Cell, Vol. 70, 1-3, July 10, 1992, Copyright
0 1992 by Cell Press
How Many Agrins Does It Take to Make a Synapse?
Minireview
Michael J. Ferns and Zach W. Hall Department of Physiology School of Medicine University of California, San Francisco San Francisco, California 94143-0444
Synaptogenesis occurs through an extended interplay between ceils-initiated by the first contact of an axonal growth cone with the surface of a postsynaptic cell (“the protoplasmic kiss” of Cajal) and culminating in the establishment of a mature synapse (“the final ecstasy of an epic love story”). The initial interaction between the pre- and postsynaptic cells triggers a series of events that locally alters the structures of both cells to form a stable synapse, which is then maintained over the lifetime of the organism and, in some cases, can be reformed efficiently after injury. At the neuromuscular junction, where the process has been best studied, the first discernible step in synaptic specialization is the clustering of acetylcholine receptors (AChRs) in the postsynaptic membrane underneath the nerve terminal. Experiments with cultured cells almost 15 years ago demonstrated that these clusters are induced by the nerve and arise in part by the redistribution of AChRs on the muscle cell surface (Anderson and Cohen, 1977; Frank and Fischbach, 1979). Later experiments have shown that increased local synthesis of the AChR also contributes to the nerve-induced clusters. Although the molecular mechanisms responsible for the redistribution of AChRs in the postsynaptic membrane are still unknown, recent experiments identify a clustering agent that is secreted by the nerve and offer tantalizing clues about how it works. Agrin, the clustering agent, was discovered after experiments on the frog neuromuscular junction had demonstrated that the synaptic basal lamina contains a factor that directs AChR clustering in regenerating muscle. When muscle fibers are damaged, the fusion of mono-
nerve terminal
Figure 1. A Diagrammatic Scheme for the Role of Agrin in the Neurally Induced Clustering of AChRs in the Muscle Fiber Membrane
nucleated cells causes new fibers to form within the old basal lamina sheath. In such fibers, AChRs accumulate at the old synaptic sites even when the nerve and its associated Schwann cell are absent, suggesting that the synaptic basal lamina contains molecules that cluster AChRs in the membrane of the newly formed muscle fibers (Burden et al., 1979). To identify these molecules, McMahan and colleagues purified from the Torpedo electric organ a protein, agrin, that clusters AChRs in cultured chick myotubes (Godfrey et al., 1984; Nitkin et al., 1987). Further, they showed that an immunologically similar protein is concentrated in the basal lamina at adult frog and chick endplates (Reist et al., 1987). Agrin is thus likely to be responsible for the clustering of AChRs in regenerating muscle and perhaps for the maintenance of clusters at adult synapses. The biological activity and synaptic location of agrin suggested that it might also be the agent that clusters AChRs at developing synapses. lmmunocytochemical and bioactivity assays have shown that agrin, or a protein related to it, is present in embryonic motoneurons and is transported to nerve terminals (Magill-Sole and McMahan, 1988, 1990). On the basis of these and other results, McMahan (1990) proposed that, during development, agrin is released by motor nerve terminals, becomes stably incorporated into the synaptic basal lamina, and induces the initial synaptic clustering of AChRs in muscle fibers (Figure 1). Acceptance of the “agrin hypothesis,” however, has been complicated by the finding that muscles and other tissues also contain proteins that are immunologically related to agrin and that have at least some AChR clustering activity (Godfrey, 1991). Agrin immunoreactivity is present in muscle prior to innervation and is colocalized with the few AChR clusters that appear in developing muscle when it is deprived of innervation (Godfrey et al., 1988; Fallon and Gelfman, 1989). These observations raise the possibility that the original clusters might be induced by musclederived agrin in a process that is directed or regulated by the nerve. Two recent cell culture experiments now make clear the importance of agrin in inducing AChR clusters at newly formed synapses, and, moreover, show that the responsible agrin is neurally derived. The most decisive result is the demonstration that antibodies specific for nerve-derived agrin block the formation of AChR clusters at neuromuscular contacts in interspecific cultures (Reist et al., 1992). When chick motor neurons were cultured with chick or rat muscles, AChR cluster formation at synaptic sites was blocked by an antibody to Torpedo agrin that appears to recognize chick but not rat agrin. In contrast, the antibody did not affect the formation of clusters on chick muscles induced by rat motor neurons. In acomplementaryfinding, another interspecific culture system (Xenopus and Rana) was used to show that nerve-derived agrin is associated with synaptic AChR clusters from the time of their first appearance, less than 2 hr after first nerve contact (Cohen and Godfrey, 1992). AChR clusters that are not associated with the nerve, however, have little or no agrin associated
