Muscle Agrin: Neural Regulation and Localization at Nerve-Induced Acetylcholine Receptor Clusters

The Journal of Neurosciencs. June 1993. U(6): 2509-2514

Muscle Agrin: Neural Regulation and Localization at Nerve-induced Acetylcholine Receptor Clusters

Erich Lieth and Justin I?. Fallon Neurobiology Group, Worcester Foundation for Experimental Biology, Shrewsbury, Massachusetts 01545

The synapse-organizing protein agrin is expressed by mus- pivotal role in orchestrating the formation of neuromuscular cle cells. In this study we begin to characterize the role of junctions, but the nature of that role remains to be elucidated. muscle agrin in synapse formation by investigating its dis- Both neural and muscle cells can synthesize agrin (Magill- tribution and expression during the formation of nerve-in- Sole and McMahan, 1988; Lieth et al., 1992) but it is not clear duced ACh receptor (AChR) aggregates in vitro. We have how agrin from these cellular sources may contribute to the used species-specific anti-agrin antibodies to show that formation of neuromuscular synapses.A simple model might muscle agrin colocalizes with nerve-induced AChR clusters state that agrin released from nerve endings causessynaptic in chimeric nerve-muscle cocultures. Furthermore, quanti- elements to redistribute on the postsynaptic cell (McMahan, tation by radioimmune assay shows that the expression of 1990). Another possibility is that axons induce the releaseof mu&e agrin by cultured chick myotubes is increased more agrin and other related molcculcs from the muscle cells they than twofold by coculture with cells from the motoneuron- contact, and that such molecules are involved in either the re- rich ventral spinal cord. These data suggest a role for muscle distribution or stabilization of synaptic elements(Lieth et al., agrin in neuromuscular junction development. 1992). Also, agrin from muscleand nerve may act synergistically [Key words: agrtn, neuromuscular junction, synaptogene- in the formation and maintenance of the neuromuscularjunc- sis, extracellular matrix, basal lamina, nerve-muscle cocul- tion. We have previously reported that muscle cells in culture ture, ACh receptor] synthesizeand releaseagrin (Lieth et al., 1992). Moreover, these molecules are localized at AChR aggregatesinduced by agrin The organization of specificmolecules at the postsynaptic mem- solubilized from Torpedo electric organ extracellular matrix. In brane is one of the crucial processesin neuromuscularjunction the present study, we sought to charactcrizc further the expres- formation. Agrin, a protein present in the synaptic basallamina sion of muscleagrin at the neuromuscularjunction to learn how of speciesranging from fish (Fallon et al., 1985) to birds and this molecule may be involved in the formation of this synapse. mammals (Reist et al., 1987; Rupp et al., 1991), induces the Neuron-myotube cocultures provide a powerful and acccs- redistribution of a number of synaptic molecules,such as ACh siblc system to study the role of agrin in synaptogenesis.Mo- receptors(AChRs), into structures resemblingpostsynaptic spe- toneuronsinduce the clustering of AChR on cultured myotubes, cializations on myotubes in virro (Nitkin et al., 1987; Wallace, forming synapsesthat functionally and structurally resemble 1989; Nastuk and Fallon, 1993). neuromuscularjunctions in vivo (Anderson and Cohen, 1977; Several lines of evidence support the hypothesis that agrin Frank and Fischbach, 1979; Cohen and Weldon, 1980; Role et directs the formation of neuromuscularjunctions. For instance, al., 1987). Other ncuronal types, such as cells of the dorsal root anti-agrin antibodies label the synaptic basallamina (Fallon et ganglia, lack this property. We cultured chick myotubes in the al., 1985; Reist et al., 1987) a structure that in the absenceof presenceof motoneurons from rat spinal cord, and usedspecies- nerve can promote the formation ofpostsynaptic specializations specific antibodies that recognize chick agrin to visualize agrin on the surfacesof regenerating myofibers (Burden et al., 1979). derived from muscle. We show that muscle agrin colocalizes Agrin immunoreactivity is present at the very earliest times that with nerve-induced AChR clusters. Furthermore, coculture with neuromuscular