A mechanism in agrin signaling revealed by a prevalent Rapsyn mutation in congenital myasthenic syndrome

Essential revisions:

Reviewer #1:

The neuromuscular junction has for decades been a model synapse for investigating basic mechanisms of synaptic transmission and synapse development.

[…] The data presented are well documented, except that numbers of items examined (AChR clusters, animals) should be given (see e.g. (4.) below). Validations by appropriate controls are convincing,.

We thank Dr. Brenner for his constructive comments that have significantly improved the manuscript.

Reviewer #2:

Essential revisions:

[…] The authors demonstrate by genetic approaches and functional studies that the N88 residue of Rapsyn is essential for NMJ formation in mouse and provide novel mechanistic insights for Rapsyn function in postsynaptic differentiation. Rapsyn N88K mutation strongly impairs functional differentiation of NMJ as early as embryonic day 14. Evidence is presented that the mutation alters the E3 ligase activity of Rapsyn and MuSK-induced Rapsyn tyrosine phosphorylation. The authors further identify Y86 site as one of the main residues phosphorylated by MuSK and show that it is critical for Rapsyn E3 ligase activity and AChR clustering presumably by promoting Rapsyn self-association. Finally, rescue experiments using AAV-mediated Rapsyn expression in N88K mutant muscle indicate that WT-Rapsyn, but not Y86F-Rapsyn ameliorate the NMJ morphological deficits of N88K mutant thigh muscle suggesting a critical role for Y86 in Rapsyn function during NMJ formation. This study provides important mechanistic insights of the molecular mechanisms underlying Rapsyn function in NMJ formation. Although overall the data support the conclusions, some of the results and mechanisms proposed appear over-interpreted and care should be given to the phrasing all throughout the manuscript for accuracy.

We thank the reviewer for their comments “This study provides important mechanistic insights on the molecular mechanisms underlying Rapsyn function in NMJ formation” and constructive critiques and suggestions that have significantly improved the manuscript.

1) The authors claim that their study led to discover a pathway by which signal is transduced from MuSK to Rapsyn. However, the sequential link between MuSK activation, Rapsyn self-association, E3 ligase activity and Rapsyn tyrosine phosphorylation events in AChR clustering is not clear and no evidence is provided to support the sequential model proposed by the authors in the discussion that may only reflect one possibility among many others. The author should clarify their signaling model and propose experimental data that support their conclusions as to the mechanistic frame of Rapsyn function in AChR clustering.

This is an insightful question. We have performed additional experiments to characterize the time courses of MuSK activation, Rapsyn tyrosine phosphorylation, Rapsyn self-association, Rapsyn E3 ligase activity in C2C12 myotubes in response to Agrin stimulation. As shown in revised Figure 6A, 6B and Figure 7I, 7J, MuSK activation occurred within 10 minutes, followed by Rapsyn tyrosine phosphorylation and self-association (CO-IP of Rapsyn-HA and Rapsyn-EGFP in C2C12 cells) (which plateaued around 20 and 90 minutes, respectively). E3 ligase activity was not peaked until 90 min after Agrin stimulation (Figure 6B). These data demonstrate that these events occur sequentially. Together with other data presented in this paper, they support a working model that MuSK stimulates AChR cluster formation by increasing Rapsyn phosphorylation and self-association and thus enhancing its E3 ligase activity. These results are now presented in subsection “Reduced Rapsn tyrosine phosphorylation by N88K mutation” and subsection “Y86 phosphorylation for E3 ligase activity and AChR clustering”.

2) The authors demonstrate that Rapsyn N88K mutation impairs NMJ formation in vivo as early as E14, where only minimal agrin/MuSK signaling is occurring. It would be important to determine whether Rapsyn is tyrosine phosphorylated by MuSK at this early stage and discuss how MuSK leads to Rapsyn tyrosine phosphorylation and self-association in absence of agrin.

As suggested, we looked at Rapsyn tyrosine phosphorylation in E14 muscles. As shown in Figure 6E, 6G, tyrosine-phosphorylation was detected in muscles of E14 WT mice; but the level was reduced in muscles from N88K mice. As demonstrated in the literature, both MuSK kinase activity and Rapsyn are necessary for prepatterning or the formation of aneural AChR clusters (Lin et al., 2001; Yang et al., 2001b); and Rapsyn is known to interact with MuSK in the absence of Agrin (Apel et al., 1997). Further, LRP4 could interact with MuSK and thus maintain its activity in the absence of Agrin (Kim et al., 2008; Zhang et al., 2008). Our data are in agreement with these observations. Together, they suggest that tyrosine phosphorylation of Rapsyn could occur in the absence of Agrin and contributes to the formation of aneural AChR clusters or muscle prepatterning. These points are now described in the revised subsection “Reduced Rapsn tyrosine phosphorylation by N88K mutation” and Discussion section.

3) N88K mutation is located in the third TPR repeat known to promote Rapsyn oligomerization. However, data previously reported that the N88K mutation does not alter the ability of Rapsyn to self-associate in heterologous cells (Ohno et al., 2002) in contrast to the authors findings. This should be included and commented in the manuscript.

