Development of the neuromuscular junction
- First Online:
- Cite this article as:
- Witzemann, V. Cell Tissue Res (2006) 326: 263. doi:10.1007/s00441-006-0237-x
- 1.2k Views
The differentiation of the neuromuscular junction is a multistep process requiring coordinated interactions between nerve terminals and muscle. Although innervation is not needed for muscle production, the formation of nerve-muscle contacts, intramuscular nerve branching, and neuronal survival require reciprocal signals from nerve and muscle to regulate the formation of synapses. Following the production of muscle fibers, clusters of acetylcholine receptors (AChRs) are concentrated in the central regions of the myofibers via a process termed “prepatterning”. The postsynaptic protein MuSK is essential for this process activating possibly its own expression, in addition to the expression of AChR. AChR complexes (aggregated and stabilized by rapsyn) are thus prepatterned independently of neuronal signals in developing myofibers. ACh released by branching motor nerves causes AChR-induced postsynaptic potentials and positively regulates the localization and stabilization of developing synaptic contacts. These “active” contact sites may prevent AChRs clustering in non-contacted regions and counteract the establishment of additional contacts. ACh-induced signals also cause the dispersion of non-synaptic AChR clusters and possibly the removal of excess AChR. A further neuronal factor, agrin, stabilizes the accumulation of AChR at synaptic sites. Agrin released from the branching motor nerve may form a structural link specifically to the ACh-activated endplates, thereby enhancing MuSK kinase activity and AChR accumulation and preventing dispersion of postsynaptic specializations. The successful stabilization of prepatterned AChR clusters by agrin and the generation of singly innervated myofibers appear to require AChR-mediated postsynaptic potentials indicating that the differentiation of the nerve terminal proceeds only after postsynaptic specializations have formed.
KeywordsNeuromuscular junctionDevelopmentAcetylcholine receptorMuSKAgrinTransgenic modelsMouse
The innervation of skeletal muscle leads to the formation of neuromuscular junctions (NMJs). Developing NMJs serve as one of the best model systems for studying synapse formation, since changes in shape, size, and molecular composition can be followed with high spatial and temporal resolution. The analysis of NMJs has shown that differentiation of the pre- and postsynaptic specializations is a multistep process requiring coordinated sequential interactions between nerve terminals and muscle (Sanes and Lichtman 2001). With the formation of muscle fibers around embryonic days (E) 12–14 in mouse, acetylcholine receptors (AChRs) are prepatterned at sites that are used by the outgrowing motor neurons to form synaptic contacts. On each muscle fiber, one synaptic site (often contacted by several different nerves) becomes stabilized, and the mature endplate pattern is established at about E17. During postnatal development, further characteristic structural and functional changes are observed. Multiple innervations of muscle fibers are reduced until motor neurons form single synaptic contacts, embryonic AChRs are replaced by aggregates of adult AChRs (Mishina et al. 1986; Witzemann et al. 1989) that acquire a characteristically shaped “pretzel-like” pattern, and the postsynaptic membrane forms numerous folds with AChRs becoming concentrated at their crests (Sanes and Lichtman 2001). Concomitantly, nuclei accumulate at the synaptic sites (Couteaux 1973; Grady et al. 2005). The mechanisms that regulate and stabilize the structural and functional specializations throughout development are largely unknown. Considerable advances have been achieved with the availability of an increasing number of transgenic mouse models in which synapse formation can be followed anatomically in vivo.
Phrenic nerve/diaphragm: a model for studying the embryonic development of neuromuscular junctions
In studies of regulatory interactions between nerve and muscle, examination of the ontogeny of the phrenic nerve/diaphragm specialization has been of especial value (Greer et al. 1999) for several reasons. The appearance of myofibers in the thin fragile primordium can be clearly visualized. The main primary and secondary axon branches approaching the diaphragm can be stained histochemically or by immunolabeling, and the establishment of synaptic contacts and the innervation pattern can be observed with high spatiotemporal resolution. Other model systems, e.g., hindlimb muscles, are more complex, since a mixed population of motor nerves innervates the various muscles. Differences in synaptic inputs, in electrophysiological properties, and in the putative trophic substances that cause morphological changes at developing synapses can be directly visualized since both nerve and muscle can be studied in parallel.
