Absence of postnatal death among motoneurones supplying the inferior gluteal nerve of the rat

Postnatal developmental profile of neurons and glia in motor nuclei of the brainstem and spinal cord, and its comparison with organotypic slice cultures

In vitro preparations of the neonatal rat spinal cord or brainstem are useful to investigate the organization of motor networks and their dysfunction in neurological disease models. Long-term spinal cord organotypic cultures can extend our understanding of such pathophysiological processes over longer times. It is, however, surprising that detailed descriptions of the type (and number) of neurons and glia in such preparations are currently unavailable to evaluate cell-selectivity of experimental damage. The focus of the present immunohistochemical study is the novel characterization of the cell population in the lumbar locomotor region of the rat spinal cord and in the brainstem motor nucleus hypoglossus at 0-4 postnatal days, and its comparison with spinal organotypic cultures at 2-22 days in vitro. In the nucleus hypoglossus, neurons were 40% of all cells and 80% of these were motoneurons. Astrocytes (35% of total cells) were the main glial cells, while microglia was <10%. In the spinal gray matter, the highest neuronal density was in the dorsal horn (>80%) and the lowest in the ventral horn (≤57%) with inverse astroglia numbers and few microglia. The number of neurons (including motoneurons) and astrocytes was stable after birth. Like in the spinal cord, motoneurons in organotypic spinal culture were <10% of ventral horn cells, with neurons <40%, and the rest made up by glia. The present report indicates a comparable degree of neuronal and glial maturation in brainstem and spinal motor nuclei, and that this condition is also observed in 3-week-old organotypic cultures.

Cell death of spinal interneurones

The occurrence of neuronal death during development is well documented for some neuronal populations, such as motoneurones and dorsal root ganglion cells, whose connecting pathways are clearly defined. Cell survival is thought to be regulated largely by target and input connections, a process that serves to match the size of synaptically linked neuronal populations. Far less is known about interneurones. It is assumed that most interneurone populations are excluded from this process because their connections are more diffuse. Recent studies on the rat spinal cord have indicated that interneurone death does occur, both naturally during development and induced following peripheral nerve injury. Here the evidence for spinal interneurone death is reviewed and the factors influencing it are discussed. There are many functional types of interneurones in the spinal cord that may differ in vulnerability to cell death, but it is concluded that for most spinal interneurones the traditional view of target regulation is unlikely. Instead it is proposed that developmental interneurone death in the spinal cord forms part of a plastic response to altered sensory activation rather than a size-matching exercise. There is also emerging evidence that interneurone death may play a more direct role in some neurodegenerative diseases than hitherto considered.

The role of apoptosis and excitotoxicity in the death of spinal motoneurons and interneurons after neonatal nerve injury

There is evidence that motoneurons which die following neonatal nerve injury in rats do so through an excitotoxic mechanism. In this study, we have investigated whether this excitotoxicity induces motoneuron death by apoptosis. Sciatic motoneurons were prelabelled at birth with the retrograde tracing agent, Fast Blue, and the sciatic nerve was crushed in one leg two days later. At intervals up to 12 days, sections of the lumbar enlargement were analysed for apoptosis using propidium iodide and terminal deoxynucleotidyl transferase biotin-14-UTP nick end labelling techniques. A significant concentration of Fast Blue-labelled apoptotic motoneurons was seen in the area of the sciatic motor pool ipsilateral to the nerve injury, with the majority occurring in the first three days. Comparison of estimates of the time-course of apoptosis with that of motoneuron survival suggest that all motoneuron death induced during the first 12 days occurs by apoptosis and that the process is only recognizable for 2 h. Treatment with the N-methyl-D-aspartate receptor antagonist, dizocilpine maleate, reduced the level of apoptosis by 60%. Taken together, these data show that motoneurons which have been affected by an excitotoxic mechanism die by apoptosis. The apoptotic study also provides evidence, for the first time, that unilateral nerve injury induces motoneuron death in the contralateral sciatic motor pool. Apoptotic interneurons were also seen on both sides of the spinal cord as a result of nerve injury.

