Neuron, Vol. 24, 567±583, November, 1999, Copyright 1999 by Cell Press
Disruption of TrkB-Mediated Signaling Induces
Disassembly of Postsynaptic Receptor Clusters
at Neuromuscular J unctions
spines (Horch et al., 1999) in cortical slices in vitro, it
remains unclear what role neurotrophic signaling plays
at the level of the synapse. The neuromuscular junction
is a good modelsystemforaddressing the role ofneuro-
trophin signaling in synaptic structure and function, ow-
ing to its simple organization and accessibility foranaly-
ses in living animals. The three cell types that comprise
junctions, the terminal Schwann cell, the presynaptic
terminals of motor neurons, and the postsynaptic mus-
cle fiber, express neurotrophins and Trk receptors in
development as well as adulthood, and this expression
is innervation dependent (cf. Funakoshi et al., 1993,
1995;Griesbeck etal., 1995).Brain-derived neurotrophic
factor (BDNF), full-length TrkC, and a truncated TrkB
isoform lacking the intracellular tyrosine kinase domain
(trkB.t1; Middlemas et al., 1991; Eide et al., 1996) are
expressed by Schwann cells (Frisen et al., 1993; Funa-
koshi et al., 1993; Offenhauser et al., 1995); BDNF, NT-
4/5, full-length TrkB, and full-length TrkC are expressed
by motor neurons (Yan et al., 1997), and NT-4/5 and
BDNF are expressed by muscle fibers (Funakoshi et al.,
1993; Koliatsos et al., 1993). It is generally believed that
postsynaptic muscle fibers release neurotrophins and
other signaling molecules that are essential for the sur-
vival of motor neurons and that these are taken up by
presynaptic motor nerve terminals, which express the
appropriate receptors and transport neurotrophin bound
to receptor retrogradely to the motor neuron cell body
(Henderson et al., 1993; Koliatsos et al., 1993, 1994;
Friedman et al., 1995; Funakoshi, 1995). Exchange of
neurotrophins and other molecules is likely to be an
rotrophins potentiate presynaptic release ofneurotrans-
mitter (Lohof et al., 1993; Wang and Poo, 1997) and are
essential for motor neuron (reviewed by Oppenheim,
1996) and Schwann cell (Trachtenberg and Thompson,
1996)survival, as wellas forthemaintenanceofpostsyn-
aptic characteristics in developing and mature muscle
(Wang etal., 1995; Wang and Poo, 1997; Xie etal., 1997).
Itis unknownwhethersignaling throughthese pathways
has functional consequences for developing or adult
neuromuscularsynapses in vivo, because many geneti-
cally altered mice lacking a Trk or neurotrophin die in
the embryonic or perinatal period (reviewed by Snider,
1994), making them not useful for exploring the role of
neurotrophin signaling at junctions after birth.
The cellular localization of neurotrophins and Trks at
neuromuscularjunctions and theirrole insynaptic main-
tenance have not been well studied. To address this
issue, we used immunostaining, Western blot analysis,
andRT-PCR todescribetheexpressionofTrkB isoforms
at neonatal and adult neuromuscular junctions. These
analyses showed thatbothtrkB.FL and trkB.t1 are local-
ized to the perisynaptic region, including the postsynap-
tic muscle fiber membrane and Schwann cells. Given
that TrkB isoforms were localized to the postsynaptic
membrane at AChR-rich regions, we asked what role
postsynaptic TrkB-mediated signaling might play in
Michael Gonzalez, Francis P. Ruggiero, Qiang Chang,
Yi-J un Shi, Mark M. Rich, Susan Kraner,
and Rita J . Balice-Gordon*
Department of Neuroscience
University of Pennsylvania School of Medicine
Philadelphia, Pennsylvania 19104
Neurotrophins and tyrosine receptor kinase (Trk) re-
ceptors are expressed in skeletal muscle, but it is
unclear what functional role Trk-mediated signaling
plays during postnatal life. Full-length TrkB (trkB.FL)
as well as truncated TrkB (trkB.t1) were found to be
localized primarily to the postsynaptic acetylcholine
receptor± (AChR-) rich membrane at neuromuscular
junctions. In vivo, dominant-negative manipulation of
TrkB signaling using adenovirus to overexpress trkB.t1
in mouse sternomastoid muscle fibers resulted in the
disassembly of postsynaptic AChR clusters at neuro-
muscular junctions, similar to that observed in mutant
trkB?/?mice. When TrkB-mediated signaling was dis-
rupted in cultured myotubes in the absence of motor
nerve terminals and Schwann cells, agrin-induced
AChR clusters were also disassembled. These results
demonstrate a novel role for neurotrophin signaling
through TrkB receptors on muscle fibers in the ongo-
ing maintenance of postsynaptic AChR regions.
Neurotrophins and theirreceptors, the tyrosine receptor
kinases (Trks),play numerous roles during development,
regulating neuronal survival (cf. Oppenheim, 1996) and
cell morphology (cf. McAllister et al., 1995, 1997; Horch
et al., 1999), as well as synaptic connectivity (cf. Cabelli
et al., 1995, 1997). Neurotrophin and/or Trk manipula-
tions in a number of systems (reviewed by Barbacid,
1994; Snider, 1994) have suggested that synaptic con-
nections are maintained and modulated by the retro-
grade and anterograde exchange of these signals be-
tween pre- and postsynaptic cells. This modulation can
be acute, occurring within minutes of application in tis-
sue culture (Lohof et al., 1993), as well as chronic, for
example, by maintaining immature patterns of synaptic
connections in skeletal muscle (cf. Nguyen et al., 1998)
and visual cortex (Cabelli et al., 1997) for as long as
excess neurotrophin is present.
The role that neurotrophins and Trk receptors play in
modulating synaptic structure and function is poorly
understood. While ligands that signal through TrkB and
TrkC have antagonistic effects in modulating dendritic
outgrowthand remodeling (McAllisteretal., 1995, 1997),
and have been shown to destabilize dendrites and
*To whom correspondence should be addressed (e-mail: rbaliceg@
Figure 1. TrkB Expression at Adult and Neonatal Mouse Neuromuscular J unctions
TrkB immunoreactivity (left), R?BTX staining of postsynaptic AChRs (middle), and an overlay (right; TrkB, green; AChRs, red) of neuromuscular
junctions from adult innervated (A, C, and G) and denervated (B, D, and H) and neonatal P8 innervated (E) and denervated (F) mouse
(A and B) Anti-trkB.FL immunoreactivity at innervated (A) and denervated (B) neuromuscular junctions is colocalized with AChR-rich areas.
Faint immunoreactivity is also present in preterminal axons (asterisk).
(C and D) Anti-trkB.t1 immunoreactivity at innervated (C) and denervated (D) neuromuscularjunctions is interdigitated among AChR-rich areas;
some faint immunoreactivity is colocalized with AChRs.
(E±H) Immunoreactivity at innervated (E) and denervated (F) P8 and innervated (G) and denervated (H) adult neuromuscular junctions, revealed
with an anti-TrkB antibody that recognizes both trkB.FL and trkB.t1, is localized within and surrounding AChR-rich regions. Inset in (G) shows
the motor nerve terminal and axon immunostaining used to demonstrate that junctions were innervated. Inset in (H) shows the absence of
this staining, which was used to determine that junctions were denervated. Scale bar, 20 ?m.
synaptic maintenance at neonatal and adult mouse
neuromuscular junctions in vivo. Adenovirus-mediated
somatic gene transfer provided the opportunity to ma-
nipulate the molecular environment of individual neuro-
muscular junctions in an otherwise normal background
and to assess the effects in living animals over time.
We reasoned that this approach would be useful for
manipulating TrkB-mediated signaling, as mutant mice
lacking TrkB die at birth (Klein et al., 1993) and would
be more generally usefulforspatialand temporalmanip-
ulation of other molecules of interest.
Signaling through trkB.FL was decreased in a domi-
nant-negative fashion by overexpressing trkB.t1 in ster-
nomastoid muscle fibers using recombinant adenovirus
that also drove the expression of the nonbiologically
active fluorescent protein GFP (Chalfie, 1995), using an
internal ribosome entry site (IRES) sequence (J ang et
al., 1989; Tomanin et al., 1997). GFP expression allowed
infected and noninfected muscle fibers to be imaged
in vivo, and neuromuscular junctions on infected and
control muscle fibers were then analyzed with immuno-
staining and confocal microscopy. We report that post-
synaptic AChR regions at neuromuscular junctions are
disassembled when TrkB-mediated signaling is dis-
rupted in a dominant-negative fashion by overexpres-
sion of trkB.t1, but not of a truncated trkA receptor, and
that AChR regions also appear to be disassembled in
mutant mice expressing half as much full-length TrkB
as wild-type mice. An in vitro agrin-induced AChR clus-
tering assay showed that agrin-induced AChR clusters
were disassembled on myotubes overexpressing trkB.t1.
This experiment suggested that postsynaptic AChR
clusterdisassembly occurs whenTrkB-mediated signal-
ing is disrupted in the muscle fiber membrane, even in
the absence of motor nerve terminals and Schwann
cells. Our results indicate that neurotrophin signaling
TrkB-Mediated Signaling at Neuromuscular J unctions
Figure 2. Specificity of Anti-Trk Antibodies and RT-PCR Amplification of TrkB.FL mRNA from Mouse Myotubes
(A±C) Western blot of membrane protein preparations from brain and skeletal muscle probed with a rabbit polyclonal anti-trkB.FL-specific
antibody that revealed a single ?140 kDa band (A); a rabbit polyclonal anti-trkB.t1-specific antibody that revealed a single ?95 kDa band (B);
or a rabbit polyclonal anti-TrkB antibody that recognizes both trkB.FL and trkB.t1 which revealed two bands, of ?140 kDa and ?95 kDa (C).
This antibody detected less trkB.FL than trkB.t1 in muscle.
(D) RT-PCR amplification of trkB.FL mRNA from brain and mouse primary myotubes using nested primers resulted in detection of a 740 base
pair band from brain and primary myotube cDNA, as expected, but not from water or myotube RNA negative control lanes.
(E) Western blot of brain and skeletal muscle membrane protein preparations probed with a rabbit polyclonal anti-TrkA-specific antibody
revealed a single ?140 kDa band from brain but not muscle.
(F) Confirming previous results (Klein et al., 1993), Western blot of brain membrane preparations from trkB?/?mice shows about half as much
trkB.FL and somewhat less trkB.t1 expression compared with trkB?/?mice.
through TrkB receptors on muscle fibers is necessary
for the ongoing maintenance of postsynaptic AChR re-
gions at neonatal as well as adult neuromuscular junc-
tions. A portion of these results have appeared in ab-
stract form (F. Ruggiero et al., 1998, Soc. Neurosci.,
abstract; M. Gonzalez et al., 1999, Soc. Neurosci., ab-
the extracellular epitope of trkB that recognizes both
the full-length and truncated isoforms of trkB (Huang et
al., 1999)showed thattrkB was localized to the postsyn-
aptic muscle fiber membrane at neuromuscular junc-
tions from postnatal day 8 (P8) mice (Figure 1E), and
this localization also persisted after denervation (Figure
1F). A similar pattern of trkB localization was observed
at innervated (Figure 1G) and denervated (Figure 1H)
adultneuromuscularjunctions, using this antibody.TrkA
immunoreactivity was not detected with a polyclonal
anti-trkA antibody(Clary etal.,1994)ateitherneonatalor
adult neuromuscular junctions (data not shown). These
results indicate that trkB.FL and trkB.t1 are localized
primarily postsynaptically, to regions that contain AChRs.