synapsescan be identified in vivo (Godfrey et spinal cord cells stimulates agrin expressionby myotubes. We al., 1988; Fallon and Gelfman, 1989) and anti-agrin antisera postulate that agrin derived from musclecells contributes to the can block the formation of nerve-induced AChR clusters on formation of stable postsynaptic specializations at the neuro- myotubes in culture (Reist et al., 1992). Also, recombinant agrin muscular junction. These results demonstrate the existence of has been shown to induce the aggregation of muscle AChR complex interactions between neurons and muscle cells that (Campanelli et al., 1991; Tsim et al., 1992). Taken together, regulate the expressionand spatial distribution of synaptic or- these observations provide strong evidence that agrin plays a ganizing molecules during neuromuscular junction development. The simple model of AChR clustering by neuronally re- leasedagrin is not sufficient to explain thesefindings. It seems Received Aug. 18, 1992; revised Nov. 19. 1992; accepted Dec. 17, 1992. more likely that both muscle and neuronal cells contribute ac- We are grateful to B. McKechnie for expert technical assistance. We also thank Dr. K. Buckley for a gift of antibody. This work was supported by a postdoctoral tively to the formation and maintenance of junctions by se- fellowship from the Muscular Dystrophy Association (JZL.), NIH HD23924, and creting synapseorganizing molecules. a Basil O’Connor Starter Scholar Research Award from the March of Dimes (J.R.F.). Correspondence should be addressed to Erich Lieth, Department of Neuro- Materials and Methods biology, Harvard Medical School, 220 Longwood Avenue, Boston, MA 02 I IS. Cell cuhre. Embryonic day 12 (E12) chick pectoral muscle was dis- Copyright @ 1993 Society for Neuroscience 0270-6474/93/132509-06$05.00/O sected,dissociated, and plated as describedpreviously (Lieth et al., Figure 1. Coculture with rat ventral spinal cord neurons induces the patching of AChRs on the surfaces of chick myotubes. Four days after the addition of rat ventral spinal cord cells to chick muscle cell cultures, specimens were fixed and stained with rh-BTX and antibodies to the synaptic vesicle protein SV2. The same field is shown viewed in fluorescein and rhodamine optics to visualize neurites (a) and AChR clusters (b), respectively. Over 90% of AChR clusters were within 5 pm of a neurite. Scale bar, 20 pm.

1992). One day later, El7 rat spinal cords were dissected and divided any dimension) per 370 pm myotube segment in each of 10 randomly along the lateral sulcus into ventral and dorsal strips. Tissues were selected microscope fields was determined under rhodamine optics. The incubated in 0.25% trypsin for 30 min at 37”C, mechanically dissociated, colocalization of agrin immunoreactivity at these clusters was then de- and transferred into a plating medium consisting of Dulbecco’s modified termined by switching to the fluorescein-selective optics. The total num- Eagle’s medium supplemented with 10% fetal bovine serum (FBS), pen- ber of AChR clusters and the proportion that colocalized with agrin icillin, and glutamine. In order to maintain low baseline levels of muscle were determined for at least two coverslips for each day in culture. agrin expression, chick embryo extract was not used in these experi- To compare the growth of neurites in cocultures of myotubes with ‘merits. Cells were plated on preformed layers of chick myogenic cells either spinal cord or cortical cells, the number of myotube segments at 100,000 cells/well in 96-well plates for radioimmune assays or 50,000 that exhibited zero, one, two, three, or four or more contacts with SV2- cells/well on glass coverslips for immunocytochemistry. After 3 d, cul- labeled neurites were counted. For each condition, data gathered over tures were fed by replacement with the same media. E 17 cerebral cortex the entire course of the experiment were pooled to identify differences cultures were prepared exactly as described above. All animals were between the culture conditions. housed and handled in accordance with NIH guidelines. Solid-phaseradioimmune assay. Radioimmune assays to measure Antibodies.The anti-agrin antibodies used h&e have been character- muscle agrin expression were carried out as described (Lieth et al., 