During metamorphosis the leg neuromuscular system of the mothManduca sexta undergoes an extensive remodeling as the larval muscles degenerate and are replaced by new muscles in the adult. The terminal processes of persistent leg motoneurons undergo severe regression followed by regrowth (Consoulas et al., 1996), accompanied, as shown here, by the loss and re-establishment of functional presynaptic specializations. Before and shortly after the degeneration of the larval muscle, immunoreactivity for the vesicular protein synaptotagmin was localized to the presynaptic varicosities of the motoneurons. Similarly localized were distinct sites of Ca2+-dependent uptake of the fluorescent dye FM1-43. During myoblast migration and accumulation about the re-expanding motor axons, synaptotagmin immunoreactivity was widely distributed in axons, and specific FM1-43 staining revealed vesicle exocytosis in distal axon branches. During myoblast proliferation and fusion, and myotube formation, synaptotagmin staining remained widely distributed in nerve branches, whereas FM1-43 staining was more localized to subdomains of these nerve branches. These initial presynaptic active sites were transient and were replaced by new sites in more distal nerve processes as the muscle anlage increased in size and additional myotubes formed. After myotube separation, synaptotagmin staining disappeared from primary branches but remained distributed within secondary and high-order nerve branches. FM1-43 staining was detected in high-order branches only. During muscle fiber striation, growth, and maturation, both FM1-43 staining and synaptotagmin immunoreactivity became localized to terminal varicosities. Thus, presynaptic function can persist after the loss of the target and occurs transiently in axon shafts before becoming restricted to terminal domains as the underlying muscle fibers mature.
Functional signal transmission at the neuromuscular junction requires the precise alignment and differentiation of specialized regions of both the presynaptic and postsynaptic cells. This alignment is maintained in the mature organism through cell contact and diffusible signals between nerve and muscle (Hall and Sanes, 1993; Connor and Smith, 1994; Keshishian et al., 1996). Postsynaptic receptors that are initially diffusely distributed on the myotube surface become localized to the postsynaptic region because of signals from the motoneuron (Broadie and Bate, 1993a,b; Hall and Sanes, 1993). Similarly, the alignment of presynaptic active sites and the clustering of synaptic vesicles at those sites depend on signals derived from the muscle (Sanes et al., 1978; Hall and Sanes, 1993; Noakes et al., 1995).
Although the synthesis and distribution of synaptic vesicle proteins in growing neurites occurs before synaptic contact, the restriction to the presynaptic terminals is initiated after the initial growth cone–muscle fiber contact (Lupa and Hall, 1989; Littleton et al., 1993). Genetic elimination of target muscles in Drosophiladoes not prevent the formation of presynaptic zones, suggesting that their initial assembly is a process autonomous to the motoneurons, but correct localization is target-dependent (Prokop et al., 1996). Similarly, synaptic vesicles present in the axons of cultured hippocampal neurons can aggregate and undergo Ca2+-dependent exocytosis in the absence of postsynaptic contacts (Kraszewski et al., 1995). Whether the aggregation of synaptic vesicles and their capacity for exocytosis in immature axons has a functional significance in vivo is not known. Similarly unclear are the mechanisms that ensure precise localization of presynaptic specializations once contact with an appropriate target is achieved or that allow this localization to be modified during postembryonic synaptic plasticity.
The leg neuromuscular system of the moth Manduca sextaundergoes a dramatic remodeling during metamorphosis that provides a useful model system for addressing the localization of presynaptic specializations. The larval legs and associated muscles degenerate and are replaced by a new set of legs and muscles in the adult (Kent et al., 1995; Consoulas et al., 1997). Both sets of muscles are innervated by the same population of motoneurons (Kent and Levine, 1988). The axons and terminal processes of these motoneurons remain in the periphery throughout metamorphosis but undergo extensive regression and growth (Consoulas et al., 1996). In the present study the functional remodeling of presynaptic motor terminals was investigated by following the redistribution of synaptotagmin, an integral membrane protein known to play a role in docking and fusion of synaptic vesicles (Perin et al., 1990), and the capacity for Ca2+-dependent synaptic vesicle recycling, as assessed with the fluorescent dye FM1-43 (Betz and Bewick, 1992). Ca2+-dependent vesicle exocytosis continues within regressed axons after muscle degeneration in the absence of a target during the early stages of motor terminal remodeling. As muscle fibers mature, sites of vesicular recycling shift within growing axons, progressively becoming restricted to mature presynaptic terminals. This remarkable example of synaptic remodeling provides a natural model for further exploration of cellular interactions that ensure proper neuromuscular function.
