Congenital myasthenic syndromes (CMS) are heterogeneous disorders in which the safety margin of neuromuscular transmission is compromised by one or more specific mechanisms. Using Sanger and exome sequencing in a CMS patient, we identified two heteroallelic mutations, p.Glu1233Lys and p.Arg1277His, in LRP4 coding for the postsynaptic low-density lipoprotein receptor-related protein 4. LRP4, expressed on the surface of the postsynaptic membrane of the neuromuscular junction, is a receptor for neurally secreted agrin, and LRP4 bound by agrin activates MuSK. Activated MuSK in concert with Dok-7 stimulates rapsyn to concentrate and anchor AChR on the postsynaptic membrane and interacts with other proteins implicated in the assembly and maintenance of the neuromuscular junction. LRP4 also functions as an inhibitor of Wnt/beta-catenin signaling. The identified mutations in LRP4 are located at the edge of its 3rd beta-propeller domain and decrease binding affinity of LRP4 for both MuSK and agrin. Mutations in the LRP4 3rd beta-propeller domain were previously reported to impair Wnt signaling and cause bone diseases including Cenani–Lenz syndactyly syndrome and sclerosteosis-2. By analyzing naturally occurring and artificially introduced mutations in the LRP4 3rd beta-propeller domain, we show that the edge of the domain regulates the MuSK signaling whereas its central cavity governs Wnt signaling. We conclude that LRP4 is a new CMS disease gene and that the 3rd beta propeller domain of LRP4 mediates the two signaling pathways in a position-specific manner.
Congenital myasthenic syndromes (CMS) are diverse disorders in which the safety margin of neuromuscular transmission is compromised by deficiency or abnormal function of an endplate (EP)-associated protein. To date, mutations in no fewer than 17 genes coding for proteins expressed at EP are known to cause CMS (1): ALG2 (MIM 607905), ALG14 (MIM 612866), CHRNA1 (MIM 100690), CHRNB1 (MIM 100710), CHRND (MIM 100720), CHRNE (MIM 100725), AGRN (MIM 103320), CHAT (MIM 118491), DPAGT1 (MIM 191350), GFPT1 (MIM 138292), LAMB2 (MIM 150325), PLEC (MIM 601282), MUSK (MIM 601296), RAPSN (MIM 601592), COLQ (MIM 603033), SCN4A (MIM 603967) and DOK7 (MIM 610285). We here describe our findings in a novel CMS caused by mutations in the low-density lipoprotein receptor-related protein 4 encoded by LRP4.
EP development is triggered by the binding of agrin released from the nerve terminal to LRP4, a transmembrane protein of the postsynaptic membrane (2,3). LRP4 bound to agrin forms a ternary complex with the postsynaptic transmembrane muscle-specific receptor tyrosine kinase (MuSK). In this complex, the 3rd β-propeller domain of LRP4 is important for association with MuSK (4), although the interacting conformations remain unresolved. Binding of LRP4 to MuSK triggers phosphorylation and activation of the MuSK intracellular kinase domain. Activated MuSK in concert with Dok-7 stimulates rapsyn to concentrate and anchor AChR in the postsynaptic membrane and to interact with other proteins implicated in the assembly and maintenance of the NMJ (5). LRP4 was recently reported to provide a retrograde signal for presynaptic differentiation at neuromuscular junction (6,7). In addition, autoantibodies directed against LRP4 were recently recognized to cause a form of autoimmune myasthenia gravis (8–10).
In addition to its specific role at the NMJ, LRP4 is also a well-characterized inhibitor of the Wnt-signaling pathway. LRP4 signaling is involved in skeleton formation and kidney development. Mutations in LRP4 have been reported in Cenani–Lenz syndactyly syndrome (CLSS) (11), sclerosteosis-2 (12), and low bone mineral density in human (13) and mice (14). Similarly, Lrp4 mutations cause mule foot disease in cow (15) and kidney and limb defects in mouse (16). In addition, a missense SNP rs2306029 in LRP4 is associated with 4.17-fold increase in the risk of developing Richter syndrome (17). To date, no report has implicated LRP4 as a CMS disease gene.
