Nerve-induced disruption and reformation of β1-integrin aggregates during development of the neuromuscular junction

Mechanisms of Development 67 (1997) 125–139

Nerve-induced disruption and reformation of b1-integrin aggregates during development of the neuromuscular junction M. John Anderson a ,*, Zhong Qiao Shi b, Saul L. Zackson b a

b

Department of Anatomy, The University of Calgary, Calgary, Alberta, T2N 4N1 Canada Department of Medical Biochemistry, The University of Calgary, Calgary, Alberta, T2N 4N1 Canada Received 27 February 1997; received in revised form 19 May 1997; accepted 19 June 1997

Abstract The earliest biochemical change detected during synaptogenesis is a local elimination of muscle basal lamina proteins. To explore whether this provides signal(s) that regulate postsynaptic differentiation, we examined the effects of innervation on the distribution of b1integrins, which were initially present in scattered aggregates complexed with basal lamina ligands. These b1-integrin aggregates disappear along paths of nerve contact as their basal lamina ligands are eliminated. New accumulations of these proteins then form during assembly of the postsynaptic apparatus. The new b1-integrin aggregates at developing synapses form partly via a redistribution of mobile molecules on muscle surface. We thus consider whether (a) the removal of integrins’ basal lamina ligands alters their cytoplasmic ligand-interactions, causing the dissociation of integrin clusters, and (b) this receptor modulation helps to transduce local changes in pericellular protease activity into cytoplasmic signals that control postsynaptic differentiation.  1997 Elsevier Science Ireland Ltd. Keywords: Integrin; Synaptogenesis; Redistribution; Proteolysis

1. Introduction The postsynaptic apparatus of the neuromuscular junction contains dense accumulations of several specialized proteins, such as acetylcholinesterase and the acetylcholine receptor (AChR), that mediate synaptic transmission between motor neurons and skeletal muscle fibers. These postsynaptic components are attached to elaborate transmembrane complexes that contain a variety of basal lamina (Moody-Corbett and Cohen, 1981; Sanes, 1982; Anderson and Fambrough, 1983; Sanes et al., 1986; Parthasarathy and Tanzer, 1987; Reist et al., 1987; Hunter et al., 1989; Ohlendieck et al., 1991; Bewick et al., 1992; Cifuentes-Diaz et al., 1994; Kramarcy et al., 1994) and cytoskeletal proteins (Froehner et al., 1981; Bloch and Hall, 1983; Froehner, 1984; Shear and Bloch, 1985; Walker et al., 1985; Bloch and Morrow, 1989; Sealock et al., 1986; Froehner et al., 1987; Ervasti and Campbell, 1991; Sealock, 1991; Turner et al., 1991; Bewick et al., 1992; Peters et al., 1994), stably anchored to dense aggregates of b1-integrins (Bozyczko et * Corresponding author. Tel.: +1 403 2840903; e-mail: [email protected]

0925-4773/97/$17.00  1997 Elsevier Science Ireland Ltd. All rights reserved PII S0925-4773 (97 )0 0094-4

al., 1989; Anderson et al., 1996) and ‘dystrophin-related proteins’ (Ohlendieck et al., 1991; Ervasti and Campbell, 1993). These complexes resemble the integrin-linked networks of surface proteins that form at adhesion sites generated by other cell types, including the hemidesmosomes of epithelial cells and the focal adhesions established by fibroblasts and platelets (Dustin and Springer, 1991; Romer et al., 1992). Despite an increasing understanding of the biochemical organization of the neuromuscular junction, there is still little direct evidence concerning the molecular mechanisms through which nerve and muscle cells interact to direct the assembly of specialized synaptic sites. Some insight into these mechanisms has, however, been provided by the discovery that the neural signals which control synapse formation can be mimicked with impressive fidelity by interactions with simple, untreated polystyrene microspheres, in the absence of any neural factors (Anderson et al., 1991; Baker et al., 1992). For example, contact with such agents evokes both a redistribution of mobile AChR to sites of bead-muscle contact, and the dispersal of AChR clusters elsewhere on the myocyte surface (see also Peng et al., 1981). The ability to mimic neural inductive signals

126

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139

does not depend upon the beads’ prior acquisition of bioactive substances, and instead correlates with their ability to desorb proteins away from adjacent surfaces (Anderson et al., 1991). This raises the interesting possibility that synapse induction is an enzymatic rather than hormonal process, since developing motor neurites also cause the focal elimination of basal lamina proteins from the adjacent muscle surface at the initial stage of synapse formation, immediately before the onset of postsynaptic differentiation (Anderson, 1986; Anderson et al., 1996). This remodeling of the muscle surface appears to result from local increases in protease activity that occur on the adjacent nerve and muscle surfaces (Champaneria et al., 1992). Taken together, such observations suggest that the focal degradation of basal lamina proteins may act as an inductive signal that causes the muscle cell to assemble an organized postsynaptic apparatus in the affected region of cell surface. For the elimination of muscle basal lamina proteins to serve as an inductive signal, there would need to be stable ligand-receptor complexes present on the muscle surface that can respond to the degradation and/or removal of their extracellular ligands, transducing this change in ligand-occupancy into cytoplasmic signals that direct the assembly of the postsynaptic apparatus. One widely expressed receptor family, the b1-integrins, is already known to form dense aggregates scattered over developing myocytes, some of which are closely associated with the AChR aggregates that form on non-innervated muscle cells and at synapses (Damsky et al., 1985; Bozyczko et al., 1989; Anderson et al., 1996). These integrin aggregates are stably associated with several extracellular matrix (ECM) proteins, including established integrin ligands such as laminins, perlecan and fibronectins (Anderson et al., 1996). Integrins are also known to be highly unconventional ‘receptors’ that can transmit regulatory signals in both directions across plasma membranes, and contribute to stable structural complexes that anchor the cytoskeleton to the adjacent ECM. Through these processes integrins play key roles in the assembly of specialized adhesive junctions in several motile cell types (for reviews see Dustin and Springer, 1991; Ginsberg et al., 1992; Calvete, 1994; Fox, 1994; Faull and Ginsberg, 1995). Since the initial proteolytic degradation of the muscle basal lamina clearly disrupts stable laminin-integrin complexes (Anderson et al., 1996), b1-integrins could help to transduce this proteolytic action into cytoplasmic signals that mediate postsynaptic differentiation. To explore this possibility further, we have examined how the initial enzymatic elimination of muscle basal lamina proteins during synaptogenesis affects the distribution of nearby b1-integrins. Our results show that integrin signaling pathways on the muscle surface respond promptly to the degradation of integrins’ basal lamina ligands, eliminating nearby integrin clusters at the initial stage of synaptogenesis. Mobile b1-integrins then aggregate within the

affected region of cell surface, and acquire dense new aggregates of basal lamina components and other postsynaptic proteins.

