Introduction

The synaptic cleft of the neuromuscular junction (NMJ) is made of a functionalized extracellular matrix (ECM), facing on the one side the presynaptic motor nerve terminal, and on the other side the muscular postsynaptic compartment [63]. Synaptic ECM cues are essential for the establishment of functional NMJ, being actively involved in regulating synaptic terminal development and maturation [24, 41, 42], in defining the precise juxtaposition of pre- to post-synaptic areas [1, 6, 44], and in guaranteeing a permissive environment for synaptic transmission [4, 35, 48]. The synaptic basal lamina contains agrin (the main player driving acetylcholine receptor aggregation and stabilization of acetylcholine receptor (AChR) aggregates at the postsynaptic membrane), laminin, nidogen, collagen IV, perlecan, and heparan sulfate glycosaminoglycans [56]. Some of these molecules, such as agrin, specific laminin isoforms and collagen IV variants, are only present at the synaptic cleft. Other molecules, such as perlecan and nidogen, display a broader deposition, being found also in the extrasynaptic muscle basement membrane, where they can differentially partner with other proteins [20, 45]. A number of these molecules, such as collagen IV, perlecan and biglycan, are actually interacting molecules of one of the most abundant ECM proteins of skeletal muscle, ColVI [11, 31, 69].

Collagen VI (ColVI) is a protein of the basement membrane of skeletal muscle [31]. It is composed of three main subunits encoded by distinct genes, where the most abundant form is made by the stoichiometric association of the α1(VI), α2(VI) and α3(VI) chains, while other three minor chains can substitute for α3(VI) in specific tissues [11, 23, 52]. Mutations of ColVI genes are causative for a broad group of inherited muscle diseases, including Ullrich congenital muscular dystrophy (UCMD), Bethlem myopathy (BM) and myosclerosis myopathy [8]. Studies in Col6a1 null mice and in BM/UCMD patients showed that ColVI plays a key role in muscle homeostasis and regeneration [7, 27, 67]. Interestingly, our recent work in Col6a1 null mice revealed that ColVI ablation affects peripheral nerve structure and function [13]. However, the potential role of ColVI at the NMJ and whether its deficiency affected neuromuscular transmission remained unknown.

Here we demonstrate that ColVI is found within the synaptic cleft of NMJs and is required for proper maintenance of the postsynaptic apparatus, as well as for preserving in vivo neuromuscular transmission in mice. Furthermore, we show that ColVI is able to modulate synaptic gene transcription, promoting AChR clustering and mimicking agrin effects. These findings highlight a novel role for ColVI at the NMJ, and point at novel and previously unforeseen features of BM/UCMD pathology.

Materials and methods

Animals

Wild-type and Col6a1−/− mice in the C57BL/6N background were used in this study [30]. All studies were carried out in 5- to 6-month-old mice. Animal procedures were approved by the Ethics Committee of the University of Padova and authorized by the Italian Ministry of Health.

Immunofluorescence on tissues

TA muscles were dissected and rapidly frozen in liquid nitrogen. Cross (7 μm) and longitudinal (40 μm) sections were collected on glass slides and stored at − 20 °C until use. Samples were fixed and permeabilized with cold 50% methanol/50% acetone at − 20 °C for 10 min. After drying, samples were washed in phosphate buffered saline (PBS) solution and incubated for 1 h with 10% goat serum (Sigma) in PBS. Alternatively, when mouse-derived primary antibodies were used, samples were incubated for 2.5 h with 4% IgG free bovine serum albumin (BSA; Sigma) in PBS at room temperature and blocked for 30 min with anti-mouse IgG Fab fragment (5 mg/ml; Jackson Immunoresearch). Samples were incubated overnight at 4 °C with the primary antibodies listed in Suppl. Table S1. After washing in PBS, samples were incubated with fluorescent secondary antibodies (Suppl. Table S2). AChR was detected with tetramethylrhodamine-conjugated α-bungarotoxin (1:200; αBT, Invitrogen) and nuclei were stained with Hoechst 33528 (Sigma). Glass slides were mounted in 80% glycerol and analyzed by confocal fluorescence microscopy. Denervated NMJ quantification was performed as described in Supp. Material.

Electron microscopy

EDL muscles were dissected from mice and fixed with 4% paraformaldehyde (PFA) in PBS for 30 min, washed and labeled with αBT. Tissue areas containing NMJs were dissected and embedded in London Resin White (London Resin Company) for immunoelectron microscopy [51] or fixed with 2.5% glutaraldehyde in cacodylate buffer and embedded in Epon812 for ultrastructural analysis. Immunoelectron microscopy analysis of ColVI was performed on London White resin ultrathin sections with a rabbit polyclonal anti-ColVI antibody (1:50; 70-XR95, Fitzgerald) and revealed with a 5-nm colloidal gold conjugated goat anti-rabbit antibody (Sigma). Sections were stained with uranyl acetate and lead citrate and observed with a Jeol JEM-1011 electron microscope operating at 100 kV.

