Forced expression of muscle specific kinase slows postsynaptic acetylcholine receptor loss in a mouse model of MuSK myasthenia gravis

Abstract We investigated the influence of postsynaptic tyrosine kinase signaling in a mouse model of muscle‐specific kinase (MuSK) myasthenia gravis (MG). Mice administered repeated daily injections of IgG from MuSK MG patients developed impaired neuromuscular transmission due to progressive loss of acetylcholine receptor (AChR) from the postsynaptic membrane of the neuromuscular junction. In this model, anti‐MuSK‐positive IgG caused a reduction in motor endplate immunolabeling for phosphorylated Src‐Y418 and AChR β‐subunit‐Y390 before any detectable loss of MuSK or AChR from the endplate. Adeno‐associated viral vector (rAAV) encoding MuSK fused to enhanced green fluorescent protein (MuSK‐EGFP) was injected into the tibialis anterior muscle to increase MuSK synthesis. When mice were subsequently challenged with 11 daily injections of IgG from MuSK MG patients, endplates expressing MuSK‐EGFP retained more MuSK and AChR than endplates of contralateral muscles administered empty vector. Recordings of compound muscle action potentials from myasthenic mice revealed less impairment of neuromuscular transmission in muscles that had been injected with rAAV‐MuSK‐EGFP than contralateral muscles (empty rAAV controls). In contrast to the effects of MuSK‐EGFP, forced expression of rapsyn‐EGFP provided no such protection to endplate AChR when mice were subsequently challenged with MuSK MG IgG. In summary, the immediate in vivo effect of MuSK autoantibodies was to suppress MuSK‐dependent tyrosine phosphorylation of proteins in the postsynaptic membrane, while increased MuSK synthesis protected endplates against AChR loss. These results support the hypothesis that reduced MuSK kinase signaling initiates the progressive disassembly of the postsynaptic membrane scaffold in this mouse model of MuSK MG.


Introduction
Most cases of myasthenia gravis (MG) are caused by autoantibodies that target the nicotinic acetylcholine receptor (AChR) in the postsynaptic membrane of the neuromuscular junction (NMJ; Berrih-Aknin et al. 2014). The major pathogenic mechanisms involve bivalent AChR autoantibodies cross-linking adjacent AChR to cause accelerated AChR degradation (antigenic modulation) and complement-mediated damage to the postsynaptic membrane (Baggi et al. 2012;T€ uz€ un and Christadoss 2013;Ban and Phillips 2015). However, a proportion of MG patients instead express autoantibodies that target the extracellular domains of muscle-specific (tyrosine) kinase (MuSK; Hoch et al. 2001;McConville et al. 2004;Reddel et al. 2014) or its co-receptor, low-density lipoprotein receptor-related protein 4 (LRP4; Higuchi et al. 2011;Pevzner et al. 2012;Zhang et al. 2012;Zisimopoulou et al. 2014). The mechanisms through which MuSK autoantibodies cause the loss of postsynaptic AChR at the NMJ in vivo remain to be fully defined.