Regulatory signals of neuronal origin
Nerve-muscle interactions are generally agreed to regulate the formation of synapses. ACh is the main neuronal signal that elicits responses from the postsynaptic AChRs and thus regulates postsynaptic differentiation. During the search for additional neuronal factors, agrin and neuregulin have been discovered over the last 25 years. Initially, the biochemical and structural organization of the subsynaptic membrane in regenerating muscle, and possibly the differentiation of the nerve terminal, was observed to be directed by components associated with the basal lamina (Sanes et al. 1978; Burden et al. 1979). Further analysis led to the identification of agrin, a synaptic basal lamina protein. Agrin is a heparin proteoglycan expressed by motor neurons and by muscle. The neuronally derived isoform is 1000-fold better at inducing AChR clusters than the muscle isoform (Gesemann et al. 1995) supporting the hypothesis that agrin plays a key role in the accumulation of AChR at synaptic sites (McMahan 1990).
Genetically engineered mice lacking agrin die at birth because neuromuscular function is disrupted (Gautam et al. 1996). This demonstrates that neuronally released agrin is required for synapse formation. The AChR clusters in the mutant mice are, however, prepatterned around E13/14, as in wild-type mice (Lin et al. 2001). Although the prepatterned region in agrin mutant mice is broader than that in wild-type mice, the observation shows that agrin is dispensable for the initial prepatterning process in muscle. The lack of agrin has severe effects on subsequent development of the NMJ. The motor nerves are unable to interact with prepatterned AChRs and to establish stable synaptic contacts. Instead, they continue to grow and branch extensively (Lin et al. 2001; Yang et al. 2001). AChR clusters, not stabilized at nerve-muscle contact zones, appear scattered as small aggregates all over the diaphragm muscle. Thus, the physiological role of agrin is not to induce postsynaptic specializations on developing myofibers, but rather to stabilize synaptic contacts, to prevent dispersion of AChR clusters, and to control axon growth. Agrin complexed with a postsynaptic protein, MuSK (see below), has been suggested to form part of a motor neuron stop signal (Dimitropoulou and Bixby 2005) and to act as a presynaptic regulator for axonal branching (Campagna et al. 1995; Gautam et al. 1996).
The finding that agrin does not induce the initial steps of synapse formation has been a big surprise. Even more astonishing is the observation that the muscle prepattern process does not even require the presence of motor neurons. Mouse mutants lacking the motoneuron-specific transcription factor Hb9 display axon guidance defects and fail to form the phrenic nerve but develop an apparently normal diaphragm muscle without innervation (Arber et al. 1999; Lin et al. 2001; Thaler et al. 1999; Yang et al. 2001). Surprisingly, AChR clusters are still prepatterned in these mutant mice, although, as in agrin mutants, the prepatterned band is broader than that in wild-type animals. However, unlike in agrin mutant mice, AChR clusters are maintained in the central region throughout embryonic development until mutant mice die at birth from respiratory paralysis. Additional support for the muscle-intrinsic patterning mechanisms has been unexpectedly obtained from mice in which the DNA topoisomerase IIβ gene has been disrupted genetically. In these mutants, motor axons fail to innervate the diaphragm, although AChRs are prepatterned in the central region of diaphragm muscle fibers (Yang et al. 2000), as in the Hb9 and agrin mutants. The defects in motor axon growth in IIβ mutant mice also results in a breathing impairment and death of the pups shortly after birth.
The observation that AChR clusters are dispersed in agrin knockout mutants but are localized mainly to the central regions of the diaphragm in the absence of motor neurons suggests that some neuronal factor, possibly ACh itself, contributes to the rearrangement of AChR clusters in innervated muscle (Lin et al. 2001; Yang et al. 2001) and supposedly promotes the differentiation of the innervated postsynaptic apparatus. The regulatory role of ACh during embryonic development has been studied by the generation of mice lacking choline acetyltransferase (ChAT). Since these mice cannot synthesize neurotransmitter, they are unable to induce postsynaptic endplate potentials and die at birth (Misgeld et al. 2002; Brandon et al. 2003). The embryo mutants display a complex phenotype. First, AChR clusters are prepatterned at E14.5 as in the wild-type, although, with development, the number of AChR clusters is increased, and the distribution is scattered over a significantly wider region than that observed in wild-type mice. Second, nerve growth and branching at E13 are increased compared with that in wild-type. Third, aberrant nerve growth and excessive branching continues throughout embryonic development. Finally, multiple endplates are formed on single myofibers that are innervated by different nerves possibly because of the absence of neurotransmission. The determination of whether these contacts can be turned into functional synapses by rescuing ACh synthesis will be of interest. Thus, synapse formation and the final differentiation of mature synapses depends on neurotransmission.