Whole-mount staining on diaphragms

Diaphragms were dissected and cut in order to divide the left half of the tissue from the right one. Samples were permeabilized at 36 °C for 15 min in a solution containing trypsin–EDTA (50 mM; Gibco) and collagenase I (1 mg/ml; Worthington) and fixed with 4% PFA in PBS for 30 min. Then, diaphragms were soaked in a blocking/permeabilizing solution (3% BSA, 10% goat serum, 0.1% Triton X-100) at room temperature for 2 h, washed in PBS and incubated overnight at 4 °C with primary antibodies (see Suppl. Table S1) in a PBS solution containing 3% BSA and 0.5% Triton X-100 in PBS. After PBS washes, samples were incubated with secondary antibodies (see Suppl. Table S2) and αBT (1:1000) in a 3% BSA/PBS solution. Samples were finally washed in PBS and mounted in 80% glycerol before analysis by confocal fluorescence microscopy. Denervated NMJ quantification was performed as described in Supp. Material.

3D imaging and analysis

Soleus and EDL muscles were dissected and fixed with 2% PFA in PBS for 2 h at 4 °C. Muscle bundles containing 5–10 fibers were prepared and stained with αBT (1:2500) for 1 h at room temperature. Stained bundles were washed in PBS and embedded in Mowiol (Sigma). Additional staining with neurofilament and synaptophysin was performed as described in Cheusova et al. [14]. Thin bundles of teased muscles were blocked in 100 mM glycine for 15 min; permeabilized in 0.5% Triton X-100, 5% BSA, and 1% FCS for 1 h; incubated with both, rabbit anti-neurofilament antibody (1:500; Millipore) and rabbit anti-synaptophysin antibody (1:500; DAKO). Secondary antibodies conjugated to Alexa-488 (Molecular Probes) were applied together with αBT. Diaphragms were dissected and fixed with 4% PFA in PBS for in 30 min at 4 °C, washed in Tris-buffered saline (TBS), incubated in blocking solution (4% BSA, 0.1% Triton X-100 in TBS) overnight at 4 °C, and stained with αBT (1:1000) overnight at 4 °C. After rinsing again in TBS, diaphragms were mounted in 80% glycerol. Muscles were analyzed by confocal fluorescence microscopy, and 3D images of NMJs were taken with 40× oil objective (Zeiss Examiner E1) at 55 ms exposure time. Images were deconvoluted and analyzed using different modules in AxioVision software. The following parameters were determined for each NMJ: volume, surface, sum fluorescence, mean fluorescence, number of fragments.

C2C12 culture and differentiation

C2C12 cells were seeded on 12-well plates or glass coverslips and cultured in DMEM (Gibco) supplemented with 20% fetal bovine serum (FBS, Gibco). Myotube differentiation was initiated by switching to DMEM supplemented with 2% horse serum (Gibco). After 7 days myotubes were left untreated or treated for 20 h with agrin (10 ng/ml; R&D), or with native murine ColVI (1 μg/ml) [30] purified by gel filtration chromatography [51] in the absence or presence of anti-ColVI antibody (40 μg/ml; sc-47712, Santa Cruz). After treatments, myotubes were lysed with either Trizol (Life Technologies), protein lysis buffer, or fixed with 4% PFA.

Real-time qRT-PCR in murine muscles and myotubes

For RNA extraction, muscles and myotube cultures were lysed in Trizol reagent (Invitrogen) and processed according to manufacturer instructions. Reverse transcription was performed using M-MLV reverse transcriptase, and cDNA products were analyzed by real-time PCR with the Rotor Gene SYBR PCR kit (Quiagen). The primers used are listed in Suppl. Table S3. Data were normalized to Gapdh expression.