2. Results 2.1. Changes in b1-integrin organization during synapse formation In order to determine the time course of integrin accumulation during synaptogenesis, we compared the distribution of b1-integrins with that of the nicotinic AChR. In Xenopus nerve-muscle cultures motor neurites induce a gradual AChR redistribution where mobile receptors initially form small microaggregates, which gradually increase in number and fuse, eventually forming long interrupted bands stretching along the path of cell contact (Anderson et al., 1977; Anderson and Cohen, 1977). This process is closely paralleled by the accumulation of basal lamina HSPG (sometimes called perlecan; see Timpl, 1993) and laminin (Anderson and Fambrough, 1983; Anderson et al., 1984; Anderson et al., 1996), and can be divided into five distinct stages, based upon the size and morphology of the developing postsynaptic complexes. Since these stages correlate closely with increases in the levels of spontaneous synaptic activity and ultrastructural differentiation (Anderson et al., 1979; Kidokoro et al., 1980; Anderson et al., 1984), it is possible to determine the sequence of changes in b1-integrins, simply by comparing their distributions with established postsynaptic markers like HSPG or the AChR on individual cells in culture. Earlier studies have also shown that b1-integrins form dense aggregates scattered over the surfaces of skeletal muscle cells developing in culture, and that AChR aggregates are associated with nearly congruent b1-integrin clusters (Damsky et al., 1985; Anderson et al., 1996). While synaptic AChR clusters in vivo co-distribute with accumulations of b1-integrins (above), it has not been determined when synaptic integrins accumulate, relative to other synaptic markers, or whether they contribute to the assembly of other postsynaptic components. To explore these issues, we first stained 10 Xenopus nerve-muscle cultures with fluorescent derivatives of a-bungarotoxin (aBGT) and a monoclonal mouse anti-b1-integrin (8C8; Gawantka et al., 1992; see Section 4), 1–5 days after adding neural tube cells. These experiments revealed that all AChR aggregates are 1

Previous reports have already determined that this antibody does not cause rearrangements of b1-integrins on any of the cells in these cultures. Indirect labeling of b1-integrins does, however, cause a rapid internalization of the probe on epithelial cells and fibroblasts, but not on myocytes (Anderson et al., 1996). 2 While faint b1-integrin staining with Cy3–8C8 was also seen along some neurites, this was much lower in intensity than the staining on adjacent myocytes. Even if neural b1-integrins were present, therefore, they could not be detected by these methods at sites of nerve-muscle contact.

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139

associated with nearly congruent accumulations of b1-integrins, both on non-innervated myocytes (not shown) and at all stages of synapse development (Fig. 1).1 Synaptic b1-

127

integrin aggregates were often more numerous and slightly larger, but were very similar in shape to the corresponding AChR clusters at all stages of synapse formation (Figs 1 and 2), indicating that the accumulation of b1-integrins closely parallels, and probably slightly precedes that of the AChR.2 In addition to the conspicuous similarities in the sizes and shapes of synaptic b1-integrin and AChR aggregates, there were striking changes in b1-integrin organization along paths of nerve-muscle contact at immature synapses (stages II–IV; Anderson et al., 1984), relative to the remaining muscle surface. At these stages of synapse development small integrin aggregates were routinely interspersed with conspicuous (3–7-mm wide) dark bands, that were devoid of the b1-integrin aggregates scattered over the adjacent muscle surface (Fig. 2). In fact, similar dark bands extended across muscle surfaces on other innervated cells that had not yet developed any detectable synaptic AChR accumulations (Fig. 3), even though many other nerve-contacted myocytes showed no evidence of postsynaptic differentiation or disruption of b1-integrin aggregates near the nerve fibers (see below). Taken together, these observations show that developing motor neurites have two opposing effects on the distribution of b1-integrins on developing myocytes, first causing the dispersal or removal of existing b1-integrin clusters, and then inducing the formation of new b1-integrin aggregates in the developing postsynaptic apparatus. 2.2. Correspondence between b1-integrins and basal lamina proteins during synaptogenesis The sequential elimination and reformation of b1-integrin aggregates at developing synapses closely resembles the removal and replacement previously observed with muscle HSPG and laminin accumulations (Anderson, 1986; Swenarchuk et al., 1990; Anderson et al., 1996). This suggested that there might be a close relationship between the enzymatic actions that eliminate muscle basal lamina proteins, and the analogous changes that occur in the distribution of b1-integrins. We thus examined whether the time course of b1-integrin elimination and replacement was directly related to the analogous changes occurring in the distribution of basal lamina HSPG and laminin. When Xenopus muscle cultures are stained for both b1integrins and HSPG (or laminin), using monoclonal antibodies labeled with contrasting fluorochromes, there is a very close correspondence between the distributions of these

Fig. 1. Accumulations of b1-integrins at and away from developing synaptic sites. This field shows a highly differentiated synaptic site in a nervemuscle culture that had been stained for integrin with Cy3–8C8 (A) and for AChR with biotin-aBGT and Cy2-streptavidin (B). Note the dense accumulations of integrins (A) along the path of the nerve fiber (marked by arrows in the phase contrast image in C), and the congruent aggregates of junctional AChR (B). Note also the abundance of integrin aggregates over the remaining myocyte surface that is now devoid of AChR clusters. The bar in C signifies 20 mm.

128

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139

lamina accumulations do not require the sustained presence of adjacent b1-integrin aggregates. In fact, organized muscle basal laminae remain intact even after the complete degen-

Fig. 2. Distribution of b1-integrins at a less advanced stage of synapse development. This shows a nerve-contacted myocyte from another culture stained as in Fig. 1. Note again the congruent integrin (A) and AChR (B) clusters along the nerve fiber (marked by arrows in C) and the intervening regions of nerve-muscle contact completely devoid of integrin clusters, unlike the remainder of the muscle surface. The bar in C is 20 mm.

basal lamina proteins and b1-integrins on non-innervated myocytes (Anderson et al., 1996; also below). On older myocytes the vast majority of b1-integrin clusters are associated with almost congruent accumulations of HSPG and laminin, even though different basal lamina deposits vary substantially in intensity, and thus in laminin/HSPG sitedensity. Infrequently, discrete accumulations of HSPG and laminin occur without corresponding aggregates of b1integrins in the adjacent sarcolemma; this shows that basal Fig. 3. Distribution of b1-integrins at the earliest stage of synapse development, before the appearance of junctional AChR clusters. The culture was stained as noted in Fig. 1. Note that integrin clusters (A), scattered over the entire muscle surface before nerve contact, have been eliminated along the nerve fiber (marked by arrows in C), before any AChR clusters (B) have formed in the developing postsynaptic apparatus. Instead the muscle surfaces display a few scattered AChR aggregates (B) away from nerve fibers, likewise associated with large integrin aggregates (A). Note also the scattering of small integrin microclusters along the nerve fiber that have not yet acquired AChR aggregates. The bar in C represents 20 mm.