Using Sanger and exome-capture resequencing, we identified two heteroallelic missense variants in LRP4, p.Glu1233Lys (c.3697G > A) and p.Arg1277His (c.3830G > A). Both variants are located at the edge of the 3rd β-propeller domain of LRP4. We show that each variant impairs binding ability of LRP4 for both agrin and MuSK as well as subsequent agrin-mediated phosphorylation and activation of MuSK, but neither mutation affects the Wnt-signaling pathway. Finally, by analysis of mutations in other diseases and by examining effects of artificially engineered mutations into the 3rd LRP4 β-propeller domain, we show the edge of this domain mediates the MuSK signaling, whereas its central cavity mediates the Wnt signaling.
The patient was born after 42 weeks of gestation with fetal distress and with Apgar scores of 3 and 6 at 1 and 5 min, respectively. Immediately after birth, she had a respiratory arrest and required hospitalization for feeding and respiratory support until the age of 6 months. She started to walk at 18 months but could only walk short distances. During childhood, she fatigued abnormally, could not climb step and was partially wheelchair dependent. Examination at the Mayo Clinic at ages 9 and 14 years revealed mild eyelid ptosis, slight limitation of lateral eye movements, moderately severe proximal greater than distal muscle weakness and hypoactive tendon reflexes. Repetitive nerve stimulation of the spinal accessory nerve showed a 13–16% decrement of the fourth compared with the first compound muscle action potential evoked from the trapezius, biceps and rectus femoris muscles with no significant decrement in the anterior tibial muscle. The decremental responses were transiently improved by edrophonium chloride, a fast acting cholinergic agonist. However, therapy for a few days with pyridostigmine at age 12 markedly worsened the patient's weakness. There was no history of similarly affected family members.
An intercostal muscle specimen was obtained from the patient at age 17 years. Routine histologic examination revealed type 1 fiber preponderance. Synaptic contacts, examined in face-on views of glutaraldehyde-fixed AChE-reacted teased muscle fibers revealed multiple irregularly arrayed synaptic contacts that varied in shape and size (Fig. 1). None of the EPs had a normal pretzel shape. Electron microscopy examination of 54 EP regions of 29 EPs showed that the structural integrity of the nerve terminals and postsynaptic regions was preserved but quantitative analysis revealed that the size of the nerve terminals was reduced to 60%, and that of the postsynaptic region to 48%, of the corresponding control value (Table 1).
|Nerve terminal area (μm2)||2.45 ± 0.30 (54)||3.88 ± 0.39 (63)||<0.005|
|Postsynaptic area (μm2)||5.51 ± 0.45 (54)||10.6 ± 0.79 (59)||<0.001|
|[125I]α-bgt-binding sites/EP||8.7 × 106||12.82 ± 0.79 × 106 (13) Range: 9.3–18.7|
|Nerve terminal area (μm2)||2.45 ± 0.30 (54)||3.88 ± 0.39 (63)||<0.005|
|Postsynaptic area (μm2)||5.51 ± 0.45 (54)||10.6 ± 0.79 (59)||<0.001|
|[125I]α-bgt-binding sites/EP||8.7 × 106||12.82 ± 0.79 × 106 (13) Range: 9.3–18.7|
Values represent mean ± SE. More than one region can occur at an EP. Values in parenthesis represent number of EP regions except for [125I]α-bgt-binding sites where they indicate number of subjects. P-values are based on t-test.
Immunofluorescence microscopy examination of frozen sections revealed normal expression of AChR and AChE at patient EPs. Ultrastructural localization of AChR at the EPs with peroxidase-labeled α-bungarotoxin (bgt) demonstrated normal distribution and density of AChR on the junctional folds. However, the total number of AChRs per EP fell slightly below the range of control values (Table 1).
The MEPP amplitude and the number of quanta released by nerve impulse fell in the normal range, and patch-clamp recordings from10 EPs revealed normal kinetics of the AChR channel (data not shown).
