Innervation of developing intrafusal muscle fibers in the rat

The chronology of development of spindle neural elements was examined by electron microscopy in fetal and neonatal rats. The three types of intrafusal muscle fiber of spindles from the soleus muscle acquired sensory and motor innervation in the same sequence as they formed--bag2, bag1, and chain. Both the primary and secondary afferents contacted developing spindles before day 20 of gestation. Sensory endings were present on myoblasts, myotubes, and myofibers in all intrafusal bundles regardless of age. The basic features of the sensory innervation--first-order branching of the parent axon, separation of the primary and secondary sensory regions, and location of both primary and secondary endings beneath the basal lamina of the intrafusal fibers--were all established by the fourth postnatal day. Cross-terminals, sensory terminals shared by more than one intrafusal fiber, were more numerous at all developmental stages than in mature spindles. No afferents to immature spindles were supernumerary, and no sensory axons appeared to retract from terminations on intrafusal fibers. The earliest motor axons contacted spindles on the 20th day of gestation or shortly afterward. More motor axons supplied the immature spindles, and a greater number of axon terminals were visible at immature intrafusal motor endings than in adult spindles; hence, retraction of supernumerary motor axons accompanies maturation of the fusimotor system analogous to that observed during the maturation of the skeletomotor system. Motor endings were observed only on the relatively mature myofibers; intrafusal myoblasts and myotubes lacked motor innervation in all age groups. This independence of the early stages of intrafusal fiber assembly from motor innervation may reflect a special inherent myogenic potential of intrafusal myotubes or may stem from the innervation of spindles by sensory axons.     

Motor innervation of intrafusal fibers in rat muscle spindles: incomplete separation of dynamic and static systems

Distributions of 53 motor axons to different types of intrafusal fibers were reconstructed from serial 1-micron-thick transverse sections of 13 poles of spindles in the rat soleus muscle. The mean number of motor axons that innervated a spindle pole was 4.1. Approximately 60% of motor axons lost their myelination prior to or shortly after entry into the periaxial fluid space of spindles. Motor innervation to the juxtaequatorial portion of nuclear bag fibers (particularly the bag1) consisted of groups of short, synaptic contacts that were terminations of thin, unmyelinated axons. In contrast, motor endings on both the bag1 and bag2 fibers were platelike in the polar intracapsular region. Chain fibers had a single midpolar platelike ending. The ratio of motor axons that innervated the bag1 fiber exclusively to axons that innervated bag2 and/or chain fibers was 1:1. However, one-fourth of motor axons coinnervated the dynamic bag1 fiber in conjunction with static bag2 and/or chain fibers. Thus the complete separation of motor control of the dynamic bag1 and static bag2 intrafusal systems observed in cat tenuissimus spindles is neither representative of the pattern of motor innervation in all other species of mammals nor essential to normal spindle function.     

Innervation of regenerated spindles in muscle grafts of the rat

Features of the nerve supply and the encapsulated fibers of muscle spindles were assessed in grafted and normal extensor digitorum longus (EDL) muscles of rats by analysis of serial 10-microns frozen transverse sections stained for enzymes which delineated motor and sensory endings, oxidative capacity and muscle fiber type. The number of fibers was significantly more variable, and branched fibers were more frequently observed in regenerated spindles than in control spindles. Forty-eight percent of regenerated spindles received sensory innervation. Spindles reinnervated by afferents had a larger periaxial space than did spindles which were not reinnervated by afferents. Regenerated fibers innervated by afferents had small cross-sectional areas, equatorial regions with myofibrils restricted to the periphery of fibers, unpredictable patterns of nonuniform and nonreversible staining along the length of the fiber for 'myofibrillar' adenosine triphosphatase (mATPase) after acid and alkaline preincubation. In contrast, regenerated fibers devoid of sensory innervation resembled extrafusal fibers in that they usually exhibited myofibrils throughout the length of the fiber, no central aggregations of myonuclei, uniform staining for mATPase and a reversal of staining for mATPase after preincubation in an acid or alkaline medium. Approximately thirty percent of encapsulated fibers devoid of sensory innervation stained analogous to a type I extrafusal fiber, a pattern of staining never observed in intrafusal fibers of normal spindles. Groups of encapsulated fibers all exhibiting this pattern of staining reflect that either these fibers may have been innervated by collaterals of skeletomotor axons that originally innervated type I extrafusal fibers or that fibers innervated by only fusimotor neurons express patterns of staining for mATPase similar to extrafusal fibers in the absence of sensory innervation. Sensory innervation may also influence the reestablishment of multiple sites of motor endings on regenerated intrafusal fibers. Those regenerated fibers innervated by afferents had more motor endings than did regenerated fibers devoid of sensory innervation. Differences in size, morphology, and patterns of staining for mATPase and numbers of motor endings between fibers innervated by afferents and fibers devoid of sensory innervation reflect that afferents can influence the differentiation of muscle cells and the reestablishment of motor innervation other than during the late prenatal/early postnatal period when muscle spindles form and differentiate in rats.     

