Activity-dependent interactions between motoneurones and muscles: Their role in the development of the motor unit

Progressin NeurobiologyVol. 41, pp. 93 to 124, 1993 Printed in Great Britain. All rights reserved

0301-0082/93/$24.00 © 1993 Pergamon Press Ltd

ACTIVITY-DEPENDENT INTERACTIONS BETWEEN MOTONEURONES A N D MUSCLES: THEIR ROLE IN THE DEVELOPMENT OF THE MOTOR UNIT ROBERTO NAVARRETE* and GERTA VRBOV.~ Department of Anatomy and Developmental Biology, Centrefor Neuroscience, University College London, Gower Street, London WCIE 6BT, U.K. *Department of Anatomy and Cell Biology, Chafing Cross and Westminster Medical School, University of London, Fuiham Palace Road, London W6 8RF, U.K.

(Received 4 June 1992)

1. 2.

3.

4.

5. 6.

CONTENTS Abbreviations Introduction Development of components of the motor unit 2.1. Motoneurone development 2.1.1. Early development: cell lineage and expression of the cholinergic phenotype 2.1.2. Development of electrical excitability 2.1.3. Development of functional specializatioti 2.1.4. Target-dependence of motonenrones 2.2. Muscle development 2.2.1. Emergence of fibre type diversity 2.2.2. Mechanisms involved in generating fibre type diversity 2.2.3. Acetylcholine receptors and acetylcholinesterase 2.2.4. Development of ion channels and passive electrical properties Establishment of neuromuscular connections and their reorganization during development 3.1. Characteristics of early neuromuscular transmission 3.2. Reorganization of synaptic contacts during postnatal development 3.2.1. Activity and regulation of neuromuscular contacts: relevance to motor unit size 3.2.2. Mechanisms involved in the regulation of synaptic contacts Development of motor activity 4.1. Embryonic development of motor function 4.1.1. Electrophysiologicai correlates 4.2. Postnatal development of motor function Mechanisms influencing the development of the motor unit: role of activity 5.1. Early layout of motor unit territory 5.2. Activity induced modifications of muscle fibre properties Summary and conclusions Acknowledgements References ABBREVIATIONS ACh AChE AChR ADP AHP ALD BAPTA-AM BTX BoTx CANP CGRP ChAT CNS E EDL EMG EPP GABA gCl -

i^

acetylcholine acetylcholinesterase acetylcholine receptor afterdepolarization afterhyperpolarization anterior latissimus dorsi acetoxymethyl ester of BAPTA 1,2-bis (2-aminophcnoxylethaneN l N,N; Nm-tetraacetic acid); cellpermeant a-bungarotoxin botulinum toxin calcium-activated neutral protease calcitonin gene-related peptide choline acetyltransferase central nervous system embryonic extensor digitorum longus electromyography end#ate potential y-aminobutyric acid chloride conductance A-type transient potassium current 93

93 94 94 94 94 95 96 99 100 100 101 102 104 106 106 106 107 107 109 109 110 !11 113 113 117 118 119 119

94

R. NAVARRETEand G. VRBOVA

~AHP & MEPP MHC MUP NMDA P PLD TA STX TTX

sustained calcium-activated potassium current sustained potassium current miniature endplate potential myosin heavy chain motor unit potential N-methyl o-aspartate postnatal posterior latissimus dorsi tibialis anterior saxitoxin tetrodotoxin

1. INTRODUCTION The motor unit is a term coined by Liddeil and Sherrington (1925) to describe the motoneurone with all its axonal branches and the muscle fibres it supplies. The term was intended to emphasize the functional unity between two separate cell types, i.e. the motoneurone and muscle fibres. It is this unit which, when all integrative processes that occur in the CNS are accomplished, executes the final task of movement. The way in which these two cell types interact during development so as to develop into a functional unit is still not clear. In this article the role of activity in shaping the properties of motoneurones and muscle fibres will be considered, with particular focus on the consequences of activity related interactions between motoneurones and muscle fibres which contribute to the emergence of the motor unit and its functional specialization. Our motivation for attempting to review this subject was prompted by the fact that the two lines of research, i.e. the ontogeny of motor activity and the cellular aspects of motoneurone and muscle differentiation, are usually considered independently. Yet activity is known to be important in many aspects of the development and maintenance of the neuromuscular system, and it is therefore important to relate developmental changes of activity to the differentiation of particular components of the motor unit. Motoneurones and muscle fibres arise from embryologieally distinct cell populations and initially develop independently of each other. At a certain stage of development they form synaptic contacts, and it is after this time of their first encounter that they become critically dependent upon each other The motoneurone gradually acquires synaptic contacts with a large number of muscle fibres and after a transient period during which individual muscle fibres are shared by several motoneurones, the emergence of the motor unit as known in the adult mammal occurs. After this transitional stage each muscle fibre becomes committed to only a single motoneurone. This allows the muscle fibres belonging to a motor unit to develop phenotypic characteristics that are matched to the function imposed upon them by their motoneurone, and in this way become an independent functional unit. The gradual emergence of heterogeneous populations of motor units and the role of activity in this process will be discussed. The development of motoneurones and their functional specialization will be considered first. Next the evidence for the intrinsic capacity of muscle fibres to develop some of their phenotypic characteristics independently of neuronal influences and the possible role

of this early diversity in the formation of the motor unit will be discussed. We will then review the possible mechanisms involved in the e s t a b l i s ~ n t of synaptic contacts and their maintenance during development with specific emphasis on the role of activity. Finally, the development of motor activity and the influence of emerging patterns of motoneurone activity on the development of distin~ populations of motor units will be discussed.

2. DEVELOPMENT OF C O ~ OF THE MOTOR UNIT 2.1. MOTONEURONEDEVELOPMENT 2.1.1. Early development: cell lineage and

expression of the cholinergic phenotype Motoneurones are derived from neuroepithelial cell precursors in the ventricular zone of the neural tube (Langman and Haden, 1970, for review see Bennett, 1983). After their final mitosis, neuroblasts begin to migrate from the ventricular zone towards their final location in the motor columns of the ventral horn. This migration is believed to occur along pathways provided by radial gtial cells (Hendrikson and Vaughn, 1974; Leber et a£, 1990). These initial stages of motoneurone development are independent of influences from the target muscles, since both proliferation and subsequent migration are unaffected by early limb bud removal (Hamburger, 1958; Hughes and Tschumi, 1958). There is little information as to the timing of the generation of motoneurone pools. Early studies indicated that motoneurones born first settle medially whereas those born later migrate more laterally (Hollyday and Hamburger, 1977). Whether this is related to the emergence of distinct motoneurone pools is not clear. It has been proposed that motoneurones are clonally related to the muscle fibres they innervate (Moody and Jacobson, 1983). However, results from recent studies on chick embryos using retroviral labelling of motonenrone progenitors do not support this proposal. In these experiments multicelhilar dunes derived from s/ngle labetled neuroblasts were found to contain motoneurones as well as other cell types, such as interneurones, glial and ependymal cells (Leber et aL, 1990). These latter results indicate that motoneurone precursors are multipotential and e#genetic factors may be involved in determining their final phenotype. Moreover, the formation of distinct motoneurone pools does not appear to be lineage-dependent since eionally related

95

ACTIVITY-DEPENDENT l~rl~xCl'lOlqS BETWEENMOTONEURONE$ AND MUSCLES

motoneurones were not restricted to a single motor pool (Leber et aL, 1990). Similar results were obtained by Vogel et aL (1988) using a chimeric analysis for studying motoneurone lineage in the mouse. One of the first events that distinguishes motoneurones from other spinal cord neurones is the appearance~of the neurotransmitter-related enzymes choline acetyltransferase (CHAT) and acetylcholinesterase (ACHE). These enzymes are present in motoneurones soon after their final mitosis, at the time when their axons have not yet reached their target muscles (Phelps et ai., 1990). There are indications that once cholinergic characteristics are established, interaction with the target muscle is necessary in order to induce and maintain a high level of expression of the enzymes involved in ACh synthesis (GiacobiniRobecchi et al., 1975; Betz et al., 1980; reviewed by Vaca, 1988). During embryonic and early postnatal development the level of ChAT increases (Burr, 1975; O'Brien and Vrbov~., 1978; Pilar et al., 1981; Phelps et al., 1984). Electrical activity or membrane depolarization seems to be important for the up-regulation of CHAT since tetrodotoxin (TTX) blockade of action potentials in spinal cord cultures leads to a reduction or delayed development of ChAT activity (Brenneman et al., 1983). Conversely, chronic membrane depolarization resulting from exposure to high K + or activation of N M D A receptors induces an increase in CHAT activity (Ishida and Deguchi, 1983; Brenneman et al., 1990)

Several types of Ca ++ currents with different voltage-dependence, kinetics and pharmacological sensitivity have been described in adult mammalian neurones (see Llin~is, 1988). In general, high threshold Ca ++ conductances (L and N types) are activated by strong depolarization which normally occurs during spike generation, whereas low threshold conductances (T type) are activated near the resting potential. Recent studies indicate that developing motoneurones possess different types of Ca + + conductances. It is however not clear whether in the adult different types of Ca + + conductances are present and confined to particular motoneurone subpopulations (reviewed by Schwindt and Crill, 1984; K.iehn, 1991). With development the contribution of different types of Ca ++ conductances to the action potential changes and Fig. 1 illustrates this. It shows that motoneurones from FA chick embryos have predominantly low threshold (or T-type) Ca + + currents and these decrease during subsequent development in vitro. In contrast, the high threshold Ca + + currents (L and N type) increase in magnitude after E4 and reach peak values at E11 (McCobb et al., 1989). Neonatal rat motoneurones have both high and low threshold Ca ++ conductances (Harada and ~) e

L

N

2.1.2. Development o f electrical excitability

The number, type and topographical distibution of ion channels and neurotransmitter receptors in the motoneurone membrane undergo developmental changes that will contribute to determine the firing pattern of the motoneurone. Most of the information available on the development of ion channels and receptor molecules is derived from studies of acutely dissociated motoneurones from chick embryos which continued their development in tissue culture and were examined with whole cell patch clamp techniques (O'Brien and Fisehbach, 1986a-d; Fruns et aL, 1987; McCobb et aL, 1989, 1990). It remains to be shown whether results obtained with this technique represent an accurate picture of developmental events occurring in vivo. The recently developed methods for whole cell patch clamp recordings from identified motoneurones in spinal cord slices (Takahashi, 1990; Manabe et aL, 1991) or isolated spinal cords of the chick embryo (Sernagor and O'Donovan, 1991) may be more appropriate. In motoneurones from acutely dissociated embryonic day 4 (E4) chick embryos, i.e. shortly after axons reach the undifferentiated limb bud, the somatodendritic action potential was generated by inward currents carried by both Na + and Ca + + ions (McCobb et aL, 1989, 1990). With development the magnitude of the Na + currents increased and as a consequence of this the amplitude, overshoot and rate of rise of the action potential increased. Thus, Na + channels are present in the soma and axon of the motoneurone from the time of onset of electrical excitability and their numbers increase with age (Ziskind-Conhaim, 1988; McCobb et al., 1990). JPN 41! l ~

|

][--

o 2

I

I

I

I--ITI

4

6

8

I0

12

(b)

!-

90

7O

IA

!-

6O

40

/

20

|o

I

~

I

I

I

I

2

4

6

8

~

12

Embryonic oge (Doys)

FIG. I. Developmental changes of Ca ++ (A) and K + (B) currents in embryonic chick motoneurones are plotted against age. Mean values (SEM) of current densities (pA/pF) measured in motoneurones after I d in culture isolated at EA, 6 and II are shown. (A) Ca ++ currents through T, N and L type channels. (B) Sustained K + currents (IK) and transient K + currents (I^). Reproduced from McCobb et al. (1989, 1990)with kind permission from the authors, Cell Press and Journal of Neuroscience.

96

R. NAVARRETEand G. Vgsovk

Takahashi, 1983; Walton and Fulton, 1986; Walton and Llinfis, 1986; Berger and Takahashi, 1990). The high threshold conductance is responsible for the afterdepolarization (ADP) seen during the falling phase of the action potential (Walton and Fulton, 1986). Low threshold Ca ÷+ conductances are particularly prominent during the first few days after birth and are associated with spontaneous oscillations o f the membrane potential (Walton and Llinfis, 1986). Low threshold Ca + ÷ conductances are also prominent in other neuronal types during early development (Carbone and Lux, 1987; Yaari et al., 1987; Fedulova et al., 1991). It has been proposed that the presence of low threshold Ca ÷+ conductances in developing neurones contributes to the generation of spontaneous autorhythmic activity characteristic of developing neuronal networks (Llinfis, 1984, 1988). Another role of the low threshold Ca + + conductance may be in the generation of high frequency "doublet" or burst firing as observed in neonatal rat motoneurones in vitro (Navarrete and Walton, 1989; Viana et al., 1991). This pattern of firing was commonly observed during the first 5-6 days after birth, but not in older animals (7-12 days old). "Doublet" firing was insensitive to Na ÷ channel blockade but was eliminated by removing Ca ÷ ÷ from the extracellular medium or by addition of inorganic Ca +÷ blockers (Navarrete and Walton, 1989). The presence of "doublet" or burst firing in young motoneurones is consistent with the predominance of phasic E M G activity pattern observed in neonatal muscles in vivo (Navarrete and Vrbovfi, 1983) and may be relevant to the functional matching of motoneurone firing patterns to muscle contractile properties in immature animals. A number of voltage dependent and neurotransmitter gated K ÷ currents have been described in adult neurones including motoneurones. This diversity of K ÷ channels provide for a rich control of neuronal excitability and firing patterns (see Schwindt and Crill, 1984; Rudy, 1988; Kiehn, 1991). Developmental changes of these outward K ÷ currents are likely to contribute to the changes of motoneurone firing patterns during maturation. The classical voltage dependent sustained K ÷ current (I~ or delayed rectifier), which is mainly responsible for the repolarization of the action potential, is already present at relatively high levels in very young (E4) chick embryo motoneurones and undergoes a moderate increase with development (McCobb et al., 1990). In contrast, the density of the transient (IA) K ÷ current is very low at E4 and increased markedly during development (McCobb et M., 1990), accounting for the shortening in the duration of the action potential with age. Figure 1 summarizes this and allows the comparison of changes in K + and Ca ÷+ conductances with age. In neonatal rat motoneurones both IK and IA types of K ÷ currents are present. In addition, Ca ÷ ÷ dependent K ÷ currents are also seen (Takahashi, 1990). These Ca ÷÷ dependent K ÷ currents largely determine the magnitude and duration of the afterhyperpolarization (AHP) of the action potential, an electrophysiological parameter that in adults distinguishes motoneurones to slow and fast muscles. In addition this current may also contribute to the repolarization of the action potential (Takabashi,

1990; Manabe et al., 1991). Whether changes of these Ca + + dependent K + currents are associated with the duration of A H P in developing animals is presently unclear (Manabe et aL, 1991). Synaptic activation of motoneurones is mediated by neurotransmitter gated ion channels located in their somatodendritic membrane. There is little information as to the time course of development of the membrane chemosensitivity to the various neurotransmitters released by afferent inputs to motoneurones. Best studied are the responses to excitatory amino acids which are known to he released by primary afferent terminals and interneurones (Jahr and Yoshioka, 1986; Lee et al., 1988; ZiskindConhaim, 1990). The acquisition of chemosensitivity to excitatory amino acid neurotransmitters was studied in tissue cultures composed either exclusively of motoneurones or in mixed spinal cord cell populations obtained from E6 chick embryos (O'Brien and Fischbach, 1986a, b). The sensitivity to glutamate was apparently modulated by the presence of spinal interneurones since it was greater in cultures that contained several cell types than in those composed exclusively of motoneurones (O'Brien and Fischbach, 1986c). The response to glutamate was not homogeneous along the somatodendritic surface of individual neurones. Areas of high sensitivity ("hot spots,') alternated with areas of low sensitivity (see Fig. 2). This arrangement is analogous to the distribution of AChR on adult skeletal muscle fibres, where at the synaptic site the density of AChR is much higher than outside it (see Section 2.2.3). In summary, our knowledge of developmental changes of ion channels and neurotransmitter receptors in the somatodendritic membrane of the motoneurone has expanded rapidly in the past few years. Results from these studies indicate that, in general, inward Na ÷ and Ca ÷ ÷ currents predominate in embryonic motoneurones while outward K + currents are relatively small at this stage and increase rapidly during late embryonic and early postnatal development. As a consequence of this, the duration of the action potential in embryonic motoneurones is much longer than in mature motoneurones. The long duration of the action potential in immature motoneurones may, in turn, result in greater Ca + ÷ entry via voltage-dependent Ca ÷ ÷ channels. Furthermore, because of the central role of Ca ÷ ÷ as an intracellular second messenger these activity-related events may influence the growth and differentiation of the motoneurone. In addition, these changes in K ÷ and Ca ÷ ÷ currents are also likely to contribute to the developmental modifications of motoneurone firing patterns and thus affect muscle fibre differentiation 2.1.3. Development o f functional specialization In adults, motoneurones are differentiated with respect to their intrinsic electrophysioiogical properties and recruitment order (see reviews by Burke, 1981; Kernell, 1990). Motoneurones to slow muscles have a much longer AHP and are recruited more readily than those supplying fast muscles (Eccles et aL, 1958). Studies of electrophysiological properties of unidentified motoneurones in foetal a n d newborn kittens (Naka, 1964; Kellerth et al., 1971) and

FIo. 2. Glutamate chemosensitivity in motoneurones from chick embryos. Example* of glutamate currents evoked at several sites on the soma and major processes of a sorted motoneurone (A) and an unsorted motoneurone grown with other spinal cord cells (B). The sensitivity declines steadily with distance from the sorted motoneurone soma, but it remains high along the process of the unsorted motoneurone. Note the 2 hot spots (E and H) along the latter process. Both cells were examined 6 d after plating. Calibration bars: 20 pA, 100 msec. Reproduced from O'Brien et al. (1986d) with kind permission from the authors and Oxford University Press.

4mv[ t

Hi

!

FIG. 6. Histological and electrophysiological illustration of the pattern of innervation of endplates of rat soleus muscle fibres during postnatal development. On the left side, cholinesterase silver stained longitudinal sections from a muscle of (a) a 9-day-old rat (note three axons contacting the endplate) and (b) a 14-day-old rat (only one axon in contact with endplate) are shown. On the fight side, records of endplate potentials from soleus muscle fibres innervated by three axons (three stepwise increments) taken from a 9-day-old rat (top) and mononeuronal innervation (no stepwise increment) taken from a 14-day-old rat (bottom) are shown. Reproduced from Vrbovfi (1987) with kind permission from Manchester University Press.