Vitamin E affects quantitative age changes in lumbar motoneurons and in their peripheral projections

Vitamin E deficiency was previously found to induce plastic changes in the number of primary sensory neurons and in motoneuron peripheral field projections. In this work, quantitative changes in motoneurons of lumbar segments, in nerve fibres constituting ventral roots and in innervating leg motor fibres were studied in normal and vitamin E deficient rats from 1 to 5 months of age. The number of lumbar motoneurons was found to decrease, while there were no changes in the number of ventral root fibres. An increase in the number of innervating leg motor fibres was observed during ageing in control rats; in vitamin E deficient rats the number of fibres in the ventral roots did not change, as occurred in controls, but the decrease in the number of motoneurons was smaller and the number of innervating leg motor fibres increased further in comparison to the controls. The findings are consistent with the idea that vitamin E deficiency causes a decrease in motoneuron death or, alternatively, that it induces some process partially compensating naturally occurring motoneuron death.

Evidence that spinal interneurons undergo programmed cell death postnatally in the rat

Programmed cell death has been demonstrated in several specific neuronal populations as a mechanism for modulating the population size following differentiation, but its applicability to all neuronal types is unclear. Evidence for programmed cell death in some populations such as the numerous spinal interneurons has been lacking. We have studied the incidence of apoptosis in the rat spinal cord with three different methods and found a previously undocumented wave of apoptosis occurring in spinal grey matter shortly after birth. The apoptotic morphology was confirmed ultrastructurally. Dying cells were identified as neurons by immunocytochemical labelling for neuronal markers and had an anatomical distribution which indicated that most of the apoptotic cells were interneurons not motoneurons. This wave of apoptosis has the characteristics of a discrete developmental process and occurs later than that of either ventral horn motoneurons or dorsal root ganglion cells, to which most spinal interneurons are connected. These findings indicate that interneurons do undergo programmed cell death, and we suggest that this occurs in response to the earlier reduction in size of their main synaptic targets.

Cellular transglutaminases in neural development

Enzymes of the transglutaminase family catalyze the Ca(2+)-dependent covalent cross-linking of peptide-bound glutamine residues of proteins and glycoproteins to the epsilon-amino group of lysine residues to create inter- or intramolecular isopeptide bonds. Transglutaminases can also covalently link a variety of primary amines to peptide-bound glutamine residues giving rise to two possibilities; firstly, where the primary amine has two or more amino groups, further catalysis can result in the formation of cross-linked bridges between glutamine residues, and secondly, where the primary amine is a monoamine, glutamine residues are rendered inert to further modification. The products are therefore in the main, homo- or heterodimers, or extensive, metabolically-stable multimeric complexes or matrices. Ca(2+)-dependent transglutaminase activity is present in the mammalian peripheral and central nervous systems and transglutaminase-catalyzed cross-linking of endogenous substrates has been demonstrated in neurons of Aplysia and the mammalian brain. Transglutaminase activity increases in the brain during development, principally owing to the increasing preponderance of glial cell activity. In a few regions including the cerebellar cortex, activity is also high in early development. Cellular transglutaminases occur widely in differentiating cells and tissues in mammals, with more than one transglutaminase frequently associated with a single cell type. The primary protein sequences of three cellular transglutaminases have been fully determined in different species, together with that of a mammalian protein homologue (band 4.2) which shares extensive sequence homologies with transglutaminases, but lacks the active site cysteine residue. The upstream sequences of two mammalian cellular transglutaminase genes (C and K) contain numerous regulatory sites, and an invertebrate transglutaminase, annulin, is spatially regulated within homeodomains. Multiple molecular forms of transglutaminase C and possibly other cellular transglutaminases exist in mammalian brain. The emerging picture is one of a family of cytosolic and membrane-bound proteins central to several regulatory pathways whose functions is to stabilize the cellular and intercellular superstructure in growing organisms. The targeted formation of glu-lys isopeptide bonds between proteins is central to this function. Cytoskeletal proteins, membrane-associated receptors, enzymes in signal transduction pathways and extracellular glycoproteins are candidate substrates as are polyamines, but few cellular proteins have been identified as components of naturally-occurring covalently-bonded matrices. Transglutaminases participate in the programme of neuronal differentiation in some but not all classes of neurone. Both neuronal and non-neuronal expression of transglutaminases may be important for guidance of migrating neurons or growth cones and sustainment of cell shape and coordinates during development.(ABSTRACT TRUNCATED AT 400 WORDS)

Localization and activity of transglutaminase, a retinoid-inducible protein, in developing rat spinal cord

