Neuron, Vol. 46, 569–579, May 19, 2005, Copyright ©2005 by Elsevier Inc.DOI 10.1016/j.neuron.2005.04.002
Neurotransmitter Acetylcholine Negatively
Regulates Neuromuscular Synapse Formation
by a Cdk5-Dependent Mechanism
Weichun Lin,1Bertha Dominguez, Jiefei Yang,
Prafulla Aryal, Eugene P. Brandon,
Fred H. Gage, and Kuo-Fen Lee*
The Salk Institute
10010 North Torrey Pines Road
La Jolla, California 92037
postsynaptic apparatus. However, recent genetic studies
suggest that postsynaptic differentiation of the neuro-
muscular synapse is initiated by a nerve-independent
mechanism and is prepatterned within the muscle (Lin
et al., 2001; Yang et al., 2001; Yang et al., 2000). During
the initial stage of synaptogenesis, the nascent, post-
synaptic acetylcholine receptor (AChR) clusters are
formed along a narrow central region of muscle and are
not closely apposed by nerve terminals (Lin et al., 2001;
Lupa and Hall, 1989; Yang et al., 2001). We have pos-
tulated that, at subsequent stages of synaptogenesis,
the navigating nerve simultaneously provides both pos-
itive and negative signals to regulate proper differentia-
tion and connection of the postsynaptic apparatus with
a nerve terminal (Lin et al., 2001). The positive signals
promote apposition and stability of the postsynaptic
apparatus with a nerve terminal, whereas the negative
signals disperse the postsynaptic apparatus that fails
to make the connection. As a result of the interplay of
these positive and negative signals, individual postsyn-
aptic apparatuses on each muscle fiber are innervated
What is the molecular nature of these positive and
negative signals that maintain or disperse the postsyn-
aptic apparatus during embryonic development? Act-
ing in part through the muscle-specific kinase (MuSK)
receptor complex, agrin plays a positive role in post-
synaptic differentiation by inducing and maintaining
AChR clustering in vitro and in vivo (Ferns et al., 1992;
Gautam et al., 1996; Lin et al., 2001; McMahan, 1990;
Nitkin et al., 1987; Ruegg et al., 1992; Yang et al., 2001).
Interestingly, analysis of agrin and MuSK mutant mice
revealed that MuSK, but not agrin, is required for initiat-
ing postsynaptic differentiation (Herbst et al., 2002; Lin
et al., 2001; Yang et al., 2001). However, agrin is pos-
tulated to play a role in stabilizing the nascent postsyn-
aptic apparatus that it encounters as well as in inducing
new postsynaptic sites (Lin et al., 2001; Yang et al.,
2001). Consistent with this idea, AChR clusters are ini-
tially formed in agrin-deficient mice but become dis-
persed at subsequent stages. We have hypothesized
that negative signals from the nerve and/or accompa-
nying Schwann cells induce the dispersion of the post-
synaptic apparatus in agrin mutant mice (Lin et al.,
2001). Both trophic factors and neural activity may play
a role in negatively regulating postsynaptic differentia-
tion by disassembling the postsynaptic apparatus in
cultures (Bloch, 1979; Bloch, 1986; Loeb et al., 2002;
Trinidad and Cohen, 2004; Wells et al., 1999). Agrin has
been shown to cause inhibition of motor axon growth
and to promote the assembly of presynaptic specializa-
tion in vitro (Campagna et al., 1997; Campagna et al.,
1995). Interestingly, both MuSK and agrin mutant mice
display excess axon growth along muscle fibers (DeChi-
ara et al., 1996; Gautam et al., 1996; Lin et al., 2001).
Thus, agrin functions as a positive signal to promote
both presynaptic and postsynaptic development. Fi-
nally, several lines of evidence from studies using phar-
macological inhibitors of cytoplasmic protein kinases
and phosphatases suggest that these molecules are in-
Synapse formation requires interactions between pre-
and postsynaptic cells to establish the connection of
a presynaptic nerve terminal with the neurotransmitter
receptor-rich postsynaptic apparatus. At developing
vertebrate neuromuscular junctions, acetylcholine
receptor (AChR) clusters of nascent postsynaptic
apparatus are not apposed by presynaptic nerve ter-
minals. Two opposing activities subsequently pro-
mote the formation of synapses: positive signals sta-
bilize the innervated AChR clusters, whereas negative
signals disperse those that are not innervated. Al-
though the nerve-derived protein agrin has been sug-
gested to be a positive signal, the negative signals
remain elusive. Here, we show that cyclin-dependent
kinase 5 (Cdk5) is activated by ACh agonists and is
required for the ACh agonist-induced dispersion of
the AChR clusters that have not been stabilized by
agrin. Genetic elimination of Cdk5 or blocking ACh
production prevents the dispersion of AChR clusters
in agrin mutants. Therefore, we propose that ACh
negatively regulates neuromuscular synapse forma-
tion through a Cdk5-dependent mechanism.