Cdl 2
with them. Thus, neural rather than muscle agrin appears to be critical for inducing AChR clusters at newly formed synapses in culture. The function of agrin proteins in muscle and in other nonneural tissues remains obscure. A puzzling observation is that agrin in these tissues appears to be much less active in clustering AChRs than agrin derived from neural tissue (Godfrey, 1991). Recent experiments indicate that these differences may result from the expression of alternatively spliced forms of agrin that differ in biological activity. The recent cloning of rat and chick agrin cDNAs (Rupp et al., 1991; Tsim et al., 1992) shows that the encoded proteins are large (- 200 kd) with repeated domains that are homologous to domains in protease inhibitors, epidermal growth factor, and laminin (Figure 2). Analysisof chick and rat cDNAs reveals several sites of alternate RNA splicing (Tsim et al., 1992; Ruegg et al., 1992; Rupp et al., 1991,1992). Chick agrin has two short inserts of 4 and 11 aa near the C-terminus (Figure 2). In the rat, two contiguous inserts of 8 and 11 aa generate four different forms at the site of the 11 aa insert in the chick. All these inserts apparently affect the biological activity of the agrin forms (see below). When constructs comprising the C-terminal half of the chick protein were expressed in transfected cells, the secreted, soluble proteins that resulted were assayed on chick myotubes (Tsim et al., 1992; Ruegg et al., 1992), and only those forms containing both the 4 and 11 aa inserts were active. For the rat, fulllength constructs of splicing variants were expressed in transfected cells and assayed by coculture with myotubes from several sources (Campanelli et al., 1991; Ferns et al., 1992). In this case, although all forms were active on rat primary and C2 muscle cell line mpotubes, only those containing the 8 aa insert were active on a C2 variant with
defective proteoglycans. When assayed on chick myotubes, rat variant forms with the 8 aa insert were most active, while those with the 11 aa insert alone conferred weak activity. It is difficult to assess the exact relationship between inserts in the two species, because further splicing variants may be discovered (see legend to Figure 2) and because of the differences in assay conditions. Both studies clearly demonstrate, however, that the AChR clustering activity of agrin can be regulated by alternative RNA splicing, with the inclusion or exclusion of very small regions having dramatic effects on biological activity. Expression of the alternatively spliced agrin forms in the chick appears to be cell and tissue specific. Purified motor neurons express chiefly the most biologically active form (containing both 4 and 11 aa inserts) (Tsim et al., 1992). In contrast, embryonic muscle expresses forms;fhat lack the 11 aa insert and are devoid of AChR clustering activity in the assay for chick forms described above (Ruegg et al., 1992). The observed expression of the most active form by the nerve fits well with the agrin hypothesis but leaves open the question of function for the less active forms in muscle. One interesting observation is that the addition of exogenous Torpedo agrin (splicing forms unknown) to chick muscle causes the aggregation not only of AChRs but also of muscle agrin (Lieth et al., 1992); muscle agrin also accumulates at nerve-induced clusters. Exogenous agrin induces the accumulation of other synaptic basal lamina components, such as acetylcholinesterase and heparan sulfate proteogl$an, apparently by a mechanism that does not involve redistribution of preexisting molecules but requires new protein synthesis (Wallace, 1989). Neural agrin may thus act catalytically, inducing the accumulation of a larger amount of muscle agrin. The muscle agrin could augment the clustering activity of neural agrin
Rat
12345676
Chick aarin
cl
Figure 2. A Schematic Chick Agrin Molecules
Protease Inhilzilor-like domains Laminin-like domains
Ea
EGF-like domains
q
put&e
t ,200
signal sequence
Potential N-linked glycosylation sites aa ,
Diagram
of Rat
and
The diagram for rat agrin is based on those found in Rupp et al. (1991, 1992); the diagram for chick agrin is based on that found in Tsim et al. (1992). as modified by K. Tsim, L. Patthy, and U. J. McMahan. The arrow marks the N-terminus of the truncated forms of chicken agrin assayed for AChR clustering activity. Recent experiments indicate that the rat agrin has a 4 aa insert corresponding to that found in chicken agrin and that chicken agrin has a 19 aa insert corresponding to that found in rat agrin (R. Scheller and U. J. McMahan, personal communication).