Western blotting

Diaphragms were dissected, stained with αBT and the NMJ-rich band was cut from the rest of the tissue under a fluorescence stereomicroscope. Samples were homogenized with a pestle in a lysis buffer (50 mM Tris HCl, pH 7.5, 150 mM, NaCl, 1 mM EDTA, 10% glycerol) containing 0.5 mM DTT, 2% SDS, 1% Triton X-100, phosphatase inhibitors (Cocktail II, Sigma) and protease inhibitors (complete EDTA free, Roche). Myotubes were lysed in a lysis solution (50 mM Tris–HCl, pH 7.5, 150 mM NaCl, 20 mM EDTA, 0.5% NP-40) supplemented with phosphatase inhibitors (Cocktail II, Sigma) and protease inhibitors (complete EDTA free, Roche). Proteins were quantified with BCA Protein assay kit (Pierce), and separated by SDS-PAGE. Quaternary structure of ColVI used for myotube treatment was assessed running it in a 2.5% acrylamide/0.5% agarose SDS-PAGE in non-reducing conditions [38], and contamination by other ECM proteins was excluded using laminin (L2020, Sigma) and fibronectin (F1141, Sigma) as positive controls. Proteins were then blotted onto PVDF membranes (Millipore) at 20 or 30 V (for non-reduced ColVI) for 2 h. Membranes were incubated overnight at 4 °C with primary antibodies (see Suppl. Table S1). After washing in TBS-T, membranes were incubated for 1 h at room temperature with horseradish peroxidase-conjugated secondary antibodies (see Suppl. Table S2). Detection was performed by chemiluminescence (Thermo-Scientific).

Nerve–muscle preparations for electrophysiology

Isolated diaphragm–phrenic nerve and soleus–sciatic nerve preparations were pinned to the bottom of a Sylgard (Sigma)-coated petri dish and maintained in Liley’s solution gassed with 95% O2–5% CO2 at room temperature [33]. The recording chamber was perfused at a rate of 1 ml/min. Nerve stump was stimulated with suction electrode connected to a stimulator (Grass S88, Grass) through a stimulus isolation unit (SIU5, Grass). Single supramaximal stimulation (0.1 ms duration) was delivered at 0.5 Hz in order to prevent synaptic fatigue. To block muscle action potentials, so that EPP and EPCs could be recorded [47, 49], µ-conotoxin GIIIB (µ-CTX, 2 µM; Alomone Labs) was added to Liley’s solution. In some experiments, the effect of the toxin wore off after 1–2 h and contractions resumed in response to nerve stimulation. These preparations were then exposed for a second time to the toxin. At the same time, AChRs were labeled by adding rhodamine-conjugated αBT (5 nM; Life Technologies) to the bath, as described before [17, 70].

Intracellular recordings in isolated diaphragm–phrenic nerve preparation

Two intracellular electrodes (resistance 10–15 MΩ) were inserted within 50 µm of the NMJs under visual inspection. Current was passed through one electrode to maintain the membrane potential within 2 mV of − 75 mV, while voltage transients were recorded with the other electrode. Signals were amplified by an Axoclamp 900A and digitized at 40 kHz by a Digidata 1440A under the control of pCLAMP 10 (Molecular Devices). Voltage records were filtered at 3 kHz and current recorded at 1 kHz (8-pole Bessel filter). Current transients were recorded using the two-electrode voltage-clamp facility of the Axoclamp 900A. 50–100 spontaneous quantal events were recorded during a period of 1 min. Records were analyzed using pCLAMP 10. Events recorded from each NMJ were averaged and the amplitude, rise time and single exponential decay time constant determined.

Extracellular recordings in isolated diaphragm–phrenic nerve preparation

The preparation was placed on the stage of a Zeiss Axio Examiner Z1 microscope fitted with incident light fluorescence illumination with filters for red (Zeiss filter set 20) fluorescing fluorophore (Carl Zeiss MicroImaging). At the beginning of the experiment, CMAP was recorded using a micropipette with a tip diameter of about 10 µm, filled with bathing solution. The electrode was positioned so that the latency of the major negative peak was minimized. The electrode was then positioned 100 µm above the surface of the muscle and CMAP was recorded. For curare experiments in ex vivo diaphragm preparations, the experimental conditions were established as previously reported [55]. The recording chamber was filled with 2 ml (500 or 800 nM) of d-tubocurarine chloride (Sigma). During curare treatment, trains of 25 repetitive nerve stimulations (5 Hz) were performed at 2 min intervals, and the ratio of CMAP amplitudes (mean (20th − 25th)/2nd) was calculated.

Intracellular recordings in soleus nerve–muscle preparation

Spontaneous and evoked excitatory postsynaptic potentials were intracellularly recorded from single muscle fibers using borosilicate glass microelectrodes (inner diameter 0.86 mm, outer diameter 1.5 mm; 15 MW resistance) (Science Products). Microelectrodes were inserted within 1 mm of the point of entrance of the nerve in the soleus muscle. For each recording, resting membrane potential was measured and set at − 70 mV for the entire duration of the recording. Spontaneous neurotransmitter release (miniature end plate potentials, mEPP) was initially recorded for 120 s. Subsequently, evoked neurotransmitter release (excitatory postsynaptic potentials, EPP) was recorded following nerve stimulation as described above for 30 s. For repetitive nerve stimulation, soleus nerve was stimulated at 10 Hz. Intracellularly recorded signals were amplified using an intracellular amplifier (SEC NPI Electronic) in current-clamp condition. Amplified signals were then sent to an A/D converter (National Instruments) and fed to a personal computer. Digitized recordings were analyzed offline using the WinEDR and Pclamp software for electrophysiology (Strathclyde and Pclamp6, Axon). mEPP amplitudes were computed by analyzing the first 120 s of each recording. EPP amplitudes were computed by analyzing the first 10 evoked responses stimulating the nerve at 0.5 Hz, in order to prevent the onset of synaptic fatigue. For tetanic nerve stimulations at 10 Hz, the ratio between the amplitude of each evoked response and the amplitude of the first one was calculated.