Westernblotanalysis ofmembrane preparations from
brain and skeletal muscle was used to determine the
specificity of the antibodies used for immunostaining.
A single ?140 kDa band was detected in brain and
skeletalmuscle,using theanti-trkB.FL-specific antibody
(Figure 2A), and a single ?95 kDa band was detected in
brainand skeletalmuscle, using theanti-trkB.t1-specific
antibody (Figure 2B). Two bands were detected in brain
and skeletal muscle, using the anti-trkB antibody that
recognizes both trkB.FL and trkB.t1 (Figure 2C); this
antibody detected less trkB.FL than trkB.t1 in muscle.
As expected from immunostaining observations, full-
length trkA was detected as a ?140 kDa band in brain
but not muscle (Figure 2E).
To further evaluate whether trkB.FL was expressed
by skeletal muscle fibers as opposed to Schwann cells
and/or motor axons and nerve terminals, RT-PCR was
performed onmRNA frompurified primary myotube cul-
tures and brain. By using nested primers designed to
specifically amplify the kinase domainoftrkB.FL, a band
of the expected size (740 base pairs) was detected in
primary myotubes, as well as in brain (Figure 2D), but
Full-length and Truncated TrkB Are Localized
to the Postsynaptic Membrane
in Neuromuscular J unctions
Northern and Western blot analyses have been used to
describe TrkB expression in skeletal muscle and sciatic
nerve (Funakoshi et al., 1993; Griesbeck et al., 1995),
but these include different cell types, and the cellular
localization of TrkB at neuromuscular junctions has not
been resolved. Immunostaining in muscle whole mounts,
using a rabbit polyclonal antibody against the kinase
domain of trkB.FL (Yan et al., 1997) and rhodamine-
conjugated ?-bungarotoxin(R?BTX), followed by confo-
cal microscopy, showed that trkB.FL was localized to
postsynaptic AChR-richregions (Figure 1A).While some
faint trkB.FL immunostaining was also observed in pre-
terminal axon branches (Figure 1A, asterisk), the per-
sistence of trkB.FL immunoreactivity after removal of
presynaptic motor nerve terminals by denervation con-
firmed its primarily postsynaptic localization(Figure 1B).
By using an anti-trkB.t1-specific antibody, trkB.t1 was
found to be localized to the postsynaptic membrane
within and surrounding AChR-rich areas (Figure 1C),
which also persisted after denervation (Figure 1D). Nei-
therantibody detected TrkB immunostaining inneonatal
neuromuscular junctions. However, an affinity-purified
polyclonal anti-trkB antibody raised in rabbit against
Figure 3. Adenovirus Constructs and Analysis of Transgene Expression
(A) Plasmids containing a trkB.t1ha, t-trkAha, or LacZ cDNA, an IRES sequence, and GFP were transfected into 293 cells with a plasmid
containing the adenovirus genome (dl327). After homologous recombination, screening, and plaque purification, high-titer adenovirus was
(B) Western blot analysis of transgene expression after infection of 293 cells with AdtrkB.t1ha-gfp (anti-trkB.t1, anti-HA, anti-GFP), Adt-trkAha-
gfp (anti-TrkA, anti-HA, anti-GFP), or AdlacZ-gfp (anti-?-gal, anti-GFP) compared with expression after transfection with the plasmid used to
generate adenovirus and endogenous trkB.t1 and TrkA from mouse brain. In AdtrkB.t1ha-gfp-infected cells, a band is detected at ?95 kDa,
the appropriate molecular mass for trkB.t1 (Klein et al., 1990), using anti-trkB.t1 and anti-HA antibodies; markers indicated by black lines are
116 and 83 kDa. Anti-GFP staining revealed a band at ?24 kDa, similar to purified recombinant GFP (last lane); markers, 46 and 30 kDa. Anti-
?-gal staining revealed a band at ?116 kDa, similar to purified recombinant ?-gal (last lane); markers, 116, 80, and 49.5 kDa. In Adt-trkAha-
gfp-infected cells, a band is detected at ?95 kDa, the predicted molecular mass generated by the hybrid, truncated TrkA construct, using
anti-TrkA and anti-HA antibodies, and a ?140 kDa band is detected in brain, the appropriate molecular mass for full-length TrkA; markers,
116 and 83 kDa. Anti-GFP staining revealed a band at ?24 kDa, similar to purified recombinant GFP (last lane); markers, 46 and 30 kDa.
(C) Gfp fluorescence and immunostaining of AdtrkB.t1ha-gfp- or Adt-trkAha-gfp-infected myotubes. Anti-HA immunoreactivity is localized in
the membrane of GFP?but not in GFP?myotubes infected with recombinant adenovirus. Anti-trkB.t1 immunoreactivity is colocalized with
anti-HA immunoreactivity inthe membrane ofGFP?myotubes;weak trkB.t1immunoreactivity is also presentinuninfected myotubes, consistent
with the localization of trkB.t1 in the postsynaptic membrane of neonatal and adult muscle fibers (Figure 1). Anti-TrkA staining shows t-trkA
is colocalized with anti-HA immunoreactivity in the cell membrane of GFP?but not in GFP?myotubes. Scale bar, 10 ?m.
(D) In a PC12 cell line stably expressing trkB.FL (PC12-trkB?), neurite outgrowth in response to 50 ng/ml BDNF was significantly reduced in
AdtrkB.t1ha-gfp-infected cells (p ?0.001, Student's ttest)compared withAdlacZ-gfp-infected and uninfected cells, whichwere notsignificantly
(E) No significant differences in neurite outgrowth were observed in infected and uninfected PC12-trkB?cells treated with 50 ng/ml NGF.
(F) In PC12 cells expressing TrkA, neurite outgrowth in response to 50 ng/ml NGF was significantly reduced in Adt-trkAha-gfp-infected cells
(p ? 0.001, Student's t test) compared with AdlacZ-gfp-infected and -uninfected cells, which were not significantly different. Bars indicate
mean ? SEM.
TrkB-Mediated Signaling at Neuromuscular J unctions
Figure 4. Transgene Expression and Lack of Inflammation following Adenovirus Infection of Mouse Sternomastoid Muscle
Shown are sternomastoid muscle cross sections taken through the region containing neuromuscular junctions 1 month after infection on P2
with AdtrkB.t1ha-gfp (A±C) or Adt-trkAha-gfp (D±F). GFP fluorescence (A and D) and anti-HA staining (B and E) reveal transgene localization
in the membrane of GFP?muscle fibers. Overlay (C and F) shows that both transgenes are expressed in muscle fibers with high fidelity. In
(G) and (H), no evidence of inflammation or infiltration of mononucleated cells, evidence of immune system involvement, is observed in H and
E±stained cross sections of muscle 2 weeks after AdtrkB.t1ha-gfp infection at P2 (G) or 1 month after infection of adult SCID mice (H). Similar
observations were made after infection with Adt-trkA-gfp and AdlacZ-gfp (data not shown). In (I), no evidence of myopathy was observed in
muscles from trkB?/?mice. Scale bar, 40 ?m.
not in the negative control lanes (water, myotube RNA
as templates; Figure 2D). While myotube cultures may
containoccasionalfibroblasts and Schwanncells, these
cells do not express trkB.FL (cf. Funakoshi et al., 1993)
and thus would not contribute to the RT-PCR amplifi-
Taken together, these results show that both trkB.FL
and trkB.t1, but not trkA, are localized to the postsynap-
tic muscle fibermembrane and not, as expected, exclu-
sively to presynaptic motor nerve terminals. We thus
asked what role postsynaptic TrkB-mediated signaling
might play in synaptic maintenance at neonatal and
adult mouse neuromuscular junctions. trkB?/?mice die
at birth (Klein et al., 1993), so we attempted to decrease
signaling through trkB.FL by overexpressing trkB.t1 in
muscle fibers, using recombinant adenovirus. Because
BDNF and NT-4/5 bind to both trkB.FL and trkB.t1, but
only trkB.FL homodimers can transduce neurotrophin
signals, we reasoned that this would result in a domi-
nant-negative manipulation whose effect could be as-
sayed at neuromuscular junctions in vivo.
Characterization of Adenovirus-Mediated
Transgene Expression In Vitro
Adenoviruses were generated thatdrove the expression
oftwotransgenes underthe controlofthecytomegalovi-
rus (CMV) promoter and an IRES sequence to allow
translation of a bicistronic message (J ang et al., 1989;
Tomanin et al., 1997). AdtrkB.t1ha-gfp encoded mouse
trkB.t1, which was epitope tagged with hemagglutinin
(HA) at its carboxyl terminus and the biologically inac-
tive, fluorescentproteinGFP (Figure 3A).Adt-trkAha-gfp
encoded a truncated, epitope-tagged trkA consisting of
the mouse trkA extracellular and transmembrane do-
mains fused to the short cytoplasmic tail of trkB.t1 and
GFP (Figure 3A). This virus was used to evaluate the
Table 1. J unction and Muscle Fiber Morphometry in Infected and Uninfected Muscle Fibers
Number of Discrete
AChR RegionsJ unction Area (?m2) Muscle Fiber Diameter
AdlacZ-gfp7 ? 0.5b
21 ? 0.5e
6.4 ? 0.5
7 ? 1
7 ? 1
7 ? 1
7 ? 0.5
499 ? 24 481 ? 3734 ? 4c
33 ? 9
465 ? 3135 ? 636 ? 7
Adt-trkAha-gfp480 ? 35 451 ? 34 35 ? 533 ? 8
Uninfected 511 ? 2835 ? 7
AdlacZ-gfp8 ? 0.7
16 ? 1e
8 ? 0.8
8 ? 1
7 ? 1
8 ? 1
7 ? 1
22 ? 1f
7 ? 1
539 ? 49 521 ? 5061 ? 957 ? 5
569 ? 51 535 ? 5658 ? 9 53 ? 7
Adt-trkAha-gfp561 ? 37547 ? 29 62 ? 6 58 ? 10
Uninfected 538 ? 58 60 ? 4
360 ? 17f
54 ? 6
506 ? 2458 ? 7
aInfection at P2, evaluation 2±4 weeks later.
bAll values reported as mean ? SEM (n junctions/n mice). The number of discrete AChR regions and junction area were determined only in
junctions that were entirely en face.
cAt least 25 fibers in each of three muscles were measured.
dInfection at 8±12 weeks, evaluation 2±4 weeks later.
eSignificantly different from GFP?and uninfected muscle fibers; Student's t test, p ? 0.005.
fSignificantly different from wild-type muscle fibers; Student's t test, p ? 0.005.
possibility that overexpression of any membrane-asso-
ciated protein might cause structural disruption of junc-
tions and to evaluate the specificity of trkB-mediated
interactions. AdlacZ-gfp encodes the biologically inac-
tive protein ?-galactosidase (?-gal) and GFP (Figure 3A)
and was used as a control for nonspecific effects of
viral infection, transgene expression, and other factors,
suchas inflammationdue to the mounting of animmune
response to viral proteins.
