zhu & Vrbovai, 1992; Navarette &Vrbova, 1993).The possibility that the peri-junctional region of the motor endplate might constitute

some sort of 'terminal exclusion zone', constraining terminals to the endplate andintensifying competition, could be further explored experimentally by determining whetherectopic synapses will fail to form in this region even when the native nerve supply isremoved by denervation. It would also be interesting to know whether growth conecollapse (Kapfhammer & Raper, 1987), a useful indicator of growth inhibiting activity,occurs when a foreign axon is presented ectopically to an innervated muscle, compared withan ectopic presentation to a previously denervated muscle. It would additionally beinteresting to know whether Schwann cells play a role in regulating the activity of proteasesand protease inhibitors, like that proposed for astrocytes and growing axons in the CNS.Repeated observations of identified terminals (Lichtman et al. 1987) or of selectivelylabelled Schwann cells (e.g. Mirsky & Jessen, 1984) in muscles treated with proteases orprotease inhibitors could provide a useful approach here.

481




AN 'INDUCED-FIT MODEL OF NEUROMUSCULAR SYNAPTIC COMPETITION

While most experiments are consistent with a mechanism of competitive synapseelimination based on control of access by terminals for a spatially restricted resource, anumber of unanswered questions remain about the qualitative nature of this mechanism.For example, how can neuromuscular synapse elimination occur under some circumstancesin the absence of neural activity? Why do no muscle fibres become completely denervatedduring the period of synapse elimination? Why, after partial denervation, do regeneratingaxons not recover innervation of many of the muscle fibres they once innervated? How isit that some observations suggest that particular terminals are advantaged through selectivesynapse formation, while other data suggest that innervation is non-selective butelimination is selective?

In an attempt to reconcile observations such as these, we should like to propose amechanism based on induction of selective adhesion as a plausible alternative to otherhypotheses. By analogy with enzyme kinetic studies, and recent insights into antigen-antibody interaction, we will describe our hypothesis as an 'induced-fit' model.

There are four principal elements to the hypothesis. The first is that motor nerveterminals may induce a selective change in the nature of adhesion molecules in the musclefibres they contact at motor endplates. The goodness-of-fit for nerve terminals provided bythese molecules could be determined by the expression of unique isoforms, generated byalternative splicing and/or differential glycosylation, or by a conformational changeinduced by the binding of a specific ligand on the nerve terminal to a complementaryadhesion molecule expressed on the surface of the muscle fibre (Fig. 7). The second elementof the hypothesis is that competing axons may be allowed to occupy synaptic space becausethere is partial fit between dissimilar isoforms or conformations of the adhesion moleculesinitially expressed at an endplate. In other words, weak ('noisy') fits would be allowedbetween different, competing terminals and a muscle fibre. The third element of thehypothesis is that with time, a conformational change in a. particular isoform is induced tobecome more permanent. This would lead to preferential adhesion by one of the terminalsinnervating an endplate and a progressive loss of adhesion and withdrawal of theremainder. This process would represent a form of positive reinforcement of a terminal,mediated by the muscle fibre, based on an initial asymmetry in the strengths of adhesionsof competitors, as required of a mechanism based on spatial competition. The fourthelement of the hypothesis allows for activity in competing terminals to accelerate theconformational change or synthesis of the appropriate complementary isoform, therebyproviding the competitive edge that is normally observed when active and inactiveterminals are pitted against one another.

There are precedents for some of the elements of the induced-fit hypothesis in othersystems. First, it has long been recognized that the goodness-of-fit between the reactive siteof an enzyme and its substrate can be altered by a conformational change in the enzyme.In fact the term 'induced fit' was first introduced in the context of enzyme-substratereaction kinetics (Koshland, Nemeth & Filmer, 1966; Koshland, 1973). Second, it hasrecently been shown that antigens can change their conformation when antibodies bind tothem, in a way that facilitates the antigen-antibody binding (Rini, Schulze-Gahmen &Wilson, 1992). Third, studies on T-lymphocytes show that binding of these to theintercellular adhesion molecule ICAM- 1 is enhanced by an induced fit between theadhesion molecule and the cell surface ligand LFA- 1, and this is facilitated by anintracellular signalling process activated by binding of different antigens to other T-cellreceptors (Cabanas & Hogg, 1993).

482




Normal Denervation

innervation Reinnervation Sprouting

Induced fit

I N/X

Elimination

Co-existenceDominance

NYAN N'0i

Fig. 7. An 'induced fit' model of synaptic competition, as applied to the competitive reinnervation of a partiallydenervated adult skeletal muscle. The initial state is indicated in the upper left diagram. Complementaryadhesion molecules ensure a tight bond between the motor nerve terminal and the muscle fibre. Afterdenervation of the fibre, the endplate may become reinnervated by sprouts bearing a different, characteristicisoform of the adhesion molecule. The initially weak interactions between these adhesion molecules and thosealready present may result in a conformational change, which is signalled to the endplate nuclei and translatedinto the expression of a more appropriate isoform in the muscle fibre membrane (upper right). If regeneratingaxons return soon enough, the transition in expression of isoforms may be incomplete and the terminals co-exist(lower right). Disuse of the muscle might maintain the state of non-specific expression of adhesion moleculeisoforms in the muscle membrane. If reinnervation is delayed, however, the adhesion molecule isoformexpressed may be completely transformed to that appropriate for the sprouted nerve.

How might an induced-fit mechanism work in relation to, say, reinnervation of partiallydenervated muscle? When an adult muscle is partially denervated, reactive changes leadingto motor nerve sprouting occur, and denervated endplates become occupied by sprouts.Functional contact could be allowed because the denervated endplates permit a 'noisy fit',perhaps as a result of change in the isoform of adhesion molecule expressed or theconformation of the existing isoforms. With time, the sprouted terminals would transformthe expression of the complementary adhesion molecules by an activity-dependentinduction of gene expression in the endplate nuclei. This could be mediated by intracellularsecond messengers. It is known, for example, that second messengers play a crucial role indetermining which major isoform of acetylcholine receptor is expressed at an endplate, andthis is mediated in part by neural activity (reviewed by Laufer & Changeux, 1989).Regenerating axons returning to the muscle may reinnervate their old sites, but theirgoodness-of-fit may depend on how long the sprouts had had to respecify the fit on at themotor endplate, and on how active the sprouted terminal had been. Thus in a paralysedmuscle more 'noisy fits' by regenerating axons would be allowed compared with an activemuscle, and multiple terminals would persist for longer. But the longer the interim between

483




nerve injury and the return of the regenerating axons, the more likely it is that the natureof the adhesion molecules would be transformed and the less the likelihood that a noisy fitby regenerating axons would be allowed.The induced-fit hypothesis seems to us to provide a plausible basis for explaining why

terminals do not normally co-exist at an endplate and why one and only one terminalultimately persists in an essentially invariant form. Testing the hypothesis is feasible,although difficult, mainly because it is not clear which molecular isoforms would be the bestcandidates for mediating the induced fit. However, as a start, it would be interesting toknow whether muscles normally express alternatively spliced isoforms of N-CAM orAChR in proportion to the number of motor units they contain, and whether partialdenervation at birth (which permanently reduces the numbers of motor units) reduces thenumbers of isoforms expressed.