1992). ized extensively. Monoclonal antibodies (mAbs) 5Bl and 1 ID2 were Briefly, protein A-purified anti-agrin mAbs were labeled to 1.0-I .5 x raised against -Torpedoagrin and cross-react with chick agrin. Their 10’ cpmlpg IgG with lzsI. The amount of agrin was measured by in- binding to agrin has been established using tissue sections, solubilized cubation with l&ml ‘2SI-anti-agrin. Nonspecific binding was deter- extracts, cell culture, immunoblots, and recombinant protein (Reist et mined by pre- and coincubation with loo-fold excess unlabeled anti- al., 1987; Lieth et al., 1992; Ruegg et al., 1992). Both antibodies have body. Each measurement is the average of six simultaneous been reported to be unreactive with rat agrin as judged by immunola- determinations, adjusted for amount of protein in the individual cul- beling of tissue sections (Reist et al., 1987). We confirmed this lack of tures. cross-reactivity in rat spinal cord cultured alone or togcthcr with primary rat muscle cells. In no case was any specific labeling observed. As these Results cultures were carried out in the presence of FBS, the lack of staining by mAbs 5B 1 and 11 D2 also demonstrates either that FBS is free of agrin Rat ventral spinal cord neurons induce AChR clusterson chick or that these antibodies do not recognize serum-borne bovine agrin. myotubes Thus, these antibodies are specific for agrin and suitable for the selective The aggregation ofAChRs on myotube surfacesby motoneurons detection of chick agrin in rat/chick chimeric cultures. in culture is commonly usedas an indicator for the formation Zmmunocytochemisty.Immunolabeling was performed exactly as described by Lieth et al., (1992). Briefly, cultures grown on glass coverslips of nerve-muscle contacts (Anderson et al., 1977; Frank and were incubated in 5Bl hybridoma supematant (approximately 20 llg! Fischbach, 1979; Role et al., 1987). To induce the formation of ml specific IgG) followed by a fluorescein-conjugated goat anti-mouse postsynaptic specializations on chick myotubes, we added dis- Ig. AChRs were labeled using 5 x lo-’ M rhodamine-conjugated cu-bun- sociated rat spinal cord cells. Figure 1, a and b, showsthe colo- garotoxin (rh-BTX; Molecular Probes, Eugene, OR) in the second layer. Cells were then fixed in -20°C MeOH, mounted in Citifluor (Pella), calization of AChR clusters with neurites. We quantified this and examined through a Zeiss Axioplan microscope equipped with epi- induction by measuring the number of AChR aggregatesin co- fluorescence optics. We obtained results indistinguishable from those cultures with the motoneuron-rich ventral spinal cord as com- reported here using another anti-agrin antibody, 1 lD2, whose species pared to cocultures with the motoneuron-poor cortex. Through- cross-reactivity is the same as that of mAb 5Bl but that recognizes a different epitope. In some cultures, neurites were labeled with the anti- out the culture period, a greater number of AChR clusters is synaptic vesicle antibody SV2 (Buckley et al., 1983). These cultures observed in cocultures with ventral cord neurons(Fig. 2). At 4 were-first labeled with rd-BTX as des&bed above, fixed for 5 min in and 5 d in coculture, ventral spinal cord induced three to five 1% HCHO. and oermeabilized with 0.05% saoonin in PBS. The cells times more AChR clusters as compared to cortex. The total were then labeled with anti-SV2 followed by flboresceinated secondary number of clusters observed in ventral spinal cord cocultures antibody as above. Quantitationof immunocytochemicalstaining.To determine the colo- was up to seven times greater than in muscle cultured alone calization of muscle agrin with AChR clusters, cultures were double (data not shown). We also quantitated the number of contacts labeled as described above. The number of AChR clusters (~2 pm in between neuritesand myotubes in eachculture condition. Table The Journal of Neuroscience, June 1993, 13(6) 2511

Table 1. The number of neurite-myotube contacts is similar in myotubes cocultured with either cerebral cortical or spinal cord cells (see Materials and Methods) W Coriexcoculture Numberof neurite Number of myotube segments 2 contacts Ventral spinal cord Cortex 0 4 11 1 23 12 2 20 24 3 10 12 24 21 21

L1 Four or more contacts per myotube segment.

