Muscle-specific expression of LARGE restores neuromuscular transmission deficits in dystrophic LARGE^myd mice

Abstract

Mutations in several glycosyltransferases underlie a group of muscular dystrophies known as glycosylation-deficient muscular dystrophy. A common feature of these diseases is loss of glycosylation and consequent dystroglycan function that is correlated with severe pathology in muscle, brain and other tissues. Although glycosylation of dystroglycan is essential for function in skeletal muscle, whether glycosylation-dependent function of dystroglycan is sufficient to explain all complex pathological features associated with these diseases is less clear. Dystroglycan glycosylation is defective in LARGE^myd (myd) mice as a result of a mutation in like-acetylglucosaminyltransferase (LARGE), a glycosyltransferase known to cause muscle disease in humans. We generated animals with restored dystroglycan function exclusively in skeletal muscle by crossing myd animals to a recently created transgenic line that expresses LARGE selectively in differentiated muscle. Transgenic myd mice were indistinguishable from wild-type littermates and demonstrated an amelioration of muscle disease as evidenced by an absence of muscle pathology, restored contractile function and a reduction in serum creatine kinase activity. Moreover, although deficits in nerve conduction and neuromuscular transmission were observed in myd animals, these deficits were fully rescued by muscle-specific expression of LARGE, which resulted in restored structure of the neuromuscular junction (NMJ). These data demonstrate that, in addition to muscle degeneration and dystrophy, impaired neuromuscular transmission contributes to muscle weakness in dystrophic myd mice and that the noted defects are primarily due to the effects of LARGE and glycosylated dystroglycan in stabilizing the endplate of the NMJ.

INTRODUCTION

The muscular dystrophies are a heterogeneous group of genetic diseases characterized by muscle degeneration, progressive weakness and often a reduced lifespan. Several severe forms of muscular dystrophy, such as Walker–Warburg syndrome (WWS) or muscle–eye–brain disease (MEB), can also include hypotonia, mental retardation and eye malformations (1). WWS, MEB, Fukuyama CMD, MDC1C and several forms of milder limb-girdle muscular dystrophy (LGMD 2I, 2K, 2M, 2N) share a defect in the post-translational processing of the cell-surface protein dystroglycan and are sometimes termed ‘dystroglycanopathies’ (2). Dystroglycan is encoded by the DAG1 gene, producing a single-polypeptide sequence that is cleaved to form two functional subunits (α and β) which remain associated at the plasma membrane (3). α-Dystroglycan is heavily glycosylated and functions as a receptor for several components in the extracellular matrix, including laminin (4,5), agrin (6) and neurexin (7). α-Dystroglycan is anchored to the extracellular face of the plasma membrane through its non-covalent association with β-dystroglycan, a type I membrane protein (8). β-Dystroglycan, in turn, binds to dystrophin and the rest of the dystrophin glycoprotein complex, thereby creating a transmembrane link that is critical for dystroglycan function. In order to function as an extracellular matrix receptor, glycosylation of α-dystroglycan is essential and has been shown to be reduced or absent in tissues of dystroglycanopathy patients (9). Mutations in the DAG1 gene are rare, and nearly all causative mutations that result in disease have been identified in genes that are thought to encode glycosyltransferases (10). Consequently, these mutations disrupt the function of dystroglycan as a receptor for extracellular ligands in the various tissues where dystroglycan is expressed, which is thought to underlie the broad clinical spectrum observed in patients.

In addition to dystrophin, dystroglycan associates with other proteins within the dystrophin–glycoprotein complex (DGC), and mutations in several DGC components have been shown to disrupt the normal assembly or function of the entire complex, resulting in multiple forms of muscular dystrophy. Muscle fibers undergo a significant degree of mechanical stress, and the DGC is hypothesized to function, at least in part, in the stabilization of the sarcolemma during cycles of contraction and relaxation (5). Within the DGC, dystroglycan functions as a transmembrane bridge between the basal lamina surrounding each muscle fiber (11), via binding laminin, and the intracellular cytoskeleton, through associations with dystrophin, thereby providing structural support to the sarcolemma. Although the importance of this complex in skeletal muscle is unequivocal, dystroglycan is ubiquitously expressed and its functions in non-muscle tissues are not as well understood.

The targeted gene deletion of dystroglycan is embryonic lethal (12) and several studies have utilized tissue-specific deletions to dissect discrete functions of dystroglycan in neural cell types. Dystroglycan is expressed in a specialized DGC in Schwann cells of peripheral nerves (13,14), where it can serve as a receptor for both laminin (15) and agrin (16). Schwann cell-specific deletion of dystroglycan leads to the development of a progressive neuropathy as evidenced by reduced nerve conduction velocity, altered structure and reduced staining of voltage-gated sodium channels at nodes of Ranvier and dysmyelination defects (17). Dystroglycan is also expressed at the neuromuscular junction (NMJ), and mice chimeric for dystroglycan expression demonstrate impaired organization/structure of NMJs (18). Agrin is an essential organizer of the NMJ during muscle development and although dystroglycan can serve as an agrin receptor in muscle, interactions between agrin and dystroglycan are dispensable for acetylcholine receptor aggregation at the postsynaptic membrane during formation of this synapse (19,20). Rather, dystroglycan appears to function both in the assembly of the synaptic basement membrane (21) and in the localization of additional DGC components to the synapse (22), which contribute to the maintenance of the synapse in adult muscle.