Good point. However, we would like to point out that in the paper by Ohno and colleagues (2002), “self-association” was defined as a morphological event (i.e., aggregate formation). In fact, as shown in Figure 4A, N88K Rapsyn could form aggregates in cultured cells, in agreement with the previous work. However, the N88K mutation reduced the co-precipitation of Rapsyn-HA and Rapsyn-GFP, suggesting that N88 is critical to self-association (Figure 7K-N). We revised subsection “Impaired ability of N88K Rapsn in AChR clustering” to indicate that our results were in agreement with the previous report in terms of “aggregates” formation.

4) CMS patients with Rapsyn N88K mutation are either homozygous for N88K or heteroallelic for N88K and a second mutation in the Rapsyn coding region. Strikingly, clinical reports suggest that disease severity of patients with Rapsyn compound allelic mutations is greater than that of patients with homozygous N88K Rapsyn mutation. The authors should discuss their findings based on available clinical data on the effect of N88K mutation on the variability and severity of Rapsyn-related CMS.

Good point. This point is now discussed in revised Discussion section.

CMS patients carrying the N88K mutation show variable phenotypes, from serious, early-onset symptoms to mild, late-onset symptoms (Cossins et al., 2006; Milone et al., 2009; Muller et al., 2003), even in patients with homozygous N88K mutation. In severe cases, N88K heteroallelic with c.966 + 1GT, L14P, or a frameshift mutation may cause postnatal death (Maselli et al., 2003; Milone et al., 2009; Richard et al., 2003), indicating N88K mutation could have seriously detrimental effects on NMJ formation or maintenance in human subjects. Overall, the symptoms of CMS patients with homozygous N88K mutation are milder than that of patients carrying heteroallelic N88K mutation with a second mutation. These may be due to that second mutations may exert more harmful effects than that of N88K mutation. Actually, among these second mutations, coding sequence deletions or insertions are often observed, such as c.1177-1178del AA, p.V50-S55del, p.K373del, c.553-554insGTTCT (Milone et al., 2009), which may cause frameshift of coding sequence and totally disrupt Rapsn’s function.

We also note that phenotypes of N88K mt mice seem to be more severe than symptoms of CMS patients. Almost all N88K mt mice died within 24 hours of birth. A simple explanation of these observations may be genetic difference. Gene mutations identified in patients with various diseases such as Lesch-Nyhan syndrome, Lowe syndrome, galactosemia diseases often failed to reproduce clinic symptoms in mouse models (Elsea and Lucas, 2002). On the other hand, a knock-in mutation of AQP2, a gene implicated in hereditary non-X-linked nephrogenic diabetes insipidus, causes more severe phenotypes in mice than in human subjects (Yang et al., 2001a). Even in same species, effects of a mutation could have variable impacts on different strains of mice, for example, the Agrin mutation, Agrin^nm380 (Bogdanik and Burgess, 2011). Furthermore, it is possible that some patients with the N88K mutation are unable to survive and thus be examined by physicians.

5) A well-known function of Rapsyn is to anchor postsynaptic proteins in the subsynaptic cytoskeleton including AChR. Interactions between Rapsyn with microtubule and actin binding proteins have been reported in the literature. It is somehow surprising that the authors did not investigate (immunostaining) whether N88K mutation impairs Rapsyn association with cytoskeleton components.

Good question. To this end, we used two methods – Rapsyn-Actin association

(Walker et al., 1984) and Triton extraction (Moransard et al., 2003). First, as shown in Figure 5B, the amount of Actin and Rapsyn associated with surface AChR was similar between N88K and WT myotubes, suggesting the mutation has little effect on actin association. Second, actin-anchored Rapsyn-AChR complexes are resistant to low concentration of detergents (Moransard et al., 2003). We found that the amount of Rapsyn that could be solubilized by Triton X-100 was similar between N88K and WT (Figure 5—figure supplement 1C and D). These results suggest that the N88K mutation may not impair Rapsyn association with Actin cytoskeleton components. These points are now described in subsection “Reduced E3 ligase activity of N88K mt Rapsn”.

Figure 1: The authors analyzed the NMJ phenotype of mice carrying Rapsyn N88K mutation. For a meaningful comparison of the NMJ phenotypes it is important to compare WT, heterozygous and homozygous mice for the mutation. To support the conclusions of the NMJ staining and EM data, the authors could perform ex vivo electrophysiological recording at P0 to evaluate the neurotransmission defects. This may be challenging since the mutant mice died at birth, but worth to try.

As suggested, we have performed electrophysiological recordings to analyze synaptic transmission in four genotypes (WT, heterozygous N88K/+, homozygous N88K and N88K/-). As shown in Figure 3 and in subsection “Aberrant NMJ formation in N88K mt mice”, there was no difference in resting membrane potentials among the genotypes; however, fibers with mEPP, the amplitude and frequency of mEPPs were reduced in N88K mt and N88K/- mt, compared with WT or N88K/+. These results support the conclusions of morphological studies.