CONCLUDING DISCUSSION

Neurotrophic theory in its conventional form (e.g. Purves, 1988) has been highly successfulin providing a framework for thinking about the control of innervation of targets wherethese receive a distributed innervation, as in the case of smooth muscle, where NGF andrelated neurotrophins have been identified as important neurotrophic resources. However,the mode of competition presumed to occur in smooth muscle is perhaps too restrictive toaccount for stabilization and maintenance of connections which are initially focused inmore restricted areas of cell surfaces.

In our view, the terminology used by Keddy (1989) to describe competition in anecological context helps us to focus on the options to be considered for understandingneuromuscular synaptic competition. Specifically, the terminology and conceptualframework which he (and we) recommends help us to view the competitive interactionsbetween motor nerve terminals from a number of different standpoints, and placecontemporary approaches to the study of skeletal muscle innervation into perspective. Thisis a necessary exercise if the challenge of integrating molecular, cellular and organismicphysiology is to be met (Boyd & Noble, 1993). For example, adopting the definition ofcompetition that we propose not only underscores the importance of identifying the natureof the molecular resources for which motoneurones compete; it also indicates that successin this endeavour will not provide all the answers to the problem of synaptic competition.Nonetheless it would be helpful to know whether nerve terminals consume neurotrophicmolecules, or whether the competition is based on control of access to specific isoforms ofadhesion molecules that are spatially restricted in the region of the endplate. We need toknow whether there is more than one type of chemical factor driving the interactionsbetween terminals - for example, whether there are both endogenous growth-promotingand endogenous growth-inhibiting molecules in muscle - and to what extent other celltypes (like Schwann cells, for example) might influence the outcome of competition. Also,it still remains important to establish how and to what extent the use and disuse (activity)of nerve terminals or muscle fibres are able to limit the availability or access toneurotrophic resources and to regulate the number and disposition of terminal boutons ata motor endplate; and to establish the role of activity-dependent competitive remodellingof neuromuscular junctions in the overall function of motor units in the control ofmovement.

Finally, although these ideas about consumptive and spatial competition may be appliedto the problem of synaptic competition in the peripheral nervous system, it is important to

484




ask whether they are also likely to be of value in understanding the control of convergenceand divergence in the central nervous system. In contrast to skeletal muscle, each neuronein the CNS normally receives many different kinds of synaptic input, and there isburgeoning evidence that a membrane protein not expressed in skeletal muscle, theN-methyl-D-aspartate (NMDA) receptor, plays a critical role in many forms of synapticplasticity in the CNS (Daw, Stein & Kox, 1993; Malenka & Nicoll, 1993). Activity-dependent, competitive elimination of convergent, homologous inputs is common duringcritical periods of development in the CNS, for instance in visual cortex, auditory nuclei,and in the cerebellum (Jackson & Parks, 1982; Shatz, 1990; Rabbachi, Bailly, Delhaye-Bouchaud & Mariani, 1992). In adults, even profound lesions to the CNS may bring aboutonly a minor partial denervation of individual target cells, because of the plethora ofsynaptic inputs to any given neurone by local interneurones. For example, complete spinaltransection brings about only a minor partial denervation of lower spinal motoneurones,because the large number of synapses provided by local interneurones and peripheralsensory afferents are unaffected by the lesion. It seems probable that at least some of thesesynapses undergo continuous remodelling (cf. Purves et al. 1987), and sprouting of axonsin the injured spinal cord is now quite well documented (Schnell & Schwab, 1993; Schwab,1993). Some form of competition will surely occur when (if?) regenerating axons are ableto reform connections on neurones partially denervated by a spinal lesion. So there maywell be closer parallels than realized hitherto between, say, the competitive reinnervationof partially denervated skeletal muscle and the processes required to re-establishfunctionally appropriate innervation of partially denervated cells in the injured CNS. In thepast, the neuromuscular junction has admirably served neurobiologists seeking explanationsfor a number of general neurobiological phenomena (such as the mechanism of chemicalsynaptic transmission; Katz, 1969; Eccles, 1990). Perhaps a clearer understanding ofthe principles and mechanisms of synaptic competition at neuromuscular junctions willtherefore make a significant contribution towards explaining how the numbers anddisposition of connections elsewhere in the nervous system are regulated and maintained.

We are extremely grateful to Drs William Betz, Jan Jansen, David Price, Dale Purves, Antony Ridgeand David Willshaw for helpful comments on earlier versions of the manuscript, and to Stephen Wattfor useful discussions. Our research is supported by grants from Action Research, Sports AidingResearch for Children (SPARKS), and from The Wellcome Trust.

REFERENCES

ANDERSON, A. J. & RIBCHESTER, R. R. (1993). Redistribution ofjunctional acetylcholine receptors ondissociated mouse skeletal muscle fibres in culture. Journal of Physiology 459, 49P.

ANDERSSON, A. M., OLSEN, M., ZHERNOSEKOV, D., GAARDSVOLL, H., KROG, L., LINNEMANN, D. &BOCK, E. (1993). Age-related changes in expression of neural cell adhesion molecule in skeletalmuscle: a comparative study of newborn, adult and aged rats. Biochemical Journal 290, 641-648.

ANDREOSE, J. S., Xu, R., LoMo, T., SALPETER, M. M. & FUMAGALLI, G. (1993). Degradation of2 AChR populations at rat neuromuscular junctions: regulation in vivo by electrical stimulation.Journal of Neuroscience 13, 3433-3438.

ARAKAWA, Y., SENDTNER, M. & THOENEN, H. (1990). Survival effect of ciliary neurotrophic factor(CNTF) on chick embryonic motoneurones in culture: comparison with other neurotrophic factorsand cytokines. Journal of Neuroscience 10, 3507-3515.

BALICE-GORDON, R., CHUA, C. K., NELSON, C. A. & LICHTMAN, J. W. (1993a). Gradual loss ofsynaptic cartels precedes axon withdrawal at developing neuromuscular junctions. Neuron 11,801-815.

485




BALICE-GORDON, R., GARRIGA, C. & LICHTMAN, J. W. (1993b). Local control of ACh receptorstability and location within neuromuscular junctions. Societyfor Neuroscience Abstracts 19, 526.2.

BALICE-GORDON, R. & LICHTMAN, J. W. (1990). In vivo visualization of the growth of pre- andpost-synaptic elements of neuromuscular junctions in the mouse. Journal of Neuroscience 10,894-908.

BALICE-GORDON, R. & LICHTMAN, J. W. (1991). Motor nerve terminals are selectively eliminated fromregions of neuromuscular junctions focally blocked with a-bungarotoxin. Societyfor NeuroscienceAbstracts 17, 514.3.