Western blot analysis of 293 cells infected with either
AdtrkB.t1ha-gfp, Adt-trkAha-gfp, orAdlacZ-gfp showed
that trkB.t1ha, t-trkA, LacZ, and GFP were expressed
at the predicted molecular weight (Figure 3B). Immuno-
staining in infected 293 cells and rat myoblasts showed
that trkB.t1ha or t-trkAha and GFP were coexpressed
in infected cells (Figure 3C). Of GFP?cells, 87% ? 2%
also expressed trkB.t1ha, 91% ? 3% of GFP?cells also
expressed t-trkAha, and 92% ? 2% of GFP?cells also
expressed ?-gal (n ? 3 wells, ?100 cells per well for
each condition, mean ? SEM). This suggested that GFP
fluorescence can be used as a high-fidelity indicator of
adenovirus-mediated expression of the first transgene
1998). After infection of PC12 cells expressing trkB.FL
(PC12-TrkB?) with AdtrkB.t1ha-gfp, significantly fewer
GFP?cells extended neurites when exposed to BDNF
compared withuninfected controlcells (Figure 3D). Sim-
ilar results were observed with NT-4/5 treatment (data
not shown). No difference in neurite extension was ob-
served betweenPC12-TrkB?cells infected withAdlacZ-
gfp anduninfected cells (Figure3D).Regardless ofinfec-
tion, allPC12-TrkB?cells responded withsimilarneurite
outgrowth to nerve growthfactor(NGF; Figure 3E). After
infectionofPC12cells expressing full-lengthtrkA (PC12)
with Adt-trkA-gfp, significantly fewer GFP?cells ex-
tended neurites when exposed to NGF compared with
uninfected control cells (Figure 3F). This demonstrates
that adenovirus-mediated overexpression of trkB.t1ha
or t-trkAha results in a dominant-negative decrease in
neurotrophin signaling through endogenous full-length
Trk receptors, as has been reported in other systems
(Eide et al., 1996; Li et al., 1998).
In Vivo Overexpression of Truncated TrkB in
Neonatal Muscle Fibers Induces Disruption
of Neuromuscular J unctions
To determine the role of neurotrophin signaling through
postsynaptically localized TrkB at neuromuscular junc-
tions in vivo, the sternomastoid muscle of P2 mouse
pups was exposed to 10 ?l of 1 ? 1011plaque-forming
units (pfu) of AdtrkB.t1ha-gfp, Adt-trkAha-gfp, or Ad-
lacZ-gfp and allowed to survive for2±4weeks. Examina-
tion of cross sections through the endplate band of
Overexpression of TrkB.t1 Has a Dominant-Negative
Effect on BDNF- and NT-4/5-Mediated Signaling
A PC12 cell culture neurite extension assay was used
to evaluate whether viral expression of truncated Trks
could interfere with neurotrophin signaling in a domi-
nant-negative manner (cf. Eide et al., 1996; Li et al.,
TrkB-Mediated Signaling at Neuromuscular J unctions
Figure 5. In Vivo Observation of GFP?Mus-
cle Fibers and SubsequentAnalysis by Immu-
nostaining and Confocal Microscopy
Two to fourweeks afterAdlacZ-gfp infection,
the sternomastoid muscle in living mice was
exposed and imaged.
(A) 4-Di-2-Asp-stained motor nerve terminals
(yellow green) were visible on GFP?muscle
fibers (green, two fibers).
(B) R?BTX-stained AChRs were also imaged;
red patches are out-of-focus junctions.
(C)Afterremovaland fixation, the sternomas-
toid muscle was immunostained, using anti-
neurofilament and anti-SV2 antigen antibod-
ies, and confocal microscopy was used to
visualize nerve terminals.
(D) The same junctions were relocated based
on the unique pattern of AChR-rich regions.
Scale bar, 40 ?m.
TrkA antibodies demonstrated that in the majority of
fibers that expressed GFP, TrkB.t1ha (Figures 4A±4C)
or t-trkA (Figures 4D±4F) was also detected. TrkB.t1ha
and t-trkAha were localized to the muscle fiber mem-
brane (Figures 4B and 4E), including in and around neu-
romuscular junctions revealed by R?BTX staining (data
not shown). Thus, adenovirus-mediated transgene ex-
pressioncould be readily detected afterinvivoinfection.
Examination of hematoxylin and eosin± (H and E±)
stained sections of infected and control neonatal mus-
cles showed no evidence of inflammation or muscle
fiberde- orregeneration, such as foci of small, atrophic,
or angulated fibers or fibers with central nuclei (Figure
4G). No evidence of atrophy or hypertrophy was ob-
served in GFP?compared with GFP?muscle fibers in
infected muscles, oramong infected muscles compared
with uninfected control muscles (Table 1). Thus, adeno-
virus-induced myopathy was not likely to indirectly in-
duce structural changes in neuromuscular junctions.
In a similarset of experiments, animals were anesthe-
tized, the sternomastoid muscle exposed, and muscle
fibers examined forGFP fluorescence in vivo with meth-
ods previously described (Balice-Gordonand Lichtman,
1993). GFP was observed in 5%±30% of muscle fibers
on the superficial muscle surface. GFP or expression of
the other transgenes was never observed in connective
tissue, terminal Schwann cells, blood vessels, or other
tissues. Vital staining and imaging of motornerve termi-
nals and AChR-rich regions were used to visualize neu-
romuscular junctions on GFP?muscle fibers (Figures
5A and 5B). Muscles were then processed for immuno-
histochemistry, and the same junctions were relocated,
using the unique pattern of AChR-rich areas and confo-
cal microscopy (O'Malley et al., 1999; Figures 5C and 5D).
In 35 of 37 junctions on AdlacZ-gfp-infected fibers
that were GFP?, the spatial relationship among terminal
Schwanncells, motornerve terminals, and postsynaptic
AChR-rich regions was indistinguishable from those at
junctions on GFP?, uninfected fibers in the same mus-
cles, fibers fromuninfected muscles, orfibers frommus-
cles exposed to vehicle (n ? at least 10 junctions from
each of three mice in each condition; Figures 6A±6C).
Few ifany muscle fibers (2 of 37)appeared to have been
damaged and regenerated. Among GFP?and GFP?fi-
bers in muscles exposed to virus, and muscle fibers in
uninfected control muscles, no differences in muscle
fiber diameter or junction area (Table 1) were observed.
These data show that neuromuscular junctions on ade-
novirusinfected muscle fibers expressing biologically in-
active marker proteins such as ?-gal or GFP are not
Incontrast, neuromuscularjunctions on GFP?muscle
fibers infected with AdtrkB.t1ha-gfp 2±4 weeks pre-
viously showed several characteristic alterations when
compared with junctions on GFP?muscle fibers, which
were similar to junctions from uninfected control mus-
cles. In ?85% of junctions, the postsynaptic membrane
contained many small, punctate AChR regions (Figures
6D±6F). In ?50% of these junctions, these punctate re-
gions consisted of areas of normal receptor density in-
terspersed with fainter areas (Figures 6D±6I, asterisks).
Figure 6. Disassembly of Postsynaptic AChR Regions at Neuromuscular J unctions on AdtrkB.t1ha-gfp-Infected Neonatal Muscle Fibers
(A±C) Immunostained nerve terminals ([A], green) and R?BTX-stained AChRs ([B], red), and the alignment of pre- and postsynaptic specializa-
tions ([C], overlay) at neuromuscular junctions on GFP?muscle fibers from sternomastoid muscles infected with AdlacZ-gfp at P2. J unction
area and pre- and postsynaptic alignment are similar to those observed in age-matched uninfected muscles.
(D±L) Neuromuscular junctions on GFP?muscle fibers from muscles infected with AdtrkB.t1ha-gfp at P2.
(D±F) In ?85% of junctions on GFP?muscle fibers, the postsynaptic membrane contained many small, punctate AChR regions (E), which in
many cases were completely occupied by motor nerve terminals ([D], overlay in [F]).
(G±I) In ?50% of these junctions, regions of high and low receptor density were observed. Faint AChR regions are a hallmark of synaptic
sites in the process of being lost (cf. Balice-Gordon and Lichtman, 1993); one faintly stained region of AChRs is indicated with an asterisk,
although several are present.
(J ±L) In ?30% of junctions, no motor nerve terminal staining was detected over small, punctate, disrupted AChR regions. The diffuse green
staining in (J ) is due to residual GFP plus nonspecific staining due to the secondary antibody, commonly seen in whole-mount preparations.
(M±O) Motor nerve terminals (M) and AChRs (N), and the alignment of pre- and postsynaptic specializations ([O], overlay) at neuromuscular
junctions on GFP?muscle fibers from sternomastoid muscles infected with Adt-trkA-gfp at P2. J unction area and pre- and postsynaptic
alignment are similar to those observed in uninfected muscles. Scale bar, 20 ?m.
(P) Histogram of the number of discrete AChR regions per junction from uninfected control sternomastoid muscle (upper) and AdlacZ-gfp-
(upper middle), AdtrkB.t1ha-gfp- (lower middle), or Adt-trkA-gfp- (lower) infected muscles. The distribution from AdtrkB.t1ha-gfp-infected
muscle fibers is significantly different from control or muscle fibers infected with other adenoviruses (p ? 0.005, Komolgorov-Smirnoff test).
Histograms from control and muscle fibers infected with other adenoviruses are not significantly different. The number of discrete AChR
regions counted in the junction in (B) (left to right), 12, 6, and 7; in (E), 23; in (H), 19; in the three junctions shown en face in (K), ?35, ?35,
and 14; and in (N), 3 and 4.
TrkB-Mediated Signaling at Neuromuscular J unctions
Figure 7. Disassembly of Postsynaptic AChR Regions at Neuromuscular J unctions on AdtrkB.t1ha-gfp-Infected Adult Muscle Fibers
(A±C) Anti-S-100 immunostaining of terminal and myelinating Schwann cells (A), immunostained nerve terminals (B), and R?BTX-stained
AChRs (C) at neuromuscular junctions on GFP?muscle fibers from adult SCID mouse sternomastoid muscle infected with AdlacZ-gfp 2 weeks
previously. The size and alignment of pre- and postsynaptic specializations are similar to those seen in age-matched uninfected muscles.
(D±I) Neuromuscular junctions on GFP?muscle fibers infected with AdtrkB.t1ha-gfp 2 weeks previously.
(D±F) In about half of the junctions on GFP?fibers, Schwann cells are present (D), but motor nerve terminals (E) only partially occupy punctate
and disrupted postsynaptic AChR regions (F) that contain faint AChR regions (one of several indicated with an asterisk).
(G±I) In about half of the junctions on GFP?fibers, Schwann cells (G) and motor nerve terminals (H) are present over punctate postsynaptic
AChR regions (I) that contain faint AChR regions (one of several indicated with an asterisk).
(J ±L) Schwann cells (J ), motor nerve terminals (K), and AChRs (L) appear normal at neuromuscular junctions on GFP?muscle fibers infected
with Adt-trkAha-gfp 2 weeks previously. J unction area and pre- and postsynaptic alignment are similar to those observed in age-matched
uninfected muscles. Scale bar, 20 ?m.
(M) Histogram of the number of discrete AChR regions per junction from AdlacZ-gfp- (upper), AdtrkB.t1ha-gfp- (middle), and Adt-trkA-gfp-
(lower) infected muscles. The distribution from uninfected control mice is shown in Figure 8J . The distribution from AdtrkB.t1ha-gfp-infected
muscle fibers is significantly different from control muscle fibers or muscle fibers infected with other adenoviruses (p ? 0.005, Komolgorov-
Smirnoff test). Histograms from control muscle fibers and muscle fibers infected with other adenoviruses are not significantly different. The
number of discrete AChR regions counted in the junction in (C) (left to right), 8, 5, and 5; in (F), 15; in (I), 23; and in (L), 8 and 7.
