Colchicine-induced differential sprouting of the endplates on fast and slow muscle fibers in rat extensor digitorum longus, soleus and tibialis anterior muscles

The patterns of sprouting of motor endplates were examined in fast extensor digitorum longus and slow soleus muscles and in tibialis anterior muscles containing fast and slow muscle fiber types. A histochemical technique combining nerve silver impregnation and endplate cholinesterase staining was developed for this task. Temporal examination of the innervation was conducted 3, 7 and 10 days after either a 45 or 90 min application of the ipsilateral sciatic nerve with 5 mM colchicine. This dosage of drug did not cause detectable axon or muscle fiber degeneration, unlike 60 mM which was highly neurotoxic. At 3 days following treatment with the lower concentration, there were no significant differences in the percentages of intranodal, preterminal and ultraterminal sprouts between the normal (non-treated), sham-treated, contralateral systemic-control and drug-treated groups of muscles. By 7 and 10 days, the muscles on the drug-treated side exhibited significant increases in the 3 types of sprouts. Collateral sprouting was uncommon: most outgrowths remained on the muscle fibers innervated by the parent axons. Endplates in the tibialis anterior muscles of the control and drug-treated groups were classified Complex, Intermediate or Simple according to the relative degrees of branching of the terminal arbors. The occurrence of endplate classes and muscle fiber types was correlated in the superficial and deep regions of this muscle. Complex endplates innervated fast glycolytic fibers, Intermediate endplates supplied fast oxidative glycolytic fibers, and Simple endplates served slow oxidative fibers. In response to colchicine, the endplates of the slow muscles sprouted more than those of fast muscles while the innervation of slow fiber types sprouted less than that of fast fiber types. Furthermore, intranodal sprouts were more prevalent in slow muscles and ultraterminal sprouts more numerous in fast muscles whereas intranodal sprouts predominated on fast fiber types and ultraterminal sprouts were characteristic of slow fiber types. These apparently contradictory results were reconciled when it was noted that soleus endplates were mostly Complex and Intermediate, and the extensor digitorum longus contained more Simple endplates. Thus, consistency of sprouting patterns among endplate types of the 3 muscles was recognized when the pre-existing branching patterns were considered. This indicated that the patterns of sprouting were determined by the motor neurons rather than the muscle fibers. The observed sprouting responses supported the hypothesis that colchicine treatment of motor axons caused muscle fibers to elaborate a diffusible sprout-inducing factor.     

Changes in sarcomere length following tenotomy in the rat

The medial belly of the gastrocnemius and the extensor digitorum longus muscles of rats were tenotomized. One day following tenotomy, the mean sarcomere length of the fast medial gastrocnemius was 1.8 microns, a value comparable to that of tenotomized slow soleus. The mean sarcomere length of the tenotomized extensor digitorum longus, however, was 2.0 microns, a figure which differed significantly from the values obtained for both the soleus and the gastrocnemius. Histological preparations showed the presence of central core degeneration in slow fatigue-resistant fibers of the tenotomized gastrocnemius comparable to that seen in the soleus. No changes were found in the fibers of the tenotomized extensor digitorum longus. The fact that central core lesions were produced in the fibers of soleus and medial gastrocnemius but not in the extensor digitorum longus may be related to the lesser reduction in sarcomere length following tenotomy of the latter muscle.     

Spinal transection and the postnatal differentiation of slow myosin isoenzymes

Newborn rats underwent cordotomy, and the myosin composition of individual muscles was investigated 3 months postoperatively. The results indicate that, after cordotomy, the myosin composition in the extensor digitorum longus and tibialis anterior muscles is normal, whereas in the soleus muscle the myosin has catalytic and molecular properties intermediate between those of adult fast and slow myosin. Together with histochemical data these results indicate that spinal transection causes a developmental arrest in the soleus muscle, at a stage corresponding to a mixed fiber population.     

Is fast fiber innervation responsible for increased acetylcholinesterase activity in reinnervating soleus muscles?