1 lD2 or 5Bl. Both antibodies have previously been reported 2 3 4 5 to bind chick, but not rat agrin (Reist et al., 1987). We confirmed Days in coculture this lack of cross-reactivity by labeling rat spinal cord cultured alone or together with primary rat muscle cells. In no casewas Figure 2. Ventral spinal cord neurons induce a greater number of any specific labeling observed (not shown). Thus, in these ex- AChR aggregates on myotubes than do cerebral corticalneurons. Neural periments only agrin secretedby muscle is visualized. Figure 3 cells were plated onto l-d-old muscle cultures and the preparations shows the colocalization of muscle agrin with nerve-induced stained 2-5 d later with rh-BTX. The number of AChR clusters per 370 pm myotube segment was counted as described in Materials and Meth- AChR clusters. Muscle agrin is concentrated at the large,mature ods.‘Thenumber of nerve-inducedAChR clusters was determined by AChR clusters. Note that the AChR microclusters(I 2 pm) are subtracting the number of AChR clusters observed in same-age cultures not colocalized with muscleagrin. The accumulation of muscle grown without nerve (spontaneousclusters). Each determination is the agrin at the more developed AChR clusters is consistent with meanof threecultures (?SEM). the hypothesis that it plays a role in the maturation or stabilization of thesepostsynaptic specializations (seeDiscussion). 1 showsthat there were equivalent numbers of neurite contacts To quantify the appearanceof muscle agrin at nerve-induced in myotubes cultured with either spinal cord or cerebral cortical AChR clusters, we compared the accumulation of agrin im- cells. The culture conditions usedhere thus provide comparable munoreactivity at AChR clusters in cocultures containing mo- support for the growth of both neuronal types. Therefore, these toneurons (ventral cord) to control cocultures without moto- resultsshow that rat motoneuronsselectively induce AChR clus- neurons(cerebral cortex). The number of agrin-colocalizedAChR ters on chick myotubes. clusters increasessteadily over time in coculture with ventral cord, but remainsrelatively unaltered in control cocultures (Fig. Muscle agrin accumulates at nerve-induced AChR clusters 4). By 5 d of culture up to 1O-fold more agrin-colocalized clusters We previously showed that myogenic cells in culture secrete are observed in ventral cord cocultures. The proportion of all agrin-related molecules(Lieth et al., 1992). Here we studied the AChR clustersthat colocalize with muscle agrin is consistently distribution of muscle agrin with respect to motoneuron-in- high in ventral cord cocultures, particularly after 3 d in vitro duced AChR aggregates.Chick myotubes and rat spinal cord (Fig. 4, numbers in bars). In cocultures with control neural cells neuronswere cocultured and then stainedwith anti-agrin mAb this proportion varied between culture platings, similar to the

Figure 3. Muscle agrin colocalizes with nerve-induced AChR aggregates. Chick muscle cells were cultured with rat ventral spinal cord neurons for 3 d, and then double labeled with anti-agrin antibody 5B 1, which does not recognize rat agrin (a) and rh-BTX (b). Muscle agrin is concentrated at 87% (see Fig. 4) of the large, mature AChR clusters (arrows). Note that the AChR microclusters (52 pm, arrowheads) are not colocahzed with muscle agrin. Scale bar, 10 pm. 2512 Lieth and Falbn l Muscle Agrin and Synapse Formation

n Ventral SC. ccculture - T q Cortexccculture

+ Vent. Spiel Cord + cerebral Cortex