or may be important for organizing other components of the synaptic basal lamina. The pathways by which agrin acts on muscle ceils are unknown. Two recent observations may provide clues. Agrin binds to a membrane protein in muscle and electric organ that is distinct from the AChR (Nastuk et al., 1991). Second, agrin induces phosphorylation of the p subunit of the AChR (Wallace et al., 1991). The f3subunit appears to be in close contact with the 43 kd protein. The 43 kd protein is a cytoplasmic peripheral membrane protein present in 1 :l stoichiometry with the AChFl and has been implicated in cluster formation (Froehner et al., 1990; Phillips et al., 1991). One attractive hypothesis is that agrin acts through its receptor to stimulate tyrosine phosphorylation of the AChR and other proteins and that these events cause AChRs to cluster. This scheme seems a promising one for chick muscle, in which tyrosine phosphorylation is evident from the time of first formation of clusters; however, a more complicated scheme may be required for mammalian endplates, where phosphotyrosine is not detectable until the second postnatal week (Qu et al., 1990). It remains to be determined whether phosphotyrosine is present but undetectable earlier, or whether a different pathway is used, perhaps involving a second receptor. At any rate, the availability of the cloned agrin forms and the biochemical identification of an agrin receptor promise rapid progress in unraveling the molecular mechanism of this fundamental event in synaptogenesis. References Anderson, 757-773.
M. J., and Cohen,
Burden, S. J., Sargent, 82, 412-425.
M. W. (1977).
P. B., and McMahan,
Campanelli, J. T., Hoch, W., Rupp, (1991). Cell 67, 909-916. Cohen, Fallon,
J. Physiol.
M. W., and Godfrey, J. R., and Gelfman.
U. J. (1979).
F., Kreiner,
E. W. (1992). C. E. (1989).
Frank,
E., and Fischbach,
G. D. (1979). C. W., Scotland,
Godfrey,
Exp. Cell Res.
E. W. (1991).
E. W., Siebenlist, D. L., and Yorde,
Lieth, E., Cardasis, 54. Magill-Sole, 1833. Magill-Solo McMahan, 407-418.
J. Cell Biol.
J. Neurosci.,
R. H.
in press.
J. Cell Biol. 708, 1527-1535. F., Hall, Z. W., and
J. Cell Biol. 83, 143-158. P. B., and Patrick,
J. (1990).
795, 99-109.
Godfrey, E. W., Nitkin, R. M., Wallace, B. G., Rubin, han, U. J. (1984). J. Cell Biol. 99, 615-627. Godfrey, Bolender,
268,
T., and Scheller,
Ferns, M., Hoch, W., Campanelli, J. T., Rupp, Scheller, R. H. (1992). Neuron 8, 1079-1086. Froehner, S. C., Leutje, Neuron 5, 403-410.
(Lond.)