During reinnervation of the completely denervated rat hind limb we observed previously a temporary overproduction of acetylcholinesterases in the soleus but not in the extensor digitorum longus muscle. In the present study, we investigated whether the predominantly slow soleus, which is low in AChE activity, is initially reinnervated by axons that originally innervated fast muscle fibers with high AChE activity, such as those of the extensor digitorum longus. Local denervation of the rat soleus was carried out to eliminate reinnervation by axons destined for other muscles. This produced an overshoot in AChE activity that was qualitatively similar to that observed with high sciatic crush. Local denervation of the soleus in the guinea pig was done because this muscle is composed solely of slow (type I) fibers, thereby virtually eliminating the possibility of homologous muscle fast fiber innervation. The overshoot in this preparation was qualitatively similar to that seen with distal denervation in the guinea pig and local and distal denervation in the rat. Thus, initial fast fiber innervation is not responsible for the patterns of change in AChE activity seen with reinnervation in the soleus. We concluded that the neural control of AChE is different in these two muscles and may reflect specific differences in the characteristics of AChE regulation in fast and slow muscle. How these neural influences are translated into muscle synthesis and degradation remains unknown.     

Fetal-type creatine kinase in rat fast and slow muscles during denervation and reinnervation

Creatine kinase (CK) has three forms of isozymes; CK-BB, CK-MB, and CK-MM. In adult rats they show a specific tissue distribution: the BB form in the brain, the MB form in the heart, and the MM form in skeletal muscle. In embryonic skeletal muscles only the BB and MB forms are found. Adult slow-twitch muscles contain more fetal type creatine kinase (CK-B) than do fast-twitch muscles. In the present experiment the effect of denervation and reinnervation on the CK-B concentration was investigated in rat fast (extensor digitorum longus)- and slow (soleus)-twitch muscles by a highly sensitive immunoassay. Denervation of these muscles produced a progressive increase in CK-B concentration in both muscles. When the sciatic nerve was cut and immediately sutured, the CK-B concentration in both muscles showed a gradual reduction after an initial increase. By the 34th postoperative week the CK-B concentration in the soleus was about one-half that of the contralateral control, whereas that in the extensor digitorum longus was nearly normal. After cross union of the nerves innervating the muscles, the CK-B concentration in the soleus was reduced at 35 weeks to about one-half normal, but that in the extensor digitorum longus was always higher than the control value. After self-reunion of the nerves, the CK-B concentration at the 20th week was approximately normal in the extensor digitorum longus and significantly increased in the soleus. We suggest that the motoneurons normally innervating the extensor digitorum longus have a greater capability in suppressing the production of CK-B than do the soleus motoneurons.     

Localization of the pool of G4 acetylcholinesterase characterizing fast muscles and its alteration in murine muscular dystrophy

The distributions of acetylcholinesterase and its molecular forms within muscles of normal and dystrophic 129/ReJ mice were established by a concomitant cytochemical and biochemical study performed on 1-mm serial sections of three predominantly fast muscles, i.e., anterior tibialis, extensor digitorum longus, and sternomastoid, as well as the slow-twitch soleus. This comparative study showed the following main findings. 1) In every muscle of both normal and dystrophic mice a) the three asymmetric forms were confined to the motor zone where they systematically codistributed with the endplates, and b) all globular forms, including G4, were concentrated at the motor zone from which they extended over the entire muscle length along a concentration gradient. 2) In the normal muscles, the perijunctional sarcoplasmic cytochemical reaction exhibited by individual fibers was grouped into a well-defined cojunctional acetylcholinesterase compartment in which the endplates were embedded. The overall intensity of the cojunctional cytochemical reaction was either high or low according to whether the muscle was predominantly fast or slow. 3) This cojunctional acetylcholinesterase compartment varied in close parallelism with G4 and thus appeared as the cytochemical correlate of the G4 molecules concentrated around the endplates. In particular, as the shape of the motor zone progressively increased in complexity along with the intricacy of the muscle fiber organization, from sternomastoid to extensor digitorum longus to anterior tibialis, so did both the relative volume occupied by the cojunctional acetylcholinesterase compartment and the proportion of G4. 4) The motor zone of the normal fast-twitch muscles characteristically differed from that of the soleus by the presence of a G4-rich environment around the endplates, which was cooperatively provided by the surrounding fibers. 5) In dystrophic muscles, this cojunctional G4-rich compartment was lost: the cojunctional cytochemical compartment was no longer discernable, while G4 was reduced to a minimal low level similar to that characteristic of the normal soleus.     

