Journal of Neuroscience Methods 138 (2004) 51–56

A morphological technique for exploring neuromuscular topographyexpressed in the mouse gluteus maximus muscle

S.J. Lampaa,b, S. Potlurib, A.S. Nortonb, M.B. Laskowskia,b,∗a Program in Neuroscience, Veterinary Comparative Anatomy, Pharmacology and Physiology, Washington State University,

Pullman, WA 99164-6520, USAb WWAMI Medical Program, University of Idaho, Moscow, ID 83844-4207, USA

Received 14 January 2004; received in revised form 27 February 2004; accepted 4 March 2004

Abstract

Motor neuron pools innervate muscle fibers forming an ordered topographic map. In the gluteus maximus (GM) muscle, as well as additionalmuscles, we and others have demonstrated electrophysiologically that there exists a rostrocaudal distribution of axon terminals on the musclesurface. The role of muscle fiber type in determining this topography is unknown. A morphological approach was designed to investigate thisquestion directly. We combined three different methods in the same muscle preparation: (1) the uptake of activity-dependent dyes into selectedaxon terminals to define the spinal segmental origin of a peripheral nerve terminal; (2) the fluorescent labeling of nicotinic acetylcholinereceptors to determine motor endplate size; (3) the immunocytochemical staining of skeletal muscle to determine fiber subtype. We appliedthese methods to the mouse GM muscle to determine the relationship between muscle fiber type and the topographic map of the inferiorgluteal nerve (IGN). Results from this unique combination of techniques in the same preparation showed that axon terminals from more rostralspinal nerve segments of origin are larger on rostral muscle fibers expressing myosin heavy chain (MyHC) IIB epitope than caudal type IIBfibers. Because type IIB fibers dominate the GM, this suggests that for these rostral axons terminal size is independent of fiber type. How thisaxon terminal size is related to the topographic map is the next question to be answered.© 2004 Elsevier B.V. All rights reserved.

Keywords:Gluteus maximus; Neuromuscular junction; Axon terminals

1. Introduction

The neuromuscular junction has been a very useful modelto study both synaptic structure and function. Componentsof the neuromuscular junction include the presynapticaxon terminal, terminal Schwann cells, and the postsynap-tic acetylcholine receptors residing on the muscle fibers(Sanes and Lichtman, 1999). A variety of staining tech-niques have been used to provide a structural correlate ofneuromuscular transmission. Early techniques to identifyneuromuscular junction morphology relied on staining axonterminals with silver, or staining acetylcholinesterase tolocate motor endplates (Alderson et al., 1989; Karnovskyand Roots, 1964; Pestronk and Drachman, 1978). Later,lipophilic dyes were used to anterogradely label nerveterminals in vertebrate neuromuscular junctions. Two of

∗ Corresponding author. Tel.:+1-208-885-6696;fax: +1-208-885-7910.

E-mail address:mlaskow@uidaho.edu (M.B. Laskowski).

the more common lipophilic dyes are 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI)and 4-(4-diethylaminostyryl)-N-methylpyridinium iodide(4-Di-2-Asp) (Magrassi et al., 1987). Among the problemsusing these techniques were slow diffusion of the dyesalong axons and the inability to determine the relative ros-trocaudal origin of the axons of interest (Balice Gordon andLichtman, 1993; Lichtman et al., 1987; Marques and Neto,1998; Rich and Lichtman, 1989).

Betz et al. (1992)showed that a class of styryl dyes,whose original use was as membrane potential sensors,could label active vertebrate axon terminals. When neuro-muscular preparations are stimulated either electrically orwith high potassium, these styryl dyes bind to the activelyexocytosed cell membranes, and then during the processof endocytosis are taken up by active axon terminals (Betzet al., 1996; Cochilla et al., 1999; Reid et al., 1999).A toxin derived from snake venom, alpha bungarotoxin(�-bungarotoxin), binds specifically to the�-subunit ofthe pentameric nicotinic acetylcholine receptors (nAChRs).

0165-0270/$ – see front matter © 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.jneumeth.2004.03.012

52 S.J. Lampa et al. / Journal of Neuroscience Methods 138 (2004) 51–56

Alpha bungarotoxin now has been fluorescently conju-gated (Ravdin and Axelrod, 1977) and is a useful agentfor labeling and then visualizing nAChRs (Balice Gordonand Lichtman, 1993; Rich and Lichtman, 1989). Fur-ther studies have used both activity-dependent dyes andfluorescent-labeled�-bungarotoxin, to identify specific mo-tor endplates (Barry and Ribchester, 1995; Costanzo et al.,1999; Rafuse et al., 2000; Ribchester et al., 1994). Othershave looked at the link between the motor axon and themuscle fiber that it innervates using receptor labeling alongwith immunocytochemistry for both the presynaptic axonterminals and specific muscle fiber types (Prakash et al.,1996).