L. L., and McMa-
R. E., Wallskog, P. A., Walters, L. M., D. E. (1988). Dev. Biol. 730, 471-486.
C. A., and Fallon,
J. R. (1992).
Dev. Biol. 149,41-
C., and McMahan,
U. J. (1988).
J. Cell Biol. 707, 1825-
C., and McMahan,
U. J. (1990).
J. Exp. Biol. 153, l-10.
U. J. (1990).
Cold Spring
Harbor
Nastuk, M. A., Lieth, E.. Ma, J., Cardasis, McKechnie. B. A., and Fallon, J. R. (1991).
Symp.
Quant.
Biol. 55,
C. A., Moynihan, E. B., Neuron 7, 807-818.
Nitkin, R. M., Smith, M. A., Magill, C., Fallon, J. R., Yao, Y.-M. M., Wallace, B. G., and McMahan, U. J. (1987). J. Cell Biol. 105, 24712478. Phillips, W. P., Kopta, C., Blount, and Merlie, J. P. (1991). Science
P., Gardner, P. D., Steinbach, 257, 568-570.
J. H.,
Qu, Z., Moritz,
E., and Huganir,
Reist, N., Magill, 2457-2469. Reist, 868. Ruegg, Gensch,
N. E., Werle,
R. L. (1990).
C., and McMahan,
Neuron
U. J. (1987).
M. J., and McMahan,
4, 367-378.
J. Cell Biol. 705,
U. J. (1992).
Neuron
8,865-
M. A., Tsim, K. W. K., Horton, S. E., Kriiger, S., Escher, E. M., and McMahan, U. J. (1992). Neuron 8,691-699.
Rupp, F.. Payan, D. G., Magill-Solo R. H. (1991). Neuron 6, 811-823.
C., Cowan,
Rupp, F., Ozcelik, T. H., Linial, M., Peterson, Scheller, R. (1992). J. Neurosci., in press. Tsim, K. W. K., Ruegg, M. A., Escher, U. J. (1992). Neuron 8, 677-689. Wallace,
B. G. (1989).
Wallace,
B. G., Qu, Z., and Huganir,
J. Neurosci.
G., Kriiger,
G.,
D. M., and Scheller, K., Francke,
U., and
S., and McMahan,
9, 1294-1302. R. L. (1991).
Neuron
6, 869-878.
0 1992 by Cell Press
How Many Agrins Does It Take to Make a Synapse?
Minireview
Michael J. Ferns and Zach W. Hall Department of Physiology School of Medicine University of California, San Francisco San Francisco, California 94143-0444
Synaptogenesis occurs through an extended interplay between ceils-initiated by the first contact of an axonal growth cone with the surface of a postsynaptic cell (“the protoplasmic kiss” of Cajal) and culminating in the establishment of a mature synapse (“the final ecstasy of an epic love story”). The initial interaction between the pre- and postsynaptic cells triggers a series of events that locally alters the structures of both cells to form a stable synapse, which is then maintained over the lifetime of the organism and, in some cases, can be reformed efficiently after injury. At the neuromuscular junction, where the process has been best studied, the first discernible step in synaptic specialization is the clustering of acetylcholine receptors (AChRs) in the postsynaptic membrane underneath the nerve terminal. Experiments with cultured cells almost 15 years ago demonstrated that these clusters are induced by the nerve and arise in part by the redistribution of AChRs on the muscle cell surface (Anderson and Cohen, 1977; Frank and Fischbach, 1979). Later experiments have shown that increased local synthesis of the AChR also contributes to the nerve-induced clusters. Although the molecular mechanisms responsible for the redistribution of AChRs in the postsynaptic membrane are still unknown, recent experiments identify a clustering agent that is secreted by the nerve and offer tantalizing clues about how it works. Agrin, the clustering agent, was discovered after experiments on the frog neuromuscular junction had demonstrated that the synaptic basal lamina contains a factor that directs AChR clustering in regenerating muscle. When muscle fibers are damaged, the fusion of mono-
nerve terminal
Figure 1. A Diagrammatic Scheme for the Role of Agrin in the Neurally Induced Clustering of AChRs in the Muscle Fiber Membrane
nucleated cells causes new fibers to form within the old basal lamina sheath. In such fibers, AChRs accumulate at the old synaptic sites even when the nerve and its associated Schwann cell are absent, suggesting that the synaptic basal lamina contains molecules that cluster AChRs in the membrane of the newly formed muscle fibers (Burden et al., 1979). To identify these molecules, McMahan and colleagues purified from the Torpedo electric organ a protein, agrin, that clusters AChRs in cultured chick myotubes (Godfrey et al., 1984; Nitkin et al., 1987). Further, they showed that an immunologically similar protein is concentrated in the basal lamina at adult frog and chick endplates (Reist et al., 1987). Agrin is thus likely to be responsible for the clustering