Human samples

All the human muscle biopsies used in these studies were provided by the Neuromuscular Bank of Tissues and DNA samples of the Telethon Network of Genetic BioBanks. Six subjects were selected from a cohort of BM and UCMD patients followed at the Neuromuscular Center of the University of Padova. All patients carried previously characterized mutations on ColVI genes. Relevant data concerning genetic analysis, age at the biopsy, age at the last clinical evaluation, and disease progression are reported in Fig. 5a. Control and patient samples were obtained by open muscle biopsy from vastus lateralis, with the exception of P4, whose biopsy derived from the femoral quadriceps. Four other dystrophic patients were included in the study as pathological controls: two affected by Duchenne muscular dystrophy (DMD1, DMD2) and two by polymyositis (PM1, PM2). Both DMD patients were 1 year old at the age of biopsy, carried characterized dystrophin mutations and both biopsies derived from the femoral quadriceps. PM1 biopsy was collected when the patient was 80 years old, and was derived from femoral quadriceps; PM2 biopsy was collected at the age of 69 years old, and was derived from vastus lateralis. Both PM biopsies were collected before glucocorticoid treatment.

Analysis of gene expression in human samples

Total RNA was isolated from muscle biopsies using Trizol Reagent (Life Technologies). First-strand cDNA synthesis was performed using High-Capacity cDNA Reverse Transcription Kit (Life Technologies) and transcript levels were quantified by SYBR Green Real-Time PCR (Life Technologies) using the ABI PRISM 7000 sequence detection system. The primers used are listed in Suppl. Table S4.

Immunofluorescence on human samples

8-µm-muscle cryosections were collected on Superfrost slides, fixed with 4% PFA, treated with 0.5% Triton X-100, blocked with 10% FBS in PBS for 30 min and then incubated overnight at 4 °C with primary antibodies (see Suppl. Table S1) diluted in blocking solution. After washing, samples were incubated with secondary fluorescent antibodies (see Suppl. Table S2) for 1 h at room temperature together with αBT (1:200). Slides were mounted using Vectashield medium with DAPI stain (Vector) and examined on a Leica TCS SP5 confocal microscope. Muscle specimens were stained with hematoxylin and eosin to check histopathology integrity, and myofibrillar ATPase with preincubation at pH 4.3 to assess muscle fiber type composition. Slides were examined on upright microscope (BX60, Olympus) equipped with a CCD camera (DP70, Olympus). Images were post-processed using ImageJ suite. Type I and type II muscle fibers were counted. On average, 1248 (range 524–4082) myofibers per patient were evaluated.

Statistical analysis

Data are presented as mean ± s.e.m., except where indicated. The statistical significance was determined by unpaired two-tailed Student’s t test or by unpaired one-tailed Mann–Whitney test as indicated. A P value < 0.05 was considered as a significant difference.

Results

ColVI is associated to the neuromuscular synaptic ECM

To investigate the presence of ColVI at the NMJ, we performed immunostaining for ColVI together with fluorophore-coupled α-bungarotoxin (αBT), which specifically binds AChRs. Immunolabeling of both 7-μm-thick cross sections (Fig. 1a) and 40-μm-thick longitudinal sections (Fig. 1b) of mouse tibialis anterior (TA) muscle showed a close localization of ColVI with AChR clusters. Single stack confocal planes helped in analyzing at higher resolution the continuity of ColVI with laminin and AChR clusters (Fig. 1c). It is well known that ColVI is composed of three different α chains, and that the most frequent α3(VI) can be alternatively substituted by α4(VI), α5(VI) or α6(VI) in its association to α1(VI) and α2(VI) chains in specific tissues [11]. Interestingly, α5(VI) was previously suggested to be present at the NMJ [22]. Therefore, we compared α3(VI) and α5(VI) immunostaining on longitudinal muscle sections, detecting only partial colocalization of these two ColVI chains (Suppl. Fig. S1a). While immunolabeling data suggested that both ColVI chains are found in association with NMJs, comparison between the two staining patterns in single stack confocal planes indicated that α3(VI) is closer to the AChR clusters than α5(VI) (Suppl. Fig. S1b).

