Mechanism of Agrin-Induced Acetylcholine Receptor Aggregation Bruce G. Wallace Department of Physiology, University of Colorado School of Medicine, Denver, Colorado 80262

SUMMARY Agrin induces the formation of specializations on chick myotubes in culture at which several components of the postsynaptic apparatus accumulate, including acetylcholine receptors ( AChRs). Agrin also induces AChR phosphorylation. Several lines of evidence suggest that agrininduced phosphorylation of tyrosine residues in the ,f3 subunit of the AChR is an early step in receptor aggregation: agrin-induced phosphorylation and aggregation have the same dose dependence; treatments that prevent aggregation block phosphorylation; phosphorylation begins before any detectable change in receptor distribution, reaches a maximum hours before aggregation is complete, and declines slowly together with the disappearance of aggregates after agrin is withdrawn; agrin slows

the rate a t which receptors are solubilized from intact rnyotubes by detergent extraction; and the change in receptor extractability parallels the change in phosphorylation. A model for agrin-induced AChR aggregation is presented in which phosphorylation of AChRs by an agrin-activated protein tyrosine kinase causes receptors to become attached to the cytoskeleton, which reduces their mobility and detergent extractability, and leads to the accumulation of receptors in the vicinity of the activated kinase, forming an aggregate. (c 1992 John Wiley & Sons, Inc.

Keywords: acetylcholine receptor, agrin, tyrosine phosphorylation, neuromuscular junction, postsynaptic membrane, protein tyrosine kinase. ~~

INTRODUCTION

Synapses throughout the nervous system are characterized by cytoplasmic, membrane, and extracellular matrix specializations, many of which are known to play a crucial role in transmission. Nowhere are these specializations more evident than at the vertebrate skeletal neuromuscular junction. The motor nerve terminal has specialized active zones, which consist of clusters of synaptic vesicles, aggregates of ion channels, and presynaptic densities, that mediatc release of the transmitter acetylcholine. The postsynaptic apparatus includes aggregates of transmitter receptors in the muscle fiber membrane that detect acetylcholine released from nerve terminals, unique cytoskeletal components in the muscle cytoplasm that are thought to mainReccived March 3, 1992: accepted April 6, 1992 Journal of Neurobiology, Vol. 23. No. 5 , pp. 592-604 (1992) G’ 1992 John Wiley & Sons. Inc. CCC 0022-3034/92/050592- 13$04.00

592

tain synaptic structure, and a high concentration of acetylcholinesterase, the enzyme that terminates the action of acetylcholine, in the basal lamina that lines the synaptic cleft. Determining how these complex synaptic specializations are assembled and maintained at the neuromuscular junction is of fundamental importance to understanding the formation, maintenance, and modification of synapses throughout the nervous system. The formation of postsynaptic specializations in developing and regenerating muscles relies on communication between motoneurons and myofibers. Numerous studies over the last two decades have shown that during development the axon terminal provides signals that direct the muscle fiber to assemble a postsynaptic apparatus at the site of nerve muscle contact (Dennis, 198 1 ; Schuetze and Role, 1987). An important clue to the identity of one such signal came from experiments on regenerating muscles in the adult. The basal lamina in the synaptic cleft of adult junctions was found to contain molecules that direct regenerating muscle

,4grin-lnduced AChR Aggregation

fibers to form aggregates of acetylcholine receptors ( AChRs) and acetylcholinesterase ( AChE) (Bur-

den et al., 1979; McMahan and Slater, 1984; Anglister and McMahan, 1985). Experiments aimed at identifying and characterizing the basal lamina molecules that induce the aggregation of AChRs and AChE led to the discovery of agrin, a protein that appears to play a central role in this communication process. Identification of Agrin

Agrin was originally isolated from a basal laminacontaining fraction ofthe Torpedo electric organ, a tissue with a particularly high density of cholinergic synapses. When applied to chick myotubes in culture, agrin induces the formation of specializations at which various components of the postsynaptic apparatus accumulate, including AChRs, AChE, butyrylcholinesterase ( BuChE), a heparan sulfate proteoglycan, laminin, and a 43-kD AChRassociated protein (Godfrey, Nitkin, Wallace, Rubin, and McMahan, 1984; Nitkin et al., 1987; Wallace, 1986, 1989; McMahan and Wallace, 1989; Nitkin and Rothschild, 1990). Results of a variety of experiments suggest that the formation of the postsynaptic apparatus at developing and regenerating neuromuscular junctions is triggered by agrin that is released from motor axon terminals and becomes stably bound lo the synaptic basal lamina ( McMahan, 1990; Campanelli, Hoch, Rupp, Kreiner, and Scheller, 199 I ). For example, motoneuron-enriched fractions of spinal cord contain agrin-like AChR- and AChE-aggregating molecules, antibodies to agrin label motoneurons, as do antisense probes for agrin mRNA, agrin-like molecules are transported in motor axons to muscle, agrin is stably bound to synaptic basal lamina, and anti-agrin antibodies block basal lamina- and nerve-induced AChR aggregation at developing and regenerating neuromuscular junctions. cDNAs encoding agrin and agrin-related proteins have been cloned from marine ray, chick, and rat ( McMahan, 1990: Rupp, Payan, Magill-Solc, Cowan, and Scheller, 199 1; Tsim, Ruegg, Escher, Kroger, and McMahan, 1992). In each species, alternative splicing appears to generate a family of agrin-related messages. In chick, agrin-related transcripts encoding proteins that cause AChR aggregation were found in brain and spinal cord, while the only transcripts found in muscle encode an agrin-related protein that does not cause receptor aggregation (Ruegg et al., 1992). The function of these agrin-related proteins is not clear. They

