|Home | About | Journals | Submit | Contact Us | Français|
It had long been thought that motor axons approach muscles that are regionally unspecialized and induce postsynaptic differentiation by releasing signals that focally initiate transcriptional and post-translational responses in muscle. This neuro-centric view of synapse formation, however, has been challenged by recent experiments, which showed that AChR clusters are concentrated in the central region of muscle independent of innervation. This nerve-independent prepattern of AChR expression requires MuSK, a receptor tyrosine kinase that is critical for synapse formation. How muscle prepatterning is established and whether motor axons recognize this prepattern are not known. Here, we show that MuSK itself is prepatterned in muscle and that high, ectopic MuSK expression is sufficient to promote ectopic motor axon growth and synapse formation, indicating that the muscle prepattern is recognized by motor axons and promotes synapse formation in the central region of developing mammalian muscle. Further, we provide evidence that early expression of MuSK in developing myotubes is sufficient to re-establish muscle prepatterning independent of the MuSK promoter. Moreover, we show that ectopic MuSK expression stimulates synapse formation in the absence of Agrin and rescues the neonatal lethality of agrin mutant mice, demonstrating that MuSK, independent of Agrin, is sufficient to direct presynaptic and postsynaptic differentiation. In contrast to a neuro-centric view for synapse formation, these data demonstrate that the postsynaptic cell can play a dominant role in regulating synapse formation.
Once axons project to their target area, they branch within their target field in a highly stereotyped manner. This organizational feature of neurons is critical to form synapses on appropriate target cells and to establish functional neuronal circuits. The mechanisms that direct and confine axon branching and synapse formation to a defined target area, however, are poorly understood. In mammals, neuromuscular synapses are located within a narrow endplate band in the central region of the muscle, marked by presynaptic nerve terminals, postsynaptic clusters of acetylcholine receptors (AChRs) and elevated levels of AChR gene expression (for reviews see (Burden, 1998; Sanes and Lichtman, 1999, 2001). A cardinal question in vertebrate synapse formation is how this characteristic and stereotyped pattern of innervation is established and which roles nerve and muscle play in shaping this topographic arrangement.
It had long been thought that motor axons approach a muscle that is regionally unspecialized and induce postsynaptic differentiation by releasing signals that focally initiate transcriptional and post-translational responses in the muscle (Bennett and Pettigrew, 1974, 1976; Burden, 1998; Kandel et al., 2000; Sanes and Lichtman, 2001; Schaeffer et al., 2001). These studies implicated motor neuron-derived Agrin, a glycosylated proteoglycan, as the critical neural signal for inducing postsynaptic differentiation and the muscle-specific receptor tyrosine kinase, MuSK, a component of the Agrin receptor, as the transducer for clustering AChRs and activating AChR gene expression at synaptic sites (Burden, 1998; Glass et al., 1996a; Glass et al., 1996b; Glass and Yancopoulos, 1997; Sanes and Lichtman, 2001; Schaeffer et al., 2001; Valenzuela et al., 1995). This neuro-centric view of neuromuscular synapse formation, however, has been challenged by recent experiments, which showed that mammalian muscle is spatially patterned independent of innervation. In mice that lack DNA topoisomerase IIβ, motor axons fail to innervate diaphragm and limb muscles, yet AChR clusters are concentrated in the central region of these muscles (Yang et al., 2000). Subsequent studies, which examined the roles of Agrin and MuSK in nerve-independent muscle patterning, demonstrated that MuSK, but not Agrin is required to cluster AChRs in the central region of the muscle (Lin et al., 2001; Yang et al., 2001). This nerve-independent prepattern of AChR expression is then modified and sharpened by two neural signals: Agrin, which stabilizes AChR clusters, and acetylcholine (ACh), which extinguishes AChR clusters (Lin et al., 2005; Misgeld et al., 2005). As a consequence, AChR clusters are selectively maintained at nascent synapses and dispersed at non-synaptic sites. These genetic studies in mice have been complemented by live-imaging of neuromuscular synapse formation in zebrafish (Flanagan-Steet et al., 2005; Panzer et al., 2006), which showed that this pattern of AChR expression is established prior to innervation and suggested that motor axon growth cones grow toward this prepatterned zone (Panzer et al., 2006).
Important aspects of muscle pre-patterning, however, remain poorly understood. The muscle intrinsic mechanisms that are responsible for establishing regional differences in muscle prior to innervation are not known. Further, it is not known whether this pre-existing regional specialization in the muscle is recognized by approaching motor axons and has a role in confining axon branching and synapse formation to the appropriate target field within the muscle.