Spinal cord motor neuron pools project axons that formordered topographic maps of connectivity onto the surface ofskeletal muscles. Of key interest in our laboratory is the phe-nomenon of neuromuscular topography. We and others haveobserved a physiologic existence of this topographic mapin the diaphragm, serratus anterior (SA), and gluteus max-imus (GM) muscles (Brown and Booth, 1983; Feng et al.,2000; Laskowski and Sanes, 1987; Laskowski and High,1989; Laskowski and Owens, 1994). These electrophysio-logical studies have defined the topographic relationship be-tween position of spinal motor neurons within the pool andthe muscle fibers they innervate. A central question in thesestudies is how muscle fiber type influences the formation ofthis topographic map. Both motor axons and muscle fibersmay be necessary at the neuromuscular junction in shapingthis topography (Prakash et al., 1996). Moreover, differencesin nAChR areas have been correlated with topographic dif-ferences in the rat serratus anterior muscle (Potluri et al.,2002).

To further advance our understanding of the mecha-nisms underlying topography, a morphological method wasneeded that could correlate the spinal segmental origin ofthe motor neuron, the size of the axon terminals and themuscle fiber type innervated. Here we utilized a confocalmicroscope to explore the complex relationship existingat the neuromuscular junction that forms the basis of thistopographic map. First, we used an activity-dependent dyeto map specifically activated axon terminals from a givenspinal motor pool. Second, we labeled motor endplateswith rhodamine-conjugated�-bungarotoxin and mappedthe selectively stained terminals onto corresponding re-ceptors. Third, we identified muscle fiber type immuno-cytochemically, and superimposed the map of active axonterminals on these identified muscle fibers. Finally, wemeasured receptor size and plotted this as a function ofmuscle fiber type and spinal root of origin. Our resultsfor the first time provide quantitative morphological as-sessment of the neuromuscular topographic map. Withthis technique we are now able to identify specific ac-tive axon terminals and their corresponding muscle fibersubtypes in the same muscle. Preliminary results have ap-peared in abstract form (Potluri et al., 2002; Lampa et al.,2003).

2. Methods

2.1. Experimental animals

C57-Bl6mice (Jackson Laboratories) in the age range ofpostnatal days 17–21 (P17–P21) were used in this study. Thisage was selected because it is after the period of synapseelimination (Jansen and Fladby, 1990; Zoubine et al., 1996).Animals were given a mixture of ketamine and xylazine(K–X) administered at 0.1 cc/10 g intraperitoneally to euth-anize them. The animals were cared for and sacrificed inaccordance with animal use policies set forth by the Univer-sity of Idaho Animal Care and Use Committee.

2.2. Gluteus maximus muscle preparation

Using electrophysiological methods our lab previously re-ported in another muscle, the rat serratus anterior, that thereis a rostrocaudal distribution of axon terminals onto the sur-face of the muscle (Laskowski and High, 1989; Laskowskiand Sanes, 1987; Laskowski et al., 1998; Potluri et al.,2002). We confirmed this topography within the mousegluteus maximus muscle using the same method (Potluriet al., 2002). After euthanizing an experimental animal, thespinal cord along with the ventral roots of lumbar spinalnerves (L3 and L4) was exposed via a dorsal laminectomyof the lumbar vertebrae. The GM muscle was then isolated;muscle fibers from the GM have their origins on the iliaccrest, the sacroiliac joint, as well as the spinous processesof sacral vertebrae. From their origin the GM muscle fibersextend distally to insert onto the gluteal tuberosity of thefemur (Brown and Booth, 1983; English, 1990). Two sep-arate nerves branching from the sciatic nerve innervatethe GM, the superior and inferior gluteal nerve (SGN andIGN), respectively (Fig. 1). From earlier electrophysiologicstudies based on muscle fiber twitches and compound ac-tion potentials it was noted that the majority of the GM wassupplied by the IGN (English, 1990; Feng et al., 2000).