The distribution of the retinoid-inducible enzyme, tissue transglutaminase (tTG) in developing rat spinal cord was determined by enzyme assay and immunocytochemistry. tTG activity was at its highest in the forebrain in late foetal development. In hindbrain and spinal cord, elevated activity persisted until after birth. In spinal cord only, a second peak of activity occurred during the first week post partum (P3). tTG was associated with both the cytosolic and particulate tissue fractions throughout spinal cord development, but the particulate component was more prominent in the early postnatal period. tTG was more concentrated during this period in the ventral horn, where the particulate-associated enzyme activity was highest. In spinal cord at 3 days post partum, particulate tTG could be solubilized with lubrol-PX, dithiothreitol and potassium thiocyanate. Both soluble and particulate-associated tTG coeluted with guinea-pig liver transglutaminase C by DEAE-sephacel chromatography. The first peak of tTG activity during late foetal life coincided with the transient localization of the enzyme by immunocytochemistry in vascular endothelia throughout the spinal cord. The second peak of activity at 3 days post partum, by which time vascular immunoreactivity was absent, coincided with the occurrence of small numbers of intensely immunoreactive motor neurones in the ventral horn. Immunoreactive motor neurones were seen predominantly at two levels: the lower thoracic segments and lumbar enlargement. The abnormal appearance of many immunoreactive neurones suggested degenerative changes were occurring. tTG was also present in central canal cluster cells from birth onwards. No neuronal immunoreactivity was seen throughout foetal development. A proportion of motor neurones prepared from E15 spinal cord and grown in coculture with spinal cord astrocytes, were immunoreactive for tTG. All immunoreactive neurones showed signs of degeneration. Addition of myotube-conditioned medium (a source of cholinergic differentiation factor, CDF) reduced the proportion of tTG-immunoreactive neurons in the cultures. Schwann cell-conditioned medium (a source of ciliary neurotrophic factor, CNTF) had a similar but less potent effect on the numbers of immunoreactive neurones. The possibility that tTG is a marker for late, but not early-phase programmed cell death in the developing rat spinal cord is discussed in the light of a proposed role for tTG in the mechanism of natural cell death by apoptosis.

Postnatal maturation of the dendritic fields of motoneuron pools supplying flexor and extensor muscles of the distal forelimb in the rat

In the rat cervical spinal cord the corticospinal projection on motoneurons either direct or indirect (via interneurons) comes about postnatally making it accessible for experimental research. Therefore, the postnatal developmental changes of motoneurons and in particular their dendritic fields were examined. Motoneurons innervating the two antagonistic muscles in the distal forepaw, the m. flexor digitorum profundus and the m. extensor digitorum communis, were retrogradely labelled by intramuscular injections of cholera toxin subunit B conjugated with horseradish peroxidase in rats of various postnatal ages. Following a 48–72 hour survival period the motoneurons and their dendritic fields were studied in the seventh and eighth cervical spinal cord segments. Both the number and the position of motoneurons were found to remain constant throughout postnatal development. Extensor motoneurons were positioned dorsolaterally in the ventral horn at the border of grey and white matter, flexor motoneurons were in general medial to extensor motoneurons. The results on the dendritic field demonstrate firstly, that during postnatal development the extension of the dendrites of both flexor and extensor motoneurons changes from spreading out in all directions at postnatal day 2 to spreading in only a few, specific directions from postnatal day 21 onwards, with the restriction that both motoneuron pools follow a different time scale to achieve this. Secondly, both pools have a temporal dendritic component extending into the white matter of the lateral funiculus.(ABSTRACT TRUNCATED AT 250 WORDS)

Developmentally regulated expression of calcitonin gene-related peptide at mammalian neuromuscular junction

Using indirect immunofluorescence techniques, we have found that calcitonin gene-related peptide-like immunoreactivity is present in the neuromuscular junctions of somatic muscles as well as in almost all motor neurons of the lumbar enlargement of 1-week-old rats. It gradually decreases in both motor neurons and motor nerve endings as the animal grows up and completely disappears from the neuromuscular junctions in adult rats, persisting only in the motor nerve endings on the intrafusal fibers. In situ hybridization experiments have shown that the down-regulation of calcitonin gene-related peptide-like immunoreactivity is strictly related to a reduction in CGRP mRNA levels in the spinal motor neurons. These results indicate that the expression of CGRP is developmentally regulated in spinal cord alpha motor neurons. They also suggest that the peptide may play an important role at the immature neuromuscular junction.