Fig. 1
figure 1

ColVI is present in the NMJ synaptic cleft. a Immunofluorescence for ColVI (green) in 7-μm-thick cross sections of wild-type and Col6a1−/− TA muscles, showing ColVI colocalization with AChRs in wild-type animals. AChRs were stained with αBT (red), and nuclei were stained with Hoechst (blue). Arrowheads point at some NMJs, and the inset shows higher magnification of one NMJ in wild-type muscle. Scale bar, 100 μm. b Confocal analysis of ColVI immunostaining (green) in 40-μm-thick longitudinal sections of wild-type TA, revealing that ColVI staining encases αBT-labeled AChR clusters (red) and is in close contact with them. Laminin immunostaining (grey) is provided as marker of basal membrane. Scale bar, 40 μm. c Serial stacks, corresponding to single focal planes, from the immunostaining analysis in b are provided, showing at higher resolution the tight contact between ColVI and AChR clusters (red). Scale bar, 40 μm. d Immunogold labeling for ColVI in wild-type EDL, showing the presence of ColVI, identified by gold particles (arrowheads), in the ECM associated with the postsynaptic membrane within the synaptic cleft. Presynaptic compartment is indicated in transparent blue. The right panel shows a higher magnification of the squared area. Scale bar, 300 nm

Full size image

Ultrastructural analysis by immunoelectron microscopy confirmed ColVI deposition in the ECM associated with the synaptic cleft (Fig. 1d), thus identifying ColVI as a novel constituent of the synaptic basal lamina.

Motor endplates are altered in the absence of ColVI in vivo

To explore the in vivo role of ColVI at the NMJ, we performed whole-mount αBT labeling of adult wild-type and Col6a1−/− mouse muscles. Postsynaptic boutons of Col6a1−/− muscles appeared less regular in shape, being frequently fragmented (Fig. 2a). Moreover, parts of presynaptic regions were not covered by αBT-positive clusters in Col6a1−/− muscles, in correspondence with areas of NMJ fragmentation, when stained with neurofilament and synaptophysin (Fig. 2b). Fragmentation was confirmed by morphometric analysis of extensor digitorum longus (EDL), soleus and diaphragm muscles, revealing a marked fragmentation of NMJs in Col6a1−/− muscles when compared to wild-type muscles (Fig. 2c and Suppl. Fig. S2). In soleus and EDL, NMJ fragmentation was accompanied by changes in three-dimensional (3D) morphometric parameters, with an increase of the surface area, volume and fluorescence intensity of NMJs (Suppl. Fig. S3).

Fig. 2
figure 2

ColVI ablation induces AChR clusters fragmentation. a Representative micrographs of wild-type and Col6a1−/− EDL and soleus endplates, obtained by whole-mount staining with αBT (red). Arrowheads indicate the presence of AChR cluster fragmentation. b Representative images of wild-type and Col6a1−/− EDL and soleus endplates, stained for AChR clusters with αBT (red) and immunostained for neurofilament and synaptophysin (green), showing regions of nerve terminals not completely covered by subsynaptic AChR aggregates (arrowheads), corresponding to areas of cluster fragmentation. Images in a and b are deconvoluted Z-stack micrographs presented as 2D. c Morphometric analysis of whole-mount αBT staining of EDL (left panel) and soleus (right panel) from wild-type and Col6a1−/− mice. Muscles lacking ColVI display a significantly higher fragmentation of NMJs, when compared to wild-type muscles (***P < 0.001; unpaired two-tailed Student’s t test; n = 115 NMJs, wild-type EDL; n = 106 NMJs, Col6a1−/− EDL; n = 69 NMJs, wild-type soleus; n = 84 NMJs, Col6a1−/− soleus). NMJs were sampled from three mice per genotype. Error bars indicate s.e.m