Synapses are specialized structural units for cellular
communication in the nervous system and therefore are
essential for executing complex nervous system func-
tions. A central question in neurobiology is how the pat-
terns of synaptic connection are established and main-
tained (Cohen-Cory, 2002; Goda and Davis, 2003; Shen
et al., 2004). One of the key features during patterning
and formation of synapses is the close apposition of
individual neurotransmitter receptor-rich postsynaptic
apparatus with a specialized nerve terminal. The verte-
brate neuromuscular junction (NMJ) is a well-studied
model for elucidating the mechanisms underlying pat-
terning and formation of synapses (Burden, 1998;
Sanes and Lichtman, 1999; Wyatt and Balice-Gordon,
Over the last several decades, a neurocentric model
of NMJ synaptic formation has been espoused, sug-
gesting that motor nerves are largely responsible for
both initiating and maintaining differentiation of the
1Present address: Center for Basic Neuroscience, UT Southwest-
ern Medical Center, 6000 Harry Hines Boulevard, Dallas, Texas
volved in the maintenance and dispersion of AChR
clusters (Cheng and Ip, 2003; Dai and Peng, 1998;
Ferns et al., 1996; Madhavan et al., 2005; Wallace,
In the present study, we demonstrate that cyclin-
dependent kinase 5 (Cdk5), a cytoplasmic serine/threo-
nine kinase, plays a role in the dispersion of AChR clus-
ters. We show that cholinergic agonists activate Cdk5
and disperse AChR clusters in myotube cultures that
are previously induced by agrin. Blockade of Cdk5 ac-
tivity by a selective inhibitor, roscovitine, ameliorates
ACh-induced dispersion of AChR clusters in cultures.
Moreover, blocking Cdk5 activity by in utero treatment
with roscovitine or by genetic deletion of the Cdk5 gene
in agrin mutant embryos leads to the maintenance of a
significant number of AChR clusters that would other-
wise have been dispersed. Similar results were ob-
served when production of the neurotransmitter ACh
was blocked in agrin mutant embryos. These combined
pharmacological and genetic results suggest a Cdk5-
dependent pathway by which ACh regulates the forma-
tion of synapses by disassembling the postsynaptic
apparatus that fails to make correct connections with
the presynaptic terminals.
Figure 1. Maintenance of AChR Clusters in Agrin Mutants following
Injection of Cdk5-Selective Inhibitor
In utero injection of the Cdk5-specific inhibitor roscovitine dis-
solved in solution containing DMSO. E16.5 diaphragm muscles
were collected from embryos of pregnant females injected daily
with DMSO (A and B) or roscovitine (C and D) starting at E14.5 and
stained for AChR clusters. In contrast to a few AChR clusters in
AGD mutants (B), numerous AChR clusters were present in agrin
(AGD) mutants treated with roscovitine (D). Scale bar, 50 ?m.