593

may be involved in synapse formation; immunohistochemical experiments have shown that musclederived agrin-like proteins accumulate together with AChRs at developing neuromuscular junctions in vivo and at agrin-induced specializations in cultured myotubes ( Lieth, Cardasis, and Fallon, 1992). It has been suggested that muscle-derived agrin-related proteins play a role in the formation or stabilization of AChR aggregates (Lieth et al., 1992), yet this appears unlikely. Hybrid nervemuscle co-culture experiments have demonstrated that antibodies specific for nerve-derived agrin block nerve-induced receptor aggregation, whereas antibodies specific for muscle-derived agrin-related proteins do not (Cohen and Godfrey, 1992; Reist, Werle, and McMahan, 1992). Moreover, although muscle-derived agrin-like molecules accumulate at agrin-induced aggregates in cultured myotubes, they do not stabilize the aggregates; if agrin is removed from the medium, then the aggregates disperse (Wallace, 1988: B. G. Wallace, unpublished observations). Formation of a mature neuromuscular junction no doubt involves signals in addition to agrin. For example, electromechanical activity is known to regulate the extent to which AChE accumulates at synaptic sites (Dennis, 198 1 ). There is also evidence that molecules, such as ascorbate, acetylcholine receptor-inducing activity (ARIA), and calcitonin gene-related peptide (CGRP), may be released by axon terminals to alter the subunit composition and increase the rate of synthesis of AChRs (New and Mudge, 1986; Fontaine, Karsfeld, and Changeux, 1987; Harris et al., 1988, 1989; Horovitz, Knaack, Podleski, and Salpeter, 1989; Witzemann, Brenner, and Sakmann, 1991), and that endogenous factors may cause AChRs to be inserted preferentially at synaptic sites (Role, Matussian, O’Brien, and Fischbach, 1985; Dubinsky, Loftus, Fischbach, and Elson, 1989). Moreover, a variety of factors in addition to agrin have been shown to cause the formation of AChR aggregates on myotubes in culture (Wallace, 1988; McMahan and Wallace, 1989). Some of these are clearly nonphysiological, such as positively charged latex beads (Peng and Cheng, 1982). Understanding the mechanism of action of any of these might shed light on how AChRs accumulate at neuromuscular junctions. However, given the evidence that agrin is the signal released by motor axon terminals to induce differentiation of the postsynaptic apparatus in developing and regenerating muscles, it seems particularly important to understand agrin’s mechanism of action.

594

M’ullucc.

Characteristics of Agrin-Induced AChR Aggregation The formation of agrin-induced specializations on chick myotubes in culture resembles events at developing neuromuscular junctions. When agrin is added to the medium bathing the myotubes, small aggregates of AChRs begin to appear within 2 h and increase rapidly in number until 4 h (Wallace, 1988). Over the next 12-20 h, the total area occupied by aggregated receptors continues to increase, and the mean size of each aggregate grows to approximately I5 pm’. AChRs accumulate into agrin-induced specializations through lateral migration of receptors already on the surface of the myotube by an energy-dependent process (Godfrey et al., 1984; Wallace, 1988). The dependence on metabolic energy indicates that agrin-induced AChR aggregation is a more complex process than the simple cross-linking of surface molecules by a multivalent ligand, which does not rely on energy metabolism. Moreover, for every molecule of agrin added to the culture medium, an estimated 160 AChRs are induced to aggregate (Nitkin et al., 1987). This stoichiometry, together with the finding that agrin-induced specializations contain several components of the postsynaptic apparatus in addition to AChRs, suggests that agrin acts catalytically, for example, by binding to a receptor on the surface of myotubes and triggering a cascade of intracellular events that results in the formation of a postsynaptic specialization (Nitkin et al., 1987). Recent experiments have tentatively identified such an agrin receptor (Nastuk et al., 199 1 ). Agrin induces aggregation of AChRs in the presence of inhibitors of protein synthesis (Wallace, 1988), consistent with the finding that agrin-induced accumulation of AChRs occurs by lateral migration of preexisting receptors. The accumulation of components of the extracellular matrix, such as heparan sulfate proteoglycan, would seem less likely to occur by lateral migration, and, therefore, might depend on the release of newly synthesized molecules. Indeed, formation of aggregates of heparan sulfate proteoglycan is prevented by inhibitors of protein synthesis (Wallace, 1989). Thus, different components of the postsynaptic apparatus accumulate in agrin-induced specializations by different mechanisms. Role of Phosphorylation in AChR Aggregation The finding that agrin-induced AChR aggregation does not require protein synthesis indicates that

formation of AChR aggregates is mediated by posttranslational modification of existing proteins. One of the primary mechanisms for posttranslational regulation of protein function is phosphorylation, and activation of protein kinases in muscle cells has been shown to influence AChR aggregation. For example, treating chick myotubes with phorbol 1 2-myristate 13-acetate (TPA ), an activator of protein kinase C, blocks the formation of agrin-induced AChR aggregates (Wallace, 1988 ), and causes spontaneous and agrin-induced receptor aggregates to disperse (Wallace, 1988; Ross, Rapuano, and Prives, 1988 ). Myotubes transformed with the Rous sarcoma virus do not have spontaneously occurring AChR aggregates, nor do they respond to aggregating factors such as agrin. These effects of transformation are mediated by the src gene product, a protein tyrosine kinase (Anthony, Schuetze, and Rubin, 1984). On the other hand, beads coated with basic fibroblast growth factor (bFGF) induce AChRs to aggregate in cultured Xenupus myocytes at the site of bead muscle contact (Peng, Lauren, and Chen, 1991). The receptor for bFGF is a protein tyrosine kinase, and tyrosine kinase inhibitors block the aggregation of AChRs induced by bFGF-coated beads. AChRs are themselves substrates for several protein kinases. The nicotinic AChR in electric organ and skeletal muscle is a pentameric complex of four subunits, a, p, y (or c ) , and 6. in the stoichiometry a&6 (Changeux, 1984). When purified from electric organ, each of the four subunits is phosphorylated (Vandlen, Wu, Eisenach, and Raftery, 1979; Hopfield, Tank, Greengard, and Huganir, 1988). In addition, several protein kinases that phosphorylate AChRs are found in postsynaptic membranes: a CAMP-dependent protein kinase, protein kinase C, and a protein tyrosine kinase (Huganir and Greengard, 1990). Phosphorylation sites have been identified on each of the subunits and are highly conserved (Huganir and Miles, 1989). The number of sites and the diversity of kinases for which AChRs are substrates suggest that phosphorylation of AChRs may modulate receptor function in several ways. Somewhat surprisingly, phosphorylation by several kinases at various sites on different subunits all cause similar changes in the rate of receptor desensitization (Huganir and Greengard, 1990). One explanation for such an apparent convergence of regulatory mechanisms is that phosphorylation may modulate other properties of the receptor, such as aggregation, assembly (Ross, Rapuano, Schmidt, and Prives, 1987) or turnover (Shyng, Xu, and Salpeter,