MATERIALS AND METHODS
Animals. M. sexta (L.) were obtained from a laboratory culture reared on an artificial diet (Bell and Joachim, 1976) under a long-day photoperiod regimen (17/7 hr light/dark cycle) at 26°C and ∼60% relative humidity. Both chronological and morphological criteria were used for the staging of animals (Nijhout and Williams, 1974; Bell and Joachim, 1976;Reinecke et al., 1980; Tolbert et al., 1983; Consoulas et al., 1996, 1997). In summary, L0, L1, L2, and L3 represent the days of the last (fifth) larval instar, W0 signifies the first day of wandering, and W1 to W4 represent the remaining days before pupation. After pupation, stage P0 indicates the day of the pupal molt, and P1 through P18 are the next stages of adult development.
Nerve staining techniques. Biocytin filling was used to reveal the details of the peripheral branching of leg motoneurons (Horikawa and Armstrong, 1988; Consoulas et al., 1996). To fill the peripheral axons of the leg motoneurons in the orthograde direction, the animals were first anesthetized by chilling on ice. After removing the head and abdomen, the thoracic segments were dissected along the dorsal midline and pinned down on a Sylgard-coated Petri dish in saline. The whole prothoracic ganglion with intact nerves was isolated in a Vaseline pool to allow the infusion of a biocytin solution (3% w/v biocytin in distilled water; Sigma, St. Louis, MO). The preparations were stored at 7°C. After biocytin infusion for a maximum of 2 d, the preparations were dissected and fixed in freshly prepared solution containing 4% paraformaldehyde, 0.15% glutaraldehyde, and 0.2% saturated picric acid in 0.1 mphosphate buffer, pH 7.4, overnight (Sun et al., 1993). They were subsequently dehydrated in ethanol, permeabilized in xylol or propylene, rehydrated in ethanol, and washed in 10 mmPBS, pH 7.4, three times for 15 min each and in PBS containing 1% Triton X-100 (PBSX) three times for 15 min each. To block nonspecific staining the preparations were incubated in 10% normal goat serum (NGS; Jackson ImmunoResearch, West Grove, PA) and 3% bovine serum albumin (BSA; Boehringer Mannheim, Indianapolis, IN) in PBSX for 1 hr. They were then incubated in Cy3-conjugated streptavidin (Jackson ImmunoResearch) for 5–12 hr in 7°C. The preparations then were washed several times with PBS, dehydrated in ethanol, and cleared in methyl salicylate.
Synaptotagmin immunostaining. The distribution of immunoreactivity for the presynaptic protein synaptotagmin was examined in whole-mount preparations with a polyclonal antibody raised againstDrosophila synaptotagmin (DSYT2; Littleton et al., 1993; generously provided by H. J. Bellen, Howard Hughes Medical Institute, Baylor University, Houston, TX). The protocol was similar to that used in a previous study of the Manduca leg system (Consoulas et al., 1996). Developing and adult leg muscles were dissected in cold saline and fixed in freshly made 4% paraformaldehyde for 1 hr at room temperature. After rinsing in 10 mm PBS and 10 mm PBS with 0.2% Triton X-100 (PBSX) for 2 hr, they were blocked for 1 hr (six times for 10 min each) in a Tris-HCl buffer, pH 7.0, containing 10% Triton X-100, 1% Na azide, 0.25% BSA, and 2% NGS. The preparations were then incubated overnight in primary antiserum (1:1000) made up in the blocking buffer. After washing in PBSX and PBS for 2 hr they were incubated in Cy3-conjugated secondary antibody for 4–8 hr at 4°C. The preparations were then rinsed in PBSX and PBS, dehydrated, and cleared in methyl salicylate. The results presented here are based on a minimum of 10 preparations from each developmental stage. No staining was observed at any developmental stage in parallel preparations in which no primary antibody was used. We have repeated with identical results the immunostaining of stages L2, P18, and critical intermediate stages (P4–P10) using a polyclonal antibody recently generated against Manduca synaptotagmin (kindly provided by S. H. Dubuque and L. P. Tolbert, Division of Neurobiology, University of Arizona, Tucson, AZ). The synaptotagmin protein shows a high degree of homology between the two insect species (Dubuque et al., 1997).