Slow-tonic MHC expression in paralyzed hindlimbs of fetal rats.

Whether nerve activity and active contraction of myotubes are essential for the assembly and initial differentiation of muscle spindles was investigated by paralyzing fetal rats with tetrodotoxin (TTX) from embryonic day 16 (E16) to E21, prior to and during the period when spindles typically form. TTX-treated soleus muscles were examined by light and electron microscopy for the presence of spindles and expression of myosin heavy chain (MHC) isoforms by the intrafusal fibers. Treatment with TTX did not inhibit the formation of a spindle capsule or the expression of a slow-tonic MHC isoform characteristic of intrafusal fibers, but did retard development of spindles. Spindles of TTX-treated E21 muscles usually consisted of one intrafusal fiber (bag2) only rather than two fibers (bag1 and bag2) typically present in untreated (control) E21 spindles. Intrafusal fibers of TTX-treated spindles also had only one sensory region supplied by multiple afferents, and were devoid of motor innervation. These features are characteristic of spindles in normal E18-E19 muscles. Thus, nerve and/or muscle activity is not essential for the assembly of muscle spindles, formation of a spindle capsule, and transformation of undifferentiated myotubes into the intrafusal fibers containing spindle-specific myosin isoforms. However, activity may promote the maturation of intrafusal bundles, as well as the maturation of afferent and efferent nerve supplies to intrafusal fibers.     

Multiple-bag-fiber muscle spindles in tenuissimus muscles of the cat

Over 300 complete and incomplete cat muscle spindles were examined in serial transverse sections of tenuissimus muscles in search of spindles with more than two nuclear bag intrafusal muscle fibers. Several histochemical and histological stains were used to identify the intrafusal fibers and assess their motor and sensory innervation. About 13% of the spindles contained either three or four bag fibers rather than the usual two. Every multiple-bag-fiber spindle possessed at least one nuclear bag1 and one nuclear bag2 fiber. The supernumerary bag fibers were either another bag1 and/or bag2 fiber, or a mixed bag fiber. The extra bag fibers had the usual morphologic and histochemical properties of cat nuclear bag fibers. All multiple-bag spindles received primary sensory innervation, and most had secondary sensory endings in addition. Their motor pattern was similar in the number, appearance and disposition of intrafusal motor endings to that of the usual two-bag-fiber spindles. Bag fibers of the same kind shared motor nerve supply in three multiple-bag spindles in which tracings of individual motor axons were obtained histologically. It is unclear whether any functional advantage is conveyed to a muscle spindle by its having more than one bag1 and one bag2 fiber.     

Aggregation of myonuclei and the spread of slow-tonic myosin immunoreactivity in developing muscle spindles.

The pattern of regional expression of a slow-tonic myosin heavy chain (MHC) isoform was studied in developing rat soleus intrafusal muscle fibers. Binding of the slow-tonic antibody (ATO) began at the equator of prenatal intrafusal fibers where sensory nerve endings are located, and spread into the polar regions of nuclear bag2 and bag1 fibers but not nuclear chain fibers during ontogeny. The onset of the ATO reactivity coincided with the appearance of equatorial clusters of myonuclei (nuclear bag formations) in bag1 and bag2 fibers. Moreover, the intensity of the ATO reaction was strongest in the region of equatorial myonuclei and decreased with increasing distance from the equator of bag1 and bag2 fibers at all stages of prenatal and postnatal development. The polar expansion of ATO reactivity continued throughout the postnatal development of bag1 fibers, but ceased shortly after birth in bag2 fiber coincident with innervation by motor axons. Thus, afferents that innervate the equator might induce the slow-tonic MHC isoform in bag2 and bag1 fibers by regulating the myosin gene expression by equatorial myonuclei, and efferents or twitch contractile activity might inhibit the spread of the slow-tonic MHC isoform into the poles of bag2 but not bag1 fibers. Absence of ATO binding in chain fibers suggests that chain myotubes may not be as susceptible to the effect of afferents as are myotubes that develop into bag2 and bag1 fibers. The different patterns of slow-tonic MHC expression in the three types of intrafusal fiber may therefore result from the interaction of three elements: sensory neurons, motor neurons, and intrafusal myotubes.     