Blockade of Cdk5 Activity Retains
AChR Clusters in Agrin Mutants
It has been previously shown that AChR clusters are
present during the early stages of development and
dispersed subsequently in agrin mutant embryos (Lin
et al., 2001; Yang et al., 2001). The nerve then provides
negative signals to elicit pathways to disperse AChR
clusters in the absence of agrin (Lin et al., 2001). What
might be the molecular basis of such negative signal-
ing? Several lines of evidence suggest that protein ki-
nases and phosphatases may play a role in clustering
or dispersing AChRs. For example, protein kinase and
phosphatase inhibitors profoundly affect the formation
and stability of AChR clusters induced by agrin (Dai and
Peng, 1998; Ferns et al., 1996; Madhavan et al., 2005;
Wallace, 1995). Several cytoplasmic phosphatases and
kinases have been implicated in the formation and sta-
bility of AChR clusters, including phosphatase Shp-2
and members of the Src kinase family (Dai and Peng,
1998; Madhavan et al., 2005; Smith et al., 2001). Recent
results suggest that Cdk5 may be activated by nerve-
derived dispersion factors to antagonize agrin-medi-
ated AChR clustering. First, Cdk5 and its coactivator
p35 are expressed in muscle (Fu et al., 2001). Second,
Cdk5 can be activated by factors that have been shown
to induce the dispersion of AChR clusters in vitro, in-
(Tokuoka et al., 2000; Wells et al., 1999) and neuregu-
lin-1 (NRG-1) (Fu et al., 2001; Fu et al., 2003; Trinidad
and Cohen, 2004). Finally, activation of Cdk5 reduces
the clustering of the scaffolding protein postsynaptic
density-95 (PSD-95) and ion channels in heterologous
cells, and Cdk5-deficient cortical neurons display in-
creased PSD-95 cluster size (Morabito et al., 2004),
suggesting a negative role for Cdk5 in regulating as-
sembly of the postsynaptic apparatus. Therefore, we
asked whether Cdk5 is required for dispersing AChR
clusters in agrin mutant embryos.
We began our analysis by pharmacologically block-
ing Cdk5 activity in agrin (AGD allele) mutants in utero.
Pregnant E14.5 female mice from an intercross of AGD
heterozygotes were injected daily with the Cdk5-spe-
cific inhibitor roscovitine to block the Cdk5 activity
(Meijer et al., 1997). Embryos were collected at E16.5,
two days after the start of roscovitine injection, and
were processed for AChR clustering analysis. Since
roscovitine was dissolved in DMSO solution prior to in-
jection, DMSO-containing solution was injected as a
control. As shown in Figure 1, few AChR clusters were
present in AGD mutant muscles treated with DMSO
alone (Figure 1B). In contrast, numerous AChR clusters
were present in AGD mutants treated with roscovitine
(Figure 1D). Control littermates treated either with
DMSO alone (Figure 1A) or roscovitine (Figure 1C) did
not exhibit a marked alteration in AChR clustering.
Quantitatively, the average areas of individual AChR
clusters were much greater in AGD mutants treated
with roscovitine than those observed in AGD mutants
treated with DMSO alone (Table 1).
Although roscovitine has been widely used to block
Cdk5 activity, it is possible that the inhibitor may have
other nonspecific effects in embryos. We therefore took
a genetic approach to further determine whether Cdk5
plays a role in the dispersion of AChR clusters in agrin
mutants. We analyzed NMJ development in Cdk5 mu-
tants (Xie et al., 2003) and agrin (AGD)/Cdk5 double
ACh Negatively Regulates Synaptogenesis via Cdk5
Nikolic, M., Chou, M.M., Lu, W., Mayer, B.J., and Tsai, L.H. (1998).
The p35/Cdk5 kinase is a neuron-specific Rac effector that inhibits
Pak1 activity. Nature 395, 194–198.
Nitkin, R.M., Smith, M.A., Magill, C., Fallon, J.R., Yao, Y.-M.M., Wal-
lace, B.G., and McMahan, U.J. (1987). Identification of agrin, a syn-
aptic organizing protein from Torpedo electric organ. J. Cell Biol.
O'Malley, J.P., Moore, C.T., and Salpeter, M.M. (1997). Stabilization
of acetylcholine receptors by exogenous ATP and its reversal by
cAMP and calcium. J. Cell Biol. 138, 159–165.
Rotzler, S., and Brenner, H.R. (1990). Metabolic stabilization of ace-
tylcholine receptors in vertebrate neuromuscular junction by mus-
cle activity. J. Cell Biol. 111, 655–661.
Ruegg, M.A., Tsim, K.W.K., Horton, S.E., Kröger, 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 distribu-
tion. Neuron 8, 691–699.
Sahlgren, C.M., Mikhailov, A., Vaittinen, S., Pallari, H.M., Kalimo, H.,
Pant, H.C., and Eriksson, J.E. (2003). Cdk5 regulates the organiza-
tion of Nestin and its association with p35. Mol. Cell. Biol. 23,
Sanes, J.R., and Lichtman, J.W. (1999). Development of the verte-
brate neuromuscular junction. Annu. Rev. Neurosci. 22, 389–442.
Shen, K., Fetter, R.D., and Bargmann, C.I. (2004). Synaptic specific-
ity is generated by the synaptic guidepost protein SYG-2 and its
receptor, SYG-1. Cell 116, 869–881.