Preparations from different developmental stages were processed together in the same dish and viewed with a confocal microscope (see below). Background staining varied from stage to stage because of changes in the types of surrounding tissues (e.g., intact muscle, developing muscle, and epidermis). The background level on a gray scale was held constant among preparations to adjust for this variability. Thus, the aperture and neutral-density settings remained constant, whereas the gain and black-level settings were adjusted over a small range ( of the full scale). These minor changes in microscope settings did not alter the apparent distribution of immunoreactivity as tested by varying the settings over this range while acquiring images of mature and developing stages.
FM1-43 staining.The fluorescent dye FM1-43 (Molecular Probes, Eugene, OR) was used to monitor synaptic vesicle exocytosis and recycling (Betz and Bewick, 1992). The legs were dissected from the animal and pinned to a Sylgard-coated Petri dish that was attached firmly with wax to a microscope slide. The leg nerve 2a was stimulated with a Grass Instruments S 88 stimulator via a saline-filled suction electrode with 1–5 Hz pulses for 5–10 min in the presence of 4 μm FM1-43 in saline (see Table1 for specific details). The strength of stimulation was adjusted to recruit all of the pretarsal flexor motoneurons (see below). The saline consisted of (in mm): 140 NaCl, 5 KCl, 4 CaCl2, 28 glucose, and 5N-2-hydroxyethylpiperzeine-N′-ethanesulfonic acid, pH 7.4 (Trimmer and Weeks, 1989). The evoked postsynaptic responses were intracellularly recorded via glass electrodes filled with 2 m potassium acetate (tip resistance, 40–60 MΩ). The signals were amplified with an Axoclamp-2A (Axon Instruments) amplifier and recorded on an eight-channel video recording system (Vetter 3000A) and subsequently transferred to a computer (acquisition sample rate, 10 Khz) for analysis using Data-Pac II software (Run Technologies).
For imaging, preparations were transferred to a Bio-Rad (Cambridge, MA) 600 krypton/argon confocal laser scanning microscope and viewed through a Zeiss 40× water immersion objective (488 nm excitation filter). The same area could be viewed repeatedly, after FM1-43 unloading, reloading, and synaptotagmin staining, by aligning the slide at the same x,y coordinates on the microscope stage, in addition to using landmarks such as trachea. The specific pattern of staining that was observed was consistent with previous observations of insect neuromuscular junctions (Ramaswami et al., 1994). Experiments that were performed to ensure the specificity of staining are described in Results.
Muscle-staining techniques. In many cases the state of internal structures within the legs was examined in serial longitudinal sections. The legs were removed from the animals, fixed in alcoholic Bouin’s fixative for 2–3 d, embedded in paraffin (paraplast), and serially sectioned (10–12 μm). After deparaffinization and rehydration, the sections were stained with hematoxylin-eosin.
5-Bromodeoxyuridine labeling. To reveal the number and locations of nuclei undergoing DNA replication cells, 50 μg/gm body weight 5-bromodeoxyuridine (BrdU, Sigma) dissolved in distilled water was injected into the animals at specific developmental stages 12 hr before their dissection. The prothoracic legs then were fixed for 2 d in alcoholic Bouin’s or Carnoy’s fixative, embedded in paraffin, and sectioned. After deparaffinization, rehydration, and extensive washing in PBS and PBSX (0.1% Triton X-100), the DNA was denatured by treatment with 2N HCl in PBS for 15 min. Nonspecific activity was blocked with 10% NGS in PBS for 30 min. The sections were then incubated for 2 hr in the primary antibody against BrdU (Becton Dickinson, Mountain View, CA) diluted 1:100 in PBS with 5% NGS. After washing the sections in PBSX and PBS for 1 hr, they were incubated in goat anti-mouse secondary antibody diluted 1:200 (Cy3- or fluorescein-conjugated, Sigma) in PBS for 1 hr.
Terminal deoxynucleotidyl transferase-mediated dUTP nick end label staining. Apoptotic nuclei of degenerating muscles were revealed with the terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling (TUNEL) technique following the instructions of the manufacturer (Boehringer-Mannheim).
Labeling of filamentous actin and nuclei. To determine the presence of intact or differentiated muscles in whole mounts and sectioned preparations, filamentous actin within the muscles was labeled with BODIPY FL phallacidin or Oregon Green phalloidin (both from Molecular Probes). Whole-mount preparations were fixed in 4% paraformaldehyde in PBS for 4 hr to overnight. After washing with PBS and PBSX (1% Triton-100 in PBS) several times, nonspecific activity was blocked with 10% NGS in PBSX for 30 min, and then the preparations were incubated in 66 nm BODIPY FL phallacidin or Oregon Green phalloidin in PBS overnight in 4°C. In many cases, the nuclei of all classes of cells were revealed by incubating in 25 μm propidium iodide (Sigma) in PBS for 7 min (Sun et al., 1993) without pretreatment with RNase to eliminate the free distributed RNA in the cytoplasm of the cells. Therefore, in addition to the nucleus, the cytoplasm was lightly stained.