In this study we show that MuSK itself is expressed preferentially in the central region of muscle independent of innervation and that this spatial pattern of MuSK expression dictates where motor axons grow and where synapses will form. Moreover, surprisingly, ectopic MuSK stimulates the formation of ectopic synapses independent of Agrin, demonstrating that MuSK activation, independent of Agrin, is sufficient to stimulate presynaptic and postsynaptic differentiation. Further, we provide evidence that early expression of MuSK in developing muscle, conferred by the skeletal α-actin promoter, is sufficient to initiate muscle prepatterning in mice that lack endogenous MuSK and motor neurons, consistent with the idea that early expression of MuSK, together with the pattern of muscle growth and positive feedback by MuSK, is sufficient to establish muscle prepatterning.
AChR gene expression and AChR clusters are concentrated in the central region of skeletal muscle in the absence of innervation (Lin et al., 2001; Yang et al., 2001; Yang et al., 2000). This nerve-independent patterning of AChR expression, termed muscle prepatterning, is dependent upon MuSK (Lin et al., 2001; Yang et al., 2001), a receptor tyrosine kinase that is activated by Agrin and essential for neuromuscular synapse formation (DeChiara et al., 1996; Strochlic et al., 2005). To determine whether MuSK itself is prepatterned in muscle lacking motor axons, we examined the pattern of MuSK expression in muscles from mice that lack motor axons. Mice that lack motor neurons were generated by crossing HB9cre mice, which express Cre recombinase selectively in developing motor neurons, with Isl2loxP-stop-loxP-DTA mice, which, following Cre-mediated deletion of a translational stop sequence, express diphtheria toxin A (DTA) from the isl2 gene (Yang et al., 2001). In mice carrying both alleles, DTA is expressed in developing motor neurons, leading to death of motor neurons before they differentiate and extend axons into muscle (Yang et al., 2001). We found that MuSK mRNA, like AChR mRNA, is concentrated in the central region of muscles from HB9cre; Isl2loxP-stop-loxP-DTA (Isl2DTA) mice (Figure 1B, D). Thus, MuSK expression is itself restricted to the central muscle region independent of innervation, suggesting that the pattern of MuSK expression defines the region of AChR expression in the absence of motor axons.
To determine whether this restricted pattern of MuSK expression dictates the pattern of AChR clustering in muscle lacking motor neurons, we generated transgenic mice that express MuSK uniformly throughout the muscle, under the control of the human skeletal α-actin (HSA) regulatory region (Figure 1E). We analyzed two HSA::MuSK transgenic lines, MuSK-L and MuSK-H, which overexpress MuSK by 3- and 20-fold, respectively (Figure 1F, G, H, I).
To determine whether ectopic MuSK expression induces ectopic AChR clusters in muscle lacking motor axons, we generated HB9cre; Isl2DTA mice that carry either the MuSK-L or MuSK-H transgene, and we stained for AChRs with Alexa-594-α-bungarotoxin (α-BGT). We found that uniform expression of MuSK from the high-expressing MuSK-H transgene, but not the low-expressing MuSK-L transgene, is sufficient to induce AChR clusters throughout the muscle of mice lacking motor neurons (Figure 2D, F, G). Thus, high ectopic MuSK expression is sufficient to perturb muscle prepatterning and induce ectopic AChR clustering.
We next examined whether disruption of the muscle prepattern perturbed motor axon growth. We stained whole mounts of diaphragm muscles with antibodies to Neurofilament (NF) and Synaptophysin (Syn) to visualize axons and nerve terminals, respectively, and with α-BGT to mark postsynaptic AChRs. We found that uniform expression of MuSK from the high-expressing MuSK-H transgene, but not the low-expressing MuSK-L transgene, is sufficient to induce motor axon growth throughout the muscle (Figure 3C, E). Similar exuberant axon growth was found in limb muscles (data not shown). Indeed, quantification of axon growth shows that motor axons in MuSK-H mice extend to cover nearly the entire (~90%) muscle (Figures 3G, S1). Thus, high, ectopic MuSK expression, sufficient to alter muscle prepatterning and induce AChR clustering, is sufficient to promote motor axon growth, suggesting that muscle prepatterning normally restricts motor axons to this pre-specialized region of the muscle.