BALICE-GORDON, R. & LICHTMAN, J. W. (1993). In vivo observations of pre- and postsynapticchanges during the transition from multiple to single innervation at developing neuromuscularjunctions. Journal of Neuroscience 13, 834-855.

BALICE-GORDON, R. & THOMPSON, W. (1988). The organization and development of compartmentalinnervation in rat extensor digitorum longus muscle. Journal of Physiology 398, 211-231.

BANKS, R. W. (1981). A histological study of the motor innervation of the cat's muscle spindle.Journal of Anatomy 133, 572-591.

BARKER, D. (1974). The morphology ofmuscle receptors. In Handbook ofSensory Physiology, MuscleReceptors, ed. HUNT, C. C., pp. 1-190. Springer-Verlag, Berlin.

BARRY, J. A. & RIBCHESTER, R. R. (1994). Effects of recovery from nerve conduction block onelimination of polyneuronal innervation in partially denervated and reinnervated rat muscle.Journal of Physiology 476.P, 62P.

BARTHELS, D., VOOPER, G., BONED, A., CREMER, H. & WILLE, W. (1992). High degree of NCAMdiversity generated by alternative RNA splicing in brain and muscle. European Journal ofNeuroscience 4, 327-337.

BEKOFF, A. & BETZ, W. J. (1977). Properties of isolated adult rat muscle fibres maintained in tissueculture. Journal of Physiology 271, 537-547.

BENNETT, M. & LAVIDIS, N. (1986). Topographic projections of segmental nerves to the frogglutaeus muscle during loss of polyneuronal innervation. Journal of Physiology 375, 303-326.

BENNETT, M. R., DAVIES, A. M. & EVERETT, A. W. (1989). The development of topographical mapsand fibre types in toad (Bufo marinus) glutaeus muscle during synapse elimination. Journal ofPhysiology 409, 43-61.

BENNETT, M. R. & Ho, S. (1988). The formation of topographical maps in developing ratgastrocnemius muscle during synapse elimination. Journal of Physiology 396, 471-496.

BENNETT, M. R. & LAVIDIS, N. (1984). Development of the topographical projection of motorneurons to a rat muscle accompanying loss of polyneuronal innervation. Journal of Neuroscience4, 2204-2212.

BENNETT, M. R. & PETTIGREW, A. G. (1974). The formation of synapses in striated muscle duringdevelopment. Journal of Physiology 241, 515-545.

BENOIT, P. & CHANGEUX, J.-P. (1975). Consequences of tenotomy on the evolution of multi-innervation in developing rat soleus muscle. Brain Research 99, 354-358.

BERKEMEIER, L. R., WINSLOW, J. W., KAPLAN, D. R., NIKOLICS, K., GOEDDEL, D. V. & ROSENTHAL,A. (1991). Neurotrophin-5 -a novel neurotrophic factor that activates trk and trkb. Neuron 7,857-866.

BETZ, W. J. & BEWICK, G. (1993). Optical monitoring of transmitter release and synaptic vesiclerecycling at the frog neuromuscular junction. Journal of Physiology 460, 287-309.

BETZ, W. J., CALDWELL, J. H. & RIBCHESTER, R. R. (1979). The size of motor units during post-nataldevelopment of rat lumbrical muscle. Journal of Physiology 297, 463-478.

BETZ, W. J., CALDWELL, J. H. & RIBCHESTER, R. R. (1980 a). The effects of partial denervation at birthon the development of muscle fibres and motor units in rat lumbrical muscle. Journal ofPhysiology303, 265-279.

BETZ, W. J., CALDWELL, J. H. & RIBCHESTER, R. R. (1980b). Sprouting of active nerve terminals inpartially inactive muscles of the rat. Journal of Physiology 303, 265-279.

BETZ, W. J., CHUA, M. & RIDGE, R. M. A. P. (1989). Inhibitory interactions between motoneuroneterminals in neonatal rat lumbrical muscle. Journal of Physiology 417, 25-51.

BETZ, W. J., MAO, F. & BEWICK, G. (1992). Activity-dependent fluorescent staining and destainingof living vertebrate motor nerve terminals. Journal of Neuroscience 12, 363-375.

BIXBY, J. L. (1981). Ultrastructural observations on synapse elimination in neonatal rabbit skeletalmuscle. Journal of Neurocytology 10, 81-100.

486




BIXBY, J. L. & VAN ESSEN, D. C. (1979). Competition between foreign and original nerves in adultmammalian skeletal muscle. Nature 282, 726-728.

BOEKE, J. (1916). Studien zur Nervenregeneration. I. Die Regeneration der motorischen Nerven-elemente und die Regeneration der Nerven der Muskelspindeln. Verhandlingen der KoninklijkeAkademie van Wetenschaffen (Amsterdam) 18, 1-120.

BOEKE, J. (1921). The innervation of striped muscle fibres and Langley's receptive substance. Brain44, 1-22.

BOOTH, C. M., KEMPLAY, S. K. & BROWN, M. C. (1990). An antibody to neural cell adhesionmolecule impairs motor nerve terminal sprouting in a mouse muscle locally paralyzed withbotulinum toxin. Neuroscience 35, 85-91.

BOYD, C. A. R. & NOBLE, D. (1993). The Logic of Life: The Challenge of Integrative Physiology.Oxford University Press, Oxford.

BROWN, M. C., HOLLAND, R. L. & HOPKINS, W. G. (1981 a). Motor nerve sprouting. Annual Reviewof Neuroscience 4, 17-42.

BROWN, M. C., HOLLAND, R. L. & HOPKINS, W. G. (1981 b). Restoration of focal multipleinnervation in rat muscles by transmission block during a critical stage of development. Journal ofPhysiology 318, 355-364.

BROWN, M. C., HOPKINS, W. G. & KEYNES, R. J. (1982). Comparison of the effects of denervationand botulinum toxin paralysis on muscle properties in mice. Journal of Physiology 327, 29-37.

BROWN, M. C. & IRONTON, R. (1978). Sprouting and regression of neuromuscular synapses inpartially denervated mammalian muscles. Journal of Physiology 278, 325-348.

BROWN, M. C., JANSEN, J. K. S. & VAN ESSEN, D. C. (1976). Polyneuronal innervation of skeletalmuscle in new-born rats and its elimination during maturation. Journal ofPhysiology 261, 387-422.

BUCHANAN, J., SUN, Y. A. & Poo, M.-M. (1989). Studies of nerve muscle interactions in Xenopuscell-culture - fine-structure of early functional contacts. Journal of Neuroscience 9, 1540-1554.

CABANAS, C. & HOGG, N. (1993). Ligand intercellular adhesion molecule 1 has a necessary role inactivation of integrin lymphocyte function-associated molecule 1. Proceedings of the NationalAcademy of Sciences of the USA 90, 5838-5842.