Full size image

Endplate fragmentation is often due to aging or muscle denervation [3, 50]. Although we did not detect denervation defects in the absence of ColVI (Suppl. Fig. S4a,b), Col6a1−/− muscles displayed a remarkable upregulation of AChRγ and AChRα, but not AChRε, mRNA transcripts (Fig. 3a), thus confirming NMJ alterations and pointing at an abnormal neuromuscular transmission [71]. On the other hand, the presynaptic compartment did not display any overt alteration in Col6a1−/− muscles (Suppl. Fig. S4c). Interestingly, some ColVI interactors, such as perlecan, collagen IV and biglycan [11], were reported to be critical in stabilizing the postsynaptic apparatus [2, 4, 21]. In order to evaluate the impact of ColVI ablation on the extracellular and cell surface NMJ protein network, we performed immunofluorescence for a panel of NMJ-enriched proteins (Fig. 2d). This analysis revealed that perlecan, known to engage the dystrophin–glycoprotein complex (DGC) [4, 46], is abnormally increased in the NMJs of Col6a1−/− muscles, and a similar increased labeling was displayed by other DGC components, such as utrophin, dystrophin and NOS1, but not δ-sarcoglycan (Fig. 3b and Suppl. Fig. S5a). Collagen type IV, an ECM protein involved in NMJ maturation [21], was also upregulated in the NMJs of Col6a1−/− muscles (Fig. 3b). The MuSK/LRP4 heterodimer acts as an agrin receptor and was shown to be contacted by biglycan, thus playing a key role on synapse stability [2]. Although MuSK immunolabeling appeared similar between wild-type and Col6a1−/− NMJs, LRP4 displayed increased labeling in Col6a1−/− NMJs (Fig. 3b and Suppl. Fig. S5a). In agreement with the increased immunolabeling of NMJ-associated proteins in ColVI-deficient muscles (semi-quantitative analysis is provided in Fig. 3c), real-time PCR analysis of Col6a1−/− diaphragms showed altered expression of postsynaptic genes, with significant upregulation of MuSK, LRP4 and utrophin mRNA transcripts (Fig. 3d). In addition, MuSK, NOS1 and AChRγ protein levels resulted upregulated in Col6a1−/− isolated synaptic diaphragms, as revealed by semi-quantitative western blot analysis (Suppl. Fig. S5b, c).

Fig. 3
figure 3

Absence of ColVI causes NMJ defects and affects the postsynaptic compartment. a qRT-PCR analysis for the Chrng, Chrna and Chrne genes in diaphragm (top panel) and TA (lower panel) of wild-type and Col6a1−/− mice, showing a significant increase of Chrng and Chrna transcripts in Col6a1−/− muscles (***P < 0.001; **P < 0.01; unpaired two-tailed Student’s t test; diaphragm, n = 6 mice, each genotype; TA, n = 8 mice, each genotype). Error bars indicate s.e.m. b Representative immunolabeling for NMJ-associated proteins (green) in cross sections of wild-type and Col6a1−/− TA muscles stained with αBT staining (red). The localization at the NMJ of collagen IV, perlecan, utrophin, NOS1 and LRP4 was analyzed in at least three animals per genotype. Scale bar, 20 μm. c Quantification of immunofluorescence intensity from assays in b, showing altered expression of the proteins at the Col6a1−/− NMJ. a.u.: arbitrary units (*P < 0.05; unpaired one-tailed Student’s t test; n = 3 mice, each genotype; a minimum of 20 NMJs per genotype was analyzed). Error bars indicate s.e.m. d qRT-PCR analysis for postsynaptic genes in diaphragms (DIA) and TA muscles of wild-type and Col6a1−/− mice (***P < 0.001; *P < 0.05; unpaired two-tailed Student’s t test; diaphragm, n = 6 mice, each genotype; TA, n = 8 mice, each genotype). Error bars indicate s.e.m

Full size image

Impaired NMJ function in the absence of ColVI

Previous studies demonstrated decreased muscle strength in Col6a1−/− mice, with lower locomotor activity during running wheel or treadmill exercise [28] and normal activity on rotarod [13]. However, NMJ-related functional aspects were never evaluated in ColVI-deficient mice. When measuring the latency to fall from an upside-down rotated horizontal grid, we found that Col6a1−/− mice hung for significantly shorter times than wild-type animals (Fig. 4a). Ex vivo electrophysiological recordings showed that both nerve-evoked endplate potentials (EPP) and nerve-independent miniature endplate potentials (mEPP) had significantly reduced amplitudes in soleus (a muscle with predominance of slow twitch fibers) [31] and diaphragm (a muscle with predominance of mixed fast twitch fibers) [60] of Col6a1−/− mice when compared to wild-type animals (Fig. 4b, c). The reduced EPP and mEPP amplitudes could be due either to decreased quantal size or to a reduced postsynaptic sensitivity to individual quanta. Notably, both the above-mentioned neuromuscular transmission defects were observed, as the mean quanta were significantly reduced in Col6a1−/− muscles, together with decreased endplate currents (Suppl. Table S5) and lower input resistance (Fig. 4c).