Effect of denervation on glucose uptake in rat soleus and extensor digitorum longus muscles

From 1 to 14 days after denervation, glucose uptake in the fast extensor digitorum longus and slow soleus muscles in rats was investigated and compared with that of the corresponding intact contralateral muscles. Denervation-induced atrophy in soleus was greater than that in extensor digitorum longus muscle. Glucose uptake in extensor digitorum longus muscle increased significantly, but that in soleus decreased significantly.     

Neural regulation of muscle activation

The effects of temperature and repetitive stimulation on isometric twitch contractions of normal, self-innervated and cross-innervated fast extensor digitorum longus and slow soleus muscles of the rat were studied in situ. Normal and self-innervated extensor digitorum longus showed a 2-fold increase in twitch tension following either a train of 200 stimuli at 35 C (post-tetanic potentiation) or lowering of temperature from 35 C to 20 C (cooling potentiation), while under these conditions the twitch tensions of normal and self-innervated slow soleus fell by about 15%. Post-tetanic and cooling potentiation were virtually abolished in cross-innervated extensor digitorum longus but appeared in cross-innervated soleus. The degree of potentiation in these and control muscles was inversely proportional to contraction time. These experiments suggest that mammalian motor nerves exert a controlling influence on the degree of activation of skeletal muscle during a twitch. An hypothesis of the molecular mechanism of post-titanic and cooling potentiation is proposed. According to this hypothesis, neural regulation of muscle activation results from the transformation of myosin after nerve cross-union.     

Upregulation of MHC class I in transgenic mice results in reduced force‐generating capacity in slow‐twitch muscle

Expression of major histocompatibility complex (MHC) class I in skeletal muscle fibers is an early and consistent finding in inflammatory myopathies. To test if MHC class I has a primary role in muscle impairment, we used transgenic mice with inducible overexpression of MHC class I in their skeletal muscle cells. Contractile function was studied in isolated extensor digitorum longus (EDL, fast‐twitch) and soleus (slow‐twitch) muscles. We found that EDL was smaller, whereas soleus muscle was slightly larger. Both muscles generated less absolute force in myopathic compared with control mice; however, when force was expressed per cross‐sectional area, only soleus muscle generated less force. Inflammation was markedly increased, but no changes were found in the activities of key mitochondrial and glycogenolytic enzymes in myopathic mice. The induction of MHC class I results in muscle atrophy and an intrinsic decrease in force‐generation capacity. These observations may have important implications for our understanding of the pathophysiological processes of muscle weakness seen in inflammatory myopathies. Muscle Nerve, 2008     

Effect of vinblastine on neural regulation of metabolism in rat skeletal muscle.

Chronic application of vinblastine, a substance known to disrupt axoplasmic flow, to nerves innervating the fast extensor digitorum longus and slow soleus muscles of the rat, produces electrophysiological signs of denervation (depolarization and extrajunctional acetylcholine sensitivity), but does not alter motor activity. We therefore examined the effects of vinblastine treatment on those metabolites and enzymes that are known to change after denervation of fast and slow skeletal muscle. A silastic cuff containing 0.1% vinblastine was placed around sciatic nerves of adult rats for 5 days. Glucose-6-P decreased 68% in the extensor, but did not change in soleus muscles. Phosphocreatine also decreased slightly, but significantly, in the extensor. Thus, intracellular levels of glucose-6-P and phosphocreatine in the extensor may be controlled, in part, by a factor (s) transported to the muscle by axoplasmic flow. Other metabolites known to change 5 days after denervation, namely glucose, glycogen, lactate, and α-ketoglutarate, were not altered in extensor and soleus muscles innervated by vinblastine-treated nerves. The activities of glucose-6-phosphate dehydrogenase and hexokinase increased in denervated extensor muscles, but not in denervated soleus muscles. Thus, metabolism in extensor muscles may be more readily altered after disruption of neural influences than is metabolism in soleus muscles. In contrast to denervation, exposure of sciatic nerves to vinblastine did not alter enzyme activities. These results provide evidence that certain metabolic processes, as well as membrane properties in skeletal muscle, are influenced by separate and distinct neural factors.     