Figure 5. Coculture with ventral spinal cord stimulates the expression of agrin by muscle cells. Chick myotubes were cultured either alone, with rat ventral spinal cord cells, or with cerebral cortex cells for 5 d. 2 3 4 5 Agrin was measured by radioimmunoassay using mAb 5BI, which in Days in coculture these cultures recognizes only agrin expressed by muscle. Each dcter- mination represents the mean of six cultures normalized for total protein/culture (?SD). Ventral spinal cord culture increased the amount Figure 4. At all days in vitro ventral spinal cord cells induce a greater of muscle agrin by twofold as compared to coculture with cerebral cortex number ofagrin-colocalizing AChR clusters than do cortical cells. Chick or myotubesalone. Ventral spinal cord versus myotube alone,p < 0.00 I; myotubes were cultured for 2-5 d in the presence of rat ventral spinal ventral spinal cord versus cortical coculture, p i 0.0 I (Student’s f test). cord or cerebral cortex cells. Preparations were stained with rh-BTX and mAb 5B I, and the number of AChR patches colocalizcd with muscle agrin counted. Each determination is the mean (?SEM), adjusted for the number of agrin-colocalizing spontaneous clusters. Numbers Gelfman, 1989; Lieth et al., 1989, 1992; Ruegg et al., 1992). inside bars are the percentages of all AChR clusters in the cultures that However, in none of these studies could it be determined if were also colocalized with muscle agrin. muscle agrin is expressed at the neuromuscular junction. We therefore designed a chimeric nerve-muscle coculture system variable colocalization with spontaneous AChR clusters ob- to addressthis question. Dissociatedrat ventral spinal cord cells served in cultures of myotubes alone (Lieth et al., 1992). In induce robust AChR clustering on chick myotubes, but cerebral cocultures containing motoneurons, the percentageof colocal- cortical cells (Figs. I, 2), dorsal spinal cord, or sensoryganglia ized AChR clusters was always at least 70% by 3 d in vitro. (E. Lieth and J. R. Fallon, unpublished observations) do not. Thus, agrin derived from muscle accumulates at a high pro- This clustering is similar to that seen when chick spinal or portion of AChR clusters induced by neurons of the ventral parasympathetic motoneurons are cocultured with myotubes spinal cord. (Frank and Fischbach, 1979; Role et al., 1987). In the present study. greater than 90% of clusters in ncrvc-muscle coculturcs Spinal cord neurons stimulate the e.~pression of muscle agrin wcrc associatedwith ncurites (Fig. I), indicating that neurons The colocalization of muscle agrin with nerve-induced AChR induced AChR clusters in our cultures. As shown in Figure 3, aggregatesopens the possibility that muscle responds to the muscle agrin is localized at the nerve-induced AChR clusters. presenceof motoneuronsby secretingincreased amounts of agrin. In a recent study using a chimeric frog system, Cohen and We therefore measuredthe effect of spinal cord coculture on Godfrey (I 992) did not dctcct the expressionof muscle agrin at the accumulation of muscleagrin. Muscle agrin levels were de- ncurite contacts. It is not known whether the failure to detect termined by radioimmunoassay. &culture with rat ventral spi- muscleagrin in this caseis due to insufficiently sensitive detec- nal cord increasedthe expressionof muscleagrin by more than tion methods or the lack of muscle agrin expressionby these twofold as compared to myotubcs cultured alone (Fig. 5). In amphibian cells. Significantly, muscle agrin was not observed contrast, only a small increasewas observed after coculture with under any circumstancesin the frog experiments. It is important cerebralcortical cells.The overall protein levels in thesecultures to determine whether this contrasting result is due to technical was not affected by coculture with neural cells (data not shown). or speciesdifferences. It should be noted that agrin protein and These results indicate that the motoneuron-rich spinal cord se- mRNA are expressedin avian and mammalian muscle,as well lectively increasesthe expressionof muscle agrin. as in elasmobranchelectric organ (Lieth et al., 1992; Ruegg et al., 1992; Smith et al., 1992). Thus, agrin expression by the Discussion targets of motoneurons is well conserved in evolution, and the The studies reported here were designed to assess the expression failure of frog muscle to express agrin would be a departure from of muscle agrin at nerve-induced AChR clusters. WC show that this observation. On the other hand, the detection of muscle rat spinal cord cells induce AChR clusters on cultured chick agrin at nerve-induced AChR clusters is in accord with earlier myotubes, and that agrin synthesized