Sanger and exome-capture resequencing analysis
We analyzed the patient DNA using the exome-capture resequencing analysis. The number of SOLiD tags was 90.7 × 106 spanning 4.53 Gb, and 72.8 × 106 tags (80.2%) spanning 3.47 Gb (76.6%) were mapped to the human genome hg19/GRCh37. As the SureSelect probes span 38 Mb, the mean coverage became 91.4. Among the 38-Mb SureSelect probe regions, 3.9% of nucleotides were not covered by any tags. Search for SNVs and indels using Avadis NGS with default parameters detected 46 555 SNVs/indels. We eliminated SNVs/indels registered in dbSNP137 and those with unreliable calls (Avadis decibel score ≤200), and obtained 4074 SNVs/indels. Restriction of our analysis to non-synonymous, frameshifting and splicing SNVs/indels in 33 candidate genes that are essential for the neuromuscular signal transmission yielded three SNVs. Among the three SNVs, a heterozygous c.1148C > G SNV predicting p.Ser376Arg in SNTB2 encoding syntrophin β2 was observed in a patient with periodic paralysis among our cohort of 31 patients other than CMS. In addition, 13 other missense SNPs and two frameshifting SNPs are registered in 539 codons encoded by SNTB2 in dbSNP137. Thus, p.Ser376Arg in SNTB2 was unlikely to be pathogenic. The two other SNVs were heterozygous c.3697G > A (chr11: 46 897 357) and c.3830G > A (chr11: 46 897 102) in LRP4, which predicted p.Glu1233Lys (EK mutation) and p.Arg1277His (RH mutation), respectively. Sanger sequencing of the patient's genomic DNA and mRNA confirmed both mutations and sequencing of cloned mRNA indicated that the mutations were heteroallelic. The father was heterozygous for p.Glu1233Lys. A half brother carried no mutation. No DNA was available from the mother. According to the PolyPhen-2 (18), SIFT (19) and Mutation Taster (20) algorithms, the predicted consequences of the EK mutation were ‘benign’, ‘tolerated’ and ‘disease causing with P > 0.99999’, respectively. Those of the RH mutation were ‘probably damaging’, ‘affect protein function’ and ‘disease causing with P > 0.99999’, respectively. LRP4 is a transmembrane protein with large extracellular domains (Fig. 2A). These mutations were in the 3rd low-density lipoprotein receptor (LDLR) type B repeat, known as β-propeller-like structure. The EK and RH mutations are downstream of the 4th and 5th YWTD motifs, respectively (Fig. 2B). The YWTD motifs are predicted to form the second β sheet below the surface β sheet of each blade of the 3rd β-propeller domain, and the mutations are at the linker between the surface β sheet and the second β sheet.
The EK and RH mutations in LRP4 impair the MuSK signaling pathway
During the NMJ formation, binding of agrin to LRP4 induces phosphorylation and activation of MuSK, which activates ATF2 downstream of JNK and induces clustering of AChR. To study effects of the mutations in this signaling pathway, we used a JNK-responsive ATF2-luciferase (ATF2-luc) reporter (21), which specifically monitors MuSK-dependent stimulation in transfected HEK293 cells. Overexpression of LRP4 and MuSK activated the ATF2-luc reporter in the absence of agrin, as previously reported (22). The EK mutant minimally and the RH mutant moderately impaired LRP4-induced activation of ATF2-luc (Fig. 3A). Addition of agrin to the medium markedly enhanced the ATF2-luc activity for the wild-type LRP4, whereas both mutants completely abolished responsiveness to agrin (Fig. 3A).
We also examined effects of the mutations on MuSK phosphorylation that occurs immediately after formation of the agrin/LRP4/MuSK complex. Consistent with its effects on the signaling activity, agrin enhanced MuSK phosphorylation in the presence of wild-type LRP4 but not in the presence of mutant LRP4 in HEK293 cells (Fig. 3B) and in Lrp4-downregulated C2C12 myoblasts (Fig. 4B). Similarly, wild-type LRP4, but not EK and RH mutants, rescued AChR clustering in Lrp4-downregulated C2C12 myotubes (Fig. 4C). These results support the notion that the EK and RH mutations compromise agrin-mediated activation of MuSK and AChR clustering.
LRP4 has also been known as an extracellular antagonist for Wnt signaling. Wnt signaling is involved in tissue development including limb, bone and kidney. Indeed, previously reported mutations of LRP4 in human, mouse and cow exhibit structural abnormalities in limb, bone and/or kidney. We thus examined the effects of our LRP4 mutations on Wnt-signaling pathway using the TOPFLASH reporter. Wild-type LRP4 suppressed the Wnt3a-mediated TOPFLASH activity, and the EK and RH mutations retained similar suppressive effects (Fig. 3C). Lack of limb, bone and kidney symptoms in our patient can be attributed to EK and RH only affecting agrin-induced activation of MuSK but having no effect on Wnt signaling.