Confocal microscopy. The stained preparations were viewed with a confocal microscope (MRC-600 with a Nikon Optiphot-2 microscope and a krypton/argon laser light source; Bio-Rad). In cases in which two dyes were used, the images were merged by using different pseudocolors (red for Cy3-conjugated streptavidin or propidium iodide; green for streptavidin-dichlorotriazinyl amino fluorescein, Oregon Green phalloidin, FM1-43, or BODIPY FL phallacidin). Images were prepared using Confocal Assistant (Bio-Rad) and Corel Draw 6 (Corel) and printed on a Tektronix dye-sublimation printer.
Remodeling of the pretarsal flexor muscle during metamorphosis
General features of the remodeling of the Manduca leg neuromuscular system have been described previously (Consoulas et al., 1997). The present study focused on the pretarsal flexor muscle (PrtFlx) of the prothoracic legs to understand the relationship between presynaptic and postsynaptic differentiation. In the larva, this muscle has closely packed fibers bearing different sizes of nuclei (Fig.1 A). During the initial days of the last larval instar (L0–L3) the muscle was intact and responded to mechanical or electrical stimulation. PrtFlx muscle degeneration occurred between stages W0 and W3. At the same time, the unguitractor tendon, onto which the adult PrtFlx muscle will later attach, began to develop from an invagination of the epidermis near the tip of the imaginal (developing adult) leg. Muscle degeneration on day W0 was marked by the appearance of gaps between the fibers and, by stage W2, many large vacuoles in the cytoplasm and blistering of the sarcolemma (data not shown). By the end of stage W2, the cross-striations had disappeared, and within a few hours many of the larval myonuclei became apoptotic (Fig. 1 B). After pupation, a homogeneous population of spindle-shaped cells (imaginal myoblasts; see Consoulas and Levine, 1997), originating mainly in coxa, migrated and became distributed within the leg segments (Fig.1 C). By the end of stage P2, myoblasts began to accumulate close to the unguitractor tendon in which the terminal processes of the PrtFlx motoneurons grew, became aligned at a 30° angle with the tendon, and fused (Fig. 1 D; Consoulas et al., 1996;Consoulas and Levine, 1997). During subsequent pupal stages (P3–P8) the initial muscle anlage increased in size because of the continued accumulation and proliferation of myoblasts (Fig.1 E,F). Myotube formation began by stage P4, separation of myotubes from each other began by stage P6, and by stage P8 all of the fibers were clearly separate. Throughout the same period, free myoblasts continued to accumulate, proliferate, and fuse with the muscle fibers that had already formed (Fig. 1 G). Myoblast proliferation declined by the time the fibers became striated (stage P9). During subsequent stages of adult development, the muscle fibers increased further in diameter and became well-separated from each other, with myonuclei distributed near the outer surface of the fibers (Fig. 1 H).
Remodeling of PrtFlx muscle innervation during metamorphosis
The larval leg motoneurons persist during metamorphosis to innervate the new adult leg muscles (Kent and Levine, 1988). The peripheral processes of these persistent motoneurons first retract after the loss of their larval target muscles and then re-expand to innervate the adult muscles (Consoulas et al., 1996). The PrtFlx muscle is supplied by three persistent excitatory motoneurons (K. Oanh-Phan and U. Rose, personal communication). In both stages, individual fibers are either singly innervated by one fast motoneuron or dually innervated by one fast and one slow motoneuron, which were not distinguished in this study. To reveal the developmental fate of the motoneuron terminal axon branches, the main leg nerve 2a was filled with biocytin. In the larva, the motoneuron axons run through nerve 2a to supply the PrtFlx muscle fibers (Fig.2 A). During the early phase of PrtFlx muscle degeneration (stage W2), the high-order motor branches began to retract. After the breakdown of the muscle (stage W3) several secondary and high-order branches, as well as many terminal varicosities, disappeared, whereas the remaining branches occupied a central position within the femorotibial segment of the imaginal leg (Fig. 2 B). By the end of the larval life (stage W4) the imaginal leg had grown considerably, and the retracted PrtFlx motor branches began to re-expand. During stage P2 the retracted nerve axons continued to expand over the inner surface of the imaginal epidermis in contact with accumulating imaginal myoblasts (Fig. 2 C). By stage P3, the secondary and high-order nerve axons were expanding over the developing PrtFlx muscle anlage, as thin processes started to appear along these branches. Extensive nerve growth was apparent during the next days of pupal development (stages P5–P10; Fig.2 D,E). By stage P12 the innervation pattern had most of the features of the adult pattern, but distal nerve processes were still growing (Fig. 2 F). The remaining stages of pupal development (P12–P18) were devoted to the establishment of adult terminal varicosities (see below).