Our understanding of the pathogenesis of MuSK autoantibodies depends upon animal and cell culture experiments . Rabbits, rats, and mice that were actively immunized with the extracellular domain of MuSK developed MuSK autoantibodies. They showed impairment of neuromuscular transmission associated with loss of postsynaptic AChR and other changes to the NMJ structure Shigemoto et al. 2006;Xu et al. 2006;Richman et al. 2011;Chroni and Punga 2012;Viegas et al. 2012;Patel et al. 2014;Ulusoy et al. 2014). Passive transfer of IgG from MuSK MG patients into mice also caused loss of postsynaptic AChR and synaptic impairments Viegas et al. 2012). Most of the autoantibodies from MuSK MG patients are of the IgG4 subtype (McConville et al. 2004), and it has been demonstrated that administering the IgG4 fraction from MuSK MG plasma to mice was sufficient to cause myasthenia gravis (Klooster et al. 2012). However, antibodies of the IgG4 subtype are considered functionally monovalent and would not be expected to cross-link MuSK or activate complement (Tao et al. 1991;Aalberse et al. 2009). These observations suggest that MuSK autoantibodies drive the development of pathology via an alternative mechanism. A proteoglycan called neural agrin (secreted by the motor nerve) can activate MuSK by binding to its co-receptor, LRP4 Zhang et al. 2008). Upon agrin-induced activation of MuSK, nonreceptor tyrosine kinases including Abl and Src become phosphorylated, leading to tyrosine phosphorylation of the AChR b-subunit (Mohamed et al. 2001;Smith et al. 2001;Mittaud et al. 2004). Phosphorylation of the AChR recruits the cytoplasmic protein rapsyn (Borges et al. 2008). Hypothetically, this extra rapsyn might stabilize endplate AChR by: (1) cross-linking adjacent AChR (Zubera and Unwin 2013); (2) anchoring AChR to the cytoskeleton (Lo et al. 1980;Phillips et al. 1993); (3) recruiting protein kinase A, which fosters the local recycling of AChR (Choi et al. 2012); and (4) inhibiting calpain and caspase-3, which may protect the AChR cluster from calcium-mediated disassembly mechanisms (Chen et al. 2007;Wang et al. 2014). In cell culture experiments, MuSK autoantibodies disrupted the binding of LRP4 to MuSK, thereby preventing the activation of MuSK by neural agrin (Huijbers et al. 2013;Koneczny et al. 2013). These in vitro findings raise the hypothesis that MuSK autoantibodies disrupt maintenance of the adult NMJ in vivo simply by suppressing MuSK signaling function within the postsynaptic membrane. Here, we examine the early pathophysiological changes at the NMJ and the influence of MuSK expression on disease progression in a mouse model of MuSK MG.
The passive transfer mouse model has been used to examine the effects of MuSK autoantibodies on the NMJ in vivo. Mice administered daily injections of IgG from MuSK MG patients displayed progressive declines in endplate AChR density and endplate potential amplitude, culminating in failure of neuromuscular transmission after 12-15 days (Morsch et al. 2012(Morsch et al. , 2013. Src kinase (Y418) and the b-subunit of the AChR (b-subunit Y390) are key targets of the tyrosine kinase cascade initiated following MuSK activation (Mittaud et al. 2004). After 14 daily injections of IgG from MuSK MG patients, motor endplate immunolabeling for the phosphorylated Src-Y418 (pSrc) and AChR b-subunit-Y390 (pAChR) were both diminished, consistent with the idea that anti-MuSK IgG suppresses MuSK kinase activity within the postsynaptic membrane . However, at this late stage of the disease process, postsynaptic immunofluorescence for MuSK and AChR were also diminished (Cole et al. 2010;Ghazanfari et al. 2014). Developmental studies have suggested that structural interactions of MuSK with other proteins might play a role in organizing the specialized postsynaptic membrane (Apel et al. 1997;Zhou et al. 1999;Bromann et al. 2004). Thus, it is possible that the loss of endplate AChR results either from reduced MuSK kinase activity or physical displacement of MuSK from the postsynaptic membrane. Here, we show that a single injection of MuSK MG patient IgG caused reductions in endplate levels of phosphorylated Src and AChR within 24 h of a single injection of MuSK MG patient IgG. Repeated daily injections of MuSK MG patient IgG caused progressive loss of AChR from the endplate. However, forced expression of recombinant MuSK provided protection against this AChR loss, despite the continually circulating activation-blocking autoantibodies.