Neural organization of spindles in three hindlimb muscles of the rat.

The neuroanatomical organization of the dynamic (bag1) and static (bag2 and chain) intrafusal systems was compared by light and electron microscopy of serial sections among 71 poles of muscle spindle in soleus (SOL), extensor digitorum longus (EDL), and lumbrical (LUM) muscles in the rat. Eighty-four percent of 195 fusimotor (gamma) axons to the spindles innervated either the dynamic bag1 fiber or the static bag2 and/or chain fibers. Sixteen percent of the gamma axons coinnervated the dynamic and static intrafusal fibers. Some of these nonselective axons were branches of effernts that also gave rise to axons selective to either the dynamic or static types of intrafusal fibers in one or more spindles. Thus activation of individual stem gamma efferents might not have a purely dynamic or purely static effect on the integrated afferent outflow from spindles of a hindlimb muscles in the rat. In addition, primary afferents in all muscles had terminations that cross-innervated the dynamic bag1 and static bag1 and/or chain intrafusal fibers in individual spindles, an arrangement that may enhance the mixed dynamic/static behavior of afferents when different intrafusal fibers are activated concurrent. Spindles of the slow SOL and fast EDL muscles had similar features, whereas differences were observed in the organization of the proximal (SOL and EDL) and distal (LUM) muscles. Spindles in LUM muscles had fewer static intrafusal fibers, a higher ratio of dynamic to static gamma axons, and a higher incidence of skeletofusimotor (beta) innervation to intrafusal fibers than spindles in the SOL or EDL muscles. Thus, the relative contribution of dynamic and static systems to muscle afferent outflow may differ among spindles located in different segments of the rat hindlimb. However, the dynamic and static intrafusal systems of spindle were less sharply demarcated in each of the three hindlimb rat muscles than in the cat tenuissimus muscle.     

Innervation of regenerated spindles in muscle grafts of the rat

Summary Features of the nerve supply and the encapsulated fibers of muscle spindles were assessed in grafted and normal extensor digitorum longus (EDL) muscles of rats by analysis of serial 10-m frozen transverse sections stained for enzymes which delineated motor and sensory endings, oxidative capacity and muscle fiber type.The number of fibers was significantly more variable, and branched fibers were more frequently observed in regenerated spindles than in control spindles. Forty-eight percent of regenerated spindles received sensory innervation. Spindles reinnervated by afferents had a larger periaxial space than did spindles which were not reinnervated by afferents. Regenerated fibers innervated by afferents had small cross-sectional areas, equatorial regions with myofi-brils restricted to the periphery of fibers, unpredictable patterns of nonuniform and nonreversible staining along the length of the fiber for myofibrillar adenosine triphosphatase (mATPase) after acid and alkaline preincubation. In contrast, regenerated fibers devoid of sensory innervation resembled extrafusal fibers in that they usually exhibited myofibrils throughout the length of the fiber, no central aggregations of myonuclei, uniform staining for mATPase and a reversal of staining for mATPase after preincubation in an acid or alkaline medium. Approximately thirty percent of encapsulated fibers devoid of sensory innervation stained analogous to a type I extrafusal fiber, a pattern of staining never observed in intrafusal fibers of normal spindles. Groups of encapsulated fibers all exhibiting this pattern of staining reflect that either these fibers may have been innervated by collaterals of skeletomotor axons that originally innervated type I extrafusal fibers or that fibers innervated by only fusimotor neurons express patterns of staining for mATPase similar to extrafusal fibers in the absence of sensory innervation. Sensory innervation may also influence the reestablishment, of multiple sites of motor endings on regenerated intrafusal fibers. Those regenerated fibers innervated by afferents had more motor endings than did regenerated fibers devoid of sensory innervation.Differences in size, morphology, and patterns of staining for mATPase and numbers of motor endings between fibers innervated by afferents and fibers devoid of sensory innervation reflect that afferents can influence the differentiation of muscle cells and the reestablishment of motor innervation other than during the late prenatal/early postnatal period when muscle spindles form and differentiate in rats.     