ACTIVITY-DEPENDENTINTERAC'rIONSBETWEENMOTONEURONESANDMUSCLES neonatal rats (Fulton and Walton, 1986) indicate that their intrinsic properties undergo changes with age that lead to their differentiation into functional subclasses. Unlike in adults, in young kittens (2-3 weeks of age), the duration of AHP was short in motoneurones to both future "slow" and "fast" muscles (Huizar et al., 1975). With further development, the AHP duration increased in motoneurones to slow muscles while it remained short in motoneurones to fast muscles. Similar results were reported by Hammerberg and Kellerth (1975). More recently, the properties of functionally identified motoneurones of the developing rat were studied using an /n vitro hemisected spinal cord-hindlimb preparation. Analysis of the input resistance and action potential characteristics indicated that presumptive "fast" flexor motoneurones (TA/EDL) could be separated from predominantly "slow" extensor motoneurones (SOL/LG) on the basis of these parameters. There were however significant postnatal changes in these parameters which resulted in further diversification of the two cell populations (Navarrete et al., 1988; R. Navarrete and K. Walton, in preparation). These results therefore suggest that, in rats, phenotypic differences between motoneurones to slow and fast muscles may already be present at birth. Such differences could be due to a systematic variation in the number, type and/or distribution of ion channels in the somato-dendritic membrane, as well as the geometry of the cells. Indeed, recent studies using whole cell patch clamp recordings in neonatal spinal cord slices have disclosed the existence of significant differences in the magnitude of transient K + (IA) currents between sciatic motoneurones and the hormone-sensitive bulbocavernosus motoneurones. No such differences were found between these two motoneurone sub-populations with respect to the sustained Ca+ +-dependent K + (IAHP) current which is considered to be responsible for the generation of the spike AHP (Manabe et al., 1991). One may expect that further studies applying these techniques to developing "slow" and "fast" motoneurones may reveal quantitative differences in ion channels between these sub-populations. The control of motoneurone firing patterns depends not only on their intrinsic electrophysiological properties but also on the source and distribution of the synaptic input that activates them. There is evidence for continued reorganization of synaptic inputs to motoneurones during the first few weeks of postnatal development in mammals (Conradi and Ronnevi, 1975; Ronnevi, 1979; Gilbert and Stelzner, 1979; see Stelzner, 1982). In particular, brainstem descending noradrenergic and serotonergic pathways undergo developmental changes in the distribution and transmitter synthesis of their terminals in the spinal cord (Loizou, 1972; Commissiong, 1983; Aramant et ai., 1986; Rajaofetra et al., 1989). The period of maturation of these noradrenergic and serotonergic inputs in the lumbosacral cord corresponds to the onset of antigravity functions and locomotor activity of the hindlimbs when postural tonic EMG activity can first be recorded from the slow extensor soleus muscle (Navarretc and Vrbov/t, 1983). Thus, developmental changes in synaptic inputs to motoneurones have to be considered in order to account for the

99

differentiation in motoneurone firing patterns. It may indeed be that changes in afferent synaptic activity associated with the maturation of descending and interneuronal inputs to motoneurones may regulate the functional specialization of the motoneurone 2.1.4. Target-dependence o f motoneurones

Early studies on the embryonic interactions between motoneurones and their target muscles suggested that a quantitative relationship exists between the number of motoneurones in the spinal cord and the amount of muscle tissue available for innervation (Shorey, 1909; Hamburger, 1934, 1958). Hamburger (1934) removed a limb bud in 2½-day-old chick embryos and subsequently found a decrease in the number of motoneurones in the lateral motor column on the side of the missing limb bud. This result was the first to suggest that there is a retrograde influence of the target on embryonic motoneurones. It was originally thought that the proportion of surviving motoneurones roughly corresponds to the amount of target available (Hamburger, 1977). However not all studies agree with this conclusion. A substantial number of motoneurones still die even in what appear to be favourable conditions of target availability, while it is possible for an excess number of motoneurones to be supported by relatively little muscle tissue (for review see Oppenheim, 1991). Nevertheless, in spite of these quantitative discrepancies there is substantial evidence to show that some kind of interaction with the target is important for regulating motoneurone survival. It has been suggested that motoneurone survival is dependent on a t r o p h i c factor released from the muscle. Some support for the presence of trophic factors in muscle have come from studies in which muscle extracts in various stages of purification have been shown to improve the survival of embryonic motoneurones both in vivo and/n vitro (Oppenheim, 1991). The idea of a specific substance present in the target that regulates the number of motoneurones which survive is conceptually simple and therefore attractive. However, the mechanism of action of such a substance on motoneurones is not yet known, and may well be different from that of nerve growth factor which acts on sensory and sympathetic neurones (see Purees and Lichtman, 1985). An alternative to the protective effects of targetderived neurotrophic molecules is the possibility that, as a result of its interaction with the target, the phenotype of the developing motoneurone changes so that it becomes less target dependent. One part of the motoneurone where radical changes are seen to occur as a result of contact with the target muscle is the nerve terminal, which is transformed from a growing into a secreting structure. A typical feature of this transition is a rapid increase in transmitter release from the growth cone and a cessation of neurite elongation (Xie and Poo, 1986). Such transformation in the growth status of the terminal could be induced by simple biophysical changes in the microenvironment of the growth cone as result of the response of the muscle cells to the released transmitter. The fact that many terminals of a single motoneurone undergo this change, may stimulate the cell body to increase

100 A

R. NAVARRETEand G. VRBOV~. Early Oevelo~nent

Late development

neurones, and then indirectly the size of the future motor unit.

syna~otic

,o#'~ '%

2.2. MUSCLEDEVELOPMENT 2.2.1. Emergence of fibre type diversily

ii

intar~tlorl with target

B

4b.

i

withtarget

Fro. 3. A diagrammatic representation of possible mechanism for postnatal motoneurone death. As a result of interaction between the rnotoneurone and muscle at the neuromuscular junction, a retrograde signal is sent to the cell body. This signal may be atrophic factor or a phenotypic change in motoneurone metabolism resulting from events associated with neurotransmission. (A) In the presence of the retrograde signal, the motoneurone develops normally and is able to withstand increased synaptic activation. (B) If interaction with the target is prevented, whether by axotomy or by transmission block, the retrograde signal is diminished and the target-deprived motoneurone is less able to cope with increased activation and eventually degenerates. Reproduced from Lowrie and Vrbov~ (1992) with kind permission from Elsevier Science Publishers.

the synthesis of the enzyme responsible for the production of ACh, and reduce the synthesis of growthassociated proteins. The cell is thus transformed into a transmitting neurone, and in this way the target muscle induces the motoneurone to modify some features of its phenotype. It could be that associated with this change are other modifications of the cell's phenotype which allow the motoneurone to survive. It is possible that the motoneurone in order to survive has to become competent to withstand the everincreasing afferent input that ovcurs in the developing spinal cord during the period when motor function matures. The interaction with the target may be required for the motoneurone to withstand these functional demands (Lowrie and VrbovL 1992). Figure 3 diagrammatically illustrates this proposal. Whatever the nature of the influence o f the target, the existing evidence clearly shows that the developing motoneurone depends for its survival on contact with the target muscle. Thus it could be that the target regulates the number of surviving moto-

The existence of different types of muscle fibres in adult vertebrates is well established and has been reviewed recently (Pette and Staron, 1990; Pette and Vrbov~t, 1992). Different types of muscle fibres can be distinguished using several criteria. These are: (a) histochemical evaluation of the activities of enzymes related to the energy metabolism of the muscle fibre, (b) histochemical identification of the myosin ATPase, and immunocytochemical distinction of myosin heavy chain (MHC) isoforms, (c) the analyses of single fibres and the precise identification of quantitative differences between their various intracellular components. These criteria allow the distinction of various muscle fibre types based on differences of isozymic forms of myosin, and other muscle specific proteins. These approaches taken together revealed a much greater diversity of muscle fibre types than hitherto expected. The demonstration of greater heterogeneity of muscle fibres indicates that the concept of a motor unit composed of completely identical muscle fibre types as proposed by Edstr6m and Kugelberg (1968) and Nemeth el al. (1981) is unlikely to be true, Indeed, several recent results showed that not all muscle fibres belonging to the same motor unit are identical and that the size, as well as the precise activity of oxidative enzymes, varies among different fibres belonging to the same motor unit (Albani el al., 1988; Martin et al., 1988). Thus, some degree of heterogeneity is present within the adult motor unit. This is an important point because it illustrates that the motor nerve is not the only factor that determines muscle fibre properties and that the same axon can successfully interact with a varied population of muscle fibres. Interestingly, not only in adults, but also during early development, before distinct motor units are formed, there is heterogeneity of muscle fibres. Its existence was first demonstrated by Butler el al. (1982) and by Phillips and Bennett (1984) for the embryonic chick wing musculature. Their results showed that the appearance of muscle fibres containing different isoforms of MHC can proceed even in the absence of innervation. Similar results were obtained on developing rat muscles by Condon el al. (1990). These results led to a number of proposals regarding the regulation of muscle fibre heterogeneity during early development, and the role of this early heterogeneity in the formation of distinct motor units. The heterogeneity of developing muscles is based predominantly on observations of MHC isoforms in muscle fibres. Early in the development of the rat limb bud the majority of primary myotubes react with an antibody to the adult slow MHC, while other fibres react only to an antibody raised against embryonic MHC. Many of the primary myotubes contain both types of myosin, and the situation gets more complicated when, during further development, new

ACTIVITY-DEPENDENTINTERACTIONSBETWEEN MOTONEURONESAND MUSCLES

muscle fibres are formed (Condon et al., 1990; Sheard et al., 1991). Many investigators imply that the muscle fibres which contain only slow MHC are the ones destined to become the adult slow fibres. Since the formation and distribution of these fibres is independent of innervation, they suggest that muscle fibre types are predetermined and different types are derived from separate clones (Miller and Stockdale, 1987; Condonet al., 1990; Miller, 1991). Recently the sequence of induction of different MHC isoforms in developing muscles was studied at the mRNA level (Lyons et ai., 1990). These studies show that during the initial stages of muscle development the sequence of expression of the myosin gene in mouse somites proceeds in a preprogrammed fashion in which the mRNAs coding for cardiac MHC is replaced by that coding for embryonic MHC, subsequently followed by mRNA for perinatal MHC. This sequence is apparently similar in all myogenic cells. Only later, in the mouse, at EIS, can differences in the expression of myosin genes between developing muscle fibres be seen (Lyons et al., 1990). Recently the proposal that muscle fibre types are predetermined by their lineage has been directly put to the test under more physiological conditons of postnatal rat muscle development in situ. In these experiments myoblast clones were labelled by retroviral infection and the fate of their progeny was studied using a number of monoclonal antibodies to MHC isoforms (Hughes and Blau, 1992). Results from this study indicate that single clones of myoblasts fused with muscle fibres containing different combinations of MHC isoforms. Thus, the progeny of an individual myoblast appears to contribute randomly to different muscle fibre types (Hughes and Blau, 1992). These experiments are important because they indicate that while myoblasts are clearly capable of expressing intrinsic programmes of differentiation if isolated from influences of the environment they would normally encounter in situ (see Miller, 1991), this program can be overriden after fusion with pre-existing muscle fibres. The extent to which this plasticity of muscle phenotype is also present at earlier stages of development, at the time when environmental signals such as innervation do not appear to play as important a role, is still open to experimental analysis. From these results it appears that all myogenic cells undergo a preprogrammed developmental sequence and differentiate only after its completion. This findings are not compatible with the proposal of predetermined clones. Different isozymic forms of other components of the myofibril, such as troponin, have also been described. Different isozymic forms of troponin T are expressed in rat muscle during the embryonic than during the postnatal period (Sabry and Dhoot, 1991) but the regulation of this protein has not been explored to the same extent as that of myosin. Developmental changes of enzymes associated with energy metabolism have also been studied. These studies show that there is a difference between future slow and fast muscles in the chick, some time before any functional differences with regard to their contractile properties can be seen. In the adult chicken, the anterior latissimus dorsi (ALD) muscle is composed of slow tonic fibres and the posterior latissimus

101

dorsi (PLD) muscle of fast twitch fibres. These 2 muscles differ in the adult and during development in a number of ways (see Vrbovfi et aL, 1978). For instance, in the 15-day-old chick embryo, the ALD muscle has much higher levels of oxidative e ~ e s than the fast PLD muscle which contains higher levels of phosphorylase, an enzyme associated with glycogen breakdown (Gordon et al., 1977b). The isozymic forms of sarcomeric proteins, i.e. myosin, actin and the troponin complex, are often thought to be associated with the contractile properties of the muscle fibre. This association was inspired by the findings of Bfir/my (1967) and B/rainy and Close (1971), who demonstrated that the speed of force development was correlated with the ATP-ase activity of the myosin molecule. Since the myosin ATP-ase is associated with the MHC, the different isozymic forms of MHC may confer on to the muscle fibres a particular contractile speed. Unlike in adults, where there is a correlation between different sarcomeric proteins and contractile speeds, in developing muscles such correlations have not been demonstrated. During early embryonic life all chick muscles are slow contracting, no matter whether they will later become slow or fast contracting (Gordon et aL, 1975). In kittens and rats too, both future slow and fast muscles are initially slow contracting and relaxing (Buller et al., 1960; Close, 1964). In rats, the increase in contraction speed seen during development correlates better with the development of the cytosolic Ca + + buffering protein parvalbumin (Leberer and Pette, 1986) than the changes of sarcomeric proteins. Denervation of neonatal muscles or paralysis of embryonic muscles reduces the rate at which the muscle speed increases, but does not altogether prevent this increase (Brown, 1973; Lewis, 1973), thus indicating that innervation-independent processes can bring about some degree of maturation of the muscle cell. However, complete differentiation into slow and fast muscles does not occur in either denervated or inactivated muscles (Brown, 1973; Buller et al., 1960; Gutmann et ai., 1974; Lewis, 1973). 2.2.2. Mechanisms involved in generating fibre type diversity

The observations of early diversification of muscle fibres, which is independent of innervation, led to the proposal that myoblasts are intrinsically different, and those populations originating from a particular clone give rise to a particular muscle fibre type. These ideas originated from/n vitro experiments in which those myotubes derived from a single cloned myoblast ancestor isolated from chick embryos at a certain stage of its development, gave rise to myotubes of a similar phenotype. Myotubes derived from ancestors isolated later in development differed from those derived earlier, thus suggesting a diversity of the original ancestral myoblasts (Rutz et al., 1982). These experiments were further extended to show that it is possible to isolate three distinct "colonies" of myotubes each derived from a single myoblast lineage: (i) one that expresses only the fast MHC isoforms (70%), (ii) another colony that expresses both fast and slow MHC (30%) and (iii) in very few instances 0 - 2 % ) a colony where each myotube

102

R. NAVARRETEand G. VRBOV,~

expresses only slow MHC. The proposal of a predetermined fate of muscle fibres was largely based on these in vitro findings (Miller and Stockdale, 1987). To which extent/n vitro results are representative of events occurring during in vivo development is not certain, and while these findings are interesting, muscle development in the whole animal may be different (see Section 2.1.1). The fact that diversification of the isozymic forms of myosin in embryonic muscles occurs in the absence of innervation (Butler et al., 1982; Phillips and Bennett, 1984) may be taken to support the idea that muscle fibre types are predetermined. Nevertheless this initial independence of innervation does not necessarily mean that the development of early myogenic cells is preprogrammed. For instance, mechanical factors related to the stretch imposed on particular sets of muscle fibres may influence their gene expression. Differences in the anatomical position of particular myotubes at a given developmental stage and their involvement in force production are also likely to play a role in inducing particular isoforms of MHC. Thus myotubes that span from tendon to tendon would participate in movement in a mechanically different manner from those that are short and do not make a substantial contribution to force production. Indeed, studies on the lumbrical muscles have revealed that primary myotubes which usually contain slow myosin span from tendon to tendon, whereas secondary myotubes do not (Harris et al., 1989). This finding taken together with the observation that stretch can influence the development of muscle fibres in vitro (Vandenburgh et al., 1990) and exerts an important influence upon them suggests that it may contribute to the development of a heterogeneous population of myotubes. Consistent with such a proposal are results showing that neonatal muscles are influenced by mechanical conditions. In the soleus muscle of the rat the number of muscle fibres containing slow MHC increases after birth, as the animal is beginning to use the muscle for postural support. If the supporting function of this muscle is interfered with, the number of muscle fibres with slow MHC fails to increase (Lowrie et al., 1989). Thus whatever the origin of the muscle fibres, environmental factors have a strong influence on the expression of a particular phenotype. The importance of epigenetic influences is also clear from results of experiments on reinnervated muscles following nerve crush in the perinatal period. In reinnervated rat TA and EDL muscles, the distribution of slow MHC containing fibres was different from that seen in normal animals. Such a situation could be explained if on reinnervation axons of "slow" motor units colonized muscle fibres that were destined to become fast (Lowrie et al., 1988). These results on reinnervated neonatal fast muscles are consistent with those of Soha et al. (1989) on the neonatal rabbit soleus muscle, where they also found non-selective reinnervation of muscle fibres. However, they differ from those of Soileau et al. (1987) who claimed that in the rat soleus after neonatal nerve injury, there is no rearrangement of muscle fibre types because the reinnervating axons return to the muscle fibres they originally supplied. The discrepancy between these results is most likely due to

the fact that Soileau et al., (1987) examined their muscles 2-3 weeks after the nerve injury, a time interval which is too short to allow the completion in myosin transition. This was shown by Lowrie et al. (1988) who studied the muscles at different time intervals following recovery from the nerve injury. Initially no rearrangement of muscle fibre types was apparent, and for the rearrangement of fibre types to be complete intervals of 6-8 weeks were necessary (Lowrie et al., 1988). These results taken together show that (a) in neonatal animals axons from slow motor units can make connections with muscle fibres that would otherwise be fast, and (b) that the muscle fibres destined to become fast or slow have the ability to express different types of myosin isoforms. Thus, muscle fibre type of developing animals can be regulated by external factors such as neuronal activity or mechanical conditions of the growing muscle cell. 2.2.3. Acetylcholine receptors and acetylcholinesterase Apart from intraceUular muscle specific proteins of which the best explored ones are the components of the myofibril, some membrane-associated proteins are also muscle specific and unique for skeletal muscle fibres. Molecules that are part of the surface membrane could also contribute to the establishment o f specific connections between motor nerves and muscles and are therefore of special interest to neurobiologists. Of these, the acetylcholine receptor (AChR) is of particular importance because it is the molecule that enables the muscle to become functionally connected to its innervation and respond to the acetylcholine (ACh) released from the motor nerve ending. In view of the unique position of this molecule it is not surprising that much research has been carried out to reveal its structure, function, and the developmental regulation of its synthesis and distribution along skeletal muscle fibres. Excellent reviews have recently been published on the development of the AChR (Schuetze and Rote, 1987; Changeux, 1991). The AChR is a pentamer comprised of 4 distinct subunits with the stoichiometry of ct (2), ~ (1), or E (1) and ~ (1) (see Schuetze and Role, 1987). The receptor complex has two binding sites for ACh, both on the a subunits. Each of the four subunits of the AChR is translated from a separate mRNA, inserted into the endoplasmic reticulum, glycosylated and assembled (Salpeter and Loring, 1985; Anderson, 1986; Merlie and Smith, 1986). A few hours after completion of their synthesis the assembled AChR molecules are inserted into the membrane (see Fambrough, 1979). AChRs incorporated into the surface membrane can later be either stabilized and become "junctional AChRs" or as is the case with extrajunctional AChRs, become internalized and degraded by lysosomal enzymes. AChRs outside the endplate region and at immature endplates have a metabolic half life of about 1 day which increases 10fold at mature endplates (Fambrough, 1979). The appearance of the AChR in the membrane of mononucleated myoblasts is one of the first signs of their commitment to a myogenic lineage, and precedes myoblast fusion. The presence of AChRs in myoblasts and their sharp increase after fusion has

ACTIVITY-DEPENDENTINTERACnONSBETWEENMOTONEURONESANDMUSCLES

FIt. 4. Schematicrepresentation of a muscle fibre close to its place of contact with its motor nerve. The thick line on the musclefibre illustrates the accumulation of acetylcholine receptors (AChR) at the place of nerve-muscle contact. The action potentials outside this site down-regulate the AChR synthesis by extra-junctional nuclei, possiblyvia increasesof intracellular Ca ++. Calcitonin gene-related peptide (CGRP) released from the nerve may play a role in maintaining the AChRs at the site of nerve-muscle contact. been demonstrated in vitro in the absence of innervation (Fambrough and Rash, 197 I; Sytkowski et al., 1973; Dryden et al., 1974; see Fambrough, 1979). After fusion and further development of the myotube, AChRs form clusters. Cluster formation can be induced by various stimuli applied to the surface of muscle fibres both in adults (Jones and VrbovL 1974) and in developing muscles (Peng et al., 1981). Although the clusters become associated with the area of contact with the motor nerve, they are also found in aneural areas of the muscle fibre and develop in the absence of innervation. However, following the establishment of neuromuscular contacts the only clusters to remain on the muscle fibre are at the site of these contacts. Those AChRs that are localized outside the region of the neuromuscular contact disappear. During development the disappearance of the extrajunctional AChRs is brought about by muscle activity (see Fig. 4). If, in chick embryos, activity is prevented by temporary paralysis of the muscle, then the reduction of ACh sensitivity, i.e. the reduction in AChRs outside the endplate region usually seen during development, fails to occur (Gordon et al., 1974; Burden, 1977). Thus nerve induced activity plays a crucial role in regulating the distribution of AChRs. Even in adult muscle fibres, AChRs at extrajunctional sites can reappear after denervation and following complete paralysis of the muscle. Although both during development and in adults activity is an important factor that controls the distribution of AChRs, there are other factors, such as contact phenomena or putative substances released by them, e.g. calcitonin gene related peptide (CGRP), that may play an important regulatory role (see Changeux, 1991). In developing muscle fibres AChR clusters found at newly formed synapses are easily disrupted by removal of extracellular Ca ++, chronic carbachol exposure, and elevated external K + (Bloch and Steinbach, 1981). In contrast, mature synaptic contacts are resistant to these treatments. Similarly, AChR clusters at developing synapses disperse within hours or days after denervation (Slater, 1982; Kuromi and Kidokoro, 1984), but clusters at mature endplates remain intact for 2 weeks or more after removal of the nerve (Frank et al., 1975; Ko et al., 1977;