by the muscle cells is work demonstrating muscle agrin localization at spontaneous concentrated at these postsynaptic specializations. In addition, and 7brpedoagrin-inducedAChR clusters(Fallon and Gelfman, coculture with spinal cord cells induces an increase in the ex- 1989; Licth et al., 1992).Thus, muscle agrin is localized at AChR pressionof muscle agrin. These findings suggestthat the agrin clusters that arise in at least three different contexts. localized at the neuromuscularjunction is synthesized at least The stimulation of muscle agrin accumulation by spinal cord in part by the muscle cell, and indicate that the expressionof cells (Fig. 5) is of considerable interest since it suggeststhe muscleagrin is regulated by interaction with neurons. motoncuron may regulate the expressionof target-derived agrin Agrin synthesisby musclecells is well established(Fallon and at the synapse.Although the mechanismof this stimulation is The Journal of Neuroscience, June 1993. 13(6) 2513 unknown, one attractive and economical model is suggestedby This exogenous,neuronally derived agrin would then signal the recent data. We have previously demonstratedthat the cell sur- cxprcssion of muscle agrin at thesespecializations. This model face receptor for exogenousTorpedo agrin is distinct from the is consistent with the observation that muscle agrin alone in- AChR (Nastuk et al., 1991). Both this receptor as well as the ducts only low levels of AChR clustering (Godfrey, 1991; Lieth AChRs redistribute in responseto incubation with exogenous et al., 1992) and that antiseraspecific to neuronal agrin incom- agrin. Moreover, newly synthesized muscleagrin selectively ac- pletely block nerve-induced clustering (Rcist et al., 1992). As cumulatesat AChR clusters induced by Torpedo agrin (Lieth ct discussedpreviously, this cndogcnous muscle agrin could be al., 1992). These observations, coupled with the tyrosine phos- part of the machinery intrinsic to the myoliber that is necessary phorylation ofthe AChRs seenupon agrin stimulation (Wallace for it to organize its postsynaptic apparatus. For example, mus- et al., 1991) suggestthat exogenousagrin activates a seriesof cle agrin could serve to stabilize the postsynaptic specialization intracellular events that mediate the formation of postsynaptic (Lieth ct al., 1992). A key to understanding the respective func- specializations. It is tempting to speculate that the stimulation tions of agrin derived from thesedifferent cells is to identify the of agrin expressionin muscle cells is one of the events evoked isoforms present at each stagein synapsedevelopment. Such a by neurally derived agrin during the formation ofneuromuscular temporal characterization is essentialsince, for example, tyro- junctions. It is of course also possiblethat electrical activity or sine phosphorylation of the AChR is only observed after birth ncuronally derived factors such as ACh receptor-inducing ac- in the rat, even though AChR clustering and agrin accumulation tivity (ARIA) calcitonin gene-related pcptide could induct the arc seena week carlicr (Qu et al., 1990; Rupp ct al., 1991). It expressionof muscle agrin (New and Mudge, 1986; Usdin and is also important to characterize fully the agrin receptor at these Fischbach, 1986; Rutishauscr and Landmesser, 1991). Fur- developing junctions; work on this problem is in progress(Nas- thermore, while there is good evidence that motoneurons reg- tuk et al., 199 1). Finally. additional assaysneed to be brought ulate the expressionof muscle-derivedagrin, we cannot rule out to bear in order to probe selectively the different stagesof AChR the possibility that non-neuronal cells from the spinal cord may formation and stabilization. contribute to this regulation. We arc currently testing thesehy- In summary, we have shown that the expressionand local- pothcscs. ization of muscleagrin arc subject to neural regulation. The dual Recent molecular analysis of agrin has revealed that agrin sourceof agrin at the neuromuscularjunction, coupled with the isoforms arise by alternative mRNA splicing. Studies usingrc- existence