Ectopic motor axons in MuSK-H mice terminate and form specializations in peripheral regions of the muscle, which appear identical to synapses in the central region of muscle from wild-type mice (Figure 4A, B, C). In wild-type mice, synapses form a band that occupies the central ~10% of the muscle, whereas in MuSK-H mice the synaptic zone covers nearly 90% of the muscle, similar to the distribution of motor axons in these muscles (Figures 4D, S2).
In mammalian muscle, nearly all muscle fibers bear a single synaptic site (Sanes and Lichtman, 1999). We counted the number of synapses within a defined area of the diaphragm muscle from wild-type and MuSK-H mice and found that the number of synapses is 28.67 ± 3.7 % greater in muscle from MuSK-H mice (Figures 4E, S2). We sought to determine the source of these additional synapses and found that diaphragm muscles from wild-type and MuSK-H mice contain a similar number of muscle fibers (Figures 4F, G, S2), indicating that a subset of myofibers, possibly as high as ~30% of the total, contain more than one synaptic site in MuSK-H mice. We counted the number of Isl1/2-positive motor neurons in the cervical, thoracic and lumbar regions of the spinal cord from wild-type and MuSK-H mice and found that MuSK-H mice contain a normal number of motor neurons (Figures 4H, I, S3), indicating that the additional synapses arise from increased branching of a fixed number of motor neurons rather than from increased motor neuron survival. To determine whether these additional synapses are generated from branches that arise from axons or terminals, we counted the number of terminal sprouts in diaphragm muscles from wild-type and MuSK-H mice. Although we detect a modest level of terminal sprouting in MuSK-H mice (4 terminal sprouts/495 synapses examined in MuSK-H mice and no terminal sprouts/401 synapses in wild-type mice; Figure 4J, K, L), the extent of terminal sprouting cannot account for the increase in the number of synapses. Thus, most of the additional synapses in MuSK-H mice appear to arise from collateral branching of motor axons.
Nascent neuromuscular synapses undergo a complex process of maturation during the first few weeks after birth. One of the more striking changes is the elaboration of a plaque-shaped form at birth to a pretzel-shaped form postnatally (Sanes and Lichtman, 1999; Slater, 1982a, b). This transition occurs normally in MuSK-H mice, since synaptic AChR clusters become perforated and highly branched, like in wild-type mice (Figure 5D, E). In addition to AChR clusters that are associated with nerve terminals, muscles from MuSK-H mice bear numerous ectopic, non-synaptic AChR clusters (Figures 5F, K, L, N, S4), which are maintained postnatally (Figures 5F, S1). Unlike synaptic AChRs, these non-synaptic AChRs do not mature postnatally but instead retain the typical oval plaque-shape of nascent synaptic AChR clusters (Figure 5F). These findings indicate that spatially restricted, nerve-derived signals are required for this aspect of postsynaptic maturation in vivo.
During normal fetal development, individual synaptic sites are innervated by multiple axons, which are gradually eliminated during the first few weeks after birth so that each synaptic site is ultimately innervated by only a single motor axon (Nguyen and Lichtman, 1996). Synaptic sites in MuSK-H mice are likewise multiply innervated during fetal development (Figure 5H), and although the rate of synapse elimination is delayed (Figure 5I), by P30, all synaptic sites in MuSK-H mice are singly innervated (data not shown).
Our data demonstrate that MuSK expression is patterned independent of innervation and that this prepatterning has an important role in regulating where synapses form in muscle. We next sought to determine how MuSK prepatterning is established. Previously, we raised the possibility that the temporal pattern of muscle growth might have an important role in clustering AChRs in the central region of muscle (Yang et al., 2000). Notably, because myoblasts fuse to developing myotubes at their growing ends, the central region of muscle is more mature than the distal ends of the muscle (Williams and Goldspink, 1971; Zhang and McLennan, 1995). Further, MuSK is expressed shortly after myoblasts fuse to form myotubes, and MuSK activation stimulates MuSK protein clustering and MuSK expression (Jones et al., 1999; Lacazette et al., 2003; Moore et al., 2001; Sander et al., 2001). These data raise the possibility that stochastic activation of MuSK in early myotubes, followed by both positive feedback loops, might be sufficient to initiate and sustain MuSK signaling in the central region of the muscle, where it is first expressed and activated (Figure 6H) (Arber et al., 2002).