CALDWELL, J. H. & MILTON, R. L. (1988). Sodium channel distribution in normal and denervatedrodent and snake skeletal muscle. Journal of Physiology 401, 145-161.

CALDWELL, J. H. & RIDGE, R. M. A. P. (1983). The effect of deafferentation and spinal cordtransection on synapse elimination in developing rat muscles. Journal of Physiology 339, 145-159.

CALLAWAY, E., SOHA, J. M. & VAN ESSEN, D. C. (1987). Competition favouring inactive over activemotor neurons during synapse elimination. Nature 328, 422-426.

CALLAWAY, E. M. & VAN ESSEN, D. C. (1989). Slowing of synapse elimination by bungarotoxinsuperfusion of the neonatal rabbit soleus muscle. Developmental Biology 131, 356-365.

CARONI, P. & SCHWAB, M. F. (1988a). Two membrane fractions from rat central myelin withinhibitory properties of neurite growth and fibroblast spreading. Journal of Cell Biology 106,1281-1288.

CARONI, P. & SCHWAB, M. F. (1988 b). Antibody against myelin-associate inhibitor of neurite growthneutralizes non-permissive substrate properties of CNS white matter. Neuron 1, 85-96.

CHAPRON, J. & KOENIG, J. (1989). In vitro synapse maturation. Neuroscience Letters 106, 19-22.COLMAN, H. & LICHTMAN, J. W. (1992 a). 'Cartellian' competition at the neuromuscular junction.

Trends in Neurosciences 15, 197-199.COLMAN, H. & LICHTMAN, J. W. (1992b). Reply to R. R. Ribchester. Trends in Neurosciences 15,

389-390.COLMAN, H. & LICHTMAN, J. W. (1993). Interactions between nerve and muscle: Synapse elimination

at the developing neuromuscular junction. Developmental Biology 156, 1-10.CONNOLD, A. L., EVERS, J. V. & VRBOVA, G. (1986). Effects of low calcium and protease inhibitorson synapse elimination during postnatal development in the rat soleus muscle. Developmental BrainResearch 28, 99-107.

COVAULT, J. & SANES, J. R. (1986). Distribution of N-CAM in synaptic and extrasynaptic portionsof developing and adult skeletal muscle. Journal of Cell Biology 102, 716-730.

DAN, Y. & POO, M.-M. (1992). Hebbian depression of isolated neuromuscular synapses in vitro.Science 256, 1570-1573.

DAVIES, J., COOK, G. M. W., STERN, C. D. & KEYNES, R. J. (1990). Isolation from chick somites ofa glycoprotein fraction which causes collapse of dorsal root ganglion growth cones. Neuron 4,11-20.

487




DAW, N. W., STEIN, P. S. G. & Kox, K. (1993). The role of NMDA receptors in informationprocessing. Aniuiiil Reriew*ofVNeluroscienee 16, 207 222.

DECHANT. G., BIFFO, S., OKAZAWA, H., KOLBECK, R., POFTGIESSER, J. & BARDF, Y. A. (1993 a).Expression and binding characteristics of the BDNF receptor chick trkb. Development 119,545 558.

DECHANT, G., RODRIGUEZ-TEBAR, A., KOLBECK, R. & BARDF, Y. A. (1993 b). Specific high-affinityreceptors for neurotrophin-3 on sympathetic neurons. Jolzs1bl of Neuroscience 13, 2610-2616.

DING, R., JANSFN, J. K. S., LAING, H. G. & TONNESEN, H. (1983). The innervation of skeletal musclesin chickens curarized during early development. Journal of Neurocytologv 12, 887-919.

DOHRMANN, U., EDGAR, D. & TIHOENEN, H. (1987). Distinct neurotrophic factors from skeletalmuscle and the central nervous system interact synergistically to support the survival of culturedembryonic spinal motor neurons. Developmenital Biology 124, 145-152.

DONAHUE, S. P. & ENGLISH, A. W. (1987). The role of synapse elimination in the establishment ofneuromuscular compartments. Develompmenital Biology 124, 481 -489.

DONAHUE, S. P. & ENGLISHl, A. W. (1989). Selective elimination of cross-compartmental innervationin rat lateral gastrocnemius muscle. Jolurnlal of Neluoscienece 9, 1621-1627.

DUXSON, M. J. (1982). The effect of post-synaptic block on development of the neuromuscularjunction in postnatal rats. Journial ol Neurocv'tologv 11, 395-408.

ECCLES, J. (1990). Developing concepts of the synapse. Journal of Neuroscience 10, 3769-3781.EDELMAN, G. (1987). Neur-al Darltinismn. The Theory ol'f Neuronlal Group Selectioni. Basic Books, New

York.EDELMAN, G. (1992). Bright Air, Brilli(anit Fire. Basic Books, New York.FAWCETT, J. W. & HOUSDEN, E. (1990). The effects of protease inhibitors on axon growth through

astrocytes. Development 109, 59-66.FLADBY, T. (1987). Postnatal loss of synaptic terminals in the normal mouse soleus muscle. ActcPhy.siologica Scandinavica 129, 229-238.

FLADBY, T. & JANSEN, J. K. S. (1987). Postnatal loss of synaptic terminals in the partially denervatedmouse soleus muscle. Acta Phvsiologica Scantlinavica 129, 239-246.

FLADBY, T. & JANSEN, J. K. S. (1990). Development of homogeneous fast and slow motor units in theneonatal mouse soleus muscle. Development 109, 723-732.

FRANK, E., JANSEN, J. K. S., LoMo, T. & WESTGAARD, R. H. (1975). The interaction between foreignand original motor nerves innervating the soleus muscle of the rat. Joulrna(il of Physiology 247,725-743.

FUNAKOSTIi, H., FRISEN, J., TIMMUSK, T., ZACHRISSON, O., VERGF, V. M. K. & PERSSON, H. (1993).Differential expression of mRNAs for neurotrophins and their receptors after axotomy of thesciatic nerve. Joulss101 of Cell Biology 123, 455-465.

GA-ES, H. & RIDGE, R. M. A. P. (1992). The importance of competition between motoneurones indeveloping rat muscle: effects of partial denervation at birth. Jolurnlal of Physiolog, 445, 457-472.

GORDON, H. & VAN ESSEN, D. C. (1985). Specific innervation of muscle fibre types in adevelopmentally polyinnervated muscle. Developmental Biology 111, 42-50.

GORDON, T., PERRY, R., TUFFEREY, A. R. & VRBOVA, G. (1974). Possible mechanisms determiningsynapse formation in developing skeletal muscles of the chick. Cell anlud Tissule Research 155, 13-25.

GORIO, A., CARMIGNOTO, G., FINESSO, M., POLATO, P. & NUNZI, M. G. (1983). Muscle reinnervation.Sprouting, synapse formation and repression. Neuroscienice 8, 403-416.