Agrin-Induced AChR Aggregution 1991 ), and that at least some of the effects on desensitization may be secondary, due to relatively nonspecific electrostatic interactions arising from adding a negatively charged residue on the cytoplasmic surface of the protein (Perozo and Bezanilla, 1990). The broad array of proteins phosphorylated by activation of protein kinase C, cell transformation, and activation of the bFGF receptor, and the many changes in cell morphology and metabolism that result make it difficult to determine if the associated effects on AChR aggregation are a direct consequence of receptor phosphorylation. The effects of agrin on chick myotubes in culture, on the other hand, are quite selective. While agrin induces the formation of specializations at which several components of the postsynaptic apparatus accumulate, including AChRs, AChE, and BuChE, agrin has no detectable effect on the number or size of the myotubes or their DNA or protein content, on the number of surface AChRs or their rate of degradation, or on the amount, rate of synthesis, accumulation and release, or molecular forms of AChE and BuChE (Godfrey et al., 1984; Wallace, 1989). Thus, agrin-induced AChR aggregation may allow a more rigorous investigation of the role of AChR phosphorylation in the regulation of receptor distribution. In chick myotubes in culture, the 0, y,and 6 AChR subunits are phosphorylated, as determined by incubating cultures overnight with 32P-inorganic phosphate, isolating surface AChRs, and analyzing them by SDS-polyacrylamide gel electrophoresis and autoradiography (Fig. 1). Overnight incubation with agrin causes a threefold increase in phosphorylation of the p subunit and a 20-30% increase in phosphorylation of the y and 6 subunits. These effects are selective; agrin treatment causes no detectable change in either the incorporation of 32Pinto proteins precipitated by trichloroacetic acid (TCA) or the pattern of phosphoproteins revealed by 2D isoelectric focussing-SDSpolyacrylamide gel electrophoresis. Agrin-induced phosphorylation of the y and 6 AChR subunits is blocked by 1- [ 5-isoq1~inolinylsulfonyl]-2-methylpiperazine (H-7), an inhibitor of several protein serine kinases (Hidaka, Inagaki, Kawamoto, and Sasaki, 1984), indicating that phosphorylation of these subunits might be due to the activity of serine kinases such as protein kinase C or a cyclic nucleotide-dependent protein kinase. On the other hand, H-7 does not block agiin-induced phosphorylation of the p subunit. Moreover, only protein tyrosine kinases are known to phosphorylate the p subunit

595

3 2 ~ - p h ~ ~ p h a t ephosphotyrosine

con ag BTx cur

BTx con ag

Figure 1 Agrin-induced phosphorylation of AChRs in cultured chick myotubes. AChRs were isolated from chick myotubes and the subunits were separated by SDSpolyacrylamide gel electrophoresis. [ 32P]-phosphate: Autoradiogram of AChRs isolated from myotube cultures labelled with [ 32P]H,PO,. The 0,y,and 6 subunits were labelled in control cultures. Overnight incubation with agrin caused an increase in phosphorylation of all three subunits. phosphotyrosine: Autoradiogram of AChRs isolated, transferred to nitrocellulose, and probed with anti-phosphotyrosine antibodies and [ '251]protein A. The only subunit labelled was the 6 subunit, and agrin caused a marked increase in the level of phosphotyrosine it contained. (con): AChRs isolated from control cultures. (ag): AChRs isolated from myotubes treated overnight with agrin. (BTx, cur): Samples from agrin-treated cultures in which isolation of AChRs was specifically blocked with a-bungarotoxin or curare. The receptor subunits were identified using subunit-specific monoclonal antibodies (Wallace, 1991; Wallace et al., 1991 ); the position and appearance of the bands corresponding to the y and 6 subunits suggests that these subunits were partially degraded during the isolation procedure.

(Huganir and Miles, 1989). These findings suggest that agrin-induced phosphorylation of the subunit might be occurring on tyrosine residues. Western blot analysis of isolated AChR subunits with anti-phosphotyrosine antibodies confirmed that

596

Uullace

agrin causes a selective increase in tyrosine phosphorylation of the p AChR subunit (Fig. 1 ). Agrin-induced changes in tyrosine phosphorylation of the AChR p subunit are correlated with changes in receptor distribution, while changes in phosphorylation of the y and 6 subunits are not. Thus, H-7 blocks agrin-induced phosphorylation of the y and 6 subunits, but does not block agrininduced AChR aggregation. Conversely, three inhibitors of agrin-induced receptor aggregation, low pH, TPA, and polyanions (Wallace, 1988, 1990) also block agrin-induced tyrosine phosphorylation of the fl subunit (Wallace et al., 199 1 ). In addition, agrin-induced receptor aggregates stain with antiphosphotyrosine antibodies (Wallace, Qu, and Huganir, 199 I ). These findings led to the hypothesis that, in chick myotubes, agrin-induced AChR aggregation is the result of activation of a protein tyrosine kinase that phosphorylates the AChR subunit, and that, as a consequence of phosphorylation, the properties of AChRs change such that they tend to become attached to the cytoskeleton and so accumulate in the vicinity of the activated kinase, forming an aggregate (Wallace et a]., 1991). In order to test this hypothesis, we have examined in greater detail the relationship between agrin-induced tyrosine phosphorylation of AChRs and receptor aggregation.

METHODS Chick Myotube Cultures Myotube cultures were prepared from hindlimb muscles of 1 I - to 12-day-old White Leghorn chick embryos by the method of Fischbach (1972) with minor modifications (Wallace, 1989). Experiments were made on 5- to 6-day-old myotube cultures.

Agrin Experiments were made with partially purified agrin (Cibacron pool fraction) prepared from the electric organ of Torpedo cnlifomica as previously described ( Nitkin et al., 1987). With the exception of the dose-response experiments, 4 U of agrin were added to each culture dish.

Assay of Tyrosine Phosphorylation of Surface AChRs Surface AChRs were labelled with biotinylated-n-bungarotoxin, and the toxin-receptor complexes were solubilized with detergent and isolated on streptavidin-conjugated Sepharose beads as previously described (Wallace

et al., 199 I ). The extent of phosphotyrosine incorporation was measured by incubating the beads bearing the toxin-receptor complexes with anti-phosphotyrosine antibodies and '251-protein A (Wallace et al., 1991).