LRP4 directly binds to agrin and MuSK through its extracellular domain. To test the effects of the LRP4 mutations on binding to MuSK and agrin, we performed cell surface-binding assays. We first confirmed that the EK and RH mutants had no effect on LRP4 expressions in cell body and cell membrane in COS7 cells (Supplementary Material, Fig. S1). MuSKect-mycAP (Fig. 5A) and agrin-mycAP (Fig. 5B) bound efficiently to wild-type LRP4 expressed on the surface of COS7 cells. The RH and EK mutants compromised binding of both MuSKect-mycAP and agrin-mycAP (Fig. 5A and B). We also analyzed direct binding of LRP4ecd-Flag to purified MuSKect-mycAP (Fig. 5C) and agrin-mycAP (Fig. 5D) by in vitro plate-binding assays (Fig. 5E and F). We found that the mutations decreased binding affinities of LRP4 for agrin-mycAP (Fig. 5E) and MuSKect-mycAP (Fig. 5F). The RH mutant compromised binding of agrin and MuSK more than the EK mutant in the cell surface-binding assay but not in the in vitro plate-binding assay, which may be accounted for by some other molecules expressed on the cell surface, beneath the cell membrane or secreted from the cells.
Mutations in LRP4 causing sclerosteosis-2 have No effect on MuSK signaling
In contrast to our EK and RH mutations in the 3rd β-propeller domain, p.Arg1170Trp (abbreviated as RW) and p.Trp1186Ser (WS) mutations identified in a patient with sclerosteosis-2 impair Wnt-suppressing activity of LRP4 (12). We first confirmed that wild-type LRP4 and the two mutants are similarly expressed on the plasma membrane in HEK293 cells (Fig. 6A) as we observed in COS7 cells (Supplementary Material, Fig. S1). We then analyzed the effects of the two mutations on MuSK (Fig. 6B) and Wnt (Fig. 5C) signaling. Consistent with the previous report (12), the RW and WS mutations abrogated the Wnt-suppressing activity of LRP4 (Fig. 6C) but had no effect on agrin-induced MuSK signaling (Fig. 6B). Cell surface-binding assay also confirmed that the two mutants retained their ability to bind to agrin and MuSK (Fig. 6D).
As the four mutations affecting MuSK or Wnt signaling were all in the 3rd β-propeller domain, we scrutinized the positions of mutations by homology modeling of the 3rd β-propeller domain of human LRP4 using the 1st β-propeller domain of human LRP6 (PDB ID: 3SOV). The 3rd β-propeller domain contains six blade-like structures and displays a YWTD motif at the second β sheet below the surface of each blade (Fig. 7A and Supplementary Material, Movie S1). In this model, the EK and RH mutations, which only affect agrin/LRP4/MuSK signaling, are located on edge of the 5th and 6th blades, respectively (Fig. 7B and Supplementary Material, Movie S1). In contrast, the RW and WS mutations, which only affect Wnt signaling, are located in a central cavity of the propeller (Fig. 7C and Supplementary Material, Movie S1).
Artificially engineered Lrp4 mutations at the edge of the 3rd β-propeller domain affect MuSK signaling and those in the central cavity affect Wnt signaling
To further confirm that the edge of the 3rd β-propeller domain mediates MuSK signaling and that the central cavity mediates Wnt signaling, we introduced four other artificial mutations based on structural modeling of the 3rd β-propeller domain (Fig. 7A and Supplementary Material, Movie S1). As RH and EK mutations affected two exposed amino acids on the 2nd β sheet, we introduced an alanine into two neighboring amino acids to make IA and VA mutations (Fig. 7B and Supplementary Material, Movie S1). Similarly, as WS and RW mutations were facing the central cavity of the 3rd β-propeller domain, we introduced an alanine into the corresponding amino acids in the other blades to make YA and NA mutations (Fig. 7C and Supplementary Material, Movie S1). We then examined the effects of each mutation on MuSK and Wnt signaling. As expected, the VA and IA mutations at the edge of the 3rd β-propeller domain affected MuSK signaling (Fig. 8A), but not Wnt signaling (Fig. 8B). In contrast, the NA and YA mutations in the central cavity normally activated MuSK signaling (Fig. 8A), but lost Wnt-suppressive activity (Fig. 8B). Similarly, the cell surface-binding assay showed that the VA and IA mutations reduced binding of agrin and MuSK (Fig. 8C and D). Thus, the artificial mutations further underscore the differential signaling roles of the edge and the central cavity of the 3rd β-propeller domain.