2.3. Axon terminal topography

Prior recordings of endplate potentials showed that theIGN consisted mainly of contributions from ventral rootsof L3 and L4 (data not shown). In a small Sylgard (DowCorning Corp.) dish the spinal cord, intact ventral roots,and the GM were pinned out and superfused with Ringer’ssolution containing (in mM): 144 NaCl, 4 KCl, 1 KH2PO4,1 MgCl2, 4 HEPES, 11 glucose, and 5 CaCl2. In eachexperiment the lumbar ventral root of either L3 or L4was taken up into suction electrodes. Only a single rootwas stimulated in each experiment. After equilibration,axon terminals were loaded with an activity-dependent dye(N-(3-triethylammoniumpropyl)-4-(4-dibutylamino)styryl)pyridinium dibromide (FM 1–43) (Molecular Probes). Thepreparation was stimulated in a solution with 8�M of dis-solved FM 1–43 in Ringer’s at 3 Hz, and at 10–12 V for

S.J. Lampa et al. / Journal of Neuroscience Methods 138 (2004) 51–56 53

Fig. 1. A schematic representation of the mouse gluteus maximus (GM)muscle and its major innervation by the inferior gluteal nerve (IGN) andminor innervation by the superior gluteal nerve (SGN). R and C indicatethe rostral and caudal extent of the origin of the GM. L3 and L4 dorsalroot ganglia are labeled to show the lumbar ventral root connections tothe sciatic nerve. Note drawing is not to scale.

10 min. This method has been shown to label active axon ter-minals (Betz et al., 1992). Following dye loading the prepa-ration was washed for 1 h in Ringer’s solution to remove theexcess FM 1–43. The rostral-most border of the GM wasarbitrarily set as the point where the SGN and IGN overlap,an anatomical landmark that is easily discernable under adissection microscope. The GM caudal to this point wasdivided into two equal sections that were named “rostral”and “caudal” accordingly. After axon terminals of L3 or L4were labeled with FM 1–43, 20�l rhodamine-conjugated�-bungarotoxin (Molecular Probes) at 1:350 was addedfor 40 min to label the nAChRs (Costanzo et al., 1999).Sections were mounted between coverslips and glass slidesusing a fluorescent mounting medium, Vectashield (Vec-tor Labs). Axon terminals that incorporated FM 1–43along with their corresponding receptors labeled withrhodamine-conjugated�-bungarotoxin were imaged withan inverted Nikon Diaphot 200 epi-fluorescent microscopeusing a Nikon Plan Apo 60X objective. Images werescanned and acquired using a BioRad 1024 MRC confocalsystem. In order to co-localize receptors and axon terminalsthe sections were excited with a krypton-argon laser usingboth the 488 and 568 nm wavelengths. Using photomulti-plier tubes 1 and 2 in the confocal system, the emissionswere visualized with 585 EFLP and 522 DF22 filters, re-spectively. Co-localized sequential images were taken at1.0�m through the motor endplate. Assembled images weresaved, and the location of the receptors was noted for later

re-mapping of receptors on specific myosin heavy chain fibertypes.

2.4. Muscle fiber typing protocol

A map was drawn of the endplates that had activity-labeledaxon terminals. After obtaining images of axons terminalsand receptors, the sections were fixed in 2% paraformalde-hyde for 20 min, and then placed in 0.1 M glycine in PBSfor 1 h. Finally, the sections were placed in a 4◦C refrig-erator overnight in a blocking solution that consisted of2% normal goat serum, 2% bovine serum albumin, and0.1% Triton X-100 in PBS. Next, a primary monoclonalantibody for myosin heavy chain subtype IIB (Schiaffinoand Reggiani, 1994; Schiaffino et al., 1989), HB 287,(Anti-MyHC-IIb IgM) (gifts from Joshua Sanes, Washing-ton University) was added for at least 12 h, followed bya secondary antibody goat anti-mouse IgM Alexa Fluor488 (Molecular Probes, Eugene) for 8–10 h. The steps forthe primary and secondary antibodies were carried out at4◦C. The same sections were re-mounted, and using previ-ously recorded landmarks, those same receptors that wereidentified as being co-localized with L3 or L4 active axonterminals were relocated and re-imaged. The receptors andmuscle fibers were imaged using the 568 nm wavelengthfor the rhodamine-labeled receptors and 488 nm for theAlexa 488-labeled muscle fiber subtypes. Images weresaved and projected using Confocal Assistant 2.55 (ToddClark Brelje). Only receptors that were both co-localizedwith FM 1–43 from earlier acquired images and were onpositive immunoreactive muscle fibers after immunocyto-chemistry were used for calculation of receptor areas onmuscle fiber subtype. Receptor areas were measured fromthresholded images, and the calibration value was deter-mined by measuring a known distance using a micrometer.The final step required Metamorph 5.0 software (UniversalImaging Corp.) to calculate the average endplate areas inboth the rostral and caudal portions of the GM. Statisticalanalysis used a two-way analysis of variance to comparethe relative differences of the separate muscle sections, aswell as the difference in the source of innervation betweenventral roots and within the same ventral root of origin.