of AChRs in regenerating muscle and perhaps for the maintenance of clusters at adult synapses. The biological activity and synaptic location of agrin suggested that it might also be the agent that clusters AChRs at developing synapses. lmmunocytochemical and bioactivity assays have shown that agrin, or a protein related to it, is present in embryonic motoneurons and is transported to nerve terminals (Magill-Sole and McMahan, 1988, 1990). On the basis of these and other results, McMahan (1990) proposed that, during development, agrin is released by motor nerve terminals, becomes stably incorporated into the synaptic basal lamina, and induces the initial synaptic clustering of AChRs in muscle fibers (Figure 1). Acceptance of the “agrin hypothesis,” however, has been complicated by the finding that muscles and other tissues also contain proteins that are immunologically related to agrin and that have at least some AChR clustering activity (Godfrey, 1991). Agrin immunoreactivity is present in muscle prior to innervation and is colocalized with the few AChR clusters that appear in developing muscle when it is deprived of innervation (Godfrey et al., 1988; Fallon and Gelfman, 1989). These observations raise the possibility that the original clusters might be induced by musclederived agrin in a process that is directed or regulated by the nerve. Two recent cell culture experiments now make clear the importance of agrin in inducing AChR clusters at newly formed synapses, and, moreover, show that the responsible agrin is neurally derived. The most decisive result is the demonstration that antibodies specific for nerve-derived agrin block the formation of AChR clusters at neuromuscular contacts in interspecific cultures (Reist et al., 1992). When chick motor neurons were cultured with chick or rat muscles, AChR cluster formation at synaptic sites was blocked by an antibody to Torpedo agrin that appears to recognize chick but not rat agrin. In contrast, the antibody did not affect the formation of clusters on chick muscles induced by rat motor neurons. In acomplementaryfinding, another interspecific culture system (Xenopus and Rana) was used to show that nerve-derived agrin is associated with synaptic AChR clusters from the time of their first appearance, less than 2 hr after first nerve contact (Cohen and Godfrey, 1992). AChR clusters that are not associated with the nerve, however, have little or no agrin associated
Cdl 2
with them. Thus, neural rather than muscle agrin appears to be critical for inducing AChR clusters at newly formed synapses in culture. The function of agrin proteins in muscle and in other nonneural tissues remains obscure. A puzzling observation is that agrin in these tissues appears to be much less active in clustering AChRs than agrin derived from neural tissue (Godfrey, 1991). Recent experiments indicate that these differences may result from the expression of alternatively spliced forms of agrin that differ in biological activity. The recent cloning of rat and chick agrin cDNAs (Rupp et al., 1991; Tsim et al., 1992) shows that the encoded proteins are large (- 200 kd) with repeated domains that are homologous to domains in protease inhibitors, epidermal growth factor, and laminin (Figure 2). Analysisof chick and rat cDNAs reveals several sites of alternate RNA splicing (Tsim et al., 1992; Ruegg et al., 1992; Rupp et al., 1991,1992). Chick agrin has two short inserts of 4 and 11 aa near the C-terminus (Figure 2). In the rat, two contiguous inserts of 8 and 11 aa generate four different forms at the site of the 11 aa insert in the chick. All these inserts apparently affect the biological activity of the agrin forms (see below). When constructs comprising the C-terminal half of the chick protein were expressed in transfected cells, the secreted, soluble proteins that resulted were assayed on chick myotubes (Tsim et al., 1992; Ruegg et al., 1992), and only those forms containing both the 4 and 11 aa inserts were active. For the rat, fulllength constructs of splicing variants were expressed in transfected cells and assayed by coculture with myotubes from several sources (Campanelli et al., 1991; Ferns et al., 1992). In this case, although all forms were active on rat primary and C2 muscle cell line mpotubes, only those containing the 8 aa insert were active on a C2 variant with