Explanation for the labeling of cervical motoneurons in young rats following the introduction of horseradish peroxidase into the calf

This study was carried out to determine whether cervical motoneurons, labeled following the introduction of horseradish peroxidase into the rat hind leg, belong to the cutaneous trunci motoneuron pool. The cutaneous trunci is a superficial muscle that extends from the axilla, over the flank, and into the thigh. Its nerve supply is derived from the brachial plexus. In experimental animals, horseradish peroxidase was either injected directly into the right gastrocnemius muscles, or applied to gelfoam and implanted over the calf muscles in the right leg of 5-, 10-, 15-day-old and adult rats. In control animals the cutaneous trunci was denervated prior to the administration of horseradish peroxidase. Labeled cervical motoneurons were present in the 5-, 10-, and 15-day-old but not the adult experimental groups and were located within the predetermined confines of the cutaneous trunci motoneuron pool. No labeling of cervical motoneurons was observed in any of the control groups in which the cutaneous trunci muscle was denervated. The most likely explanation for the labeling of cervical motoneurons in young rats was the local diffusion of horseradish peroxidase from the calf to the thigh, where it entered the cutaneous trunci muscle and was taken up by some of its motoneurons. The absence of such labeling in adult rats was probably due to the presence of connective tissue barriers to diffusion and to the greater distance between the site of horseradish peroxidase application and the cutaneous trunci muscle, which prevented the tracer from reaching the cutaneous trunci muscle and labeling its motoneurons.

Postnatal development of peroneal motoneurons in the kitten

In 1- to 72-day-old kittens, motoneurons of the 3 peroneal muscle nuclei were labeled by retrograde axonal transport of horseradish peroxidase from individual muscles. At birth, the locations of peroneal nuclei were similar to those of the adult cat. Counts of motoneurons at different ages indicated that postnatal cell death does not occur in peroneal motor nuclei. Primary dendrites were as numerous in motoneurons of newborn kittens as in adult motoneurons but they were thinner, shorter and poorly ramified. The number of recurrent axon collaterals was higher in the first postnatal week than at later stages. The growth of motoneurons followed similar rates in the 3 peroneal nuclei. Distributions of cell body diameters and volumes were unimodal at birth and became bimodal between 15 and 20 days postnatal. The separation of peroneal motoneurons in two size subgroups, presumably corresponding to alpha and gamma populations, was followed by an increase in growth rate which became faster for alpha than for gamma motoneurons.

Loss of motor neurons from the median nerve motor nucleus of the mutant mouse 'wobbler'

This paper describes the location and number of motor neurons in the median nerve pool of wobbler mice and normal littermates as determined by retrograde labelling of the cut median nerve with horseradish peroxidase (HRP) in animals from 3 weeks to 1-year-old. The median nerve motor nucleus is located in spinal segments C5-T1, and in normal animals contains 199 (6) (mean (SEM] motor neurons. Three-week-old wobbler mice have the same number of labelled neurons as control animals, and this number falls to 75% of normal values by 4 weeks of age, and to approximately 60% by 6 weeks of age and older. Numerous swollen, pale and frequently vacuolated perikarya are present in the same 3-6-week-old mice. In the 3-week-old mutants these comprise on average 17% of the total large (greater than 20 microns) neuronal cell bodies counted in segments C5-T1. By 6 weeks this figure has fallen to 10%, and to less than 4% in adult wobblers. We conclude that the most active period in the expression of the wobbler phenotype is from 3 to 6 weeks of age.

Age changes in axon number along the cervical ventral spinal nerve roots in rats

Axon counts were made at two standardised levels of C7 ventral spinal nerve roots from 46 female rats representing nine ages between birth and 500 days. The objective was to provide a definitive account of proximodistal changes in axon numbers and of age changes in axon numbers both during postnatal development and at several stages during maturity. At each age there is a proximodistal increase in the numbers of axons in all categories examined (myelinated, promyelin, transitional, and fetal) between levels midway along the subarachnoid course of the root and where it is apposed to but separate from the dorsal root ganglion. During maturation and throughout maturity axon totals change similarly at both levels: After a slight increase immediately postnatum, they decline sharply between 4 and 20 days due to a marked loss of unmyelinated axons. A gradual decline in myelinated axon numbers continues to 500 days. While these changes are occurring, axon numbers in all categories show a proximodistal increase throughout. The magnitude of this increase lessens with age for all but the transitional category due to a preferential decrease in numbers distally. Though these observations do not differentiate between axon branching and looping of sensory axons into the ventral root as a cause of the proximodistal increase in numbers, they tend to support the former. At each age during maturation axon proportions at proximal and distal levels correspond well for each animal, indicating that axon segregation proceeds at related rates within each root. Age changes in axon proportions within the transitional and fetal categories indicate that the postnatal stage of axon segregation results from axon loss, rather than Schwann cell proliferation.