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139

eration of the entire muscle cell in vivo (Sanes et al., 1978). Earlier studies have also shown that motor neurites induce the elaboration of dense postsynaptic accumulations of HSPG and laminin during synaptogenesis. These accumulations are often slightly larger, but are identical in shape to the corresponding junctional AChR aggregates, and undergo similar changes in organization during synapse maturation (Anderson and Fambrough, 1983; Anderson et al., 1984; Anderson et al., 1996). To determine whether these changes in muscle HSPG and laminin distribution are coupled to changes in b1-integrin organization, we examined nerve-muscle cultures, 1–5 days after adding dissociated neural tube cells. These cultures were stained for b1-integrins and HSPG (or laminin), using monoclonal antibodies conjugated with contrasting fluorochromes (see Section 4). Such observations revealed that synaptic integrin accumulations are consistently associated with congruent aggregates of both HSPG (Figs 4 and 5) and laminin (Fig. 6). There were, however, subtle differences in the extent of HSPG (or laminin) and b1-integrin co-distribution within individual aggregates, particularly at early stages of synapse development. Some microclusters of synaptic b1-integrins displayed little or no discernible laminin or HSPG (Figs 5 and 6), and integrin aggregates often appeared slightly larger than the corresponding HSPG/laminin accumulations (Figs 4 and 5). This was similar to the situation at integrin aggregates on non-innervated myocytes (Anderson et al., 1996). At later stages of synaptic HSPG and laminin accumulation, the relative staining intensities changed, with HSPG/laminin intensity significantly exceeding that of adjacent b1-integrin clusters (Figs 4 and 5). Since monoclonal antibodies bind stoichiometrically to unique epitopes on their antigens, these observations indicate that synaptic b1integrin clustering and the focal deposition of HSPG and laminin vary slightly in time course, but are closely coupled processes. In fact, integrin clusters may provide templates for muscle basal lamina assembly, both at and away from synapses. 2.3. Changes in b1-integrin distribution before the onset of postsynaptic differentiation: Since these observations revealed a close association between the accumulation of HSPG/laminin and b1-integrins at developing synapses, we were particularly interested in determining whether the initial proteolytic elimination of muscle HSPG/laminin accumulations (Anderson, 1986; Anderson et al., 1996) was also related to the early disappearance of muscle integrin clusters (above). This issue is of particular importance, because some biochemical action at this initial stage of nerve-muscle interaction provides the inductive signals that evoke subsequent postsynaptic differentiation. We therefore compared the distributions of b1integrins and HSPG (or laminin) at early stages of synapse development, including the initial stage where focal lami-

129

nin/HSPG elimination can be detected, but postsynaptic HSPG/laminin accumulation is not yet visible. On older muscle cultures (maintained 3–5 weeks at 10°C) scattered HSPG/laminin accumulations usually extended over most of the muscle surface, allowing their elimination at developing synapses to be detected readily by fluorescent staining (Anderson, 1986; Anderson et al., 1996). When such cultures were stained for both b1-integrin and HSPG or laminin (as noted above) it became evident that there was a very close association between the distributions of HSPG/ laminin and b1-integrins, where all were eliminated virtually synchronously along paths of nerve-muscle contact (Figs 5 and 6). In fact, many striking examples were found where integrin dispersal had occurred along with the elimination of its basal lamina ligands, before the onset of postsynaptic differentiation (Fig. 7). Previous observations have shown that the focal removal of muscle HSPG and laminin occur at developing neuromuscular junctions, but not at contacts with the non-cholinergic neurites that are also present in these cultures (Anderson, 1986; Anderson et al., 1996). This cellular specificity is significant, because it helps to distinguish the biochemical actions that ultimately lead to synapse formation from other phenomena that may be common to all contacts with growing nerve fibers. It was thus important to note that many sites of nerve-muscle contact in Xenopus nerve-muscle cultures showed no evidence of focal b1integrin, HSPG or laminin elimination (Fig. 8). These observations provide further evidence that the focal elimination of HSPG, laminin and b1-integrin accumulations is restricted to developing synapses. Based upon these observations, it seems reasonable to conclude that the earliest stage of synapse development is a transient phase where cell surface proteases eliminate any HSPG and laminin accumulations that were present before the arrival of the nerve fiber, causing the dispersal or removal of their associated b1-integrin molecules.3 2.4. Redistribution of b1-integrins during synapse development Earlier studies have shown that the nerve-induced accumulations of AChR and HSPG arise at developing junctions through quite different mechanisms. AChR are integral membrane proteins that can migrate, dissociate and aggregate within the plasma membrane (Anderson and Cohen, 1977; Anderson et al., 1984), while HSPG is a basal lamina component that is deposited locally on the cell surface, and then remains immobile (Anderson, 1986). Integrins are also integral membrane proteins, but form relatively stable 3 It was also a formal possibility that the initial absence of 8C8 staining at developing synapses results from a proteolytic damage to its epitope on b1integrin molecules, rather than from the elimination of the affected integrin clusters. This possibility is excluded by later experiments which examine the redistribution of 8C8-integrin complexes at developing synapses.

130

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139

aggregates that are closely associated with immobile basal lamina ligands on the muscle surface (Anderson et al.,

1996). It thus remains unclear whether synaptic integrin accumulation occurs via a redistribution of existing surface molecules and/or by a locally targeted deposition of recently synthesized molecules. These alternatives were examined by determining the fate of the b1-integrin molecules that were already present on the muscle surface before the addition of neural tube cells to the cultures. In these experiments muscle cultures were exposed overnight to biotin-conjugated 8C8 antibody, rinsed, and then placed in a ‘chase’ containing digoxygenin-labeled 8C8, immediately before adding neural tube cells (see Section 4). After the development of synapses (over 4–5 days at 15°C) the cultures were rinsed and then stained overnight with streptavidin and anti-digoxygenin (35/20; see Hunter et al., 1982) labeled with the contrasting dyes, Cy3 and Cy2. This treatment resulted in the differential staining of new and old b1-integrin molecules on each muscle cell. In fact, when young, yolk-laden myotomal myocytes were added to older muscle cultures after the initial antibody treatment, they subsequently displayed staining only with the antidigoxygenin antibody (not shown). When 12 nerve-muscle cultures were examined after such labeling, developing synapses could readily be distinguished by the shapes of their associated b1-integrin aggregates, and by the intervening dark bands devoid of integrin clusters (noted above). From these observations it became clear that developing synaptic accumulations of b1-integrin routinely contained both fluorochromes (Fig. 9), indicating that nerve-induced accumulations of b1-integrins arise, at least in part, via the redistribution and aggregation of integrin molecules that were present on the muscle surface before the establishment of nerve-muscle contact. Unlike earlier results with labeled AChR (Anderson and Cohen, 1977; Anderson et al., 1984), however, there were often subtle but distinct differences in the distributions of the separate labels on individual muscle cells. These were usually difficult to resolve by simple visual inspection, without superimposing the respective fluorescent images (Fig. 10). The presence of adjacent integrin aggregates containing primarily new or old integrin molecules indicates that the bulk of the b1-integrin molecules on these myocytes are organized within stable surface aggregates, where they are not free to equilibrate with integrin molecules present in adjacent regions of plasmalemma. These findings suggest that the control of integrin mobility and aggregation is more complex than that of either the AChR or HSPG, and may involve (a) the redistribution of existing surface molecules, (b) a locally targeted deposition

Fig. 4. Association of b1-integrin clusters with corresponding basal lamina accumulations. This nerve-muscle culture was stained for integrins with Cy3–8C8 (A) and for HSPG with RG-labeled 2AA5, 2BA5 and 2BD5 antibodies (B). Note the virtually congruent accumulations of integrin (A) and HSPG (B) over the entire muscle surface, and the particularly dense aggregates in the differentiated postsynaptic apparatus (marked by arrows in C). The bar in C is 20 mm.