Previous work has shown that faint AChR regions are in
the process of being eliminated from junctions (Balice-
Gordon et al., 1993; Rich et al., 1994). In many of these
cases, postsynaptic AChR regions were completely
(Figures 6D±6F) or partially (Figures 6G±6I) occupied
by motor nerve terminals. In ?30% of junctions, the
disruptionofpostsynaptic AChR regions and withdrawal
of nerve terminals were extreme (Figures 6J ±6L). The
postsynaptic AChR regions at such junctions contained
many small, grape-like clusters of receptors (Figure 6K),
and these junctions were typically not occupied by mo-
tor nerve terminals (Figure 6J ) and were thus dener-
vated.This rangeofpostsynaptic AChR disruptionprob-
ably reflects the range of expression levels of trkB.t1ha,
as cross-sectional analyses showed that some muscle
fibers expressed high levels, while others expressed
relatively low levels, of transgene (Figure 4). In ?15%
of GFP?fibers, no abnormalities in postsynaptic AChR-
rich regions or in motor nerve terminals or axons were
noted, probably because of relatively low levels of
trkB.t1ha expressionatjunctions.Because inmostjunc-
tions, punctate AChR regions were still occupied by
motor nerve terminals and terminal Schwann cells, we
infer that on fibers overexpressing trkB.t1 postsynaptic
regions were disassembled and that this was followed
by the retraction of motor nerve terminals and terminal
The disassembly of postsynaptic AChR regions was
evaluated by counting the number of discrete receptor
regions in junctions from infected compared with con-
trol muscle fibers. Neuromuscular junctions on GFP?
fibers fromAdtrkB.t1ha-gfp-infected muscles contained
a 3-fold greater number of small, discrete, punctate
AChR regions (Figure 6P, lower middle) compared with
junctions from GFP?fibers from AdlacZ-gfp-infected
muscles (Figure 6P, uppermiddle) oruninfected control
muscles (Figure 6P, upper; Table 1). The 3-fold increase
in AChR region number in AdtrkB.t1ha-gfp-infected fi-
bers was not accompanied by an increase in junction
area (Table 1). These results indicate that the overex-
pression of trkB.t1 in neonatal muscle fibers leads to
the disassembly and loss ofpostsynaptic AChR regions.
To determine whether disruption of postsynaptic
AChR regions was specific to overexpression of TrkB
ormight be due to nonspecific disruption resulting from
overexpression of any membrane protein, junctions
from GFP?muscle fibers from Adt-trkAha-gfp muscle
fibers were also evaluated. Neuromuscular junctions
from GFP?fibers in Adt-trkAha-gfp-infected muscles
(Figures 6M±6O) were indistinguishable from junctions
of GFP?fibers, junctions in AdlacZ-gfp-infected mus-
cles, or junctions in uninfected control mice in terms
of the alignment of motor nerve terminals and AChR
regions, junctionarea, and the numberofdiscrete AChR
regions (Figure 6P, lower; Table 1). Thus, neitheroverex-
pression of a membrane protein nor disruption of TrkA-
mediated signaling perse affects the structural integrity
of postsynaptic AChR regions. These observations indi-
cate that the specific disruption of TrkB-mediated sig-
naling in neonatal muscle fibers leads to the disassem-
bly and loss of postsynaptic AChR regions.
markerproteins, such as ?-gal or GFP, are not structur-
In contrast, 2 weeks after infection with AdtrkB.t1ha-
gfp, 74% (86 of 116 total junctions evaluated) of junc-
tions on GFP?fibers had AChR regions that were punc-
tate and contained regions of heterogeneous receptor
density(Figures 7D±7I;examples offaintregions marked
withasterisks)and inthis respectwere similarinappear-
ance to neonatal junctions on muscle fibers overex-
pressing trkB.t1. In about half of these cases, postsyn-
aptic AChR regions were partially occupied by motor
nerve terminals and terminalSchwanncells (Figures 7D±
7F), while in the other half, receptor regions were more
completely occupied (Figures 7G±7I). As in neonatal
junctions, these observations suggest that the overex-
pression of trkB.t1 in adult muscle fibers leads to the
disassembly of postsynaptic AChR regions, which in
turn may lead to changes in overlying nerve terminals
and Schwann cells.
Neuromuscular junctions on GFP?fibers from Ad-
trkB.t1ha-gfp-infected muscles contained a ?2-fold
greater number of small, discrete, punctate AChR re-
gions (Figure 7M, middle) compared with junctions from
GFP?fibers from AdlacZ-gfp-infected muscles (Figure
7M, upper) or uninfected control muscles (Figure 8J ,
white bars; Table 1). The 2-fold increase in AChR region
number in AdtrkB.t1ha-gfp infected fibers was not ac-
companied by an increase in junction area (Table 1).
Neuromuscular junctions from GFP?fibers in Adt-
trkAha-gfp-infected muscles, however, were indistin-
guishable from their counterparts in AdlacZ-gfp-infected
muscles or from junctions in uninfected control mice
(Figures 7J ±7L) in terms of the alignment of synaptic
components, junction area, and the number of discrete
AChR regions (Figure 7M, lower; Table 1). Thus, neither
overexpression of a membrane protein nor disruption
of TrkA-mediated signaling per se affects adult neuro-
muscular synaptic maintenance. Taken together, these
observations indicatethatTrkB-mediatedsignaling modu-
lates the maintenance of postsynaptic AChR regions
not only in development but also throughout adulthood.
Overexpression of TrkB.t1 Induces Disassembly
of Adult Neuromuscular J unctions
Given that several lines of evidence suggest that neona-
tal neuromuscular synapses and muscle fibers are de-
pendent on trophic interactions for maintenance, but
that trophic interactions become relatively less impor-
tant in adult animals, we asked whetheroverexpression
of trkB.t1 affected postsynaptic AChR regions at neuro-
muscular junctions from adult mice. Immune-compro-
mised (SCID) adult mice (Bosma et al., 1983) were used
to avoid an immune response that is prominent when
adultanimals are exposed to adenovirus (cf. Kass-Eisler
et al., 1994). Eight 12-week-old SCID mice were infected
with adenovirus and evaluated 2±4 weeks later. No evi-
dence of inflammation or muscle fiber de- or regenera-
tion was observed, such as foci of small, atrophic, or
angulated fibers orfibers with central nuclei, in infected
muscles (Figure 4H); thus, adenovirus-induced myopa-
thy was not likely to induce structural changes in neuro-
Neuromuscular junctions on GFP?muscle fibers in
sternomastoid muscles infected with AdlacZ-gfp were
similar in appearance (Figures 7A±7C) to junctions on
GFP?muscle fibers and to junctions from uninfected
control muscles. Between infected and uninfected con-
trol muscle fibers, no differences in muscle fiberdiame-
terorjunction area (Table 1) were observed. These data
show thatadultneuromuscularjunctions onadenovirus-
infected muscle fibers expressing biologically inactive
Postsynaptic AChR Regions Are Punctate
The experiments described above show that when sig-
naling through endogenous TrkB receptors is disrupted
in a dominant-negative fashion, using adenovirus to
overexpress trkB.t1, postsynaptic AChR regions aredis-
assembled.Thus, itwas ofinterestto determinewhether
micedeficientinTrkB expressionshow similarstructural
disruptions of neuromuscular junctions. Mutant mice
that lack TrkB (trkB?/?) do not survive past the time of
birth and thus were not useful for addressing this issue.
We asked whether synaptic maintenance was altered
in heterozygous mutant mice expressing half as much
trkB.FL as wild-type controls (Kleinetal., 1993).Western
blot analysis confirmed that trkB.FL expression was re-
duced in trkB?/?compared with trkB?/?mouse brain
(Figure 2F). As reported by Klein et al. (1993), a more
modest reduction in trkB.t1 was also observed. Neuro-
muscular junctions in trkB?/?mice and in age- and
sex-matched trkB?/?controls ofthe same genetic back-
ground were stained, imaged, and analyzed with confo-
cal microscopy and interactive software. No evidence
TrkB-Mediated Signaling at Neuromuscular J unctions
Figure 8. Neuromuscular J unctions in trkB?/?
Mice Have Numerous Punctate Postsynaptic
(A±F) In neuromuscular junctions from the
sternomastoid muscle of trkB?/?mice, termi-
nal Schwann cells (A and D) and motor nerve
terminals (B and E) are present overpostsyn-
aptic AChR regions (C)thatcontainnumerous
punctate AChR regions (C and F). (G±I)Termi-
nal Schwann cells (G), motor nerve terminals
(H), and postsynaptic AChR regions (I) from
age- and sex-matched wild-type mice. The
number of discrete AChR regions counted in
the junction in (C), 8; in (F), 48; and in (I), 26.
Scale bar, 10 ?m.
(J )Histogramofthe numberofdiscrete AChR
regions per junction in trkB?/?mice (black
bars) is significantly different from that from
age- and sex-matched trkB?/?mice (white
bars; p ? 0.001, Komolgorov-Smirnoff test).
of de- or regeneration of muscle fibers was observed,
such as foci of small, atrophic, or angulated fibers or
fibers with central nuclei, from examination of H and
E±stained cross sections (Figure 4I), and muscle fiber
diameter in trkB?/?was similar to that found in trkB?/?
mice (Table 1). Thus, no evidence of myopathy, which
itself canresult in structural disruptions of neuromuscu-
lar junctions, was apparent in trkB?/?mice.
Neuromuscular junctions were, however, strikingly
different in trkB?/?compared with trkB?/?adult mice.
Whereas junctions from trkB?/?animals consist of con-
tiguous branches of postsynaptic AChRs (Figures 8A±
8C), junctions from trkB?/?mice contained many small,
discrete AChR regions (Figures 8D±8I). Motornerve ter-
minals and terminal Schwann cells were present over
postsynaptic AChR regions intrkB?/?mice, as intrkB?/?
mice.J unctions fromtrkB?/?sternomastoid musclecon-
tained a 3-fold greater number of small, discrete, punc-
tateAChR regions compared withthosefromtrk?/?mice
(Figure8J ;Table1).Theincrease inAChR regionnumber
was accompanied by a decrease in junction area in
trkB?/?mice (Table 1). The increase in the number of
discrete AChR regions per junction in trkB?/?mice is
similar to the punctate appearance of receptor regions
at junctions on muscle fibers overexpressing trkB.t1.
naling through trkB.FL is decreased, either by adenovi-
rus-mediated somatic gene transfer or by gene tar-
geting, postsynaptic regions are disassembled.
Overexpression of TrkB.t1 Results in
the Disassembly of Agrin-Induced
AChR Clusters in Myotubes
Because effects on motor nerve terminals and terminal
Schwann cells appeared to be variable in AdtrkB.t1ha-
gfp-infected and trkB?/?mice, we asked whether TrkB-
dependent disassembly of postsynaptic AChR regions
requires the presence of terminal Schwann cells and/or
motor neurons. Adenovirus was used to overexpress
trkB.t1 in an in vitro agrin-induced AChR clustering
Figure 9. Disassembly ofAgrin-InducedAChR
Clusters on Myotubes Overexpressing TrkB.t1
(A) Three days of agrin treatment results in
the formation of large AChR clusters ?4 ?m
in length, visualized by R?BTX staining (red;
three clusters on edge are present). Inset
shows a large, contiguous en face AChR
(B) In AdlacZ-gfp-infected myotubes, large
AChR clusters (red; one on edge cluster is
present), similar to those observed in unin-
fected myotubes, are present.