GRINNELL, A. D., LETINSKY, M. S. & RHEUBEN, M. B. (1979). Competitive interaction betweenforeign nerves innervating frog skeletal muscle. Jolle-1cl of Phy.siology 289, 241-262.

GURNEY, M. E., YAMAMOTO, H. & KWON, T. (1992). Induction of motor neuron sprouting in vivo byciliary neurotrophic factor and basic fibroblast growth factor. Jolurniial of Neuro.scienee 12,3241 -3247.

GUIH, L. (1962). Neuromuscular fuinction after regeneration of interrupted nerve fibres into partiallydenervated muscle. Ex-perimental Nelurology 6, 129- 141.

HAIMANN, C., MALLART, A., FERRE, J. T. 1. & ZILBER-GACIHELIN, N. F. (1981). Interaction betweenmotor axons from two different nerves reinnervating the pectoral muscle of Xeniopu s laevi.s. Journalof Ph_vsiologji 310, 257 272.

HARDMAN, V. & BROWN, M. C. (1987). Accuracy of reinnervation of rat intercostal muscles by theirown segmental nerves. Journal of Neuroscilence 7, 1071-1036.

HARDMAN, V. J. & BROWN, M. C. (1985). Absence of postnatal death among motoneurones supplyingthe inferior gluteal nerve in the rat. Developmlental Brain Research 19, 1-9.

488




HARISH, 0. E. & Poo, M.-M. (1992). Retrograde modulation at developing neuromuscular synapses:involvement of G-protein and arichidonic acid cascade. Neuron 9, 1201-1209.

HENDERSON, C. E. (1988). The role of muscle in the development and differentiation of spinalmotoneurones: in vitro studies. CIBA Symposium 138, 172-191.

HENDERSON, C. E., CAMU, W., METTLING, C., GouIN, A., POULSEN, K., KARIHALOO, M., RULLAMAS,J., EVANS, T., MCMAHON, S. B., ARMANINI, M. P., BERKEMEIER, L., PHILLIPS, H. S. & ROSENTHAL,A. (1993). Neurotrophins promote motor neuron survival and are present in embryonic limb bud.Nature 363, 266-270.

HEUMANN, R., KORSCHING, S., SCOTT, J. & THOENEN, H. (1984). Relationship between levels of nervegrowth factor (NGF) and its messenger RNA in sympathetic ganglia and peripheral nerves. EMBOJournal 3, 3183-3189.

HEUMANN, R. & THOENEN, H. (1986). Comparison of the time course of changes in nerve growthfactor protein levels and those of its messenger RNA in the cultured rat iris. Journal of BiologicalChemistry 261, 9246-9249.

HOLLAND, R. L. & BROWN, M. C. (1980). Postsynaptic transmission block can cause terminalsprouting of a motor nerve. Science 207, 649-651.

HUME, R. I. & PURVES, D. (1988). Geometry of neonatal neurones and the regulation of synapseelimination. Nature 293, 469-471.

HUNTER, D. D., SHAH, V., MERLIE, J. P. & SANES, J. R. (1989). A laminin-like adhesive proteinconcentrated in the synaptic cleft of the neuromuscular junction. Nature 338, 229-234.

JACKSON, H. & PARKS, T. N. (1982). Functional synapse elimination in the developing avian cochlearnucleus with simultaneous reduction in cochlear nerve branching. Journal of Neuroscience 2,1736-1743.

JANSEN, J. K. S. & FLADBY, T. (1990). The perinatal reorganization of the innervation of skeletalmuscle in mammals. Progress in Neurobiology 34, 39-90.

JANSEN, J. K. S., LoMo, T., NICOLAYSEN, K. & WESTGAARD, R. H. (1973). Hyperinnervation ofskeletal muscle fibres: dependence on muscle activity. Science 181, 559-561.

JONES, S. P., RIDGE, R. M. A. P. & ROWLERSON, A. (1987 a). The non-selective innervation of musclefibres and mixed composition of motor units in a muscle of the neonatal rat. Journal of Physiology386, 377-394.

JONES, S. P., RIDGE, R. M. A. P. & ROWLERSON, A. (1987b). Rat muscle during post-nataldevelopment: evidence of no interconversion between fast and slow twitch fibres. Journal ofPhysiology 386, 395-406.

JORDAN, C. L., LETINSKY, M. S. & ARNOLD, A. P. (1989). The role of gonadal hormones inneuromuscular synapse elimination in rats. II: multiple innervation persists in the adult levator animuscle after juvenile androgen treatment. Journal of Neuroscience 9, 239-247.

KAPFHAMMER, J. P. & RAPER, J. A. (1987). Collapse of growth cone structure on contact with specificneurites in culture. Journal of Neuroscience 7, 201-212.

KATZ, B. (1969). The Release of Neural Transmitter Substances. Liverpool University Press,Liverpool.

KEDDY, P. (1989). Competition. Chapman & Hall, London, New York.KEYNES, R. J. & STERN, C. D. (1984). Segmentation in the vertebrate nervous system. Nature 310,

786-789.Ko, C.-P. (1987). A lectin, peanut-agglutinin, as a probe for the extracellular matrix in living

neuromuscular junctions. Journal of Neurocytology 16, 567-576.KOLIATSOS, V. E., CLATTERBUCK, R. E., WINSLOW, J. W., CAYOUETTE, M. H. & PRICE, D. L. (1993).

Evidence that brain-derived neurotrophic factor is a trophic factor for motor neurons in vitro.Neuron 10, 359-367.

KORNELIUSEN, H. & JANSEN, J. K. S. (1976). Morphological aspects of the elimination of polyneuronalinnervation of skeletal muscle fibres in newborn rats. Journal of Neurocytology 5, 591-604.

KORSCHING, S. (1993). The neurotrophic factor concept- a reexamination. Journal of Neuroscience13, 2739-2748.

KOSHLAND, D. E. (1973). Protein shape and biological control. Scientific American 229, 52-64.KOSHLAND, D. E., NEMETH, G. & FILMER, D. (1966). Comparison of experimental binding data and

theoretical models in proteins containing subunits. Biochemistry 5, 365-385.KUFFLER, D. P., THOMPSON, W. & JANSEN, J. K. S. (1977). The elimination of synapses in multiple-

innervated skeletal muscle fibres of the rat: dependence on distance between end-plates. BrainResearch 138, 353-358.

489




KUFFLER, D. P., THOMPSON, W. & JANSEN, J. K. S. (1980). The fate of foreign endplates on cross-innervated rat soleus muscle. Proceedings of the Royal Society B 208, 189-222.

LAMB, A. H. (1980). Motoneurone counts in Xenopus frogs reared with one bilaterally innervatedhindlimb. Nature 284, 347-350.

LANCE-JONES, C. (1982). Motoneuron cell death in the developing lumbar spinal cord of the mouse.Developmental Brain Research 4, 473-479.