3. Results

In a single experiment individual axon terminals de-rived from the ventral root of either L3 or L4 were firstvisualized and imaged using the activity-dependent dye,FM 1–43 (Fig. 2A). Application of rhodamine-conjugated�-bungarotoxin labeled corresponding nAChRs on theclosely apposed muscle fibers (Fig. 2B). The co-localizationof receptors and axon terminals allowed for identificationof specific activated terminals. Images were merged andprojected to see the co-localization of active axon terminalsand receptors (Fig. 2C). This mapping of FM 1–43 stained

54 S.J. Lampa et al. / Journal of Neuroscience Methods 138 (2004) 51–56

Fig. 2. Images showing active axon terminals (AT) and postsynaptic nicotinic acetylcholine receptors (nAChR) on the gluteus maximus (GM). A illustratesnerve terminals filled with FM 1–43 (using fluorescein optics), B shows endplates stained with rhodamine-conjugated�-bungarotoxin (under rhodamineoptics). C is a merged composite image showing a co-localization of both AT and nAChR. Scale bar: 20�m.

terminals with�-bungarotoxin stained endplates, was nec-essary because axon terminal staining is lost during theprocess of immunocytochemistry. The digital co-localizedimages taken from 12 animals were captured and storedfrom the rostral and caudal portions of the GM and werelater used for morphometric analysis of motor endplates onmuscle fiber subtype.

An earlier study showed that the rat gluteus muscle con-tained about 85% fast muscle fibers (Armstrong and Phelps,1984). We confirmed this observation. Rostral and caudal20�m serial cross sections through the mouse GM showeda similar distribution of fibers labeled for MyHC type IIstaining (Potluri et al., 2002). Here we asked which subtypeof fast fibers dominated the GM. Cross sections through therostral and caudal portions were used to test whether there

Fig. 3. Re-localization of identified endplates on type IIB muscle fibers. A shows the merged images with FM 1–43 and rhodamine-conjugated-bungarotoxin.B shows MyHC type IIB labeling with Alexa 483 on the earlier localized receptors. Scale bar: 20�m.

was a larger proportion of MyHC IIB than IIA fiber sub-types. These results showed that overall the GM muscle was70% (±2.6%) type IIB and 13% (±1.7%) type IIA. Basedon these cross sectional studies, muscle fiber type immuno-histochemistry was performed on whole mounts using thetype IIB subtype antibody (HB 277). Because the dominantfiber type in the GM is type IIB we focused on this specificsubtype for the measurement of receptor areas. We askedwhether endplate areas on type IIB fibers differed in ros-tral versus caudal halves of the muscle, and did they dif-fer based on innervation by either L3 or L4. To comparethe active terminals from L3 or L4 and the co-localized re-ceptors, images were obtained after performing immuno-cytochemistry of both muscle fibers and receptors.Fig. 3shows a merged example of those receptors some of which

S.J. Lampa et al. / Journal of Neuroscience Methods 138 (2004) 51–56 55

Fig. 4. Receptor areas on active axon terminals from either L3 or L4 onMyHC type IIB fibers. Values are mean± S.E.M; (∗) indicates signifi-cant difference (P < 0.05). Data show that receptor areas innervated byterminals from L3 are significantly larger on rostral muscle fibers thanon caudal muscle fibers.

are on MyHC IIB subtype muscle fibers. Receptors wereimaged at threshold, and a total of 60 receptors were mea-sured from rostral and caudal sections of the GM muscle forboth L3 and L4 axon terminals using the Metamorph soft-ware. These results showed that endplates innervated by L3axon terminals showed a significant difference from rostralto caudal, 199.47 ± 4.97�m2 versus 185.64 ± 4.33�m2,respectively. No rostrocaudal difference was observed in ar-eas of L4 receptors, 190.16±6.00�m2 for rostral endplatesand 190.52 ± 4.41�m2 for caudal endplates (Fig. 4). Theresults suggest that with this technique motor endplate sizecan be used as a morphological tool to investigate the basisof neuromuscular topography.