(C) In AdtrkB.t1ha-gfp-infected myotubes,
AChR clusters consisted of aggregations of
many small, punctate regions, as opposed
to the large, contiguous clusters seeninunin-
fected control myotubes (inset in [A]) or in
AdlacZ-gfp-infected myotubes. Scale bar,
(D) AdtrkB.t1ha-gfp-infected myotubes have
significantly fewer AChR clusters ?4 ?m in
length compared with uninfected or AdlacZ-
gfp-infected myotubes (p ? 0.001, Student's
t test). Because agrin-induced clusters were
present prior to infection, these results sug-
gest that overexpression of trkB.t1 induces
clusters. Bars indicate mean ? SEM.
assay in myotube cultures in which motor neurons and
Schwanncells were notpresent. Myotubes were treated
with neural agrin for 3 days, which induced large AChR
clusters on most myotubes (contiguous AChR cluster
length, 4.1 ? 0.5 ?m, mean ? SEM), similar to those
On day 2 of agrin treatment, myotubes were treated
withvehicle, AdlacZ-gfp orAdtrkB.t1ha-gfp, and cluster
number, size, and appearance were assayed after one
more day of agrin treatment.
On day 3, the numberand size of agrin-induced AChR
clusters was similar in uninfected (2 ? 0.2 clusters per
myotube, 4 ? 0.3 ?m in length) or AdlacZ-gfp-infected
(1.9 ? 0.2 clusters per myotube, 4 ? 0.2 ?m in length)
myotubes (not significantly different, Student's t test,
p ? 0.50; Figure 9D). However, in GFP?myotubes over-
expressing trkB.t1, existing agrin-induced AChR clus-
ters were disrupted. The numberof large AChR clusters
was significantly reduced (0.5 ? 0.1) compared with
GFP?myotubes in the same cultures (1.9 ? 0.3) or with
uninfected orAdlacZ-gfp-infected myotubes (p ?0.001,
Student's t test; Figure 9D). Moreover, the appearance
of remaining clusters on GFP?myotubes (Figure 9C)
was strikingly different from those on GFP?, uninfected
(Figure 9A) or AdlacZ-gfp-infected (Figure 9B) myo-
tubes. The agrin-induced clusters on GFP?myotubes
were punctate and disrupted in appearance (Figure 9C)
compared with the clusters on control myotubes, which
consisted of a contiguous patchof uniformly distributed
AChRs (Figure 9A, inset). Thus, agrin-induced AChR
clusters were disrupted on myotubes overexpressing
trkB.t1, and this disruption occurred in the absence of
motornerve terminals and Schwann cells. These obser-
vations show that autocrine-mediated signaling in mus-
cle fibers can affect the maintenance of postsynaptic
We report that TrkB-mediated signaling plays an impor-
tant role in the ongoing maintenance of neuromuscular
junctions during postnatal life. Immunostaining, West-
ern blot analysis, and RT-PCR were used to show that
trkB.FL is localized to the postsynaptic muscle fiber
membrane. Dominant-negative manipulation of TrkB-
mediated signaling by overexpressionof trkB.t1 in mus-
cle fibers resulted in the disassembly of postsynaptic
AChR regions in neonates as well as adults. A similar
postsynaptic disassembly was apparent in neuromus-
cular junctions from mutant mice expressing half as
much full-length TrkB as wild-type mice. The specificity
of disruption of TrkB-mediated signaling in inducing the
disassembly of postsynaptic AChR regions was sup-
ported by the observation that junctions on Adt-trkAha-
gfp-infected muscle fibers were normal in appearance.
In myotube-only cultures, overexpression of trkB.t1 re-
sulted in the disassembly of agrin-induced AChR clus-
ters, suggesting thatTrkB-mediated signaling canaffect
postsynaptic AChR clusters even in the absence of mo-
tor nerve terminals and Schwann cells. Taken together,
our results suggest that neurotrophin signaling through
TrkB receptors on muscle fibers is necessary for the
maintenance of postsynaptic AChR regions at neonatal
as well as adult neuromuscular junctions.
Postsynaptic Localization of TrkB Isoforms
Previous work utilizing NorthernorWesternblotanalysis
showed that trkB.t1, BNDF, and NT-4/5 are expressed
in skeletal muscle and spinal cord, and trkB.FL was
expressed in spinal cord but not muscle (Funakoshi et
al., 1993; Griesbeck et al., 1995; Shelton et al., 1995;
Yamamoto et al., 1996). However, the cellular localiza-
tion of TrkB or TrkB ligands cannot be determined from
TrkB-Mediated Signaling at Neuromuscular J unctions
these approaches. In the present work, Western blot
analyses were used to confirmthe specificity ofantibod-
ies used to detect TrkB isoforms in skeletal muscle and
brain. Somewhat different levels of trkB.FL and trkB.t1
were detected in muscle, depending on the antibody
used, but expression was not quantified. Immunostain-
ing showed that trkB.FL and trkB.t1 are localized to
the postsynaptic membrane and are concentrated at
neuromuscular junctions at and around AChR-rich re-
gions (Figure 1). Given that trkB.FL is expressed only at
neuromuscular junctions, the fact that junctional area
is ?0.5% of the muscle fiber surface area might have
precluded its detection by whole-tissue approaches.
While expression persisted after short-term denerva-
tion, we have observed a decrease inTrkB expressionat
longer intervals after denervation (M. G. and R. J . B.-G.,
unpublished data), suggesting thatTrkB levels inmuscle
fibers may be regulated by innervation. We were sur-
prised to observe that relatively little trkB.FL immunore-
activity was detected presynaptically in motoraxon and
terminal branches, given that motor neurons express
trkB.FL mRNA and protein(Funakoshietal., 1993;Gries-
beck et al., 1995). It is possible that only a small fraction
of the trkB.FL made by motor neurons is transported to
the periphery or, alternatively, it is not stably located in
motor axons or nerve terminals. In primary myotubes
without motor innervation, RT-PCR amplified trkB.FL
mRNA from mouse primary myotubes; while such cul-
tures may contain a few fibroblasts and Schwann cells,
these cell types do not express trkB.FL (Funakoshi et
al., 1993). The observation that trkB.FL mRNA and pro-
tein are expressed in muscle cells is consistent with a
recent report of trkB.FL expression by primary chick
myotubes, using RT-PCR (Wells et al., 1999). Taken to-
gether, this work suggests that TrkB is expressed by
muscle fibers and is localized to neuromuscular junc-
immunostaining, the disassembly of postsynaptic AChR-
rich regions appeared to vary in severity. The disassem-
bly of receptor regions appears to be followed in some
cases by retraction of motor nerve terminals. Thus, in
addition to muscle fiber±specific effects, whichare con-
sidered below, there may also be anterograde and/or
retrograde interactions among terminal Schwann cells,
motor nerve terminals, and muscle fibers via TrkB lo-
cated pre- and postsynaptically; these in turn may
modulate different aspects of synaptic structure and
function.Thatweobserved a similardisruptionofneuro-
muscular junctions in both neonates and adults argues
that neurotrophinsignaling throughTrkB plays anongo-
ing role in the maintenance of postsynaptic AChR re-
In trkB?/?mice, there is a reduction of trkB.FL, as well
as, to a lesser extent, trkB.t1 (Klein et al., 1993). While
the decrease in trkB.t1 may contribute to the junction
changes we observed, no signaling role has beenidenti-
fied for this isoform. The structural abnormalities seen
in trkB?/?animals are similar to those in AdtrkB.t1ha-
gfp-infected muscle fibers, in that junctions contain ag-
gregations of many small, punctate AChR regions as
opposed totheelongated branches ofreceptors thatare
present in junctions from wild-type mice. The simplest
explanation for this similarity is that the disassembly of
postsynaptic receptor regions occurs when signaling
through trkB.FL is decreased. However, trkB?/?junc-
tions are different, in that faint AChR levels within junc-
tions are not observed, and the number of discrete re-
gions is somewhat greater than in AdtrkB.t1ha-gfp
muscles. This is probably because the developmental
history of trkB?/?animals is different fromthat ofadeno-
virus-infected animals; the former have reduced TrkB
throughout development, while the latter have experi-
enced reduced TrkB signaling only after developmental
events, such as synapse formation, have occurred in a
normal background. Motor neurons may also be af-
fected in trkB?/?mice, and this may in turn affect syn-
apse formation and maintenance apart from effects on
muscle fibers, although motor neuron numberis normal
in trkB?/?mice (Klein et al., 1993). Despite these differ-
ences, the disruptionofjunctions inbothcircumstances
highlights the ongoing requirement for TrkB-mediated
signals in neuromuscular synaptic maintenance.
Given the likelihood that there is multidirectional sig-
naling mediated by neurotrophins and other molecules
at neuromuscular junctions, as at other synapses (re-
viewed by Altar and DiStefano, 1998), it is interesting
that overexpression of trkB.t1 on myotubes in vitro
caused the disruption of existing agrin-induced AChR
clusters in the absence of motor neurons and Schwann
cells. This argues strongly that TrkB-mediated signaling
inmuscle fibers perse is required forthe maintenance of
AChR clusters. Recent work from Fallon and colleagues
(Wells et al., 1999) showed that both BDNF and NT-4/5
caninhibitthe formationofagrin-induced AChR clusters
in chick myotubes in culture and that direct activation
of muscle trkB.FL by anti-TrkB antibodies mimicked
neurotrophin inhibition of AChR clustering. While the
work by Wells et al. (1999) on cluster formation in vitro
and the work we present here on the disassembly of
existing clusters in vivo and in vitro appear contradic-
tory, they are likely to be two aspects of the same pro-
cess: neurotrophins can modulate agrin's ability to form
TrkB-Mediated Signaling Modulates Postsynaptic
AChR Clusters at Neuromuscular J unctions
The postsynaptic localization of trkB.FL raises several
interesting implications for Trk-mediated signaling at
neuromuscular junctions. An extensive literature sup-
ports the classic notion of target-derived trophin affect-
ing presynaptic neurons that express the trophin recep-
tor. Because some TrkB expression was detected
presynaptically, it is likely that retrograde signaling from
muscle fibers to motor nerve terminals does play a tro-
phic role in nerve±muscle maintenance. The expression
of trkB.t1 in Schwann cells (Funakoshi et al., 1993) and
our observations that trkB.FL and trkB.t1 are localized
to postsynaptic AChR regions make it plausible that
there are multidirectionalsignals exchanged among ter-
minalSchwanncells, motornerve terminals, and muscle
fibers that are essentialforsynaptic maintenance. When
this signaling is disrupted, as inthe presentexperiments
by adenovirus-mediated somatic gene transfer or by
disrupting endogenous TrkB by mutagenesis, a robust
disassembly of postsynaptic AChR regions at neuro-
muscular junctions is observed. While we were unable
to correlate the extent of disruption with the extent of
trkB.t1ha expression on a junction-by-junction basis,
owing to difficulty in detecting the anti-HA immunoreac-
tivity in whole mounts necessary for viewing junction
Cruz), an anti-TrkB rabbit polyclonal antibody that recognizes both
trkB.FL and trkB.t1, an anti-trkB.FL-specific antibody, or an anti-
TrkA antibody. After incubation with secondary antibodies (alkaline
phosphatase± [AP-] conjugated goat anti-rabbit immunoglobulin G;
Tropix PE Biosystems) and chemiluminescent detection (CEP-Star
Kit, Tropix PE Biosystems), blots were exposed to film.