defective proteoglycans. When assayed on chick myotubes, rat variant forms with the 8 aa insert were most active, while those with the 11 aa insert alone conferred weak activity. It is difficult to assess the exact relationship between inserts in the two species, because further splicing variants may be discovered (see legend to Figure 2) and because of the differences in assay conditions. Both studies clearly demonstrate, however, that the AChR clustering activity of agrin can be regulated by alternative RNA splicing, with the inclusion or exclusion of very small regions having dramatic effects on biological activity. Expression of the alternatively spliced agrin forms in the chick appears to be cell and tissue specific. Purified motor neurons express chiefly the most biologically active form (containing both 4 and 11 aa inserts) (Tsim et al., 1992). In contrast, embryonic muscle expresses forms;fhat lack the 11 aa insert and are devoid of AChR clustering activity in the assay for chick forms described above (Ruegg et al., 1992). The observed expression of the most active form by the nerve fits well with the agrin hypothesis but leaves open the question of function for the less active forms in muscle. One interesting observation is that the addition of exogenous Torpedo agrin (splicing forms unknown) to chick muscle causes the aggregation not only of AChRs but also of muscle agrin (Lieth et al., 1992); muscle agrin also accumulates at nerve-induced clusters. Exogenous agrin induces the accumulation of other synaptic basal lamina components, such as acetylcholinesterase and heparan sulfate proteogl$an, apparently by a mechanism that does not involve redistribution of preexisting molecules but requires new protein synthesis (Wallace, 1989). Neural agrin may thus act catalytically, inducing the accumulation of a larger amount of muscle agrin. The muscle agrin could augment the clustering activity of neural agrin
Rat
12345676
Chick aarin
cl
Figure 2. A Schematic Chick Agrin Molecules
Protease Inhilzilor-like domains Laminin-like domains
Ea
EGF-like domains
q
put&e
t ,200
signal sequence
Potential N-linked glycosylation sites aa ,
Diagram
of Rat
and
The diagram for rat agrin is based on those found in Rupp et al. (1991, 1992); the diagram for chick agrin is based on that found in Tsim et al. (1992). as modified by K. Tsim, L. Patthy, and U. J. McMahan. The arrow marks the N-terminus of the truncated forms of chicken agrin assayed for AChR clustering activity. Recent experiments indicate that the rat agrin has a 4 aa insert corresponding to that found in chicken agrin and that chicken agrin has a 19 aa insert corresponding to that found in rat agrin (R. Scheller and U. J. McMahan, personal communication).
or may be important for organizing other components of the synaptic basal lamina. The pathways by which agrin acts on muscle ceils are unknown. Two recent observations may provide clues. Agrin binds to a membrane protein in muscle and electric organ that is distinct from the AChR (Nastuk et al., 1991). Second, agrin induces phosphorylation of the p subunit of the AChR (Wallace et al., 1991). The f3subunit appears to be in close contact with the 43 kd protein. The 43 kd protein is a cytoplasmic peripheral membrane protein present in 1 :l stoichiometry with the AChFl and has been implicated in cluster formation (Froehner et al., 1990; Phillips et al., 1991). One attractive hypothesis is that agrin acts through its receptor to stimulate tyrosine phosphorylation of the AChR and other proteins and that these events cause AChRs to cluster. This scheme seems a promising one for chick muscle, in which tyrosine phosphorylation is evident from the time of first formation of clusters; however, a more complicated scheme may be required for mammalian endplates, where phosphotyrosine is not detectable until the second postnatal week (Qu et al., 1990). It remains to be determined whether phosphotyrosine is present but undetectable earlier, or whether a different pathway is used, perhaps involving a second receptor. At any rate, the availability of the cloned agrin forms and the biochemical identification of an agrin receptor promise rapid progress in unraveling the molecular mechanism of this fundamental event in synaptogenesis. References Anderson, 757-773.