The generation of double-knockout mice deficient in both ChAT and agrin (Lin et al. 2005; Misgeld et al. 2005) has increased the complexity of the mutant mouse model. The mutant mice die at birth, but surprisingly, NMJs display pre- and postsynaptic-like specializations. AChRs, although reduced in number and more widely distributed than those in the ChAT knockout mice, remain clustered even without the cluster-stabilizing agrin. How can such a phenotype be interpreted? Lin et al. (2005) and Misgeld et al. (2005) hypothesize that the dispersion of AChR clusters that is observed in absence of agrin does not occur when nerves are missing or when ACh is not released. Based on these observations, the function of agrin seems to be to stabilize nascent postsynaptic specializations and to prevent the destabilization of AChR clusters and their dispersion by neuronally released ACh (Lin et al. 2005; Misgeld et al. 2005). ACh-mediated dispersion of AChRs may occur by endocytosis causing the loss of AChRs and by a redistribution of AChRs in the plane of the membrane (Misgeld et al. 2005). ACh may signal via a cyclin-dependent kinase 5 (cdk5)-dependent mechanism that could destabilize AChR clusters directly or impair assembly and recycling of AChR clusters (Lin et al. 2005).
The release of neurotransmitter is induced by interactions between the so-called SNAREs and the sec1/Munc18-related protein, munc 18-1 (Rizo and Sudhof 2002). Blockade of neurotransmitter release has been achieved in munc18-1 null mutant mice that consequently lack both spontaneous and evoked neuromuscular transmission throughout embryonic development and that die immediately after birth, probably because they cannot breathe (Verhage et al. 2000). At first sight, one might expect a phenotype comparable to that observed in the ChAT knockout animals. However, at E14, axonal targeting and the main branching of the phrenic nerve are normal in this mutant. At birth, munc18-1 mutant muscles contain an elaborate and disorganized meshwork of neuronal processes, probably resulting from axonal defasciculation and aberrant axonal sprouting, whereas AChR clusters remain uninnervated. The phenotypical differences of the munc 18-1 and the ChAT knockout mice may be attributable to the finding that the neurons in munc 18-1 mutants do not receive any synaptic input and thus lack the prerequisites for normal differentiation (Heeroma et al. 2003). Nevertheless, this mouse model supports the view that initial axon pathfinding does not depend on neurotransmitter secretion or transmitter-mediated postsynaptic specializations, whereas the maintenance of the motor neuronal system and the organization of nerve terminal branches and stabilization of synapses does indeed require synaptic activity.
The search for putative neuronal factors that contribute to postsynaptic differentiation has led to the identification of “AChR-inducing activity” (ARIA; Falls et al. 1993). ARIA (or neuregulin-1) turns out to be a member of the neuregulin family, proteins containing an epidermal growth factor (EGF)-like motif that activates membrane-associated tyrosine kinases related to the EGF receptor (known as erbB1). Neuregulin/erbB perform many functions during neural development in the central and peripheral nervous system, heart, breast, and other organ systems (Buonanno and Fischbach 2001; Garratt et al. 2000; Schaeffer et al. 2001). Recent evidence suggests that neuregulin signaling plays important roles in many neurological disorders (Esper et al. 2006). Attempts to clarify its role in the differentiation of NMJs in vivo has been difficult because ablation of neuregulin-1 (Meyer and Birchmeier 1995) and receptors, erbB2 (Lee et al. 1995) and erbB4 (Gassmann et al. 1995), causes death mainly because of heart defects before myelination in the central nervous system begins. ErbB3 knockout embryos survive until birth, and postsynaptic differentiation reveals no significant differences from that of wild-type animals (Riethmacher et al. 1997). Mutants that rescue erbB2 defects in heart development and thus survive until birth indicate that neuregulin/erbB signaling is not involved in NMJ formation during embryonic development (Morris et al. 1999; Woldeyesus et al. 1999). This view has been confirmed by transgenic animals in which either erbB2 and erbB4 (Escher et al. 2005) or neuregulin-1 (Jaworski and Burden 2006) are conditionally deleted. The results demonstrate that neuregulin signaling to muscle is dispensable for the formation of NMJ.
The initiation of postsynaptic specialization, i.e., AChR prepatterning, apparently requires no nerve-derived signals and proceeds in the absence of motor nerves. One protein that is required for clustering is the 43K protein (Neubig et al. 1979; Sobel et al. 1977), now also known as rapsyn (receptor associated protein of the synapse; Musil et al. 1989). This membrane-associated cytoplasmic protein was isolated by virtue of its tight association with the AChR. AChR/rapsyn complexes aggregate and are stabilized in the postsynaptic membrane (Froehner 1993). These aggregates are necessary for all forms of AChR clustering as is clearly demonstrated in rapsyn-deficient mice (Gautam et al. 1995) in which no AChR clusters form on the muscles.