ForPCR analysis of TrkB mRNA expression in primary myotubes,
hindlimb skeletal muscle was removed from embryonic day 18 (E18)
C57/Bl6J mice, and myotube cultures were prepared (O'Malley et
al., 1997).After6days inculture, myotubes were harvested,and total
RNA was extracted (RNEasy kit, Qiagen, Chatsworth, CA). Myotube
RNA was reverse transcribed with murine leukemia virus reverse
transcriptase and oligo (dT) primers (Advantage RT-for-PCR kit,
ClonTech). The cDNA was PCR amplified by using primers specific
for the trkB.FL receptor. The upstream primer was 5?-GATAGTAGT
CGGTGCTGTAC-3?, and the downstream primer was 5?-GGAGCAT
CTCTCGGTCTATG-3?. The product of this PCR reaction was ampli-
fied furtherwith a nested upstream primer(5?-TGTTGGGACGCCAG
GTAGAC-3?) and a downstream primer (5?-AGTTGGCGAGACATTC
CAAG-3?). The 20 ?lPCR mixture contained 2 ?loffirst strand cDNA
or PCR product from the first PCR round, 2.0 ?l of 10? PCR buffer
(GIBCO-BRL), 2.5 mM MgCl2, 0.2 mM dNTP, 100 ng of each primer,
0.2 units of Taq polymerase (GIBCO-BRL), and sterile water. The
cycling conditions were 94?C for 4 min, 94?C for 30 s, 55?C for 30
s, and 72?C for 1 min for 30 cycles, followed by 72?C for 10 min. In
each case, total RNA and water were used as negative control
templates, and cDNA from brain was used as a positive control.
as well as to maintain AChR clusters. The mechanism
by which this occurs is at present unclear. TrkB.t1 over-
expression may disrupt the localization or signaling of
other molecules, for example, MuSK, which plays an
important role in agrin-induced AChR clustering (cf.
Glass et al., 1996; reviewed by Sanes and Lichtman,
1999). That the overexpression of a truncated TrkA with
the same short cytoplasmic tailas trkB.t1 does not itself
lead to disruption of AChR clusters argues that these
effects are TrkB specific, perhaps owing to interactions
with MuSK or other molecules that do not occur with
TrkB-mediated signaling may also play a role in the
loss of AChR regions that occurs during neonatal syn-
apse elimination (Balice-Gordon et al., 1993) and during
the synapse elimination that occurs during adult rein-
nervation (Rich and Lichtman, 1989). As innervation by
several motor axons is pruned to the mature pattern
of innervation by a single motor neuron, postsynaptic
AChRs are lost from beneath inputs that will be elimi-
nated but not from beneath the input that will remain.
The gradual disappearance of postsynaptic AChR re-
gions is followed by the retraction of overlying motor
nerve terminals and Schwann cells (Balice-Gordon et
al., 1993; Culican et al., 1998) and culminates in the
withdrawal of an input from the junction. Because the
disruption of TrkB-mediated signaling can induce loss
ofAChR regions fromneonatalas wellas adultjunctions,
it is possible that TrkB ligands play a role in the disas-
sembly of receptor regions that occurs during synapse
elimination. If this is the case, we speculate that overex-
pression of trkB.FL in muscle fibers might prevent or
attenuatetheloss ofAChR regions during synapseelimi-
nation in development and reinnervation. This will be
one aspect of future work.
Generation and Evaluation of Replication-Defective,
The pAdLink plasmid, containing the CMV promoter/enhancer, an
SV40 polyadenylation sequence, and flanking adenovirus backbone
sequences, was obtained fromDr. J . Wilson, the Institute forHuman
Gene Therapy, University of Pennsylvania. pAdLink was modified
by inserting multiple cloning sites, an IRES from pLIGns, (Lillien,
1995), and GFP (codon-corrected cDNA; GIBCO-BRL). cDNAs en-
coding othertransgenes were thencloned into this plasmid.Recom-
binant, replication-defective adenovirus was generated by homolo-
gous recombination with the viral Ad5, E1a-deleted dl327 backbone
in human embryonic kidney 293 stem cells that are permissive for
viral replication using standard methods (Precious and Russell,
The Escherichia coli lacZ gene encoding ?-gal and the gene for
GFP were cloned into pAdLink, and adenovirus was generated.
AdlacZ-gfp was used as a control for nonspecific effects of viral
infection. A mouse truncated TrkB cDNA (trkB.t1; (Klein et al., 1990;
obtained from Dr. M. Barbacid, Bristol Myers Squibb) was epitope
tagged at the carboxyl terminus of the protein with HA (Pati, 1992),
and this gene and the gene for GFP were cloned into the modified
pAdLink plasmid. Because no endogenous truncated TrkA isoforms
have been identified, a fragment of the full-length mouse trkA gene
(obtained from Dr. G. Yancopoulos, Regeneron Pharmaceuticals)
containing the extracellular, NGF-binding, and transmembrane do-
mains was fused in-frame to the cytoplasmic tail of the mouse
trkB.t1 cDNA. This construct was epitope tagged at the carboxyl
terminus of the protein with HA. This hybrid t-trkAha cDNA was
cloned into the modified pAdLink plasmid. Adt-trkAha-gfp virus was
used to evaluate the possibility that overexpression of a membrane
protein could nonspecifically induce structural disruption of neuro-
muscularjunctions and to determine the specificity of Trk-mediated
effects on junctions.
Purified virus was generated afterthree rounds ofplaque selection
by a limiting dilution method in 293 cells (Precious and Russell,
1985). The integrity of the viral genome was examined by Southern
blot, and the absence of wild-type Ad5 virus was confirmed by PCR
using primers specific to the deleted E1a region (Woo et al., 1997).
Virus was resuspended in HEPES-buffered saline (HBS [pH 7.8]) ?
10% glycerol, particle density was measured spectrophotometri-
cally at OD260, and pfu was determined by plaque assays on agar
overlays using a limiting dilution method. Virus aliquots of 1 ? 1012
pfu/ml were stored at ?70?C for ?4 months, and viral stocks were
stored in liquid N2.
To assess transgene expression, subconfluent HeLa cells or
Localization of Trks to Neuromuscular J unctions
Muscles were denervated by resectionof the sternomastoid muscle
nerve; 3 days later, denervation was confirmed by the absence of
motor axon and terminal immunostaining as described below (cf.
O'Malley et al., 1999). Whole mounts of innervated and denervated
muscles were stained with R?BTX to localize AChRs at junctions
and were also stained with one of several anti-TrkB antibodies and
an appropriate fluorescently conjugated secondary antibody: an
affinity-purified, rabbit polyclonal anti-TrkB antibody (Huang et al.,
1999; gift of Dr. L. Reichardt) that recognizes both truncated and
full-length isoforms of TrkB; an affinity-purified, rabbit polyclonal
anti-trkB.FL-specific antibody raised against amino acids 606±619
of the tyrosine kinase domain (Yan et al., 1997; gift of Dr. S.
Feinstein); or an affinity-purified, rabbit polyclonal anti-trkB.t1-spe-
cific antibody raised against a peptide from the carboxyl terminus
of TrkB (Yan et al., 1997; gift of Dr. S. Feinstein). In the cases of the
latter two antibodies, antibody incubation with the peptide used
as antigen prior to application to tissue sections resulted in no
detectable staining. J unctions were also stained with a rabbit poly-
clonal anti-TrkA raised against the ectodomain of rat TrkA (Clary et
al., 1994; gift of Dr. L. Reichardt).
Western blot analysis was performed with membrane prepara-
tions from brain, muscle, and cultured mouse primary myotubes
using standard methods. Protein (20 ?g) from each sample was
separated ona 10% SDS±PAGE geland transferred to nitrocellulose
membrane overnight (30 V, 40?C). Membranes were blocked for 1
hr at room temperature in 0.4% I-block casein and 0.1% Tween
20 in phosphate-buffered saline (PBS) and incubated in a primary
antibody: an anti-trkB.t1-specific rabbit polyclonal antibody (Santa
TrkB-Mediated Signaling at Neuromuscular J unctions
mouseprimary myotubes wereexposedtoAdlacZ-gfp,AdtrkB.t1ha-
gfp, Adt-trkAha-gfp, or vehicle (HBS ? 1% glycerol). Cells were
subsequently viewed with epifluorescence microscopy for GFP ex-
pression, processed histologically for ?-gal, with X-gal as a sub-
strate (Sanes et al., 1986), or processed for immunostaining with
antibodies against HA (mouse monoclonal, Boehringer-Mannheim),
TrkB, or TrkA and fluorescently conjugated secondary antibodies.
Western blot characterization of transgene expression was per-
formed using transfected and infected 293 cells and mouse primary
myotubes infected with AdtrkB.t1ha-gfp, Adt-trkAha-gfp, AdlacZ-
gfp, orvehicle, with specific antibodies against ?-gal (mouse mono-
clonal, Boehringer-Mannheim), GFP (mouse monoclonal, Clontech),
TrkB, TrkA, or HA. Mouse brain was used as a positive control for
TrkB and TrkA, recombinant GFP or ?-gal protein were used as
positive controls for adenovirus-driven GFP and ?-gal expression,
and unmanipulated cells were used as negative controls. Immuno-
positive bands onWesternblots were visualizedwithAP-conjugated
secondary antibodies, using the WesternStar kit (Tropix PE Bio-
possibility that adenovirus infection might induce inflammation or
other nonspecific cellular changes. The superficial surface of the
sternomastoid muscle was exposed to virus or vehicle in P1±P3
mice; 3 days to 2 months later, mice were sacrificed, muscles were
dissected and frozen in isopentane cooled in liquid N2, and 10±20
?m frozen sections were obtained (n ? at least 3 mice for each
virus at each time point; 3 animals infected with Adt-trkAha-gfp
examined 2±4 weeks later). Transgene expression in infected mus-
cles was evaluated by immunostaining as described above. Serial
sections were left unstained to examine the extent of expression of
GFP orstained withH and E to determine whetherinflammationwas
present. Sections were imaged and muscle fiberdiametermeasured
with interactive software.
Examination of H and E±stained sections showed no evidence of
inflammation or degeneration of infected or vehicle-treated muscle
fibers atany time pointexamined. We found thatintroducing adeno-
virus into P1±P3 C57/Bl6J mice resulted in transgene expression in
more than 30% of surface fibers and that transgene expression
persisted for as long as 2 months, the last time point examined.
Because there are more large, type IIB fibers on the superficial
surface of the mouse sternomastoid muscle than small, type IIA
fibers, somewhat more large fibers were GFP?than small fibers.
However, no difference in muscle fiber diameter was observed as
a result ofinfection ortransgene expression (Table 1). No qualitative
differences were observed in transgene expression across fiber
types. In contrast, mice of the CD1 strain or Wistar rats exposed to
adenovirus at the same ages, or C57/Bl6J mice exposed to virus at
P7 or as adults, did not show persistent expression, and extensive
inflammation and degeneration of infected muscle fibers was ob-
served. It is likely that host tolerance is more readily achieved in
the perinatal period in inbred C57/Bl6J mice than in the two other
commonly used rodent strains examined, and this is likely animpor-
tant factor in achieving persistent transgene expression. C57/Bl6J
mice were thus used for all neonatal experiments.