Factors affecting muscle fiber transformation in cross. Reinnervated muscle

Nerves of two fast muscles [peroneus longus (PL) and extensor digitorum longus (EDL)], having different type 2 muscle fiber compositions, were used to cross-reinnervate the slow soleus muscle in the rat. Contraction characteristics, histochemical muscle fiber type compsotions and myosin heavy chain (MHC) isoform compositions were determined for the reinnervated muscles. Shortening velocity increased in soleus muscles crossreinnervated with EDL nerve [X-SOL(EDL)] but not in muscles cross-reinnervated with PL nerve [X-SOL(PL)]. Type 2A MHC isoform content was increased in X-SOL(EDL) but not in X-SOL(PL), where MHC isoform composition remained similar to normal soleus. The complement of type 1 (slow) muscle fibers was reduced and that of type 2 (fast) fibers increased in both types of X-SOL muscle, but this change was significantly greater in X-SOL(EDL); the majority of the type 2 fibers in X-SOL muscles were of type 2A. Results show that “the type 2 composition” of the reinnervating motoneuron pool is an important factor in determining the transformation of a target slow muscle after cross-reinnervation. © 1993 John Wiley & Sons, Inc.     

Appearance in slow muscule sarcolemma of specializations characteristic of fast muscle after reinnervation by a fast muscle nerve.

We utilized quantitative freeze-fracture electron microscopy to study the plasticity of orthogonal clusters of 60-Å particles (the “square array”) found in the sarcolemma of fast-twitch muscle. The membrane macromolecular composition of normally slow-twitch rat soleus muscle was examined 1 year after surgical reinnervation by the nerve from fast-twitch extensor digitorum longus muscles. The isometric contraction times and histochemical profiles were monitored and it was confirmed that conversion of fiber types had occurred. The sarcolemma of the switched “fast” soleus developed square arrays of 60-Å particles characteristic of fast-twitch muscle whereas the sarcolemma of the contralateral control, the “slow” soleus, contained only random particles. Square array density per square micrometer in cross-reinnervated fast soleus fibers resembled that of normal fast extensor digitorum longus muscles and varied as a function of distance from the neuromuscular junction. This experiment demonstrates that the appearance of these unusual clusters of 60-Å particles is neuronally regulated. We further suggest that these macromolecules are under the influence of the same subtle aspects of innervation that regulate the differentiation of myosin adenosine triphosphatase and thereby the contractile behaviors of fast- and slow-twitch muscle. The function of the sarcolemmal square array is unknown; however, a correlation with a membrane property that is more highly developed in fast-twitch muscle is to be expected.     

Improved Motor Function of Denervated Rat Hindlimb Muscles Induced by Embryonic Spinal Cord Grafts

Loss of motoneurons results in a decrease in force production by skeletal muscles and paralysis. Although it has been shown that missing motoneurons of rats can be replaced by embryonic homotopic neurons, attempts to guide their axons to their target muscles that have lost their innervation have been unsuccessful. In this study attempts were made to guide axons from grafted embryonic motoneurons to their target via a reimplanted ventral root. Adult hosts that received an embryonic graft prelabelled with 5-bromo-2′-deoxyuridine had their L4 ventral root avulsed and reimplanted into the spinal cord. Three to six months later, neurons that had their axons in the L4 ventral ramus were retrogradely labelled with fast blue and diamidino yellow. In five animals that had received an embryonic graft 116 ± 16 cells were retrogradely labelled, and of these at least 15% were of graft origin, since they were positive for 5-bromo-2′-deoxyuridine. In five animals that had their L4 ventral root reimplanted but did not receive a graft, only 12 ± 1.3 cells were retrogradely labelled. However, meaningful functional recovery could be achieved only if the regenerating axons of embryonic motoneurons found in the L4 ventral ramus were able to reverse the loss of force of muscles that had lost their innervation. This study shows that axons of embryonic motoneurons grafted into an adult rat spinal cord, as well as some axons of host origin, can be guided to denervated hindlimb muscles via reimplanted lumbar ventral roots. In normal rats ～30 motor axons innervated the extensor digitorurn longus and 60 innervated the tibialis anterior via the L4 ventral root. In rats that did not receive a graft only 3.7 ± 1.2 axons reached the extensor digitorum longus and 3.5 ± 0.4 reached the tibialis anterior muscle via the implanted L4 ventral root. In animals that had an embryonic graft, 7.6 ± 0.5 axons innervated the extensor digitorum longus and 8.5 ± 0.5 reached the tibialis anterior muscle via the implanted root. In rats without a transplant the maximum tetanic tension elicited by stimulating the implanted L4 root was 16 ± 7 g for the extensor digitorum longus and 53 ± 36 g for the tibialis anterior muscle, whereas the corresponding muscles in animals that had an embryonic graft developed 82 ± 16 and 281 ± 95 g respectively. Thus it appears that the grafted motoneurons contributed to the innervation and functional recovery of the denervated muscles.     