ofthis regulatory network, suggeststhat neuronalagrin combinant fragmentsof chick agrin indicate that only the iso- alone dots not direct all aspectsof postsynaptic differentiation. forms expressedby motoneuronsinduce AChR clustering(Ruegg More likely, agrin from both the neuron and musclecollaborate et al., 1992). On the other hand, all full-length isoforms of rat to organize the ncrvc-muscle synapse. agrin can stimulate AChR aggregation(Ferns et al., 1992). While the reasonsfor this apparent discrepancy remain to be cluci- dated, it is clear that isoforms of agrin expressedby muscle are References capableof inducing AChR aggregation. AndersonMJ, CohenMW (1977) Nerve-inducedand spontaneous It has recently been reported that neuronally derived agrin is redistributionof acetylcholincreceptors on cultured musclecells. J present at AChR clusters in Xenopus/Runu coculturcs (Cohen Physiol(Lond) 268757-773. and Godfrey, 1992). In preliminary studies with chick ciliary Buckley KM, Schweitzer ES, Miljanich GP. Clift OL, Kushncr PD, Reichardt LF, Kelly RB (1983) A synaptic vesicle ant&n is re- ganglion neuronscultured on rat myotubes, we have found that stricted to the junctional region of the presynaptic plasma membrane. neural agrin is localized at AChR clusters (Lieth and Fallon, Proc Nat1 Acad Sci lJSA 80:7342-7346. unpublished observations). In addition, motoneurons synthe- Burden SJ, Sargent PB. McMahan UJ (1979) Acetylcholine receptors size agrin and transport it anterogradely (Magill-Sole and in regeneratingmuscle accumulate at original synaptic sites in the McMahan, 1988) and antisera specific for chick agrin partially absence of the nerve. J Cell Biol 82:412+25. Campanelli JT, Hoch W, Rupp F, Kreiner T, Scheller RH (199 1) Agrin block AChR cluster induction in chick ciliary ganglion/rat mus- mediates cell contact-induced acctylcholine receptor clustering. Cell cle cultures (Reist et al., 1992). Taken together with the present 67~909-9 16. results, thesedata indicate that them is a dual cellular origin of Cohen MW, Godfrey E (1992) Early appearance of and neuronal con- agrin at the neuromuscularjunction. tribution to agrin-like molecules at embryonic frog nerve-muscle synapses formed in culture. J Neurosci 12:2982-2992. We proposethat this dual origin is a reflection and a mediator Cohen MW, Weldon PR (I 980) Localization ofacctylcholine receptors of the critical cellular interactions underlying synaptogenesis. and synapticultrastructure at nerve-musclecontacts in culture: dc- The formation of the neuromuscularjunction isa lengthy, com- pendenccon nerve type. J Cell Biol 86:388-401. plex process.Over 7 weekselapse between the time that ncrve- Dahm LM. Landmesser LT (1991) The regulation of synaptogenesis induced AChR clusters first appear to the point at which the during normaldevelopment and followingactivity blockade.J Neu- rosci 11:238-255. synapse is fully mature (reviewed in Salpeter, 1987; see also Fallon JR, Gelfman CE (I 989) Agrin-related molecules arc concen- Dahm and Landmesser,1991). Several major events occur in trated at acetylcholine receptor clusters in normal and aneurai de- the intervening period including motoneuron cell death, for- velopingmuscle. J Cell Biol 108:1527-l 535. mation and innervation of secondary myotubcs, elimination of Fallon JR, Nitkin RM, Reist NE, Wallace BG, McMahan UJ (1985) Acetylcholine receptor-aggregating factor is similar to molecules con- polyneuronal innervation, and metabolic stabilization of AChR centratedat neuromuscularjunctions. Nature 3 I5:57 l-574. turnover. Thus, there arc numerouspoints during synaptogcne- FernsMJ, Hoch W, Rupp F, Kreiner T, SchellerRH (1992) RNA sis in which muscle agrin may play a role, and thcsc cvcnts are splicing regulates agrin-mcdiatcd acctylcholine receptor clustering ac- not limited to the initial events in nerve-induced AChR clus- tivity on culturedmyotubes. Neuron 8:1079-1086. tering. Frank E, Fischbach GD (1979) Early events in neuromuscularjunction formation in vitro: induction of acetylcholinc receptor clusters in the A model for agrin expression suggestedby these studies is postsynaptic membrane and morphology of newly formed synapses. that neurally derived agrin is rclcasedonto the muscle ccl1 sur- J Ccl1 Biol 83:143-158. face and inducesthe initial formation ofpostsynaptic structures. Godfrey EW (I 99 1) Comparison of agrin-like proteins from the ex- 2514 Lieth and Fallcm l Musde Agrin and Synapse Formation