Fig. 4
figure 4

ColVI ablation alters neuromuscular synapse function and affects safety factor. a The time for which wild-type and Col6a1−/− mice clung upside down to a horizontal grid was measured, showing impaired performance in Col6a1−/− animals (***P < 0.001; unpaired two-tailed Student’s t test; n = 10 mice, each genotype). Error bars are s.e.m. b, c Measurement of the EPP and mEPP amplitudes in wild-type and Col6a1−/− soleus (b) (n = 126 EPP, wild-type, n = 238 EPP, Col6a1−/−; n = 362 mEPP, wild-type; n = 416 mEPP, Col6a1−/−) and diaphragm (c) (n = 58 EPP, wild-type; n = 63 EPP, Col6a1−/−; n = 78 mEPP, wild-type; n = 104 mEPP, Col6a1−/−), revealing alterations of the electrophysiological features in Col6a1−/− muscles. The input resistance (c, left panel) (n = 76, wild-type; n = 82, Col6a1−/−) was also significantly lower in Col6a1−/− diaphragms (***P < 0.001; **P < 0.01; *P < 0.05; unpaired two-tailed Student’s t test). All measurements were obtained from 4 mice per genotype. Error bars indicate s.d. (b) or s.e.m. (c). d EPP amplitude after repetitive stimulation at 5 or 10 Hz for 25 s in wild-type and Col6a1−/− diaphragms, showing enhanced reduction of postsynaptic stimulation after stress in Col6a1−/− animals (***P < 0.001; unpaired two-tailed Student’s t test; n = 123 EPP, 5 Hz wild-type, n = 152 EPP, 5 Hz Col6a1−/−; n = 116, 10 Hz wild-type; n = 168, 10 Hz Col6a1−/−). Measurements were obtained from 8 mice per genotype. Error bars indicate s.d. e EPP amplitude intracellularly recorded after tetanic stimulation (10 Hz) for 25 s in wild-type and Col6a1−/− soleus fibers (n = 8 fibers, each genotype). Measurements were obtained from 2 mice per genotype. Data are percentage to baseline EPPs evoked at 0.5 Hz before tetanic stimulation. Error bars indicate s.d. f CMAP measurements in wild-type and Col6a1−/− diaphragms under untreated conditions and in the presence of increasing concentrations (500 and 800 nM) of d-tubocurarine. Col6a1−/− animals show a significantly higher decrease of CMAP, which is already displayed at 500 nM d-tubocurarine, thus unmasking a reduced safety factor in the absence of ColVI (***P < 0.001; **P < 0.01; *P < 0.05; unpaired two-tailed Student’s t test; n = 6–11 mice, each genotype). Error bars indicate s.e.m

Full size image

We next assessed neuromuscular transmission by repetitive stimulation of the phrenic nerve for 25 pulses with 10 Hz (Fig. 4d). When analyzing the EPP amplitudes after 10 Hz stimulation, the decrement in the diaphragm of Col6a1−/− mice was higher than that of wild-type littermates, and a similar difference was found in soleus (Fig. 4e). We finally evaluated the safety factor, which reflects the fact that the threshold required to generate a muscle action potential is exceeded by the excitatory effect generated by nerve stimulation [72]. Towards this aim, we carried out compound muscle action potential (CMAP) measurements on wild-type and Col6a1−/− diaphragms in the presence of increasing concentrations of d-tubocurarine, in order to monitor the effect of a partial block of AChRs. Treatment with d-tubocurarine led to a strong decrease of CMAP amplitudes in response to repetitive stimuli in Col6a1−/− muscles but not in wild-type muscles (Fig. 4f), pointing at a reduced safety factor in ColVI-deficient animals. These results indicate that lack of ColVI impairs neuromuscular synaptic transmission, by affecting both sustained neurotransmitter release and safety factor.

Detection of NMJ alterations in UCMD patients

Next, we assessed whether the NMJ defects displayed by Col6a1−/− mice also occur in BM and UCMD patients. Towards this aim we investigated muscle biopsies already available in the Neuromuscular Bank of Tissues and DNA samples of the Telethon Network of Genetic BioBanks, which were isolated from six patients carrying different mutations in ColVI genes and displaying a range of clinical phenotypes, from typical BM to severe UCMD (Fig. 5a). The features of these muscle biopsies were evaluated in agreement with previous studies [59]. Histological analysis showed a prevalence of type I vs type II fibers in UCMD biopsies, when compared to BM and control biopsies (Fig. 5a and Suppl. Fig. S6). Immunofluorescence analysis for the myosin heavy chain (MHC) slow and fast isoforms largely confirmed the fiber type distribution reported in Fig. 5a for BM and UCMD patients (Suppl. Fig. S7). Interestingly, MHC immunolabeling performed in DMD and polymyositis biopsies did not show any overt disproportion in fiber type (Suppl. Fig. S7), thus sustaining the specificity of the results concerning patients affected by ColVI-related myopathies. Immunofluorescence showed close localization of ColVI labeling with αBT staining in control biopsies, whereas in BM and UCMD biopsies ColVI labeling appeared discontinuous and often detached from αBT staining (Fig. 5b). Moreover, and in keeping with the defect displayed by Col6a1−/− mice, immunostaining for collagen IV and utrophin appeared strongly upregulated at the level of NMJs in BM and UCMD patients’ biopsies, when compared to unaffected control biopsies (Fig. 5c). Finally, we evaluated the expression of genes coding for the different AChR subunits. Similarly to Col6a1−/− mice, UCMD biopsies showed increased mRNA levels of CHRNG and CHRNA genes, when compared to unaffected control biopsies (Fig. 5d). On the contrary, the mRNA levels of CHRNG and CHRNA genes in BM biopsies were, respectively, decreased or similar to those of control biopsies (Fig. 5d). This points at the presence of more overt NMJ defects in UCMD patients, compared to BM patients. Finally, we performed utrophin (Fig. 5e) and collagen IV (Fig. 5f) immunostaining also in two DMD and two polymyositis pathological controls, in order to gain further insight in the specificity of the postsynaptic defects detected in ColVI-related myopathies. This analysis did not show any major alteration in utrophin distribution in DMD or polymyositis biopsies, when compared to control biopsies (Fig. 5e). Collagen IV was abundant in DMD biopsies but not at the level of the NMJ, as instead displayed by UCMD and BM biopsies (Fig. 5f). Also compared to polymyositis biopsies, the deposition of collagen IV in UCMD/BM patients appeared higher (Fig. 5f), thus supporting the concept that the observed alterations are dependent on ColVI defects at the NMJ in BM and UCMD patients.