Figure 2: The staining of AChR clusters at E14 is of poor quality. The authors should provide a more suitable confocal image that reflects their quantification (almost no AChR clusters are visible in the mutant).

As suggested, better representative images are provided in revised Figure 2I-L.

Figure 5C: The authors should explain why a 2H treatment of agrin was used to evaluate the effect of N88K mutation in Rapsyn tyrosine phosphorylation in muscle cells.

Agrin activates MuSK within 10 minutes of stimulation and tyrosine phosphorylation of rapsyn that plateaus around 20 minutes (Figure 6A). Agrin induces the formation of AChR clusters beginning around 2 hours of stimulation (Ngo et al., 2012). Therefore, we studied the effect of N88K on Rapsyn tyrosine phosphorylation at 2 hours of agrin stimulation. We revised Materials and methods section to explain 2 hours treatment of Agrin for evaluating tyrosine phosphorylation of Rapsyn (subsection “C2C12 culture and Generation of Gene-modified C2C12 cells”).

Reviewer #3:

[…] Overall, the paper is well done and it reports on an interesting mechanism. Having said this, I find it a bit disturbing that others have reported much milder effects of the N88K mutant (see for example Cossins et al., 2006) in similar cell assays. These contributors also reported that the N88K-mutant rapsyn affects the stability of agrin-induced AChR clusters. Finally, the N88K mutant also has a much milder phenotype in human patients. Hence, a more thorough discussion and assessment what could account for these differences would strengthen the paper. One possibility that could explain the difference in the severity might be that levels of the N88K rapsyn mutant were much lower in the current study than in those studies conducted by others. Hence, a proper dose-response comparison between mutant and wild-type rapsyn may help address this difference.

We thank Dr. Ruegg for his constructive comments that have significantly improved the manuscript.

Regarding “much milder effects of the N88K mutant (see for example Cossins et al., 2006) in similar cell assays” – These points are now discussed in the revised Discussion section –. Pathological mechanisms of the N88K mutation have been examined previously. It was shown to reduce Rapsn’s ability to induce AChR clusters in HEK and TE671 cells (Cossins et al., 2006; Ohno et al., 2002). In agreement, the N88K mutant was shown here to reduce Rapsn’s ability to induce AChR clusters in HEK293T cells (Figure 4A). In N88K knock-in mt mice and in CRISPR/Cas9-generated C2C12 myotubes, AChR clusters were almost diminished.

However, N88K seemed to be able to induce AChR clusters in Rapsn mt myotubes (Cossins et al., 2006). The cause of this discrepancy is unclear. In the current study, the impact of the N88K mutation was inferred mainly by loss-of-function approaches, both in vitro and in vivo, without changing endogenous Rapsn level (Figure 1 and Figure 4E and F). Rapsn is known to induce AChR clusters at low concentrations, but inhibit AChR clusters at high level (Han et al., 1999; Yoshihara and Hall, 1993). This inverted U-function could complicate the interpretation of results of overexpressing WT or mt Rapsn. These points are now discussed in the revised Discussion section.

Regarding “much milder phenotypes in human patients” –These points are now discussed in the revised Discussion section –. CMS patients carrying the N88K mutation show variable phenotypes, from serious, early-onset symptoms to mild, late-onset symptoms (Cossins et al., 2006; Milone et al., 2009; Muller et al., 2003), even in patients with homozygous N88K mutation. In severe cases, N88K heteroallelic with c.966 + 1GT, L14P, or a frameshift mutation may cause postnatal death (Maselli et al., 2003; Milone et al., 2009; Richard et al., 2003), indicating N88K mutation could have seriously detrimental effects on NMJ formation or maintenance in human subjects. Overall, the symptoms of CMS patients with homozygous N88K mutation are milder than that of patients carrying heteroallelic N88K mutation with a second mutation. These may be due to that second mutations may exert more harmful effects than that of N88K mutation. Actually, among these second mutations, coding sequence deletion or insertion are often observed, such as

c.1177-1178del AA, p.V50-S55del, p.K373del, c.553-554insGTTCT (Milone et al., 2009), which may cause frameshift of coding sequence and totally disrupt Rapsn’s function. We also note that phenotypes of N88K mt mice seem to be more severe than symptoms of CMS patients. Almost all N88K mt mice died within 24 hours of birth. A simple explanation of these observations may be genetic difference. Gene mutations identified in patients with various diseases such as Lesch-Nyhan syndrome, Lowe syndrome, galactosemia diseases often failed to reproduce clinic symptoms in mouse models (Elsea and Lucas, 2002). On the other hand, a knock-in mutation of AQP2, a gene implicated in hereditary non-X-linked nephrogenic diabetes insipidus, causes more severe phenotypes in mice than in human subjects (Yang et al., 2001a). Even in same species, effects of a mutation could have variable impacts on different strains of mice, for example, the Agrin mutation, Agrin^nm380 (Bogdanik and Burgess, 2011). Furthermore, it is possible that some patients with the N88K mutation are unable to survive and thus be examined by physicians.