AChR clustering at endplates also requires a muscle-specific receptor tyrosine kinase, MuSK; this was initially found concentrated in postsynaptic membranes (Jennings et al. 1993; Valenzuela et al. 1995). Mice lacking MuSK fail to form NMJs, show aberrant nerve growth, and die at birth (DeChiara et al. 1996). In contrast to mouse models, such as agrin, ChAT, munc 18-1, neuregulin/erbB2, or cdk5 knockout mice or knockout mutants lacking motor nerves, MuSK mutants, like rapsyn mutants, express no AChR clusters at all. Initially, MuSK was assumed to be activated by neuronally released agrin to induce AChR clusters (Glass et al. 1996); however, the molecular basis for agrin-induced MuSK signaling is not known. Since initial AChR clustering and muscle prepatterning does not require agrin and occurs in the absence of motor nerves, the agrin/MuSK signaling theory has had to be revised. Potential downstream signaling molecules include the Rac/JNK kinases (Weston et al. 2000) and Dishevelled (Dv1) interacting with PAK1 kinase (Luo et al. 2002), and geranylgeranyltransferase 1 (Luo et al. 2003).
How might AChR prepatterning be achieved? Cell culture experiments indicate that MuSK mediates both the agrin-dependent activation of a complex signal transduction pathway and the agrin-independent association of the kinase with other postsynaptic, but as yet unidentified, components (Herbst and Burden 2000; Zhou et al. 1999). Thus, MuSK signaling may contribute directly or indirectly to synapse-specific gene transcription (DeChiara et al. 1996). Direct gene transfer experiments expressing transgenic MuSK in single muscle fibers of mice or rats also show that MuSK kinase activity is sufficient to induce the formation of AChR clusters (Hesser et al. 1999; Jones et al. 1999). In combination with the conditional inactivation of endogenous MuSK in vivo, clear evidence exists for ligand-independent MuSK activation that induces postsynaptic-like specializations and the expression of AChR ε subunits (Sander et al. 2001). The expression of ε subunit transcripts in muscle of wild-type mice is normally restricted to subsynaptic nuclei producing synapse-specific transcripts at adult NMJs (Brenner et al. 1990). Muscle prepatterning thus may be achieved as follows. Centrally localized nuclei of the developing myofibers are the most advanced in their differentiation, since they are developmentally the oldest nuclei. MuSK concentrations around these nuclei may reach the critical threshold for increased spontaneous kinase activity (Arber et al. 2002). At these sites, a positive feedback loop activates the expression not only of MuSK, but also of AChR. MuSK signaling acts only locally because of the restricted subcellular distribution of MuSK (Sander et al. 2001). AChR complexes aggregated and stabilized by rapsyn are thus prepatterned independently of neuronal signals in developing myofibers.
The role of MuSK as a synaptic organizer affecting synaptic gene transcription gains further support from mouse mutants in which MuSK is inactivated postnatally by Cre recombinase-mediated recombination of musk (Hesser et al. 2006). The results show that MuSK is required to supply adult endplates continuously with AChRs and to maintain the postsynaptic apparatus. Synapse disassembly in adult muscle has also been observed upon small-interfering-RNA-mediated inactivation of MuSK (Kong et al. 2004). Inactivation of MuSK in adult muscle resulting in the loss of AChRs and the consequent loss of postsynaptic activity induces aberrant nerve growth and branching (Hesser et al. 2006).
Stabilization of synaptic contacts requires postsynaptic signal transduction
AChRs mediate neuromuscular transmission via a transient inward current flow following ACh release from the motor nerve terminal (Colquhoun and Sakmann 1998). In mammalian muscle, they are expressed in a fetal and an adult isoform (Sakmann and Brenner 1978) that differ in their functional properties and have a different subunit composition (Mishina et al. 1986). The fetal form, AChRγ, comprised of α2βγδ subunits, is replaced during development by the adult form, AChRε, in which the ε subunit replaces the γ subunit; thus, the adult-type receptors are comprised of α2βεδ subunits. Expression of the γ and ε subunit genes is regulated differently (Numberger et al. 1991), and ACh-induced muscle activity strongly reduces the expression levels of fetal but not adult AChR subtypes (Duclert and Changeux 1995). Although the biological relevance of the postnatal γ/ε subunit switch in the neonatal muscle cannot be explained at the molecular level, both AChR subtypes are clearly essential for normal synaptic development and function: γ subunit knockout animals die at birth (Takahashi et al. 2002), whereas ε subunit knockout mice die prematurely because of severe muscle weakness (Witzemann et al. 1996; Missias et al. 1997).