LRP4 mutations cause a novel CMS
Because mutations in AGRN (25) and MUSK (26) have been known to cause CMS and because LRP4 was shown to be a coreceptor for agrin that mediates activation of MuSK (Fig. 7D) (2,3). LRP4 has been a candidate gene for a CMS for a number of years. In this communication, we show that mutations in LRP4 cause a CMS and that the identified mutations affect MuSK signaling by compromising binding of LRP4 to both agrin and MuSK.
Our data predict that the identified LRP4 mutations interfere with agrin-LRP4-MusK signaling and thereby hinder concentration of AChR on the junctional folds as well as normal development and maintenance of the entire junction. Our EP studies show that the synaptic contacts are dysplastic (Fig. 1) and that the nerve terminal and postsynaptic areas at the EP regions are hypoplastic (Table 1), but the AChR content of EPs and the synaptic response to ACh are not appreciably reduced. Because the patient has a myasthenic disorders by clinical and EMG criteria, we attribute sparing of the intercostal muscles to different expressivity of the genetic defect in different muscles. Sparing of selected muscles can occur in both autoimmune and congenital myasthenias and is indicated by absence of muscle weakness or a decremental EMG response from some muscles in either disorder. For example, in MuSK antibody-positive myasthenia, EPs in intercostal and biceps brachii muscles have a normal AChR content, generate normal MEPP amplitudes, and have well preserved junctional folds (27,28). Analysis of Musk and Lrp4 expressions in mouse muscles by quantitative RT–PCR revealed that expressions of Musk in omohyoid and trapezius muscles were less than those in the other muscles (Supplementary Material, Fig. S2). The least expression of Musk in mouse omohyoid has been recently reported (29). In contrast, Musk expression in intercostal muscle was ∼three times more compared with omohyoid. Although human specimens of various muscles were not available for our studies, high and low MuSK expressions in intercostal and trapezius muscles may partly account for spared and impaired NMJ signal transmissions in intercostal and trapezius muscles, respectively, in our patient with LRP4 mutations.
Position-specific disease phenotypes of LRP4 mutations
The extracellular domain of LRP4 is known to bind to several proteins: agrin (2,3), MuSK (4), Wnt ligands (30), dkk1 (31), a Wnt inhibitor (32,33), sclerostin (31), another Wnt inhibitor (32,33) and possibly apoE (34). Specific binding domains of LRP4 have been dissected: agrin binds to the LDLa repeats 6–8, EGF-like domains, and the 1st β-propeller domain (Fig. 7B and Supplementary Material, Movie S1) (4,23); MuSK binds to the 4th/5th LDLa repeats and the 3rd β-propeller domain (4); sclerostin binds to the 3rd β-propeller domain (12); and apoE binds to LDLa (34). As for Wnt ligands, the precise molecular mechanisms how LRP4 suppresses Wnt signaling remain elusive, although the Wnt-suppressive effect of LRP4 is well established (35).
The diverse array of binding partners enables LRP4 to play an essential role in multiple biological processes including limb and kidney morphogenesis (16,31); bone development through cell fate decision and migration (35); and synaptogenesis (6,34). The multiple phenotypes caused by mutations in the 3rd β-propeller domain prompted us to scrutinize different regions of this domain, and we found that the edge mediates MuSK signaling and the central cavity mediates Wnt signaling. That a single missense mutation in the 3rd β-propeller domain compromises LRP4 binding to MuSK supports a previous observation that MuSK is bound to the 3rd β-propeller domain of LRP4 (Fig. 7B and Supplementary Material, Movie S1) (4). In contrast, reduced binding of our LRP4 mutants to agrin was unexpected because agrin binds to the EGF-like domain, LDLa repeats 6–8 and the 1st β-propeller domain (4,23). The 2nd and 3rd β-propeller domains, however, enhance binding to agrin to ∼170% of the truncated LRP4 lacking these domains (4). Accordingly, mutations in the 3rd β-propeller domain in our patient are likely to compromise the enhancing effect conferred by the 3rd β-propeller domain of LRP4.
MATERIALS AND METHODS
All human studies were in accord with and approved by the Institutional Review Boards of the Mayo Clinic and Nagoya University Graduate School of Medicine. The patient's father gave informed consent for the patient to participate in the study. Venous blood sample was obtained from the patient and his father and genomic DNA was isolated with the QIAamp Blood DNA kit (Qiagen) according to the manufacturer's recommendations.