Analysis of Dominant-Negative Effect of Truncated
An in vitro assay was used to determine whether virally expressed
trkB.t1ha could decrease BDNF orNT-4/5 signaling through endog-
enous, full-length TrkB in a dominant-negative fashion (Eide et al.,
1996; Li et al., 1998). We used a stably transfected PC12 cell line
that expresses full-length TrkB (PC12-trkB; gift of Dr. M. Chao);
these cells extend neurites inthe presence of BDNF orNT-4/5. Cells
were plated at low-passage number and maintained in RPMI with
10% horse serum, 5% fetal bovine serum, penicillin (100 units)/
streptomycin (100 ?g), and 0.4 mg/ml G418 at 37?C in 5% CO2. One
day aftersplitting, cells were infected withAdtrkB.t1ha-gfp, AdlacZ-
gfp (2 ? 108pfu/104cells), or vehicle. Three days later, 1±100 ng/
ml BDNF, NT-4/5, NGF, or PBS was added to the medium for 5
days. Neurite extension by PC63 cells expressing TrkA (obtained
from Dr. R. Pittman and grown in medium lacking G418) to 1±100
ng/ml NGF was used as a positive control. PC63 cells were also
in response to NGF was determined to evaluate nonspecific effects
of viral infection. A similar assay was performed using PC12 cells
that endogenously express full-length TrkA to determine whether
virally expressed t-trkAha could decrease NGF signaling in a domi-
nant-negative fashion. At least 50 cells in each of four dishes were
evaluated for each condition.
In Vivo Delivery of Adenovirus to Adult Sternomastoid Muscle
Because adenovirus induces a robust immune response in adult
animals, mice with severe combined immunodeficiency (SCID) were
used. SCID mice lack a T and B cell±mediated immune response
(cf. Bosma et al., 1983) and thus do not mount an immune reaction
to viral proteins or transgenes encoding foreign proteins; they are
also a widely used mouse model for adult adenovirus infection (cf.
Le Gal Le Salle et al., 1993). Two to four weeks after applying 10±20
?l of 1 ? 1011pfu/ml AdlacZ-gfp, Adt-trkAha-gfp, AdtrkB.t1ha-gfp,
or vehicle to the superficial surface of the sternomastoid muscle
in 2- to 3-month-old SCID mice, mice were imaged and muscles
immunostained as described below.
In a second series of experiments, transgene expression was
examined 1 week to 2 months after infection in muscle cross sec-
tions as described above. We found that introducing adenovirus
into P1±P3 C57/Bl6J mice resulted intransgene expressionin?25%
of surface fibers and that transgene expression persisted for as
long as 2 months, the last time point examined. As in neonates,
because there are more large, type IIB fibers on the superficial
surface of the mouse sternomastoid muscle than small, type IIA
fibers, somewhat more large fibers were GFP?than were small
fibers. However, no difference in muscle fiber diameter was ob-
served as a result of infection or transgene expression (Table 1).
No qualitative differences were observed in transgene expression
across fiber types.
To determine whetherneuromuscularjunctions in the sternomas-
toid muscle of SCID mice were similar in theirstructure and mainte-
nance to those fromwild-type C57/Bl6J mice, superficialneuromus-
cular junctions were vitally stained and imaged twice at 2±4 week
intervals and then immunostained as described below. A compari-
son of vital images taken 2±4 weeks apart showed that SCID neuro-
muscularjunctions are as stably maintained as neuromuscularjunc-
tions from wild-type C57/Bl6J mice. No differences were observed
in terminal Schwann cells, presynaptic motor nerve terminals, or
postsynaptic AChR regions between views over these intervals.
In Vivo Delivery of Adenovirus to Neonatal
The superficial surface of the sternomastoid muscle in the ventral
neck of anesthetized P1±P3 C57/Bl6J mice was exposed as de-
scribed previously (Balice-Gordon and Lichtman, 1993). Ten to
twenty microliters of1 ? 1011pfu/mlofAdtrkB.t1ha-gfp, Adt-trkAha-
gfp, AdlacZ-gfp in HBS ? 1% glycerol, or vehicle was applied to
the surface of the muscle, as opposed to being injected into the
muscle. Muscle injection induced muscle fiber damage that itself
resulted in synaptic disassembly (M. G. et al., unpublished data;
see also Rich and Lichtman, 1989). Topical application of virus also
maximized the number of infected fibers on the superficial muscle
surface that could subsequently be analyzed.
Following virus application,the skinwas sutured,and whenrecov-
ered, pups were returned to their mother. Two to four weeks later,
animals were anesthetized, the sternomastoid muscle was exposed,
and superficial neuromuscular junctions were vitally stained and
imagedas describedpreviously (Balice-Gordonand Lichtman,1993;
Balice-Gordon, 1998). Presynaptic motor nerve terminals were
stained with4-Di-2-ASP, and postsynaptic AChRs were stained with
R?BTX. Images of vitally stained neuromuscular junctions on GFP?
and GFP?fibers were obtained withan epifluorescence microscope
(Leica DMRX), a low-light level SIT camera (Dage-MTIVE1000), and
a PC-based image processing system and interactive software
(MetaMorph; Universal Imaging). These images were used to relo-
cate junctions of interest following immunostaining (see below).
In one series of experiments, we determined the persistence of
adenovirus expression inneonatal C57/Bl6J mice and evaluated the
Neuromuscular J unction Immunostaining
Sternomastoid muscles were removed and processed as whole
mounts for immunostaining as previously described (Son and
Thompson, 1995; O'Malley et al., 1999). The primary antibodies
used were a mouse monoclonal anti-neurofilament SMI31 antibody
(Sternberger Monoclonals) and a mouse monoclonal anti-SV2 anti-
body (Developmental Studies Hybridoma Bank, Iowa) for motor
nerve terminals, and a rabbit polyclonal anti-S-100 antibody (Dako)
for Schwann cells, followed by incubation with fluorescently conju-
gated secondary antibodies. Insome experiments, anti-HA antibod-
ies were also used, and the expression of virally encoded Trk in
analyzed fibers was evaluated qualitatively. Because GFP fluores-
cence oftendid notpersistafterfixationand permeabilizationproce-
dures required for immunostaining, junctions on GFP?fibers were
relocated, using the unique pattern of postsynaptic AChRs imaged
prior to fixation (cf. O'Malley et al., 1999; Figure 4).
Immunostained neuromuscularjunctions were analyzed by confo-
cal microscopy (Leica TCS 4D system; 100?, 1.4 n.a. oil objective).
The figures show single plane projections of confocal stacks of
images. J unctions were evaluated based on the extent of alignment
between presynaptic motornerve terminals and postsynaptic AChR
regions, the presence of faint AChR regions, the number of discrete
AChR regions, and junction AChR area. Alignment between pre-
and postsynaptic specializations was evaluated by comparing im-
munostaining in overlays. The presence of faint AChR regions was
determined from images mapped between 1 and 254 gray levels by
measuring the average pixel intensity in different junction regions;
regions were classified as faint if their average intensity was less
than or equal to half of the total junction average. The number of
discrete AChR regions per junction was determined in a subset of
R?BTX-labeled junctions that were completely en face in orienta-
tion; regions were counted as discrete if the outer perimeterdid not
contact other regions within that junction. Normal junctions contain
long, uniformly stainedbranches ofAChR regions (cf.Balice-Gordon
and Lichtman, 1993; O'Malley et al., 1999), as opposed to numerous
small, punctate, discrete AChR clusters. J unction AChR area was
measured from en face images with interactive software. Similar
immunostaining, and analyses of junction area and number of dis-
crete AChR regions, were performed on the sternomastoid muscle
from trkB?/?mice and wild-type controls (C57/Bl6J background),
followed by confocal microscopy. Muscle fiber morphology and
diameter were evaluated in H and E±stained cross sections.
the National Institutes of Health (NS34373, AG13329), a McKnight
Neuroscience Scholar award to R. B.-G., and a National Institutes
of Health National Research Service Award (NS10745) to M. G.
Received J une 14, 1999; revised September 24, 1999.
Altar, C.A., and DiStefano, P.S. (1998). Neurotrophin trafficking by
anterograde transport. Trends Neurosci. 21, 433±437.
Balice-Gordon, R.J . (1998). In vivo approaches to neuromuscular
structure and function. Methods Cell Biol. 52, 323±348.
Balice-Gordon, R.J ., and Lichtman, J .W. (1993). Invivo observations
ofpre- and postsynaptic changes during the transitionfrommultiple
to single innervationat developing neuromuscularjunctions. J . Neu-
rosci. 13, 834±855.
Balice-Gordon, R.J ., Breedlove, S.M., Bernstein, S., and Lichtman,
J .W. (1990). Neuromuscular junctions shrink and expand as muscle
fiber size is manipulated: in vivo observations in the androgen-
sensitive bulbocavernosus muscle of mice. J . Neurosci. 10, 2660±
Balice-Gordon, R.J ., Chua, C.K., Nelson, C.C., and Lichtman, J .W.
(1993). Gradual loss of synaptic cartels precedes axon withdrawal
at developing neuromuscular junctions. Neuron 11, 801±815.
Barbacid, M. (1994). The Trk family of neurotrophin receptors. J .
Neurobiol. 25, 1386±1403.
Bosma, G.C., Custer, R.P., and Bosma, M.J . (1983). A severe com-
bined immunodeficiency mutation in the mouse. Nature 301,
Cabelli, R.J ., Hohn, A., and Shatz, C.J . (1995). Inhibition of ocular
dominance column formation by infusion of NT-4/5 or BDNF. Sci-
ence 267, 1662±1666.
Cabelli, R.J .,Shelton, D.L.,Segal, R.A.,and Shatz,C.J .(1997).Block-
ade ofendogenous ligands oftrkB inhibits formationofoculardomi-
nance columns. Neuron 19, 63±76.
Chalfie, M. (1995).Greenfluorescentprotein. Photochem.Photobiol.
Clary, D.O., Weskamp, G., Austin, L.R., and Reichardt, L.F. (1994).
TrkA cross-linking mimics neuronal responses to nerve growth fac-
tor. Mol. Biol. Cell 5, 549±563.
Culican, S.M., Nelson, C.C., and Lichtman, J .W. (1998). Axon with-
drawal during synapse elimination at the neuromuscular junction is
accompanied by disassembly ofthe postsynaptic specializationand
withdrawal of Schwann cell processes. J . Neurosci. 18, 4953±4965.
Eide, F.F., Vining, E.R., Eide, B.L., Zang, K., Wang, W.Y., and Rei-
chardt, L.F. (1996). Naturally occurring truncated trkB receptors
have dominant inhibitory effects on brain-derived neurotrophic fac-
tor signaling. J . Neurosci. 16, 3123±3129.
Friedman, B., Kleinfeld, D., Ip, N.Y., Verge, V.M., Moulton, R., Boland,
P., Zlotchenko, E., Lindsay, R.M., and Liu, L. (1995). BDNF and
NT-4/5 exert neurotrophic influences on injured adult spinal motor
neurons. J . Neurosci. 15, 1044±1056.
Frisen, J ., Verge, V.M., Fried, K., Risling, M., Persson, H., Trotter,
J ., Hokfelt, T., and Lindholm, D. (1993). Characterization of glial trkB
receptors: differential response to jury in the central and peripheral
nervous systems. Proc. Natl. Acad. Sci. USA 90, 4971±4975.
Funakoshi, H., Frisen, J ., Barbany, G., Timmusk, T., Zachrisson,
O., Verge, V.M., and Persson, H. (1993). Differential expression of
mRNAs for neurotrophins and their receptors after axotomy of the
sciatic nerve. J . Cell Biol. 123, 455±465.