LANCE-JONES, C. (1988). Development of neuromuscular connections: guidance of motoneuroneaxons to muscles in embryonic chick hindlimb. CIBA Symposium 138, 97-115.

LANCE-JONES, C. & LANDMESSER, L. T. (1980). Motoneurone projection patterns in the chick hindlimb following early partial reversals of the spinal cord. Journal of Physiology 302, 581-602.

LANDMESSER, L. T. (1984). The development of specific motor pathways in the chick embryo. Trendsin Neurosciences 7, 337-339.

LANDMESSER, L. T., DAHM, L., TANG, J. & RUTISHAUSER, U. (1990). Polysialic acid as a regulator ofintramuscular nerve branching during embryonic development. Neuron 4, 655-667.

LASKOWSKI, M. B. & HIGH, J. A. (1989). Expression of nerve-muscle topography during develop-ment. Journal of Neuroscience 8, 3094-3099.

LASKOWSKI, M. B. & SANES, J. R. (1988). Topographically selective reinnervation of adult mammalianskeletal muscle. Journal of Neuroscience 8, 3094-3099.

LAUFER, R. & CHANGEUX, J.-P. (1989). Activity-dependent regulation of gene expression in muscleand neuronal cells. Molecular Neurobiology 3, 1-53.

LAVIDIS, N. & BENNETT, M. R. (1992). Probabilistic secretion from visualized sympathetic nervevaricosities of mouse vas deferens. Journal of Physiology 454, 9-26.

LEvI-MoNTALCINI, R. (1987). The nerve growth factor 35 years later. Science 237, 1154-1162.LICHTMAN, J. W. & BALICE-GORDON, R. J. (1990). Understanding synaptic competition in theory and

in practice. Journal of Neurobiology 21, 99-106.LICHTMAN, J. W., MAGRASSI, L. & PURVES, D. (1987). Visualization of neuromuscular junctions over

periods of several months in living mice. Journal of Neuroscience 7, 1215-1222.LICHTMAN, J. W. & WILKINSON, R. (1987). Properties of motor units in the transversus abdominis

muscle of the garter snake. Journal of Physiology 393, 355-374.LICHTMAN, J. W., WILKINSON, R. S. & RICH, M. M. (1985). Multiple innervation of tonic end-plates

revealed by activity-dependent uptake of fluorescent probes. Nature 314, 357-359.LINDSAY, R. M., ALDERSON, R. F., FRIEDMAN, B., HYMAN, C., FURTH, M. E., MAISONPIERRE, P. C.,

SQUINTO, S. P. & YANCOPOULOS, G. D. (1991). The neurotrophin family of ngf-related neurotrophicfactors. Restorative Neurology and Neuroscience 2, 211-220.

Lo, Y. J. & Poo, M. M. (1991). Activity-dependent synaptic competition in vitro: heterosynapticsuppression of developing synapses. Science 254, 1019-1022.

LOHOF, A. M., Ip, N. Y. & Poo, M.-M. (1993). Potentiation of developing neuromuscular synapsesby the neurotrophins NT-3 and BDNF. Nature 363, 350-353.

LUBISCHER, J. L., JORDAN, C. L. & ARNOLD, A. P. (1992). Transient and permanent effects ofandrogen during synapse elimination in the levator ani muscle of the rat. Journal of Neurobiology23, 1-9.

LUFF, A. R. & TORKKO, K. (1990). Long-term persistence of enlarged motor units in partiallydenervated hindlimb muscles of cats. Journal of Neurophysiology 64, 1261-1269.

LuIzzI, F. J. & LASEK, R. J. (1987). Astrocytes block axonal regeneration in mammals by activatingthe physiological stop pathway. Science 237, 642-645.

LUPA, M. T. & CALDWELL, J. H. (1991). Effect of agrin on the distribution of acetylcholine receptorsand sodium channels on adult skeletal muscle fibres in culture. Journal of Cell Biology 115,765-778.

LUPA, M. T., KRZEMIEN, D. M., SCHALLER, K. L. & CALDWELL, J. H. (1993). Aggregation of sodiumchannels during development and maturation of the neuromuscular junction. Journal of Neuro-science 13, 1326-1336.

McARDLE, J. J. (1975). Complex end-plate potentials at the regenerating neuromuscular junction ofthe rat. Experimental Neurology 49, 629-638.

MAGCHIELSE, T. & MEETER, E. (1986). The effect of neuronal activity on the competitive eliminationof neuromuscular junctions in tissue culture. Developmental Brain Research 25, 211-220.

MALENKA, R. C. & NICOLL, R. A. (1993). NMDA-receptor dependent synaptic plasticity: multipleforms and mechanisms. Trends in Neurosciences 16, 521-527.

490




MASSOULIE, J. & BON, S. (1983). The molecular forms of cholinesterase and acetylcholinesterase invertebrates. Annual Review of Neuroscience 5, 57-106.

MERLIE, J. P. & SANES, J. R. (1985). Concentration of acetylcholine receptor mRNA in synapticregions of adult muscle fibres. Nature 317, 66-68.

MIRSKY, R. & JESSEN, K. R. (1984). A cell surface protein of astrocytes, Ran-2, distinguishes non-myelin forming Schwann cells from myelin-forming Schwann cells. Developmental Neuroscience 6,304-316.

MONARD, D. (1988). Cell derived proteases and protease inhibitors as regulators of neuriteoutgrowth. Trends in Neurosciences 11, 541-544.

NAVARETTE, R. & VRBOVA, G. (1993). Activity-dependent interactions between motoneurones andmuscles: their role in the development of the motor unit. Progress in Neurobiology 41, 93-124.

NELSON, P. G., FIELDS, R. D., Yu, C. & LIu, Y. (1993). Synapse elimination from the mouseneuromuscular junction in vitro: a non-Hebbian activity-dependent process. Journal ofNeurobiology 24, 1517-1530.

O'BRIEN, R. A. D., OSTBERG, A. J. C. & VRBOVA, G. (1978). Observations on the elimination ofpolyneuronal innervation in developing mammalian skeletal muscle. Journal of Physiology 282,571-582.

O'BRIEN, R. A. D., OSTBERG, A. J. C. & VRBOVA, G. (1984). Protease inhibitors reduce the loss ofnerve terminals induced by activity and calcium in developing rat soleus muscles in vitro.Neuroscience 12, 637-646.

OLSON, L. & MALMFORS, T. (1970). Growth characteristics of adrenergic nerves in the adult rat.Fluorescence, histochemical and 3H-noradrenaline uptake studies using tissue transplanted to theanterior chamber of the eye. Acta Physiologica Scandinavica Supplementum 348, 1-111.

OPPENHEIM, R. W. (1986). The absence of significant postnatal motoneuron death in the brachial andlumbar spinal cord of the rat. Journal of Comparative Neurology 246, 281-286.

OPPENHEIM, R. W. (1991). Cell death during development of the nervous system. Annual Review ofNeuroscience 14, 453-501.