103

Braithwaite and Harris, 1979; Slater, 1982; Labovitz and Robbins, 1983). Thus, continued action of the nerve is required for cluster maintenance at newlyformed endplates but not at mature contacts. Presumably other factors maintain the AChR distribution in the absence of the nerve at the adult neuromuscular junction (Slater, 1982). It could be that during maturation an extracellular matrix material such as synapse-specific basal lamina or specific molecules such as "agrin" are acquired and these play a role in the stabilization of the AChR (reviewed by Sanes, 1989). Alternatively the AChRs may become anchored to the membrane by cytoskeletal elements (reviewed in Froehner, 1986). Not only does the distribution of muscle AChRs change during development but also their functional properties become transformed. The existence of different types of AChR was first described on denervated adult muscles, where AChR associated channels at the endplate region were found to be different from those that appeared extrajunctionally. When activated by ACh, endplate channels conduct larger currents but close more quickly than extrajunctional channels (Neher and Sakmann, 1976; Sakmann, 1978). During endplate maturation AChR channel properties change from "extrajunctional-like" to "endplate-like'. AChR channels in immature endplates have a longer opening time than at mature endplates. Interestingly AChRs in extrajunctional regions of developing muscle fibres change in parallel with those at endplates. Therefore, the common practice of describing AChR channels as either "extrajunctional type" or "junctional type" is misleading. Rat muscle fibres can express at least two types of AChR channels, an embryonic type associated with a long opening time and an adult one with a short opening time (Sakmann and Brenner, 1978). It is now known that the channel opening time of the AChR is associated with one of its subunits. The AChR with fast channel characteristics contain the ~ subunit, and those with slow channel properties the ~, subunit (for review see Schuetze and Role, 1987). The proportion of slow embryonic channels at rat endplates decreases during early postnatal life (see Schuetze and Role, 1987). In the developing soleus endplates the fraction of slow channels decreases steadily from virtually 100% around birth to less than 20% three weeks later (Sakmann and Brenner, 1978; Fischbach and Schuetze, 1980; Michler and Sakmann, 1980). During this period there is a corresponding increase of mRNA coding for the ~ subunit and a reciprocal loss of that coding for the 7 subunit of the AChR receptor (Witzeman et al., 1989; Brenner et al., 1990). The synthesis of AChRs is also regulated by the nerve and its activity (reviewed by Changeux, 1991). Shortly alter birth when slow channels predominate at the endplate, denervation blocks the normal developmental appearance of fast channels. Studies of endplate formation in adults (Brenner and Sakmann, 1983) suggest that denervation exerts these effects via the loss of muscle activity rather than the absence of a specific neurotrophic factor. Apparently neither the nerve nor muscle activity is required for the continued expression of fast channels

104

R. NAVARRETEand G. VR~3VA

in adult endplates. When adult rat muscles are denervated, at least half of the channels at former endplates are fast, even 18-20 days after denervation (Brenner and Sakmann, 1983), although most of these receptors presumably were replaced at least once (cf. Levitt and Salpeter, 1981). The combined results of the denervation studies on neonatal and adult muscles suggest that the appearance of fast AChR channels requires muscle activity, but not necessarily the continued presence of the nerve. The possible mechanisms by which activity induces the alteration of the AChR molecule have been recently reviewed (Changeux, 1991). Figure 4 illustrates schematically the proposed regulation of AChR synthesis in the junctional and extrajunctional membrane. Another important step in endplate development and specialization is the accumulation of the enzyme acetylcholinesterase (ACHE). This enzyme appears in myogenic cells very early in development, well before ventral root axons reach the myotome. The areas of AChE activity on myogenic cells increase in the vicinity of growth cones that invade the myotome (Tennyson et al., 1971, 1973). Gradually, as innervation proceeds, the AChE accumulates at the place of contact between the motor nerve ending and muscle fibre (Kupfer and Koelle, 1951; Lentz, 1969; Trth and Karcsfi, 1979; Brzin et al., 1981). This accumulation of AChE is one of the indicators of the morphological differentiation of the subsynaptic sareolemma and muscle fibre (Kelly and Zacks, 1969; Bennett and Pettigrew, 1974; Brzin et al., 1981). In developing muscle fibres, patches of AChE on the sarcolemma outside the endplate disappear at the time when the enzyme accumulates at the endplate region. AChE in skeletal muscles can be present in two distinct molecular forms, globular and asymmetric (Hall, 1973; Brzin et al., 1983; Toutant and Massoulie, 1987). The asymmetric form is present in endplates of mature muscles (Hall, 1973; Sketelj and Brzin, 1985) and is dependent on innervation (Vigny et aL, 1976), for it becomes reduced after denervation. Innervation and activity are known to exert an important influence on the accumulation of AChE at the neuromuscular junction. Following neonatal denervation A C h E fails to accumulate at the neuromuscular junction (Zelenfi and Szentagothai, 1957). In embryonic development, even in muscles with intact innervation AChE does not accumulate at the place of contact between the motor nerve ending and nerve terminal in inactive muscles paralysed with curare (Gordon et aL, 1974). However, if more sensitive methods are used and denervation carried out at birth small amounts of AChE do accumulate at the site of former contact with the nerve (Sketelj et aL, 1991). This continued development of AChE on newly formed endplates is concomitant with the persistence of AChRs (Slater, 1982) and some degree of growth of synaptic folds (Brenner et al., 1983). Moreover it is regulated by the activity of the muscle fibre. It has been suggested that the contact with the nerve induces some "trace", which allows the endplate region of the muscle to continue to develop. There are several possibilities as to the nature of the "trace". The presence of a synapse organizing factor

in the basal lamina under the nerve endings has been proposed to be responsible for the continuous specialization of the endplate region (Nitkin el a/., 1987). Such a molecule called "'agrin", causes aggregation of AChRs in cultured, aneural myogenic cells (Fallon and Geifman, 1989). Interestingly, this substance is present in the motoneurone soma (MagillSole and McMahan, 1988). In summary, profound changes occur during the development of skeletal muscle fibres both in the expression of isoforms and distribution of the two major proteins that are involved in cholinergic transmission, i.e. the AChR and ACHE, and these changes depend, in part, on the activity of the muscle fibre. 2.2.4. Development o f ion channels and passive electrical properties

Developmental changes in membrane associated characteristics such as voltage gated ion channels and passive membrane properties of the muscle fibre have been described. Most studies have concentrated on ion channels of myoblasts and myotubes in tissue culture. In addition, developmental transitions in channel properties have been demonstrated in muscles developing in situ during postnatal development. Here we will concentrate on those studies which specifically address the rote of innervation, and neuromuscular activity in the regulation of ion channels of the muscle membrane. The early appearance of Na + currents was demonstrated in a rat myoblast cell line L6 (Kidokoro, 1975) and subsequently in myoblasts obtained from rat primary cultures (Gonoi et aL, 1985; Weiss and Horn, 1986). These studies showed that Na ÷ currents are initially resistant to pharmacological blockage by TTX. Although TTX-sensitive Na + currents are detectable in some myoblasts in the absence of innervation (Gonoi et al., 1985; Weiss and Horn, 1986) the motor nerve appears to play an important role in the developmental induction of TTX-sensitive Na + currents. In adult innervated mammalian muscle fibres the majority of Na + channels have a high affinity to TTX which completely blocks the generation of the action potential (Redfern and Thesleff, 1971; Harris and Thesleff, 1971). However, if adult muscles are denervated or paralysed by botulinum toxin (BoTx), a new type of Na + channel appears, making the action potential partly resistant to the action of TTX (Mathers and Thesleff, 1978; Bambrick and Gordon, 1987). In newborn rat muscles too, TTX-resistant action potentials are normally found in fully innervated muscle (Harris and Marshall, 1973). The TTXinsensitive Na + channels present in denervated muscle fibres and during early development have a different molecular structure and gating properties compared to those present in adult innervated muscle. TTX-resistant Na + channels open at more negative membrane potentials and have longer activation and inactivation decay times, and smaller single channel conductance compared to adult innervated muscle fibres (Gonoi et al., 1989; Kallen et al., 1990). During development the proportion of TTXresistant Na + channels decreases, and this decrease does not take place if the muscles are denervated or paralysed (Sherman and Caterall, 1984; Gonoi et al.,

105

ACTIVITY-DEPENDENTINTERACTIONSBETWEENMOTONEURONESANDMUSCLES 1989). This developmental transition from TTXinsensitive to TTX-sensitive Na + channels is similar to that previously discussed for the AChR. Experiments using radiolabelled Na ÷ channel antagonists such as TTX or saxitoxin (STX), which bind specifically to a receptor site on the extracellular face of the channel (see Caterall, 1991), show that the number of TTX-sensitive Na ÷ channels increases rapidly during the first three weeks of postnatal development in the rat (Sherman and Caterall, 1982; Lombet et al., 1983; Bambrick and Gordon, 1988). The most rapid incorporation of TTX-sensitive Na + channels into the skeletal muscle membrane occurred during the second postnatal week, coinciding with a period of increasing neuromuscular activity (see Section 4.2). This developmental increase can be inhibited by neonatal denervation (Sherman and Caterall, 1982; Gonoi et al., 1989) or if the muscle is paralysed with BoTx (Bambrick and Gordon, 1988). Thus, it appears that muscle activity induces a decrease in TTX-insensitive channels and an increase in TTX-sensitive Na ÷ channels. Voltage gated Ca ÷ ÷ channels are present at high levels in the T-tubular system of adult skeletal muscle. These high threshold (L-type) Ca ÷÷ channels are sensitive to the pharmacological action of dihydropyridines and mediate long lasting inward Ca +÷ currents in response to strong membrane depolarizations. During prenatal and early postnatal development both L- and T-type Ca ÷ ÷ channels are present in skeletal muscle fibres (Kano et al., 1987; Gonoi and Hagesawa, 1988). The presence of T-type Ca +÷ channels in immature muscle fibres is transient and it is replaced by the L-type Ca ÷÷ channels during postnatal development (Kano et al., 1989; Gonoi and Hasegawa, 1988). As discussed in Section 2.1.2, the presence of T-type Ca ÷+ channels appears to be a general property of developing excitable cells. Unlike L-type Ca ÷ ÷ channels, T-type Ca ÷ ÷ channels open at membrane potentials close to the resting potential, so that even small fluctuations of the membrane potential of the immature muscle fibre may trigger significant Ca ÷ ÷ entry via T-type Ca ÷÷ channels and thus influence a number of Ca ÷÷ activated processes related to muscle differentiation. During the first three postnatal weeks in the mouse, Ca ÷ ÷ currents carried by L-type channels increased four-fold, whereas those mediated by T-type channels decreased and became undetectable (Gonoi and Hasegawa, 1988). The postnatal increase in L-type Ca ÷÷ currents was prevented by neonatal denervation, whereas the decrease in T-type currents was unaffected by this procedure (Gonoi and Hasegawa, 1988). Thus, innervation dependent and independent factors appear to control the expression of different types of Ca + ÷ channels. Changes in expression of chloride channels also occur during muscle development and these are innervation-dependent (Conte-Camerino et al., 1989). In adult mammalian skeletal muscle fibres the chloride conductance (gC1-) contributes up to 80% of the total membrane conductance of the muscle fibres (Bretag, 1987). This high CI- conductance is not yet present in muscle fibres from newborn rats and increases with development. In chick fast twitch muscle fibre (but not in tonic ones) this increase

occurs during the last few days of embryonic develop ment (Poznansky and Steele, 1984). In rats during the first week after birth the gC1- is very low and increases sharply during the second postnatal week (Conte-Camerino et al., 1989). During the same period of development there is also an increase in transcription of the gene coding for the CI- channel in mouse muscles (Steinmayer et al., 1991). The postnatal increase in gCl- is critically dependent on neuromuscular activity or some other nerve-related factor operating during this period, for it fails to occur after neonatal denervation (Conte-Camerino et al., 1989). It is surprising that little attention has been given to developmental changes of passive membrane properties of muscle fibres since the amplitude and spread of the depolarization during the endplate potential depends largely on these properties. In adult avian muscle there are large differences in membrane resistance and time constant between slow tonic ALD muscle and the fast twitch PLD muscle (Fedde, 1969; Gordon et al., 1977a). These differences are not yet established in embryos at E14-16. The membrane resistance of PLD declined rapidly after El7 whereas that of ALD continued to increase until the second month after hatching (Gordon et al., 1977a). Similarly, the time and space constants were initially similar and changed in a different way in the two muscles during embryonic development. In the ALD both time and space constants increased with development, whereas in the PLD fibres the time and space constant became reduced during embryonic development (Gordon et al., 1977a). In PLD muscle fibres the input resistance which is an important factor that determines the size of the response to a given excitatory input, decreases sharply with age (Fig. 5). This indicates that the response of an immature muscle fibre to the same excitatory input will be greater than that of an adult muscle fibre. In conclusion, the developmental changes in ion channels and other membrane-associated proteins such as the AChR, AChE listed here may be of great importance for the differentiation of the muscle fibre for they will determine the size and duration of the response of the muscle fibre to the excitatory action

=E v

8

8

J1, _c J

4

15

I le

~ 1

i 111

I 111

i

i

,.j

Embryonicdays FIG. 5. Changes of input resistance of muscle fibres from chicken PLD are plotted against age. Hatching is indicated by the interrupted line, and one value after hatching is taken from a PLD muscle of a 42-day-old chicken. Data are taken from Gordon et al., (1977a).

106

R. NAVARRISTEand G. VRBOVA

of ACh released by the ingrowing nerve. As a result of the long channel opening time of the immature AChR channels and the high input resistance of the muscle fibre, even small amounts of ACh released from the growth cone or immature nerve terminal will evoke large synaptic potentials capable of initiating propagated action potentials (Jaramillo et al., 1988). These membrane characteristics might be involved in the matching of nerve terminals to particular myotubes (Section 5.1)

3. ESTABLISHMENT OF NEUROMUSCULAR CONNECTIONS AND THEIR REORGANIZATION DURING DEVELOPMENT There are several recent reviews which describe in detail the development of the neuromuscular junction both in situ and in tissue Culture (Salpeter and Loring, 1985; Peng, 1987; Changeux, 1991). Here we will specifically review those aspects of neuromuscular junction development that depend on pre- and/or postsynaptic activity and may contribute to the establishment of the adult motor unit topography. 3.1. CHARACTERISTICSOF EARLYNEUROMUSCULAR TRANSMISSION In the chick embryo, electrical stimulation of motor axons first elicits muscle contraction at stage 27-28 (Landmesser and Morris, 1975), at a time when most presynaptic axon profiles are not yet closely associated with AChR clusters in the muscle fibre (Dahm and Landmesser, 1991). The fact that initial neuromuscular transmission does not require synaptic specializations has also been demonstrated in tissue culture studies which show that synaptic potentials can be evoked within a few minutes after the growth cone contacts a receptive myotube (Frank and Fischbach, 1979; Xie and Poo, 1986). It appears that the growth cones can produce spontaneous and evoked ACh release even before the terminals form a physical contact with a myotube (Young and Poo, 1983; Hume et al., 1983). Without contact with a myogenic cell, growth cones release very little ACh, but within 20 minutes of contact with the muscle membrane both the amount and frequency of the spontaneous ACh release increased (Xie and Poo, 1986) suggesting that contact with the target may specifically induce the release of neurotransmitter. These initial ACh induced depolarizations of the muscle membrane resemble miniature endplate potentials (MEPPs) observed in the mature neuromuscular junction, in that their frequency is reduced by low Ca + + and they can be blocked by curare or BTX (Xie and Poo, 1986). However, in contrast to MEPPs in the mature neuromuscular junction their amplitude can be very large and in some cases trigger muscle contractions (Jaramillo et al., 1988). It is possible that during the initial stages of development of the neuromuscular junction the presence of sub-threshold spontaneous synaptic depolarizations constitute the first form of interaction

between motoneurones and muscle fibres. This initial form of interaction may be largely independent of action potentials in either nerve or muscle, but may nevertheless induce muscle contraction and differentiation. It is interesting that similar "giant" spontaneous MEPPs are observed in adult muscles following experimentally induced inactivity of the neuromuscular junction by chronic TTX blockade of action potentials in the nerve (Gundersen, 1990) or after botulinum toxin paralysis (see Thesleff and Molgo, 1983). Transmission at newly formed neuromuscular junctions is much less efficient than in the adult. Most endplate potentials (EPPs) recorded from embryonic muscles have a slow rise time and small amplitude which decreases rapidly in response to repetitive stimulation (Dennis et al., 1981; Pilaf et al., 1981; Sheard et al., 1991). Both the frequency of spontaneous MEPPs and the quantal content of the EPP increase markedly during embryonic and early postnatal development (Diamond and Miledi, 1962; Letinsky, 1974; Dennis et al., 1981; Kelly, 1978). This is associated with an increase in the amount of enzymes involved in ACh synthesis (O'Brien and Vrbovfi, 1978; Pilar et al., 1981). With development, the efficiency of neuromuscular transmission improves gradually as pre- and postsynaptic components of the neuromuscular junction become integrated. The developmental stage of the muscle fibre itself may influence its ability to respond to the transmitter. The size of the response may depend on the type and density of AChRs. In addition, the size of the muscle cell will influence its excitability by virtue of the fact that smaller cells have higher input resistance and are therefore more excitable. 3.2. REORGANIZATIONOF SYNAPTICCONTACTS DURING POSTNATALDEVELOPMENT Even after contacts between motoneurones and muscle fibres have been made, the emergence of the motor unit in its adult form is not yet complete. In adults each muscle fibre is supplied by a single axon (see Fig. 6b), and is thus activated by only one motoneurone. Each motoneurone in turn supplies many muscle fibres and the number of muscle fibres supplied by a single motoneurone is referred to as "the innervation ratio". During late embryonic and early postnatal life the situation is very different, in that individual muscle fibres are contacted by several axons. Figure 6 illustrates examples of electrophysiologically and histologically identified endplates from neonatal animals with polyneuronal innervation (Fig. 6a) and adult endplates with one axon to each muscle fibre (Fig. 61)) (Redfern, 1970). Thus in young animals each motoneurone supplies many more muscle fibres than in the adult (Bagust et al., 1973), i.e. it has a much higher innervation ratio. This point is schematically illustrated in Fig. 7. In this section the role of activity in the regulation of the innervation ratio will be discussed first, since it has a profound influence on the motor unit topography. The mechanism by which activity could exert this influence will be considered later.