The skeletal muscle alpha-actin and MuSK promoters share similar temporal patterns of expression, as the actin promoter, like the MuSK promoter, is activated in developing myotubes (Brennan and Hardeman, 1993; Minty et al., 1982). However, unlike the MuSK gene, the actin gene is expressed uniformly throughout the myotube (Fontaine et al., 1988). To determine whether early expression of MuSK might be sufficient to pattern AChR clustering in muscle, we crossed the low-expressing HSA::MuSK-L transgene, which does not perturb AChR pre-patterning (Figure 2F), into mice that lack motor neurons and endogenous MuSK. Despite uniform expression from the actin promoter, we considered the possibility that early expression of MuSK protein, followed by stochastic activation, sustained by the post-translational feedback loop, might be sufficient to concentrate and activate MuSK in the central region of the muscle (Figure 6I).
We generated mice that lack motor neurons and endogenous MuSK but carry the MuSK-L transgene (HB9cre; Isl2DTA; MuSK−/−; HSA::MuSK-L), as well as control mice that lack either motor neurons (HB9cre; Isl2DTA) or motor neurons and endogenous MuSK (HB9cre; Isl2DTA; MuSK−/−), and we examined the pattern of AChR clustering in muscle from E13.5 mice. As shown previously, AChR clusters are patterned in the central region of muscle lacking motor neurons but not in mice lacking both motor neurons and MuSK (Lin et al., 2001; Yang et al., 2001) (Figure 6A, D). However, in mice lacking motor neurons and endogenous MuSK, but carrying the MuSK-L transgene, AChR clusters are patterned in the central region of muscle (Figure 6C), and the width of this zone is indistinguishable from control mice (Figure 6G). These data demonstrate that the actin promoter can replace the MuSK promoter and re-establish muscle prepatterning. Thus, early but uniform MuSK expression is sufficient to pattern the central region of the muscle. These findings are consistent with the idea that this early MuSK expression leads to stochastic MuSK activation, which is reinforced by feedback mechanisms that further cluster and activate MuSK (see Discussion).
Nonetheless, the MuSK promoter is required to maintain muscle prepatterning, since AChR clusters are ultimately scattered throughout most of the muscle by E18.5 in mice that carry the HSA::MuSK-L transgene and lack motor neurons and endogenous MuSK (Figure 6F). These data indicate that different mechanisms initiate and maintain muscle prepatterning, as regulatory elements in the MuSK promoter are required to maintain but not to establish muscle prepatterning.
We demonstrated that high, ectopic MuSK expression dictates where motor axons grow and form synapses. We considered the possibility that Agrin, released from motor axons, as they approach and explore muscle in MuSK-H mice, was required to activate ectopic MuSK and stimulate postsynaptic and presynaptic differentiation. To test this idea, we crossed the MuSK-H or MuSK-L transgene into mice that lack either neural Agrin or all isoforms of Agrin, and we examined AChR clustering and motor innervation. As described previously (Gautam et al., 1996), few AChR clusters are present in agrin mutant mice at E18.5, and most of these AChR clusters are not innervated (Figure 7A, B, G, H). In contrast, in agrin mutant mice that carry either the MuSK-H (Figure 7C, D) or MuSK-L transgene (Figure 7E, F, I, J), AChR clusters are abundant and found throughout the muscle of E18.5 mice. Further, a large number of these AChR clusters are contacted by motor axons (Figure S5), which terminate and arborize in a manner that appears identical to motor axon terminals in wild-type mice (inset in Figure 7D, F, J). Importantly, both the MuSK-H and MuSK-L transgenes rescue the neonatal lethality of mutant mice that lack neural Agrin (Figure 7K) or all Agrin isoforms (data not shown), demonstrating that these synapses are functional. Although the rescued agrin mutant mice are runted, they breathe and survive for several weeks postnatally (Figure 7K). These findings indicate that increasing the level of MuSK expression, as little as 3-fold, is sufficient to form synapses independent of Agrin and to counteract dispersion of AChR clusters by ACh (see Discussion).
Different classes of neurons exhibit a characteristic pattern of axon branching within their target field, leading to a stereotyped pattern of innervation within the appropriate layer or lamina. The roles of incoming presynaptic axons and postsynaptic target neurons in establishing this stereotyped pattern of axon growth and synapse formation are poorly understood. Our findings raise the possibility that regional differences in target fields, established prior to innervation, may have a general role in shaping axon growth and branching in the peripheral and central nervous systems.