Ethical approval
All mouse experiments were conducted with the approval of the University of Sydney Animal Ethics Committee in compliance with the NSW Animal Research Act 1985 and the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes 8th Edition NHMRC 2013. Six-week-old female C57BL/6J mice (approximately 19 g) were obtained from the Animal Resources Centre, Western Australia. Mice were fed ad libitum and had continuous access to water. A total of 35 mice were used in the study and the group size for each experiment was 3, unless otherwise stated. The authors understand the ethical principles under which this journal operates and confirm that the work described in this paper conforms to the journal animal ethics checklist. Concerning the use of human IgG, informed patient consent was obtained in accordance with the Declaration of Helsinki. The project was approved by the Human Research Ethics Committee of Sydney South West Area Health Service.

Adeno-associated viral vector expression constructs and injections
Full-length murine MuSK coding sequence fused in-frame at its 3 0 (cytoplasmic) end to enhanced green fluorescent protein (MuSK-EGFP) was described previously . The 3.4 kb MuSK-EGFP open reading frame was blunt cloned into the plasmid backbone, pAAV-CMV-MCS-SV40pA prior to packaging into recombinant adeno-associated virus serotype 6 capsid (rAAV) as described (Blankinship et al. 2004). The 1.9 kb rapsyn-EGFP open reading frame (Gerv asio and Phillips 2005) was similarly blunt-end cloned into the above vector plasmid and packaged into rAAV.
For anesthesia, mice were first placed in an induction chamber supplied with a continual source of 4% isoflurane in oxygen, until rendered unconscious. The mouse was then transferred to a procedure space and a facemask was fitted, to maintain a plane of anesthesia at which the foot withdrawal reflex remained fully suppressed (1.5-2% isoflurane). The skin of the lower hind limbs was shaved and cleaned with 70% ethanol prior to a surgical incision to expose the lateral surface of the tibialis anterior (TA) muscle. A Hamilton syringe with a 26-gauge needle was used to inject 10 lL of 0.9% sterile NaCl solution containing rAAV-MuSK-EGFP or rAAV-rapsyn-EGFP into the belly of the right TA muscle. The dose was 2 9 10 9 viral genomes (vg), unless otherwise specified. In a pilot experiment, this dose yielded MuSK-EGFP fluorescence at the majority of endplates ( Figure S1). The left TA muscle (contralateral control) received the same dose of empty rAAV vector (no gene inserted). The incisions were suture closed and the mice were administered injections of buprenorphine (0.03 mg/kg subcutaneously; Reckitt Benckiser, Australia) immediately, and again 24 h postsurgery for analgesia.

Passive transfer of MuSK MG IgG
The passive transfer model of MuSK MG has been detailed previously Morsch et al. 2012;Phillips et al. 2015). Figure 1 illustrates the experimental design for this study. Mice received daily i.p. injections of IgG derived from therapeutic plasma exchange of anti-MuSK-positive MG patients (described below). To assess the immediate impact of the autoantibodies, a subset of mice was killed 24 h after a single i.p. injection of MuSK MG patient IgG. To assess the longer term effects of anti-MuSK, another subset of mice were killed after 11 daily i.p. injections of patient IgG. Overt weakness does not develop in this model until day 13-14. To suppress any active immune response to the human proteins, the latter mice received a single i.p. injection of cyclophosphamide monohydrate (300 mg/kg; Sigma, St Louis MO; 10 mg/ mL in 0.9% NaCl) 24 h after the first IgG injection (Toyka et al., 1975). On each day of the IgG injection series, mice were weighed, inspected for general health, and were graded for whole body weakness, before and after a standard exercise regime Phillips et al. 2015). Mice were terminated if they either developed grade 2 weakness (mouse lying prone before exercise) or lost 15% of its original body weight. None of the mice in this study reached either of these criteria. Mice were killed with pentobarbitone (30 mg i.p.; Cenvet Australia).
The method of patient IgG purification and the batch numbering scheme was described previously Ghazanfari et al. 2014). Patient 4 was at Myasthenia Gravis Foundation of America (MGFA) grade 3B when batch AM4.5 was collected. Patient 2 was at MGFA grade 4B for batch AM2. Patient 7 was at MGFA grade 3B for AM7.2. Control IgG was purified from pooled plasma arising from therapeutic venesection of hemochromatosis patients and was screened negative for both anti-AChR and anti-MuSK by standard clinical immunoprecipitation assays. The specific batches and amounts of IgG injected for each experiment are specified in the figure legends.