4. Discussion

The technique we describe in this report was designed toinvestigate the basis for topography by comparing musclefiber type, spinal nerve origin and one parameter of topo-graphic difference, the motor endplate area. We did thisby exploring the morphology of the motor endplates usingactivity-dependent dyes and immunocytochemistry. Thismethod is novel in its ability to identify for the first time thesegmental origin of axon terminals and their correspondingnAChR receptor areas with the specific muscle fiber typein the same preparation. This combined method is ideal forasking morphological questions regarding the mechanismsfor topographic map formation. In this study we askedspecifically whether fiber type distribution could explain thepreviously observed neuromuscular topography. With thisapproach we showed that both spinal nerves L3 and L4 sup-ply the dominant fiber type of the GM muscle, type IIB, butdo so differently. Receptors supplied by axon terminals of L3motor neurons, the more rostral of the two IGN nerve roots,are larger on rostral than caudal IIB fibers. No difference

in receptor area could be detected among those receptorsinnervated by L4 terminals. L4 is the minor contribution tothe GM and therefore a differential distribution in endplatesize may be of lesser importance biologically. The diameterof IIB GM fibers is the same both rostrally and caudally (un-published data). Thus for the projecting L3 motor neurons,their terminals are larger when situated on type IIB fibers intheir topographically matched target area. Differences in re-ceptor areas have been correlated with topography in anothermuscle the rat SA (Potluri et al., 2002). In that study rostralaxon terminals were larger in the rostral territories than inthe caudal territories of the muscle. Conversely, caudal ax-ons terminals were larger on the caudal portion of the SAmuscle. If other muscles show a similar correlation betweenthe endplate size and segmental origin of spinal nerve, it mayindicate that positionally matched motor neurons have largermotor endplates. One possible explanation is that theselarger terminals may release more neurotransmitter and thushave a greater influence in their appropriate topographicallymatched target area. This would be especially important ontype II muscle fibers where tetanic activity of motor neuronsproduce endplate potentials with reduced quantal content(Reid et al., 2003). Although the difference in nerve terminalsize is small, a greater release of transmitter could provide agreater safety factor for transmission during bursts of nerveaction potentials (Wood and Slater, 2001). The combina-tion of techniques reported here will enable more detailedstudies of the structural and functional basis for neuromus-cular topography. Particularly important will be the use ofthis combined technique in exploring the mechanism ofselective reinnervation (Laskowski and Sanes, 1988).

Acknowledgements

This research was supported by NIH grant NS27024(MBL) and P20RR16454 from the BRIN Program of theNational Center for Research Resources (MBL).

References

Alderson K, Pestronk A, Yee WC, Drachman DB. Silver cholinesteraseimmunocytochemistry: a new neuromuscular junction stain. MuscleNerve 1989;12:9–14.

Armstrong RB, Phelps RO. Muscle fiber type composition of the rathindlimb. Am J Anat 1984;171:259–72.

Balice Gordon RJ, Lichtman JW. In vivo observations of pre- and postsy-naptic changes during the transition from multiple to single innervationat developing neuromuscular junctions. J Neurosci 1993;13:834–55.

Barry JA, Ribchester RR. Persistent polyneuronal innervation in partiallydenervated rat muscle after reinnervation and recovery from prolongednerve conduction block. J Neurosci 1995;15:6327–39.

Betz WJ, Mao F, Bewick GS. Activity-dependent fluorescent stain-ing and destaining of living vertebrate nerve terminals. J Neurosci1992;12:363–75.

Betz WJ, Mao F, Smith CB. Imaging exocytosis and endocytosis. CurrOpin Neurobiol 1996;6:365–71.

56 S.J. Lampa et al. / Journal of Neuroscience Methods 138 (2004) 51–56

Brown MC, Booth CM. Postnatal development of the adult pattern ofmotor axon distribution in rat muscles. Nature 1983;304:741–2.

Cochilla AJ, Angleson JK, Betz WJ. Monitoring secretory membrane withFM1–43 fluorescence. Annu Rev Neurosci 1999;22:1–10.

Costanzo EM, Barry JA, Ribchester RR. Co-regulation of synaptic effi-cacy at stable polyneuronally innervated neuromuscular junctions inreinnervated rat muscle. J Physiol 1999;521:365–74.

English AW. Development of compartmentalized innervation of the ratgluteus maximus muscle. J Comp Neurol 1990;301:104–13.

Feng G, Laskowski MB, Feldheim DA, Wang H, Friesen J, FlanaganJG, et al. Roles for ephrins in positionally selective synaptogenesisbetween motor neurons and muscle fibers. Neuron 2000;25:295–306.