Recent studies have demonstrated that muscle is pre-specialized in the central, prospective synaptic region prior to and independent of innervation, and these findings have led to a revised view of the steps and mechanisms that regulate neuromuscular synapse formation (Flanagan-Steet et al., 2005; Lin et al., 2001; Panzer et al., 2006; Yang et al., 2001; Yang et al., 2000). Key features of postsynaptic differentiation are established in a muscle-autonomous, MuSK-dependent manner and motor neurons refine and sharpen, rather than induce this prepattern of postsynaptic differentiation.
Here, we find that MuSK expression is itself prepatterned in muscle, suggesting that the pattern of MuSK activity may establish the muscle prepattern and regulate the pattern of innervation. We tested this idea by overexpressing MuSK uniformly throughout the muscle and found that high, ectopic MuSK expression alters the muscle prepattern and induces AChR clustering throughout the muscle. Moreover, we show that motor axons recognize the disturbed prepattern by growing and forming synapses throughout the muscle. Thus, the muscle prepattern is recognized by approaching motor axons, shaping their growth and promoting synapse formation in the central region of the muscle. Further, we demonstrate that ectopic MuSK expression is sufficient to induce synapse formation independent of Agrin and to rescue the neonatal lethality of agrin mutant mice. Thus, MuSK is sufficient to regulate the pattern of axon growth and to induce presynaptic and postsynaptic differentiation. Finally, our experiments provide insight into the mechanisms that establish muscle prepatterning. We find that a lower level of exogenous MuSK expression, conferred by a heterologous promoter that is activated early during muscle differentiation but ultimately expressed uniformly in muscle, is sufficient to establish muscle prepatterning. These data suggest that the timing and level, rather than the pattern of MuSK expression have a critical role in initiating muscle prepatterning. Below, we discuss the implications of these findings: first, the role of muscle prepatterning in shaping motor axon growth and regulating the pattern of innervation; second, how MuSK may direct synapse formation in the absence of Agrin; third, how the timing and level of MuSK expression may play a critical role in establishing and sustaining the muscle prepattern.
We found that high levels of ectopic MuSK expression in developing muscle induces ectopic AChR clusters, consistent with previous studies which demonstrated that ectopic expression of MuSK in adult myofibers induces ectopic AChR clustering and transcription (Fu et al., 2001; Hesser et al., 1999; Jones et al., 1999; Sander et al., 2001). Here, we find that ectopic overexpression of MuSK alters the pattern of motor axon growth and synapse formation, since motor axons grow and form synapses aberrantly throughout the muscle of mice that carry the MuSK-H transgene. These findings demonstrate that MuSK misexpression disrupts the AChR prepattern and indicate that approaching motor axons normally recognize and respond to the MuSK-dependent muscle prepattern. Consistent with these findings, live imaging of neuromuscular synapse formation in zebrafish suggested that motor axons sense and turn toward prepatterned AChR clusters (Flanagan-Steet et al., 2005; Panzer et al., 2006); our studies demonstrate that motor axons recognize a MuSK-dependent prepattern and are re-routed when the pattern of MuSK expression is altered. Further studies will be necessary to identify the signaling events downstream from MuSK that lead to the production of cell surface and/or secreted proteins that shape motor axon growth and promote synapse formation.
Motor axons provide two signals that act in an opposing manner to modify and sharpen the AChR prepattern (Lin et al., 2001; Yang et al., 2001). Agrin stimulates MuSK and stabilizes postsynaptic differentiation, whereas ACh stimulates AChRs and destabilizes AChR clusters that are not apposed by nerve terminals (Lin et al., 2005; Misgeld et al., 2005). In agrin mutant mice, motor neurons fail to provide Agrin, so prepatterned AChR clusters are extinguished by ACh (Lin et al., 2005; Misgeld et al., 2005). In ChAT mutant mice, which fail to synthesize ACh, non-synaptic AChR clusters are not dispersed, so non-synaptic, prepatterned AChR clusters persist while synaptic AChR clusters are stabilized. The mechanisms by which these two signaling pathways converge are poorly understood, but the combination of these two signals, one promoting and one extinguishing AChR clusters, leads to the selective stabilization of postsynaptic differentiation at nascent synapses (Lin et al., 2005; Misgeld et al., 2005). Our studies show that MuSK overexpression alters this balancing act, since non-synaptic AChR clusters are not extinguished but are maintained in mice that lack Agrin and overexpress MuSK. Taken together with experiments showing that the AChR prepattern is maintained in agrin mutant mice that fail to synthesize ACh (Lin et al., 2005; Misgeld et al., 2005), these studies demonstrate a fine balancing act between the signaling pathways mediated by Agrin/MuSK, which stabilizes the muscle prepattern, and ACh/AChRs, which destabilizes the prepattern.