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139

131

Fig. 5. Correspondence between b1-integrin and HSPG accumulations at all stages of synapse development. The culture was stained as noted in Fig. 4. Note again the congruent integrin (A) and HSPG (B) aggregates at a differentiated synaptic site (see large arrows in C), and the elimination of previous clusters at an adjacent nerve contact on the same cell (small arrows in C). Note also the presence of a few small integrin microclusters (A) near the latter nerve contact, without corresponding HSPG aggregates (B). The bar in C represents 20 mm.

132

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139

of new integrin molecules, and (c) local differences in the rates of integrin association, dissociation and turnover. These results do show, however, that synaptic integrin accu-

mulation does not result simply from a nerve induced expression of a new ‘synaptic’ integrin isoform, that is intrinsically targeted to the neuromuscular junction.

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139

In addition, the results shown in Fig. 8 clearly indicate that 8C8 antibody molecules remain stably complexed for several days, even when their associated integrin molecules migrate into and aggregate at developing synapses. The 8C8-binding epitopes on the b1-integrin molecules accumulating at developing synapses are thus not noticeably affected by the proteolytic enzymes that become active at these sites. It is unlikely, therefore, that the adjacent regions devoid of 8C8 staining result simply from proteolytic damage to the 8C8-epitope. Instead, the absence of 8C8 staining at these sites reflects the elimination of integrin clusters that were initially positioned near the paths of nerve contact.

3. Discussion The integrins constitute an extended family of transmembrane proteins that is present on virtually all cell types. They are often called ‘ECM receptors’ because their interactions with ECM-ligands (including laminin, fibronectin, collagen and others) help regulate the adhesion, spreading, migration and differentiation of many cells (for reviews see above). Integrins are not, however, classical ‘receptors’ as this name applies to established hormone, neurotransmitter and cytokine receptors. Even though regulatory roles of integrins remain only partly understood, it is already clear that they depart from the classical receptor paradigm in several ways. It is known, for example, that integrins can establish stable interactions with both their ECM ligands and the cytoplasmic proteins that couple them to the cytoskeleton becoming stable constituents of organized transmembrane complexes that link elaborate webs of ECM proteins to equally elaborate complexes in the adjacent cytoskeleton. In addition to this structural capacity, however, integrins can also function as ‘receptors’ that convey regulatory signals in both directions across plasma membranes (Ginsberg et al., 1992; Humphries et al., 1993). Our current results show that nerve-muscle interaction has remarkable effects on the distribution of b1-integrins during synapse formation, eventually leading to their aggregation within the stable postsynaptic complexes that bind cholinergic nerve terminals to the muscle surface at synapses (see above). Since these changes in muscle b1integrin distribution are restricted to cholinergic nerve contacts that are undergoing synapse formation, it is clear that the signaling pathways which control muscle integrin organization are modulated by the unknown regulatory signals through which motor neurons evoke synaptic formation.

133

These results also reveal that the earliest detectable change in muscle surface organization during synapse formation is the focal elimination of b1-integrin clusters and at least some of their associated basal lamina proteins, which were originally present as organized ligand-complexes scattered along the path of nerve-muscle contact. In fact, this removal of integrin clusters extends throughout the 3–7-mm wide perineural zone of muscle surface within which postsynaptic structures later form. The elimination of organized integrin complexes is transient, however, and soon followed by the reappearance of discrete new integrin-ECM accumulations along the nerve, together with accumulations of postsynaptic components like the AChR. During this redistribution, AChR dissociate from pre-existing integrinECM complexes, while mobile receptors are aggregating at the newly formed integrin-ECM accumulations developing at the site of innervation (Anderson et al., 1977; Anderson and Cohen, 1977; Anderson et al., 1984). While it is not yet obvious what biochemical signals initiate and regulate these striking changes in the distribution of muscle integrins and basal lamina proteins, it is evident that muscle basal lamina complexes are normally very stable (Anderson, 1986), and will even survive the death and degeneration of the entire myofiber (Sanes et al., 1978). In particular, the survival of muscle laminin and HSPG deposits is not dependent upon the presence of adjacent integrin clusters (Anderson et al., 1996). It seems highly unlikely, therefore, that the initial disappearances of muscle HSPG and laminin at developing synapses result simply from cytoplasmic changes that could, in principle, remove muscle integrin clusters. Instead, it is far more likely that the elimination of HSPG and laminin results from focal changes in the activity of pericellular proteases that occur on the adjacent nerve and muscle surfaces (Anderson, 1986; Champaneria et al., 1992; Anderson et al., 1996). It is also clear that the enzymatic degradation of muscle basal lamina proteins is closely coupled, spatially and temporally, to the disappearance of their associated b1-integrin clusters; this strongly suggests that basal lamina degradation affects the cytoplasmic pathways that control the distribution of muscle b1-integrins. Integrin clustering in several motile cell types is known to be controlled by complex processes which employ interactions between a variety of cytoplasmic proteins and protein kinases. This web of molecular interactions forms orderly signaling pathways that modulate (and are modulated by) interactions between integrin molecules and several cytoskeletal-associated proteins (such as talin, vinculin, a-actinin and paxillin), as well as several cytoplasmic enzymes such as protein kinase C,

Fig. 6. Codistribution between b1-integrin and laminin accumulations at developing junctions. Integrins were stained with Cy3–8C8 (A) while laminins were stained with biotin-6BB5 followed by Cy2-streptavidin (B). Note that the entire surface of the muscle cell is covered with congruent accumulations of integrins (A) and laminins (B), both of which have become particularly concentrated in the developing postsynaptic apparatus (marked by large arrows in C). Note also the disappearance of integrin clusters (A) and laminin accumulations (B) along another, less differentiated path of nerve contact on the same cell (see small arrows in C). The bar in C is 20 mm.