In summary, the axons and terminal processes of the PrtFlx motoneurons undergo extensive remodeling after the larval muscle degeneration (stages W2–W4), characterized by a phase of axonal retraction and the loss of terminal varicosities, followed by a phase of axonal growth, during which myoblasts migrate and accumulate close to the terminal processes of the motoneurons (stages W4–P3). A third phase of rapid and extensive nerve growth over the developing PrtFlx muscle anlage follows (stages P3–P12), with higher-order branch growth and maturation of the adult terminal varicosities marking the final phase.
As an initial step in determining the fate of presynaptic specializations during the retraction and re-expansion of PrtFlx motoneuron terminals, the distribution of immunoreactivity for the synaptic vesicle membrane protein synaptotagmin was examined (Figs.3, 4). During the first days of the last larval instar (stages L0–W2b-late), when the PrtFlx muscle is intact, synaptotagmin immunoreactivity was restricted to terminal varicosities (Figs. 3 A,4 A). After muscle degeneration (stages W3 and W4), terminals became enlarged, reminiscent of the “retraction bulbs” seen in vertebrate muscles during synapse elimination (Riley, 1977). These enlarged nerve endings were immunopositive for synaptotagmin; there was no staining in the preterminal axons (Figs. 3 B,C,4 B). During the early stages of pupal development (stages P0–P2-late), the retracted terminals began to lose their varicose appearance and were replaced by thin processes that grew in contact with the epidermis and migrating myoblasts. During this phase of nerve outgrowth, synaptotagmin immunoreactivity became widely distributed, not only within the high-order terminal processes, but also within the primary and secondary PrtFlx nerve branches and axons throughout the main leg nerve (Fig. 3 D,E). Although widely distributed within these processes, the staining was punctate rather than uniform. Between stages P2-late and P6, thick primary and secondary branches with thin high-order terminal processes grew over the developing PrtFlx muscle anlage (Fig. 4C1). Synaptotagmin immunoreactivity remained distributed in a punctate manner within all of these processes (Figs. 3 F,G; 4C2,D,E). During stages P6–P8, synaptotagmin immunoreactivity disappeared from primary nerve branches but remained distributed in secondary and high-order branches, especially at branch points and in the thin terminal processes that grew over the surface of the myotubes (Figs. 3 H, 4 F,arrows). During subsequent stages of muscle development (stages P8–P18), synaptotagmin immunoreactivity became gradually restricted to high-order terminal branches over the developing muscle fibers and finally to terminal varicosities over the mature adult muscle fibers (Figs. 3 I–L, 4 G,H).
Synaptic vesicle recycling during the remodeling of PrtFlx muscle innervation
The styryl dye FM1-43, which allows the direct study of synaptic vesicle exocytosis and recycling (Betz and Bewick, 1992, Betz et al., 1992), was used to correlate the distribution of synaptotagmin immunoreactivity with the functional maturation of the neuromuscular transmission during muscle remodeling. Motoneurons in dissected leg preparations from different developmental stages were electrically stimulated in the presence of 4 μm FM1-43 in normalManduca saline (4 mm Ca2+). In unwashed preparations, the dye caused staining of all membranes. After washing the preparations in Ca2+-free saline for 20–95 min, depending on the stage (Table 1), labeling remained only in the terminals of the stimulated PrtFlx motoneurons. Nonstimulated terminals over other muscles were devoid of staining after washing. No staining was observed in terminals that had been stimulated in the presence of FM1-43 in Ca2+-free saline or in washed terminals after they had been exposed to FM1-43 but not stimulated (data not shown). Terminals could be loaded after they had been exposed to FM1-43 in high-K+ saline (data not shown), but the staining was generally weaker than in terminals of motor axons that had been electrically stimulated. To ensure that FM1-43 labeled the presynaptic sites specifically, the following protocol was used for most of the developmental stages (L2, W4, P10, P14, and P18): (1) to load the PrtFlx motor terminals with the dye, nerve 2a (which contains the axons of PrtFlx motoneurons) was electrically stimulated in the presence of FM1-43 in normal saline; the preparation was then washed in Ca2+-free saline, and confocal images were taken with the minimum possible exposure to fluorescent light; (2) to unload the terminals, the same preparation was restimulated in the absence of FM1-43 in normal saline, washed in Ca2+-free saline, and imaged; (3) restimulation of the same terminals in the presence of FM1-43 in normal saline led to a second loading with the dye; and (4) after taking images from the reloaded terminals, the preparation was fixed and processed for anti-synaptotagmin immunostaining (Figs.5, 6 A,B,7 E–H). This protocol was modified, as described below, for experiments performed on stage P1–P8 animals (Figs.6 C,D, 7 A–D; Table 1) because both the developing nerves and muscle fibers were fragile and degraded rapidly. Where possible, the postsynaptic responses to nerve stimulation were also monitored by recording intracellularly from muscle fibers to confirm functional synaptic transmission.