Detergent Extraction Surface AChRs were labelled with ' 2 5 1 - ~ ~ - b ~ n g a r ~ t o ~ i n , the cultures were rinsed at room temperature twice with medium, twice with buffer containing 0.286 M sucrose, 50 m M NaCI, 1 ink! MgCI,, 10 m M HEPES, pH 7.4 (buffer A), and incubated in buffer A supplemented with 0.05% Triton X-100 (Prives, Fulton, Penman. Daniels, and Christian, 1982). The detergent-containing buffer was replaced at 2-min intervals. and the fraction of receptors solubilized was determined by counting the extracts in a gamma counter. At the end of the incubation period, the cultures were incubated for 1-2 h with 1 ml of 1 N NaOH, and the extracts were counted to measure residual AChRs.

Quantitation of Aggregation Myotube cultures were labelled with 2 X lo-* M rliodamine-conjugated n-bungarotoxin, rinsed, and fixed as prcviously described (Wallace, I Y8Y ) . Cultures were examined with a Nikon Optiphot microscope equipped for phase contrast and epifluorescencc using a 40X or lOOX oil-immersion phase-contrast objective. Myotube segments (200 or 85 pm in length) were imaged with a Star I chilled CCD camera (Photometrics Ltd), and the images were analyzed using WHIP Virtual Image Processing Software (G. w . Hannaway and Associates, Boulder, CO). The extent of receptor aggregation was computed by selecting by eye a threshold pixel intensity that discriminated between receptor aggregates and diffusely labelled portions of thc myotube surface. This threshold value was subtracted from all pixels, and the remaining fluorescence intensity was integrated. Assays were done on triplicate cultures; for each culture dish 14 myotube segments were analyzed.

RESULTS Dose Dependence of Agrin-Induced Aggregation and Phosphorylation

As increasing amounts of agrin are added to chick myotube cultures, the number of AChR aggregates induced during an overnight incubation increases and then reaches a plateau (Godfrey et al., 1984). If AChR aggregation is driven by tyrosine phosphorylation, then agrin-induced tyrosine phosphorylation might be expected to vary in a similar way with agrin concentration. The extent of tyrosine phosphorylation of AChRs was determined by

**O

r

1

&aggregation - - - phosphorylation

0 ' 0

I

I

2

I

I

4

I

I

6

I

I

8

units of agrin

Figure 2 Dose dependence of agrin-induced AChR aggregation and phosphorylation. Myotube cultures were incubated overnight with the indicated concentration of agrin and the extent of AChR aggregation and tyrosine phosphorylation determined. The results are expressed as the mean & S.E.M. ( n = 8 for phosphorylation, n = 12 for aggregation).

isolating receptors from cultured chick myotubes, labelling the isolated AChRs with anti-phosphotyrosine antibodies, and measuring the extent of antibody binding using '251-labelled protein A ( Wallace et al., 199 l ) . AChR aggregation was analyzed in sister cultures by labelling AChRs with rhodamine-conjugated a-bungarotoxin, acquiring fluorescence images of the myotubes with a chilled CCD camera, and measuring the integrated fluorescence intensity of all aggregated receptors. Figure 2 shows that as the amount of agrin added during an overnight incubation was increased, the extent of aggregation and tyrosine phosphorylation of AChRs increased in parallel. Time Course of Changes in Aggregation and Phosphorylation To examine further the relationship between receptor phosphorylation and aggregation, we analyzed the time course of agrin-induced tyrosine phosphorylation of AChRs. As illustrated in Figure 3 , an increase in tyrosine phosphorylation was detected within 30 min of adding agrin to the myotube cultures and reached a plateau by 3 h. Previous studies have shown that AChR aggregation cannot be detected until 2 h after the addition of agrin and continues for at least the next 16 h. Thus, agrin-induced AChR tyrosine phosphorylation be-

gan before any detectable change in AChR distribution and reached its maximum hours before AChR aggregation was complete. These findings are consistent with the hypothesis that agrin-induced tyrosine phosphorylation of the 0subunit of the chick AChR is an early step in agrin-induced receptor aggregation. Agrin must be present for AChR aggregates on chick myotubes to be maintained; if agrin is removed, then aggregation stops and previously induced aggregates slowly disappear ( Wallace, 1988). If aggregates form by the immobilization of phosphorylated AChRs, then the extent of tyrosine phosphorylation might be expected to decrease at the same rate as aggregated receptors disperse. To test this prediction, myotubes were treated overnight with agrin, rinsed, and placed in fresh medium, incubated for various lengths of time, and analyzed for AChR aggregation and tyrosine phosphorylation (Figure 4). Receptor phosphorylation decreased with the same slow time course as the disappearance of aggregated receptors, as expected if aggregation was a consequence of phosphorylation.

Changes in AChR Extractability Fluorescence recovery after photobleaching experiments indicate that aggregated AChRs are rela-

--m ---t

OO

4

8

12 time (hr)

phosphorylation aggregation

16

20

24

Figure 3 Time course of agrin-induced aggregation and tyrosine phosphorylation. Cultures were treated with agrin for the indicated length of time, and the extent of aggregation and tyrosine phosphorylation were determined. Results are normalized to the values observed after 24 h treatment and are expressed as mean f S.E.M. For measurements of phosphorylation each sample contained AChRs from two cultures ( n = 3 ) ; the data for aggregation is from previous experiments in which aggregation was measured as the mean area occupied by aggregated receptors on eight, 360-pm long myotube segments from each culture dish [ n = 5 , (Wallace, 1988) I .