Heterogeneity of spindle units in the cat tenuissimus muscle

Three tandem spindles and their nerve supplies, reconstructed by light microscopy of serial transverse sections of the cat tenuissimus muscle, were compared to single spindle units. Each tandem spindle consisted of one large unit containing a dynamic bag1, a static bag2, and several static chain fibers (b1b2c unit) linked by the bag2 fiber to a small unit containing only a bag2 and chain fibers (b2c unit). Most features of primary afferents, secondary afferents, and motor neurons were qualitatively and quantitatively similar in both single and tandem b1b2c units. However, b1b2c units of tandem spindles had a lower density of skeletofusimotor innervation than did single b1b2c spindles. The b2c spindle units differed greatly from single or tandem b1b2c units. The b2c spindle units had fewer intrafusal fibers and incoming axons than either the tandem or single b1b2c units. The motor innervation of b2c units was typified by nonselective gamma axons that coinnervated both bag2 and chain fibers, in contrast to the regular occurrence of both selective and nonselective motor axons in b1b2c spindle units. The afferent located at the equator of b2c units differed in size, branching pattern, and intrafusal distribution of its ending from both the primary and secondary sensory axons of b1b2c units and, therefore, might represent a third category of spindle afferent. Thus, cat tenuissimus muscles contain three types of spindle units that differ in the number and organization of muscular and neural elements. These differences in structure and neural organization among tenuissimus spindle units may be a source for generation of different sensory signals in response to common mechanical or fusimotor stimuli.     

One-bag-fiber muscle spindles in tenuissimus muscles of the cat

Over 150 complete and 139 incomplete single muscle spindles were examined in serial transverse sections of cat tenuissimus muscles in search for spindles lacking one of the two types of nuclear bag intrafusal fiber. Several histochemical reactions were used to type the intrafusal muscle fibers and assess the spindle motor and sensory innervation. One complete spindle lacked a bag1 fiber, and another spindle lacked a bag2 fiber. Several incomplete spindles also lacked bag1 fibers. In addition, ten double tandem spindles contained one capsular unit each that lacked the bag1 fiber, and one triple tandem spindle had two such capsules. All one-bag-fiber spindles had primary sensory innervation, but none had secondary sensory innervation. Their motor innervation was similar to that of the usual two-bag-fiber spindles in the number and disposition of intrafusal motor endings. It is unclear whether the one-bag fiber spindles, either single or tandem-linked, are products of an aberrant spindle development or represent a true anatomical and functional subcategory of the cat muscle spindle.     

Development of immunohistochemical characteristics of intrafusal fibres in normal and de-efferented rat muscle spindles

Intrafusal muscle fibres in adult muscle spindles differ in their myosin composition. After selective motor denervation intrafusal muscle fibres develop mature ultrastructural characteristics. In order to evaluate the role of fusimotor innervation on the maturation of the myosin composition of intrafusal muscle fibres we have examined with immunohistochemical techniques i) the postnatal development of muscle spindles in new-born rats and in 7-21 day old rats; ii) muscle spindles in the EDL of 21-day-old rats de-efferented at birth. For the characterization of myosins in intrafusal fibres we used three myosin antisera: antipectoral myosin, antiheart myosin and antiheart myosin adsorbed with muscle powder from the soleus muscle of guinea pig. We show in this study that during development intrafusal fibres change immunoreactivity and that in the absence of motor innervation bag fibres do not fully develop the myosin characteristics of control spindles. We conclude that the maturation of bag1 and bag2 fibres apparently requires next to the inductive influence of sensory axon terminals the presence and activity of fusimotor axons.     

Nonselective motor innervation of nuclear bag1 intrafusal muscle fibers in the cat

The motor nerve supply to cat nuclear bag1 intrafusal muscle fibers was reconstructed from light and electron microscopy of serial transverse sections of spindles in the tenuissimus muscle. Twenty-six of thirty poles of bag1 fibers that were examined received motor innervation. Every innervated bag1 pole received at least one (range 1-3) selective motor axon that supplied this fiber type only. Four of the innervated bag1 poles (15%) received additional motor supply from a nonselective motor axon that also innervated one nuclear chain fiber in the same spindle pole. The chain fibers co-innervated with bag1 fibers were among the longest chain fibers although they were shorter than two long chain fibers also present in the spindle poles. In cross-sections stained with toluidine blue they displayed 1-3 equatorial nuclei side by side, and there were fewer intermyofibrillar granules in their polar regions than in most of the other chain fibers. The endings of nonselective motor axons on the bag1 and chain fibers were morphologically and ultrastructurally dissimilar. It is suggested that instances of common innervation of the (dynamic) bag1 fiber and a (static?) chain fiber represent an integral and, presumably, functionally meaningful part of the motor pattern in some cat spindles.     