OPPENHEIM, R. W., HAVERKAMP, L. J., PREVETTE, D., MCMANAMAN, J. L. & APPEL, S. (1988).Reduction of naturally occurring motor neuron death in vivo by a target-derived neurotrophicfactor. Science 240, 919-922.

OPPENHEIM, R. W., MADERUT, J. L. & WALLS, D. J. (1982). Reduction of naturally occurring celldeath in the thoracolumbar preganglionic cell column of the chick embryo by nerve growth factorand hemicholinium 3. Developmental Brain Research 3, 134-139.

OPPENHEIM, R. W., QIN-WEI, Y., PREVETTE, D. & YAN, Q. (1992). Brain derived neurotrophic factorrescues developing avian motor neurons from cell death. Nature 360, 755-757.

PATTERSON, P. H. (1988). On the importance of being inhibited, or saying no to growth cones. Neuron1, 263-267.

PURVES, D. (1988). Body and Brain: A Trophic Theory of Nerve Connections. Harvard, Boston.PURVES, D. & LICHTMAN, J. W. (1985). Principles of Neural Development. Sinauer, New York.PURVES, D., VOYVODIC, J. T., MAGRASSI, L. & YAWO, H. (1987). Nerve terminal remodelling

visualized in living mice by repeated examination of the same neuron. Journal of Neuroscience 7,1215-1222.

RABBACHI, S., BAILLY, Y., DELHAYE-BOUCHAUD, N. & MARIANI, J. (1992). Involvement of the N-methyl-D-aspartate (NMDA) receptor in synapse elimination during cerebellar development.Science 256, 1823-1825.

REDFERN, P. A. (1970). Neuromuscular transmission in new-born rats. Journal of Physiology 209,701-709.

RIBCHESTER, R. R. (1988a). Activity dependent and independent synaptic interactions duringreinnervation of partially denervated rat muscle. Journal of Physiology 401, 53-75.

RIBCHESTER, R. R. (1988b). Competitive elimination of neuromuscular synapses. Nature 331, 21.RIBCHESTER, R. R. (1992). Cartels, competition and activity-dependent synapse elimination. Trends

in Neurosciences 15, 389.RIBCHESTER, R. R. (1993). Co-existence and elimination of convergent motor nerve terminals in

reinnervated and paralysed adult rat skeletal muscle. Journal of Physiology 466, 421-441.RIBCHESTER, R. R., MAO, F. & BETZ, W. J. (1994). Optical measurements of activity-dependentmembrane recycling in motor nerve terminals of mammalian skeletal muscle. Proceedings of theRoyal Society B 255, 61-66.

491




RIBCHESTER, R. R. & TAXT, T. (1983). Motor unit size and synaptic competition in rat lumbricalmuscles reinnervated by active and inactive motor axons. Journal of Physiology 344, 89-111.

RIBCHESTER, R. R. & TAXT, T. (1984). Repression of inactive motor nerve terminals in partiallydenervated rat muscle after regeneration of active motor axons. Journal of Physiology 347,497-511.

RICH, M. M. & LICHTMAN, J. W. (1989). In vivo visualization of pre- and postsynaptic changes duringsynapse elimination in reinnervated mouse muscle. Journal of Neuroscience 9, 1781-1805.

RIDGE, R. M. A. P. & BETZ, W. J. (1984). The effect of selective chronic stimulation on motor unitsize in developing rat muscle. Journal of Neuroscience 4, 2614-2620.

RIDGE, R. M. A. P. & ROWLERSON, A. (1900). Sprouting motoneurones break as well as makecontacts in partially denervated rat muscle. Journal of Physiology 425, 15P.

RILEY, D. (1977). Spontaneous elimination of nerve terminals from the endplates of developingskeletal myofibers. Brain Research 134, 279-285.

RINI, J. M., SCHULZE-GAHMEN, U. & WILSON, I. A. (1992). Structural evidence for induced fit as amechanism for antibody-antigen recognition. Science 255, 959-965.

RODRIGUEZ-TEBAR, A., DECHANT, G., GOTZ, R. & BARDE, Y. A. (1992). Binding of neurotrophin-3to its neuronal receptors and interactions with nerve growth-factor and brain-derived neurotrophicfactor. EMBO Journal 11, 917-922.

Ross, J. J., DUXSON, M. J. & HARRIS, A. J. (1987). Neural determination of muscle fibre numbers inembryonic rat lumbrical muscles. Development 100, 395-409.

RYAN, T. A., REUTER, H., WENDLAND, B., SCHWEIZER, F. E., TSIEN, R. W. & SMITH, S. J. (1993). Thekinetics of synaptic vesicle recycling measured at single presynaptic boutons. Neuron 11, 713-724.

SAFFELL, J. L., WALSH, F. S. & DOHERTY, P. (1992). Direct activation of second messenger pathwaysmimics cell adhesion molecule-dependent outgrowth. Journal of Cell Biology 118, 663-670.

SANES, J. R., JOHNSON, Y. R., KOTZBAUER, I. T., MUDD, J., HANLEY, T., MARTINOU, J. C. &MERLIE, J. P. (1991). Selective expression of an acetylcholine lac Z transgene in synaptic nuclei ofadult muscle fibres. Development 113, 1181-1191.

SANES, J. R., SCHACHNER, M. & COVAULT, J. (1986). Expression of several adhesive macromolecules(N-CAM, LI, JI, NILE, uvomorulin, laminin, fibronectin, and a heparan sulfate proteoglycan) inembryonic, adult and denervated adult skeletal muscle. Journal of Cell Biology 102, 420-431.

SCHALLER, K. L., KRZEMIEN, D. M., MCKENNA, N. M. & CALDWELL, J. H. (1992). Alternativelyspliced sodium channel transcripts in brain and muscle. Journal of Neuroscience 12, 1370-1381.

SCHNELL, L. & SCHWAB, M. E. (1990). Axonal regeneration in the rat spinal cord produced by anantibody against myelin associated neurite growth inhibitors. Nature 343, 269-272.

SCHNELL, L. & SCHWAB, M. E. (1993). Sprouting and regeneration of lesioned corticospinal tractfibres in the adult rat spinal cord. European Journal of Neuroscience 5, 1156-1171.

SCHWAB, M. E. (1993). Experimental aspects of spinal cord regeneration. Current Opinions inNeurology and Neurosurgery 6, 549-553.

SCHWAB, M. & CARONI, P. (1988). Oligodendrocytes and CNS myelin are non-permissive substratesfor neurite growth and fibroblast spreading in vitro. Journal of Neuroscience 9, 1997-2010.

SCOTT, L. J. C., BALSAMOMO, J., SANES, J. R. & LILIEN, J. (1990). Synaptic localization and neuralregulation of an N-acetylgalactosaminyl transferase in skeletal muscle. Journal ofNeuroscience 10,346-350.

