Examination of intrafascicular muscle fiber terminations: implications for tension delivery in series-fibered muscles.

Mammalian skeletal muscles with long fascicle lengths are predominantly composed of short muscle fibers that terminate midbelly with no direct connection to the muscle origin or insertion. The manner in which these short fibers terminate and transmit tension through the muscle to their tendons is poorly understood. We made an extensive morphological study of a series-fibered muscle, the guinea pig sternomastoid, in order to define the full range of structural specializations for tension transmission from short fibers within this muscle. Terminations were examined in single fibers, teased small bundles of fibers, and in sections at both the light and electron microscopic level. In many cases, sites of fiber termination were defined by reactivity for the enzyme acetylcholinesterase, which also marks myotendinous junctions. Additionally, transport of the lipophilic fluorescent dye, DiI, or injection of Lucifer Yellow were used to visualize undisturbed fiber terminations in whole muscles using confocal and fluorescence microscopy. At the light microscopic level, we find that intrafascicularly terminating fibers end about equally often in either a long progressive taper, or in a series of small or larger blunt steps. Combinations of these two morphologies are also seen. However, when analyzed at higher resolution with confocal or electron microscopy, the apparently smooth progressive tapers appear also to be predominantly composed of a series of fine stepped terminations. Stepwise terminations in most cases join face-to-face with complementary endings of neighboring muscle fibers, some via an extended collagenous bridge and others at close interdigitating myomyonal junctions. These muscle-to-muscle junctions show many of the features of myotendinous junctions, including dense subsarcolemmal plaques in regions of myofibrillar termination and we suggest that they serve to pass tension from fiber to fiber along the longitudinal axis of the muscle. In addition, we observe regions of apparent side-to-side adhesion between neighboring fibers at sites where there is no apparent fiber tapering or structural specialization typical of myofibril termination. These sites show acetylcholinesterase reactivity, and large numbers of collagen fibers passing laterally from fiber to fiber. These latter connections seem most likely to be involved in lateral transmission of tension, either from fiber to fiber, or from fiber to endomysium. Overall, our results suggest that tension from intrafascicularly terminating fibers is likely to be passed along the muscle to the tendon using both in-series and in-parallel arrangements. The results are discussed in light of current theories of tension delivery within the series-fibered muscles typical of large, nonprimate mammals.

Differences in pituitary cell number but not cell type between genetically lean and fat Coopworth sheep.

Coopworth sheep selected for low backfat (lean genotype) have been shown to have heavier pituitary glands than those selected for high backfat (fat genotype). This paper investigated whether this difference was due to an increase in pituitary cell number or cell size and whether the relative proportions of different pituitary cell types differed between the genotypes. In three separate trials, ram lambs aged 6 to 8 months were slaughtered and the pituitary glands were processed for stereological or immunocytochemical studies. The pituitary glands of lean genotype sheep were between 30 and 60% heavier than those of the fat sheep. Lean sheep had a significantly (P<0.05) larger cross-sectional area of the pituitary fossa (96.6 vs. 81.7 mm(2)) than fat genotype sheep. The pituitaries from lean sheep contained significantly more total cells than fat sheep (Trial 1: 290 vs. 183 million cells, P<0.01; Trial 2: 353 vs. 239 million cells, P <0.05). The volume of individual cells did not differ between the genotypes. Trial 3 showed that there was no difference between lean and fat sheep in the percentage of cells staining positive for the five pituitary hormones studied. It is concluded that the larger pituitary glands of lean compared to fat genotype sheep are a result of a nonspecific increase in the size of the whole gland through increased cell numbers, with no change in cell size or the relative proportion of different cell types.

Early development of the myotome in the mouse.