Dystroglycan appears to have important functions in both peripheral nerve and the NMJ, but whether these functions are dependent upon its ability to bind extracellular ligands is not well understood. The myd mouse model has a mutation in the like-acetylglucosaminyltransferase (LARGE) gene (23,24), a bifunctional glycosyltransferase, and displays a muscular dystrophy similar to that observed in patients with LARGE mutations (25,26). The structure of the NMJ in myd muscle is also abnormal (27,28), resembling defects observed in dystroglycan-null NMJs (18), and appears to result from impaired NMJ maintenance rather than initial assembly (21). Additionally, peripheral nerve conduction velocity is reduced in both myd and the allelic variant LARGE^enr animal models and coincides with defects in distal nerve myelination (27). These data suggest that dystroglycan glycosylation is essential not only in skeletal muscle but also for peripheral nerve function and NMJ stabilization.

Here, we used muscle-specific overexpression of LARGE to rescue extracellular matrix receptor function of dystroglycan exclusively in myd muscle to determine the degree to which dystroglycan glycosylation in non-muscle tissues contributes to neuromuscular dysfunction in muscular dystrophy. We demonstrate that rescue of dystroglycan function exclusively in striated muscle is sufficient to rescue several symptoms of muscular dystrophy in myd animals including amelioration of muscle pathology, restoration of force production and extension of lifespan. Furthermore, we demonstrate that structural defects at the NMJ and functional deficits in neuromuscular transmission observed in myd mice can be completely restored via rescued glycosylation of dystroglycan in muscle fibers.

RESULTS

Muscle-specific expression of LARGE results in hyperglycosylation of dystroglycan in myd skeletal muscle

Transgenic mice were generated that express human LARGE exclusively in striated muscle tissues, using a muscle creatine kinase (MCK) promoter/enhancer sequence. MCK-LARGE mice demonstrate hyperglycosylation of dystroglycan in skeletal muscle concomitant with significantly enhanced laminin-binding activity. Additionally, MCK-LARGE transgenic (TG) animals demonstrate muscle-specific expression of the myc-tagged LARGE protein as shown by reactivity with the anti-myc 9E10 antibody in whole lysates from both skeletal and cardiac muscle (Fig. 1A). LARGE-myc protein was not detected in non-muscle tissues, including brain, spinal cord and lung. To reveal aspects of the disease in myd mice that result from altered dystroglycan function in non-muscle tissues, MCK-LARGE transgenic mice were crossed onto the myd strain to generate MCK-LARGE/LARGE^myd (TG-myd) mice, transgenic animals homozygous for the LARGE mutation. Western blot analysis of wheat germ agglutinin (WGA)-purified lysates isolated from wild-type (WT), myd and TG-myd animals demonstrated that the transgene was capable of glycosylating α-dystroglycan selectively in cardiac and skeletal muscle (Fig. 1B). Staining with the glycosylation-specific IIH6 antibody demonstrated that even though glycosylation was absent in all myd tissues, muscle from TG-myd animals exhibited significant levels of glycosylation, above that observed in WT muscle. Native dystroglycan runs as a broad band between 120 and 156 kDa, and similar to what is observed in MCK-LARGE mice (not shown), α-dystroglycan from TG-myd muscle ran at a much higher molecular weight, demonstrating that LARGE overexpression is capable of adding or extending additional glycans on α-dystroglycan. Dystroglycan glycosylation was not detected in non-muscle tissues of TG-myd mice, such as brain and peripheral nerve (Fig. 1C and D), which confirmed that the transgene was not active in these tissues. Furthermore, levels of β-dystroglycan expression were similar among all genotypes in all tissues analyzed (Fig. 1B–D, lower panels), which is consistent with previous reports demonstrating that LARGE overexpression in muscle does not alter the composition or level of expression of other DGC components, including dystrophin (29). As a consequence of reduced glycosylation of dystroglycan in myd animals, laminin-binding activity of WGA-enriched samples from both skeletal muscle (Fig. 1E) and brain (Fig. 1F) was also reduced. However, TG-myd animals showed a considerable increase in laminin-binding activity in muscle above that observed in WT muscle (Fig. 1E), whereas laminin-binding activity in the brain was not different from myd mice (Fig. 1F). The fold increase in laminin-binding activity in transgenic skeletal muscle was 4.18 ± 0.77 versus WT skeletal muscle (n = 6 independent experiments, mean ± SEM). Because myd mice exhibit neuronal migration defects in the brain (9), sections of cerebellum were stained to confirm the loss of dystroglycan function in non-muscle tissues of TG-myd animals. Sagittal sections stained with IIH6 and 4′,6-diamidino-2-phenylindole (DAPI) demonstrated a lack of dystroglycan glycosylation in both myd and TG-myd brain (Fig. 2A). Additionally, groups of densely stained granule cells were detected in the molecular layer, indicating a neuronal migration failure during cerebellar development (Fig. 2B). IIH6-stained gastrocnemius muscle sections from the same animals demonstrated that although dystroglycan glycosylation was absent in myd animals, this glycosylation was restored at the sarcolemma in TG-myd mice (Fig. 2C). These data demonstrate that the LARGE transgene is able to selectively glycosylate and restore dystroglycan function in skeletal muscle, whereas non-muscle tissues remain impaired.

Figure 1.