The generation of mice in which fetal-type AChRγ is substituted by adult-type AChRε throughout embryonic development (Koenen et al. 2005) demonstrates that myoblast fusion, muscle formation, and endplate differentiation are not significantly altered. However, the innervation pattern of the diaphragm is significantly affected. Expression of AChRε during embryonic development leads to a much broader distribution of endplates than in wild-type mice. Broadening of the innervated region in this case does not result from the absence of ACh as hypothesized for ChAT-deficient mice (Brandon et al. 2003; Misgeld et al. 2002) or aneural mice (Lin et al. 2001; Yang et al. 2000, 2001) but is a consequence of the altered ion conductance properties of the endplate AChRs. Thus, ACh alone is not sufficient for the precise targeting of endplates; successful innervation also needs the presence postsynaptic AChRs. ACh released from the branching motor nerves (Young and Poo 1983) could activate prepatterned AChRs, thereby inducing signals that promote precise target recognition and determine the size of the final endplate region. Activity might decrease the area over which individual myotubes are susceptible to innervation and prevent the formation of additional endplates on the same myofiber. Whether the functional ion conductance properties of the AChRs actually determine the localization of endplate can be tested by the generation of additional mutant mice with altered AChR properties.
Synaptic input regulates neuronal survival
During normal NMJ development, myofibers are often innervated by more than one motor neuron. As postsynaptic development progresses, multiple contacts are reduced in favor of singly innervated myofibers; these changes are probably based on activity-dependent mechanisms (Buffelli et al. 2003). With maturation, approximately half of the motor neurons degenerate. In the rapsyn, MuSK, agrin, and ChAT knockout mice, more motor neurons appear to survive than in wild-types (Banks et al. 2001; Brandon et al. 2003; Terrado et al. 2001). This abnormally high survival is probably attributable to the disruption of NMJs in these mutants and the resultant lack of activity-based competition. This suggests that specializations of the neuromuscular synapse that ensure efficient synaptic transmission and muscle contraction are important for motor axon branching and neuronal survival.
Muscle prepatterning: MuSK concentrations around centrally localized nuclei of the developing myofibers may reach the critical threshold for increased spontaneous kinase activity earlier than more distally localized nuclei. A positive feedback loop activates not only the expression of MuSK but also of AChR. AChR complexes aggregated and stabilized by rapsyn are thus prepatterned independently of neuronal signals in developing myofibers.
ACh/AChR-dependent stabilization of nerve-muscle contact sites: ACh released by branching motor nerves causes AChR-induced postsynaptic potentials that appear to determine endplate localization. Thus, ACh positively regulates the localization and stabilization of developing synaptic contacts. These “active” contact sites may prevent AChRs clustering in non-contacted regions and counteract the establishment of additional contacts. Aberrant development is observed in absence of ACh as AChR clustering continues and endplate-like specializations are formed in increased numbers; these are contacted by nerves displaying presynaptic-like specializations. Different nerves can contact multiple endplates aberrantly localized on the myofibers. Similar pre- and postsynaptic differentiations, although in significantly reduced numbers, are observed, even in absence of ACh and agrin, supporting the view that ACh-induced signals lead to the dispersion of non-synaptic AChR clusters and possibly the loss of excess AChRs. The lack of synaptic activity in these cases causes extensive uncontrolled nerve growth.
Stabilization of synaptic contacts by agrin: agrin released from the branching motor nerve may form a structural link specifically to the ACh-activated endplates thereby enhancing MuSK kinase activity and AChR accumulation and preventing the dispersion of postsynaptic specializations. At developing endplates, synapse-specific gene transcription contributes to the differentiation and stabilization of the postsynaptic membrane. The precise molecular mechanisms that regulate synapse-specific gene transcription are still unknown. The successful stabilization of prepatterned AChR clusters by agrin and the generation of singly innervated myofibers appear to require AChR-mediated postsynaptic potentials supporting the view that the differentiation of the nerve terminal proceeds only after postsynaptic specializations have formed.
I thank Dr. D.J. Haydon-Wallace for critically reading the manuscript. I also thank Drs. M. Koenen and F. Chevessier for discussions and U. Mersdorf for technical assistance.