The structure and development of the myotome has been extensively studied in birds and amphibians but few studies have systematically addressed its development in mammals. We have used a transgenic mouse carrying an nLacZ marker coupled to a myosin light chain 3F promoter to describe the structure of the developing mammalian myotome. Through studies of transgene expression pattern, coupled with immunohistochemistry for the muscle structural proteins desmin and slow myosin heavy chain we describe a gradient of maturity for the cells within the developing myotome. Our results show that the earliest myocytes of the mammalian myotome span the rostrocaudal extent of the somite and have single large nuclei which localise centrally within the myotome. Throughout the period of study the myotome is more mature ventrally than dorsally and cells comprising the medial aspect of the myotome are younger than those lying laterally. Immunohistochemistry for the earliest expressed muscle regulatory factor (myf-5) is used to define areas of the myotome contributing new myogenic cells. In the early myotome small, round, myf-5-expressing cells are found extensively within the dorsomedial aspect of the dermamyotome and also within the entire rostral and caudal dermamyotomal lips. They subsequently appear within the central zone of the myotome, adjacent to the medially curled rostral and caudal dermamyotomal lips, and there begin to elongate symmetrically. As the myotome enlarges, myf-5 expression is always restricted to the most medial aspect of the myotome, adjacent to the least mature myocytes, marking the site of addition of new myogenic cells. Together, these results allow development of a model of mammalian myotome formation where growth occurs medially by addition of new cells from both rostral and caudal dermamyotome lips, while more mature myocytes are displaced laterally. Furthermore, early myotomal myocytes differentiate in the absence of MyoD expression, unlike later myotomal myocytes. This, along with their distinct morphology, suggests these cells may form a separate lineage of pioneer myogenic cells.

Alpha motoneurons are present in normal numbers but with reduced soma size in neurotrophin-3 knockout mice.

Neurotrophin-3 (NT-3) is essential for survival of proprioceptive and mechanoreceptor neurons, but its role in motoneuron development in vivo has seemed slight. Recent evidence however has suggested that NT-3 may be involved in motoneuron maturation. Here, we quantitatively assess the number and state of development of motoneurons within the lumbar lateral motor column (LLMC) of newborn NT-3 (-/-), (+/-) and (+/+) mutant mice. We find that the number of alpha motoneurons in the LLMC is the same in all genotypes, but soma size is significantly reduced in (-/-) mutants. This suggests that NT-3 is not required for normal production and survival of alpha motoneurons, but is essential for their full maturation and/or maintenance. We also confirm a previous report that gamma motoneurons are absent in NT-3 (-/-) mice.

Variations in oxidative enzyme type profiles among prenatal rat lumbar motoneurons.

We have used cytochrome oxidase histochemical staining to evaluate whether immature rat lumbar motoneurons show intrinsic separation into high or low oxidative enzyme types. Relative oxidative enzyme levels are frequently used to help differentiate between muscle fibres of various types and to differentiate between mature neurons. Here we show a wide variation in motoneuron cytochrome oxidase levels from prenatal times, although the range of staining levels as measured densitometrically is greater for mature than for prenatal animals. We find variation in cytochrome oxidase levels among motoneurons prior to the formation of mature patterns of connectivity or electrical activity, and conclude therefore that this differentiation is unlikely to have arisen by differential usage and probably arose as a function of cell lineage.

The transient existence of 'en passant' nerve terminals in normal embryonic rat skeletal muscle.

In a recent examination of labeled normal embryonic rat nerve terminals we have found evidence of synaptic specialisation at non-terminal regions of axons ('en passant' terminals). The specialisation typically consists of a vesicle-filled axonal swelling lying in close apposition to a zone of thickened sarcolemma, with the two structures separated by a basal lamina-filled cleft. Such 'en passant' terminals are never seen in normal mature rat muscle, where all synaptic sites form at the true termination of axonal branches. These terminals thus appear to be a transient feature of developing neuromuscular relations. Analogous structures may arise in mature muscle when axons undergo ultraterminal sprouting in response to paralysis or partial denervation. We discuss our finding in relation to the role such terminals may play in normal development and in partially denervated mature muscle.

Composition of newly forming motor units in prenatal rat intercostal muscle.

We have examined the composition of rat intercostal motor units during the period of late gestation, when most muscle fibres are formed, in order to see the pattern of the contacts initially made between single motoneurons and myotubes. At this early stage, the muscle contains two types of myotubes, primary and secondary myotubes, and a major aim was to see whether individual motoneurons preferentially made contact with a particular myotube type. The technique used to define myotubes contacted by a single motoneuron was anterograde labelling of the neuron, followed by electron microscopic detection of labelled terminals and their postsynaptic targets. We find that prenatal motor units are inhomogeneous with respect to their primary/secondary myotube composition. Most individual motoneurons show many permutations of contact with primary myotubes, secondary myotubes, and undifferentiated cells, including single nerve terminals which contact both primary and secondary myotubes. Our results are interpreted in terms of changes to the composition of both the muscle and of the motor units during the final 5 days of gestation. We demonstrate that motoneurons necessarily make their initial contacts on primary myotubes, but that these are surprisingly sparse. As secondary myotubes appear and become innervated, motor units are at first all similar and all heterogeneous. However, primary myotubes are represented more often in motor units than in the muscle as a whole. This probably reflects the relative densities of polyinnervation of primary vs. secondary myotubes. By embryonic day 20, motor units have become divergent in composition, with some dominated by primary myotubes and others by secondaries. We propose that motoneurons initially establish contacts at random on either myotube type, but then begin to express preference for one type or the other and reorganise their periphery. Refining of motor unit composition towards homogeneity in the postnatal period probably involves other elements, such as mutability of muscle fibre and/or motoneuron characteristics as a function of usage and muscle position, perhaps influenced by sensory feedback mechanisms.