Neuromuscular junction studies
Intercostal muscle specimens were obtained from the patient and from control subjects without muscle disease undergoing thoracic surgery. Cryosections were used to colocalize the acetylcholine receptor (AChR) and acetylcholine esterase (AChE) as described (36). AChE was also visualized on teased, glutaraldehyde-fixed muscle fibers cytochemically (37). EPs were localized for electron microscopy (38) and quantitatively analyzed (39) by established methods. Peroxidase-labeled α-bgt was used for the ultrastructural localization of AChR (40). The number of AChRs per EP was measured with [125I]α-bgt. The amplitude of the miniature EP potential (MEPP) and EP potential (EPP) amplitudes and estimates of the quantal content of the EPP (m) were measured as previously described (41,42). Single-channel patch-clamp recordings were performed as previously described (43,44).
Exome-capture resequencing analysis
We enriched exonic fragments using the SureSelect human all exon v2 (Agilent Technologies) and sequenced 50 bp of each fragment in a single direction using ABI SOLiD4 system (Life Technologies). The sequencing fragments were mapped to the human genome hg19/GRCh37 using BioScope 1.4 (Life Technologies). SNVs/indels were called by Avadis NGS 1.3.1 (Agilent) using default parameters. We restricted our analysis to 34 candidate genes that were known to be essential for neuromuscular signal transmission: ABL2, ACHE, AGRN, APP, CHAT, CHRNA1, CHRNB1, CHRND, CHRNE, CHRNG, COLQ, CSNK2B, DAG1, DOK7, DTNA, DVL1, ETV5, GFPT1, LAMB2, LRP4, MUSK, NRG1, NTF4, PAK1, PGGT1B, PHLDB2, PLEC, RAPSN, SCN4A, SLC18A3, SNTB2, STNA1 and SYNC.
Primer pairs for each exon and the flanking intron regions of LRP4 were designed with Primer 3. PCR products were directly sequenced with ABI PRISM 3730 DNA Analyzer and BigDye Terminator Cycle Sequencing Kit version 1.1 according to the manufacturer's protocols (Applied Biosystems).
Full-length human LRP4 cDNA (Open Biosystems) was cloned into the EcoRI site of the pcDNA3.1 mammalian expression vector. The extracellular domain (amino acid 1–1722) of human LRP4 cDNA was cloned into the HindIII and XbaI sites upstream of a 3xFlag epitope of a mammalian expression vector p3xFlag-CMV-14 to generate hLRP4ecd-Flag for the plate-binding assay. The mouse Musk cDNA in pExpress-1 was purchased from Open Biosystems and was used for the luciferase assay in HEK293 cells. Human MUSK cDNA with a Flag-tag at the N-terminal end was cloned into the EcoRI and XbaI sites of the p3xFlag-CMV-14 to generate Flag-MuSK, and was used for the co-immunoprecipitation assay. For vector expressing shRNA against Lrp4, double-stranded oligonucleotides (sense, 5′-GATCCCGGAAGTTTCCTGACATAAATTCAAGAGATTTATGTCAGGAAACTTCCTTTTGGAAA-3′ and 5′-AGCTTTTCCAAAAAGGAAGTTTCCTGACATAAATCTCTTGAATTTATGTCAGGAAACTTCCGG-3′) were cloned into a lentiviral vector pLenti-CMV-GFPx2-DEST, which was kindly provided by Dr Eric Campeau at the University of Massachusetts Medical School. The extracellular domain of mouse Musk cDNA fused to a myc-tag and alkaline phosphatase (MuSKect-mycAP) was kindly provided by Dr Lin Mei. To generate rat agrin-mycAP that retains potency to facilitate AChR clustering as a neural agrin, we cloned amino acid 1141–1937 of rat Agrn cDNA (M64780.1) into pAPtag-5 (GenHunter) at the HindIII and SnaBI sites, so that the Igk-originated signal peptide is attached upstream of the insert and the myc-tag/alkaline phosphatase downstream. Mutant LRP4 plasmids carrying p.Arg1170Trp, p.Trp1186Ser, p.Asn1214Ala, p.Glu1233Lys, p.Val1252Ala, p.Tyr1256Ala, p.Arg1277His and p.Ile1287Ala were generated by the QuikChange Site-Directed Mutagenesis kit (Stratagene). Lack of PCR artifacts was verified by sequencing the entire inserts. Super 8x TOPFLASH plasmid (Addgene), ATF2-Luc (21) and phRL-TK Renilla luciferase vector (Promega) were used for the luciferase reporter assay.