Selective expression of LARGE-myc in striated muscle results in hyperglycosylation and increased laminin binding activity in TG-myd mice. (A, upper panel) Triton-X whole lysates from WT and TG mice stained with the 9E10 anti-myc antibody show LARGE-myc protein expression exclusively in skeletal and cardiac muscle. (A, lower panel) Equal loading of the gel is indicated by Ponceau S staining of the blot before immunostaining. WGA-enriched lysates demonstrate a loss of α-dystroglycan glycosylation in myd animals by a lack of reactivity with the glycosylation-specific IIH6 antibody. (B–D, upper panels) TG-myd mice exhibit hyperglycosylation of α-dystroglycan selectively in cardiac and skeletal muscle as demonstrated by a dramatic increase in the molecular weight of α-dystroglycan. Glycosylation of α-dystroglycan was absent in brain and sciatic nerves of TG-myd animals due to the lack of transgene expression. (B–D, lower panels) Although the extent of α-dystroglycan glycosylation varied between the three different genotypes, expression level of dystroglycan was unchanged as evidenced by similar reactivity with an anti-β-dystroglycan antibody using stripped blots. (E and F) Absence of dystroglycan glycosylation resulted in a reduction in laminin-binding activity compared with WT animals in both myd skeletal muscle and brain. Although laminin-binding activity is similarly reduced in TG-myd brain (F), laminin-binding activity in TG-myd skeletal muscle is significantly enhanced beyond that observed in WT muscle as a result of selective hyperglycosylation of dystroglycan. Binding data shown are representative of binding experiments performed in triplicate. The fold increase in laminin-binding activity in transgenic skeletal muscle was 4.18 ± 0.77 versus WT skeletal muscle (n = 6 independent experiments, mean ± SEM).

Figure 2.

TG-myd mice demonstrate the absence of functional dystroglycan in non-muscle tissue. (A) Parasagittal sections of cerebellum from WT, myd and TG-myd mice stained with IIH6 demonstrate that only the WT brain expresses glycosylated dystroglycan. (B) The absence of glycosylation in myd and TG-myd cerebellum resulted in neuronal migration failure during development as indicated by densely DAPI-stained cells (white arrows) within the molecular layer. Although abnormal dystroglycan glycosylation and function in the cerebellum were evident in both myd and TG-myd brains, (C) skeletal muscle from TG-myd animals contained glycosylated dystroglycan at the sarcolemma.

Hyperglycosylation of skeletal muscle dystroglycan in myd mice ameliorates muscle disease

myd mice demonstrate a progressive muscle disease characterized by ongoing cycles of muscle degeneration and regeneration and an increase in fibrosis resulting from reduced interactions with the extracellular matrix that predispose the muscle to contraction-induced injury (11,24). Therefore, we hypothesized that the selective rescue of dystroglycan function in differentiated muscle would restore interactions with the extracellular matrix and protect skeletal muscle from mechanical injury. Hematoxylin and eosin-stained gastrocnemius sections showed that myd muscle exhibited several signs of muscle pathology, including fibers of heterogeneous size, infiltration of inflammatory cells and multiple regenerating fibers, whereas muscle from TG-myd animals was indistinguishable from WT muscle (Fig. 3A). Sirius red staining showed an increase in fibrosis in myd muscle that was not evident in TG-myd littermates (Fig. 3B). Furthermore, evidence of the ongoing regeneration in myd animals was no longer present in TG-myd muscle (Fig. 3C). The frequency of myofibers containing internalized nuclei in myd mice was significantly elevated [52 ± 3% centrally nucleated fibers in extensor digitorum longus (EDL) muscles, 39 ± 3% in soleus, n = 4 with >280 fibers counted per sample], whereas both WT and TG-myd muscle contained minimal numbers of centrally nucleated fibers (<1% in both WT and TG-myd EDL muscles, 4 ± 1% in WT soleus, 2 ± 0.2% in TG-myd soleus, n = 4 with >200 fibers per sample). Because muscle pathology was no longer evident, we hypothesized that TG-myd animals would also have lower levels of serum creatine kinase activity. Although creatine kinase activity was nearly 3-fold higher in myd animals compared with WT levels, expression of LARGE in skeletal muscle was sufficient to reduce creatine kinase activity to levels equivalent to WT littermates (Fig. 3D). Additionally, although myd mice lose significant body weight in the later stages of life as a consequence of muscle wasting (Fig. 3E), TG-myd animals maintain body weight and are similar in size as WT littermates.

Figure 3.

Muscle disease is ameliorated in TG-myd animals. (A) Hematoxylin and eosin-stained skeletal muscle sections demonstrate amelioration of dystrophy in TG-myd muscle (TG-myd) as indicated by an absence of inflammatory cells and a normalization of fiber size. (B) Muscle sections stained for Sirius Red/picric acid demonstrate extensive fibrosis in myd muscle that is not present in WT and TG-myd mice. (C) Muscle sections co-stained with the anti-laminin L-9393 antibody and DAPI demonstrate restoration of normal muscle architecture and a dramatic reduction in fibers undergoing degeneration or regeneration as indicated by a reduction in fibers with internalized nuclei. (D) Plasma creatine kinase levels were significantly reduced in TG-myd mice to levels of WT animals (n = 16, 13, 15 for WT, myd, TG-myd, respectively). (E) Although body weights of myd animals were significantly reduced due to the progressive loss of muscle mass, weights of TG-myd mice were not different from WT animals (n = 8). P < 0.05 compared with WT (*) by one-way ANOVA with a Dunnett post-test. Data are presented as means ± SEM.