ACTIVITY-DEPENDENT INTERACTIONS BETWEEN MOTONEUKONESAND MUSCLES

107

A

II

ij t I

/

J 0 D D

o 0

D FI~. 7. Schematic illustration of the formation of the mammalian motor unit. The spinal cord and two of its motonenrones contacting muscle fibres during different stages of embryonic and early postnatal development are represented, In (A) the terminals of the two embryonic motonenrones converge onto few primary myotubes. (B) The number of primary myotubes increases and secondary myotubes connected via gap junctions to the primary myotubes are shown (smaller size fibres). Terminals from the two motoneurones converge onto all the muscle fibres. (C) illustrates the situation in the adult, where each muscle fibre is contacted by a single nerve terminal from one motoneurone. 3.2.1. Activity and regulation o f neuromuscular contacts: relevance to motor unit size In the adult animal the distribution of motor units within a given muscle is such that the most active motor units have the smallest innervation ratio, while the largest motor units are usually least active. Consistent with this, are the results obtained by Callaway et al. (1987). These authors found that motor units whose activity was reduced during the neonatal period by blocking the action potential by TTX had larger territories than those whose axons remained active. This finding is at odds with results obtained in adult lumbrical muscles of the rat during reinnervation, where the territories of the inactive motor units were smaller than those of the active ones (Ribehester and Taxt, 1983, 1984). However, comparisons between these two sets of experiments are difficult, for they were done in one case on an adult fast muscle (Ribchester and Taxt, 1983, 1984), and in the other case on a neonatal rabbit slow muscle (Callaway et al., 1987). In contrast to results obtained when muscle activity is prevented by applying TTX to the nerve, treatment of soleus muscles of newborn rats with at-BTX which abolishes completely the response of the muscle to ACh, led to a considerable loss of synaptic contacts (Green~mith and Vrbov/t, 1991). The discrepancy between results from experiments where activity was reduced by TTX and those where it was entirely prevented by a-BTX could be explained in the following manner. Preventing impulse activity from reaching the muscle by applying TTX to the nerve may still allow the maturation of the neuromuscular junction.

It is known that the spontaneous release of transmitter is increased in axon terminals treated with TTX (Gunderson, 1990). Such spontaneously released transmitter may be sufficient to bring about the maturation of the developing neuromuscular junction and it could be that under these conditions synaptic contacts are maintained. However, postsynaptic differentiation is arrested when the AChR is blocked by curare or BTX (Gordon et al., 1974) and in this situation neuromuscular contacts are lost. These results indicate that the maturation of the postsynaptic membrane is critical for the maintenance of nerve-muscle contacts and therefore for the establishment of the innervation ratio of the motor unit. In view of this, the cellular mechanisms that are involved in the activity-regulated maintenance of contacts between axon terminals and muscle fibres will be of great importance for the development of the motor unit 3.2.2. Mechanisms involved in the regulation o f synaptic contacts In rat hindlimb muscles the emergence of the adult motor unit size achieved by elimination of polyneuronal innervation takes place during the second and third week of life when the amount of neuromuscular activity of the animal suddenly increases (Navarrete and Vrbov~, 1983). Indeed the elimination of polyneuronal innervation is critically dependent on neuromuscular activity. If neuromuscular activity is increased by chronically applied electrical stimulation, the rate of loss of synaptic contacts is

108

R. NAVARRETEand G. VRBOVA Physiology 10o

Histology

...............................................................................................................................

.c .c

~D

0

ACh

FIG. 8. Effect of ACh and Ca ++ on synapse elimination at the neuromuscular junction. The block diagram shows the reduction of polyneuronal innervation induced in vitro by ACh, and the influence of BAPTA-AM, Nifedipine or leupeptin on this ACh induced loss of neuromuscular contacts, The results are expressed as a percentage change where the untreated contralateral muscles are taken as 100%. Electrophysiology is based on data from Zhu and Vrbov~i (1991) and histology on O'Brien et al. (1980). dramatically increased (O'Brien et al., 1978). Conversely, if neuromuscular activity is reduced either by applying TTX to the motor nerve, or by surgical means such as tenotomy or spinal cord section, the rate of synapse elimination is reduced (Benoit and Changeux, 1975; Caidwell and Ridge, 1983; Callaway et al., 1987). Such results could be explained in terms of a competitive interaction between the axons contacting the endplate region without an active participation of the target muscle, or it could be that the activity of the muscle influences its own innervation. This question has been addressed in experiments where motoneurone activity was not interfered with, but the activity of the target muscle was manipulated pharmacologically. This was achieved by blocking the enzyme that hydrolyscs the transmitter, ACHE. Under these conditions the loss of contacts between nerve terminals and muscle fibres was greater than normally seen (O'Brien et al., 1982; Duxson and Vrbov~, 1985; Greensmith and Vrbov~i, 1991). M oreover, if during the period of rapid synapse elimination the response of the muscle is prevented by blocking the AChR using curare or BTX, the loss of neuromuscular contacts is temporarily arrested (Srihari and Vrbowi, 1978; Duxson, 1982). The mechanism by which the target muscle regulates its synaptic inputs is not entirely clear, but there is evidence that: (a) the depolarization of the endplate region is an important step in this process, and (b) reducing the Ca + + concentration of the extracellular compartment during the period of synapse elimination slows the disappearance of polyneuronal innervation (Connold et al., 1986). Thus Ca + + ions are involved in regulating neuromuscular contacts. They seem to do so through a Ca + ÷ activated neutral protease (CANP), calpain, which is known to be involved in the breakdown of cytoskeletal proteins (see review by Croall and DeMartino, 1991). In vitro studies demonstrated that prolonged depolarization of the endplate region leads to disruption of many neuromuscular contacts (O'Brien et al.,

1982, 1984). Figure 8 illustrates that exposure of the soleus muscles from 8-10oday-old animals to ACh for as short a time as 1-2 hours leads to a reduction of neuromuscular contacts (Zhu and Vrbovfi, 1992). The figure also shows that this effect of ACh was much reduced when the muscles were pre-incubated in BAPTA-AM (a compound that reduces Ca +` transients in nerve endings) before being exposed to ACh, or when L-type Ca + ÷ channels were blocked by nifedipine (Zhu and Vrbov~i, 1992). In addition, Fig. 8 also shows results obtained from a different series of experiments where the decrease of nerve muscle contacts brought about by ACh in the developing soleus was reduced by inhibiting the CANP by leupeptin (O'Brien et al., 1982, 1984; Connold e{ al., 1986). We are therefore proposing the model as summarized in Fig. 9 to account for synapse elimination. During activity the skeletal muscle fibre releases K + which is known to accumulate in the synaptic cleft (for review see Lowrie et al., 1988). Such accumulation of K ÷ almost certainly causes depolarization of the nerve terminal, which in turn will lead to opening of Ca ++ channels and allow entry of Ca + + from the extracellular compartment. If the Ca ÷~ transients are

FIG. 9. Mechanism leading to elimination of polyneuronal innervation. Top drawing shows an endplate contacted by two terminals of unequal size and illustrates the entry of Ca + + into them (see arrows). In the smaller terminal Ca + + reaches high enough concentrations to disrupt nc-urofilaments as illustrated in the middle drawing. The :bottom drawing shows the consequence of this, i.e. withdrawal o f the small terminal.

ACTIVITY-DEPENDENTINTERACTIONSBETWEENMOTONEURONF~AND MUSCLES

B

A

C

,.If0

I

109

,o ol' I,

oj

0~

1,01

.,.--..q

~j0

~

C

0

FIG. 10. Schematic illustration of nerve terminal transfer from pnmary to secondary myotubes during myogenesis. In (A) the primary myotube is contacted by three terminals of unequal size, in (B) the smallest of these shares contact with both the primary and small secondary myotube (note close contact between the two myotubes), in (C) the secondary myotube is separated from the primary myotube and has the small terminal. After Duxson et al. (1986). prolonged and large enough, then the concentration of Ca ++ in the nerve terminal may become high enough to activate the CANP present and lead to the breakdown of cytoskeletal proteins. This breakdown will subsequently lead to a loss of contacts between the nerve terminal and muscle fibre. According to this proposal, small axon profiles would be more vulnerable for two reasons: (a) because of their surface-tovolume ratio, the Ca + + transients can be expected to be greater in small than in large terminals, and (b) because of their smaller size, they probably contain less cytoskeletal protein. It has indeed been shown that in situations where loss of terminals is known to take place, contacts between small terminals and muscle fibres are reduced more than those between large terminals and muscle fibres (O'Brien et al., 1984; Duxson and Vrbov~i, 1985). Consistent with the proposal that CANP is involved in the reduction of neuromuscular contacts are results showing that blocking CANT either in vitro or/n vivo reduces loss of neuromuscular contacts (Connold et al., 1986; O'Brien et al., 1982, 1984). Thus any event that prolongs or increases depolarization of the postsynaptic membrane may have a deleterious effect on small nerve terminals that contact it. The possibility that the degree of maturity of the endplate may be important for the maintenance of syuaptic contact between a nerve terminal and muscle fibre (discussed in Section 3.2.1) is an important concept, for it indicates another dimension by which matching between motoneurones and muscle fibres can be achieved. Primary myotubes that are larger and more mature than newly formed secondary myotubes will provide more favourable conditions for nerve terminal contacts, and it is indeed these fibres that initially are able to accept a much higher degree of polyneuronal innervation than the secondary, smaller fibres (Sheard et al., 1991). The transfer of terminals from primary to secondary myotubes which follows, and is illustrated in Fig. 10, may be regulated by the mechanisms described in this section. This point will be discussed in detail in section 5.1.

There are other possibilities by which muscle activity may regulate its synaptic input, such as the release of specific motoneurone (or nerve terminal) survival factors, but one would have to postulate that the less active the endplate, the more survival factors it releases. Such a mechanism, though attractive by its simplicity, has not so far gained much experimental support (for review see Jansen and Fladby, 1990).

4. DEVELOPMENT OF MOTOR ACTIVITY 4.1. EMBRYONICDEVELOPMENTOF MOTOR FUNCTION The motor behaviour of the embryo reflects the early activity of the nervous system. The study of motor behaviour attracted much interest, particularly during the first half of this century when many studies were aimed at correlating the emergence of spontaneous and reflexly elicited embryonic movements with the anatomical development of the CNS (for reviews see: Coghill, 1929; Carmichael, 1926; Windle, 1940; Hamburger, 1963; Provine, 1976; Bekoff,1981; Prechtl, 1984). More recently, studies of embryonic development in experimental animals have been able to directly correlate behavioural observations with electrophysiological recordings of the underlying neuronal activity (see below). Taken together, these studies show that from the outset, embryonic movements are neurogenic, i.e. they reflect motoneurone activity that activates the target muscles. With further development, embryos and neonates show increasingly more complex and coordinated activity patterns that reflect the maturation of the neuronal networks that generate this activity. More than a century ago, Preyer (1885) made the important observation that spontaneous movements in the chick embryo appear at 3½days of incubation, that is, several days before any responses can be elicited by skin stimulation (7 days). The first spontaneous movements consist of movements of the head soon followed by those of the trunk (Hamburger and

1 l0

R. NAVARRETE and G. VRBOVA

I00 25

•~"

/

~

~5C

/

\

C

I

5

I

6

I

7

I

8

I

9

I

I

II 13 Age (doys)

20 .~_

I

15

I

17

I

I

19 20

~)

FIG. l 1. Changes of motor activity in the chick embryo are plotted against age. Solid line represents the percentage of time during which polyneuronal burst discharges were present at different stages of development. The short vertical lines represent the standard errors of these values. Percentage of time during which polyneuronal burst activity is present is compared with the mean number of body movements (interrupted line). Reproduced from Provine (1972b) with kind permission from the author and Elsevier Science Publishers. Balaban, 1963). At about 6 days of incubation the limbs begin to move. At this stage movements are jerky and uncoordinated. A general characteristic of embryonic movements is their periodicity; phases of high motor activity alternate with inactive phases (Hamburger and Balaban, 1963). A marked i n ~ in overall activity takes place at about E9, reaching a sustained plateau of almost continuous activity between E13-17. This high level of activity decreases towards the time of hatching (Hamburger et al., 1965). Figure 11 provides an illustration of the changes in the amounts of activity with embryonic ages. More specific changes were revealed by electromyographic (EMG) recordings. They show that at E7, flexor and extensor muscles of the leg are activated synchronously for brief periods of time, and that coordinated, reciprocal activation of antagonists emerges gradually after E9-10 (Bekoff, 1976; Landmesser and O'Donovan, 1984; O'Donovan and Landmesser, 1987). Observation of embryonic movements after experimental manipulations have provided evidence that for much of embryonic life, motor activity is generated in the spinal cord and is independent of segmental afferent or descending input (Hamburger et aL, 1966; Oppenheim, 1975). Extensive deafferentation of the lumbosacral spinal cord in E2 embryos and isolation of the lumbrosacral cord from the brain did not alter markedly embryonic movements up to E15-17. After this time, however, the amount of activity decreased sharply and the embryos were unable to perform coordinated hatching movements, indicating that these more complex movements are dependent on supraspinal influences (Hamburger et al., 1966; Oppenheim, 1975).

Spontaneous movements with similar characteristics to those observed in the chick embryo, are seen in mammalian foetuses (Graham-Brown, 1915; Angulo y GonzAlez, 1932; Windle and Baxter, 1936; Narayanan et o2., 1971). Graham-Brown (1915)was the first to observe that exteriorized foetal kittens with an intact umbilical cord can perform coordinated alternating limb movements which resemble the stepping movements of adult cats. He suggested that these movements are "automatic" and represent a prenatal antecedent to adult locomotion (Graham-Brown, 1915). The advent of non-invasive ultrasound techniques has recently allowed the study of the s e q ~ of appearance of movements in the undisturbed human foetus (Reinold, 1976; de Vries et al., 1982). The first movements appear at 7-8 weeks of gestation and consist of slow movements of limited amplitude lasting for only 1-2 sec (de Vries et al., 1982, 1985). The total incidence of foetal movements increased rapidly between 8-13 weeks, reached a sustained plateau from 13-20 weeks and decreased during the second half of gestation. 4.1. I. Electrophysiological correlates

In the past few years our knowledge of the spinal neuronal circuits involved in the generation of embryonic movements has advanced considerably duo to the introduction of #1 vitro preparations of the spinal cord and bralustem. These/n vitro preparatior~ can produce patterns of neuronal activity similar to those recorded/n eve from chick embryos. The first direct electrophysiological demomtmtion of the origin of embryonic movements in the chick was provided by Provine and colleagues (see Provine,

ACTIVITY-DEPENDENTINTE~CT~ONSBL~rw~ MOTOh~UgO~SANDMUSCLES

1976). Extracellular recordings in situ from the lumbosacral spinal cord in normal embryos revealed that neuronal activity in the spinal cord is concomitant with observed limb movements (Ripley and Provine, 1972). This activity was generated endogenously in the spinal cord and was not affected by blocking neuromuscular activity. These experiments indicated that movement-induced afferent signals are not necessary for the production of embryonic movements and that these movements are generated by intrinsic circuits within the spinal cord (Provine, 1976). Neuronal activity in the spinal cord during episodes of embryonic motility (at E6) is characterized by a high degree of synchrony. Synchronous neuronal discharges could be recorded from two separate electrodes located in different parts of the same segment or even from widely separate segments of the spinal cord (Provine, 1971). More recently this synchronous activity has been investigated by recording the output of functionally distinct motoneurone pools. EMG recordings from antagonist leg muscles at these early stages of development (stage 30) show that episodes of motor activity begin with a highly synchronized burst in both agonist and antagonist muscles (Landmesser and O'Donovan, 1984; O'Donovan and Landmesser, 1987). At later stages (stage 31), the synchronous burst is followed by a delayed discharge which is longer in flexor than in extensor muscles, thus accounting for the reciprocal activation of the antagonist muscles. Several recent lines of evidence indicate that the physiology, pharmacology and anatomical connectivity of the spinal neuronal networks involved in the generation of embryonic motor activity undergo considerable developmental changes which extend well into postnatal life. Firstly, reciprocal excitation of agonists and antagonists has been previously observed during embryonic and early .ppstnatal life in various species including humans (~nggard et al., 1961; Ekholm, 1967; Gottlieb et al., 1982; Navarrcte et al., 1987; Lee et al., 1988; O'Sullivan et al., 1991). The synchronous synaptic depolarization of agonist and antagonist motoneurones which characterizes embryonic motor behaviour has not been described in adults (see Baldissera et al., 1981; Jordan, 1983). This suggests that pre-motor interneurones may initially establish excitatory connections to both flexor and extensor motoneuroncs and that the inappropriate connections may become eliminated, or rendered ineffective, during further development. Secondly, developmental changes in neuronal properties such as a decrease in input resistance with age (Fulton and Walton, 1986; Navarrete et al., 1988) or changes in dendritic structure (Kudo and Yamada, 1987; Ulfhake and Cullheim, 1988; Jacobs and Weeks, 1990) may alter the responses of motoneurones or interneurones to synaptic depolarization. Finally, the developmental changes of the pharmacological properties of voltage and ligand gated ion channels will affect the response of the cells to these transmitters. 4.2. POSTNATALDEVELOPMENTOF MOTORFUNCTION Many features characteristic of embryonic motor behaviour persist after birth in the rat and other mammalian species including humans (see Prechtl,

111

1974). For instance, episodes of phasic motor activity are often observed in the newborn rat, usually in association with paradoxical sleep and are thought to represent a postnatal continuation of spontaneous embryonic movements (see Corner, 1977). However, after birth the requirement to cope with gravity imposes new demands on the motor system. This is particularly evident for motoneurones which innervate muscles concerned with the maintenance of posture. Distinct patterns of activity of motoneurones to slow and fast muscles emerge, in part, after birth, as descending systems concerned with antigravity postural function begin to exert a controlling action over the spinal cord. As indicated in the previous section, some aspects of spinal motor function, such as the reciprocal activation of antagonists, can be seen during prenatal life. These features of spinal motor function can also be elicited under appropriate experimental conditions in the neonate. In newborn rats which do not yet show spontaneous locomotor movements, coordinated reciprocal activation of antagonists can be triggered by exogenous administration of the monoamine precursor L-DOPA (Navarrete and Vrbov/L, 1985; Iwahara e t a / . , 1991). The most pronounced changes of locomotor activity in the rat are seen during the first three weeks of postnatal development (Altman and Sudarshan, 1975; Stelzner, 1971; Westerga and Gramsbergen, 1990). Simultaneous EMG recordings from hindlimb flexor and extensor muscles in freely moving rat pups were used to characterize the changes in activation patterns of the motoneurones to these muscles during this period (Navarrete and Vrbovfi, 1983; R. Navarrete and G. Vrbov/~, unpublished observations). During the first postnatal week, forward progression is effected mainly by "pivoting" movements of the forelimbs whilst the hindlimbs are unable to lift the pelvis off the ground. At this time the EMG activity of soleus, a hindlimb extensor muscle involved in postural support, is strikingly different from that of the adult. As shown in Fig. 12, the 7-day-old soleus muscle is mostly silent at rest, and during spontaneous locomotion the soleus is activated only for relatively brief periods (Fig. 12Aa). At this stage of development, the aggregate EMG activity of the soleus muscle is not significantly different from that of the flexor TA and EDL muscles (see Fig. 12B). Another interesting feature of the EMG activity pattern in neonatal rats is the presence of clear instances of co activation between ankle flexor (TA) and extensor (soleus) muscles (Fig 13Aa). This coactivation of antagonists may serve to stabilize the hindlimb during attempts at quadrupedal locomotion (see also Bradley and Smith, 1988a). This immature form of postural support is probably due to insufficient tonic descending drive to the extensor motoneurones since such co-activation is not usually observed during L-DOPA-induced locomotion in neonatal rats (Navarrete and Vrbov/L, 1985; R. Navarrete unpublished observations). At about 10-12 days of age, motoneurones to the soleus muscles undergo a fundamental change in their pattern of activity. At this time, tonic discharges of a few motor units are first detected in the resting animal. By 14 days, postural EMG activity in soleus

R. NAVAR~'rEand G. Vgaov.~

112

A

B = Spont

Rest

locomotion

=SOL

~TA/EDL

a

i

TA

t

7d SOL

b TA L

,l h [III i

!

' ' 'I,ll v

14d.

_=

'

O

w

C TA

. . . .

