- Email: [email protected]
Spatial organization within rat motoneuron pools
Neuroscience Letters, 60 (1985) 325-329
325
Elsevier Scientific Publishers Ireland Ltd.
NSL 03555
SPATIAL ORGANIZATION WITHIN RAT M O T O N E U R O N POOLS
V.J. H A R D M A N and M.C. BROWN
University Laboratory of Physiology, Parks Road, Oxford, OX1 3PT (U.K.) (Received May 24th, 1985; Revised version received July 10th, 1985; Accepted July 1 lth, 1985)
Key words." motoneuron pool - somatotopy - horseradish peroxidase - motoneuron - intercostal muscle -
gluteus muscle - rat
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 [12] which supply individual muscles [2--4, 14]. We have used horseradish peroxidase as a retrograde tracer to confirm the topographical relationship between the rostro-caudal location o f motoneuron cell bodies and the antero-posterior motor unit distribution in the rat gluteus maximus muscle [3]. We also provide evidence for a further axis o f 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 [7], but project within that muscle to a predictable location according to their position in the motoneuron pool. The presence o f such topographical maps suggests that motoneurons are subject to greater developmental constraints than previously thought.
The general experimental procedure was as follows: adult rats were anesthetized with 3.5~/o chloral hydrate (1 ml/100 g) and the muscles (gluteus maximus and external or internal intercostals) exposed. Quantities of 1~o horseradish peroxidaselabelled wheat germ agglutinin (WGA-HRP) (0.1 ktl), in 0.9~o saline containing 1~o Fast green, were injected i.m. with a 1-#1 Hamilton syringe. Different areas of homologous muscles were injected on the left and right to allow comparison of the position of cells in the longitudinal axis in a single animal and the horizontal position of cells in single sections. The skin and any overlying muscles were sutured and the animals allowed to recover. Forty-eight hours later they were reanesthetized deeply, and perfusion fixed (1X paraformaldehyde--1.25~ glutaraldehyde in 0.1 M phosphate buffer, pH 7.4). The relevant portions of spinal cord were removed, sectioned and histochemically processed for HRP according to the method of Mesulam [11]. The results for the gluteus maximus are shown in Fig. 1. The diagram in Fig. la illustrates the gross anatomy of the muscle. We confined our study to the region of muscle supplied by the inferior gluteal nerve and cut the superior nerve to prevent labelling of its axons following a rostral injection. Fig. 1b shows camera lucida drawings of the positions of the motoneurons labelled following an injection made rostrally in the muscle (on the left) and caudally (on the right). The two groups clearly 0304-3940/85/$ 03.30 © 1985 Elsevier Scientific Publishers Ireland Ltd.
326
a
Medial
Caudal I--i---~- ! inferior gluteal nerve
//~-~-
r~ Rostral
-~-: --- R ~ l - _ 7 ,,,t'~ / I [ / / / ~; ' ',, \ ~ I / / / / /
I maximus , /I/ ~ _ II 'Fd /LGluteus Lateral
b 4
1 mm
superior gluteal nerve
L_--
m,
1 mn
rostral injection
caudal injection
Fig. 1. a: diagram of the gluteus maximus muscle, viewed from its inner surface to show the position of the inferior and superior gluteal nerves and the sites of H R P injection (R and C). T F L , tensor fascia latae muscle, b: typical example of rostro-caudal organization of m o t o n e u r o n s in the inferior gluteal pool from camera lucida drawings of serial 100-/lm sections. Left: 0.1 Id injection of H R P into the rostral part of the muscle at position R (see a). Right: 0.1/~l injection into caudal region of muscle at C (b). The superior gluteal nerve was cut to prevent labelling of its axons. The number of cells labelled on the left was 14 and on the right 16. The size of the injection site estimated a few minutes after injection from the spread of the dye was about 1.5 m m , and the distance between injection sites was l0 m m (approximately the length of the inferior gluteal endplate region). F r o m other work it is known that the mean width of muscle occupied by each motor unit is around 15~o of the muscle length [3[ and the number of 0t motoneurons is approximately 40 [6[. If the motor units are assumed to be equally spaced across the muscle [3], the n u m b e r of cells expected to be labelled by an injection site of 1.5 m m diameter is given by: (injection diamet e r + m e a n motor unit length) (number of ct m o t o n e u r o n s - l ) / ( l e n g t h of m u s c l e - m e a n motor unit length) = 14. This compares very well with the actual numbers labelled and suggests that the effective area from which H R P uptake could occur was similar to the apparent area injected.