Permanent changes in muscle and motoneurones induced by nerve injury during a critical period of development of the rat

The sciatic nerve was crushed in rats at different times during the first two weeks after birth. Following reinnervation the recovery of the fast and slow muscles and their motoneurones was compared. The main factor affecting recovery of muscle weight and tension was the age at which the nerve was crushed; the earlier the injury the greater the impairment. However, recovery also depended upon muscle type. The fast muscles, tibialis anterior and extensor digitorum longus, always recovered less well than the slow soleus muscle. The greatest difference in recovery was seen when the nerve was crushed between 3 and 6 days of age. The fatigue resistance of fast muscles was markedly increased after nerve injury at any time during the first two postnatal weeks and was greatest when the nerve crush was done soon after birth. However, this change was not just related to muscle weakness as the increase in fatigue resistance after nerve crush at 5 and 12 days was similar regardless of the difference in recovery of the muscles. Retrograde labelling of motoneurones with HRP demonstrated that about 60-70% of motoneurones innervating fast or slow muscles were lost following sciatic nerve crush at birth. It is concluded that motoneurone loss probably accounts for most of the impairment of soleus after postnatal nerve crush but only partly explains the poor recovery of fast muscles.     

Reciprocal changes in myosin isoform mRNAs of rabbit skeletal muscle in response to the initiation and cessation of chronic electrical stimulation.

Changes in myosin heavy chain (MHC) mRNAs were studied in rabbit fast-twitch muscles during continuous electrical stimulation at 10 Hz for periods up to 3 weeks, and during the first 12 days of the recovery process that followed cessation of 6 weeks' stimulation. Two cDNA probes were used to detect MHC mRNAs specific to fast- and slow-twitch skeletal muscle in RNase protection assays and Northern- and slot-blot analyses. The isolation and base sequence of one of these probes, corresponding to the MHC gene expressed in soleus (slow-twitch), is described. At an early stage of the response to stimulation, fast MHC mRNA was replaced by slow MHC mRNA. During recovery, this process occurred in reverse but took longer. The time course of recovery was slightly faster in tibialis anterior than in extensor digitorum longus. The changes in mRNAs during both stimulation and recovery reflected changes in the corresponding muscle proteins.     

Temperature effects on the kinetics of force generation in normal and dystrophic mouse muscles

The kinetics of isolated extensor digitorum longus and soleus muscles from normal and genetically dystrophic (129/ReJ dy/dy) mice were studied at temperatures from 8 to 38 degrees C. The rate constants for the exponential rise of tetanic force and for the exponential decay of force during an isometric twitch or short tetanus were similar in normal and dystrophic soleus muscles, but the decay rates were significantly reduced in dystrophic extensor digitorum longus muscles. The temperature dependence for several rate constants for isometric twitches and tetani was similar in all muscles studied, suggesting that the same rate limiting processes apply to fast and slow, normal and dystrophic muscles. Thus, the contractile proteins and those in the sarcoplasmic reticulum of dystrophic muscle are probably normal. The slower relaxation phase in dystrophic extensor digitorum longus muscles is compatible with a reduction in Ca2+-pumping sites in the sarcoplasmic reticulum, perhaps secondary to a change in motor unit composition. Some changes in the temperature dependence for measured times, toward those of soleus muscles, is consistent with the increased proportion of slow twitch motor units in dystrophic extensor digitorum longus muscles.     

Histochemical changes in innervated and denervated skeletal muscle fibers following treatment with bupivacaine (marcain).

Widespread degenerative and regenerative changes were produced in the fast tibialis anterior and slow soleus muscles of rats by intramuscular injection of bupivacaine combined with hyaluronidase. The redifferentiation of muscle fiber types during the period of regeneration was followed in innervated and denervated muscles using histochemical methods. Evidence of differentiation was present in fast and slow innervated muscles 2 weeks after treatment and three fiber types could be recognized in the tibialis anterior at three weeks. By four weeks tibialis anterior and soleus showed a normal distribution of fiber types and associated differences in fiber caliber. Following denervation some enzyme activity was seen to appear in both muscles, but after four weeks fiber caliber remained markedly reduced and no differentiation of the enzyme activity of fibers was seen. It was concluded that a complete dedifferentiation of fibers had followed treatment with bupivacaine and that innervation was essential for the redifferentiation of fiber types.     