I

li

li

"

"

21(1

2see

i

8

12

16 20 AGE (days)

FIG. 12. Postnatal development of motor activity in slow and fast muscles of the rat. In A, examples of EMG records taken from soleus and TA/EDL muscle of freely moving rats aged 7, 12 and 21 days are presented. Recordings were made at rest and during spontaneous locomotion. In B, the aggregate EMG activity recorded from soleus and TA/EDL muscles is plotted against the postnatal age of the animals. Vertical lines show SEM. For explanation see text. is clearly established (Fig. 12Ab) and the aggregate activity of the muscle is almost 3 times greater than that of the fast flexor muscles (Fig. 12B). Quadrupedal locomotion, with full hindiimb support, is first observed at this time (coinciding with the time of eye opening) and there is a general improvement of motor coordination (Fig.12Ab). By 21 days, the basic qualitative EMG pattern seen during adult locomotion, characterized by reciprocal, tonic activity in soleus and phasic activity in TA/EDL, is fully established (Vrbov~i, 1963; Hni.k et al., 1981; Navarrete and VrbovL 1983, 1984; Hennig and Lomo, 1985; Vejsada et al., 1991). Further quantitative increases in soleus aggregate EMG activity take place during the following three to six months (Navarrete and VrbovL 1984) as the postural function of the muscle increases in proportion to the weight of the animal (see Kugelherg, 1976). Analysis of the firing rate of individual motor units at low levels of recruitment indicates that during the first postnatal week both prospective slow (soleus) and fast (EDL) motor units have a relatively low frequency irregular discharge. During the second postnatal week, motor unit activity in solens becomes more regular but its frequency remains low, whereas EDL motor units increase their rate of discharge (Navarrete and Vrbov~, 1983). Intracellular recordings from motoneurones /n vitro indicate that the postnatal increase in the firing rate of presumptive

"fast" motor units is consistent with a concomitant decrease in the duration of the A H P that occurs during the same period (Navarrete et al., 1988; R. Navarrete and K. Walton, in preparation). Characterization of the firing pattern of single motor units from EMG recordings during the first two postnatal weeks is complicated by the higher incidence of polyphasic motor unit potentials (MUP) and the apparent synchronous recruitment of different motor units (Bursian and Sviderskaya, 1971; Navarrete and Vrbov=i, 1983). These immature features of the MUP are probably due to two main factors. Firstly, individual muscle fibres are potyneuronally innervated and the territory of different motor units therefore overlaps (see section 3.2). Thus, activation of a given muscle fibre by more than one terminal contributes to the longer duration and polyphasic nature of its MUP. Secondly, the pr0sence of eleetrotonic or gap junctional coupling between neonatal motoneurones (Fulton et al., 1980; K a r a m l a n et al., 1988; Motorina, 1989; Becket and Navarrete, 1990; Walton and Navarrete, I991) may lead to synchronous firing of several motor units during recruitment. These characteristics of immature motor units impose limitations on the ability of an individual motoueurone to impose a uniform activity pattern on all of the muscle fibre it supplies. Only after elimination of polynvuronal innervation and the disappearance of electrotonic coupling between

ACTIVITY-DEPENDENTINTERACTIONSBETWEENMOTONEURONESANDMUSCLES

motoneurones (Walton and Navarrete, 1991) can an individual motoneurone exert an exclusive influence over the muscle fibres it supplies and only then the concept of the motor unit in the Sherringtonian sense may be said to emerge. In summary, embryonic motor activity appears as soon as neuromuscular connections are established and thus represents the first external manifestation of function in the developing CNS. This neuronal activity is channelled via the motoneurone and transmitted to the target muscle fibres. In addition, transmitter release from the motor nerve endings may not always elicit an action potential in the muscle fibre, but may nevertheless influence the cell. Thus, from the outset embryonic activity can influence the evolving motor unit in various ways. Embryonic movements follow in general, a cranio-caudal sequence which reflects the maturation of the circuitry in the corresponding segments of the spinal cord. The motor output to different muscle groups in the embryo shows evidence of coordinated function suggesting that motoneurons establish roughly appropriate central and peripheral connections early on. During postnatal development however, various aspects of the anatomy and function of the motor unit undergo refinement. This process coincides with the emergence of distinct activity patterns of motoneurones supplying slow and fast muscles, the ordered recruitment of motor units and the onset of coordinated locomotor function.

5. MECHANISMS INFLUENCING THE DEVELOPMENT OF THE M O T O R UNIT: ROLE OF ACTIVITY There are two apparently conflicting proposals concerning the mechanism by which the motor unit emerges. The first proposal is based on the many findings demonstrating that the muscle fibre phenotype can be modified by neuronal activity and adapt to the pattern of activity imposed upon them by the motor nerve. This is the case both in young and adult animals. If in young kittens the motor nerves from one muscle are transposed to a different muscle the development of the muscle's contractile properties will be regulated by the motor nerve that now supplies it (Buller et al., 1960). There is ample evidence that in adults the motonenrone regulates muscle phenotype by means of its activity (for reviews see Frischknecht et ai., 1990; Pette and Vrbov~, 1992). Given this plasticity of the muscle fibre, it was proposed that the development of the matching properties of muscle fibres to the activity pattern of the motoneurone is determined by activity (Vrbovfi, 1980). However, more recent evidence appears to indicate that even at a time when (a) motoneurones and their activity patterns are not yet fully differentiated, and (b) muscle fibres receive inputs from several axons/ motoneurones, there is already some matching between muscle fibres and the motoneurones that supply them. Using the glycogen depletion method, conflicting results have been obtained. Thompson et al. (1984) reported that in the neonatal rat soleus, individual axons supply muscle fibres that contain JPN 41/[--H

113

similar or identical MHC isoforms. Thus there appears to be a segregation of muscle fibres of a similar type within a single motor unit. In the lumbrical muscle of the neonatal rat, however, muscle fibres belonging to the same motor unit have been demonstrated to be heterogeneous (Jones et al., 1987). Perhaps the most compelling evidence that some degree of homogeneity of muscle fibres within the same motor unit is present soon after birth, before the activity patterns of motoneurones are fully differentiated is that of Fladby and Jansen (1990). These authors identified several muscle fibres supplied by a single axon electrophysiologlcally, by recording endplate potentials, and labelled these muscle fibres with Lucifer Yellow. They report a greater degree of homogeneity within fibres activated by a single axon than within randomly selected fibres. Thus there is some indication that a rough layout of motor unit territory may be established early in development. This initial layout of motor unit territory could be regulated by special molecular markers as proposed by Miller and Stockdale (1987) and Thompson et a/. (1984). In favour of this idea are findings that some cell adhesion molecules (N-CAM) are found in higher amounts in adult slow, tonic avian muscle fibres (Bleisch et al., 1989) and these may be matched to markers on growth cones of "slow" but not "fast" motoneurones. However, there is no convincing evidence to show that any of these surface markers are actually expressed differentially during synaptogenesis, so that axons of fast and slow motor units could recognize them. An alternative is that the initial layout of motor unit territory is regulated by mechanism where presynaptic terminals contact postsynaptic muscle fibres with comparable degrees of maturity. In the next section, the possible mechanisms that regulate the initial layout of motor unit territory within a given muscle are considered first. Later, events that lead to the specialization of muscle fibres into distinct types to achieve the matching of motoneurone activity pattern with muscle fibre properties are discussed 5.1. EARLYLAYOUTOF MOTOR UNIT TERRITORY By definition, the motor unit is formed by the motoneurone and all the muscle fibres it supplies. In the adult the unique feature of such a unit is that a muscle fibre can be supplied only by a single motoneurone and is not shared by others. This arrangement is necessary to enable the motoneurone and its associated muscle fibres to function as a true unit (Frischknecht et ai., 1990). During most of embryonic development such a concept of the motor unit does not exist. This is due to the fact that initially there is convergence of the full complement of motor axons onto relatively few myogenic cells. Moreover, several axons share the same myotube. During early stages of development all axons of a particular motor pool are already present among the few myogenic cells that will form a particular muscle, and the number of muscle fibres available for the axons to innervate is relatively small. The majority of these early muscle fibres are primary myotubes, which are relatively well developed, span the whole length of the muscle and have a large diameter (for review see

114

R. NAVARRETEand G. VRBov~,

Kelly, 1983). With further development small myotubes appear enclosed within the same basal lamina as the primary myotube. These are of small diameter, contain few myofilaments and are initially short. With time, these secondary myotubes fuse with other myoblasts, increase in length and gradually reach the tendons where they form independent attachments (Ross et al., 1987). In some muscles quantitative estimates of the numbers of primary and secondary myotubes has been carried out. It was reported that in the IVth himbricai muscle of the rat there are only about 100 primary myotubes (Ross et al., 1987), which is about 12% of the total number of fibres the muscle contains in its adult state (Betz et aL, 1980). Similarly, in the rat sternocostalis muscle the initial number of primary myotubes present at the time when innervation of the muscle fibres starts is less than 20% of the final number of muscle fibres (Sheard et al., 1991). The oldest, more advanced, primary myotubes are contacted by several nerve terminals derived from different axons. At this stage many new secondary myotubes are generated which are in close proximity to the primary myotubes and the axon terminals that contact them. Indeed, the contact between the primary and secondary myotubes is so intimate that they are connected by gap junctions which mediate electrical coupling between the two fibres (see Kelly, 1983). Moreover, some secondary myotubes share some of the axon terminal profiles with the large primary myotube (Duxson et al., 1986). With further development at least some of these axon terminals will lose contact with the primary myotube and innervate only their new target, the secondary myotube. Thus there is a transfer of terminals from the primary to the secondary myotube (Duxson et al., 1986). This process is illustrated diagramatically in Fig. 10. The sequence described above suggests that a selection is carried out as to which of the many terminals is to be transferred from the primary to the secondary myotube. What can influence this selection? An important factor in this selection process could be the compatibility of the secondary myotube with a particular terminal. Compatibility may depend on the characteristic properties of: (a) the new axon terminal that is to be transferred, (b) the secondary myotube that is to accept a new terminal, and (c) the interaction between the terminal that is to be transferred, with other terminals sharing contact with the primary myotube. We will deal first with the characteristics of the small secondary myotubes, since we can make certain assumptions about these that are more difficult to make about the axon terminals. The characteristics of the secondary myotube will be very different from those of the primary one, in that it will have a much higher input resistance, less mature AChRs and less AChE (see Sections 2.2.3 and 2.2.4). Thus the secondary myotube will respond with a relatively large depolarization to even a low amount of transmitter released from the terminal. Its response to higher amounts of transmitter release will be disproportionately large. In addition, the immature myotube is likely to have a prolonged endplate potential, since it has AChRs with long channel opening times, and little AChE (see Section 2.2.3). Indeed, recent results confirm this, and show that

EPPs of young myotubes are much longer than those of more mature cells. With increasing age the rise time and decline of depolarization becomes faster. A recent study on the rat sternocostalis muscle shows that at a time when the duration of the endplate potential is already shorter in the primary myotube, the less mature secondary myotube still has prolonged EPP (Sheard et al., 1991). A more mature or more active terminal would therefore produce a large and prolonged response on a secondary myotube. Such a large response can lead to the retraction of the nerve ending, by an activity-dependent mechanism such as described in Section 3.2.2. For this reason the most likely candidate for transfer would be a relatively immature, inactive terminal, which will initially produce a small depolarization and activate the myotube only rarely. The lower level of activity of a small terminal could be due both to a low quantal content of the terminal and to intermittent failure of propagation of the action potential in the preterminal axon (Pilar et al., 1981). Such a terminal will have a better chance of maintaining contact with a small secondary myotube and gradually strengthen this synaptic contact. Thus, according to this proposal an immature terminal will be best matched to establish contact with a secondary myotube. Indeed, the earliest independent contacts between secondary myotubes and axon terminals are those that initiate an EPP of long duration but small amplitude, indicating a low quantal content of the terminal (Sheard et aL, 1991). Indirect supportive evidence for such a mechanism comes from results obtained on the developing A L D and PLD muscles of the chick. In these muscles quite early in development slow tonic fibres are seggregated in the A L D and fast twitch fibres in the PLD. Most muscle fibres in the A L D are generated between E8-11 and by El2 A L D has the full complement of its muscle fibres. PLD develops later than A L D and the most pronounced increase in the number of muscle fibres takes place between El0-12, and thereafter continues at a slower rate until E15 (Oppenheim and Chu Wang, 1983). At 16 days of embryonic development the A L D muscle fibres are larger and more mature than muscle fibres of the fast PLD as judged by their content of myofibrils and by the position of the nuclei (Gordon et al., 1974, 1981). At this time of development the axons profiles on A L D muscle fibres are more numerous and mature than those on PLD muscle fibres (Srihari and Vrbov/t, 1978, 1980). While many axon profiles on El6 PLD muscle fibres are still surrounded by Schwann cell processes, most axon profiles on El6 A L D make good contact with the muscle fibre and contain a larger number of vesicles and mitochondria (Srihari and Vrbov~, 1978, 1980). Figure 13 illustrates this point. Thus the developmentally younger PLD muscle fibres which could be considered to contain secondary myotubes seem to be contacted by less mature axons. However, within the following few days the axon profiles on PLD muscle fibres grow and become much larger than those on A L D fibres. It is possible that the rapid growth of the terminals is due to relative lack of activity of the synapse. Inactivity is known to allow terminals to become large, as illustrated at the frog neuromuscular junction, where the length of the motor nerve terminal increases when

FIO. 13. Eleetronmicrographs of neuromuscular junctions from ALD (A) and PLD (B) muscles from 16day-old chick embryos are shown. Note that in ALD almost all nerve terminal profiles (T) are rich in synaptic vesicles, whereas in PLD the density of vesicles in the terminals is low. The ALD muscle fibre (M) is also more developed than the PLD muscle fibre, as judged by its high content of myofibrils. Scale = 0.5#.

115

ACTIVITY-DEPENDENTI~rnn~AcrloNsBETWEENMOTONEURONm AND MUSCLES the frogs are inactive (for review see Grinnell and Herrera, 1981). Finally, a selective transfer of less mature nerve terminals to small myotubes based on the activity of the terminal could also explain the checkerboard pattern of muscle fibre type distribution found in most adult mammalian muscles, and could be the earliest event that helps to establish the anatomical distribution of muscle fibres within a motor unit

117

spinal transection in neonates, which abolishes tonic but not phasic motoneurone activity (Bradley and Smith, 1988b), selectively prevents the development of slow contractile speed and the synthesis of slow myosin isoforms in soleus (Buller et al., 1960; Margreth et al., 1980). These specific effects of decreased tonic motoneurone activity of slow myosin expression suggest that normally the onset of antigravity function in the soleus muscle induces a change in MHC isozymes in fibres from this muscle. The role of antigravity function in this process is also supported 5.2. AC~WTYINDUCEDMODIFICATIONS OF MUSCLE by results which show that in the rat soleus, muscle FIBREPROPERTIES fast fibres which usually convert into slow ones As discussed previously (Section 2.2.1), the ex- during development fail to do so if they are not pression of various myosin isoforms is an early event exposed to their normal load and the muscle continin myogenesis. Primary myotubes from future slow ues to contain a relatively high proportion of fast and fast muscles often express slow as well as embry- fibres. Moreover many muscle fibres in such an onic forms of M H C while secondary myotubes are unloaded soleus muscle contain both fast and slow thought to contain embryonic or fast M H C (Kelly isoforms of the MHC (Lowrie et al., 1989). and Rubinstein, 1980; Narusawa et aI., 1987). These The matching of motoneurone properties to the early events are independent of the motoneurone and phenotypic characteristics of the muscle fibres may its activity since they appear to take place even in involve two stages. One has already been mentioned aneural muscles (Weydert et aI., 1987; Condonet aI., and involves the matching of particular terminals to 1990). muscle fibres of an appropriate developmental stage. The relationship of these developmental changes in The second requires that the contractile and metamyosin isoforms to the speed of contraction of bolic characteristics of the muscle fibres that form the developing muscles and the formation of motor units motor unit are matched to the functional properties with distinct contractile characteristics is far from and central connections of the motoneurone that clear. While in adults a correlation between the speed supplies them. While very little information exists in of shortening of a muscle and the rate at which relation to the firing pattern of single, identified myosin hydrolyzes A T P exists (B~r~ny, 1967), such motoneurones in embryos and neonates it is worthcorrelations have not been demonstrated in embry- while to consider the possibility that at least part of onic or neonatal muscles. Moreover, there is little the phenotypic heterogeneity found in neonatal information about the abilityof the differentisozymic muscle could reflect early differences in the function forms of M H C to hydrolyse ATP. Such information of different motor units within a muscle. The fact that is essential if the research on the developmental the conduction velocity of motor axons supplying expression of various myosin isoforms is to be given immature kitten muscles is related to the contractile functional significance. Indeed factors which deter- speed of the unit (Ridge, 1967; Bagnst et al., 1974; mine the duration of the active state during contrac- Hammarberg and Kellerth, 1975) suggests that tion, such as the abilityof the sarcotubular system to muscle fibres are matched to the properties of motorelease and sequester Ca ++ , and the levels of Ca ++ neurones that innervate them already during early binding proteins may be more important. The finding postnatal life. This is consistent with the emergence that in the rat during the firsttwo weeks of postnatal of early differences in membrane electrical properties development both future fast (EDL) and slow between predominantly "slow" extensor compared to (soleus) muscle increase their speed of contraction, "fast" flexor motoneurones in the rat (Navarrete indicates that this developmental event, c o m m o n to et al., 1988). However, it is important to stress that both types of muscles, is probably associated with the even if the firing pattern of the motoneurone shows activation process of the contractile machinery. Only some differentiation prenatally, transmission at all later in development does the direction of change of peripheral synapses made by a given motoneurone is the contractile speed differ in future slow and fast likely to be unequal. This is due to the fact that the muscles. After the first 3-4 weeks after birth in the motoneurone's peripheral field is maximally exrat, the speed of contraction of the fast E D L contin- panded and transmitter synthesis at each terminal ues to increase whereas soleus gradually becomes probably cannot keep up with functional demand. slower (Close, 1964; Kugelberg, 1976). It is interest- Thus one could envisage a situation where the "core" ing that the period during which both slow and fast of the developing motor unit is formed by those muscles are increasing their speed of contraction (the synapses that already have more efficient neuromusfirst two weeks after birth) coincides with the time cular transmission and are in the process of consoliwhen both types of muscles are subjected to a pre- dation. This part of the motor unit may have more dominantly phasic activity pattern. During the third homogeneous characteristics than its "peripheral" postnatal week, the rate of increase of the contractile portion where many synapses are weak and may not speed is less than that of EDL, and this could be be able to elicit reliably a muscle action potential. The related to the onset of tonic postural activityin some findings that the homogeneity of the motor unit motor units of the soleus muscle CNavarrete and increases during the first few days after birth (Fladby Vrbovl, 1983). The fact that a specificactivitypattern and Jansen, 1990; Gates and Ridge, 1992), suggests and not simply the presence of the nerve is required that, as development progresses, the firing pattern of for thistransformation is indicated by the finding that the motoneurone is being transmitted more faithfully

118

R. NAVARRETEand G. VRBOV,~

to a larger number of synapses, thus contributing to the greater uniformity of the phenotypic characteristics of the motor unit. Remodelling of muscle fibre characteristics occurs throughout the life of the animal. During postnatal development this phenomenon was first noticed in the rat soleus muscle where, with increasing age and weight of the animal, the number of slow myosin containing fibres increases (Kugelberg, 1976). In contrast, in the rat EDL and lumbrical muscles, the number of slow fibres decreases with age (Lowrie et al., 1988; Gates and Ridge, 1992), These age-related changes could be due to a shift of the motor unit population, i.e. in the soleus muscle whole motor units could be transformed from a fast towards a slow type. This proposal was originally made by Kugelberg (1976). Alternatively, transitions of muscle fibre phenotype may occur in those muscle fibres which contain a different MHC isoform during early development from that of the motor unit to which they finally belong. Although Kugelberg (1976) supported the first possibility, recent results indicate that the latter phenomenon might also contribute to the final "sorting out" process. If the suggestion that only whole motor units are altered during development was correct, it would have to be assumed that the original matching between a particular type of axon and the muscle fibre it supplies is indeed perfect, with no errors. This is unlikely to be the case. Results on the IVth lumbrical muscle of the rat show that there is a much greater heterogeneity of muscle fibres belonging to the same motor unit in younger animals than during later postnatal life, indicating that some muscle fibres become gradually transformed under the influence of innervation (Gates and Ridge, 1992). This result is not surprising in view of the ability of muscle fibres to adapt to different patterns of activity or mechanical stimuli (see Frisehknecht et al., 1990). Indeed it is well known that muscle fibres change their phenotype during postnatal development (Goidspink and Ward, 1979) and the relative slowness of this process may account for the conflicting interpretations concerning the homogeneity of muscle fibres within a single motor unit. It was claimed by Kugelberg and Edstrom (1968) and later by Nemeth et al. (1981) that motor units are homogeneous. This, in broad terms, is true, but if transitions with age take a long time to be completed, there may be considerable variability in the results obtained. Moreover, until recently it was possible to study only the presence of a protein but not the time course of expression of the gene coding for that protein. Particularly in the case of myosin, this may be very misleading because of the time lag involved between the change in gene expression and the presence of detectable amounts of the protein in the muscle fibre (see Pette and Vrbowi, 1992). It is likely that the use of more modern experimental tools such as in situ hybridization and analysis of gene expression, in single muscle fibres will reveal in more detail the second stage of refinement of motor unit territory during postnatal development which is regulated by activity.