327
lie in different positions, with rostral muscle innervated by more rostrally placed motoneurons than caudal muscle. This was the finding in 5 other experiments and agrees with data from electrophysiological stimulation of ventral rootlets [3]. Muscles such as the intercostals are similar to the gluteus in that they are essentially flat with end-plates lying in a single band across the muscle. In other respects, however, they differ: firstly, each muscle is innervated by a single segmental nerve and secondly, their muscle fibers are distributed not rostro-caudally as in the gluteus maximus but in an axis that runs from beside the vertebral column, laterally and ventrally towards the sternum, i.e. proximo-distally. We found (Figs. 2 and 3) that this different orientation is reflected in a different distribution of the intercostal neurons within their motor pool. Again, muscle fibers in separate regions of the muscles are innervated by motoneurons lying in different areas of the motor pool, but in this case distal muscle fibers are innervated by more dorsally located neurons. Also, there is no difference in the rostro-caudal extent of the labelled neurons within the spinal cord. The shape of the grey matter is not identical from section to section, and this must introduce error into our method of standardizing motoneuron position (see inset in Fig. 3b). In spite of this, the difference between the median positions of neurons which project proximally and distally in both muscles is highly significant. One possible explanation for the apparent spatial organization within these motor pools could be the segregation of muscle fibers of a particular type (fast fatigable, slow, etc.) into separate muscle regions and the clustering of their respective moto0
L
28 24 2O-
%
16-
12,
;r
e
~o
~o
,
3~
4'o
5'o
~
~o
Vertical position v / V Fig. 2. Histogram summarizing the positions of distally and proximally labelled internal intercostal motoneurons in the dorso-ventral axis of the cord. Their distributions in the rostro-caudal axis did not differ. The position of each neuron was normalized as shown in the inset. (n = total number of motoneurons labelled in 5 experimental animals.) - - - , proximal (n = 91); - - - , distal (n = 171).
328
neurons within each motor pool. The sizes of the labelled neurons compared within animals showed no significant difference (Mann-Whitney U-test) between proximal and distal or rostral and caudal injections. This is correlated with a fairly uniform distribution of different muscle fiber types along the length of the muscle as shown by adenosine triphosphatase histochemistry (the exception to this being the most proximal part of the external intercostal muscle which contains a much higher proportion of type I fibers). This illustrates that there is no segregation of motoneurons according to size or motor unit type as noted earlier by Burke et al. [5]. These results demonstrate that within motoneuron pools there exists a positional organization that determines the location of motor units within a muscle. Moreover, this order is found along more than one axis. The rostrocaudal axis as described here within the gluteus maximus muscle confirms previous electrophysiological data [2-4, 14] and is analogous to the known segmental selectivity shown by autonomic ganglia [10, 15]. In the latero-medial axis of brachial and lumbar regions of the spinal cord, which innervate the limbs, it is known that lateral motoneurons innervate muscles derived from the dorsal premuscle mass and more medial neurons innervate ventrally derived muscle [7]. Thirdly the dorso-ventral axis, shown here within the intercostal motor pools, represents a proximo-distal position within the muscle. This implies that, while forming their initial connections, motoneurons might possess markers
b
a lOO~
8o~
o
B0-
]
~J
/"
6o~ I
'o!
I00~
/
//
~ ~'~
~
60/
Distal 40-
2o~
I
20-
/
/
/
/
/
C .o.o, f - - -"~
Distal
O~
6-
fo
Wo
'
~o
Fig, 3. Summary of positions of all (a) internal and (b) external intercostal motoneurons labelled after distal and proximal injections into the two muscles (5 experimental animals per muscle). The positions of the motoneurons and the outline of the grey matter were measured and expressed as a percentage of the greatest width and depth of the grey matter, as measured from the middle of the the central canal (see inset). The solid line shows an outline of the ventral horn, averaged from several sections. The dotted lines show the outer limits of all the cells labelled in these experiments. External intercostals: distal injections, n = 4 3 : proximal injections, n = l l 0 . Internal intercostals: distal, n = 171; proximal, n = 9 1 . The median values and 95'~o confidence limits in the horizontal and vertical axes are also shown. The two intercostal motor pools lie in characteristic locations in the ventral horn, but they do not follow the lateromedial rule of limb motoneurons since the external muscle lies dorsal to the internal intercostal but is innervated by more medial motoneurons. However, the thoracic ventral horn does show some organization with respect to dorsal/ventral muscle origin [ 131.