Long-term effects of spinal cord transection on fast and slow rat skeletal muscle: II. Morphometric properties

Morphometric properties of rat soleus and extensor digitorum longus muscles were studied 1 year following complete thoracic spinal cord transection (spinal cord level T9). Both muscles demonstrated almost complete type 1 to type 2 muscle fiber type conversion after 1 year. Muscle fiber atrophy was observed in both muscles. Type 2 fiber atrophy occurred to about the same extent in both muscles. Atrophy was most severe for the soleus type 1 fibers (50% decrease in size). Calculations based on the fiber type and size changes observed indicate that the percentage of the muscle cross-sectional area occupied by each fiber type was almost the same for both muscles 1 year after transection. Discriminant analysis of the data indicated that the percentage of type 2 fibers present in the muscle was the best discriminator between the various groups. These morphometric data provided a basis for understanding the contractile results presented in the previous study as well as insights into the mechanism of transformation in skeletal muscle. Furthermore, inherent differences between type 1 and type 2 fibers were demonstrated between predominantly slow and predominantly fast muscles. Thus, after almost one-half a lifetime of transection, rat muscles are almost completely transformed to fast muscle, and, regardless of initial conditions, have nearly identical properties.     

Cordotomy-denervation interactions on contractile and myofibrillar properties of fast and slow muscles in the rat

Cordotomy-denervation interactions were studied on contractile and myofibrillar properties of slow (soleus) and fast (extensor digitorum longus) muscles of the rat. The spinal cord was transected midthoracically in neonatal (2-day-old) animals. Two months after birth, a unilateral transection of the sciatic nerve was carried out in both cordotomized and control animals. Five weeks after denervation, contractile properties were tested isometrically in vitro; myofibrillar properties were assessed by histochemical staining of the muscle fibers and by electrophoretic analysis of the myosin heavy chain composition. The following results were obtained: (i) In cordotomized animals the contraction time of the soleus was significantly shorter (-23.3% on average) than that in the control animals and this shortening was accompanied by a proportional slow-to-fast shift in myofibrillar properties. (ii) The extensor digitorum longus properties were not significantly different in the control and cordotomized animals. (iii) Denervation in control animals was followed by a marked increase of contraction and half-relaxation times in the extensor digitorum longus, whereas in the soleus only the half-relaxation time was significantly increased; myofibrillar properties in the soleus showed an appreciable slow-to-fast shift, whereas in the fast muscle the main change was an increase in type 2A fibers to the detriment of type 2B. (iv) In cordotomized animals, denervation caused the soleus contraction time to increase to control values, whereas myofibrillar properties shifted to an even faster pattern; in the extensor digitorum longus denervation caused the same changes seen in the control animals. The results showed that cordotomy at birth caused the soleus to develop as a faster muscle than in the control animals. The concurrent effects of cordotomy and denervation on the myofibrillar properties of the soleus suggest that the slow-to-fast change in these properties is a common consequence of the reduction in the level of motor activity. The opposite effects of the two experimental conditions in the soleus contraction time support the view that the contractile alterations that follow denervation mainly reflect alterations in the muscle activation process.     

Physiological and histochemical changes of the extensor digitorum longus and soleus muscles after lateral cordotomy in the albino rat

The physiological and histochemical effects of unilateral and bilateral cordotomy on the extensor digitorum longus and soleus muscles were studied in the albino rat. After unilateral cordotomy, the ipsilateral extensor digitorum longus and soleus muscles became slower and relatively faster, respectively, compared to the contralateral and normal muscles. No gross histological abnormalities were found, but changes in fiber diameter and typology were statistically significant in both muscle types compared to their contralateral muscles, and were in substantial agreement with the changes observed on the contractile force and kinetics of the isometric twitch. Bilateral cordotomy in adult animals, like unilateral cordotomy in immature animals, affected only the soleus muscle fiber types. The results have implications for the role of the upper motoneuron physiological and histochemical trophic influences in controlling slow and fast motor units.