To confirm that muscle-specific overexpression of LARGE in myd mice restored muscle function, soleus and EDL muscles were used to assess contractile function in vitro. Consistent with our previous report (30), soleus and EDL muscles from myd animals demonstrated a reduction in both absolute and specific force production (Fig. 4A). However, total force production in both TG-myd EDL and soleus muscle was significantly greater than myd and fully restored to values measured in WT littermates. Specific force values calculated for EDL muscle of myd mice were ∼63% of that measured in WT animals. Specific force measured in TG-myd EDL muscle was significantly higher than that measured in myd animals and was ∼85% of that measured in WT animals. Although specific forces measured in the soleus muscle were significantly reduced in myd animals, TG-myd muscle was fully rescued and not different from values measured in WT muscle.

Figure 4.

Muscle function is improved in TG-myd mice. Contractile function of EDL and soleus muscle was measured in vitro (n = 6 muscles). (A) Force production in both myd EDL and soleus muscle was significantly reduced, whereas values for TG-myd muscles were not different compared with WT. (B) Although specific force of TG-myd EDL was significantly higher than myd, values for TG-myd EDL were below that of WT muscle. Contraction-induced injury was performed by subjecting EDL muscles to a series of lengthening contractions of 30% strain. (C) myd muscle demonstrated an elevated susceptibility to contraction-induced damage as indicated by an increase in force deficit following one, two and five lengthening contractions compared with WT muscle. Muscles from TG-myd animals did not display a susceptibility to injury any more so than WT animals following two and five lengthening contractions. P < 0.05 compared with WT (*) and myd (#) by one-way ANOVA with a Bonferroni post-test. Data are presented as means ± SEM.

We previously reported that myd muscles composed of predominantly fast-twitch fibers are highly susceptible to contraction-induced injury as a result of reduced interactions between dystroglycan and laminin (30). We hypothesized that restoration of dystroglycan glycosylation specifically in skeletal muscle would be sufficient to restore extracellular matrix receptor function and reduce the susceptibility of myd muscle to contraction-induced injury. Muscle injury was performed by subjecting the EDL to a series of five lengthening contractions of 30% strain in vitro. Consistent with our previous study, force deficits were significantly greater in myd EDL muscle than in WT muscle after each successive lengthening contraction (Fig. 4C). After five lengthening contractions, force production in myd EDL muscle was ∼20% of the initial measured value. In contrast, force deficits measured in TG-myd EDL muscle were significantly lower than values measured in myd muscle after each lengthening contraction, indicating that restored glycosylation of sarcolemmal dystroglycan was sufficient to protect muscle from mechanical injury.

In addition to significant improvements in muscle function, TG-myd animals were much healthier, and in contrast to myd animals, were capable of breeding. myd animals in our colony rarely survive past 40 weeks of age, whereas TG-myd animals are presently as old as 70 weeks (Supplementary Material, Fig. S1A). Additionally, TG-myd animals do not demonstrate the hindlimb clasping behavior exhibited by myd littermates (Supplementary Material, Fig. S1B). Because glycosylation of dystroglycan remains impaired in the central and peripheral nervous system of TG-myd animals, we next wanted to assess motor performance in the absence of muscle disease.

TG-myd mice do not demonstrate deficits in neuronal function

Deficits in motor coordination as a result of either abnormal cerebellar architecture or deficits in peripheral nerve function in myd animals were first tested using an accelerating rotorod protocol. Mice were placed on a stationary rod, and the time each animal was able to remain on the rod once it began rotating was recorded daily for five consecutive days. Although fall latencies were significantly reduced in myd animals, fall latencies were not different between TG-myd and WT animals (Fig. 5A). Additionally, all three genotypes of mice were able to significantly increase the amount of time they spent on the rod by day 5. Although TG-myd mice demonstrate abnormal brain development (Fig. 2B), the performance of TG-myd animals was not different from WT mice, which suggests that this defect does not significantly affect motor function or task learning.

Figure 5.

Neuronal function is improved in TG-myd animals. Motor coordination was tested using an accelerating rotorod. Animals (n = 6) were placed on a stationary rod that rotated at a constant speed of 5 r.p.m. for 60 s and began accelerating at a rate of 0.1 r.p.m./s. The time each animal was able to stay on the rod beginning at rotation onset was recorded for three daily trials over 5 days. (A) WT and TG-myd mice stayed on significantly longer than myd mice during each day, and values measured for the two groups were not different from one another. Each data point is representative of the mean latency for all mice of each genotype. A tail flick assay was used to assess whether neurological dysfunction was evident in myd and TG-myd animals. Tail flick responses (n = 6) were measured as the time it took for each mouse to remove their tail from a heated beam of light. (B) Only myd animals demonstrated a significant delay compared with WT animals. (C) Nerve conduction velocity was measured in sciatic nerve (n = 10). Values measured in myd animals were significantly slower, whereas values in TG-myd were not different from values recorded in WT littermates. *P < 0.05 compared with WT by one-way ANOVA with a Dunnett post-test. Data are presented as means ± SEM.

Deficits in nerve structure and function have been reported in multiple strains of mice that contain mutations in LARGE, and this suggests that LARGE-mediated dystroglycan glycosylation is required for normal peripheral nerve function (27,31). Nerve function was tested to determine whether TG-myd mice similarly demonstrated abnormal nerve function as a consequence of disrupted dystroglycan glycosylation in neuronal tissues. A tail flick assay was performed that measures the ability of an animal to respond to a heated light beam stimulus focused on the tail. As expected, time latencies measured in myd animals were significantly longer than those measured in WT animals (Fig. 5B). Values obtained for TG-myd animals, however, were not different from WT animals. Because performance in a tail flick assay can be dependent upon both nerve and muscle function, nerve function was next measured directly. Although the conduction velocity measured in the sural nerve was not different between the three different genotypes tested (not shown), sciatic motor nerve conduction velocity was significantly reduced only in myd animals (Fig. 5C). However, electron microscopy of sciatic nerve from all three genotypes did not reveal any evidence of pathology (not shown).