329
which label their position in three axes. Clearly, further investigation is necessary to test this hypothesis. The innervation of different muscle fiber types by motoneurons with appropriate properties is of considerable functional importance. Whether precise spatial innervation patterns are also necessary in muscle, as they clearly are in the central nervous system for correct analysis of information, is more doubtful. In most muscles, the fibers converge on a common tendon, so losing any localized information. The relatively large, flat muscles we have used in this study may be exceptional in their possession of spatially ordered motoneuron projection patterns, as different regions of them may need to be separately activated during certain movements. However, detailed motoneuron pool organization is not a prerequisite for precise, functional synaptic arrangements, as shown by the loosely arranged amphibian motoneuron pools [8, 9] and the extensive overlap of dendritic arbours of cells belonging to different pools in all vertebrate classes (e.g. ref. 13). It seems possible that all motoneuron projections are spatially organized, simply as a consequence of developmental mechanisms universally adopted to achieve organized axonal projections, which in most neural pathways a r e functionally important. This work was funded by the M.R.C. Thanks to I.D. Thompson for critical comments. V.J.H. is holder of an M.R.C. Studentship. 1 Alberts, B., Bray, D., Lewis, J,, Raft, M., Roberts, K. and Watson, J.D., The Molecular Biology of the Cell, Garland Publishing, New York, 1983, p. 1084. 2 Bennett, M.R. and Lavidis, N.A., Development of the topographical projection of motor neurons to a rat muscle accompanies loss of polyneuronal innervation, J. Neurosci., 4 (1984) 2204-2212. 3 Brown, M.C. and Booth, C.M., Postnatal development of the adult pattern of motor axon distribution in rat muscle, Nature (Lond.), 304 (1983) 741-742. 4 Browne, K.M., The spatial distribution of segmental nerves to striate musculature of the hindlimb of the rat, J. Comp. Neurol., 93 (1950) 441-455. 5 Burke, R.E., Strick, P.L., Kanda, K., Kim, C.C. and Walmsley, B., Anatomy of medial gastrocnemius and soleus motor nuclei in cat spinal cord, J. Neurophysiol., 40 (1977) 667-680. 6 Hardman, V.J. and Brown, M.C., Absence of postnatal death among motoneurones supplying the inferior gluteal nerve of the rat, Dev. Brain Res., 19 (1985) 1-9. 7 Landmesser, L., The distribution of motoneurons supplying chick hind limb muscles, J. Physiol. (Lond.), 284 (1978) 371-390. 8 Lichtman, J.W. and Frank, E., Physiological evidence for specificity of synaptic connections between individual sensory and motor neurons in the brachial spinal cord of the bullfrog, J. Neurosci., 4 (1984) 1745-1753. 9 Lichtman, J.W., Jhaveri, S. and Frank, E., Anatomical basis of specific connections between sensory axons and motor neurons in the brachial spinal cord of the bullfrog, J. Neurosci., 4 (1984) 1754-1763. 10 Lichtman, J.W., Purves, D. and Yip, J.W., lnnervation of sympathetic neurones in the guinea-pig thoracic chain, J. Physiol. (Lond.), 298 (1980) 285-299. 11 Mesulam, M.-M., Tracing Neuronal Connections with Horseradish Peroxidase, John Wiley and Sons, Chichester, 1982, pp. 127-131. 12 Romanes, G.J., The motor pools of the spinal cord, Prog. Brain Res., 11 (1964) 93-119. 13 Smith, C.L. and Hollyday, M., The development and postnatal organisation of motor nuclei in the rat thoracic spinal cord, J. Comp. Neurol., 220 (1983) 16-28. 14 Swett, J.E., Eldred, E. and Buchwald, J.S., Somatotopic cord-to-muscle relations in efferent innervation of cat gastrocnemius, Am. J. Physiol. 219 (1970) 762-766. 15 Wigston, D.J. and Sanes, J.R., Selective reinnervation of adult mammalian muscle by axons from different segmental levels, Nature (Lond.), 299 (1982) 464--467.