134

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139

p125FAK and pp60C-SRC. The actions of these integrin signaling pathways also affect (and are affected by) integrins’ interactions with their ECM ligands (Lipfert et al., 1992; Sastry and Horwitz, 1993; Schaller and Parsons, 1994; Parsons et al., 1994; Richardson and Parsons, 1995). This ela-

borate web of regulatory interactions allows integrin signaling pathways to participate in a variety of different physiological processes, and to differentiate between interactions with soluble ECM ligands and the spatial arrays of ligands presented by organized ECMs (Nurden and Nurden, 1993; Shattil et al., 1994; Diamond and Springer, 1994; Stuiver and O’Toole, 1995). Based upon the known sensitivity of integrin signaling pathways to changes in integrin ligand-occupancy, and even to the cross-linkage of adjacent ligand molecules afforded by an organized ECM, it seems reasonable to conclude that the initial removal of muscle integrin clusters at developing synapses results primarily from the proteolytic actions that degrade and eliminate the adjacent basal lamina. These proteases could modulate integrin signaling pathways in several different ways, most conspicuously by inflicting damage to the ECM that leads to the diffusing away of soluble peptide fragments, thereby disrupting stable integrin ligand-complexes (Anderson et al., 1996). Surface proteases could also act on integrin molecules directly, or modulate integrin signaling pathways indirectly via nearby protease receptors (Coughlin et al., 1992; Coughlin, 1994) or by releasing cytokines sequestered in the ECM (Vlodavsky et al., 1991; Clark and Brugge, 1995). In this manner integrin signaling pathways in skeletal myofibers could be modulated by the proteolytic enzymes that remodel the ECM, causing first a local elimination of integrin clusters, and then the formation of new integrinbased complexes of postsynaptic proteins. In fact, since laminin and HSPG bind directly to plasminogen and plasminogen activators (Stephens et al., 1992; Moser et al., 1993), such processes would allow the fibrinolytic protease cascade to generate localized signals that modulate integrin signaling pathways in skeletal myofibers, and possibly in other cells. This would have interesting implications, since related integrin signaling pathways in platelets are already known to be controlled indirectly by the fibrinogenic protease cascade acting through thrombin receptors, leading to the formation integrin-packed platelet adhesion sites (Coughlin et al., 1992; Hynes, 1992; Coughlin, 1994). Furthermore, other platelet integrins may be modulated by proteases bound to the surfaces of adjacent leucocytes (Pidard et al., 1994). There is, in fact, another parallel between the processes that control synapse differentiation on myofibers, and those that regulate the assembly of analogous integrin-based adhesion sites of platelets. Postsynaptic differentiation can Fig. 7. Elimination of b1-integrin-laminin accumulations before the onset of postsynaptic differentiation. Integrin (A) and laminin (B) were stained as noted in Fig. 6. Note that integrin (A) and laminin (B) accumulations have both been eliminated along the entire path of nerve-muscle contact (shown by arrows in C) before the appearance of new integrin and laminin complexes at the developing synapse. Such observations strongly suggest that integrin signaling pathways are modulated by the proteolytic elimination of integrins’ ECM ligands, immediately before the start of postsynaptic differentiation. The bar in C represents 20 mm.

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139

also be induced by direct interactions between regenerating myofibers and the specialized ECM present at neuromuscular junctions (Burden et al., 1980), much as platelet adhesion sites can be induced by contact with suitable ECM complexes (Nurden and Nurden, 1993; Fox, 1994; Faull and Ginsberg, 1995). Since the formation of synapses and platelet adhesion sites both involve the accumulation of organized integrin-based complexes of ECM and cytoskeletal proteins, these similarities strongly suggest that de novo synapse formation is an analogous integrin-mediated process, controlled indirectly by proteolytic interactions between cell surface enzymes.

135

developed conspicuous widespread accumulations of ECM. These muscle cultures had been allowed to develop for 2–4 weeks at 10°C before the addition of neural tube cells (see above; Anderson, 1986; Anderson et al., 1996). Previous analysis has already established the percentages of innervated muscle cells in such cultures, and the close correlation between the extent of synaptic differentiation and the local

4. Experimental procedures 4.1. Preparation of cultures Procedures for the preparation of cultures of dissociated myotomal and neural tube cells from Xenopus laevis embryos have been described earlier (Anderson et al., 1977; Anderson et al., 1984). In brief, myotomes and neural tubes of early tail-bud stage embryos were isolated by sterile microdissection in the presence of 0.3–1.0 mg/ml collagenase (Type IV, Sigma Chemical Co., St. Louis, MO). Separated myotomes were pooled, exposed to trypsin-EGTA for 60 min, and then placed for a similar period in a ‘plating medium’ of 60% v/v Dulbecco’s Modified Eagle’s Medium (DMEM; Gibco Canada Inc., Burlington, Ontario) supplemented with 8 mM HEPES buffer (pH 7.6) and 5% v/v fetal bovine serum (FBS, Gibco). Dissociated cells were plated in sealed glass/polycarbonate culture chambers (Anderson et al., 1977) on films of rat-tail collagen that had been pretreated for 30 min with 10 mg/ml bovine serum fibronectin (Sigma). After 1 day at room temperature (20–22°C) the serum concentration was reduced to 0.5% v/v. Muscle cultures were then maintained for up to 5 weeks at 10°C before the addition of neural tube cells. To generate neuromuscular junctions in culture, neural tube cells (dissociated as above) were added to established muscle cultures, and incubated for 1–5 days at 21°C or for 2–7 days at 15°C in the presence of 10 mm d-tubocurarine (Sigma) to block muscle twitching after synapse development. Some cultures were fixed for 10 min at −10°C with 37–50 mg/ml formaldehyde in a solution containing 70% v/v TR, 10% ethanol, and 10% dimethylsulfoxide. Fixed cultures were permeabilized for 10 min with 0.1% w/v Triton X100 in TR (above) and stored for up to 2 days at 4°C in TR containing 5–15% FBS and 0.1% w/v sodium azide. In order to best display the earliest changes in surface organization during synaptogenesis, which involve the focal elimination of muscle ECM proteins and their b1integrin receptors along the path of nerve muscle contact, this study concentrated upon an examination of more than 35 examples in which growing nerve fibers were allowed to innervate ‘older’ muscle cultures, where most myocytes had

Fig. 8. Absence of laminin and b1-integrin remodeling at other sites of nerve-muscle contact, not undergoing synapse formation. This culture was stained for b1-integrins (A) and laminins (B) as described in Fig. 6. Note the lack of integrin (A) or laminin (B) elimination along the path of the nerve fibers marked by arrows in the phase contrast view shown in C. Earlier studies have demonstrated that 30–40% of the nerve-contacted myocytes in this system show no evidence of cholinergic innervation, and do not evoke focal accumulation of postsynaptic proteins (Kidokoro et al., 1980; Anderson et al., 1984, 1996; Anderson, 1986). Examples such as these indicate that the elimination and replacement of muscle integrinlaminin accumulations are not simply routine consequences of nerve-muscle contact, but are instead restricted to developing synaptic sites. The bar in C signifies 20 mm.

136

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139

accumulation of basal lamina HSPG (Kidokoro et al., 1980; Anderson et al., 1984). The present study extends these earlier observations by showing that there is a precise correspondence between the changes in HSPG/laminin organization and that of b1-integrins on all (100%) of muscle cells in such cultures.