Neurotransmission deficits present in myd animals are restored in TG-myd mice

NMJ structure is disrupted in both myd and DG-deficient muscle (18,28). Because the NMJ is formed from both presynaptic and postsynaptic components, we investigated whether rescue of sarcolemmal dystroglycan function was sufficient to restore NMJ architecture in TG-myd mice. Alexa-488-conjugated α-bungarotoxin and an antibody to neurofilament were used to label presynaptic and postsynaptic regions of NMJs in whole-fixed sternocleidomastoid muscle. NMJs in WT muscle exhibited a characteristic pretzel shape, whereas this structure in myd muscle was fragmented in appearance, consistent with previous reports (Fig. 6A) (28). However, muscle-specific expression of LARGE was sufficient to restore NMJ structure in TG-myd muscle such that NMJs were indistinguishable in appearance from those in WT muscle. This suggests a critical importance of sarcolemmal dystroglycan, as opposed to dystroglycan expressed in either presynaptic neurons or perisynaptic Schwann cells, for the normal maintenance of NMJ structure. To address the functional consequence of altered NMJ structure in myd mice, a contraction protocol was performed that utilized paired measurements of force production in the gastrocnemius muscle. This allowed for comparisons to be made in each muscle between the maximum force produced following stimulation of either the sciatic nerve or the muscle directly. If neurotransmission was impaired, direct muscle stimulation would be expected to produce higher maximal force values than when the muscle was stimulated via the nerve. Consistent with observed measurements in the EDL and soleus muscle, specific force values measured in myd gastrocnemius muscle were significantly lower than WT and TG-myd values (Fig. 6B). Additionally, forces measured following muscle stimulation were slightly lower than values obtained following nerve stimulation in both WT and TG-myd animals, whereas the opposite was observed in all myd animals tested. Direct stimulation of myd muscle resulted in a mean 16% increase in maximal force production compared with values measured during nerve stimulation (Fig. 6C). However, no such increase was observed in TG-myd muscle, likely due to the improved NMJ structure. These results suggest that aberrant structure of the NMJ as a consequence of impaired dystroglycan function causes a functional denervation in muscle fibers, and that restoration of dystroglycan function at the postsynaptic membrane of skeletal muscle alone is sufficient to restore these functional defects in neurotransmission.

Figure 6.

NMJ structure and neurotransmission defects are restored in TG-myd mice. Sternocleidomastoid muscles were fixed whole to stain the presynaptic and postsynaptic regions of the NMJ. Acetylcholine receptors were labeled with Alexa-488-conjugated α-bungarotoxin (green), and an antibody to neurofilament (red) was used to label the motor neuron. (A) Representative images of NMJs in WT and TG-myd muscle demonstrate the characteristic pretzel shape, whereas NMJs in myd muscle appear abnormal and fragmented. The white bar indicates 50 µm. Force production of gastrocnemius muscle was measured in situ and stimulated either by the tibial nerve or using a cuff electrode surrounding the muscle. (B) Specific force values measured following muscle stimulation were always slightly lower than those measured after nerve stimulation for each WT and TG-myd animal (n = 5, 4, 5 for WT, myd, TG-myd, respectively). In contrast, specific force values were always higher for myd animals following direct muscle stimulation, indicating a partial functional denervation of fibers. (C) The difference in specific force measured for either nerve or direct muscle stimulation is shown as a percentage of total specific force produced from nerve.

DISCUSSION

Although muscle disease is the prominent and shared feature of all muscular dystrophies, patients with mutations in glycosyltransferases also suffer from severe central and peripheral nervous system impairments as a result of disrupted dystroglycan function (25,32–34). Several studies have highlighted the importance of glycosylation-dependent interactions between dystroglycan and laminin at the sarcolemma that are important for providing critical structural support during muscle contractions (11,35). Disruption of this mechanical link can result in a high susceptibility to contraction-induced damage and is hypothesized to underlie the eventual decline in muscle function observed in DGC-related muscular dystrophies. However, dystroglycan is ubiquitously expressed and a mechanical role for the DGC in non-muscle tissue is less apparent.

Here we demonstrate that muscle-specific restoration of the ligand-binding activity of dystroglycan rescues several features of muscular dystrophy in myd animals. Overexpression of LARGE in myd muscle resulted in significant hyperglycosylation of sarcolemmal α-dystroglycan beyond that observed in WT animals, reestablishing dystroglycan as an extracellular matrix receptor in skeletal muscle. This resulted in a complete attenuation of the muscle pathology observed in myd animals and coincided with a recovery of muscle contractile performance. Additionally, we observed that restored function of dystroglycan at the motor endplate was sufficient to restore normal NMJ architecture, and this corresponded to a functional rescue of neurotransmission deficits observed in myd animals. These results are the first to show that muscle weakness observed in myd mice as a consequence of primary muscle dysfunction is further compounded by a failure in neurotransmission.