325
Elsevier Scientific Publishers Ireland Ltd.
NSL 03555
SPATIAL ORGANIZATION WITHIN RAT M O T O N E U R O N POOLS
V.J. H A R D M A N and M.C. BROWN
University Laboratory of Physiology, Parks Road, Oxford, OX1 3PT (U.K.) (Received May 24th, 1985; Revised version received July 10th, 1985; Accepted July 1 lth, 1985)
Key words." motoneuron pool - somatotopy - horseradish peroxidase - motoneuron - intercostal muscle -
gluteus muscle - rat
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 [12] which supply individual muscles [2--4, 14]. We have used horseradish peroxidase as a retrograde tracer to confirm the topographical relationship between the rostro-caudal location o f motoneuron cell bodies and the antero-posterior motor unit distribution in the rat gluteus maximus muscle [3]. We also provide evidence for a further axis o f 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 [7], but project within that muscle to a predictable location according to their position in the motoneuron pool. The presence o f such topographical maps suggests that motoneurons are subject to greater developmental constraints than previously thought.
The general experimental procedure was as follows: adult rats were anesthetized with 3.5~/o chloral hydrate (1 ml/100 g) and the muscles (gluteus maximus and external or internal intercostals) exposed. Quantities of 1~o horseradish peroxidaselabelled wheat germ agglutinin (WGA-HRP) (0.1 ktl), in 0.9~o saline containing 1~o Fast green, were injected i.m. with a 1-#1 Hamilton syringe. Different areas of homologous muscles were injected on the left and right to allow comparison of the position of cells in the longitudinal axis in a single animal and the horizontal position of cells in single sections. The skin and any overlying muscles were sutured and the animals allowed to recover. Forty-eight hours later they were reanesthetized deeply, and perfusion fixed (1X paraformaldehyde--1.25~ glutaraldehyde in 0.1 M phosphate buffer, pH 7.4). The relevant portions of spinal cord were removed, sectioned and histochemically processed for HRP according to the method of Mesulam [11]. The results for the gluteus maximus are shown in Fig. 1. The diagram in Fig. la illustrates the gross anatomy of the muscle. We confined our study to the region of muscle supplied by the inferior gluteal nerve and cut the superior nerve to prevent labelling of its axons following a rostral injection. Fig. 1b shows camera lucida drawings of the positions of the motoneurons labelled following an injection made rostrally in the muscle (on the left) and caudally (on the right). The two groups clearly 0304-3940/85/$ 03.30 © 1985 Elsevier Scientific Publishers Ireland Ltd.
326
a
Medial
Caudal I--i---~- ! inferior gluteal nerve
//~-~-
r~ Rostral
-~-: --- R ~ l - _ 7 ,,,t'~ / I [ / / / ~; ' ',, \ ~ I / / / / /
I maximus , /I/ ~ _ II 'Fd /LGluteus Lateral
b 4
1 mm
superior gluteal nerve
L_--
m,
1 mn
rostral injection
caudal injection
Fig. 1. a: diagram of the gluteus maximus muscle, viewed from its inner surface to show the position of the inferior and superior gluteal nerves and the sites of H R P injection (R and C). T F L , tensor fascia latae muscle, b: typical example of rostro-caudal organization of m o t o n e u r o n s in the inferior gluteal pool from camera lucida drawings of serial 100-/lm sections. Left: 0.1 Id injection of H R P into the rostral part of the muscle at position R (see a). Right: 0.1/~l injection into caudal region of muscle at C (b). The superior gluteal nerve was cut to prevent labelling of its axons. The number of cells labelled on the left was 14 and on the right 16. The size of the injection site estimated a few minutes after injection from the spread of the dye was about 1.5 m m , and the distance between injection sites was l0 m m (approximately the length of the inferior gluteal endplate region). F r o m other work it is known that the mean width of muscle occupied by each motor unit is around 15~o of the muscle length [3[ and the number of 0t motoneurons is approximately 40 [6[. If the motor units are assumed to be equally spaced across the muscle [3], the n u m b e r of cells expected to be labelled by an injection site of 1.5 m m diameter is given by: (injection diamet e r + m e a n motor unit length) (number of ct m o t o n e u r o n s - l ) / ( l e n g t h of m u s c l e - m e a n motor unit length) = 14. This compares very well with the actual numbers labelled and suggests that the effective area from which H R P uptake could occur was similar to the apparent area injected.