Shetty, K.T., Link, W.T., and Pant, H.C. (1993). cdc2-like kinase from
rat spinal cord specifically phosphorylates KSPXK motifs in neuro-
filament proteins: isolation and characterization. Proc. Natl. Acad.
Sci. USA 90, 6844–6848.
Siegelbaum, S.A., Trautmann, A., and Koenig, J. (1984). Single ace-
tylcholine-activated channel current in developing muscle cells.
Dev. Biol. 104, 366–379.
Smith, C.L., Mittaud, P., Prescott, E.D., Fuhrer, C., and Burden, S.J.
(2001). Src, Fyn, and Yes are not required for neuromuscular syn-
apse formation but are necessary for stabilization of agrin-induced
clusters of acetylcholine receptors. J. Neurosci. 21, 3151–3160.
Tan, T.C., Valova, V.A., Malladi, C.S., Graham, M.E., Berven, L.A.,
Jupp, O.J., Hansra, G., McClure, S.J., Sarcevic, B., Boadle, R.A., et
al. (2003). Cdk5 is essential for synaptic vesicle endocytosis. Nat.
Cell Biol. 5, 701–710.
Tanowitz, M., Si, J., Yu, D.-H., Feng, G.-S., and Mei, L. (1999). Regu-
lation of neuregulin-mediated acetylcholine receptor synthesis by
protein tyrosine phosphatase SHP2. J. Neurosci. 19, 9426–9435.
Tokuoka, H., Saito, T., Yorifuji, H., Wei, F., Kishimoto, T., and Hisa-
naga, S. (2000). Brain-derived neurotrophic factor-induced phos-
phorylation of neurofilament-H subunit in primary cultures of
embryo rat cortical neurons. J. Cell Sci. 113, 1059–1068.
Tomizawa, K., Sunada, S., Lu, Y.F., Oda, Y., Kinuta, M., Ohshima, T.,
Saito, T., Wei, F.Y., Matsushita, M., Li, S.T., et al. (2003). Cophosph-
orylation of amphiphysin I and dynamin I by Cdk5 regulates
clathrin-mediated endocytosis of synaptic vesicles. J. Cell Biol.
Trinidad, J.C., and Cohen, J.B. (2004). Neuregulin inhibits acetyl-
choline receptor aggregation in myotubes. J. Biol. Chem. 279,
Wallace, B.G. (1995). Regulation of the interaction of nicotinic ace-
tylcholine receptors with the cytoskeleton by agrin-activated pro-
tein tyrosine kinase. J. Cell Biol. 128, 1121–1129.
Wells, D.G., McKechnie, B.A., Kelkar, S., and Fallon, J.R. (1999).
Neurotrophins regulate agrin-induced postsynaptic differentiation.
Proc. Natl. Acad. Sci. USA 96, 1112–1117.
Weston, C., Yee, B., Hod, E., and Prives, J. (2000). Agrin-induced
acetylcholine receptor clustering is mediated by the small gua-
nosine triphosphatases Rac and Cdc42. J. Cell Biol. 150, 205–212.
Womble, M.D. (1986). The clustering of acetylcholine receptors and
formation of neuromuscular junctions in regenerating mammalian
muscle grafts. Am. J. Anat. 176, 191–205.
Wyatt, R.M., and Balice-Gordon, R.J. (2003). Activity-dependent
elimination of neuromuscular synapses. J. Neurocytol. 32, 777–794.
Xie, Z., Sanada, K., Samuels, B.A., Shih, H., and Tsai, L.H. (2003).
Serine 732 phosphorylation of FAK by Cdk5 is important for micro-
tubule organization, nuclear movement, and neuronal migration.
Cell 114, 469–482.
Yang, X., Li, W., Prescott, E.D., Burden, S.J., and Wang, J.C. (2000).
DNA topoisomerase IIβ and neural development. Science 287,
Yang, X., Arber, S., Willian, C., Li, L., Tanabe, Y., Jessell, T.M., Birch-
meier, C., and Burden, S.J. (2001). Patterning of muscle acetylcho-
line receptor gene expression in the absence of motor innervation.
Neuron 30, 399–410.
Young, S.H., and Poo, M.-M. (1983). Spontaneous release of
transmitter from growth cones of embryonic neurones. Nature 305,