Images from motor terminals over the intact larval and adult PrtFlx muscle (stages L2 and P18) that were loaded with the FM1-43 were identical to those taken after synaptotagmin immunolocalization, thus confirming that these varicosities are sites of synaptic vesicle release (Fig. 5). Staining was restricted to terminal varicosities. Synaptic vesicle recycling was still apparent after the degeneration of the larval PrtFlx muscle and was restricted to the retracted motor terminal varicosities that were synaptotagmin-immunopositive (Fig. 6 A,B; stage W3-late).
During myoblast production, migration, and accumulation at the site of the adult PrtFlx muscle formation (stages P0–P3), motor axons over the epidermis in the region where the PrtFlx anlage will form still became loaded with FM1-43 after nerve stimulation (Table 1, Fig.6 C,D; stages P1 and P2). In these early pupal stages the fragile nature of the preparations precluded our ability to demonstrate unloading of stained terminals or to fix and process FM1-43-loaded terminals for synaptotagmin immunoreactivity. However, no FM1-43 loading occurred when nerves were stimulated in Ca2+-free saline or in nonstimulated terminals after exposure to FM1-43 and washing in normal saline.
During stages P4 and P8 the ability to load and unload terminals with FM1-43 was demonstrated in one set of experiments (data not shown), and in another set of experiments FM1-43-loaded terminals were imaged and then fixed and processed for anti-synaptotagmin staining (Fig.7 A–D). Activity-dependent FM1-43 loading could readily be demonstrated at late stage P4 (Fig.7 A,B), during which enlarged secondary branches of PrtFlx motoneurons grew over the muscle anlage (Fig. 4 C1,C2). During this phase of development, as myoblasts continue to accumulate in the anlage, proliferate, and form myotubes, small excitatory junction potentials (EJPs) were successfully recorded in the central areas of the anlage, where the primary and secondary nerve branches were present, but were usually absent in the peripheral areas where high-order collaterals grew. The FM1-43-loaded presynaptic domains were localized to the primary and secondary axon branches and to a few high-order collaterals that were synaptotagmin-immunopositive (Fig.7 A,B). However, FM1-43 incorporation was absent from many regions of the nerve branches that were synaptotagmin-immunoreactive, including most of the high-order collaterals.
After myotube formation and separation was completed (stage P8), FM1-43 and synaptotagmin immunoreactivity disappeared from primary branches that were no longer in physical contact with the myotubes and became colocalized within secondary and high-order nerve branches that were growing along them (Fig. 7 C,D; also see Fig.4 F). The strongest FM1-43 and synaptotagmin staining was found at branching points.
For the remaining stages, FM1-43 loading and unloading and synaptotagmin distribution could again be examined in the same motor terminals. After the muscle fibers became striated, punctate FM1-43 staining was revealed in high-order branches and branch points where synaptotagmin immunoreactivity was colocalized (Fig.7 E,F). During subsequent stages of neuromuscular development, areas of synaptic vesicle recycling were gradually restricted to terminal varicosities that were also synaptotagmin-immunopositive (Figs. 5 E–H,7 G,H).
(1) Stages L2–W0. Functional motor terminals on the intact PrtFlx muscle fibers comprise rosettes of varicosities in which synaptotagmin and FM1-43 staining are strictly co-localized.
(2) Stages W0–W4. During larval muscle degeneration some distal motor branches disappear, whereas the remaining retract to a central region within the imaginal leg. Despite the absence of a target muscle, synaptotagmin immunoreactivity and FM1-43 loading are colocalized within enlarged larval motor terminals that remain intact.