NMJ structure is altered in myd muscle, and our results demonstrate that this defect can be reversed via selective restoration of dystroglycan function at the sarcolemma. Additionally, because the structure of the NMJ is more disrupted in myd animals than in other mouse models of muscular dystrophy (28), this suggests that the altered architecture is not simply a byproduct of degenerating muscle but rather due to the distinct requirement of glycosylated dystroglycan in this structure. Although dystroglycan is a glycosylation-dependent agrin receptor (6) and can bind rapsyn (36), interactions with these critical NMJ proteins are not essential for the initial formation of this structure. Instead, these interactions are important for the maintenance of the NMJ in adult muscle by contributing to the formation of the surrounding basal lamina and serving as a scaffold for additional proteins (22). Our results highlight the critical function of postsynaptic dystroglycan in maintaining NMJ structure, the loss of which contributes to muscle disease due to defects in both neurotransmission and primary muscle dysfunction.

In addition to an amelioration of muscle disease, an overall increase in health was observed that included improvements in longevity. Although the cause of death is unknown in myd mice, severe muscle weakness/paralysis leading to reduced food intake and cardiomyopathy (35) may be contributing factors. Unexpectedly, a complete recovery of motor performance was observed in TG-myd animals despite evidence of abnormal cerebellar and brain development that was not rescued by the transgene. Because the cerebellum and motor cortex participate in the coordination of balance and movement, the functional consequence of defective neuronal migration in TG-myd animals was tested using an accelerating rotorod. Ordinarily, comparisons between WT and myd animals using this assay would be confounded by the profound muscle weakness present in myd animals. However, because muscle function was restored in TG-myd animals, comparisons were able to be made with WT mice. Surprisingly, no differences in rotorod performance were observed and both groups of animals were able to improve significantly over the 5-day period. Although myd animals performed poorly, a significant increase in performance was observed over the course of 5 days. These results suggest that motor learning and coordination is not severely affected in myd animals despite evidence of developmental brain defects.

Deficits in rotorod performance have been documented in a Schwann cell-specific deletion of dystroglycan, concomitant with defects in nerve myelination and conduction velocity (17). In a related model of LARGE deficiency, conduction velocity in the sciatic nerve was also reduced and coincided with the presence of large clusters of unmyelinated axons (27). To determine whether the same defects existed in myd animals, sciatic nerves of 40-week-old animals were further analyzed. Because electron microscopy did not reveal any overt pathology in any of the three genotypes, this suggests that the nerve defect is either not 100% penetrant or is extremely mild in our cohort of myd animals. Nerve conduction velocity was measured in both sciatic and sural nerves and deficits were only observed in sciatic nerves of myd animals. Sural nerve conduction velocity (SNCV) was not different between the three genotypes. Because nerve function was restored in TG-myd animals, despite evidence demonstrating that dystroglycan was hypoglycosylated in peripheral nerve, these data suggest that LARGE-mediated glycosylation of dystroglycan is not critical for normal peripheral nerve function. Although this conflicts with studies demonstrating the importance of interactions between dystroglycan and laminin in the myelination of peripheral nerves (37–39), not all nerve defects reported in dystroglycan-null animals are consistent with the exclusive glycosylation-dependent function of dystroglycan in the PNS. LARGE^enr animals do not demonstrate the defects in node elongation and sodium channel clustering that have been reported for Schwann cell-specific dystroglycan-null animals (17,27), which suggests that the formation of axonal nodal domains does not require glycosylation of dystroglycan by LARGE.

To verify the tissue-specific nature of the promoter, multiple tissues were analyzed for LARGE-myc expression, and expression of the transgene was not detected in transgenic non-muscle tissues. Neuronal migration defects were also observed in the cerebellum of TG-myd animals, which confirmed that deficits caused by reduced dystroglycan function in non-muscle tissues (40) were still present in these animals. Therefore, the absence of neuronal defects in TG-myd mice is not explained by leaky expression of the transgene resulting in residual glycosylation and function of dystroglycan in additional cell types.

An interesting explanation that might account for the observed rescue of nerve function may be related to the improvement in either muscle function or restored structure of the neuromuscular synapse. If the impaired muscle function in myd animals negatively influenced motor neuron function, this might explain the reduced conduction velocity observed in the motor sciatic nerve but not in the sensory sural nerve. Maintenance of neuronal connections is dependent upon target-derived retrograde signals such as neurotrophins (41), several of which originate from skeletal muscle, including brain-derived neurotrophic factor, neurotrophin-3, neurotrophin-4 and glial cell-derived neurotrophic factor (42–46), which may contribute to motor neuron survival and/or differentiation (47). There is also evidence to suggest that expression of neurotrophins and their receptors might be altered in muscular dystrophy (48,49). Recently, neuregulin was shown to be important for the stabilization of the NMJ in a mechanism dependent upon the phosphorylation of α-dystrobrevin, a component of the DGC at the NMJ (50). It is intriguing to speculate that disruptions in neurotrophin-related signaling may contribute to muscle weakness in muscular dystrophy, but this hypothesis remains to be addressed.

In this study, we demonstrate via selective restoration of dystroglycan function in skeletal muscle that extracellular matrix receptor function of dystroglycan in striated muscle is sufficient to ameliorate several characteristics of muscular dystrophy. We show that the reported morphological defects in the myd NMJ result in a functional deficit in neurotransmission as a result of altered sarcolemmal dystroglycan function. Because peripheral nerve function defects were not observed in TG-myd animals, this suggests that impaired dystroglycan function in skeletal muscle may cause reciprocal deficits in nerve function as a consequence of either reduced communication at the NMJ or through retrograde signaling from diseased myofibers.