tracellular matrix of chicken kidney and muscle with neural agrin, a neurons induces the aggregation of acetylcholine receptors at neuro- synapse organizing protein. Exp Cell Res 19599-109. muscular junctions. Neuron 8:865-868. Godfrey EW, Siebenlist RE, Wallskog PA, Walters LM, Bolender DL, Role LW, Roufa DG, Fischbach GD (1987) The distribution of ace- Yorde DE (1988) Basal lamina components are concentrated in tylchohne receptor clusters and sites of transmitter release along chick premuscle masses and at early acetylcholine receptor clusters in chick ciliary ganglion neurite-myotube contacts in culture. J Cell Biol 104: embryo hindlimb muscles. Dev Biol 130:47 I-486. 37 l-379. Lieth E, Missias AC, Yang E, Fallon JR (1989) Agrin-related molecules Ruegg MA, Tsim KW, Horton SE, Kmeger S, Escher G, Gensch EM, at the developing neuromuscular junction. The assembly of the ner- McMahan UJ (1992) The agrin gene codes for a family of basal vous system. New York: Liss. lamina proteins that differ in function and distribution. Neuron 8:69 I- Lieth E, Cardasis CA, FaIlon JR (1992) Muscle-derived agrin in cul- 699. tured mvotubes: exoression in the basal lamina and at induced ace- Rupp F, Payan DG, Magill-Sole C, Cowan DM, Scheller RH (1991) tylcholine receptor clusters. Dev Biol 149:4 l-54. Structure and expression of a rat agrin. 6:s 1 l-823. Madll-Sole C. McMahan UJ f 1988) Motor neurons contain aarin-like Rutishauser U, Landmesser L (199 1) Polysialic acid on the surface of molecules. J Cell Biol 107: 1825-I833. axons regulates patterns of and activity-dependent innervation. Trends McMahan UJ (1990) The agrin hypothesis. Cold Spring Harbor Symp Neurosci 14:528-532. Quant Biol 50:407-t 18. Salpeter MM (1987) Development and neural control of the neuro- Nastuk MA, Fallon JR (1993) Agrin and the molecular choreography muscular junction and of the junctional acetylcholine receptor. In: of synapse formation. Trends Neurosci 16:72-76. The vertebrate neuromuscular junction (Salpeter MM, ed), pp 55- Nastuk MA, Lieth E, Ma J, Cardaisis CA, Moynihan EB, McKechnie 115. New York: Liss. BA, Fallon JR (199 I) The putative agrin receptor binds ligand in a Smith MA, Magill-Sole C, Rupp F, Yao YM, Schilling JW, Snow P, calcium-dependent manner and aggregates during agrin-induced ace- McMahan UJ (1992) Isolation and characterization of a cDNA that tylcholine receptor clustering. Neuron 7:807-S 18. encodes an agrin homologue in marine ray. Mol Cell Neurosci 3:406- New HV, Mudge AW (1986) Calcitonin gene-related peptide_ _ regulates 417. AchR synthesis. Nature 323:809-S 11. - Tsim KW, Ruegg MA, Escher G, Kroeger S, McMahan UJ (1992) Nitkin RM. Smith MA. Maaill C. Fallon JR. Yao YM. Wallace BG. cDNA that encodes active agrin. Neuron 8:677-689. McMahan UJ (1987) Ide&ication of agrin, a synaptic organizing Usdin TB, Fischbach GD (1986) Purification and characterization of protein from Torpedo electric organ. J Cell Biol 105:2471-2478. a polypeptide from chick brain that promotes the accumulation of Qu ZC, Moritz E, Huganir RL (1990) Regulation of tyrosine phos- acetylcholine receptors in chick myotubes. J Cell Biol 103:493-507. photylation of the nicotinic receptor at the rat neuromuscular junc- Wallace BG (1989) Agrin-induced specializations contain cytoplas- tion. Neuron 4~367-378. mic, membrane, and extracellular matrix-associated components of Reist NE, Magill C, McMahan UJ (1987) Agrin-like molecules at the postsynaptic apparatus. J Neurosci 9: 1294-l 302. synaptic sites in normal, denervated, and damaged skeletal muscles. Wallace BG, Qu Z, Huganir RL (199 1) Agrin induces phosphorylation J Cell Biol 1052457-2469. of the nicotinic acetylcholine receptor. Neuron 6:869-878. Reist NE, Werle MJ, McMahan UJ (1992) Agrin released by motor