SENDTNER, M., HOLTMANN, B., KOLBECK, R., THOENEN, H. & BARDE, Y. A. (1992a). Brain derivedneurotrophic factor prevents the death of motor neurons in newborn rats after nerve section.Nature 360, 757-759.

SENDTNER, M., KREUZBERG, G. W. & THOENEN, H. (1990). Ciliary neurotrophic factor prevents thedegeneration of motor neurons after axotomy. Nature 345, 440-441.

SENDTNER, M., SCHMALBRUCH, H., STOCKLI, J. A., CARROLL, P., KREUTZBERG, G. W. & THOENEN, H.(1992 b). Ciliary neurotrophic factor prevents the degeneration of motor neurons in mouse mutantprogressive neuronopathy. Nature 358, 502-504.

SHATZ, C. (1990). Impulse activity and the patterning of connections during CNS development.Neuron 5, 745-756.

SHEARD, P. W. & LAMB, A. H. (1991). Motoneuron and muscle fibre counts in normal and bilaterallyinnervated Xenopus hindlimbs. Developmental Brain Research 58, 133-142.

SHELTON, D. L. & REICHARDT, L. F. (1984). Expression of B-nerve growth factor gene correlates withthe density of sympathetic innervation in effector organs. Proceedings of the National Academy ofSciences of the USA 81, 7951-7955.

492




SLATER, C. R. (1982). Neural influence on the postnatal changes in acetylcholine receptor distributionat nerve-muscle junctions in the mouse. Developmental Biology 94, 11-22.

SMALL, S. J. & AKESON, R. (1990). Expression of the unique NCAM VASE exon is independentlyregulated in distinct tissues during development. Journal of Cell Biology 111, 2089-2096.

SMALLHEISER, N. R. & CRAIN, S. M. (1984). The possible role of "sibling neurite bias" in thecoordination of neurite extension, branching and survival. Journal of Neurobiology 15, 517-529.

SNIDER, W. (1988). Nerve growth factor promotes dendritic arborization of sympathetic ganglioncells in developing mammals. Journal of Neuroscience 8, 2628-2634.

SOHA, J. M., YO, C. & VAN ESSEN, D. C. (1987). Synapse elimination by fibre type and maturationalstate in rabbit soleus muscle. Developmental Biology 123, 136-144.

SRIHARI, T. & VRBOVA, G. (1978). The role of muscle activity in the differentiation of neuromuscularjunctions in slow and fast chick muscle. Journal of Neurocytology 7, 529-540.

STENT, G. S. (1973). A physiological mechanism for Hebb's postulate of learning. Proceedings of theNational Academy of Sciences of the USA 70, 97-1001.

TANAKA, H. & LANDMESSER, L. T. (1986). Interspecies selective motor neuron projection patterns inchick-quail chimeras. Journal of Neuroscience 6, 2880-2888.

TANG, J. & LANDMESSER, L. T. (1993). Reduction of intramuscular nerve branching andsynaptogenesis is correlated with decreased motoneuron survival. Journal of Neuroscience 13,3095-3103.

TAXT, T. (1983 a). Cross innervation of fast and slow-twitch muscles by motor axons of the sural nervein the mouse. Acta Physiologica Scandinavica 117, 331-341.

TAXT, T. (1983b). Local and systemic effects of tetrodotoxin on the formation and elimination ofsynapses in reinnervated adult muscle. Journal of Physiology 340, 175-194.

THOENEN, H. (1991). The changing scene of neurotrophic factors. Trends in Neurosciences 14,165-170.

THOMPSON, W. (1978). Reinnervation of partially denervated rat soleus muscle. Acta PhysiologicaScandinavica 103, 81-91.

THOMPSON, W. (1983). Synapse elimination in neonatal rat muscle is sensitive to pattern of muscleuse. Nature 302, 614-616.

THOMPSON, W. & JANSEN, J. K. S. (1977). The extent of sprouting of remaining motor units inpartially denervated immature and adult rat soleus muscle. Neuroscience 2, 523-535.

THOMPSON, W. J., KUFFLER, D. P. & JANSEN, J. K. S. (1979). The effect of prolonged reversible blockof nerve impulses on the elimination of polyneuronal innervation of new-born rat skeletal musclefibres. Neuroscience 4, 271-281.

THOMPSON, W. J., SUTTON, L. A. & RILEY, D. A. (1984). Fibre type composition of single motor unitsduring synapse elimination in neonatal rat soleus muscle. Nature 309, 709-711.

VOYVODIC, J. T. (1989). Peripheral target regulation of dendritic geometry in the rat superior cervicalganglion. Journal of Neuroscience 9, 1997-2010.

VRBOVA, G., LOWRIE, M. B. & EVERS, J. (1988). Reorganization of synaptic inputs to developingskeletal muscle fibres. CIBA Symposium 138, 131-151.

WALSH, F. S., FURNESS, J., MOORE, S. E., ASHTON, S. & DOHERTY, P. (1992). Use of the neural celladhesion molecule VASE exon by neurons is associated with a specific down-regulation of neuralcell adhesion molecule-dependent neurite outgrowth in developing cerebellum and hippocampus.Journal of Neurochemistry 59, 1959-1962.

WATSON, W. E. (1976). Cell Biology of Brain. Chapman & Hall, London.WERLE, M. J. & HERRERA, A. A. (1988). Synaptic competition and the elimination of polyneuronal

innervation following reinnervation of adult frog sartorius muscles. Journal of Neurobiology 19,465-481.

WERLE, M. J. & HERRERA, A. A. (1991). Elevated levels of polyneuronal innervation persist for aslong as two years in reinnervated frog neuromuscular junction. Journal ofNeurobiology 22, 97-103.

WIGSTON, D. J. & SANES, J. R. (1985). Segmental reinnervation of intercostal muscle transplantedfrom different segmental levels to a common site. Journal of Neuroscience 5, 1208-1221.

WITZEMANN, V., BREMNER, H. R. & SAKMANN, B. (1991). Neural factors regulate AChR subunitmRNA at rat neuromuscular synapses. Journal of Cell Biology 114, 125-141.

YAN, Q., ELLIOTT, J. & SNIDER, W. D. (1992). Brain derived neurotrophic factor rescues spinal motorneurons from axotomy-induced cell death. Nature 360, 753-755.

YANKNER, B. A. & SHOOTER, E. M. (1983). The biology and mechanism of action of nerve growthfactor. Annual Review of Biochemistry 51, 845-868.

493



494 R. R. RIBCHESTFR AND J. A. BARRY

ZHU, P.-H. & VRBOVA, G. (1992). The role of Ca in the elimination of polyneuronal innervation ofrat soleus muscle fibres. Europeati Jolurnlal of Neuroscience 4, 433-437.

ZORN, A. M. & KREIG, P. A. (1992). Developmental regulation of alternative splicing in the mRNAencoding Xenopus laevis neural cell adhesion molecule (NCAM). Developlmenttal Biology 149.327-337.



ribchester & barry

Documents