Synaptic rearrangements and alterations in motor unit properties in neonatal rat extensor digitorum longus muscle

1. We have used in vitro intracellular recordings and measurements of the contractile properties of single motor units to examine the changes in muscle innervation occurring during the post-natal development of a fast-twitch muscle in the hindlimb of the rat, the extensor digitorum longus (EDL). 2. Intracellular recordings of end-plate potentials evoked in response to graded stimulation of the nerve supply to the muscle indicate that during the first day after birth, each muscle fibre receives synaptic input from at least two motoneurones and that some muscle fibres receive as many as six such inputs. With subsequent development, most of this polyneuronal innervation is eliminated: the first singly innervated fibres are encountered on day 3; by day 18 fewer than 5% of the fibres remain polyneuronally innervated. These results show that there are quantitative differences in post-natal synapse elimination in EDL compared to its well-studied counterpart, the soleus. Although the great majority of fibres in both muscles become singly innervated at about 18 days, the first singly innervated fibres appear at least a week earlier in the EDL. None the less, synapses are lost from EDL at about half the rate they are lost from soleus. 3. The number of motor units, determined by counting the number of twitch increments produced by graded stimulation of ventral root filaments teased to contain only a few EDL motor axons, remains unchanged from an average of forty-one from post-natal day 1 to day 17. In addition, the number of muscle fibres counted in muscle cross-sections stained with an anti-myosin antibody increases less than 10% from birth to adulthood. Therefore, synapse elimination in EDL occurs with a largely constant population of muscle fibres as well as motoneurones. 4. Measurements of tensions generated by single motor units indicate that the average size of a motor unit declines from 6.8% of the muscle fibres at day 1 to 2.3% at 17 days. This result indicates that each motoneurone, on average, comes to innervate threefold fewer muscle fibres. Motor units derived from each of the spinal segments innervating the muscle undergo equivalent reductions in motor unit size, indicating that there is no segmental disproportion to synapse elimination in this muscle. At all ages, there is a large diversity of motor unit sizes in the muscle. Synapse elimination therefore appears to maintain rather than decrease this diversity.(ABSTRACT TRUNCATED AT 400 WORDS)

The role of synapse elimination in the establishment of neuromuscular compartments

To investigate the specificity of development of initial neuromuscular connections, we examined the compartmental distribution of synapses in neonatal rat lateral gastrocnemius (LG) muscle. Initial neuromuscular connections might be restricted to the compartmental territories present in adults; alternatively, synapse elimination could establish the compartments from a less precise pattern of innervation. We examined 46 pups of ages 0 to 14 postnatal days using a variety of techniques. The principle method was evoked electromyographic (EMG) activity in response to nerve stimulation. The nerve branch to one neuromuscular compartment was cut and the remainder of the nerve was stimulated. The presence of EMG activity was used to identify the areas of muscle contracting in response to nerve stimulation. After cutting a particular branch, EMG activity generally could not be recorded from the denervated compartment. These results indicate that the pattern of innervation at birth is essentially compartment-specific, and that neuromuscular compartments are not shaped from some less precise pattern by postnatal synapse elimination. The factors which operate prenatally to determine this high degree of specificity in neuromuscular connectivity seen at the time of birth, however, remain unknown.

Mammalian motoneuron cell death: development of the lateral motor column of a wallaby (Macropus eugenii)

We have investigated the development of the lumbar lateral motor column of the tammar as a model of mammalian motoneuron cell death that is accessible to experimental manipulation. The tammar is an Australian marsupial, belonging to the subfamily of wallabies and kangaroos. After a gestational period of 26-28 days, the pup crawls to its mother's pouch using its forelimbs. The major morphometric events that shape the formation of the hindlimb occurred between 21 days gestation and birth. At birth the premuscle masses had divided and motor nerves had begun to penetrate the muscles of the thigh and shank. The period of motoneuron cell death was biphasic and occurred entirely postnatally. During phase I, between birth and 40 days, 59% of motoneurons were lost. Cell numbers then stabilised before falling a further 24%, to give an overall loss of 70%. Most of phase II cell loss occurred between 90 and 150 days. The possibility that a second period of motoneuron cell death may be a common feature of mammals is discussed.