Surprisingly, neuromuscular synapses form and function in mice that lack Agrin and express a low- or high-expressing MuSK transgene. Moreover, unlike agrin mutant mice, which fail to form synapses and die at birth, these mice survive postnatally. These studies demonstrate that Agrin per se is not required to form functional synapses in vivo and support the idea that the role of Agrin in neuromuscular synapse formation is to enhance MuSK signaling in the context of opposing signals that act to extinguish synapses. Although we cannot exclude the possibility that a ligand, other than Agrin, activates MuSK, previous studies have shown that MuSK, like other receptor tyrosine kinases, can be activated in a ligand-independent manner in transfected cells that overexpress MuSK and in transfected adult myofibers (Hesser et al., 1999; Jones et al., 1999; Sander et al., 2001). Moreover, nerve-independent AChR clustering, unlike AChR clustering at synapses, is strikingly sensitive to MuSK dosage, consistent with the idea that muscle prepatterning is established by ligand-independent MuSK activation (Lin et al., 2001). Thus, ligand-independent activation of MuSK may be sufficient to trigger presynaptic and postsynaptic differentiation in vivo. In either case, it is striking that MuSK activation, independent of Agrin, is sufficient in vivo to restore the formation of functional synapses and to rescue the lethality of agrin mutant mice. Importantly, these studies demonstrate that synapse formation is critically dependent upon MuSK but does not require an adhesive interaction between Agrin and MuSK for motor axons to stop and differentiate (Campagna et al., 1995; Dimitropoulou and Bixby, 2005).
In the absence of motor neurons and endogenous MuSK, AChRs are expressed uniformly in skeletal muscle. Strikingly, uniform expression of MuSK from an actin::MuSK transgene restores AChR patterning to muscles that would otherwise lack any sign of regional specialization. Our findings are consistent with the idea that early MuSK expression, conferred by the α-actin promoter, leads to stochastic dimerization and activation of MuSK protein, which is reinforced by positive feedback mechanisms, leading to further MuSK activation in the central region of muscle where MuSK is first expressed (Figure 6H, I). Although our data do not exclude other models for transforming a uniform pattern of MuSK RNA expression to a spatially restricted pattern of MuSK kinase activity, this temporal model is the most parsimonious and takes account of known properties of MuSK, including positive feedback regulation, and the spatial pattern of muscle growth.
Our data indicate that transcriptional mechanisms, however, are required to maintain and stabilize the muscle prepattern, since in mice that lack motor neurons and endogenous MuSK but carry the MuSK-L transgene, AChR clusters are ultimately scattered throughout most of the muscle at later developmental stages (Figure 6F). Since activated MuSK stimulates MuSK transcription (Moore et al., 2001), these data suggest that a positive-feedback transcriptional loop may have a critical role in stabilizing the muscle prepattern. A better understanding of the transcriptional mechanisms that regulate MuSK expression will be necessary to determine how muscle prepatterning is maintained and how synapses are stabilized.
The HSA promoter/exon1/intron cassette (−2000/+234) was excised from pHSA2000CAT (Brennan and Hardeman, 1993; Muscat and Kedes, 1987) by digestion with HindIII, and the blunt-ended fragment was subcloned into the StuI site of pSG5 (Stratagene). Rat MuSK-GFP cDNA was excised from L29/MuSK-FLAG-GFP (Herbst et al., 2002) with EcoRI and subcloned into pSG5, containing the HSA promoter cassette. The HSA::MuSK sequence, as well as the linked rabbit β-globin intron and SV40 polyadenylation signal, were released from the pSG5 plasmid, using SfiI and XmnI, and the gel-purified fragment was injected into zygotic male pronuclei, as described previously (Simon et al., 1992). Founder mice and F1 progeny were genotyped by PCR. We obtained four founder mice and generated three lines. We studied two lines, a MuSK-L line, which express 3-fold more MuSK than wild-type mice, and a MuSK-H line, which overexpresses MuSK by 20-fold (Figure 1); a third line expressed the transgene at very low levels and was not studied further.