4.2. Preparation and derivatization of probes This study has used several monoclonal antibodies to label b1-integrins and basal lamina components at developing amphibian neuromuscular junctions. The preparation and characterization of mouse monoclonal antibodies to Xenopus basal lamina heparan sulfate proteoglycan (HSPG) have been reported earlier (Anderson and Fambrough, 1983). Xenopus b1-integrins were labeled by IgG produced by the 8C8 hybridoma line, generously provided by Dr. Peter Hausen (Gawantka et al., 1992). Xenopus laminins were stained with monoclonal antibodies 5AD3 or 6BB5 (Anderson et al., 1996). In each case the hybridoma lines had been cloned in soft agar, and propagated as ascitic tumors in BALB/c mice. IgG fractions were isolated from ascitic fluids by ammonium sulfate precipitation and ionexchange chromatography on DEAE-cellulose. Purified immunoglobulins were stored frozen at −70°C, and then derivatized with fluorescent dyes, biotin or digoxygenin. Protein concentrations were determined using the Folin reagent. Procedures for labeling monoclonal mouse IgG with tetramethylrhodamine isothiocyanate (TRITC) and fluorescein isothiocyanate (FITC) have been described previously (Anderson and Fambrough, 1983). Monoclonal antibodies and streptavidin were derivatized similarly, using NHSCy3, NHS-Cy2 (Research Organics, Inc., Cleveland, OH), NHS-biotin, NHS-digoxygenin, and NHS-Rhodamine Green (RG; all from Molecular Probes, Eugene, OR) at pH 8.5 in 0.1 M sodium bicarbonate buffer, or for 60 min (with NHS-Cy3 and NHS-Cy2) in pH 9.5 buffer. RG derivatives were subsequently exposed overnight at 4°C to 0.1 M hydroxlyamine to remove protective groups from their primary amines, before isolation by gel filtration in TR. Labeled probes were usually supplemented with 1 mg/ml bovine serum albumin (BSA) and 10% w/v glycerol and stored in small aliquots at −70°C. 4.3. Fluorescent staining of cultured cells Muscle AChR were labeled for 4 h (or overnight) at 15°C with TRITC- or biotin-labeled aBGT (Molecular Probes), sometimes followed by 4 h in 1–4 mg/ml FITC-, Cy2-, or RG-labeled streptavidin. To observe the distributions of b1-

Fig. 9. Redistribution and aggregation of mobile b1-integrin molecules during synaptogenesis. Surface integrin molecules on non-innervated myocytes were labeled with biotin-8C8, before the addition of dissociated nerve cells. Integrin molecules subsequently exposed on the muscle surface were then labeled by the continuous presence of digoxygenin-labeled 8C8 antibody. After synapse development, the old (A) and new (B) integrin molecules were differentially labeled with Cy3-streptavidin and Cy2–38/20 (anti-digoxygenin) respectively. Note that both old and new integrin molecules have accumulated along the path of nerve-muscle contact (see arrows in C), showing that the motor neurite induces a migration and aggregation of integrin molecules that were already on the muscle surface at the time of nerve contact. The bar in C represents 20 mm.

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139

integrin, laminin and HSPG on living cells, cultures were usually incubated overnight in culture medium at 15°C with 1–4 mg/ml of directly labeled monoclonal antibody. In some cases they were exposed first (as above) to digoxygenin- or biotin-derivatized monoclonal antibody. After six rinses over .30 min, this was followed by a .4-h exposure at 15°C to 5 mg/ml dye-labeled anti-digoxygenin IgG (from mouse hybridoma 35/20) (Hunter et al., 1982) or 1–4 mg/ ml fluorescent streptavidin. In control experiments, monoclonal antibodies were replaced by JIE7, another monoclonal IgG reactive with b-galactosidase from E. coli (Mason et al., 1987). Stock antibody solutions were thawed ,3 days before use, diluted in HEPES buffered 60% v/v DMEM (or TR), containing 1% v/v FBS (usually with 10 mM d-tubocurarine), centrifuged to remove particulates, and then sterilized by passage through 0.2-mm pore membrane filters. In some experiments muscle cultures were first labeled by overnight incubation (at 10°C) with 2–4 mg/ml biotin-conjugated 8C8 antibody to saturate existing b1-integrins on the

137

myocyte surfaces. After rinsing, the cells were then placed in a 1-mg/ml chase of digoxygenin-8C8, and freshly dissociated neural tube cells were then added (as noted above). After 3–5 days of nerve growth and synapse development (at 15°C), the cultures were rinsed again and then exposed overnight to a mixture of 2 mg/ml Cy3-streptavidin and 4 mg/ ml Cy2-labeled anti-digoxygenin (35/20; see above), to stain new and old integrin molecules on myocytes with contrasting fluorochromes. The cultures were then examined to determine whether motor neurites induce a redistribution of mobile b1-integrin molecules during synapse formation. 4.4. Fluorescence microscopy Stained cultures were usually examined alive to optimize visualization of nerve fibers in phase contrast optics. Since DMEM contains fluorescent components visible in FITC/ RG/Cy2 optics, living cultures were mounted in HEPES-

Fig. 10. Different distributions of old and new b1-integrin molecules on a non-innervated myocyte. Staining conditions were similar to those described in Fig. 9. Note that old (A) and new (B) integrin molecules displayed distinctly different distributions on many cells, but that this was often difficult to discern without superimposing the separate images (C). The red component of the image in C derives from that in A, while the green is provided by the image in B. D shows a phase contrast view of the cell. Such observations indicate that most clustered integrin molecules remain relatively immobile in stable complexes, and do not equilibrate (here over 5 days) with molecules in adjacent areas of the myocyte surface. The bar in D is 20 mm.

138

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139

buffered DMEM salts (without amino acids, indicator or vitamins) containing 10 mM d-tubocurarine. Fixed cultures were mounted in TR supplemented with 20 mM Na2CO3, 20–30 mM p-phenylene diamine, and 30–80 mM NaHSO3 (pH 9.5). Such conditions enhanced FITC fluorescence, and significantly slowed the rate of photobleaching for all probes. Fluorescently stained cultures were examined on Zeiss WL or Axioskop microscopes equipped for incident illumination with narrow-band filter-sets selective for FITC/RG/ Cy2 and TRITC/Cy3, provided by Carl Zeiss Canada Ltd. (Mississauga, Ontario) or Omega Optical Co. (Brattleborough, VT). These filters prevented bleed-through of TRITC and Cy3 signals into the FITC/RG/Cy2 channel. Micrographs were recorded on 35-mm Ilford HP-5 film that was developed for 15 min at 20°C in Ilford Microphen. Exposure times for fluorescence micrographs were from 5 to 60 s. Micrograph negatives were digitized using a Nikon Coolscan scanner, and then assembled for publication using Adobe Photoshop 3.0 software.

Acknowledgements We are indebted to Ms. Christine Renz and Mrs. Chandra Prasad for excellent technical assistance, and to Dr. J.B. Rattner for critical evaluation of the manuscript. We are also grateful to Drs. P. Hausen and M. Margolies for generous donations of hybridoma lines. A brief account of some of these findings was presented at the annual meeting of the American Society for Cell Biology (Anderson and Zackson, 1994).