327
lie in different positions, with rostral muscle innervated by more rostrally placed motoneurons than caudal muscle. This was the finding in 5 other experiments and agrees with data from electrophysiological stimulation of ventral rootlets [3]. Muscles such as the intercostals are similar to the gluteus in that they are essentially flat with end-plates lying in a single band across the muscle. In other respects, however, they differ: firstly, each muscle is innervated by a single segmental nerve and secondly, their muscle fibers are distributed not rostro-caudally as in the gluteus maximus but in an axis that runs from beside the vertebral column, laterally and ventrally towards the sternum, i.e. proximo-distally. We found (Figs. 2 and 3) that this different orientation is reflected in a different distribution of the intercostal neurons within their motor pool. Again, muscle fibers in separate regions of the muscles are innervated by motoneurons lying in different areas of the motor pool, but in this case distal muscle fibers are innervated by more dorsally located neurons. Also, there is no difference in the rostro-caudal extent of the labelled neurons within the spinal cord. The shape of the grey matter is not identical from section to section, and this must introduce error into our method of standardizing motoneuron position (see inset in Fig. 3b). In spite of this, the difference between the median positions of neurons which project proximally and distally in both muscles is highly significant. One possible explanation for the apparent spatial organization within these motor pools could be the segregation of muscle fibers of a particular type (fast fatigable, slow, etc.) into separate muscle regions and the clustering of their respective moto0
L
28 24 2O-
%
16-
12,
;r
e
~o
~o
,
3~
4'o
5'o
~
~o
Vertical position v / V Fig. 2. Histogram summarizing the positions of distally and proximally labelled internal intercostal motoneurons in the dorso-ventral axis of the cord. Their distributions in the rostro-caudal axis did not differ. The position of each neuron was normalized as shown in the inset. (n = total number of motoneurons labelled in 5 experimental animals.) - - - , proximal (n = 91); - - - , distal (n = 171).
328
neurons within each motor pool. The sizes of the labelled neurons compared within animals showed no significant difference (Mann-Whitney U-test) between proximal and distal or rostral and caudal injections. This is correlated with a fairly uniform distribution of different muscle fiber types along the length of the muscle as shown by adenosine triphosphatase histochemistry (the exception to this being the most proximal part of the external intercostal muscle which contains a much higher proportion of type I fibers). This illustrates that there is no segregation of motoneurons according to size or motor unit type as noted earlier by Burke et al. [5]. These results demonstrate that within motoneuron pools there exists a positional organization that determines the location of motor units within a muscle. Moreover, this order is found along more than one axis. The rostrocaudal axis as described here within the gluteus maximus muscle confirms previous electrophysiological data [2-4, 14] and is analogous to the known segmental selectivity shown by autonomic ganglia [10, 15]. In the latero-medial axis of brachial and lumbar regions of the spinal cord, which innervate the limbs, it is known that lateral motoneurons innervate muscles derived from the dorsal premuscle mass and more medial neurons innervate ventrally derived muscle [7]. Thirdly the dorso-ventral axis, shown here within the intercostal motor pools, represents a proximo-distal position within the muscle. This implies that, while forming their initial connections, motoneurons might possess markers
b
a lOO~
8o~
o
B0-
]
~J
/"
6o~ I
'o!
I00~
/
//
~ ~'~
~
60/
Distal 40-
2o~
I
20-
/
/
/
/
/
C .o.o, f - - -"~
Distal
O~
6-
fo
Wo
'
~o
Fig, 3. Summary of positions of all (a) internal and (b) external intercostal motoneurons labelled after distal and proximal injections into the two muscles (5 experimental animals per muscle). The positions of the motoneurons and the outline of the grey matter were measured and expressed as a percentage of the greatest width and depth of the grey matter, as measured from the middle of the the central canal (see inset). The solid line shows an outline of the ventral horn, averaged from several sections. The dotted lines show the outer limits of all the cells labelled in these experiments. External intercostals: distal injections, n = 4 3 : proximal injections, n = l l 0 . Internal intercostals: distal, n = 171; proximal, n = 9 1 . The median values and 95'~o confidence limits in the horizontal and vertical axes are also shown. The two intercostal motor pools lie in characteristic locations in the ventral horn, but they do not follow the lateromedial rule of limb motoneurons since the external muscle lies dorsal to the internal intercostal but is innervated by more medial motoneurons. However, the thoracic ventral horn does show some organization with respect to dorsal/ventral muscle origin [ 131.