MATERIALS AND METHODS

Animals

MCK-LARGE transgenic animals were generated on a C57BL/6 background and bred onto the myd strain from a house-maintained colony. WT, TG, myd and TG-myd mice used for all experiments were age/sex-matched littermates aged 20–40 weeks unless otherwise noted. Animals were housed in a specific pathogen-free barrier facility in the Unit for Laboratory Animal Medicine at the University of Michigan, and all procedures were approved by the University of Michigan Committee for the Use and Care of Animals.

Western blotting

Tissues were removed from deeply anesthetized animals and immediately frozen on dry ice until further processing. Samples of sciatic nerve were pooled from five mice per genotype. All samples were homogenized in a buffer containing TBS (120–150 mm sodium chloride, 50 mm Tris, pH 7.5) and 1% Triton X-100. Samples were cleared via centrifugation (10 000g, 10 min) and quantified using the DC Assay (Bio-Rad, Hercules, CA, USA). WGA enrichment was performed by incubating Triton-X lysates overnight with WGA-conjugated sepharose (Vector Laboratories, Burlingame, CA, USA) at 4°C while rotating. Beads were washed, and protein was eluted in batch by incubating beads with a buffer containing 500 mm N-acetylglucosamine and 0.1% Triton X-100 in TBS. All buffers contained protease inhibitors (0.5 µg/ml Pepstatin A, 2 kallikrein inhibitor units/ml Aprotinin, 1 µg/ml Leupeptin, 0.4 mm PMSF, 0.6 mm Benzamidine). Samples were separated on 3–15% gradient sodium dodecyl sulfate-polyacrylamide gels and transferred to polyvinylidene fluoride membrane (Millipore, Billerica, MA, USA). Immunoblotting was performed using a blocking/incubation buffer that contained 5% nonfat dry milk dissolved in TBS-T (TBS + 0.05% Tween-20). Final membranes were developed using enhanced chemiluminescence substrate (Thermo Scientific, Rockford, IL, USA). Primary antibodies included a rabbit polyclonal antibody to β-dystroglycan (Santa Cruz Biotechnology, Santa Cruz, CA, USA), mouse monoclonal antibody to the myc epitope (9E10, Sigma-Aldrich, St Louis, MO, USA) and glycosylated α-dystroglycan (IIH6, gift from Dr Kevin Campbell). Secondary antibodies conjugated to horseradish peroxidase were obtained from Jackson ImmunoResearch (West Grove, PA, USA).

Laminin-binding activity

WGA-enriched samples from quadriceps and brain were diluted in laminin-binding buffer (LBB) containing TBS + 1 mm CaCl₂ and coated onto 96-well polystyrene plates overnight at 4°C. Wells were blocked for 1 h in 3% bovine serum albumin diluted in LBB followed by a 2 h incubation with laminin in varying concentrations (Invitrogen/Life Technologies, Grand Island, NY, USA). Wells were washed four times with LBB and incubated with anti-laminin antibody (L-9393, Sigma-Aldrich) for 1 h. Following a wash step, wells were incubated with anti-rabbit IgG conjugated to horseradish peroxidase (Jackson ImmunoResearch) for 1 h. Plates were then washed, incubated with o-phenylenediamine-citrate phosphate buffer and stopped with 2 m H₂SO₄. Plates were read at 495 nm.

Immunofluorescent microscopy and histology

Brains were carefully removed, cut in half along the sagittal axis and immediately frozen on a plastic cover slip placed on dry ice. Muscles were removed, mounted in OCT and immediately frozen in liquid nitrogen-cooled isopentane. Frozen samples were cut into 8 µm cross-sections, using a cryostat and stored at −80°C until further processed for immunofluorescent or chemical staining. For immunofluorescent staining, slides were rehydrated with PBS and blocked for 1–2 h in an incubation buffer containing 5% BSA in PBS. Slides were incubated at room temperature with primary and secondary antibodies for 1–2 h each with 4 × 5 min washes of PBS in between incubations. Final slides were mounted in Permafluor (Thermo Scientific) and imaged with an Olympus BX-51 fluorescence microscope. Primary antibodies used were a mouse monoclonal anti-glycosylated α-dystroglycan (IIH6, gift from Kevin Campbell) and a rabbit polyclonal antibody to laminin (L-9393, Sigma-Aldrich). Nuclei were stained using DAPI (Sigma-Aldrich).

Creatine kinase activity

Serum was collected from the saphenous vein of restrained animals and stored at −80°C. Creatine kinase activity was measured in duplicate using CK NADP Reagent (Cliniqa, San Marcos, CA, USA).

In vitro contractile function measurements

For in vitro measurements, the EDL and soleus muscles were carefully dissected from deeply anesthetized mice. A 5-0 silk suture was tied to the proximal and distal tendons. One tendon was tied to a servo motor (model 300, Aurora Scientific, Aurora, ON, Canada), the other to a force transducer (model BG-50, Kulite Semiconductor Products, Leonia, NJ, USA). The muscle was bathed in Krebs mammalian Ringer solution maintained at 25°C and bubbled continuously with 95% O₂ and 5% CO₂ to stabilize pH at 7.4. The muscle was stimulated by square-wave pulses delivered between two platinum electrodes connected to a high-power biphasic current stimulator (model 701B, Aurora Scientific) and controlled via an IBM-compatible personal computer and custom-designed software (LabVIEW 7.1, National Instruments, Austin, TX, USA). Pulses were delivered with increasing voltage until maximum isometric twitch was determined, at which muscle length was adjusted and optimal muscle length (L_o) determined. L_o was measured with digital calipers and recorded. Stimulation frequency was then increased until maximum isometric force (P_o) was achieved. Muscles were held at L_o and subjected to trains of pulses to generate an isometric contraction (300 ms for EDL, 900 ms for soleus). Cross-sectional area (CSA) was estimated by dividing the muscle wet mass (mg) by the product of fiber length (L_f, mm) and the density of mammalian skeletal muscle (1.06 g/cm³). Specific force (sP_o) was calculated by dividing P_o by the total fiber CSA for each muscle.