Formation of new myotubes occurs exclusively at the multiple innervation zones of an embryonic large muscle.

This work examines the general principle of whether production of embryonic muscle fibres is invariably linked to sites of innervation, as we have previously reported in small rodent muscles (Duxson et al. [1989] Development 107:743-750). The experimental strategy has been to make a detailed electron microscopic analysis of the formation of new myotubes in a large muscle having multiple, discrete innervation zones. The particular model system is the guinea pig sternomastoid muscle, a strap-like, parallel-fibred muscle with four distinct endplate bands, both in the embryo and the adult. Primary myotubes in the developing muscle extended from tendon to tendon of the muscle and were innervated at each of the multiple endplate zones. Each point of innervation of the primary myotubes was a focus around which many new secondary myotubes formed, and each secondary myotube was approximately centred on one of the innervation sites of its supporting primary myotube. This confirms our previous report, in rat IVth lumbrical muscle, of an invariable association between sites of formation of new secondary myotubes and sites of innervation. We suggest that, in vivo, nerve terminals either directly induce the initial myoblast fusions which give rise to new secondary myotubes, or induce some precondition for fusion. An alternative hypothesis is that a common patterning influence in the muscle localizes both innervation and secondary myotube formation to the same zone. The pattern of secondary myotube production in the embryo has important implications for the size and final architecture of muscles in larger animals, and some of these are discussed.

The relationship of nerve to myoblasts and newly-formed secondary myotubes in the fourth lumbrical muscle of the rat foetus.

The formation of normal numbers of skeletal muscle fibres depends on functional innervation of the muscle before and during the period of secondary myotube formation, but little has been known about the physical relationship between nerve terminals and the myoblasts and secondary myotubes over the critical period. This paper reports the results of a serial-section electron microscopic study of the IVth lumbrical muscle of the rat hindlimb, studied on embryonic day 20 (E20), a time when all secondary myotubes are less than 24 h old, and new ones are rapidly forming. Most myoblasts lying within the endplate region of the muscle received some direct neural contact; in almost all cases, the contact originated from an extension of a differentiated nerve terminal present at the endplate of an adjacent primary myotube. At six of 15 neural contact sites on myoblasts, primitive synaptic specialization was present. The newly-formed secondary myotubes were also directly, although sparsely, innervated in nine of ten instances. One secondary myotube was never seen to be innervated, despite extensive serial tracing. Nerve terminals passing to secondary myotubes were also principally derived from the innervation zone of the earlier-formed primary myotubes. Primary myotubes were profusely innervated by multiple axons. The results suggest that most nerve terminals are initially accommodated on the primary generation of myotubes, but progressively transfer to pre-fusion myoblasts or to secondary myotubes as these appear. In general, very young secondary myotubes are innervated by only a single axon, rather than being polyneuronally innervated. The existence of some secondary myotubes which lack any direct innervation suggests that intimate nerve contact may not be obligatory for formation of new secondary myotubes.

Neuromuscular transmission to identified primary and secondary myotubes: a reevaluation of polyneuronal innervation patterns in rat embryos.