References Anderson, M.J. (1986) Nerve-induced remodeling of muscle basal lamina during synaptogenesis. J. Cell Biol. 102, 863–877. Anderson, M.J. and Cohen, M.W. (1977) Nerve-induced and spontaneous redistribution of acetylcholine receptors on cultured muscle cells. J. Physiol. (Lond.) 268, 757–773. Anderson, M.J. and Fambrough, D.M. (1983) Aggregates of acetylcholine receptors are associated with plaques of a basal lamina heparan sulfate proteoglycan on the surface of skeletal muscle fibers. J. Cell Biol. 97, 1396–1411. Anderson, M.J. and Zackson, S.L. (1994) Nerve induced disruption and reformation of b1-integrin clusters at developing neuromuscular junctions. Mol. Biol. Cell 5, A65. Anderson, M.J., Cohen, M.W. and Zorychta, E. (1977) Effects of innervation on the distribution of acetylcholine receptors on cultured amphibian muscle cells. J. Physiol. (Lond.) 268, 731–756. Anderson, M.J., Kidokoro, Y. and Gruener, R. (1979) Correlation between acetylcholine receptor localization and spontaneous synaptic potentials in cultures of nerve and muscle. Brain Res. 166, 185–190. Anderson, M.J., Klier, F.G. and Tanguay, K.E. (1984) Acetylcholine receptor aggregation parallels the deposition of a basal lamina proteoglycan during development of the neuromuscular junction. J. Cell Biol. 99, 1769–1784. Anderson, M.J., Champaneria, S. and Swenarchuk, L.E. (1991) Synaptic

differentiation can be evoked by polymer microbeads that mimic localized pericellular proteolysis by removing proteins from adjacent surfaces. Dev. Biol. 147, 464–479. Anderson, M.J., Shi, Z.Q. and Zackson, S.L. (1996) Proteolytic disruption of laminin-integrin complexes on muscle cells during synapse formation. Mol. Cell. Biol. 16, 4972–4984. Baker, L.P., Chen, Q. and Peng, H.B. (1992) Induction of acetylcholine receptor clustering by native polystyrene beads. Implication of an endogenous muscle-derived signalling system. J. Cell Sci. 102, 543–555. Bewick, G.S., Nicholson, L.V.B., Young, C., O’Donnel, E. and Slater, C.R. (1992) Different distributions of dystrophin and related proteins at nerve-muscle junctions. Neuroreport 3, 857–860. Bloch, R.J. and Hall, Z.W. (1983) Cytoskeletal components of the vertebrate neuromuscular junction: vinculin, alpha-actinin, and filamin. J. Cell Biol. 97, 217–223. Bloch, R.J. and Morrow, J.S. (1989) An unusual beta-spectrin associated with clustered acetylcholine receptors. J. Cell Biol. 108, 481–493. Bozyczko, E., Decker, C., Muschler, J. and Horwitz, A.F. (1989) Integrin on developing and adult skeletal muscle. Exp. Cell Res. 182, 72–91. Burden, S.J., Sargent, P.B. McMahan, U.J. (1979) Acetylcholine receptors in regenerating muscle accumulate at original synaptic sites in the absence of the nerve. J. Cell Biol. 82, 412–425. Calvete, J.J. (1994) Clues for understanding the structure and function of a prototypic human integrin: the platelet glycoprotein IIb/IIIa complex. Thromb. Haemost. 72, 1–15. Champaneria, S., Swenarchuk, L.E. and Anderson, M.J. (1992) Increases in pericellular proteolysis at developing neuromuscular junctions in culture. Dev. Biol. 149, 261–277. Cifuentes-Diaz, C., Nicolet, M., Goudou, D., Rieger, F. and Mege, R.M. (1994) N-cadherin expression in developing, adult and denervated chicken neuromuscular system: accumulations at both the neuromuscular junction and the node of Ranvier. Development 120, 1–11. Clark, E.A. and Brugge, J.S. (1995) Integrins and signal transduction pathways: the road taken. Science 268, 233–239. Coughlin, S.R. (1994) Molecular mechanisms of thrombin signalling. Sem. Hematol. 31, 270–277. Coughlin, S.R., Scarborough, R.M., Vu, T.K. and Hung, DT. (1992) Thrombin receptor structure and function. Cold Spring Harbor Symp. Quant. Biol. 57, 149–154. Damsky, C.H., Knudsen, K.A., Bradley, D., Buck, C.A. and Horwitz, A.F. (1985) Distribution of cell substratum attachment (CSAT) antigen on myogenic and fibroblastic cells in culture. J. Cell Biol. 100, 1528–1539. Diamond, M.S. and Springer, T.A. (1994) The dynamic regulation of integrin adhesiveness. Curr. Biol. 4, 506–517. Dustin, M.L. and Springer, T.A. (1991) Role of lymphocyte adhesion receptors in transient interactions and cell locomotion. Ann. Rev. Immunol. 9, 27–66. Ervasti, J.M. and Campbell, K.P. (1991) Membrane organization of the dystrophin-glycoprotein complex. Cell 66, 1121–1131. Ervasti, J.M., Campbell, K.P. (1993) A role for the dystrophin-glycoprotein complex as a transmembrane linker between laminin and actin. J. Cell Biol. 122, 809–823. Faull, R.J. and Ginsberg, M.H. (1995) Dynamic regulation of integrins. Stem Cells 13, 38–46. Fox, J.E. (1994) Transmembrane signalling across the platelet integrin glycoprotein IIb-IIIa. Ann. N.Y. Acad. Sci. 714, 75–87. Froehner, S.C. (1984) Peripheral proteins of postsynaptic membranes from Torpedo electric organ identified with monoclonal antibodies. J. Cell Biol. 99, 88–96. Froehner, S.C., Gulbrandsen, V., Hyman, C., Jeng, A.Y., Neubig, R.R. and Cohen, J.B. (1981) Immunofluorescence localization at the mammalian neuromuscular junction of the Mr 43 000 protein of Torpedo postsynaptic membranes. Proc. Natl. Acad. Sci. USA 78, 5230–5234. Froehner, S.C., Murnane, A.A., Tobler, M., Peng, H.B. and Sealock, R. (1987) A postsynaptic Mr 58 000 (58K) protein concentrated at acetylcholine receptor-rich sites in Torpedo electroplaques and skeletal muscle. J. Cell Biol. 104, 1633–1646.