6. SUMMARY AND CONCLUSIONS In this review article we have attempted to provide an overview of the various forms of activity-dependent interactions between motoneurones and muscles and its consequences for the development of the motor unit. During early development the components of the motor unit undergo profound changes. Initially the two cell types develop independently of each other. The mechanisms that regulate their characteristic properties and prepare them for their encounter are poorly understood. However, when motor axons reach their target muscles the interaction between these cells profoundly affects their survival and further development. The earliest interactions between motoneurones and muscle fibres generate a form of activity which is in many ways different from that seen at later stages. This difference may be due to the immature types of ion channels and neurotransmitter receptors present in the membranes of both motoneurones and muscle fibres. For example, spontaneous release of acetylcholine may influence the myotube even before any synaptic specialization appears. This initial form of activity-dependent interaction does not necessarily depend on the generation of action potentials in either the motoneurone or the muscle fibre. Nevertheless, the ionic fluxes and electric fields produced by such interactions are likely to activate second messenger systems and influence the cells. An important step for the development of the motor unit in its final form is the initial distribution of synaptic contacts to primary and secondary myotubes and their later reorganization. Mechanisms that determine these events are proposed. It is argued that the initial layout of the motor unit territory depends on the matching of immature muscle fibres (possibly secondary myotubes) to terminals with relatively weak synaptic strength. Such matching can be the consequence of the properties of the muscle fibre at a particular stage of maturation which will accept only nerve terminals that match their developmental stage. Refinements of the motor unit teritory follows later. It is achieved by activity-dependent elimination of nerve terminals from endplates that are innervated by more than one motoneurone. In this way the territory of the motor unit is established, but not necessarily the homogeneity of the physiological and biochemical properties of its muscle fibres. These properties develop gradually, largely as a consequence of the activity pattern that is imposed upon the muscle fibres supplied by a given motoneurone. This occurs when the motor system in the CNS completes its development so that specialized activity patterns are transmitted by particular motoneurones to the muscle fibres they supply. In view of the great importance of motoneurone activity upon the developing motor unit, the review deals in considerable detail with the description of developmental changes in motor function and its underlying mechanisms. It emphasizes the importance of the more generalized early embryonic motor activity for maturation of both muscle and motoneurone and introduces the reader to the later perinatal stages, when specific patterns of movement gradually

ACTIVITY-DEPENDENTINTERACTIONSBETWEENMOTOX~U~tONESAND MUSCLES

emerge. Indeed it is during this later phase of development that the motor unit emerges as a specialized entity for fine motor control. A c k n o w l e d g e m e n t s - - W e wish to thank Dr A. Hind for her editorial help in all aspects of this article. We are grateful to Susie Okyere-Dcbrah for typing successive versions of the manuscript. It was a pleasure to have Dr. L. Greensmith's advice in designing the illustrations. We would also like to thank Dr F. Vyskocil for his comments on an earlier version of the manuscript.

REFERENCES ALBANI,M., Lowers, M. B. and VRaovk, G. (1988) Reorganization of motor units in reinnervated muscles of the rat. 3.. Neurol. Sci. 88, 195-206. ALTMAN, J. and SUDA~tlAN, K. (1975) Postnatal development of locomotion in the laboratory rat. An~n. Behav. 23, 896-920. ANGULOV GONZALEZ,A. W. (1932) The prenatal development of behavior in the albino rat. 3". Comp. Neurol. SS, 395--442. ARAMAN'r,R.B., GmoN, L. T. and ZWa3LEt, M. G. (1986) Postnatal development of dopaminc-B-hydroxylaseimmunoreactive fibers in the spinal cord of the rat. Dev. Brain Res. 25, 161-171. ~NOG^RD, L., I~RSTROM,R. and BERNHARn,C. G. (1961) Analysis of prenatal spinal reflex activity in sheep. Acta Physiol. Scand. 53, 128-136. BAGUST, J., LEWlS, D. M. and WI=STERMAN,R. A. (1973) Polyneuronal innervation of kitten skeletal muscle. 3.. Physiol. (Lond.) 229, 241-255. B^GUST, J., LEvels, D. M. and W~STERMAN, R. A. (1974) The properties of motor units in a fast and a slow twitch muscle during post-natal development in the kitten. J. Physiol. (Lond.) 237, 75-90. BALDIS~ERA,F., HULTBORN,H. and ILLERT,M. (1981) Integration in spinal neuronal systems. In: Handbook of Physiology. Section 1. The Nervous System. Vol. 2. Motor Control, pp. 509-595. Ed. V. B. BROOKS.American Physiological Society, ikthesda, MD. BAMBR1CK, L. L. and GORDON,T. (1988) Neural regulation of [3H] saxitoxin binding site numbers in rat neonatal muscle. 3.. Physiol. (Lond.) 407, 263-274. B,~P.~NY, M. (1967) ATPase activity of myosin correlated with speed of shortening. 3.. Gen. Physiol. SO, 197-218. B A t ~ , M. and CLO~, R. I. (1971) The transformation of myosin in cross-innervated rat muscles. J. Physiol. (Lond.) 213, 455-474. BECKER, D. L. and NAVARRET~,R. (1990) Dye-coupling between lumbar motoneurones in the embryonic and neonatal rat spinal cord: an in vitro study. Y. Physiol. (Lond.) 423, 98P. BEKOFF, A. (1976) Ontogeny of leg motor output in the chick embryo: a neural analysis. Brain Res. 106, 271-291. BEKOF~, A. (1981) Embryonic development of the neural circuitry underlying motor coordination. In: Studies in Developmental Neurobiology. Essays in honor of Viktor Hamburger, pp. 134-170. Ed. W. M. COWAN.Oxford University Press, New York. BEN~nT, M. R. 0983) Development of neuromuscular synapses. Physiol. Rev. 63, 915-1048. BEN~TT, M. R. and l~rnOREW, A. G. (1974) The formation of synapses in striated muscles during development. J. Physiol. (Lond.) 241, 515-545. BENOIT, P. and CtlANGEtrX,J.-P. (1975) Consequences of tenotomy on the evolution of multi-innervation in developing rat solens muscle. Brain Res. 99, 354-358. B~ROES, A. J. and TAKAHASm,T. (1990) Serotonin enhances a low-voltage-activated calcium current in rat spinal motoneurons. J. Neurosci. Ifi, 1922-1928. BETZ, H., BOURGEOIS,J.-P. and CHANGEUX,J.-P. (1980) Evolution of cholincrgic proteins in developing slow and fast skeletal muscles in chick embryo. 3". Physiol. (Load.) 302, 197-218. BETZ, W. J., CALDWELL,J. H. and RmCh'ESTI~, R. R. (1980) The cffects of partial denervation at birth on the development of muscle fibres and motor units in rat lumbrical musclc. 3.. Physiol. (Lond.) 303, 265-279, BLEISCH, W., SCHARFF,C. and NOTTEBOHM,F. (1989) Neural cell adhesion molecule (N-CAM) is elevated in adult avian slow

119

muscle fibers with multiple terminals. Proc. Natl. Acad. Sci. USA 86, 64O3--64O7. BLOCH, R. $. and STEIN~ACFI, J. H. (1981) Reversible loss of acetyleholine receptor clusters at the developing rat neuromuscular junction. Dev. Biol. gl, 386--391. BP~DLEY,N. S. and SmTH,J. L. (1988a) Neuromuscular patterns of stereotypic hindlimb behaviors in the first two postnatal months. I. Stepping in normal kittens. Dev. Brain Res. 38, 37-52. BRADLEY,N. S. and SmTH,J. L. (1988b) Neuromuscular patterns of stereotypic hindlimb behaviors in the first two postnatal months. 11. Stepping in spinal kittens. Dev. Brain Res. 38, 53-67. BRAITFIWAITE,A. W. and HARRIS,A. J. (1979) Neural influence on ACh receptor clusters in embryonic development of skeletal muscles. Nature 279, 549-551. BPa~'NEM^N, D. E., FORSYTHE, L D., NlCOL, T. and NEt.SON, P. G. 0990) N-methyI-D-aspartate receptors influence neuronal survival in developing spinal cord cultures. Dev. Brain Res. 51, 63-68. BRENNEM^N,D. E., NEALE,E. A., HABIG,W. H., BOWEI~,L. M. and NELSON,P. G. (1983) Developmental and neurochemical specificity of neuronal deficits produced by dectrical impulse blockade in dissociated spinal cord cultures. Dev. Brain Res. 9, 13-27. BP,~n~ER, H. R. and SAKUA~aN,B. (1983) Neurotrophic control of channel properties at neuromuscular synapses of rat muscle. J. Physiol. (Lond.) 337, 159-171. BRENNEI~,H. R., M I ~ , T. and WIDMER,B. (1983) Early action of nerve determines motor endplate differentiation in rat muscle. Nature 305, 536-537. B~'maL~, H. R., WITZEM^NN,V. and S^KMAma,B. (1990) Imprinting of acetylcboline receptor messenger RNA accumulation in mammalian neuromuscular synapses. Nature 344, 544-547. BtETAG, A. H. (1987) Muscle chloride channels. Physiol. Rev. 67, 618-724. BROWN, M. D. (1973) Role of activity in the differentiation of slow and fast muscles. Nature 244, 178-179. Bazr~, M., SKE'rELJ,J. and KL1NAR,B. (1983) Cholincsterases. In: Handbook of Histochemistry, pp. 251-192. Ed. A. L^J'rHA.Plenum Press, New York. BRZIN, M., SKETELJ, J., TENNYSON, V. M., KIAUTA, T. and BUDn,~NKAS-SCHOENESECK,M. (1981). Activity, molecular forms and eytochemistry of cholinestcrase in developing rat diaphragm. Muscle and Nerve 4, 505-513. BULLEt, A. J., ECCLES,J. C. and ECCLES,R. M. (1960) Interactions between motoneurones and muscles in respect of the characteristic speeds of their responses. 3.. Physiol. (Lond.) 150, 417--439. BURDEN, S. (1977) Development of the neuromuscular junction of the chick embryo: the number, distribution and stability of acetylcholine receptors. Dev. Biol. 57, 317-329. BURKI~, R. E. (1981) Motor units: anatomy, physiology and functional organization. In: Handbook of Physiology, The Nervous System II, Part 1, pp. 345--422. Ed. V. B. BROOK. American Physiological Society, Bethesda. BUt~AN, A. V. and SViDERSKAYA,G. E. (1971) Studies on the activity of motor units in newborn rats and kittens. J. Esol. Biochem. Physiol. 7, 255-265. BURT, A. M. (1975) Choline acetyltransfcrase and aeetylcholinesterase in the developing spinal cord. Expl. Neurol. 47, 173-180. BUTLER, J., COSMOS,E. and Bt~'RLY, J. (1982) Differentiation of muscle fiber types in aneurogenic brachial muscles of the chick embryo. J. Exp. Zool. 224, 65-80, CALDWELL, J. H. and RtDGE, R. M. A. P. (1983) The effects of deafferentation and spinal cord transection on synapse elimination in developing rat muscles. J. Physiol. (Lond.) 339, 145-159. CALL^WAY, E. M., SO~^, J. M. and V^N ~ , D. C. (1987) Competition favouring inactive over active motor neurons during synapse elimination. Nature 328, 422-426. C^RBONE, E. and Lux, H. D. (1987) Kinetics and selectivity of a low voltage activated calcium current in chick and rat sensory neuron. 3.. Physiol. (Lond.) Ifi, 547-570. CARMICHAEL,L. (1926) The development of behavior in vertebrates experimentally removed from the influence of external stimulation. Psychol. Rev. 33, 51-58. CATTERALL, W. A. 0991) Structure and function of voltage-gated sodium and calcium channels. Curt. Op. Neurobiol. !, 5-13. CH^NOEUX,J.-P. (1991) Compartmentalized transcription of acetylcholine receptor genes during motor endplate epigenesis. New Biol. 3, 413-429.

120

R. NAVARRETE and G. VRBOVA

CLOSE, R. 1. (1964) Properties of fast and slow skeletal muscles

FRANK,E. and FISCHBACH,G. D. (1979) Early events in neuromus-

during development in the kitten. J. Physiol. (LoAd.) 173, 74-95. Ct~HILL, G. E. (1929) Anatomy and the problem of behaviour. Harrier, New York. I10 pp. COMMISSIONG,J. W. (1983) Development of catecholaminergic nerves in the spinal cord of the rat. Brain Res. 264, 197-208. CONDON, K., SmBERS~N, L., BLAU, H. M. and THOMt~ON, W. J. (1990) Differentiation of fiber types in aneural musculature of the prenatal rat hindlimb. Dee. Biol. 138, 275-295. CONNOLD, A. L., EVERS,J. V. and VRaOV.~,G. (1986) Effect of low calcium and protease inhibitors on synapse elimination during postnatal development in the rat soleus muscle, Dee. Brain Res. 28, 99-107. CONRADI, S. and RO~EVl, L. O. (1975) Spontaneous elimination of synapses on cat spinal motoneurons after birth. Do half of synapses on cell bodies dissapear? Brain Res. 92, 505-510. Co~rrE-CAMERINO,D., DE LUCA, A., MA~¢mRINI,M. and VRaOV.~,G. (1989) Membrane ionic condactances in normal and denervated skeletal muscle of the rat during development. Pfliigers Arch. 413, 568--570. CORNEa, M. A. (1977) Sleep and the begginings of behavior in the animal kingdom. Studies of ultradian motility cycles in early life. Prog. Neurobiol. & 279-295. CROALL, D. E. and DEMAgrINO, G. N. (1991) Calcium-activated neutral protease (calpain) system: structure, function and regulation. PhysioL Rev. 71, 813-847. DAHM, L. M and LANDIC,I~SS~R, L. T. (1991) The regulation of synaptogenesis during normal development and following activity blockade. J. Neurosci. 11, 238-255. DENNIS, M. J., ZISKnqD-COh~AIM, L. and HARRIS, A. J. (1981) Development of neuromuscular junctions in rat embryos. Dee. Biol. 81, 266-279. DIAMOND,J. and MILEDI, R. (1962) A study of foetal and newborn rat muscle fibres. J. Physiol. (LoAd.) 162, 393-408. DRYDEN, W. F., ERULKAR, S. D. and DE LA HARA, G. (1974) Properties of the cell membrane of developing skeletal muscle fibres in culture and its sensitivity to acetyleholine. Clin. Exp. Pharmacol. Physiol. I, 369-387. DUXSON, M. J. (1982) The effect of postsynaptic block of development of the neuromuscular junction in postnatal rats. J. NeurocTtol, ii, 395-408. DuxSoN, M. J., Ross, J. J. and HARRIS, A. J. (1986) Transfer of differentiated synaptic terminals from primary myotubes to newformed muscle cells during embryonic development in rats. Neurosci. Lett. 71, 147-152. DUXSON, M. J. and Vaaov/~, G. (1985) Inhibition of acetylcholinesterase accelerates axon terminal withdrawal. J. NeuroeytoL 14, 337-363. ECCLES, J. C., ECCLES, R. M. and Ltn~DaEgG, A. (1958) The action potentials of the a-motoneurones supplying fast and slow muscles. J. Physiol. (LoAd.) 142, 275-291. Eosrg6M, L. and KUGELB~RG,E. (1968) Histochemical composition distribution of fibres, and fatignability of single motor units. Anterior tibial muscle of the rat. J. Neurol. Neurosurg. Psychiat. 31, 424-433. EKHOLM,J. (1967) Postnatal changes in cutaneous reflexes and in the discharge pattern of cutaneous and articular sense organs. Acta Physiol. Scand. (Suppl.) 297, 1-130. FALLON, J. H. and GI/LFMA~, C. E. (1989) Agrin-related molecules are concentrated at acetylcholine receptor clusters in normal and anenral developing muscle. J. Cell Biol. I ~ , 152%1535. FAUnaOOGH, D. M. (1979) Control of acetylcholine receptors in skeletal muscle. Physiol. Rev. 59, 165--227. FAMRROUGH,D. M. and RASh, J. E. (1971)Development of acetylcholine sensitivity during myogenesis. Dry. BioL 25, 55-68. FEDDE, M. R. (1969) Electrical properties and aeetylcholine sensitivity of singly and multiply innervated avian muscle fibres. J. Gen. Physiol. 53, 624--637. FEDULOVA, S. A., KOSTYUK,P. G. and VI~IELOVSKY,N. S. (1991) Ionic mechanisms of electrical excitability in rat sensory neurons during postnatal ontogenesis. Neuroscienee 41, 303-309. FISCn~^CH, G. D. and SCHU~TZE,S. M. (1981",)A postnatal decrease in acetylcholine channel open time at rat endplates. J. Physiol. (LoAd.) 303, 125-137. FLADnY,T. and JANSEN,J. K. (1990) Development of homogeneous fast and slow motor units in the neonatal mouse soleus muscle. Development 109, 723-732.

cular junction formation in vitro: induction of ACh receptor clusters in the postsynaptic membrane and morphology of newly formed synapses. J. Cell Biol. 83, 143-158. FRANK,E., JANSEN,J. K. S., LOMO,T. and WESTGAARD,R, H. (1975) The interaction between foreign and original motor nerves innervating the soleas muscle of rats. J. Physiol. (LoAd.) 247, 725-743. FglSCHKNECnT, R., NAVAR~rE, R. and VRaov•, G. (t990) The functional anatomy of the mammalian motor unit. In: Bailliire's Clinical Endocrinology and Metabolism. Muscle Metabolism, Volume 4/Number 3, pp. 401--416. Eds. J. B. HARRISand D. M. TtrRNBULL. Bailli6re Tindall, London. FROEHNER,S. C. (1986) The role of the postsynaptic cytoskeleton in AChR organization. TINS 9, 37-41. FAUNS, M., KRIra3ER,C. and SEARS,T. A. (1987) Identification and eiectrophysiological investigations of embryonic mammalian motoneurones in culture. Neurosci. Left. 113, 82-88. FULTON, B. P., MILEDI, R. and TAKAHASJI, T. (1980) Electrical synapses between motoneurones in the spinal cord of the newborn rat. Proc. R. Soc. London Ser. B 208, 115-120. FULTON, B. P. and WALTON, K. D. (1986) Electrophysiological properties of neonatal rat motoneurones studied in vitro. J. Physiol. (LoAd.) 370, 651~78. GATES, H.-J. and Rltx;E, R. M. A. P. (1992) The importance of competition between motoneurones in developing rat muscle: effects of partial denervation at birth. J. Physiol. (LoAd.) 445, 457-472. GIACOBINI-ROBECCHI, M. G., GIACORINI, E., FILOG^I~O, G. and CHANGEUX,J.-P. (1975) Effects of type A toxin from Clostridium botulinwn on the development of skeletal muscles and of their innervation in chick embryo. Brain Res. 83, 10%121. GILaEaT, M. and STELZNLm, D. J. (1979) The development of descending and dorsal root connections in the lumbosacral spinal cord of the postnatal rat. J. Comp. Neurol. 1114, 821-838. GOLDSPINK,G. and WARD, P. S. (1979) Changes in rodent muscle fibre types during postnatal growth, undernutrition and exercise. J. Physiol. 296, 453-469. GONOI, T., StiERMAN,S. and CATERALL,W. A. (1985) Voltage clamp analysis of tetrodotoxin-sensitive and insensitive sodium channels developing in vitro. J. Neurosci. 5, 2559--2564. GONOI, T., HAGIHARA, Y., KOBAYASHI,J., NAKAMURA, H. and OmzuMi, Y. (1989) Geographutoxin-scnsitive and insensitive sodium currents in mouse skeletal muscle developing in situ. J. Physiol. (Lond.) 414, 159-177. GONOl, T. and HASEGAW^, S. (1988) Postnatal disappearance of transient calcium channels in mouse skeletal muscle: effects of denervation and culture. J. Physiol. (Land.) 155, 13-25. GOROON, T., PERRY, R., SPtmWAY, N. C. and VItaov~., G. (1975) Development of slow and fast muscles in the chick embryo. Y. Physiol. (LoAd.) 254, 24-25P. GORDON,T., PURVF.S,R. D., and Vaaovg, G. (1977a) Differentiation of electrical and contractile properties of slow and fast muscle fibres. J. Physiol. (Land.) 269, 538--547. GORDON, T., PERRY, R., Sam^gl, T. and VRaOV,~, G. (1977b) Differentiation of slow and fast muscles in chickens. Cell Tiss. Res. 180, 211-222. GORDON, T., PmtRY, R., TUFt'ERe, A. R. and Vaaov~, G. (1974) Possible mechanisms determining synapse formation in developing skeletal muscles of the chick. Cell Tins. Res. !$$, 13-25. GO'ITLIEB, G. L., MYKLEBUST,B. M., PENN, R. D. and AGARWAL, G. C. (1982) Reciprocal excitation of muscle antagonists by the primary afferent pathway. Exp. Brain Res. 46, 454--456. GRAHAM-BROWN,T. (1915) On the activ/ties of the central nervous system of the unborn foetus of the cat; with a discussion of the question whether progression (walking, etc.) is a "learnt" complex. J. Physiol. (Lond.) 49, 208-215. GR~NSMI'rH, L. and VRaov~, G. (1991) Neuromuscular contacts in the developing rat soleas depend on muscle activity. Det,. Brain Res. 62, 121-129. GRINNELL, A. D. and HEag~RA, A. A. (1981) Specificity and plasticity of neuromuscular connections: long-term regulation of motonenron function. Prog. Neurobiol. 17, 203-282. GUNDERS~, K. (1990) Spontaneous activity at long-term silenced synapses in rat muscle. J. Physiol. (Land.) 4.1, 399-418. GUTMANN, E., MELICHN^,J. and Syaov¢/, 1. (1974) Developmental changes in contraction time, myosin properties and fibre pattern of fast and slow skeletal muscles. Physiol. Bohemoslov. 23, 19-27.