M. J., and Cohen,
Burden, S. J., Sargent, 82, 412-425.
M. W. (1977).
P. B., and McMahan,
Campanelli, J. T., Hoch, W., Rupp, (1991). Cell 67, 909-916. Cohen, Fallon,
J. Physiol.
M. W., and Godfrey, J. R., and Gelfman.
U. J. (1979).
F., Kreiner,
E. W. (1992). C. E. (1989).
Frank,
E., and Fischbach,
G. D. (1979). C. W., Scotland,
Godfrey,
Exp. Cell Res.
E. W. (1991).
E. W., Siebenlist, D. L., and Yorde,
Lieth, E., Cardasis, 54. Magill-Sole, 1833. Magill-Solo McMahan, 407-418.
J. Cell Biol.
J. Neurosci.,
R. H.
in press.
J. Cell Biol. 708, 1527-1535. F., Hall, Z. W., and
J. Cell Biol. 83, 143-158. P. B., and Patrick,
J. (1990).
795, 99-109.
Godfrey, E. W., Nitkin, R. M., Wallace, B. G., Rubin, han, U. J. (1984). J. Cell Biol. 99, 615-627. Godfrey, Bolender,
268,
T., and Scheller,
Ferns, M., Hoch, W., Campanelli, J. T., Rupp, Scheller, R. H. (1992). Neuron 8, 1079-1086. Froehner, S. C., Leutje, Neuron 5, 403-410.
(Lond.)
L. L., and McMa-
R. E., Wallskog, P. A., Walters, L. M., D. E. (1988). Dev. Biol. 730, 471-486.
C. A., and Fallon,
J. R. (1992).
Dev. Biol. 149,41-
C., and McMahan,
U. J. (1988).
J. Cell Biol. 707, 1825-
C., and McMahan,
U. J. (1990).
J. Exp. Biol. 153, l-10.
U. J. (1990).
Cold Spring
Harbor
Nastuk, M. A., Lieth, E.. Ma, J., Cardasis, McKechnie. B. A., and Fallon, J. R. (1991).
Symp.
Quant.
Biol. 55,
C. A., Moynihan, E. B., Neuron 7, 807-818.
Nitkin, R. M., Smith, M. A., Magill, C., Fallon, J. R., Yao, Y.-M. M., Wallace, B. G., and McMahan, U. J. (1987). J. Cell Biol. 105, 24712478. Phillips, W. P., Kopta, C., Blount, and Merlie, J. P. (1991). Science
P., Gardner, P. D., Steinbach, 257, 568-570.
J. H.,
Qu, Z., Moritz,
E., and Huganir,
Reist, N., Magill, 2457-2469. Reist, 868. Ruegg, Gensch,
N. E., Werle,
R. L. (1990).
C., and McMahan,
Neuron
U. J. (1987).
M. J., and McMahan,
4, 367-378.
J. Cell Biol. 705,
U. J. (1992).
Neuron
8,865-
M. A., Tsim, K. W. K., Horton, S. E., Kriiger, S., Escher, E. M., and McMahan, U. J. (1992). Neuron 8,691-699.
Rupp, F.. Payan, D. G., Magill-Solo R. H. (1991). Neuron 6, 811-823.
C., Cowan,
Rupp, F., Ozcelik, T. H., Linial, M., Peterson, Scheller, R. (1992). J. Neurosci., in press. Tsim, K. W. K., Ruegg, M. A., Escher, U. J. (1992). Neuron 8, 677-689. Wallace,
B. G. (1989).
Wallace,
B. G., Qu, Z., and Huganir,
J. Neurosci.
G., Kriiger,
G.,
D. M., and Scheller, K., Francke,
U., and
S., and McMahan,
9, 1294-1302. R. L. (1991).
Neuron
6, 869-878.