Electromyography
Compound muscle action potentials (CMAPs) were recorded from the TA muscles of mice during repetitive nerve stimulation (Morsch et al. 2012;Plomp et al. 2015). Mice were anesthetized with isoflurane/oxygen as described above. The skin was prepared with abrasive skin prepping gel (Nuprep, D.O. Weaver & Co, Aurora). Two custom-made single monopolar 3 mm recording electrodes were glued to the skin: one over the ventral aspect of the TA muscle and the second electrode at the ankle of the same hind limb. Electrolyte gel (VIASYS Healthcare, Madison WI) was applied directly at the electrode sites. Stimulation of the sciatic nerve was accomplished via a 4 mm incision in the sciatic notch and by placing the nerve on a custom-made silver hook electrode (0.6 mm diameter). Animals received at least 3 trains of 10 stimuli and the recordings obtained were averaged. At the end of the CMAP recordings, the mice were killed with pentobarbitone as described above.

Immunolabeling and confocal imaging
Muscles were dissected and fixed for 2 h in 2% paraformaldehyde/phosphate-buffered saline (PBS) at room temperature. After fixation, muscles were washed three times with PBS over 30 min and were immersed in 20% sucrose/PBS overnight at 4°C. The following day, the muscles were frozen for sectioning as described (Tse et al. 2014). Sections were preincubated for 1 h in 2% bovine serum albumin (BSA) in PBS. To compare the relative abundance of MuSK at the endplate, 12 lm transverse sections were probed overnight at 4°C with affinity purified sheep anti-MuSK (1:100), which binds the extracellular domain of both endogenous MuSK and MuSK-EGFP. Its binding is not blocked by preincubation with high concentrations of MuSK MG patient IgG (Cole et al. 2010). After washing in PBS, sections were incubated with Alexa647-a-bungarotoxin (1:200; Alexa647-BGT; Molecular Probes) and affinity-purified tetramethylrhodamine isothiocyanate (TRITC)-conjugated donkey anti-sheep IgG (1:250; H+L; Jackson ImmunoResearch Laboratories, Baltimore, PA) for 1 h at room temperature. Immunolabeling for phosphorylated Src-Y418 (pSrc) and AChR b-subunit-Y390 (pAChR) were performed as described  To permit comparison of endplate staining intensities across all sections in a given experiment, all samples were stained together on the same day and optical sections were collected in the same imaging session with fixed settings on a Zeiss LSM 510 Meta confocal microscope using a 409 1.2 NA air objective. The confocal aperture was set to 1.0 Airy unit. For enface imaging of endplates in 20 lm longitudinal sections, Z-series were collected with optical section intervals of 1 lm and a scan speed of 7. Each optical section was the average of two scans.

Image analysis
ImageJ 1.31 v software was used to measure the relative fluorescence intensities using recently published protocols This results in progressive declines in endplate AChR labeling and endplate potential amplitude. Weight loss and whole body weakness develops by days 12-15 Morsch et al. 2012). In this study, the acute (immediate) impact of MuSK MG IgG upon the motor endplate was assessed 24 h after the first IgG injection. The longer term effects were assessed (in a second set of mice) at day 11. Recombinant adeno-associated viral vector encoding MuSK-EGFP or rapsyn-EGFP was injected unilaterally into the right TA muscle several weeks prior to the first IgG injection. The contralateral (control) muscle was injected with empty rAAV vector.  (Tse et al. 2014). A minimum of 20 endplates from each muscle were averaged. To assess the relative intensity of MuSK-EGFP fluorescence in the postsynaptic membrane in Figure 5, the ImageJ "plot profile" tool was applied to a line transecting the endplate in a muscle fiber cross-section. The first fluorescence peak represented the endplate while the second peak represents the relative density of MuSK-EGFP in extrasynaptic sarcolemma (on the opposite edge of the muscle fiber). The endplate/sarcolemma peak height ratio therefore reflects the degree to which MuSK-EGFP was concentrated at the endplate. The cytoplasmic EGFP fluorescence intensity was averaged for the cytoplasmic portion of the line profile. Differences in the dosage of rAAV injected (2.5, 7.5, 10, 15 & 20 9 10 8 vector genomes per muscle) did not alter the endplate/sarcolemmal EGFP intensity ratio nor the endplate/ cytoplasm ratio ( Figure S2).