tively immobile (Axelrod et al., 1976; Dubinsky et al., 1989). Several lines of evidence implicate the 43-kD AChR-associated protein and a small group of cytoskeletal proteins in receptor immobilization and maintenance of AChR aggregates (Froehner, 199 1 ). Consistent with this view are the results of experiments on intact myotubes in culture showing that aggregated receptors are more resistant than nonaggregated receptors to extraction by mild detergent treatment (Prives et al., 1982; Podleski and Salpeter, 1988). Accordingly, we examined the detergent extractability of AChRs in control and agrin-treated myotubes. Surface receptors were labelled with ‘*’I- a-bungarotoxin, the myotubes were rinsed, incubated in buffer containing 0.05Y0 Triton X-100, and the solubilization of toxin-receptor complexes was measured. As illustrated in Figure 5, the rate at which AChRs were solubilized was considerably slower in agrintreated cultures. If phosphorylation causes attachment to the cytoskeleton, then the resistance to detergent extraction would be expected to increase with the same time course as the increase in AChR phosphorylation. We took the fraction of receptors solubilized

during the first 2-min treatment with detergent as a measure of the rate of receptor extraction. As illustrated in Figure 6, the time course of the increase in resistance of AChRs to detergent extraction paralleled the increase in AChR phosphorylation, and was significantly faster than receptor aggregation. DISCUSSION The stoichiometry and dependence on metabolic energy of agrin-induced AChR aggregation suggest that agrin combines with a receptor on the surface of cultured chick myotubes to activate some mechanism that catalyzes the formation of a specialization at which several components on the postsynaptic apparatus accumulate. The results reported here suggest that agrin-induced tyrosine phosphorylation of the @ subunit of AChRs is an early event in this catalytic process. Agrin-induced tyrosine phosphorylation of the @ subunit and agrin-induced receptor aggregation have a similar dose dependence, suggesting that the same receptor triggers both events. Treatments that prevent agrin-induced AChR aggregation also block agrin-induced

Agrin-Indzrced AChR Aggregution

2ot

0 1 \

I

0

I

I

I

I

I

6 12 hours after agrin removal

I

18

I

599

I

24

Figure 4 Disappearance of AChR aggregates and decrease in tyrosine phosphorylation after removal of agrin. Myotube cultures were treated for 24 h with agrin, rinsed, and incubated in fresh medium for the indicated time. Results are normalized to the values observed after 24 h of treatment and are expressed as mean f S.E.M. [ II = 8 for aggregation (except n = 3 for 12-h point), n = 6 for phosphorylation].

AChR tyrosine phosphorylation. Agrin-induced tyrosine phosphorylation of AChRs begins within minutes of the addition of agrin to myotube cultures, before any detectable change in receptor distribution, and reaches a maximum hours before aggregation is complete, indicating that phosphorylation is not a consequence of aggregation. Finally, if agrin is removed from the culture medium, then the extent of tyrosine phosphorylation of the ,8 subunit declines with the same slow time course as the disappearance of receptor aggregates. Agrin also induces phosphorylation of serine or threonine residues on the y and 6 AChR subunits. However, the serine kinase inhibitor H-7 blocks agrin-induced phosphorylation of the y and 6 subunits without preventing AChR aggregation, indicating that phosphorylation ofthe y and 6 subunits is not an essential step in agnn-induced AChR aggregation. H-7 has little or no effect on tyrosine phosphorylation of the p subunit, as expected. How might agrin-induced phosphorylation of AChRs cause receptor aggregation? Postsynaptic aggregates of AChRs are thought to be formed and maintained by attachment to the subsynaptic cytoskeleton via a 43 kD receptor-associated protein (Froehner, 1991), although it has been reported that small AChR aggregates can be induced on fibroblasts expressing recombinant AChRs by a

mechanism that does not involve the 43 kD AChR-associated protein ( Hartman, Millar, and Claudio, 1991 ) , The 43 kD AChR-associated protein has been shown to bind to the p subunit of AChRs (Burden, DePalma, and Gottesman, 1983) and to actin (Walker, Boustrad, and Witzemann, 1984), and also to mediate the formation of receptor aggregates when expressed together with AChRs in Xenuptis oocytes and fibroblast cell lines (Froehner, Luetje, Scotland, and Patrick, 1990; Phillips et al., 199 1 ). Suppose that agrin-induced tyrosine phosphorylation of the p subunit altered its interaction with the 43 kD AChR-associated protein in such a way as to increase the affinity of the AChR-43kD complex for the cytoskeleton and so cause receptor immobilization. Phosphorylated AChRs would then tend to accumulate in the vicinity of an agrin-activated tyrosine kinase (or, alternatively, an agrin-inhibited protein tyrosine phosphatase) . Consistent with this hypothesis are the findings that AChR aggregates at developing chick neuromuscular junctions and at agrin-induced specializations on cultured chick myotubes stain with anti-phosphotyrosine antibodies ( Wallace et al., 1991), as expected if phosphorylated receptors are immobilized at these sites, and that tyrosine phosphorylation of AChRs in muscle is regulated by innervation, apparently through the

600

M iillucr 100%

80

+control

e +2

t

l hr agrin hr agrin hr aErin

+24

20

0

I

I

I

I

I

I

1

2

3

4

5

6

time (min)

Figure 5 Agrin-induced decrease in the rate of AChR solubiliration by Triton X- 100. Myotube cultures were incubated with agrin for the time indicated, and surfice AChRs were labelled with [ i251] a-bungarotoxin. Cultures were rinsed and incubated with buffer containing 0.05%Triton X-100. At 2-min intervals the buffer was collected and counted. The appearance of Iz5I i n thc detergent extracts was used to calculate the percentage of receptors remaining associated with the myotubes as a function of time of extraction. Data points are the mean of triplicate determinations from a single experiment.

release of a nerve-derived factor (Qu, Moritz, and Huganir, 1990). Moreover, it has recently been shown that a relatively small number of agiinbinding sites are scattered over the surface of chick myotubes in culture, and that small AChR aggregates form at these sites within hours of adding agrin to the culture medium (Nastuk et al., 199 1 ) . Further support for this hypothesis comes from the demonstration, in cultured Xenopus myocytes, that beads coated with basic fibroblast growth factor induce AChRs to aggregate at the site of bead muscle contact (Peng et al., 1991). The receptor for bFGF is a protein tyrosine kinase, and tyrosine kinase inhibitors block bFGF bead-induced AChR aggregation. However, it is not known if the /3 subunit of the AChR is a substrate for a bFGF-activated tyrosine kinase. In rat muscles, on the other hand, AChR aggregation and phosphorylation appear less closely linked (Qu et al., 1990). At developing rat neuroinuscular junctions, for example, strong phosphotyrosine immunofluorescence was not seen until approximately 2 weeks after receptor aggregation. Likewise, following denervation of adult muscles, junctional staining with anti-phosphotyrosine antibodies and the level of tyrosine phosphorylation of AChRs decrease dramatically, while receptors re-

main aggregated in the postsynaptic membrane. Interpretation of these results is difficult because the contribution that phosphorylated AChRs make to the junctional staining seen with antiphosphotyrosine antibodies cannot be determined. Nevertheless, they raise the possibility that, at least in rat muscle, factors in addition to agrin may regulate AChR tyrosine phosphorylation and/or aggregation, and that tyrosine phosphorylation may not be the only mechanism by which receptor aggregates are formed and maintained. In chick myotubes in culture, agrin-induced tyrosine phosphorylation of AChRs begins before there is any noticeable change in receptor distribution and reaches a plateau hours before aggregation is complete. This difference in time course suggests that agrin-induced AChR aggregation might be a two-step process. For example, agrin-induced phosphorylation might lead to the formation of microaggregates of receptors clustered around each agrin-activated kinase, a process that reaches a steady state by 3 h. A second, slower mechanism might bring the kinase-centered islands of receptors together, forming larger aggregates, over a period of 24 h. Our biochemical assays of phosphorylation and extractability would reflect the time course of formation of microaggregates. However,