HEK293, L, COS and C2C12 cells were cultured in the Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum and transfected with FuGENE 6 transfection reagent (Roche). L cells stably expressing Wnt3a were purchased from ATCC. Conditioned media were prepared by culturing Wnt3a-producing and control L cells for 4 days. The LRP4ecd-Flag, agrin-mycAP and MuSKect-mycAP proteins were produced by transfecting each plasmid into HEK293 cells in serum-free DMEM. Recombinant rat C-terminal agrin (10 ng/ml, R&D systems) was used for agrin treatment except for the binding assays. For AChR clustering assay, C2C12 myoblast were seeded on a plate coated with collagen I (BD Biosciences) and differentiated in DMEM supplemented with 2% horse serum for 5 days. The differentiated myotubes were electroporated with shLrp4 and each LRP4 cDNA construct using the NEPA21 electroporator and the CUY900-13-3-5 electrode for attached cells (NepaGene), and then treated with 2 μg/ml doxycycline for 2 days to induce shRNA expression. Cells were treated with 10 ng/ml agrin to induce AChR clusters for 12 h. Cells were stained with 10 mg/ml Axexa594-conjugated α-bungarotoxin (1:100, Invitrogen) to label AChR and fixed in 2% paraformaldehyde. Fluorescence images were observed under an Olympus XL71 fluorescence microscope and analyzed with MetaMorph software (Molecular Devices). The lengths of AChR clusters and myotubes were defined as the longest axes of Alexa594 signals and GFP signals, respectively, in the transfected cells. AChR clusters with the axis length <4 µm were excluded from the analysis.
HEK293 cells were transfected with ATF2-Luc and phRL-TK along with the MUSK and LRP4 cDNAs. Cells were cultured for 24 h in the presence or absence of agrin in the medium in a 96-well plate. Cells were lysed with the passive lysis buffer (Promega) and assayed for the luciferase activity using the Dual luciferase system (Promega). Each experiment was done in triplicate.
Biotinylation of LRP4 on plasma membrane and western blotting
HEK293 cells transfected with MuSK and LRP4 plasmids in the presence of agrin were cultured for 24 h. C2C12 myoblasts transfected with shLrp4 and human LRP4 cDNA using NEPA21 electroporator and electroporation cuvettes (NepaGene) were cultured in a differentiation medium for 2 days in the presence of 2 µg/ml doxycycline and 10 ng/ml agrin. Efficient downregulation of Lrp4 was confirmed by quantitative RT–PCR. Cells were lysed with a buffer containing 50 mm HEPES pH 7.0, 150 mm NaCl, 10% glycerol, 1% Triton X-100, 1.5 mm MgCl2, 1 mm EGTA, 100 mm NaF, 10 mm sodium pyrophosphate, 1 µg/µl aprotinin, 1 µg/µl leupeptin, 1 µg/µl pepstatin A, 1 mm PMSF, 1 mm sodium orthovanadate. Cell lysates were subjected to coimmunoprecipitation using 1 µg of anti-phosphotyrosine antibody (4G10, Upstate) attached to protein G Sepharose beads (GE Healthcare). For biotinylation of LRP4 on plasma membrane, LRP4-transfected HEK293 cells were washed twice with PBS containing 1 mm MgCl2 and 0.1 mm CaCl2 (PBS/CM), followed by incubation with 0.5 mg/ml sulfo-NHS-SS-biotin (Pierce) in PBS/CM at room temperature for 30 min. The cells were then washed once with PBS/CM and incubated with 10 mm monoethanolamine for quenching free biotin. The cells were harvested with RIPA buffer (Pierce) after several washing with ice-cold PBS and the cell lysates were incubated with streptavidin sepharose beads (GE healthcare) to purify cell membrane protein. Total or precipitated proteins were dissolved in 1× laemmli buffer, separated on a 10 or 7.5% SDS–polyacrylamide gel and transferred to a polyvinylidene fluoride membrane (Immobilon-P, Millipore). Membranes were washed in Tris-buffered saline containing 0.05% Tween 20 (TBS-T) and blocked for 1 h at room temperature in TBS-T with 3% bovine serum albumin. The membranes were incubated overnight at 4°C either with the mouse monoclonal anti-Myc 9E10 (dilution 1:500, sc-40, Santa Cruz Biotechnology), anti-Flag M2 (dilution 1:4000, F1804, Sigma-Aldrich), anti-LRP4 (dilution 1:1000, ab85697, Abcam), anti-MuSK (1:500, sc-6009, Santa Cruz Biotechnology) or anti-β-actin (dilution 1:200, sc-47778, Santa Cruz Biotechnology) antibody. The membranes were washed three times for 10 min with TBS-T and incubated with secondary goat anti-mouse IgG antibody conjugated to horseradish peroxidase (HRP) (1:6000, LNA931 V/AG, GE Healthcare) for 1 h at room temperature. The blots were detected with Amersham ECL western blotting detection reagents (GE Healthcare) and quantified with the ImageJ program.