Statistics
GraphPad Prism (GraphPad Software, CA) was used for statistical analysis. Comparison between two groups of mice was by unpaired, two-tailed Student's t-test (e.g., anti-MuSK-injected vs control mice) where n was the number of mice per treatment group. Paired t-tests were used to compare MuSK-EGFP-expressing right TA muscles with contralateral control muscles (injected with empty rAAV vector) in the same mouse. Again, n was the number of mice. Significance is indicated throughout as follows: *P < 0.05, **P < 0.01, ***P < 0.001.

Results
Immediate effects of anti-MuSK IgG upon the postsynaptic MuSK pathway Repeated daily injections of MuSK MG patient IgG into mice lead to slow progressive declines in endplate AChR density and endplate potential amplitude, culminating in failure of neuromuscular transmission after 12-15 days (Morsch et al. 2012). We used phosphospecific antibodies to examine the impact of anti-MuSK IgG at the beginning of the anti-MuSK injection series. Mice were given a single i.p. injection of MuSK MG IgG (batch AM4.5; 35 mg) or non-myasthenic control IgG and were killed for analysis 24 h later (Fig. 1). Endplates of the control mice showed intense staining for phosphorylated Src-Y418 (pSrc; Fig. 2A-C), AChR b-subunit-Y390 (pAChR; Fig. 2G-I), and total AChR (Alexa647-a-BGT binding), consistent with previous findings .
After the single injection of MuSK MG IgG, endplates revealed no measurable reduction in total AChR labeling intensity compared to control mice (Fig. 2D,J&M), con-  sistent with previous findings (Morsch et al. 2012). In contrast, immunolabeling for pSrc and pAChR were both significantly reduced (Fig. 2E,K&M). Thus, in the mouse passive IgG transfer model of MuSK MG, anti-MuSK IgG caused rapid reductions in pSrc and pAChR at the endplate, prior to any measurable decline of endplate AChR density.

Synaptic targeting of MuSK-EGFP in healthy mice
Recombinant adeno-associated viral vector (rAAV) was used to increase the synthesis of MuSK within the muscle fibers. Three weeks after i.m. injection of rAAV into the TA muscle of na€ ıve mice, MuSK-EGFP fluorescence was evident in the sarcolemma, and was concentrated at the motor endplate (Fig. 3). In a pilot dose-response experiment, the intensity of MuSK-EGFP fluorescence at the endplate reached a plateau with injection of 2 9 10 9 vg of rAAV (Fig. 4I). Some MuSK-EGFP-expressing fibers also revealed punctate intracellular EGFP fluorescence with a mean diameter of 1 lm ( Figure S3). Immunolabeling for MuSK is normally restricted to the motor endplate ( Fig. 4E-H; Cole et al. 2010;Ghazanfari et al. 2014). Muscles expressing MuSK-EGFP revealed additional MuSK immunolabeling extending right around the circumference of the muscle fiber, consistent with supraphysiological expression levels (Fig. 4B). Despite the overall increase in MuSK abundance, the average intensity of immunolabeling for MuSK at the endplate (total MuSK) did not increase, compared to endplates in contralateral muscles that were injected with empty vector (Fig. 4J "MuSK"). Thus, in healthy muscles, the endplate density of MuSK appears to have a ceiling. AM7 or AM 4.5) or control IgG and were killed for analysis 24 h later. In mice that received control IgG, MuSK-EGFP fluorescence was approximately fourfold more concentrated at the endplate compared to the extrasynaptic sarcolemma ( Fig. 5A&C open bars) and was approximately 17 times more concentrated at the endplate compared to the cytoplasm (Fig. 5D). These values were comparable to mice that had not received any IgG (Figure S2). The single injection of MuSK MG IgG (AM4.5 or AM7) did not alter the mean intensity of MuSK-EGFP fluorescence at the endplate ( Fig. 5B compare filled bars with open bar), nor the degree to which MuSK-EGFP remained concentrated in the postsynaptic membrane (Fig. 5C&D). Thus, 24 h exposure to MuSK MG IgG did not displace MuSK-EGFP from the endplate.