(3) Stages W4–P2-late. Imaginal myoblasts initially become distributed within the imaginal leg and then migrate and accumulate close to the nerve terminals to form the adult PrtFlx muscle anlage, over which motor branches begin to expand. Synaptotagmin immunoreactivity and FM1-43 loading are widely distributed in a punctate manner within nerve branches.
(4) Stages P2-late–P4. As myoblasts accumulate, proliferate, fuse, and differentiate into myotubes, synaptotagmin immunoreactivity remains distributed within axons and terminal processes. FM1-43 loading is detectable in more restricted axonal regions.
(5) Stages P4–P8. As myoblast proliferation declines and myotube separation is completed, synaptotagmin immunoreactivity begins to disappear gradually from proximal parts of the motor axons. Synaptic vesicle recycling, as indicated by FM1-43, becomes progressively localized to more distal branches.
(6) Stages P8–adult. Muscle fibers become striated but continue to grow until the end of pupation. Synaptotagmin and FM1-43 staining become tightly colocalized to terminal varicosities.
Persistence of motor terminals in the absence of muscle
Synaptotagmin immunoreactivity and sites of FM1-43 uptake and release remain colocalized within varicosities that persist after the breakdown of the larval muscle. These enlarged varicosities may represent a coalescence of many smaller terminals as the muscle fibers shrink and nerve branches retract. Enlargement of varicosities may involve some of the same mechanisms that are associated with the remodeling of motor terminals during synapse elimination and muscle growth in vertebrates (Balice-Gordon and Lichtman, 1990; Colman and Lichtman, 1993). In amphibians, target-deprived nerve terminals with intact basal lamina and associated glia can persist in synaptic sites for up to 1 year (Yao, 1988) and remain functional, in terms of vesicular release, for up to 5 months (Dunaevsky and Connor, 1995). Whether motor terminal survival in Manduca is attributed to a muscle-derived factor that remains after the muscle death, as has been suggested for other preparations (Ko, 1984;Dunaevsky and Connor, 1995), remains to be investigated.
Presynaptic function during initial phases of adult muscle development
During the initial phase of adult muscle development (stages P0–P4; Fig. 8 C), the axons of the persistent larval motoneurons undergo growth in association with imaginal myoblasts that form the adult PrtFlx muscle anlage. Indeed, nerve and muscle interactions are essential for the development of the adult muscles during insect metamorphosis (Nüesch, 1985; Currie and Bate, 1995;Hegstrom and Truman, 1996; Bayline et al., 1998). In the developing adult legs of Manduca, both the accumulation of myoblasts into the correct sites of muscle formation and the appropriate level of proliferation are dependent on innervation (Luedeman and Levine, 1996;Consoulas and Levine, 1997).
Despite the lack of a functional target during early stages of adult development, Ca2+-dependent synaptic vesicle exocytosis is maintained. Functional contact with the new target could be demonstrated as soon as initial myotubes formed. Synaptotagmin is distributed along the axons of the motoneurons, suggesting that synaptic vesicles or their precursors are relocalized or being transported, as reported in mammals and Drosophila during embryogenesis (Kelly and Zacks, 1969; Kullberg et al., 1977; Lupa and Hall, 1989; Littleton et al., 1993; Yoshihara et al., 1997) or for neurons isolated in vitro (Matteoli et al., 1992; Kraszewski et al., 1995). Synaptic vesicle exocytosis and recycling, however, requires the presence of well-orchestrated action of several synaptic vesicle and plasma membrane proteins, in addition to synaptotagmin (Südhof, 1995). The ability of persistent axon branches to undergo FM1-43 loading suggests that a degree of functional specialization of the presynaptic machinery is maintained.
During embryonic development, neuronal growth cones are capable of neurotransmitter release before contact with the target (Hume et al., 1983; Young and Poo, 1983; Chow and Poo, 1985; Xie and Poo, 1986; Sun and Poo, 1987). Functional synaptic transmission can be detected shortly after the initial neuronal growth cone and myotube contact (Bennett and Pettigrew, 1974; Blackshaw and Warner, 1976; Kullberg et al., 1977; Dennis, 1981; Kidokoro and Yeh, 1982; Broadie and Bate, 1993a). This contact between the nerve and muscle cell membranes triggers an increase in neurotransmitter release that is driven by a rise in resting presynaptic Ca2+ concentration (Chow and Poo, 1985; Xie and Poo, 1986; Funte and Haydon, 1993; Zoran et al., 1993). Retrograde signal(s) from the target may be responsible (Connor and Smith, 1994). Similarly, in the present study, the maintenance of presynaptic function after muscle degeneration may reflect an autonomous ability of the persistent motor axons to express presynaptic specializations, or persistent cues from the larval target, combined with additional signals from the muscle precursors and the steroid hormone environment (see below).