ACTIVITY-DEPENDENTINTERACTIONSBETWEEN MOTONEURONESAND MUSCLES HALL, Z. W. (1973) Multiple forms of acetylcholinestemse and their distribution in endpiate and non-andpiate regions of rat diaphragm muscle. J. Neurobiol. 4, 343-361. HAMaUIt~, V. (1934) The effects of wing bud extirpation on the development of the central nervous system in chick embryos. J. Exp. Zool. 68, 449--494. HASmtntOER, V. (1958) Regression versus peripheral control of differentiation in motor hypoplasia. Am. J. Anat. 102, 365-410. HAMaUIO~, V. (1963) Some aspects of the embryology of behavior. Quart. Roy. Biol. 3g, 342-365. HAMaUI~C,t~, V. (1977) The developmental history of the motor neuron. Neurosci. Prog. Res. Bull. I5, 1-37. HAMaU~O~, V. and BALAaAN,M. (1963) Observations and experiments on spontaneous rhythmical behavior in the chick embryo. Dee. Biol. 7, 533-545. HAMatraO~, V., BALAeAN,M., OPP~qHrJU, R. W. and WENO~t, E. (1965) Periodic motility of normal and spinal chick embryos between 8 and 17 days of incubation. J. Exp. Zool. 159, 1-13. HAMnUtO~, V., W~ZNO~,E. and Opp~rm~M, R. W. (1966) Motility in the chick embryo in the absence of sensory input. J. Exp. Zool. 162, 133-160. HAMMA~ei~G, C. and KEI.t~nTH,J.-O. (1975) The postnatal development of some twitch and fatigue properties of single motor units in the ankle muscles of the kitten. Aeta Physiol. Scand. 95, 243-257. HAtADA, Y. and TAKAHASm,T. (1983) The calcium component of the action potential in spinal motoneurones of the rat. J. Physiol. (Lond.) ~ , 89-100. HAmUS,J. B. and MARSHALL,M. W. (1973) Tetrodotoxin-resistant action potential in newborn rat muscle. Nature 243, 191-192. HAtU~ A. J., Duxsot~, M. J., Frrz~.o4o~s, R. B. and R.W~Et~, F. (1989) Myonuclear birthdates distinguish the origins of primary and secondary myotuhes in embryonic mammalian skeletal muscle. Development 107, 771-784. HAI~IS, J. B. and THESL~F,S. (1971) Studies on tetrodotoxin-resistant action potentials in denervated skeletal muscle. Acta Physiol. Scand. g3, 382-388. HENDItlKSON, C. K. and VAUGHN, J. E. (1974) Fine structural relationship between neurites and radial giial processes in developing mouse spinal cord. J. Neurocytol. 3, 659-675. HE~,nG, R. and I.~MO, T. (1985) Firing patterns of motor units in normal rats. Nature 314, 164-166. H~K, P., VV.JSADA,R. and KAS~¢KI,S. (1981) Reflex and locomotor changes following unilateral deafferentation of rat hindlimb asscsscd by chronic electromyography. Neuroscience 6, 195-203. HOLLYDAY, M. and HAMnU~OER,V. (1977) An autoradiographic study of the formation of the lateral motor column in the chick embryo. Brain Res. 132, 197-208. HUGH,, A. F. and Tsctlu~m, P. A. 0958) The factors controlling the development of the dorsal root ganglia and ventral horn in Xenopus laevis. J. Anat. 92, 498-527. HU6H~, S. M. and BLAU, H. M. (1992) Muscle fiber pattern is independent of cell lineage in postnatal rodent development. Cell, 68, 659-671. HUtZAt, P., Ktn~o, M. and Mn:^TA, Y. (1975) Differentiation of motonenrones and skeletal muscles in kittens. J. Physiol.(Lond.) 252, 465~79. Hut,~, R. I., ROLE, L. W. and F~SCHaACH,G. D. (1983) Acetylcholine release from growth cones detected with patches of acetylcholine-rich membranes. Nature 305, 632-634. ISmDA, I. and D~GUCHI,T. (1983) Effect of depolarizing agents on choline acetyl transferasc and acetylcholinesterase activities in primary cell cultures of spinal cord. J. Neurosci. 3, 1818--1823. IWAHARA, T., VAN HAKTI~VELDT, C., GARCIA-RILL, E. and SraNNE~, R. D. (1991) L-DOPA-induced air-stepping in deosrebrate developing rats. Dee. Brain Res. 58, 257-264. JAcoes, G. A. and W~KS, J. C. (1990) Postsynaptic changes at a sensory-to-motonenron synapse contribute to the developmental loss of a reflex behavior during insect metamorphosis. J. Neurosci. 10, 1341-1356. JAHt, C. E. and YOSmOKA, K. (1986) la afferent excitation of motonenrones in the in vitro new-born rat spinal cord is selectively antagonized by kynurenate. J. Physiol. (Load.) 370, 515. JANSEN, J. K. and FLADBY, T. (1990) The perinatal reorganization of the innervation of skeletal muscle in mammals. Prog. Neurobiol. 34, 39-9O

121

JA~LLO, F., VIc,~, S. and ~ , S. M. (1988) Embryonic acetylcholine receptors guarantee spontaneous contractions in rat developing muscle. Nature 335, 66-68. Joints, R. and VRaov/~, G. (1974) Two factors responsible for the development of denervation hypersensitivity. J. Physiol. (Load.) 236, 517-538. JONES, S. P., RIDGE, R. M. A. P. and R O A N , A. (1987) The non-selective innervation of muscle fibres and mixed composition of motor units in a muscle of nconatal rat. J. Physiol. (Lond.) 377-394. JORDAN,L. M. (I 983) Factors determining motonenron rhythmicity during fictive locomotion. In: Neura/origins ofrythmic movements. Society for Experimental Biology Symposium XXXVII, pp. 423-444. Eds. A. R o ~ a ~ a n d B. L. RO~'aTS.Cambridge University Press, Cambridge. KALLL~,R. G., S~mNO,Z. H., YANG,J., Crm~, L. Q., ROOART,R. B. and BAiCm, R. L. 0990) Primary structure and expression of a sodium channel characteristic of denervated and immature rat skeletal muscle. Neuron 4, 233-242. KANt, M., WAKUTA, K. and SATOU, R. (1987) Calcium channel components of action potential in chick skeletal muscle cells developing in culture. Dee. Brain Res. 32, 233-240. KANt, M., WAKU'rA,K. and SATOH,R. 0989) Two components of calcium channel current in embryonic chick skeletal muscle cells developing in culture. Dee. Brain. Res. 47, 101-112. KAP,AmAN,O. A., KOZHANOV,V. M. and C'~nfKHOVA,N. M. (1988) Interactions of motoneurons in the isolated perfused rat spinal cord. Neirnfiziologiio 20, 243-2.50. I~LLERTH, J.-O., MSLLS'ra6M, A. and SKOOLUND,S. (1971) Pustnatal excitabifity changes of kitten motoneurones. Acta Physiol. Scand. 83, 31--41. KELLY, A. M. and R~mnqs'rE~,N, A. (1980) Why are fetal muscles slow? Nature 288, 266--269. KELLY,A. M. (1983) Emergence of speciafization of skeletal muscle. In: Handbook of Physiology. Section 10, pp. 507-537. Ed. L. D. PF~. William and Wilkins, Baltimore. KELLY, A. M. and ZACr~ S. !. (1969) The fine structure of motor end plate morphogenesis. J. Cell Biol. 42, i 54-169. KELLY,S. S. (1978) The effect of age on neuromuscular transmission. J. Physiol. (Lond.) 274, 51-62. KERNELL, D. (1990) Spinal motoneurons and their muscle fibers: mechanisms and long-term consequences of common activation patterns. In: The Segmental Motor System, pp. 36-57. Eds. M. D. BINDERand L. M. M~'NDELL.Oxford University Press, New York. Kn3OKORO,Y. (1975) Sodium and calcium components of the action potential in a developing skeletal muscle cell line. J. Physiol. (Lond.) 244, 145-159. KI~-mN, O. (1991) Plateau potentials and active integration in the "final common pathway" for motor behaviour. TINS 14, 68--73. Ko, P. K., ANDEI~ON, M. J. and COHEN, M. W. (1977) Denervated skeletal muscle fibres develop discrete patches of high acetylcholine receptor density. Science 196, 540-547. KuDo, N. and YAMADA, T. (1987) Morphological and physiological studies of development of the monosynaptic reflex pathway in the rat lumbar spinal cord. J. Physiol. (Lond.) 389, 441-459. KUGELnO, E. (1976) Adaptive transformation of rat soleus motor units during growth. Histochemistry and contraction speed. J. Neurol. Sci. 27, 269-289. KUOELeL~tO, E. and EDSTR6M, L. (1968) Differential histochemical effects of muscle contractions on phosphorylase and glycogen in various types of fibres: relation to fatigue. J. NeuroL Neurosurg. Psychiat. 31, 415-423. KUpWR, C. and Ko~t~, G. B. (1951) A histochemical study of cbofinesterase during the formation of the motor endplate of the albino rat. J. Exp. Zool. 116, 399--413. KUROMI, H. and KIDOKORO, Y. (1984) Nerve disperses pre-existing acetylcboline receptor clusters prior to induction of receptor aceumulatation in Xenopus muscle cultures. Devel. Biol. 103, 53--61. LAeOVITZ, S. S. and RonmNS, N. (1983) A maturational increase in rat neuromuscular junctional acetylcboline receptors despite disuse or denervation. Brain Res. 266, 155-158. LAND t~'X~,, L. T. and MORmS, D. G. (1975) The development of functional ianervation in the hind limb of the chick embryo. J. Physiol. (Lond.) 249, 301-326.

122

R. NAVARRETE a n d G. VR~OVA

LANDMESSER,L. T. and O'DONOVAN, M. J. (1984) Activation patterns of embryonic chick hind limb muscles recorded #1 eve and in an isolated spinal cord preparation. J. Physiol. (Lend.) 347, 189-204. LANGM.~,N, J. and HADEN, C. C. (1970) Formation and migration of neuroblasts in the spinal cord of the chick embryo. J. Comp. Neurol. 138, 419-425. LEBER, S. M., BR~DLOVE, S. M. and SANES, J. R. (1990) Lineage, arrangement, and death of clonally related motonenrons in chick spinal cord. J. Neurosci. 10, 2451-2462. LEnEr.ER, E. and I~TTE, D. (1986) Neural regulation of parralbumin expression in mammalian skeletal muscle. Bioehem. J. 235, 67-73. LEE, M. T., KOEBBE, M. J. and O'DoNoVAN, M. J. (1988) The development of sensorimotor synaptic connections in the lumbosacral spinal cord of the chick embryo. J. Neurosci. 8, 2530-2543. I..,E~TZ, T. L. (1969) Development of neuromuscular junction. I. Cytological and cytochemical studies on the junction of differentiating muscle in the regenerating limb of the newt. Triturus. J. Cell Biol. 42, 431-443. LETINSKY, M. S. (1974) Physiological properties of developing frog tadpole nerve-muscle junctions during repetitive stimulation. Dry. Biol. 40, 154-161. LEVITT, T. A. and SALPET~R, M. M. (1981) Denervated endplates have a dual population of junctional acetylcholine receptors. Nature 291, 239-241. LEwis, D. M. (1973) The effect of denervation on the differentiation of twitch muscles in the kitten hindlimb. Nature 241, 285-286. LIDDELI., E. G. T. and SI-mntNGTON, C. S. (1925) Recruitment and some other factors of reflex inhibition. Prec. R. See. London Set. B 97, 488-518. LLIN/~S, R. R. (1984) Oscillation and resonance as the basic mechanism for the dynamic organization of the CNS. Trob. Inst. Cajal (Madrid). LLINAS, R. R. (1988) The intrinsic electrophysiological properties of mammalian neurons: insights into central nervous system function. Science 242, 1654-1664. Lolzot~, U A. (1972) The postnatal ontogeny of monoamine-conraining neurones in the central nervous system of the albino rat. Brain Res. 40, 395-418. LOMBET, A., KAZAZOOLOU, T., DEL~O~T, E., R~N^UD, J.-F. and LAZDUNSKt, M. (1983) Ontogenetic appearance of sodium channels characterized as high affinity binding sites for tetrndotoxin during development of the rat nervous and skeletal muscle systems. Biochem. Biophys. Res. Commun. 110, 894-901. LowRy, M. B., DnOOT, G. K. and VRaovg, G. (1988) The distribution of slow myosin in rat muscles after neonatal nerve crush. Muscle Nerve !1, 1043-1050. LOWgIE, M. B., MOORE, A. F. K. and VR~OVA, G. (1989) The effect of load on the phenotype of the developing rat soleus muscle. Pfliigers Arch. 415, 204-208. LOWRy, M. B. and Vg~ovk,, G. (1992) Dependence of postnatal motoneurones on their targets: a review and hypothesis. TINS 15, 75-80. LYONS, G. E., ON~LL, M., Cox, R., S^SSOON, D. and BUCKINGHAM, M. (1990) The expression of myosin genes in developing skeletal muscle in the mouse embryo. J. Cell Biol. 111, 1465-1476. M^GILL-SOLc, C. and MCM^H^N, U. J. (1988) Motor neurons contain agrin-like molecules. J. Cell Biol. 107, 1825-1833. MASABE, T., AgAKI, i., T a g a u ~ m , T. and Ktr~o, M. (1991) Membrane currents recorded from sexually dimorphic motoneurones of the bulbocavernosus muscle in neonatal rats. J. Physiol. (Lend.) 440, 419-43~. MARGRETH, A., DALLA LIm~RA, L., SALVIAT1, G. and ISCHIA, N. (1980) Spinal transection and the postnatal differemiation of slow myosin isoenzymes. Muscle and Nerve 3, 483-486. MARTIN, T, P., BODINE-FOWLER, S. C., ROY, R. R., ELDRED, E. E. and EDGERTON, V. R. (1988) Metabolic and fiber size properties of the rat tibialis anterior motor units. Am. J. Physiol. 255, 43-50. MATHER$, D. A. and THESL~, S. 0978) Studies on neurntrophic regulation of routine skeletal muscle. J. Physiol. (Lend.) 282, 105---I 14. McCoss, D. P., BEST, P. M. and BEAM, K. G. 0989) Development alters the expression of calcium currents in chick limb motoneutons. Neuron 2, 1633-1643.

McConn, D. P., BEST, P. M. and BEAM, K. G. (1990) The differentiation of excitability in embryonic chick limb motoneurons. J. Neurosci. 10, 2974--2984. MERLIE, J. P. and SMITH, M. M. (1986) Synthesis and assembly of acetylcholine receptor, a multisubunit membrane glycoprotein. J. Membr. 8ioL 91, 1-10. MECHLER,A. and SAKMANN,B. (1980) Receptor stability and channel conversion in the subsynaptic membrane of the developing mammalian neuromuscular junction. Devel. Biol. 80, I - I Z MILLEa, J. B. (1991) Myoblasts, myosins, MyoDs, and the diversification of muscle fibers. Neuromusc. Disor. !, 7-17. MILt.ER, J. B. and STOCKDALE, F. E. (1987) What muscle cells know that nerves don't tell them. TINS Ifi, 325--329. MOODY, S. A. and JA¢OaSOlq, M. (1983) Compartmental relationships between anuran primary spinal motoneurons and semitic muscle fibers that they first innervate..L Neurosci. 3, 1670--I 682. MOTORINA, M. V. (1989) Electronic synapses in the spinal cord of mammals. Neurosci. Behav. Physiol. 19, 72-78. NAKA, K. I. (1964) Electrophysiology of the fetal spinal cord. I. Action potentials of the motonenron. J. Gen. PhysioL 4% 1003-1022. NARAVANAN,C. H., FOx, M. W. and HAMnURGER,V. (1971) Prenatal development of spontaneous and evoked activity in the rat (Rattus norvegicus albinus). Behavior 40, 100-134. NARUSAWA, M., FITZStMONS, R. B., IZUMO, S., NADAL-GINARD,B., RUBtNSTEn~, N. A. and KELLY, A. M. (1987) Slow myosin in developing rat skeletal muscle. J. Cell Biol. 104, 447-459. NAVARP.ETE, R. and VRaovA, G. (1983) Changes of activity patterns in slow and fast muscles during postnatal development. Dry. Brain Res. 8, 11-19. NAVAXgXTE, R. and VRBOV.~, G. (1984) Differential effect of nerve injury at birth on the activity pattern of reinnervated slow and fast muscles of the rat. Y. Physiol. (Lend.) 3~1, 675-685. NAVASRE~, R. and VRBOVA, G. 0985) Precocious activation of locomotor activity by L-DOPA in the neonatal rat. Neurosci. Lett. (Suppl.) 22, $556. NAV^~RETE, R. and WALTON, K. D. (1989) Calcium conductances trigger doublet firing in neonatal rat motoneurones in vitro. J. Physiol. (Lend.)415, 70P. NAVARRETE, R., WALTON, K. D. and LuNAs, R. R. (1987) Spinal network development and its relation to hind-limb movement: an in vitro electrophysiological study in neonatal rat. Soc. Neurosci. Abstr. 13, 824. NAWRRI~TE, R., W^LTON, K. D. and LL~AS, R. R. (1988) Postnatal changes in the electrical properties of muscle- identified rat motoneurones: an in vitro study. Soc. Neurosci. Abstr. 14, 1060. NEAR, E. and SAKMA~, B. (1976) Single channel currents recorded from membrane of denervated frog muscle fibers. Nature 260, 799-802. NEMETH, P. M., ~TTE, D. and VgBov~,, G. (1981) Comparison of enzyme activities among single muscle fibres within defined motor units. J. Physiol. (Lend.) 311, 485--495. NITKIN, R. M., SMITH, M. A., MAGILL, C., FALLON, J. R., YAO, R.-M. M., WALL^CE, B. G. and McMAnAN, U. J. (1987) Identification of agrin, a synaptic organizing protein from Torpedo electric organ. J. Cell Biol. 195, 2471-2478. O'BaIEN, R. A. D., Os'rBEao, A. J. C. and VRaov/,, G. (1978) Observations on the elimination of polyneurnnal innervation in developing mammalian skeletal muscles. J. Physiol. (Lend.) 282, 571-582. O'BRIEN, R. A. D. and VanovA, G. (1978) Acetylcholine synthesis in nerve endings to slow and fast muscles of developing chicks: effect of muscle activity. Neuroseience 3, 122%1230. O'BRIEN, R. A. D., O S ~ O , A. J. C. and Va~ovA, G. (1982) The reorganization of neuromuscular junctions during development in the rat. In: Membranes in Growth and Development. pp. 24%257. Eds: J. H. HOFFMAN,G. H. GIEmSCHand L. BOLUS. Alan R Liss, New York. O'BaIEN, R. A. D., OSTaER6, A. J. C. and VRaovA, G. (1984) Protease inhibitors reduce the toss of nerve terminals induced by activity and calcium in developing rat soleus muscles in vitro. Neuroscience 3, 122%1230. O'BI~.IEN,R. J. and FtSCnaACH, G. D (1986a) Isolation of embryonic chick motoneurons and their survival in vitro. J. Neuroscl. 6, 3265-3274.