Slow-tonic MHC expression in paralyzed hindlimbs of fetal rats

Summary Whether nerve activity and active contraction of myotubes are essential for the assembly and initial differentiation of muscle spindles was investigated by paralyzing fetal rats with tetrodotoxin (TTX) from embryonic day 16 (E16) to E21, prior to and during the period when spindles typically form. TTX-treated soleus muscles were examined by light and electron microscopy for the presence of spindles and expression of myosin heavy chain (MHC) isoforms by the intrafusal fibers. Treatment with TTX did not inhibit the formation of a spindle capsule or the expression of a slow-tonic MHC isoform characteristic of intrafusal fibers, but did retard development of spindles. Spindles of TTX-treated E21 muscles usually consisted of one intrafusal fiber (bag₂) only rather than two fibers (bag₁ and bag₂) typically present in untreated (control) E21 spindles. Intrafusal fibers of TTX-treated spindles also had only one sensory region supplied by multiple afferents, and were devoid of motor innervation. These features are characteristic of spindles in normal E18–E19 muscles. Thus, nerve and/or muscle activity is not essential for the assembly of muscle spindles, formation of a spindle capsule, and transformation of undifferentiated myotubes into the intrafusal fibers containing spindle-specific myosin isoforms. However, activity may promote the maturation of intrafusal bundles, as well as the maturation of afferent and efferent nerve supplies to intrafusal fibers.     

Histochemistry of rat intrafusal muscle fibers and their motor innervation.

Muscle spindles were followed in serial transverse sections of freshly frozen rat soleus muscles. Adenosine triphosphatase (ATPase) histochemical staining reaction was used to identify nuclear bag1, nuclear bag2 and nuclear chain intrafusal muscle fibers. Regional differences in ATPase staining occurred along bag1 and bag2 fibers but not along chain fibers. Bag1 fibers displayed ultrastructural heterogenity when their intra- and extracapsular regions were compared. Simple "diffuse" and more elaborate "plate" motor nerve terminals were demonstrated histochemically along the poles of bag1 and bag2 fibers by staining for cholinesterase. One motor terminal of the "plate" appearance was present on a chain fiber pole. There was no consistent spatial correlation between the intensity of regional ATPase staining along the nuclear bag fibers and the location, number and type of motor endings. Other factors, such as intrafusal fiber sensory innervation and regional differences in active and passive functional recruitment of nuclear bag fibers during muscle activity, may contribute to the ATPase staining variability along the intrafusal fibers.     

One-bag-fiber muscle spindles in tenuissimus muscles of the cat

Summary Over 150 complete and 139 incomplete single muscle spindles were examined in serial transverse sections of cat tenuissimus muscles in search for spindles lacking one of the two types of nuclear bag intrafusal fiber. Several histochemical reactions were used to type the intrafusal muscle fibers and assess the spindle motor and sensory innervation. One complete spindle lacked a bag₁ fiber, and another spindle lacked a bag₂ fiber. Several incomplete spindles also lacked bag₁ fibers. In addition, ten double tandem spindles contained one capsular unit each that lacked the bag₁ fiber, and one triple tandem spindle had two such capsules. All one-bag-fiber spindles had primary sensory innervation, but none had secondary sensory innervation. Their motor innervation was similar to that of the usual two-bag-fiber spindles in the number and disposition of intrafusal motor endings. It is unclear whether the one-bag fiber spindles, either single or tandem-linked, are products of an aberrant spindle development or represent a true anatomical and functional subcategory of the cat muscle spindle.     

Development of immunohistochemical characteristics of intrafusal fibres in normal and de-efferented rat muscle spindles

Summary Intrafusal muscle fibres in adult muscle spindles differ in their myosin composition. After selective motor denervation intrafusal muscle fibres develop mature ultrastructural characteristics. In order to evaluate the role of fusimotor innervation on the maturation of the myosin composition of intrafusal muscle fibres we have examined with immunohistochemical techniques i) the postnatal development of muscle spindles in new-born rats and in 7–21 day old rats; ii) muscle spindles in the EDL of 21-day-old rats de-efferented at birth. For the characterization of myosins in intrafusal fibres we used three myosin antisera: antipectoral myosin, antiheart myosin and antiheart myosin adsorbed with muscle powder from the soleus muscle of guinea pig. We show in this study that during development intrafusal fibres change immunoreactivity and that in the absence of motor innervation bag fibres do not fully develop the myosin characteristics of control spindles. We conclude that the maturation of bag₁ and bag₂ fibres apparently requires next to the inductive influence of sensory axon terminals the presence and activity of fusimotor axons.     