Progressive restriction of synaptic vesicles and presynaptic function to mature synapses
During later stages of muscle development, before the establishment of final synaptic sites, synaptotagmin is distributed first within axons and then progressively to more distal processes. Calcium-dependent vesicle exocytosis and recycling is more restricted and first occurs in the shafts of motor axons rather than the higher-order branches that will later give rise to the mature presynaptic varicosities. These data are consistent with other observations that the machinery for vesicular cycling is present in developing axons before the formation of mature synapses. In cultured hippocampal neurons, for example, both Ca2+-dependent glutamate release and clustering and exocytosis of vesicles occur in the axon shafts of immature neurons before establishment of any synaptic contact (Kraszewski et al., 1995;Verderio et al., 1995). Similarly, in the Drosophila embryo, the initial formation of active zones can occur within motor axons in the absence of muscles (Prokop et al., 1996). This appearance of presynaptic specializations before the establishment of final synaptic sites may be analogous to the unlocalized distribution of glutamate or acetylcholine receptors on muscles before innervation (Broadie and Bate, 1993a,b; Hall and Sanes, 1993). Although both presynaptic and postsynaptic proteins may be expressed autonomously, the precise register of mature synaptic specializations probably requires cellular interactions in both directions (Prokop et al., 1996).
The differentiation of myotubes is an ongoing process; at any point in time different parts of the muscle anlage are in different states of development (Consoulas et al., 1997; this study). Initial presynaptic active sites in the axon shaft are transient and are gradually replaced by new sites in more distal processes as the muscle anlage grows and myotubes are formed. The correlation between the shifting location of functional sites of synaptic vesicle turnover and the progression of muscle differentiation within the enlarging anlage may reflect a retrograde cue that must be derived from myotubes once they reach the appropriate stage of development. The gradual shifting of presynaptic sites on maturing muscles represents an interesting contrast to embryonic neuromuscular junction formation in Drosophila, in which the neuronal growth cone stops as it reaches preformed myotubes and becomes transformed into a presynaptic terminal (Yoshihara et al., 1997).
Although the adult axonal branching pattern of the PrtFlx motoneurons has been established by stage P8, there is turnover of high-order branches, and the formation of new synaptic sites is continuous as muscle fibers elongate. A similar presynaptic remodeling has been observed during metamorphosis in amphibians. Postmetamorphic myogenesis and muscle fiber growth in frogs is accompanied by differential retraction, enlargement, creation, and elimination of junctional branches and synaptic sites (Sperry and Grobstein, 1983; Wernig and Herrera, 1986; Herrera and Werle, 1990; Herrera et al., 1990,1991). Addition of new branches and synaptic sites has also been observed during the growth of body wall muscles inDrosophila (Budnik et al., 1990; Gorczyca et al., 1993; Keshishian et al., 1993, 1996).
Control of motor terminal remodeling
The central dendrites and the peripheral processes of the persistent motoneurons undergo simultaneous phases of regression and re-expansion (Kent and Levine, 1993; Consoulas et al., 1996). Although the fine details and terminal stages of dendritic growth and the differentiation of motor terminals may be regulated by cellular interactions (Kent and Levine, 1993; Truman and Reiss, 1995), many aspects of this remodeling are under the control of the steroid hormone 20-hydroxyecdysone (Weeks, 1987; Truman and Reiss, 1988; Weeks and Ernst-Utzschneider, 1989; Prugh et al., 1992; Truman and Reiss, 1995). Thus, ecdysteroids may act directly on the cell body of motoneurons (Levine et al., 1986; Levine, 1989; Prugh et al., 1992) to regulate the synthesis of proteins involved in the formation and maintenance of presynaptic machinery. It is likely, however, that the precise alignment of presynaptic and postsynaptic specializations requires communication between neuron and muscle. This can readily be addressed through manipulations of the muscle precursor cells both in vivo (Consoulas and Levine, 1997) and in nerve and muscle cocultures (Luedeman and Levine, 1996).
This work was supported by Grant NS 24822 from the National Institutes of Health. C.C. was supported by Fogarty International Center Fellowship TWO 4898. We thank Maria Anezaki for assistance with histology. We also thank Dr. Mani Ramaswami for introducing us to the FM1-43 labeling technique.
Correspondence should be addressed to Dr. Christos Consoulas, Division of Neurobiology, Room 611, Gould Simpson Building, University of Arizona, Tucson, AZ 85721.