Methods
Patients
Thirty patients were clinically and electrophysiologically diagnosed as CMS in Huashan Hospital during 2009–2016. Among them, 2 patients harboring homozygous AGRN variants were enrolled in this study. Detailed clinical information was collected. Written informed consent was obtained for genetic analysis and publication. This study was approved by the Huashan Hospital (Fudan University) Institutional Review Board.
Molecular studies
Genomic DNA from blood was extracted with High Pure PCR Template Preparation Kit (Roche,Basel, CH) according to the manufacturer’s instructions. For patient 1, 17 genes (
AGRN, ALG14, ALG2, CHAT, CHRNA1, CHRNB1, CHRND, CHRNE, COLQ, DOK7, DPAGT1, GFPT1, LAMB2, MUSK, PLEC, RAPSN, SCN4A) known to cause CMS were enriched using target capture (Baylor Genetic Laboratories, Houston, Texas, USA) and subjected to sequencing on Illumina HiSeq2000 [
4]. For patient 2, a commercial next generation sequencing (NGS) panel was used (PrecisionMD, China, including
AGRN, ALG2, ALG14, CHAT, CHRNA1, CHRNB1, CHRND, CHRNE, COLQ, DOK7, DPAGT1, GFPT1, LAMB2, MUSK, RAPSN and
SCN4A) and subsequent sequencing was conducted on the Illumina MiSeq. References to nucleotides or amino acids are based upon the genomic DNA (NC_000001.11) and cDNA (NM_198576) sequence for
AGRN. SIFT and PolyPhen-2 were used to predict the pathogenicity of novel missense variants.
Functional assays
To explore the effect of two novel variants, they were respectively introduced into cDNA by site-directed mutagenesis using Quickchange kit (Stratagene, USA). HEK 293 cells were transfected with 18 μg pDNA3.1hygro(+) GFP-tagged wild type and mutant agrin. The recombinant protein contains its own signal peptide, which allows agrin to be secreted as a soluble form. pDsRed-monomer-N1 was co-transfected to verify transfection efficiency. Forty eight h following transfection, whole cell lysates and conditioned media of HEK 293 cells transfected with either wild type or mutant agrin were harvested. The level of agrin expression was detected by western blot using mouse anti-GFP antibody (ad6556, Abcam), HRP-conjugated anti-mouse secondary antibody (Dako) and ECL (GM Healthcare). DesRed was used as a marker to verify transfection efficiency and alpha-tubulin in total cell lysates served as a loading control. Densitometry of protein bands at ~250KD in conditioned media, which corresponds to the translated ‘full length’ cDNA transcript with the GFP tag, was analyzed using ImageJ software. We further determined turnover or half-life of full length wild type and mutant agrin. We treated the transfected cells with cycloheximide (20 μg/ml) to block further protein synthesis, then the media were collected at a series of time points and the level of agrin was determined by western blotting as described above.
C2C12 myotubes were also exposed to the same amount of wild type or mutant agrin-containing medium for 16 h, Medium from non-transfected HEK293 cells was set as a control. The cells were incubated with α-Butx-594(Invitrogen, USA). Images (20 random fields at 20× objective) were captured using an Olympus IX71 fluorescence microscope with Simple PCI (Digital Pixel). Size and number of AChR clusters were analyzed using the ImageJ Macro automated counting system. The size cut-off for a cluster is 2.5 μm2.
Statistical analysis
Statistical analysis was performed using GraphPad Prism. For expression of full-length mutant and wild type agrin, statistical comparison was performed by two-way ANOVA with multiple comparisons. For AChR clustering assay, data was analyzed using unpaired Student’s t-test. P value was considered to be significant when p < 0.05.
Discussion
Agrin is a heparan sulfate proteoglycan that occurs in multiple tissues as different isoforms generated by alternative splicing with diverse functions [
6]. Motoneuron-derived agrin is considered to play an indispensable role in the formation and maintenance of NMJ [
7,
8]. The protein binds to laminin via its N-terminal agrin (NtA) domains [
9] and interacts with α-dystroglycan [
10] and low-density lipoprotein receptor-related protein 4 (LRP4) through its C-terminal end of LG domains [
11]. Two specific splice variant amino acid inserts of neural agrin at the C-terminal LG2 and LG3 domain respectively, called A and B in chickens [
12] or y and z in humans [
13], are required for interactions [
14].
To date, 12 cases of CMS due to mutations of
AGRN have been reported [
15‐
19] and the mutations are distributed in LG2, LG3, NtA and follistatin-like domains (Additional file
1: Figure S4). Here we report the identification of two CMS patients carrying novel mutations in SEA and LG2 domains in agrin that further define disease-causing mutations for this disorder.
According to previously reported cases, CMS due to
AGRN mutations may show prominent distal muscle weakness and atrophy [
16]. Our patients shared several common features with five previously reported CMS patients reminiscent of distal myopathies, including marked distal weakness affecting initially the lower and later the upper limbs, sparing of axial and oculobulbar muscles and no beneficial effect to acetylcholinesterase inhibitors (Table
1). However, reduced CMAP at rest and an incremental CMAP following exercise in the previously reported 5 patients [
16], suggesting presynaptic abnormality [
3], was not found in our patients and other reported cases with
AGRN mutations. More neurophysiological studies are needed to clarify this situation.