The early morphogenesis of rat skeletal muscle is a biphasic process involving two sequentially generated populations of myotubes. A small population of primary myotubes appears early and is followed by a much larger population of secondary myotubes which appear progressively over a number of days. All previously published electrophysiological studies of developing muscle have failed to appreciate the relevance of biphasic myotube production. Here we reevaluate the status of early motor innervation, taking into account the wide range of sizes and levels of maturity within the two myotube populations. Evoked end-plate potentials (EPPs) were recorded from fibers of E17-20 rat sternocostalis muscles. Impaled fibers were then marked by ejection of HRP from the recording pipet, enabling ultrastructural identification of fibers from which recordings had been made. The average number of synaptic inputs per fiber increased to a peak at E19, and the rate of rise of the EPPs increased with age. The majority of impaled fibers (76%) were subsequently found to be primary myotubes, even at ages when secondary myotubes formed the majority of fibers in the muscle. Electrophysiological studies during early stages of secondary myotube development therefore sample largely from the more mature primary fibers and probably give the wrong impression of the extent and degree of polyneuronal innervation and of synaptic rearrangement within the muscle. In addition, the results show that very young secondary myotubes are distinguished by EPPs of longer latency, slower rate of rise, and smaller size than those of other types of myotubes. These results suggest that young secondary myotubes are predominantly activated by EPPs which originate in adjoining primary myotubes and propagate electronically to the secondary myotube. We propose a new model of early synaptic rearrangement which accommodates the biphasic nature of muscle development. We also suggest that secondary myotubes do not require direct neural input for the initiation of their development.

The origin of secondary myotubes in mammalian skeletal muscles: ultrastructural studies.

The distribution of secondary myotubes and undifferentiated mononucleated cells (presumed to be myoblasts) within foetal IVth lumbrical muscles of the rat was analyzed with serial section electron microscopy. In all myotube clusters for which the innervation zone was located, every secondary myotube overlapped the end-plate region of the primary myotube. No secondary myotubes were ever demonstrated to occur at a distance from the primary myotube innervation zone. This indicates that new secondary myotubes begin to form only in the innervation zone of the muscle. Some young secondary myotubes made direct contact with a nerve terminal, but we cannot say if this is true for all developing secondary myotubes. Myoblasts were not clustered near the innervation zone, but were uniformly distributed throughout the muscle. Myoblasts were frequently interposed between a primary and a secondary myotube, in equally close proximity to both cell membranes. We conclude that specificity in myoblast-myotube fusion does not depend on restrictions in the physical distribution of myoblasts within the muscle, and therefore must reflect more subtle mechanisms for intercellular recognition.

Myonuclear birthdates distinguish the origins of primary and secondary myotubes in embryonic mammalian skeletal muscles.

Myotubes were isolated from enzymically disaggregated embryonic muscles and examined with light microscopy. Primary myotubes were seen as classic myotubes with chains of central nuclei within a tube of myofilaments, whereas secondary myotubes had a smaller diameter and more widely spaced nuclei. Primary myotubes could also be distinguished from secondary myotubes by their specific reaction with two monoclonal antibodies (MAbs) against adult slow myosin heavy chain (MHC). Myonuclei were birth dated with [3H]thymidine autoradiography or with 2-bromo-5'-deoxyuridine (BrdU) detected with a commercial monoclonal antibody. After a single pulse of label during the 1-2 day period when primary myotubes were forming, some primary myotubes had many myonuclei labelled, usually in adjacent groups, while in others no nuclei were labelled. If a pulse of label was administered after this time labelled myonuclei appeared in most secondary myotubes, while primary myotubes received few new nuclei. Labelled and unlabelled myonuclei were not grouped in the secondary myotubes, but were randomly interspersed. We conclude that primary myotubes form by a nearly synchronous fusion of myoblasts with similar birthdates. In contrast, secondary myotubes form in a progressive fashion, myoblasts with asynchronous birthdates fusing laterally with secondary myotubes at random positions along their length. These later-differentiating myoblasts do not fuse with primary myotubes, despite being closely apposed to their surface. Furthermore, they do not generally fuse with each other, as secondary myotube formation is initiated only in the region of the primary myotube endplate.

Cellular insertion of primary and secondary myotubes in embryonic rat muscles.

Mammalian muscles develop from two populations of myotubes; primary myotubes appear first and are few in number; secondary myotubes appear later and form most of the muscle fibres. We have made an ultrastructural study to investigate how primary and secondary myotubes in embryonic rat muscles transmit tension during the period of their development. Primary myotubes extend from end to end of the muscle from the earliest times, and attach directly to the tendon. In contrast, newly formed secondary myotubes are short cells which insert solely into the primary myotubes by a series of complex interdigitating folds along which adhering junctions occur. As the secondary myotubes lengthen and mature, their insertion is progressively transferred from the primary myotube to the tendon proper. We suggest that this variable insertion of immature secondary myotubes, combined with complex patterns of innervation and electrical coupling in developing muscle, makes it difficult to predict the overall contribution of secondary myotubes to muscle tension development. This work extends other studies showing the unique relationship between a primary myotube and its associated secondary myotubes, indicating that these may constitute a developmental compartment.