Jansen JKS, Fladby T. The perinatal reorganization of the innervation ofskeletal muscle in mammals. Prog Neurobiol 1990;34:39–90.

Karnovsky MJ, Roots LA. Direct coloring method for cholinesterase. JHistochem Cytochem 1964;12:219–21.

Lampa SJ, Potluri S, Norton AS, Laskowski MB. A novel technique forexploring neuromuscular topography in skeletal muscles. Program no.40.10. Abstract viewer/itinerary planner. Washington, DC: Society forNeuroscience, 2003.

Laskowski MB, Sanes JR. Topographic mapping of motor pools ontoskeletal muscles. J Neurosci 1987;7:252–60.

Laskowski MB, Sanes JR. Topographically selective reinnervation of adultmammalian skeletal muscles. J Neurosci 1988;8:3094–9.

Laskowski MB, High JA. Expression of nerve-muscle topography duringdevelopment. J Neurosci 1989;9:175–82.

Laskowski MB, Owens JL. Embryonic expression of motor neuron to-pography in the rat diaphragm muscle. Dev Biol 1994;166:502–8.

Laskowski MB, Colman H, Nelson C, Lichtman JW. Synaptic compe-tition during the reformation of a neuromuscular map. J Neurosci1998;18:7328–35.

Lichtman JW, Magrassi L, Purves D. Visualization of neuromuscularjunctions over periods of several months in living mice. J Neurosci1987;7:1215–22.

Magrassi L, Purves D, Lichtman JW. Fluorescent probes that stain livingnerve terminals. J Neurosci 1987;7:1207–14.

Marques MJ, Neto H. Imaging neuromuscular junctions by confocalfluorescence microscopy: individual endplates seen whole muscle withvital intracellular staining of the nerve terminals. J Anat 1998;192:425–30.

Pestronk A, Drachman DB. A new stain for quantitative measurement ofsprouting at neuromuscular junctions. Muscle Nerve 1978;1:70–4.

Potluri S, Lampa SJ, Norton AS, Laskowski MB. A morphological anal-ysis of nerve muscle topography in skeletal muscles. Program no.28.3. Abstract viewer/itinerary planner. Washington, DC: Society forNeuroscience, 2002. Online.

Prakash YS, Miller SM, Huang M, Sieck GC. Morphology of di-aphragm neuromuscular junctions on different fiber types. J Neurocy-tol 1996;25:88–100.

Rafuse VF, Polo-Parada L, Landmesser LT. Structural and functional al-terations of neuromuscular junction NCAM-deficient mice. J Neurosci2000;20:6529–39.

Ravdin P, Axelrod D. Fluorescent tetramethyl rhodamine derivatives ofalpha-bungarotoxin: preparation, separation, and characterization. AnalBiochem 1977;80:585–92.

Reid B, Martinov VN, Njå A, Lømo T, Bewick GS. Activity-dependentplasticity of transmitter release from nerve terminals in rat fast andslow muscles. J Neurosci 2003;23:9340–8.

Reid B, Slater CR, Bewick GS. Synaptic vesicle dynamics in rat fast andslow motor nerve terminals. J Neurosci 1999;19:2511–21.

Ribchester RR, Mao F, Betz WJ. Optical measurements of activitydependent membrane recycling in motor nerve terminals of mam-malian skeletal muscle. Proc R Soc Lond B: Biol Sci 1994;255:61–6.

Rich MM, Lichtman JW. In vivo visualization of pre- and postsynap-tic changes during synapse elimination innervated mouse muscle. JNeurosci 1989;9:1781–805.

Sanes JR, Lichtman JW, Development of the vertebrate neuromuscularjunction. Annu. Rev. Neurosci. 1999;389–442.

Schiaffino S, Gorza L, Sartore S, Saggin L, Ausoni S, Vianello M, et al.Three myosin heavy chain isoforms in type 2 skeletal muscle fibers.J Muscle Res Cell Motil 1989;10:197–205.

Schiaffino S, Reggiani C. Myosin isoforms in mammalian skeletal muscle.J Appl Physiol 1994;77:493–501.

Wood SJ, Slater CR. Safety factor at the neuromuscular junction. Prog.Neurobiol. 2001;64:393–429.

Zoubine MN, Ma JY, Smirnova IV, Citron BA, Festoff BW. A molecularmechanism for synapse elimination: novel inhibition of locally gen-erated thrombin delays synapse loss in neonatal mouse muscle. DevBiol 1996;179:447–57.