HSA::MuSK (MuSK-H or MuSK-L) mice were genotyped by PCR, as described previously for MCK::MuSK mice (Herbst et al., 2002). HB9cre and Isl2DTA mice have been described previously (Arber et al., 1999; Yang et al., 2001) and were genotyped by PCR (HB9cre: 5′-CCGGTGAACGTGCAAAACAGGCTCTA-3′ and 5′-CTTCCAGGGCGCGAGTTGATAGC-3′; Isl2DTA: 5′-ACGACGCTGCGGGATACTCT-3′ and 5′-CAACGCTAGAACTCCCCTCA-3′). MuSK mutant mice were genotyped by PCR, as described previously (DeChiara et al., 1996; Herbst et al., 2002). AgrinΔZ mice, lacking neural Agrin (Gautam et al., 1996), were genotyped by PCR (5′-GTCAGTGGGGGACCTAGAAAC-3′ and 5′-GTTGCTCTGCAGCGCCTT-3′). Agrin null mice, lacking all Agrin isoforms (Lin et al., 2001), were genotyped by PCR (5′-GGGCTAACACCAACAACAATGCAACAAAGG-3′ and 5′-TGCCAAGTTCTAATTCCATCAGAAGCTGAC-3′), as described previously (Lin et al., 2001). All experiments described here were approved by the Animal Care and Use Committee at NYU Medical School.
Dissected diaphragm muscles were fixed (1% formaldehyde in phosphate buffered saline) (PBS) for 90 min at room temperature and further fixed (0.2% formaldehyde in PBS) overnight at 4°C. Muscles were washed three times for 15 min in PBS, incubated for 15 min with 100 mM glycine in PBS and rinsed in PBS. Overlying connective tissue was carefully removed, and muscles were permeabilized and blocked for 1 h in PBS containing 2% bovine serum albumin, 4% normal goat serum and 0.5% Triton X-100. Axons and nerve terminals were labeled by staining muscles overnight at 4°C with rabbit polyclonal antibodies against NF (1:2000; Chemicon, Temecula, CA) and Syn (1:5; Zymed, San Francisco, CA) in blocking solution. After three 1 h washes in PBS, containing 0.5% Triton X-100 (PBT), muscles were incubated at 4°C overnight with Alexa-488 goat anti-rabbit IgG (1:250; Invitrogen) and Alexa-594-conjugated-α-BGT (1:1000; Invitrogen, San Diego, CA) in blocking solution, to label AChRs. After three 1 h washes in PBT and one 20 min wash in PBS, muscles were post-fixed (1% formaldehyde in PBS) for 10 min, rinsed in PBS and mounted under glass in Vectashield (Vector Labs, Burlingame, CA).
Diaphragm muscles from P0 wild-type, MuSK-L and MuSK-H mice were stained with antibodies to NF and Syn and Alexa-594–α-BGT. Confocal images of the left hemi-diaphragm were captured on a Zeiss 510 confocal laser scanning microscope (Carl Zeiss MicroImaging GmbH, Jena, Germany). These image stacks were compiled into a reconstructed image of the left hemi-diaphragm muscle. To measure the distribution of AChR clusters and axons within the muscle, we divided this reconstructed image into nine 100 μm wide strips, oriented parallel to the main intramuscular nerve; strip 1 was the most costal strip, strip 5 was centered over the main intramuscular nerve, and strip 9 was the most medial strip. The number of pixels in each strip (green channel for axons and red channel for AChRs) was determined using Metamorph (Molecular Devices, Sunnyvale, CA). The number of pixels in the central, synaptic strip (strip 5) was assigned a value of 100%, and the values in other strips are expressed relative to strip 5. Data were collected from four or more diaphragm muscles from each genotype.
We collected stacks of overlapping, confocal images, extending from the medial to the costal edge of the left hemi diaphragm muscle in P0 wild-type and MuSK-H mice. Following assembly of these image sets, we measured the number and position of synapses in the muscle. Reconstructed images were divided into 15 strips, and the number of synapses and non-synaptic AChR clusters were determined within each strip. Data were collected from three or more diaphragm muscles from each genotype.
AChR cluster size and density was measured, as described previously (Herbst et al., 2002; Jaworski and Burden, 2006). Briefly, we stained P0 diaphragm muscle from wild-type, MuSK-H and MuSK-L mice with Alexa-594–α-BGT, and collected confocal image stacks at the same subsaturating amplifier gain for all genotypes. The data was quantified using Volocity 3D software (Improvision, Lexington, MA). At least 70 synapses in each P0 embryo from at least 4 animals per genotype were analyzed.