Aggwgat ion

0

I

I

I

1

2

3

601

time (hr)

Figure 6 Time course of agrin-induced changes in AChR aggregation, phosphorylation, and resistance to detergent extraction. Cultures were treated for the indicated length of time with agrin and analyzed as described. The results are expressed as a percentage of the steady-state changes measured after 24 h ofagrin treatment and plotted as the mean -t S.E.M. (the number of observations is given in parenthesis).

if individual microaggregates were too small to be detected in our fluorescence assays of aggregation, then our measures of the extent of aggregation would monitor only the second, slow translocation process. Such a hypothesis is consistent with the observation that agrin-induced receptor phosphorylation is more rapid than receptor aggregation. It is also consistent with the finding reported here that agrin-induced resistance to detergent extraction increases with the same time course as phosphorylation. Moreover, there is evidence to suggest that agrin receptors aggregate slowly during agrin treatment (Nastuk et al., 199 1 ) . A prediction of this hypothesis is that after 3 h of agrin treatment both aggregated and nonaggregated receptors would be immobilized by attachment to the cytoskeleton. In the phosphorylation and extraction experiments reported here no distinction was made between aggregated and nonaggregated receptors. However, others have reported that in myotubes treated with extracts that cause receptor aggregation, both aggregated and nonaggregated AChRs show a decrease in mobility (Axelrod, Bauer, Stya, and Christian, 1981; Dubinsky et al., 1989) and an increase in resistance to detergent extraction (Podleski and Salpeter, 1988) compared to nonaggregated receptors to control myotubes. Such a model would also predict that activation

of any protein tyrosine kinase that could phosphorylate the @ subunit of AChRs would cause receptor immobilization. If kinase activation was localized, as in the bFGF-bead experiments, then AChRs would aggregate at that site. If a widely distributed kinase was activated, then widely distributed microaggregates would be formed, as during the initial phase of exposure to agrin. However, if the second, slow translocation process was not activated or was inhibited, then microscopically detectable aggregation would be blocked. Such a mechanism might account for the failure of bath application of bFGF to cause AChR aggregate formation in Xenopus myocytes (Peng et al., 199 1 ) and for the inhibition of receptor aggregation in myotubes transformed with Rous sarcoma virus (Anthony et al., 1984). In both ofthese cases it would be interesting to know if the AChR /Isubunit is phosphorylated, if AChRs are immobile, and if AChRs are relatively resistant to detergent extraction. Likewise, it would be of interest to determine whether other factors that induce AChR aggregation also influence receptor phosphorylation. Such a model does not account for the accumulation of other components of the postsynaptic apparatus in agrin-induced specializations. One possibility is that the accumulation of these other components is triggered by aggregation of AChRs. However, agrin has been shown to induce aggre-

602

Wdlucc~

gates containing several components of the postsynaptic apparatus in myotubes incubated with anti-AChR antibodies, a treatment that dramatically reduces AChR aggregation by causing receptor immobilization and internalization (Nitkin and Rothschild, 1990). Alternatively, an agrin-activated protein tyrosine kinase might phosphorylate other components of the postsynaptic apparatus, causing them to accumulate by a mechanism similar to that proposed for AChRs, or phosphorylate components of the cytoskeleton such that secretory granules carrying postsynaptic proteins preferentially fuse near the site of the activated kinase. No such changes in protein phosphorylation were found in preliminary experiments (Wallace et al., 1991), but if the proteins involved are relatively rare, then more sensitive techniques might be required to detect them. Our results indicating that agrin-induced tyrosine phosphorylation may be involved in the aggregation of AChRs at the neuromuscular junction raise several questions concerning the role of tyrosine phosphorylation in the formation of postsynaptic specializations at other synapses. Are such specializations induced by the release of agrin-like proteins from axon terminals? If such agnn-like proteins exist, then do they cause changes in tyrosine phosphorylation of postsynaptic proteins? Are other transmitter receptors substrates for protein tyrosine kinases? Does tyrosine phosphorylation alter the distribution, mobility, or detergent extractability of other transmitter receptors? Answers to such questions may provide insights into the mechanism by which complex postsynaptic specializations are assembled and maintained, information that is of fundamental importance to understanding the formation, maintenance, and modification of synapses throughout the nervous system. I thank Drs. U. J. McMahan. W. Betz, and H. Gates for helpful discussions, Dr. K. Tsim for providing agrin. Dr. R. Huganir for providing anti-phosphotyrosine antibodies, and Donna Kautzman-Eades for her excellent technical assistance. These studies were funded by National Institutes of Health Grant NS29673 and by BRSG-05357 awarded by the Biomedical Research Grant Program, Division of Research Resources, National Institutes of Health.