Elevated MuSK synthesis ameliorates anti-MuSK-induced synaptic impairment
While a single injection of MuSK MG IgG did not displace MuSK-EGFP from the endplate, repeated daily injections of IgG from MuSK MG patients have been shown to reduce endplate levels of both MuSK and AChR (Cole et al. 2010;Ghazanfari et al. 2014). We tested whether MuSK-EGFP could ameliorate this loss of MuSK and AChR. The right TA muscle was administered rAAV-MuSK-EGFP while the contralateral muscle received empty rAAV vector. Mice were subsequently challenged with 11 daily injections of MuSK MG IgG (Fig. 1) so as to produce a robust reduction in endplate AChRs without the wholebody weakness that occurs after a longer injection series   (Morsch et al. 2012). Accordingly mice injected with AM4.5 IgG (35 mg/day) exhibited only a modest weight loss by day 11 while mice injected with AM2 (45 mg/day) did not show any noticeable change in body weight (Fig. 6A). None of the mice displayed impaired posture or mobility according to standard criteria . Nevertheless, compound muscle action potential (CMAP) recordings revealed evidence of "subclinical" impairment of neuromuscular transmission, as expected (Morsch et al. 2012). Recordings from the left TA muscles (empty vector) revealed a marked decrement in the CMAP amplitude during repetitive nerve stimulation (3 impulses/ sec; Fig. 6B open circles). In contrast, right TA muscles (MuSK-EGFP-expressing) from the same mice displayed significantly less decrement (Fig. 6B filled squares). Recordings of CMAPs from mice injected with AM2 IgG (which showed no weight loss) revealed qualitatively similar results, but the trend toward amelioration of the CMAP decrement in the MuSK-EGFP-expressing muscles did not reach statistical significance (Fig. 6C).
Endplates expressing MuSK-EGFP retained more of the normal complement of MuSK immunolabeling after the anti-MuSK injection series than for contralateral muscles that had been injected with empty vector (Fig. 7F,J&M). Importantly, endplates that expressed MuSK-EGFP also retained more intense AChR labeling compared to contralateral (empty vector) muscles (Fig. 7E,I&N). The loss of endplate AChR after repeated daily injections of MuSK MG IgG is often accompanied by the appearance of tiny puncta of AChR labeling in the cytoplasm beneath the endplate . Subsynaptic puncta of AChR were again evident in the left TA muscles (empty vector) after 11 daily injections of MuSK MG IgG (Fig. 8A), but were absent from endplates expressing MuSK-EGFP (Fig. 8B&C). Together, these results suggest that increased expression of MuSK can inhibit the loss of AChR from the endplate that normally results from repeated injections of MuSK MG IgG.