ACTIVITY-DEPENDENTINTERACTIONSBETWEEN MOTONEURONESAND MUSCLES O'Bm~N, R. J. and FISOmACH,G. D. (1986b) Characterization of excitatory amino acid receptors expressed by embryonic chick motoneuroncs in vitro. J. Neuroscl. 6, 3275-3283. O'BmL~, R. J. and FtSCHaACH,G. D, (1986c) Excitatory synaptic transmission between interneurons and motoneurons in chick spinal cord cell cultures. J. Ncurosci. 6, 3284-3289. O'BmEN, R. J. and FmCHBACH, G. D. (1986d) Modulation of embryonic chick motnncuron glutamate sensitivity by intcrncurons and agnnists. J. Nearosci. 6, 3290-3296. O'DONoVAN,M. J. and LANDtm~X~, L. (1987) The development of hindlimb motor activity studied in the isolated spinal cord of the chick embryo. J. Neurosci. 7, 3256-3264. OP~I4L~M, R. W. (1975) The role of supraspinal input in embryonic motility: a re*examinationin the chick. J. Comp.Neurol. 160, 37-50. OPPENtlEIM, R. W. (1991) Cell death during the development of the nervous system. Ann. Rev. Neurosci. 14, 453-501. Op~I-~IM, R. W. and Ct~-W^No, J. W. (1983) Aspects of naturnily occurring motoneurone death in the chick spinal cord during embryonic development. In: Somatic and Autonomic Nerve-Muscle Interactions, pp. 57-107. Eds. G. BURNSTOCK,R. A. D. O'Bgl~ and G. VRBov£. Elsevier, Amsterdam. O'SULLIVAN,M. C., E ~ , J. A. and MILt.~, S. (1991) Radiation of phasic stretch reflex in biceps brachii to muscles of the arm in man and its restriction during development. J. Physiol. (Lond.) 439, 529-543. PENO, H. B., ~ G , P. C. and LUT~R, P. W. (1981) Formation of A C h receptor clusters induced by positivelycharged latex beads. Nature 292, 831-834. P'~o, H. B. (1987) Development of the neuromuscular junction in tissue culture. C R C Crit. Rev. Anat. Sci. I, 91-131. F~i~, D. and STARON, R. S. (I990) Cellularand molecular diversities of mammalian skeletal muscle fibres. Rev. Physiol. Biochem. Pharmacol. 116, 1-76. Fei ~, D. and Vleov~, G. (1992) Adaptation of mammalian skeletal muscle fibers to chronic electricalstimulation. Rev. Physiol. Biochem. Pharmacol. (in Press). PHELPS, P. E., BARBER, R. P., BRENNAN, L. A., MAINES,Y. M., SALVATEtaA,P. M. and VAUOHN,J. E. (1990) Embryonic development of four different subsets of cholinergic neurons in rat cervical spinal cord. J. Comp. Neurol. 291, 9-26. Pm~Lps, P. E., BARaEX, R. P., Hous~g, C. R., C-'XAWVOXD,G. D., SALW,TEI~RA,P. M. and VAUGHN,J. E. (1984) Postnatal developmerit of neurons containing choline acetyltransferase in rat spinal cord: an immunocytochemieal study. J. Comp. Neural. 229, 347-361. PHILLIPS,W. D. and BENNETT,M. R. (1984) Differentiation of fiber types in wing muscles during embryonic development: effect of neural tube removal. Day. Biol. 106, 457-468. I~LAR, G., TUTTLE,J. and VAC^, K. (1981) Functional maturation of motor nerve terminals in the avian iris: ultrastructure, transmittar metabolism and synaptic reliability. J. Physiol. (Lond.) 321, 175-193. POZNANSKY, M. J. and STEELE,J. A. (1984) Membrane electrical properties of developing fast-twitch and slow-tonic muscle fibres of the chick. J. Physiol. (Lond.) 347, 633-640. [htr~tlTL, H. F. R. (1974) The behavioural states of the newborn infant (a review). Brain Res. 76, 185-212. PltECt~TL, H. F. R. (1984) Continuity and change in early neural development. In: Continuity of Neural Functions from Prenatal to Postnatal Life. Clinics in Developmental Medicine, Vol. 94, pp. 1-16. Ed. H. F. R. P~CHTL. Blackwell, Oxford. Pltl~f~lt, W. (1885) Spezielle Physiologie des Embryo. Grieben, Leipzig. 644 pp. PtovlNl~, R. R. (1971) Embryonic spinal cord: synchrony and spatial distribution of polynenronal burst discharges. Brain Res. 29, 155-158.

PaOVlNE, R. R. (1976) Development of function in nerve nets. In: Simpler networks and behavior, pp. 203-220. Ed. J. F~qTItl~. Sinauer, Sunderland, Mass. PURVF.S,D. and LICHTMAN,J. W. (1985) Principles of neural development. Sinauer Associates Inc., Sunderland, Massachusetts. 433 pp. RAJAOFETRA,N., SANDILLON,F., GEFFAgD,M. and PRIVAT,A. (1989) Pre- and post-natal ontogeny of serotonergic projections to the rat spinal cord. J. Neurosci. Res. 22, 305-321. REDFER~, P. and THESLEFF,S. (1971) Action potential generation in denervated rat skeletal muscle. II. The action of tetrodotoxin. Acta Physiol. Scand. 82, 70-78.

123

RZDI~Ixl~,P. A. (1970) Neuromuscular transmission in newborn rats. J. Physiol. (Lond.) 209, 701-709. REINOLD, E. (1976) Ultrasonics in early pregnancy. Diagnostic scanning and fetal motor activity. In: Contributions to Gynecology and Obstetrics. Vol I, pp. 102-148. Ed. P. J. KELLER.S. Karger, Basel. RmcHizrrat, R. R. and TAXT,T. (1983) Motor unit size and synaptic competition in rat lumbrical muscle reinnervated by active and inactive motor axons. J. Physiol. (Lond.) 344, 89-111. RmctIEST~, R. R. and TAXT,T. (1984) Repression of inactive motor nerve terminals in partially denervated rat muscle after regeneration of active motor axons. J. Physiol. (Lond.) 347, 497-51 I. RtDOE, R. M. A. P. (1992) The differentiation of conduction velocities of slow twitch and fast twitch muscle motor inncrvation in kittens and eats. Quart. J. Exp. Physiol. 52, 293-304. RIPLEy, K. L. and Paovn~, R. R. (1972) Neural correlates of embryonic motility in the chick. Brain Res. 45, 127-134. RONNEW,L. O. (1979) Spontaneous phagocytosis of C-type synaptic terminals on spinal ¢z-motoneuronsin newborn kittens. An electron microscopic study. Brain Res. 162, 189-199. Ross, J. J., Duxsos, M. J. and HARaL%A. J. (1987) Formation of primary and secondary myotubes in rat lumbrical muscles. Development 100, 383-394. Ruby, B. (1988) Diversity and ubiquity of K + channels. Neuroscience 2.$, 729-749. RUTZ, R. and HAUSCHKA,S. D. (1982) Clonal analysis of vertebrate myogenesis. VII. Heritability of muscle colony type through sequential subcional passages in vitro. Devl. Biol. 91, 399-411. SABRY,M. A. and DHOOT,G. K. (1991) Identification and patterns of transitions of cardiac, adult slow and slow skeletal muscle-like embryonic isoforms of Troponin T in developing rat and human skeletal muscles. J. Muscle Res. Cell Mot. 12, 262-270. S^KMANN, B. (1978) Acetylcholine induced ionic channels in rat skeletal muscle. Fed. Proc. 37, 2654-2659. SAKMANN,B. and Bt~VNER,H. R. (1978) Change in synaptic channel gating during neuromuscular development. Nature 276, 401--402. SALPET~, M. and LORING, R. H. (1985) Nicotinic acetylcholinc receptors in vertebrate muscle: properties, distribution and neural control. Pron. Neurobiol. 25, 297-325. S^NES, J. R. (1989) Extracellular matrix molecules that influence neural development. Ann. Ray. Neurosci. 12, 491-516. SCHUETZE,S. M. and ROLE,L. W. (1987) Developmental regulation of the nicotinic acetylcholine receptor. Ann. Rev. Neurosci. 10, 403-457. S C ~ T , P. C. and CRILL,W. E. (1984) Membrane properties of cat spinal motoneurons. In: Handbook oftbe spinal cord, pp. 199-242. Ed. R. A. DAVlDOEF.Marcel Dekker, New York. SERNAOOR, E. and O'DoNovAN, M. J. (1991) Whule-cell patch clamp recordings from rhythmically active motoneurons in the isolated spinal cord of the chick embryo. Neurosci. Lent. 128, 211-216. SHEARD,P. W., DUXSON,M. J. and HARtlS, A. J. (1991) Neuromuscular transmission to identified primary and secondary myotubes: a re-evaluation of polyneuronal innervation patterns in rat embryos. Day. Biol. 148, 459-472. S t ~ A ~ , S. J. and CAYrERALL,W. A. 0982) Biphasic regulation of development of the high affinity saxitoxin receptor by innervation in rat skeletal muscle. J. Gen. Physiol. 80, 753-768. SHERMAN,S. J. and CATn~aALL,W. A. (1984) Electrical activity and cytosolic calcium regulate levels of tetrodotoxin-sensitive sodium channels in cultured rat muscle cells. Proc. Natl. Acad. Sci. U.S.A. gl, 262-266. SHOtEY, M. L. (1909) The effects of the destruction of the peripheral areas on the differentiation of the neuroblasts. J. Exp. Zool. 7, 25-63. SKET~tJ, J. and BRZlN, M. (1985) Asymmetric molecular forms of acetylcholinesterase in mammalian skeletal muscle. J. Neurosci. Res. 14, 95-103. SKETEI.J,J., CRNE-FINDERLE,N., R1BARIC,S. and BRZIN, M. (1991) Interactions between intrinsic regulation and neural modulation of acetylcholinesterase in fast and slow skeletal muscles. Cell. Mol. Neurobiol. II, 35-54. SLATER,C. R. (1982) Postnatal maturation of nerve-musclejunctions in hindlimb muscles of the mouse. Day. Biol. 94, 11-22. SOSlA, J. M., CALLAWAY,E. M. and VAN ESSEN,D. L. (1989) Lack of fiber-type selectivity during reinnervation of neonatal rabbit solens muscle. Dev. Biol. 131, 401-415.

124

R. NAVARRETE and G. VrtBovg

SOILEAU,L. C., SILBERSTEIN,L., BLAU,H. M. and THOMPSON,W. J. (1987) Rcinncrvation of muscle fiber types in the newborn rat soleus..I. Neurosci. 7, 4176-4194. SRmARI, T, and VRnOV~,G. (1978) The role of muscle activity in the differentiation of neuromuscular junctions in slow and fast chick muscles. J. Neurocytol. 7, 529-540. SRIHARI, T. and VRsovX, G. (1980) Effects of neuromuscular blocking agents on the differentiation of nerve-muscle connections in slow and fast chick muscles. Devel. Growth Differ. 4, 645--657. Sr~WMAYER, K., ORTLAND,C. and JENTSCH,T. J. (1991) Primary structure and functional expression of a developmentally regulated skeletal muscle chloride channel. Nature 354, 301-304. STELZNER, D. J. (1971) The normal postnatal development of synaptic end-feet in the lumbosacral spinal cord and of responses in the hindlimbs of the albino rat. Expl. Neurol. 31, 337-357. STELZN~, D. J. (1982) The role of descending systems in maintaining intrinsic spinal function: a developmental approach. In: Brain Stem Control of Spinal Mechanisms, pp. 297-321. Eds. B. SJOLUND and A. BJORKLUND.Elsevier, Amsterdam. SVrKOWSKI,A. J., VOGEL,Z. and NIm~Ns~G, M. W. 0973) Development of acetylcholine receptor dusters on cultured muscle cells. Proc. Natl. Acad. Sci. USA 70, 270-274. TA~:AHASm, T. (1990) Membrane currents in visually identified motoneurones of neonatal rat spinal cord. J. Physiol. (Lond.)423, 27-46. TENNVSON, V. M., BRZIN, M. and SLOTWlNmL P. 0971) The appearance of acetylcholinesterase in the myotome of the embryonic rabbit. An electron microscope cytochemical and bioehemica! study. J. Cell Biol. 51, 703-721. TENNYSON,V. M., BRZIN, M. and KIt~MZ~.X, L. T. (1973) Acetylcholinesterase activity in myotube and muscle satellite cells of the foetal rabbit: an electron microscopic--cytoehemical and biochemical study. J. Histochem. Cytochem. 21, 634-652. THOMPSON,W. J., SUTTON,L. A. and PalLS, D. A. 0984) Fiber type composition of single motor units during synapse elimination in neonatal rat soleos muscle. Nature 309, 707-711. TOTH, V. L. and KARCSU,S. (1979) UItrnstructural localization of the acetylcholinesterase activity in the diaphragm of the rat embryo. Acta Histochem. (Jena) 64, 148-156. TOUT^NT, J. P. and MAS~,ULn~,J. (1987) Acetylcholinesterase. In: Mammalian Ectoenzymes, pp. 289-328. Ed. K£NNy and TtrRNL~. Elsevier Science Publishers BV. ULFHAKE, B., CULLHEIM, S. and FRANSON, P. (1988) Postnatal development of cat hind limb motoneurons. I: Changes in length, branching structure, and spatial distribution of dendrites of cat triceps surae motoneurons. J. Comp. NeusoL 278, 69-87. VACA,K. 0988) The development of cholinergic neurons. Brain Res. Rev. 13, 261-286. VANDENBURGH,H. H., HATFALUDY,S., KARLISCH,P. and SHANSKY, J. (1990) Mechanically-induced alterations in cultured skeletal myotube growth. In: The Dynamic State of Muscle Fibres, pp 151-164. Ed. D. P~TE. De Gruyter, Berlin, New York. VF~SADA, R., HNiK, P., NAVARRETE, R., PALECEK, J., SOUKUP, T., BO~CKA, U. and P^YNE, R. (1991) Motor functions in rat hindlimb muscles following neonatal sciatic nerve crush. Nearoscience 40, 267-275. VIAN^, F., BAWLISS,A. and B~gG~R, A. J. (1991) Ontogenesis of motoneuron electrical properties: role of calcium and calciumactivated conductances. Soc. NeuroscL Abstr. 17, 464. VIGNV, M., KOENIG,J. and RIF.OER, F. (1976) The motor endplate specific form of acetylcbolinesterase: appearance during embryogenesis and reinnervation of rat muscle. J. Nearochem. 72, 1347-1353. VOGEL, M. W., ENGLISH,A. W and HEgRtJP, K. 0988) Chimeric analysis of lineage relationships among mouse lumbar motonautons. Soc. Nearosci. Abstr. 14, 1130.

VReov~, G. (1963) Changes in the motor reflexes produced by tenotomy. J. Physiol. (Lond.) 166, 241-250. VRSOVA,G. (1980) lnnervation and differentiation of muscle fibres. In: Development and specialization of skeletal muscle, pp. 37-50. Ed. D. F. GOLDSPINK.Cambridge University Press, Cambridge. VRSOV,~, G., GORDOI~,T. and JONES,R. (1978) Nerve-muscle interaction. Chapman and Hall, London. 233 pp. VRSOV.~,G., LOWRm,M. B. and EVERS,J. (1988) Reorganization of synaptic inputs to developing skeletal muscle fibres. In: Pla,rticity of the Neuromuscular System, CIBA Foundation Symposium 138, pp. 131-142. Eds. D. EVFa~D and J. WHELAN.Wigley and Sons. VRIES,J. I. P., Vtssr~, G. H. A. and PI~CHTL, H. F. R. 0982) The emergence of fetal behaviour. I. Qualitative aspects. Early Hum. Dee. 7, 301-322. Vgms, J. I. P., VlSSER,G. H. A. and PSECHTL, H. F. R. (1985) The emergence of fetal behaviour. It. Quantitative aspects. Early Hum. Dee. 12, 99-120. WALTON, K. D. and FULTON, B. P. (1986) The ionic mechanisms underlying the firing properties of rat neonatal motoneurnns. Neuroscience 19, 669-683. WALTON, K. D. and LLIN.~S, R. R. (1986) Calcium-dependent low threshold rebound potentials and oscillatory potentials in neonatal rat spinal motoneurons in vitro. Soc. Nearosci. Abstr. 12, 386. WALTON, K. D. and NAVARn.~"rE,R. (1991) Postnatal changes in motoneurone electrotonic coupling studied in the in vitro rat spinal cord. J. PhysioL (Lond.) 433, 283-305. WEISS, R. and HORN, R. (1986) Functional differences between two classes of sodium channels in developing rat skeletal muscle. Science 233, 361-364. WI~TERGA, J. and GRAMgBERGEN,A. (1990) The development of locomotion in the rat. Dee. Brain Res. 57, 163-174. WEYDERT,A., BARTON,P., HARRIS,A. J., PINSEr,C. and BUCKINGHAM, M. (1987) Developmental pattern of mouse skeletal myosin heavy chain gene transcripts in vivo and in vitro. CeU49, 121-129. WINDLE, W. F. (1940) Physiology of the Foetus. Origin and Extent of Function in Prenatal Life. W. B. Saunders and Co., Philadelphia. 244 pp. WINDL~, W. F. and BAXTER, R. E. (t936) Development of reflex mechanisms in the spinal cord of albino rat embryos. Correlations between structure and function and comparisons with the rat and the chick. J. Comp. NearoL 63, 189-209. WITZEMANN,V., BARG,B., CRI^DO, M., STEIN,E. and SAgMANN,B. (1989) Developmental regulation of five subunit specific mRNAs encoding acetylcholine receptor subtypes in rat muscle. FEBS Lett. 242, 419-424. XIV~Z. X and Poo, M.-M (1986) Initial events in the formation of neuromuscular synapse: Rapid induction of aeetylcholine release from embryonic neuron. Proc. Natl. Acad. Sci. USA 83, 7069-7073. YAARI,Y., HAMON,B. and Lux, H. D. (1987) Development of two types of calcium channels in cultured mammalian hippocampal neurons. Science 235, 680-682. YOUNG, S. H. and Poo, M.-M. 0983) Spontaneous release of transmitter from growth cones of embryonic neurones. Nature 305, 634. ZELENA,J. and SZ~NTAGOTHAI,J. (1957) Verlagerang der Lokalisation spezifischer Cholinesterase wihrend der Entwicktung der Muskel Innervation. Acta Histochem. 3, 248-296. ZHU, P-H. and Vlteov~, G. (1992) The role of Ca ++ in the elimination of polyneuroual innervation of rat soleus muscle fibres. European J. Neurosci. 4, 433-437. ZISKIND-COSHAIM, L. (1988) Physiological and morphological changes in developing peripheral nerves of rat embryos. Dee. Brain Res. 42, 15-28. ZISKIND-CGNtlAIM,L. 0990) NMDA receptors mediate poly- and monosynaptic potentials in motoneurons of rat embryos. J. Neurosci. 10, 125-135.