M. J. Anderson et al. / Mechanisms of Development 67 (1997) 125–139 Gawantka, V., Ellinger-Ziegelbauer, H. and Hausen, P. (1992) b1-integrin is a maternal protein that is inserted into all newly formed plasma membranes during early Xenopus embryogenesis. Development 115, 595–605. Ginsberg, M.H., Du, X. and Plow, E.F. (1992) Inside-out integrin signalling. Curr. Opin. Cell Biol. 4, 766–771. Humphries, M.J., Mould, A.P. and Tuckwell, D.S. (1993) Dynamic aspects of adhesion receptor function: integrins both twist and shout. Bioessays 15, 391–397. Hunter, M.M., Margolies, M.N., Ju, A. and Haber, E. (1982) High-affinity monoclonal antibodies to the cardiac glycoside, digoxin. J. Immunol. 129, 1165–1172. Hunter, D.D., Shah, V., Merlie, J.P. and Sanes, J.R. (1989) A laminin-like adhesive protein concentrated in the synaptic cleft of the neuromuscular junction. Nature 338, 229–234. Hynes, R.O. (1992) Integrins: versatility, modulation and signalling in cell adhesion. Cell 69, 11–25. Kidokoro, Y., Anderson, M.J. and Gruener, R. (1980) Changes in synaptic potential properties during acetylcholine receptor accumulation and neurospecific interactions in Xenopus nerve-muscle cell culture. Dev. Biol. 78, 464–483. Kramarcy, N.R., Vidal, A., Froehner, S.C. and Sealock, R. (1994) Association of utrophin and multiple dystrophin short forms with the mammalian M(r) 58 000 dystrophin-associated protein (syntrophin). J. Biol. Chem. 269, 2870–2876. Lipfert, L., Haimovich, B., Schaller, M.D., Cob, B.S., Parsons, J.T. and Brugge, J.S. (1992) Integrin-dependent phosphorylation and activation of the protein tyrosine kinase pp125FAK in platelets. J. Cell Biol. 119, 905–912. Mason, P.W., McAda, P.C., Dalrymple, J.M., Fournier, M.J. and Mason, T.L. (1987) Expression of Japanese encephalitis virus antigens in Escherichia coli. Virology 158, 361–372. Moody-Corbett, F. and Cohen, M.W. (1981) Localization of cholinesterase at sites of high acetylcholine receptor density on embryonic amphibian muscle cells cultured without nerve. J. Neurosci. 1, 596–605. Moser, T.L., Enghild, J.J., Pizzo, S.V. and Stack, M.S. (1993) The extracellular matrix proteins laminin and fibronectin contain binding domains for human plasminogen and tissue plasminogen activator. J. Biol. Chem. 268, 18917–18923. Nurden, A.T. and Nurden, P. (1993) A review of the role of platelet membrane glycoproteins in the platelet-vessel wall interaction. Bailliere’s Clin. Haematol. 6, 653–690. Ohlendieck, K., Ervasti, J.M., Matsumura, K., Dahl, S.D., Leveille, C.J. and Campbell, K.P. (1991) Dystrophin-related protein is localized to neuromuscular junctions of adult skeletal muscle. Neuron 7, 499–508. Parsons, J.T., Schaller, M.D., Hildebrand, J., Leu, T.H., Richardson, A. and Otey, C. (1994) Focal adhesion kinase: structure and signalling. J. Cell Sci. 18 (Suppl.), 109–113. Parthasarathy, N. and Tanzer, M.L. (1987) Isolation and characterization of a low molecular weight chondroitin sulfate proteoglycan from rabbit skeletal muscle. Biochemistry 26, 3149–3156. Peng, H.B., Cheng, P.C. and Luther, P.W. (1981) Formation of ACh receptor clusters induced by positively charged beads. Nature (Lond.) 292, 831–838. Peters, M.F., Kramarcy, N.R., Sealock, R. and Froehner, S.C. (1994) Beta 2-Syntrophin: localization at the neuromuscular junction in skeletal muscle. Neuroreport 5, 1577–1580. Pidard, D., Renesto, P., Rabhi, S. and Chignard, M. (1994) Regulation of the structure and activity of platelet adhesion receptors by leucocyte proteinases. Nouv. Rev. Fr. Hematol. 36, S99–S101.

139

Reist, N.E., Magill, C. and McMahan, U.J. (1987) Agrin-like molecules at synaptic sites in normal, denervated, and damaged skeletal muscles. J. Cell Biol. 105, 2457–2469. Richardson, A. and Parsons, J.T. (1995) Signal transduction through integrins: a central role for focal adhesion kinase? Bioessays 17, 229–236. Romer, L.H., Burridge, K. and Turner, C.E. (1992) Signalling between the extracellular matrix and the cytoskeleton: tyrosine phosphorylation and focal adhesion assembly. Cold Spring Harbor Symp. Quant. Biol. 57, 193–202. Sanes, J.R. (1982) Laminin, fibronectin, and collagen in synaptic and extrasynaptic portions of muscle fiber basement membrane. J. Cell Biol. 93, 442–451. Sanes, J.R., Marshall, L.M. and McMahan, U.J. (1978) Reinnervation of muscle fiber basal lamina after removal of myofibers. J. Cell Biol. 78, 176–198. Sanes, J.R., Schachner, M. and Covault, J. (1986) Expression of several adhesive macromolecules (N-CAM, L1, J1, NILE, uvomorulin, laminin, fibronectin, and a heparan sulfate proteoglycan in embryonic, adult, and denervated adult skeletal muscle. J. Cell Biol. 102, 420–431. Sastry, S.K. and Horwitz, A.F. (1993) Integrin cytoplasmic domains: mediators of cytoskeletal linkages and extra- and intracellular initiated transmembrane signaling. Curr. Opin. Cell Biol. 5, 819–831. Sealock, R. (1991) Localization of dystrophin relative to acetylcholine receptor domains in electric tissue and adult and cultured skeletal muscle. J. Cell Biol. 113, 1133–1144. Sealock, R., Paschal, B., Beckerle, M. and Burridge, K. (1986) Talin is a post-synaptic component of the rat neuromuscular junction. Exp. Cell Res. 163, 143–150. Schaller, M.D. and Parsons, J.T. (1994) Focal adhesion kinase and associated proteins. Curr. Opin. Cell Biol. 6, 705–710. Shattil, S.J., Ginsberg, M.H. and Brugge, J.S. (1994) Adhesive signalling in platelets. Curr. Opin. Cell Biol. 6, 695–704. Shear, C.R. and Bloch, R.J. (1985) Vinculin in subsarcolemmal densities in chicken skeletal muscle: localization and relationship to intracellular and extracellular structures. J. Cell Biol. 101, 240–256. Stephens, R.W., Aumailley, M., Timpl, R., Reisberg, T., Tapiovaara, H., Myohanen, H., Murphy-Ullrich, J. and Vaheri, A. (1992) Urokinase binding to laminin-nidogen. Structural requirements and interactions with heparin. Eur. J. Biochem. 207, 937–942. Stuiver, I. and O’Toole, T.E. (1995) Regulation of integrin function and cellular adhesion. Stem Cells 13, 250–262. Swenarchuk, L.E., Champaneria, S. and Anderson, M.J. (1990) Induction of a specialized muscle basal lamina at chimaeric synapses in culture. Development 110, 51–61. Timpl, R. (1993) Proteoglycans of basement membranes. Experentia 49, 417–428. Turner, C.E., Kramarcy, N., Sealock, R. and Burridge, K. (1991) Localization of paxillin, a focal adhesion protein, to smooth muscle dense plaques, and the myotendinous and neuromuscular junctions of skeletal muscle. Exp. Cell Res. 192, 651–655. Vlodavsky, I., Bashkin, P., Ishai-Michaeli, R., Chajek-Shaul, T., Bar-Shavit, R., Haimovitz-Friedman, A., Klagsbrun, M. and Fuks, Z. (1991) Sequestration and release of basic fibroblast growth factor. Ann. N.Y. Acad. Sci. 638, 207–220. Walker, J.H., Boustead, C.M., Witzemann, V., Shaw, G., Weber, K. and Osborn, M. (1985) Cytoskeletal proteins at the cholinergic synapse: distribution of desmin, actin, fodrin, neurofilaments, and tubulin in Torpedo electric organ. Eur. J. Cell Biol. 38, 123–133.