Nerve and muscle development in paralysé mutant mice.

Nerve and muscle development was studied in paralysé mutant mice. The mutant phenotype is first recognizable 6-7 days after birth (PN 6-PN 7) as cessation of muscle growth and weakness and incoordination of movement. Mutant animals die between 2 and 3 weeks of age. Muscle fibers from paralysé mutants had a unimodal distribution of diameters and normal numbers and distributions of acetylcholine receptors. The only structural abnormality seen was a reduced extracellular space within muscle fascicles. Total muscle choline acetyltransferase activity was reduced compared with that of control muscles, indicating that synaptic terminal development was impaired. Light and electron microscopy showed that polyneuronal innervation was retained in mutant endplates, and the normal process of withdrawal of redundant innervation did not occur. The paralysé muscles reacted to experimental denervation with an increase in extrajunctional acetylcholine receptor numbers. Intramuscular axons failed to become myelinated in mutant animals, although sciatic nerve axons were myelinated with a normal myelin thickness/axon diameter ratio. Nodes of Ranvier were elongated and myelin lamellae in the paranodal regions were poorly fused. Sciatic nerves in mutant animals retained the neonatal unimodal distribution of axon diameters, whereas in control animals it became bimodal by 2 weeks of age. Our results are not consistent with a previous suggestion that paralysé mutant muscle endplates are progressively denervated. We conclude that the major expression of the paralysé mutant phenotype is an arrest in development of both nerve and muscle during the first week after birth. The paralysé mutant gene most likely is involved in the general support of development of many or all body tissues from 1 week of age. We found no regression of any aspect of differentiation, once achieved.

Formation of primary and secondary myotubes in rat lumbrical muscles.

Numbers of myoblasts, primary myotubes and secondary myotubes in developing rat embryo hindlimb IVth lumbrical muscles were counted at daily intervals up until the time of birth, using electron microscopy. Motoneurone death at the spinal cord level supplying the lumbricals was assessed by counting axons in the 4th lumbar ventral root. Death of the motoneurones that supply the intrinsic muscles of the hindfoot was monitored by comparing the timecourse of development of total muscle choline acetyltransferase activity in control embryos with that in embryos where motoneurone death was inhibited by chronic paralysis with TTX, and by counting axons in the mixed nerve trunks at the level of the ankle at daily intervals. Condensations of undifferentiated cells marking the site of formation of the muscle were seen on embryonic day 15 (E15). Primary myotubes began to appear on E16 and reached a stable number (102 +/- 4) by E17. Secondary myotubes first appeared two days later, on E19, and numbered 280 at the time of birth (E22). The adult total of about 1000 muscle fibres, derived from both primary and secondary myotubes, was reached at postnatal day 7 (PN7) so considerable generation of secondary myotubes occurred after birth. There was a linear correlation between the number of undifferentiated mononucleate cells in a muscle and the rate of formation of secondary myotubes. The major period of motoneurone death in lumbar spinal cord was during E16-E17, when axon numbers in the L4 ventral root fell from 12,000 to 4000, but a discontinuity in the curve of muscle ChAT activity versus time indicated that death in the lumbrical motor pool occurred during E17-E19, after all primary myotubes had formed and before generation of secondary myotubes began. We suggest that motoneurone death, by regulating the final size of the motoneurone pool, regulates the ratio of secondary to primary myotube numbers in a muscle.

Neural determination of muscle fibre numbers in embryonic rat lumbrical muscles.