Muscle injury protocol

Following measurement of maximum twitch force and P_o, muscles were stimulated (100 ms for EDL, 300 ms for soleus) and held at L_o to allow muscles to develop P_o. Immediately following the isometric contraction, muscles were stretched through a 30% strain relative to L_f at a velocity of 1 L_f/s. Total stimulation time was 400 ms for EDL muscles and 600 ms for soleus muscles. Muscles were then returned to L_o and subjected to four additional 30% lengthening contractions, each with 12 s in between, for a total of five stretches per muscle. The muscle was allowed to rest for 1min, and a final P_o was measured. The force generated after each stretch was recorded during the isometric contraction that immediately preceded the subsequent stretch. Force deficit was calculated as the decrease in P_o observed after each stretch as a percentage of initial P_o. A total of three mice and six muscles per genotype were tested between the ages of 27 and 31 weeks.

In situ contractile function measurements

Contractile function in gastrocnemius muscle was measured in situ in anesthetized mice that were placed on a temperature-controlled platform warmed to 37°C. The muscle was carefully dissected from the surrounding environment and a 4-0 silk suture was tied around the distal tendon, which was subsequently severed and tied to the lever arm of a servo motor (model 305B, Aurora Scientific). The hindlimb was tied securely to a fixed post at the knee. A continuous drip of warmed saline was administered to the muscle to maintain a temperature of 37°C for the entirety of the procedure. A bipolar platinum wire electrode was used to directly stimulate the tibial nerve, and optimal voltage, frequency and muscle length (L_o) were determined. Muscle was then held at L_o, and 300 ms trains of pulses were applied to determine maximum isometric tetanic contraction (P_o). This process was repeated with the exception that a cuff electrode was placed around the proximal and distal ends of the muscle to stimulate the muscle. When force measurements were completed, the muscle was removed and CSA and specific force were calculated as described above. A total of five mice per genotype were tested between the ages of 20 and 40 weeks.

NMJ staining

Sternocleidomastoid muscles were dissected from anesthetized animals and incubated in 1% paraformaldehyde for 20 min at room temperature. Muscles were then rinsed in PBS and incubated in 30% sucrose overnight at 4°C. A cryostat was used to create 30 µm longitudinal sections of each muscle. Immunofluorescent staining was performed similar to that described above using α-bungarotoxin conjugated to AlexaFluor-488 (Invitrogen) and a rabbit polyclonal anti-neurofilament antibody (AB1987, Millipore).

Nerve conduction velocity measurements

Mice were anesthetized with isofluorane (5% induction, 1–2% maintenance), and temperature was maintained at 34°C using a heat lamp. Sterile electrodes were placed in the ankle and the dorsum of the foot. SNCV was determined by antidromically stimulating at the ankle and recording at the foot. SNCV was calculated by dividing the distance by the onset latency. Sciatic-tibial motor nerve conduction velocity (SMNCV) was determined by placing recording electrodes at the dorsum of the foot and orthodromically stimulating at the ankle and sciatic notch. SMNCV was calculated by dividing the distances between the onset latencies. A total of 10 mice were tested per genotype between the ages of 20 and 40 weeks.

Tail flick assay

Tail flick responses were measured using an adjustable red light emitter. The time it took for the mouse to move its tail after the beam was activated was recorded electronically. The light source was set at 25°C and the temperature increased to 70°C over a period of 10 s. A threshold of 10 s was applied to prevent injury to the mice. A total of six mice were tested per genotype between the ages of 20 and 40 weeks.

Rotorod test

Animals were tested using an accelerating rotorod (Economex, Columbus Instruments, Columbus, OH, USA). Mice were placed on the rod which remained stationary for 10 s, after which the speed was set at 5 r.p.m. After rotating at a constant speed for 60 s, the rod began accelerating at a rate of 0.1 r.p.m./s and continued until the animal was unable to remain on the rod. Total time spent on the rod was recorded starting from the time the rod began rotating. Mice were given three trials daily for 5 consecutive days. In between daily trials, animals were returned to their cage for a minimum of 15 min. A total of six animals were tested for each genotype between the ages of 22 and 30 weeks.

Clasping behavior

Clasping behavior was scored blindly according to Guyenet et al. (51). Prior to genotyping, animals were briefly suspended by the base of the tail for 10 s, and the time each animal spent with one or both legs partially or completely retracted was used as the basis for a score between 0 (unaffected) and 3 (severely affected). myd animals were 12–30 weeks of age, whereas WT and TG-myd mice were as old as 62 weeks.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

FUNDING

This work was supported in a part by grants from the National Institutes of Health AG020591 to S.V.B., and HL080388 to D.E.M. D.E.M. received additional support from the Muscular Dystrophy Association. J.D.G. received support from a Rackham Predoctoral Fellowship from the University of Michigan and from the National Institutes of Health through T32 GM008322.