REFERENCES ANGLISTER, L. and MCMAHAN, U. J. ( 1985). Basal lam-

ina directs acetylcholinesterase accumulation at syn-

aptic sites in regenerating muscle. .I. Cell Biol. 101:735-743. ANTHONY,D. T., SCHUETZE,S. M., and RUBIN,L. L. ( 1984). Transformation by Rous sarcoma virus prevents acetylcholine receptor clustering on cultured chicken muscle fibers. Puoc. Null. Acud. Sci. USA 81~2265-2269. AXELROD,D., RAVDIN,P., KOPPEL,D. E., SCHLESSINGER, J., WEBB, w. w . , ELSON,E. L., and PODLESKI, T. R. ( 1976). Lateral motion of fluorescently labeled acetylcholine receptors in membranes ofdeveloping muscle fibers. Proc. Null. Acud. Sci. USA 73:4594-4598. AXELROD, D., BAUER,H. C., STYA.M., and CHRISTIAN, C. N. ( 198 1 ). A factor from neurons induces partial immobilization of nonclustered acetylcholine receptors on cultured muscle cells. J . Cell B i d . 88:459-462. BURDEN,S. J., SARGENT,P. B., and MCMAHAN,U. J. ( 1979). Acetylcholine receptors in regenerating muscle accumulate at original synaptic sites in the absence of nerve. J . Cell Biol. 82:4 12-425. BURDEN, S. J., DEPALMA, R. L., and GOTTESMAN, G. S. ( I983 1. Crosslinking of proteins in acetylcholine receptor-rich membranes: association between the B-subunit and the 43 kD subsynaptic protein. Cell 35687692. CAMPANELLI, J . T., HOCH,W., RUPP. F.. KREINER,T.. and SCHELLER. R. H. ( 199 1 ). Agrin mediates cell contact-induced acetylcholine receptor clustering. Cell 67:909-9 16. CHANGEUX, J. P.. DEVILLERS-THIERY. A., and CHEMOULLI, F'. ( 1984). Acetylcholine receptor: an allosteric protein. Scicwce 225: 1333- 1345. COHEN,M. W. and GODFREY,E. W. (1992). Early appearance of and neuronal contribution to agrin-like molecules at embryonic frog nerve-muscle synapses formed in culture. J . Ncuro.sci. (in press). DENNIS,M. J. ( 198 I ). Development ofthe neuromuscular junction: inductive interactions between cells. Annu. Rcv. Nrurosci.4:43-68. J . M., LOFTUS,D. J., FISCHBACH, G. D., and DUBINSKY, ELSON,E. L. ( 1989). Formation of acetylcholine receptor clusters in chick myotubes: migration or new insertion? J. CeII Biol. 109: 1733-1 743. G. D. ( 1972). Synapse formation between FISCHBACH, dissociated nerve and muscle cells in low density cultures. Dev. Biol. 28:407-429. FONTAINE, B., KARSFELD, A., and CHANGEUX, J. P. ( 1987). Calcitonin gene-related peptide and muscle activity regulate acetylcholine receptor a-subunit mRNA levels by distinct intracellular pathways. J . Cell Bid. 105: 1337- 1342. FROEHNER, S. C. ( 199 1 ). The submembrane machinery for nicotinic acetylcholine receptor clustering. J . Cell B i d . 114: 1-7. FROEHNER, S. C., LUETJE,C. W., SCOTLAND, P. B., and PATRICK,J. (1990). The postsynaptic 43K protein

Agrin-Indziccd A CIiR z4ggregution

clusters muscle nicotinic acetylcholine receptors in Xenopiis oocytes. Neuron 5:403-4 10. GODFREY, E. W., NITKIN, R. M., WALLACE,B. G., RuBIN, L. L., and MCMAHAN,U. J. ( 1984). Components of Torpedo electric organ and muscle that cause aggregation of acetylcholine receptors on cultured muscle cells. J . CcJNBiol. 99:6 15-627. HARRIS,D. A., FALLS,D. L., DILL-DEVOR,R. M., and FISCHBACH,G. D. ( 1988). Acetylcholine receptor-inducing factor from chicken brain increases the level of mRNA encoding the: receptor a subunit. Proc. Nutl. '4cud. Sci. USA 85:1983-1987. HARRIS,D. A., FALLS., D. L., and FISCHBACH,G. D. ( 1989). Differential activation of myotube nuclei following exposure to an acetylcholine receptor-inducing factor. Nature 337: 113- 176. HARTMAN,D. S., MILLAR,N. S., and CLAUDIO,T. ( 199 1 ). Extracellular synaptic factors induce clustering of acetylcholine receptors stably expressed in fibroblasts. J. Cell Biol. 115165-177. HIDAKA,H., INAGAKI,M., KAWAMOTO, S., and SASAKI, Y. ( 1984). Isoquinolinesulfonamides, novel and POtent inhibitors of cyclic nucleotide dependent protein kinase and protein kinase C. Biochemistry 23:50365041. HOPFIELD,J . F., TANK,D. W., GREENGARD, P., and HUGANIR,R. L. (1988). Functional modulation of the nicotinic acetylcholine receptor by tyrosine phosphorylation. Nature 336:677-680. HOROVITZ,O., KNAACK,D., PODLESKI, T. R., and SALPETER, M. M. ( 1989). Acetylcholine receptor a-subunit mRNA is increased by ascorbic acid in cloned L, muscle cells: Northern blot analysis and in sitti hybridization. J . Cell Biol. 108:1823-1832. HUGANIR,R. L. and GREENGARD, P. ( 1990). Regulation of neurotransmitter receptor desensitization by protein phosphorylation. Neuron 5:555-576. HUGANIR,R. L. and MILES, K. (1989). Protein phosphorylation of nicotinic acetylcholine receptors. Crit. Rev. Biochem. Mol. Biol. 24: 183-2 15. LIETH,E., CARDASIS,C. A., and FALLON,J. R. ( 1992). Muscle-derived agrin in cultured myotubes: expression in the basal lamina and at induced acetylcholine receptor clusters. Dev. Biol. 149:4 1-54. MCMAHAN,U. J. (1990). The agnn hypothesis. Cold Spring Hurbor Symp. Quunt. Biol. 50:407-4 18. MCMAHAN,U. J . and SLATER,C. R. ( 1984). Influence of basal lamina on the accumulation of acetylcholine receptors at synaptic sites in regenerating muscle. J . Cell Biol. 98:1453-1473. MCMAHAN,U. J. and WALLACE,B. G. (1989). Molecules in basal lamina that direct formation of synaptic specializations at neuromuscular junctions. Dev. Neurosci. 11:227-247. NASTUK,M. A., LIETH, E., MA, J., CARDASIS. C. A,, MOYNIHAN,E. B., MCKECHNIE,B. A., and FALLON, J. R. ( 199 1 ). The putative agrin receptor binds ligand in a calcium-dependent manner and aggregates during