Rapsyn-EGFP did not protect against anti-MuSK-induced AChR loss
An additional group of mice received an i.m. injection of rAAV encoding rapsyn-EGFP (2 9 10 9 vg) into their right TA muscle (with empty vector into the contralateral muscle). The mice were subsequently challenged with 11 daily injections of MuSK MG IgG (AM4.5; 35 mg/day) or non-myasthenic control IgG, as above. In mice injected with control IgG, rapsyn-EGFP targeted strongly to the endplate and to small cytoplasmic puncta (Fig. 9D-F). This is consistent with the subcellular distribution of rapsyn-EGFP found in a previous study (Gerv asio and Phillips 2005). Many of the rapsyn-EGFP puncta that occurred beneath the endplate in these healthy mice also labeled positive for AChR (Fig. 9D-E arrow). Mice that received 11 daily injections of AM4.5 IgG showed a small reduction in body weight (Fig. 9M), and a significant reduction in endplate AChR density (Fig. 9G&N), compared to mice injected with control IgG. Rapsyn-EGFP expression did not rescue endplate AChR labeling intensity compared to contralateral (empty vector-injected) muscles (Fig. 9J&O). Thus, forced expression of rapsyn-EGFP did not significantly protect against anti-MuSK-induced loss of AChR from the endplate.

Discussion
This study supports the idea that MuSK autoantibodies cause the loss of postsynaptic AChR by inhibiting the synapse maintenance function of MuSK. MuSK autoantibodies from MG patients are predominantly of the IgG4 subtype and studies in rodent models indicate that their harmful effects are not dependent upon classical complement-mediated immunopathology (McConville et al. 2004;Klooster et al. 2012;Mori et al. 2012;Phillips et al. 2015). In vitro assays show that MuSK autoantibodies can inhibit assembly of the neural agrin-LRP4-MuSK complex and thereby might prevent the "physiological" activation of MuSK (Huijbers et al. 2013;Koneczny et al. 2013;Otsuka et al. 2015). These findings suggest that interruption of agrin-MuSK signaling activity might be sufficient to cause NMJ disassembly. Our in vivo findings now confirm that MuSK autoantibodies rapidly suppress tyrosine phosphorylation of targets of the MuSK pathway within the postsynaptic membrane. However, repeated passive transfer of MuSK MG IgG into mice eventually leads to depletion of endplate MuSK immunolabeling and AChR density (Cole et al. 2010;Kawakami et al. 2011;Ghazanfari et al. 2014;Otsuka et al. 2015). Conversely, a forced increase in MuSK synthesis ameliorated these effects of MuSK autoantibodies in our mouse passive transfer model of MuSK MG. Together, these findings suggest that the autoantibodies cause synapse failure by reducing the activity of MuSK and subsequently MuSK expression at the motor endplate.  Immunolabeling for pSrc and pAChR provides an indication of potential changes in the tyrosine kinase activity of MuSK within the postsynaptic membrane. Endplates of healthy mice displayed intense immunolabeling for both pSrc and pAChR, but the intensity of both was reduced after 14 daily injections of MuSK MG IgG . We questioned whether this might have been secondary to other pathophysiological changes. Here, we show that prominent reductions in pSrc and pAChR immunolabeling of endplates were evident within 24 h of the first injection of MuSK MG IgG, prior to measurable displacement of MuSK and loss of AChR. This timing is consistent with the notion that reduced MuSK kinase activity is the primary cause of the subsequent loss of endplate AChR. There is reason to suspect that a reduction in AChR phosphorylation might contribute to the subsequent loss of endplate AChRs. When neural agrin is added to cultured muscle cells, it activates MuSK leading to phosphorylation of Src and the AChR. AChR phosphorylation recruits rapsyn and helps link AChR to the cortical cytoskeleton and to protein kinase A (Borges and Ferns 2001;Borges et al. 2008;Choi et al. 2012). Forced expression of the intracellular protein, DOK7 (which enhances the activation of MuSK), increased the phosphorylation of AChR and the size of endplate AChR clusters (Arimura et al. 2014). Conversely, knock-in mutations that precluded tyrosine phosphorylation of the