Single muscle fibers were isolated from diaphragm and extensor digitorum longus muscles from P0 wild-type and MuSK-H mice by digestion with collagenase (0.2% w/v) at 37°C for 30 min (Shefer and Yablonka-Reuveni, 2005). After washing in PBS, the dissociated muscle fibers were fixed (1% paraformaldehyde) for 30 min at room temperature, washed in PBS and stained with Alexa-594–α-BGT and Alexa-660-phalloidin, to label AChRs and myofibers. Images of the stained, individual muscle fibers were captured by confocal microscopy, and the number of AChR clusters per muscle cell was determined. We examined 200 to 400 single fibers from each muscle and at least three mice for each genotype.
Intercostal muscles from wild-type, MuSK-L and MuSK-H mice were fixed (4% formaldehyde in PBS) overnight at 4°C, dehydrated in methanol, digested for 30 min with 20 μg/ml proteinase K, hybridized with digoxigenin-labeled riboprobes directed against mRNAs encoding the AChR δ subunit (Simon et al., 1992) or MuSK (Herbst et al., 2002) and processed as described previously (Yang et al., 2001). Weak, uniform staining was observed with sense probe for AChR δ subunit and MuSK (data not shown).
Diaphragm muscles were homogenized on ice in RNA STAT-60 (Tel-Test, Friendswood, TX) using a PT 10/35 Polytron (Kinematica AG, Littau-Lucerne, Switzerland), and total RNA was isolated according to the manufacturer’s instructions. After treatment with DNase I for 15 min at 37°C, 200 ng of RNA was reverse transcribed (RT) using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA). Quantitative PCR was performed using iQ SYBR Green Supermix (Bio-Rad, Hercules, CA) on an Opticon 2 real-time thermocycler (Bio-Rad Hercules, CA), using 5 percent of each RT reaction. The primers used for PCR amplification (Figure 1F), which amplify RNA transcribed from the endogenous MuSK gene as well as the MuSK transgene, were: 5′-CCCTGCAAGTGAAGATGAAA-3′ and 5′-TTCAAGAACTGCGATTCTGG-3′ for MuSK, as described previously (Jaworski and Burden, 2006), and 5′-CGTGTCACCTCTGCTGCT-3′ and 5′-CCTTCATATTGCCTCCCTTCT-3′ for muscle creatine kinase (mck) (Jaworski, unpublished data). MuSK expression was normalized to mck expression. The linearity of the real time PCR reaction was confirmed by analyzing serial dilutions of samples, and each reaction was performed at least in duplicate. Expression levels were determined from at least four animals for each genotype.
To analyze the number and size of myofibers, diaphragm muscles from P0 wild-type and MuSK-H mice were immersion-fixed (in 1% glutaraldehyde and 4% paraformaldehyde in 0.1 M sodium phosphate, pH 7.2) for 90 min at room temperature, washed three times with isoosmotic phosphate buffer, washed three times in 0.1 M Tris, pH 7.2, and treated for 1 h in 1% osmium. After a 1 h wash in water, diaphragm muscles were dehydrated in ethanol and embedded in Epon. For light microscopy, cross sections of anatomically matched areas of the diaphragm muscle were stained with toluidine blue, and images were captured on a Sony DKC-500 camera. The cross-sectional area of individual myofibers was measured using NIH ImageJ, and the number of muscle fibers in a defined area of the diaphragm muscle was counted. Myofiber number and size were determined from at least 3 animals for each genotype.
Spinal cords from P0 wild-type and MuSK-H mice were fixed (4% formaldehyde in PBS) at room temperature for 2 h, washed three times in cold PBS, equilibrated in 15% sucrose in PBS for 2 h at room temperature, followed by equilibration in 30% sucrose in PBS overnight at 4°C. 10 μm frozen sections from the cervical, thoracic and lumbar spinal segments were collected and stained with a rabbit polyclonal antibody against Islet1/2 (CU321). The number of Islet1/2-positive neurons, located in the ventro-lateral portion of the spinal cord, from cervical, thoracic and lumbar segments was counted in every third section. We examined spinal cord sections from at least three animals for each genotype.
We are grateful to Tom Jessell for kindly providing antibodies to Islet1/2 as well as HB9cre mice and Isl2DTA mice, Josh Sanes for kindly providing agrinΔZ and agrin null mice, and Edna Hardeman for kindly providing pHSA2000CAT. We thank Jihua Fan for excellent technical assistance, and Jeremy Dasen, Wenbiao Gan and Dan Littman for their comments on the manuscript. This work was supported with funds from the NIH and the Robert Packard Center for ALS Research.
The authors do not have a conflict of interest related to this work.