Aggregation of myonuclei and the spread of slow-tonic myosin immunoreactivity in developing muscle spindles

Summary The pattern of regional expression of a slow-tonic myosin heavy chain (MHC) isoform was studied in developing rat soleus intrafusal muscle fibers. Binding of the slow-tonic antibody (ATO) began at the equator of prenatal intrafusal fibers where sensory nerve endings are located, and spread into the polar regions of nuclear bag₂ and bag₁ fibers but not nuclear chain fibers during ontogeny. The onset of the ATO reactivity coincided with the appearance of equatorial clusters of myonuclei (nuclear bag formations) in bag₁ and bag₂ fibers. Moreover, the intensity of the ATO reaction was strongest in the region of equatorial myonuclei and decreased with increasing distance from the equator of bag₁ and bag₂ fibers at all stages of prenatal and postnatal development. The polar expansion of ATO reactivity continued throughout the postnatal development of bag₁ fibers, but ceased shortly after birth in bag₂ fiber coincident with innervation by motor axons. Thus, afferents that innervate the equator might induce the slow-tonic MHC isoform in bag₂ and bag₁ fibers by regulating the myosin gene expression by equatorial myonuclei, and efferents or twitch contractile activity might inhibit the spread of the slow-tonic MHC isoform into the poles of bag₂ but not bag₁ fibers. Absence of ATO binding in chain fibers suggests that chain myotubes may not be as susceptible to the effect of afferents as are myotubes that develop into bag₂ and bag₁ fibers. The different patterns of slow-tonic MHC expression in the three types of intrafusal fiber may therefore result from the interaction of three elements: sensory neurons, motor neurons, and intrafusal myotubes.     

Treatment with beta bungarotoxin blocks muscle spindle formation in fetal rats

Sensory and motor fibers of peripheral nerves were irreversibly destroyed in fetal rats by administering beta bungarotoxin (BTX) on embryonic day 16 or 17, after assembly of primary myotubes, but before the formation of muscle spindles. Soleus muscles of toxin-treated fetuses and their untreated littermates were removed just prior to birth and were examined by light microscopy of serial transverse sections for the presence of spindles and immunocytochemical expression of several isoforms of myosin heavy chains (MHC). Untreated muscles exhibited numerous spindles that were innervated by branches of intramuscular nerves and contained muscle fibers expressing a slow-tonic MHC isoform characteristic of the intrafusal but not extrafusal fibers. Toxin-treated muscles were devoid of intramuscular nerve bundles and perineurial structures. Encapsulations of muscle fibers resembling spindles were absent and no myotubes expressed the slow-tonic MHC isoform associated with intrafusal fibers in beta BTX-treated muscles. Thus, the assembly of muscle spindles, formation of the spindle capsule, and transformation of undifferentiated myotubes into the intrafusal fibers that contain spindle-specific myosin isoforms all depend on the presence of innervation in prenatal rat muscles.     

Morphometric analysis of coated vesicles in developing rat muscle spindles.

The incidence of coated vesicles under sarcolemmal surfaces of equatorial, juxta-equatorial and polar regions in developing and adult spindles of the rat soleus muscle was examined by quantitative morphometry of transverse ultrathin sections. Coated vesicles were more numerous: 1) under primary sensory endings than under other types of neuromuscular contacts; 2) under the appositional sarcolemma between neighbouring intrafusal fibres than under free surfaces of the sarcolemma; and 3) in developing than in mature spindles. Factors such as location and age of the animal often interacted to produce an additive effect on the incidence of coated vesicles. Although there was a high incidence of coated vesicles at the postsynaptic surface under sensory terminals of bag2 fibres in 18 and 19 day gestational embryonic rats, it peaked in 4 day postnatal animals. The high incidence of coated vesicles at sensory endings supports the view that coated vesicles mediate neurotrophic interactions between afferents and intrafusal fibres during the critical late gestation and early postnatal time period, as sensory axons first contact their target fibres and exert a maximal directing influence on the differentiation of intrafusal fibre types. In addition, the preferential localization of coated vesicles under appositional rather than free surfaces of developing intrafusal fibres in 0-4 day rats suggests that they play a role in the transport of active substances among intrafusal fibres exhibiting different stages of maturity.