AChR b-subunit caused a small, but significant reduction in the density of AChR at the adult motor endplate (À10%; Friese et al. 2007). Thus, anti-MuSK-induced suppression of MuSK kinase activation could plausibly explain the loss of endplate MuSK and AChR density. Forced expression of MuSK protected endplates from AChR loss despite circulating autoantibodies that are thought to block the activation of MuSK. This paradox might be explained by earlier studies conducted in healthy mice. Forced expression of MuSK in the extrasynaptic portion of the muscle fibre induced the formation of endplate-like AChR clusters, even in experiments that precluded the interaction of MuSK with LRP4 and agrin (Sander et al. 2001;Kim and Burden 2008). Thus, even without activation by agrin and LRP4, the intrinsic kinase activity of MuSK can be sufficient to organize postsynaptic AChRs into clusters, provided the density of MuSK in the membrane is sufficiently high. Forced expression of MuSK in our healthy mice did not significantly increase the endplate density of MuSK. In mice receiving repeated injections of anti-MuSK, the early reduction in endplate protein phosphorylation (Fig. 2), was followed by reductions in the endplate density of MuSK of as much as 75% ( Fig. 7; Cole et al. 2010;Kawakami et al. 2011;Ghazanfari et al. 2014). Forced expression of MuSK-EGFP ameliorated this autoantibody-induced loss of MuSK from the endplate, by maintaining about half the healthy endplate density of MuSK (Fig. 7M&N). Thus, if the activating influence of agrin-LRP4 is interrupted by autoantibodies, retention of postsynaptic AChRs may depend upon the endplate maintaining a certain minimum density of MuSK.
The effects of rapsyn-EGFP appeared to differ from those of MuSK-EGFP. Consistent with previous findings, rapsyn-EGFP targeted precisely to the postsynaptic AChR cluster. Rapsyn also targets to trans-Golgi elements where it normally assembles with newly synthesized AChR (Marchand et al. 2002;Gerv asio and Phillips 2005). Consistent with this, AChR labeling was co-localized with some of the cytoplasmic puncta of rapsyn-EGFP ( Fig. 9D-F). Forced expression of rapsyn was found to protect endplate AChRs in a rat model of MG that involves passive transfer of monoclonal anti-AChR (Losen et al. 2005). Rapsyn-EGFP had no significant protective influence in our passive transfer model of MuSK MG (Fig. 9). Post hoc analysis the MuSK-EGFP results (Fig. 7N) suggests an effect size of 1.65 and statistical power of 0.52 for our experiments (n = 3 mice). Hence, the lack of effect of rapsyn-EGFP in Figure 9 might simply represent type II error. However, MuSK activation has two established effects on rapsyn. Firstly, it slows the proteolysis of rapsyn, thereby increasing the cytoplasmic rapsyn pool (Brockhausen et al. 2008;Luo et al. 2008). Secondly, phosphorylation of the AChR b-subunit recruits extra rapsyn from the cytoplasmic pool to bind to the AChR (Borges et al. 2008). A large pool of rapsyn-EGFP may be of little benefit when tyrosine phosphorylation of the AChR is impaired.
The present results suggest that human MuSK autoantibodies initiate a cascade of pathophysiological changes at the endplate. Inhibition of MuSK activation causes a marked reduction in the intensity of endplate immunolabeling for pSrc and pAChR, consistent with the in vitro finding that MuSK MG IgG4 blocks the activation of the MuSK complex by neural agrin (Huijbers et al. 2013;Koneczny et al. 2013;Otsuka et al. 2015). This seems to be followed by a delayed displacement of MuSK and AChR from the endplate. Forced expression of MuSK ameliorated the autoantibody-induced loss of both MuSK and AChR from the endplate. Further detailed study of the effects of MuSK MG autoantibodies should help deepen our understanding of the mechanisms of postsynaptic differentiation and may reveal new therapeutic strategies for MuSK MG. The intensity of endplate AChR labeling was reduced at the end of the AM4.5 IgG injection series, compared to mice injected with control IgG. Open bars show mean Alexa647-a-BGT fluorescence intensity AE SEM for endplates from empty vector-injected left TA muscles (unpaired t-test, n = 3 mice