603

agrin-induced acetylcholine receptor clusterjng. Ncwron 75307-8 18. NEW,H. V. and MUDGE,A. W. ( 1986). Calcitonin generelated peptide regulates muscle acetylcholine receptor synthesis. Nature 323:809-8 I 1. NITKIN,R. M. and ROTHSCHILD,T. C. ( 1990). Agrininduced reorganization of extracellular matrix components on cultured myotubes: relationship to AChR aggregation. J . Cell Biol. 11I: I 16 1- 1 170. NITKIN, R. M., SMITH, M. A,, MAGILL, C., FALLON, J. R., YAO, Y. -M. M., WALLACE,B. G., and McMAHAN, U. J. ( 1987). Identification of agnn, a synaptic organizing protein from Torpedo electric organ. J. Cell Bid. 105:247 1-2478. PENG, H. B. and CHENG,P. -C. (1982). Formation of postsynaptic specializations induced by latex beads in cultured muscle cells. J. Nezirosci. 2: 1760- 1777. PENG,H. B., LAUREN,L. P., and CHEN,Q. ( 199 1 ). Induction of synaptic development in cultured muscle cells by basic fibroblast growth factor. Neuron 6:237246. PEROZO.E. and BEZANILLA, F. ( 1990). Phosphorylation affects voltage gating of the delayed rectifier K + channel by electrostatic interactions. Neziron 5:685-690. PHILLIPS, W. D.. KOPTA, C.. BLOUNT, P., GARDNER, P. D., STEINBACH,J . H., and MERLIE, J . P. ( 1991 ). ACh receptor-rich domains organized in fibroblasts by recombinant 43-kilodalton protein. Science 25 1:568570. PODLESKI,T. R. and SALPETER,M. M. (1988). Acetylcholine receptor clustering and Triton solubility: neural effect. J. Nezirobiol. 19: 167- 1 85. PRIVES, J., FULTON, A. B., PENMAN, S., DANIELS,M. P., and CHRISTIAN, C. N. ( 1982). Interaction of the cytoskeletal framework with acetylcholine receptor on the surface of embryonic muscle cells in culture. J . Cell 3id 92:23 1-236. Qu, Z., MORITZ,E., and HUGANIR,R. L. ( 1990). Regulation of tyrosine phosphorylation of the nicotinic acetylcholine receptor at the rat neuromuscular junction. Neuron 2:367-378. REIST, N. E., WERLE, M. J., and MCMAHAN,U. J. ( 1992). Agrin released by motor neurons induces the aggregation of acetylcholine receptors at neuromuscular junctions. Neuron 8:865-868. ROLE, L. W., MATOSSIAN,R. R., O'BRIEN, R. J., and FISCHBACH, G. D. (1985). On the mechanism ofacetylcholine receptor accumulation at newly formed synapses of chick myotubes. J . Neurosci. 5:2 1972204. Ross, A., RAPUANO,M., and PRIVES,J. ( 1988). Induction of phosphorylation and cell surface redistribution of acetylcholine receptors by phorbol ester and carbamylcholine in cultured chick muscle cells. J. Cell Biol. 107:1139-I 145. ROSS,A. F., RAPUANO,M., SCHMIDT,J. H., and PRIVES, J. M. ( 1987). Phosphorylation and assembly of nicotinic acetylcholine receptor subunits in cultured chick muscle cells. J. Biol.Chem. 262: 14640- 14647.

604

M'ullucr

RUEGG. M. A , , TSIM, K. W. K., HORTON, S. E., KROGER, S., ESCHER, G.. GENSCH, E. M., and MCMAHAN,U. J. ( 1992). The agrin gene codes for a family of basal lamina proteins that differ in function and distribution. Ncwron 8:69 1-699. Rtipp, F., PAYAN,D. G., MAGILL-SOLC,C., COWAN, D. M., and SCHELLER,R. H. ( 1991). Structure and expression of a rat agrin. Nezirorz 6:8 1 1-823. SCHUETZE.S. M. and ROLE,L. W. ( 1987). Developmental regulation of nicotinic acetylcholine receptors. A nnzr. Rev. Ncwosci. 10:403-4 5 7. SHYNG,S. -L.. Xti. R., and SALPETER,M. M. ( 1991). Cyclic AMP stabilizes the degradation of original junctional acetylcholine receptors in denervated muscle. Neiiron 6:469-475. TSIM,K. W. K., RUEGG, M. A., ESCHER,G., KROGER, S.. and MCMAHAN,U. J. ( 1992). cDNA that encodes active agrin. Armron 8:677-689. VANDLEN. R. L., WU, W. C. -S.. EISENACH,J. C., and RAFTERY, M. A. ( 1979). Studies of the composition of purificd T o r p ~ d ocu1ifi)rnica acetylcholine receptor and of its subunits. Biochrmistry 18: 1845-1 853. WALKER,J. H., BOUSTRAD,C. M., and WITZEMANN, V. ( 1984). The 43-K protein, II,.associated with acetylcholine receptor containing membrane fragments is an actin-binding protein. Eh'RO. ./.3:2287-2290.

WALLACE,B. G. ( 1986). Aggregating factor from Torpedo electric organ induces patches containing acetylcholine receptors, acetylcholinesterase, and butyrylcholinesterase on cultured myotubes. J . Cell B i d . 102:783-794. WALLACE,B. G. (1988). Regulation of agrin-induced acetylcholine receptor aggregation by Ca++and phorbol ester. J . Cell B i d 107:267-278. WALLACE,B. G. ( 1989). Agrin-induced specializations contain cytoplasmic, membrane, and extracellular matrix-associated components of the postsynaptic apparatus. J . Neurosci. 9: 1294- I 302. WALLACE,B. G. ( 1990). Inhibition of agrin-induced acetylcholine-receptor aggregation by heparin, heparan sulfate. and other polyanions. J. Neiirosci. 10:3576-3582. WALLACE,B. G. ( 199 1 ). The mechanism of agrin-induced acetylcholine receptor aggregation. Phil. Truns. R. Soc. Lond. €3 331:273-280. WALLACE,B. G., Qu, Z., and HUGANIR,R. L. ( 199 I ). Agrin induces phosphorylation of the nicotinic acetylcholine receptor. Neziron 6:869-878. WITZEMANN,V., BRENNER,H. -R., and SAKMANN,B. ( 1991 ). Neural factors regulate AChR subunit mRNAs at rat neuromuscular synapses. J . c'c~llBiol. 114:125-141.