Preparation of agrin-MycAP, MuSKect-MycAP and LRP4ecd-flag proteins
Agrin-mycAP and MuSKect-mycAP in the conditioned media of transfected HEK293 cells were concentrated ∼100-fold using Amicon Ultra-4 filters (Millipore). For the cell surface-binding assay, we used the concentrated conditioned media. For the plate-binding assay, we further purified agrin-mycAP and MuSKect-mycAP using the c-myc-Tagged Protein Mild Purification Kit ver. 2 (MBL). Wild-type and mutant hLRP4ecd-Flag proteins were purified with the Anti-DYKDDDDK-tag Antibody Beads (Wako) from the conditioned medium of the transfected HEK293. Purified MuSKect-mycAP and hLRP4ecd-Flag were detected by anti-myc antibody (9E10, Abcam) or anti-Flag antibody (M2, Sigma-Aldrich), respectively. We also measured concentrations of each protein by SDS–PAGE followed by protein staining with SYBRO Ruby Protein Gel Stain (Molecular Probes) using BSA as a standard.
Cell surface- and plate-binding assays
For the cell surface-binding assay, COS cells were transfected with LRP4 using FuGENE 6 (Roche). Cells were incubated 24 h with concentrated conditioned medium containing either agrin-mycAP or MuSKect-mycAP for 1.5 h at RT. Cells were washed with HABH buffer (0.5 mg/ml bovine serum albumin, 0.1% NaN3 and 20 mm HEPES, pH 7.0, in Hank's balanced salt solution), fixed in 60% acetone for 10 min on ice followed by 4% paraformaldehyde in 20 mm HEPES (pH 7.0) in Hank's balanced salt solution for 10 min on ice. Fixed cells were washed once with 20 mm HEPES (pH 7.0) and 150 mm NaCl, incubated at 65°C for 30 min, washed with 0.1 M Tris–HCl (pH 8.0), washed with water and stained with NBP/BCIP solution (Roche). For plate-biding assay, the Immuno plate (Nunc) was coated with 0.15 µg of purified wild-type or mutant LRP4ecd-Flag at 4°C overnight and then incubated with a blocking buffer (1%BSA in PBS) at RT for 1 h. For the binding assays, 80 µl of serially diluted agrin-mycAP or MuSKect-mycAP were added to wells that were coated with wild-type or mutant LRP4ecd-Flag in a blocking buffer. The reagents were incubated for 2 h at RT, and then washed twice with PBS. Bound AP activity was measured using LabAssay ALP (Wako).
The primary sequence of the 3rd β-propeller domain of hLRP4 (accession number: AAI36669, amino acid 1048–1350) and the coordinates of crystal structure of the 1st β-propeller domain (amino acid 20–326) of human LRP6 (PDB ID: 3SOV) were loaded into the Molecular Operating Environment software (MOE, Chemical Computing Group). The primary structures of each β-propeller domain of hLRP4 and hLRP6 were sequence-aligned and manually corrected by the structure–alignment method. Molecular mechanics were calculated by an MMFF94x force field.
This work was supported by Grants-in-Aid from the MEXT and MHLW of Japan to B.O., M.I., A.M. and K.O.; and by NIH Research Grant NS6277 from the NINDS and by Research Grant from the MDA to A.G.E.
We thank Keiko Itano for expert technical assistance.
Conflict of Interest statement. None declared.