329
which label their position in three axes. Clearly, further investigation is necessary to test this hypothesis. The innervation of different muscle fiber types by motoneurons with appropriate properties is of considerable functional importance. Whether precise spatial innervation patterns are also necessary in muscle, as they clearly are in the central nervous system for correct analysis of information, is more doubtful. In most muscles, the fibers converge on a common tendon, so losing any localized information. The relatively large, flat muscles we have used in this study may be exceptional in their possession of spatially ordered motoneuron projection patterns, as different regions of them may need to be separately activated during certain movements. However, detailed motoneuron pool organization is not a prerequisite for precise, functional synaptic arrangements, as shown by the loosely arranged amphibian motoneuron pools [8, 9] and the extensive overlap of dendritic arbours of cells belonging to different pools in all vertebrate classes (e.g. ref. 13). It seems possible that all motoneuron projections are spatially organized, simply as a consequence of developmental mechanisms universally adopted to achieve organized axonal projections, which in most neural pathways a r e functionally important. This work was funded by the M.R.C. Thanks to I.D. Thompson for critical comments. V.J.H. is holder of an M.R.C. Studentship. 1 Alberts, B., Bray, D., Lewis, J,, Raft, M., Roberts, K. and Watson, J.D., The Molecular Biology of the Cell, Garland Publishing, New York, 1983, p. 1084. 2 Bennett, M.R. and Lavidis, N.A., Development of the topographical projection of motor neurons to a rat muscle accompanies loss of polyneuronal innervation, J. Neurosci., 4 (1984) 2204-2212. 3 Brown, M.C. and Booth, C.M., Postnatal development of the adult pattern of motor axon distribution in rat muscle, Nature (Lond.), 304 (1983) 741-742. 4 Browne, K.M., The spatial distribution of segmental nerves to striate musculature of the hindlimb of the rat, J. Comp. Neurol., 93 (1950) 441-455. 5 Burke, R.E., Strick, P.L., Kanda, K., Kim, C.C. and Walmsley, B., Anatomy of medial gastrocnemius and soleus motor nuclei in cat spinal cord, J. Neurophysiol., 40 (1977) 667-680. 6 Hardman, V.J. and Brown, M.C., Absence of postnatal death among motoneurones supplying the inferior gluteal nerve of the rat, Dev. Brain Res., 19 (1985) 1-9. 7 Landmesser, L., The distribution of motoneurons supplying chick hind limb muscles, J. Physiol. (Lond.), 284 (1978) 371-390. 8 Lichtman, J.W. and Frank, E., Physiological evidence for specificity of synaptic connections between individual sensory and motor neurons in the brachial spinal cord of the bullfrog, J. Neurosci., 4 (1984) 1745-1753. 9 Lichtman, J.W., Jhaveri, S. and Frank, E., Anatomical basis of specific connections between sensory axons and motor neurons in the brachial spinal cord of the bullfrog, J. Neurosci., 4 (1984) 1754-1763. 10 Lichtman, J.W., Purves, D. and Yip, J.W., lnnervation of sympathetic neurones in the guinea-pig thoracic chain, J. Physiol. (Lond.), 298 (1980) 285-299. 11 Mesulam, M.-M., Tracing Neuronal Connections with Horseradish Peroxidase, John Wiley and Sons, Chichester, 1982, pp. 127-131. 12 Romanes, G.J., The motor pools of the spinal cord, Prog. Brain Res., 11 (1964) 93-119. 13 Smith, C.L. and Hollyday, M., The development and postnatal organisation of motor nuclei in the rat thoracic spinal cord, J. Comp. Neurol., 220 (1983) 16-28. 14 Swett, J.E., Eldred, E. and Buchwald, J.S., Somatotopic cord-to-muscle relations in efferent innervation of cat gastrocnemius, Am. J. Physiol. 219 (1970) 762-766. 15 Wigston, D.J. and Sanes, J.R., Selective reinnervation of adult mammalian muscle by axons from different segmental levels, Nature (Lond.), 299 (1982) 464--467.