Table 1
Clinical features of reported AGRN-mutant CMS
2009 [15] | F | Early childhood | +a
| +/+/+/+/ND | + | G1709R | LG2 | – | – | + |
G1709R |
2009 [15] | M | Early childhood | + | +/ND/+/ND/ND | + | G1709R | LG2 | – | + | + |
G1709R |
2012 [17] | F | Early childhood | + | +/+/+/+/+ | + | V1727F | LG2 | + | – | – |
Q353X | FS |
2014[16] | F | 15y | – | −/−/+/+/ND | + | G76S | NtA | – | ND | ND |
chr1del | chr1del |
2014[16] | M | 15y | – | −/−/−/+/ND | + | G76S | NtA | – | ND | ND |
chr1del | chr1del |
2014[16] | M | 2y | + | −/−/+/+/− | + | N105I | NtA | – | – | + |
S455Q | FS |
2014[16] | F | At birth | + | −/+/ND/+/− | + | N105I | NtA | – | – | + |
S455Q | FS |
2014[16] | M | 5y | – | +/ND/+/+/ND | + | G1871R | LG3 | ND | ND | + |
G1871R |
2017 [18] | M | 1.5y | ND | +/ND/+/−/ND | + | G1675S | LG2 | + | ND | + |
G1675S |
2017[19] | M | 21y | + | −/−/+/+/− | + | A1768P | LG2 | +/−b
| ND | + |
A1768P |
2017[19] | F | 7y | ND | −/−/+/+/− | ND | A1768P | LG2 | +/−b
| ND | + |
A1768P |
2017[19] | F | ND | ND | ND | ND | A1768P | LG2 | +/−b
| ND | + |
A1768P |
P1 | M | 16y | + | −/−/+/+/− | + | L1176P | SEA | – | ND | – |
P2 | M | 9y | + | −/−/+/+/− | + | R1698C | LG2 | – | ND | + |
Four mutations in LG2 domain, pG1675S, p.G1709R, p.V1727F and p.A1768P, have been reported in 6 patients [
15,
17‐
19]. Two patients presented with ptosis and general limb weakness, one isolated case manifested with proximal weakness and head drop, and a recently reported family developed proximal and distal weakness. In our study, Patient 2 showed predominant distal weakness and atrophy. All cases showed varying responses to salbutamol (Table
1). Based on previous studies, the LG2 domain has a critical role in the activation of the LRP4-MuSK complex, as neural agrin induces MuSK phosphorylation by interacting with LRP4 via its LG2 domain and then triggers the aggregation of AChR in the postsynaptic membrane [
20]. Heparin, as well as several monoclonal antibodies, could block agrin-induced MuSK activation and AChR aggregation by binding to the LG2 domain [
21‐
23]. In addition, the LG2 domain also participates into the structuring of the basal lamina through its interaction with α-dystroglycan [
24]. Existing functional analysis of p.G1709R and p.V1727F showed different results (Additional file
1: Figure S4). The mutation, p.G1709R, in chicken mini-agrin did not reduce the activation of muscle-specific tyrosine kinase (MuSK) or affect the binding of agrin to α-dystroglycan, indicating that the mutant protein does not interfere with the induction of the postsynaptic apparatus but disturbs the maintenance of the NMJ (Additional file
1: Figure S4). While in the neural form of full length agrin, p.R1698C mutant agrin found in Patient 2, in consistent with the reported p.V1727F mutant protein, exhibited impaired ability to induce AChR clustering [
16]. We also found the p.R1698C protein degraded faster than the wild type, which predicts that the stability of the protein is impaired. The specific molecular mechanism of phenotype difference and pathogenicity of the mutations remains to be explored.
The SEA (Sperm protein, Enterokinase and Agrin) domain named after the first three proteins in which it was identified, located in the middle of agrin, is a poorly characterized protein motif found in extracellular matrix associated glycoproteins. Recombinant agrin protein without an SEA domain could achieve similar potency in AChR clustering as seen in full-length constructs [
20] and rescued
AGRN-knockout mice [
25]. It was not implicated as important for agrin function until the identification of an
AGRN mutant (nmf380-F1061S) mouse model of CMS. NMJs in the homozygous mutant mice progressively degrade postnatally and have decreased acetylcholine receptor density (Additional file
1: Figure S4). Intriguingly, substitution of Leucine by Proline found in Pt 1 (p.L1176P) is an amino acid next to the mutant point in the mouse model, which is located at the C-terminal of an alpha-helix by a structural prediction model extrapolated from the structure of Mucin16 SEA domain [
26]. Both amino acids were conserved not only among species but also SEA domains in other proteins like mucin and enterokinase. In the mouse model, the mutation did not alter the expression but impaired conformation and secretion of the protein [
27]. While as shown in our results, the expression of p.L1176P mutant agrin was reduced both in whole cell lysates and media of transfected HEK293 cells. Thus, we believe both the export of the agrin into the medium and its stability in the medium are affected. However, whether this arose through altered glycosylation resulting in abnormal trafficking to the membrane for secretion or protein misfolding leading to altered stability requires further elucidation.
In conclusion, we identify two cases of AGRN-CMS due to homozygous mutations in the LG2 and SEA domain of agrin. Functional analysis suggested impaired stability of the mutant agrin in the conditioned medium and also probably within cellular environment of the nerve, and this is likely to be the main molecular pathogenic mechanism of the mutations for these two patients.