Funakoshi, H., Belluardo, N., Arenas, E., Yamamoto, Y., Casabona,
A., Persson, E., and Iba Â n Ä ez, C.F. (1995). Muscle-derived neuro-
trophin-4 as an activity-dependent trophic signal for adult motor
neurons. Science 268, 1495±1499.
Glass, D.J ., Bowen, D.C., Stitt, T.N., Radziejewski, C., Bruno, J .,
Ryan, T.E., Gies, D.R., Shah, S., Mattsson, K., Burden, S.J ., et al.
(1996). Agrin acts via a MuSK receptor complex. Cell 85, 513±523.
Griesbeck, O., Parsadanian, A.S., Sendtner, M., and Thoenen, H.
(1995). Expression of neurotrophins in skeletal muscle: quantitative
In Vitro Agrin-Induced AChR Clustering Assay
in Mouse Myotubes
Primary myotube cultures were prepared as described above. Three
days after myotube formation (induced by addition of Dulbecco's
modified Eagle's medium and 2% horse serum), 20 pM per dish of
recombinant rat agrin (isoform 12±4±8; Wells et al., 1999; gift of Dr.
J . Fallon) was added to media for3 days. This was the half-maximal
concentration of agrin, determined in a pilot experiment in which
an agrin dose±clustering response curve was generated. On the
second day of agrin treatment, the cultures were infected with 2 ?
108pfu of AdtrkB.t1ha-gfp, AdlacZ-gfp, or vehicle. After one addi-
tional day of agrin treatment, cells were fixed in 4% paraformalde-
hyde for 10 min and rinsed with PBS, and AChRs were labeled with
10 ?g/ml R?BTX for 1 hr. Staining was visualized with confocal
microscopy, and the number and length of AChR clusters perGFP?
and GFP?myotube were determined from single plane projections
of stacks of confocal images taken with a 100?, 1.4 n.a. oil immer-
sion lens. Any contiguous area of R?BTX staining ?4 ?m in its
longest dimensionwas scored as anagrin-induced cluster(cf. Wells
et al., 1999).
We thank Dr. M. Barbacid for providing a mouse trkB.t1 cDNA; Dr.
S. Feinstein for anti-trkB.FL- and anti-trkB.t1-specific antibodies;
Dr. L. Reichardt for anti-TrkB and anti-TrkA polyclonal antibodies;
Dr. G. Yancopolous at Regeneron Pharmaceuticals for a mouse
TrkA cDNA; J . Harding and Dr. W. Snider for trkB?/?mice; Dr. M.
Chao forproviding stably transfectedPC12-trkB cells;Dr.R.Pittman
for providing PC63 cells; A. Young, J . Cardin, and E. Snyder for
preliminary immunostaining observations;A.Cannon, S. Gohari, and
H. Zhou fortechnicalassistance; and Drs. D. Kopp and K. Personius
for helpful discussions. This work was supported by grants from
TrkB-Mediated Signaling at Neuromuscular J unctions Download full-text
comparison and significance for motoneuron survival and mainte-
nance of function. J . Neurosci. Res. 42, 21±33.
Henderson, C.E., Camu, W., Mettling, C., Gouin, A., Poulsen, K.,
Karihaloo, M., Rullamas, J ., Evans, T., McMahon, S.B., Armanini,
M.P., et al. (1993). Neurotrophins promote motor neuron survival
and are present in embryonic limb bud. Nature 363, 266±270.
Horch, H.W., Kruttgen, A., Portbury, S.D., and Katz, L.C. (1999).
Destabilization of cortical dendrites and spines by BDNF. Neuron
Huang, E.J ., Wilkinson, G.A., Farin Ä as, I., Backus, C., Zang, K., Wong,
S.L., and Reichardt, L.F. (1999). Expression of trk receptors in the
developing mouse trigeminal ganglion: in vivo evidence for NT-3
activation of TrkA and TrkB in addition to trkC. Development 126,
J ang, S.K., Davies, M.V., Kaufman, R.J ., and Wimmer, E. (1989).
Initiation of protein synthesis by internal entry of ribosomes into the
5? nontranslated region of encephalomyocarditis virus RNA in vivo.
J . Virol. 63, 1651±1660.
Kass-Eisler, A., Falck-Pedersen, E., Elfenbein, D.H., Alvira, M., But-
trick, P.M., and Leinward, L.A. (1994). The impact of developmental
stage, route of administration and the immune system on adenovi-
rus-mediated gene transfer. Gene Ther. 1, 395±402.
Klein, R., Conway, D., Parada, L.F., and Barbacid, M. (1990). The
trkB tyrosine protein kinase gene codes for a second neurogenic
receptor that lacks the catalytic kinase domain. Cell 61, 647±656.
Klein, R., Smeyne, R.J ., Wurst, W., Long, L.K., Auerbach, B.A.,
J oyner, A.L., and Barbacid, M. (1993). Targeted disruption of the
trkB neurotrophin receptor gene results in nervous system lesions
and neonatal death. Cell 75, 113±122.
Koliatsos, V.E., Clatterbuck, R.E., Winslow, J .W., Cayouette, M.H.,
and Price, D.L. (1993). Evidence that brain-derived neurotrophic
factor is a trophic factor for motor neurons in vivo. Neuron 10,
Koliatsos, V.E.,Cayouette, M.H., Berkemeier, L.R.,Clatterbuck, R.E.,
Price, D.L., and Rosenthal, A. (1994). Neurotrophin 4/5 is a trophic
factor for mammalian facial motor neurons. Proc. Natl. Acad. Sci.
USA 91, 3304±3308.
Le Gal Le Salle, G., Robert, J .J ., Berrard, S., Ridoux, V., Stratford-
Perricaudet, L.D., Perricaudet, M., and Mallet, J . (1993). An adenovi-
rus vector for gene transfer into neurons and glia in the brain. Sci-
ence 259, 988±990.
Li, Y.X., Xu, Y., J u, D., Lester, H.A., Davidson, N., and Schuman,
E.M. (1998). Expression of a dominant negative trkB receptor, T1,
reveals a requirement for presynaptic signaling in BDNF-induced
synaptic potentiation in cultured hippocampal neurons. Proc. Natl.
Acad. Sci. USA 95, 10884±10889.
Lillien, L. (1995). Changes in retinal cell fate induced by overexpres-
sion of EGF receptor. Nature 377, 158±162.
Lohof, A.M., Ip, N.Y., and Poo, M. (1993). Potentiation of developing
neuromuscularsynapses by the neurotrophins NT-3 and BDNF. Na-
ture 363, 350±353.
McAllister, A.K., Lo, D.C., and Katz, L.C. (1995). Neurotrophins regu-
late dendritic growth in developing visual cortex. Neuron 15,
McAllister, A.K., Katz, L.C., and Lo, D.C. (1997). Opposing roles for
endogenous BDNF and NT-3 in regulating cortical dendritic growth.
Neuron 18, 767±778.
Middlemas, D.S., Lindberg, R.A., and Hunter, T. (1991). trkB, a neural
receptor protein-tyrosine kinase: evidence for a full-length and two
truncated receptors. Mol. Cell Biol. 11, 143±153.
Nguyen, Q.T., Parsadanian, A.S., Snider, W.D., and Lichtman, J .W.
(1998). Hyperinnervation of neuromuscular junctions caused by
GDNF overexpression in muscle. Science 279, 1725±1729.
Offenhauser, N., Bohm-Matthaei, R., Tsoulfas, P., Parada, L., and
Meyer, M. (1995). Developmental regulation of full length trkC in rat
sciatic nerve. Eur. J . Neurosci. 7, 917±925.
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.
O'Malley, J .P., Waran, M.T., and Balice-Gordon, R.J . (1999). In vivo
observations of terminal Schwann cells at normal, denervated, and
reinnervated mouse neuromuscular junctions. J . Neurobiol. 38,
Oppenheim, R.W. (1996). Neurotrophic survival molecules formoto-
neurons: an embarrassment of riches. Neuron 17, 195±197.
Pati, U.K. (1992). Novel vectors for expression of cDNA encoding
epitope-tagged proteins in mammalian cells. Gene 114, 285±288.
Precious, B., and Russell, W.C. (1985). Growth, purificationand titra-
tion of adenoviruses. In Virology: A Practical Approach (New York:
Oxford Press), pp. 193±205.
Rich, M.M., and Lichtman, J .W. (1989). In vivo visualization of pre-
and postsynaptic changes during synapse elimination in reinner-
vated mouse muscle. J . Neurosci. 9, 1781±1805.
Rich, M.M., Colman, H., and Lichtman, J .W. (1994). In vivo imaging
shows loss of synaptic sites from neuromuscular junctions in a
model of myasthenia gravis. Neurology 44, 2138±2145.
Sanes, J .R., Rubenstein, J .L.R., and Nicolas, J .-F. (1986). Use of a
recombinant retrovirus to study post-implantation cell lineage in
mouse embryos. EMBO J . 5, 3133±3142.
Sanes, J .R., and Lichtman, J .W. (1999). Development of the verte-
brate neuromuscular junction. Annu. Rev. Neurosci. 22, 389±442.
Shelton, D.L., Sutherland, J ., Gripp, J ., Camerato, T., Armanini, M.P.,
Phillips, H.S., Carroll, K., Spencer, S.D., and Lenison, A.D. (1995).
Human trks: molecular cloning, tissue distribution and expression
ofextracellulardomainimmunoadhesions. J . Neurosci. 15, 477±491.
Snider, W.D. (1994). Functions of the neurotrophins during nervous
system development: what the knockouts are teaching us. Cell 77,
Son, Y., and Thompson, W.J . (1995). Schwann cell processes guide
regeneration of peripheral axons. Neuron 14, 125±132.
Tomanin,R., Bett,A.J .,Picci, L.,Scarpa,M., andGraham, F.L.(1997).
Development and characterization of a binary gene expression sys-
tem based on bacteriophage T7 components in adenovirus vectors.
Gene 193, 129±140.
Trachtenberg, J .T., and Thompson, W.J . (1996). Schwann cell apo-
ptosis at developing neuromuscular junctions is regulated by glial
growth factor. Nature 379, 174±177.
Wang, T., Xie, K., and Lu, B. (1995). Neurotrophins promote matura-
tion of developing neuromuscular synapses. J . Neurosci. 15, 4796±
Wang, X.-H., and Poo, M.-M. (1997). Potentiation of developing syn-
apses by postsynaptic release of neurotrophin-4. Neuron 19,
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.
Woo, Y.J ., Raju, G.P., Swain, J .L., Richmond, M.E., Gardner, T.J .,
and Balice-Gordon, R.J . (1997). In vivo cardiac gene transfer via
intraplacental delivery of recombinant adenovirus. Circulation 96,
Xie, K., Wang, T., Olafsson, P., Mizuno, K., and Lu, B.(1997). Activity-
in the development of neuromuscular junctions. J . Neurosci. 17,
Yamamoto, M., Sobue, G., Yamamoto, K., Terao, S., and Misume,
T. (1996). ExpressionofmRNAs forneurotrophins (NGF, BDNF, NT-3
and GDNF) and their receptors (p75NGFR, trkA, trkB, trkC) in the
adult human peripheral nervous system and non-neural tissues.
Neurochem. Res. 21, 929±938.
Yan, Q., Radeke, M.J ., Matheson, D.R., Talvenheimo, J ., Welcher,
A.A., and Feinstein, S.C. (1997). Immunocytochemical localization
of trkB in the central nervous system of the adult rat. J . Comp.
Neurol. 378, 135±157.