Cell death of axotomized motoneurones in neonatal rats, and its prevention by peripheral reinnervation

1. Motoneurone death induced by axotomy in the rat was studied following section of the medial gastrocnemius nerve near the muscle 4 days after birth. 2. The maximum twitch tension of the medial gastrocnemius muscle achieved by motor reinnervation after section of its nerve was about 70% of that measured on the contralateral, intact side. 3. The number of motor units counted at 35-45 days of age in the animals whose medial gastrocnemius nerves had been sectioned on day 4 was 62% of that observed in normal rats. 4. The number of medial gastrocnemius motoneurones retrogradely labelled with horseradish peroxidase (HRP) 30-40 days after section of the medial gastrocnemius nerve was 77% of that labelled on the contralateral, intact side. 5. When the medial gastrocnemius nerve had been sectioned on day 4 and prevented from peripheral reinnervation, the number of medial gastrocnemius motoneurones labelled with HRP was, on the average, only 18% of that labelled on the control side. 6. Decreased number of medial gastrocnemius motoneurones labelled with HRP following prevention of peripheral reinnervation was associated with a decrease in the neurone density of the motor cell column, indicating the occurrence of motoneurone death. 7. The majority of medial gastrocnemius motoneurones axotomized 4 days after birth appear to maintain their survival for about 2 weeks without target contact. 8. The area of the compound action potential of medial gastrocnemius motor fibres once decreased after axotomy on day 4 began to recover from the 12th day after the operation if reinnervation by the cut peripheral nerve had been allowed, whereas the compound action potential continued to decrease in those axotomized motoneurones whose reinnervation had been prevented. 9. It is concluded that target dependence of motoneurone survival previously observed at embryonic stages is still present during the early post-natal period.

Postnatal loss of synaptic terminals in the normal mouse soleus muscle

Using conventional physiological techniques for measuring unitary contractions and end-plate potentials (epps), the number, size and segmental properties of motor units (MUs) in the soleus muscle of the mouse during postnatal development have been examined. The number of MUs remains constant after birth, and there is no evidence of segmental preferences in the innervation pattern, before, during or after the postnatal elimination of redundant terminals. In neonates, MU size estimates based on twitch contractions are 30% smaller than tetanic estimates. Intracellular recording of epps shows that this is caused by facilitation of epps on repetitive stimulation. The frequency of occurrence of epps in the muscle from a few, isolated motor axons shows that the average motoneuron contacts 36% of the fibres in the muscle neonatally. A substantial fraction of the contacts is subthreshold for twitch activation of their fibre. The MU size remains constant up to day 5. During the next 10 days, the MU size is reduced to the mature value of 5% of the fibres in the muscle. It is concluded that the neonatal loss of synaptic terminals in this muscle takes place without concomitant loss of entire motor neurons, and that it is independent of possible segmental preferences in the innervation of the muscle.

Spatial organization within rat motoneuron pools

Topographical maps form the basis of the organization in many projections within the central nervous system, but in the neuromuscular system such detailed spatial organization has generally been assumed to be absent and indeed unnecessary for normal function (see, for example, ref. 1). However, there is some physiological evidence for a degree of spatial organization within the discrete, longitudinal motor columns which supply individual muscles. We have used horseradish peroxidase as a retrograde tracer to confirm the topographical relationship between the rostro-caudal location of motoneuron cell bodies and the antero-posterior motor unit distribution in the rat gluteus maximus muscle. We also provide evidence for a further axis of intracolumnar organization. The motor pools of the rat intercostal muscles, whose axons lie in a single, segmental nerve, have a ventro-dorsal axis in the ventral horn on which is mapped the proximo-distal position of the motor units. This suggests that during development, not only are motoneurons specified to innervate a particular muscle, but project within that muscle to a predictable location according to their position in the motoneuron pool. The presence of such topographical maps suggests that motoneurons are subject to greater developmental constraints than previously thought.