The generation and development of muscle cells in the IVth hindlimb lumbrical muscle of the rat was studied following total or partial denervation. Denervation was carried out by injection of beta-bungarotoxin (beta-BTX), a neurotoxin which binds to and destroys peripheral nerves. Primary myotubes were generated in denervated muscles and reached their normal stable number on embryonic day 17 (E17). This number was not maintained and denervated muscles examined on E19 or E21 contained many degenerating primary myotubes. Embryos injected with beta-bungarotoxin (beta-BTX) on E12 or E13 suffered a partial loss of motoneurones, resulting in a reduced number of axons in the L4 ventral root (the IVth lumbrical muscle is supplied by axons in L4, L5 and L6 ventral roots) and reduced numbers of nerve terminals in the intrinsic muscles of the hindfoot. Twitch tension measurements showed that all myotubes in partly innervated muscles examined on E21 contracted in response to nerve stimulation. Primary myotubes were formed and maintained at normal numbers in muscles with innervation reduced throughout development, but a diminished number of secondary myotubes formed by E21. The latter was correlated with a reduction in number of mononucleate cells within the muscles. If beta-BTX was injected on E18 to denervate muscles after primary myotube formation was complete, E21 embryo muscles contained degenerating primary myotubes. After injection to denervate muscles on E19, the day secondary myotubes begin to form, E21 muscles possessed normal numbers of primary myotubes. In both cases, secondary myotube formation had stopped about 1 day after the injection and the number of mononucleate cells was greatly reduced, indicating that cessation of secondary myotube generation was most probably due to exhaustion of the supply of competent myoblasts. We conclude that nerve terminals regulate the number of secondary myotubes by stimulating mitosis in a nerve-dependent population of myoblasts and that activation of these myoblasts requires the physical presence of nerve terminals as well as activation of contraction in primary myotubes.

Transfer of differentiated synaptic terminals from primary myotubes to new-formed muscle cells during embryonic development in the rat.

Most fibres in adult muscles are derived from secondary myotubes which form around an earlier developing population of primary myotubes. We examined embryonic rat IVth lumbrical muscles at a stage when most secondary myotubes were less than 24 h old, in order to study their acquisition of synaptic terminals. Single nerve terminals were frequently seen to synapse simultaneously with a well-developed primary myotube and a new-formed secondary myotube. Reconstructions of serial section electron micrographs revealed that most of these shared terminals had synaptic specializations opposed to the primary myotube, but others appeared to transmit mainly to the secondary myotube. Primary myotubes are contacted by multiple nerve terminals before secondary myotube formation has begun, and we suggest that synaptic terminals are progressively transferred from primaries to secondaries, the observed shared terminals representing intermediate stages in this transfer.

Inhibition of acetylcholinesterase accelerates axon terminal withdrawal at the developing rat neuromuscular junction.

In developing skeletal muscles, the rate at which superfluous innervation is lost from the endplates depends on the general level of neuromuscular activity. Whether it is activity of the presynaptic or postsynaptic structures (or both) that is critical is not well established. In this work, we transitorily inhibited the AChE of soleus muscle in postnatal rats, in order to increase postsynaptic activity, without directly altering activity of the nerve terminals. We then followed the time course of disappearance of axon terminals from the endplates of treated and normal muscles, using electron-microscope techniques. Three hours after inhibition of AChE, the muscle fibres exhibited local supercontracture and ultrastructural damage in the region of the endplate, consistent with local elevation of Ca2+ levels. At the same time, small electron-opaque vesicles, apparently of muscular origin, appeared in the synaptic cleft. The nerve terminals, however, were entirely normal in number and appearance. One day after treatment, endplates of esterase-inhibited muscles showed accelerated loss of nerve terminals, compared to endplates of normally developing muscles. No further loss of nerve terminals occurred, once AChE activity returned at the endplate. These results suggest that the rate at which superfluous nerve terminals retract from the developing neuromuscular junction is regulated by the level of activation of the muscle. It seems likely that activity of postsynaptic sites may similarly regulate changes in innervation patterns, in other developing or adapting neuro-neuronal or neuro-effector systems.

The effect of postsynaptic block on development of the neuromuscular junction in postnatal rats.

The effects of reduced muscle activity on the ultrastructural development of the rat neuromuscular junction (NMJ) have been studied. Soleus muscles of rats were treated with alpha-bungarotoxin (alpha BTx) in order to produce a postsynaptic block of activity between the ages of 10 and 12 days. Muscles of litter mates were treated with saline over the same period. The development of these control and alpha BTx treated muscles was then compared to that of normal muscles from untreated litter mates. During the period between the tenth and twelfth days after birth, normal soleus NMJs show two major ultrastructural changes. 1. The average number of axon terminal profiles present at each endplate decreases. This is thought to reflect the withdrawal of superfluous axons from the endplates. 2. There is an increasing specialization of the postsynaptic structures of the junction: complexity of folding of the muscle junctional membrane increases, as does the accumulation of subjunctional sarcoplasm and muscle nuclei. In soleus muscles treated with alpha BTx, the number of axon terminal profiles observed at the endplates does not decrease, suggesting that elimination of supernumerary axons does not occur. In addition, specialization of the postsynaptic structures of the NMJ is retarded during the period of ACh receptor block.