Fig. 5
figure 5

Hallmarks of NMJ defects in UCMD patients. a Table summarizing the clinical features, genetic characterization, and percentage of myofiber type for the BM and UCMD patients included in the study. NIMV, noninvasive mechanical ventilation; WCB, wheelchair bound. b Immunofluorescence for ColVI (green) in cross sections of muscle biopsies from unaffected control donors (CTRL), two BM patients, and two UCMD patients, stained with αBT (red). The insets show NMJ details at a higher magnification. Asterisks point at loss of contacts between ColVI and αBT labeling. Scale bar, 25 μm. c Representative immunolabeling for collagen IV and utrophin (white), performed together with αBT staining (red), in cross sections of muscle biopsies from unaffected control donor (CTRL), two BM patients, and two UCMD patients. Scale bar, 20 μm. d qRT-PCR analysis for the CHRNG and CHRNA genes in muscle biopsies from three unaffected control donors (C1-C3), three BM patients (P1-P3), and three UCMD patients (P4-P6). Dotted lines represent the mean value for each group, and the indicated P values were determined by the unpaired one-tailed Mann–Whitney test. e Representative immunolabeling for collagen IV (white) performed together with αBT staining (red), in cross sections of muscle biopsies from unaffected (CTRL) and pathological control donors, including two DMD patients (DMD1, DMD2), and two polymyositis patients (PM1, PM2). Scale bar, 25 μm. f Representative immunolabeling for utrophin (white) performed together with αBT staining (red), in cross sections of muscle biopsies from BM and UCMD patients in comparison with muscle biopsies from DMD and polymyositis patients. Arrowheads indicate NMJ, when covered by the signal of collagen IV staining. Scale bar, 25 μm

Full size image

ColVI regulates AChR clustering and synaptic gene expression in vitro

To evaluate whether ColVI has a direct role at the NMJ, we tested the ability of the purified protein to affect AChR clustering in differentiated C2C12 myotube cultures, using agrin, a protein known to induce AChR clustering [12, 24], as positive control (Fig. 6a). Treatment of C2C12 myotubes with purified ColVI alone was able to induce aggregation of longer AChR clusters (Fig. 6b, c). Notably, this enhancement was completely prevented by the concurrent addition of an anti-ColVI antibody, whereas ColVI did not display any synergistic effect with agrin (Fig. 6c). In order to understand whether the effect exerted from ColVI was specific, we confirmed an extremely low endogenous expression of ColVI in differentiated myotubes, by both quantitative real-time PCR and immunofluorescence (Suppl. Fig. S8a, b). Interestingly, after ColVI was added to culture medium, the protein could be detected in close contact with AChR clusters and in tight continuity with myotube membrane (Suppl. Fig. S8b). Moreover, ColVI preparation was tested for purity, displaying almost exclusively its tetrameric form (Suppl. Fig. S8c), and showing no contamination by other ECM molecules such as laminin and fibronectin (Suppl. Fig. S8d, e).

Fig. 6
figure 6

ColVI promotes AChR clustering in differentiated myotubes. a Representative images of differentiated C2C12 myotubes under untreated conditions (Control) or treated for 20 h with purified ColVI (1 μg/ml), a mixture of purified ColVI and anti-ColVI antibodies (anti-ColVI; 40 μg/ml), or agrin (10 ng/ml). αBT staining was performed to assess AChR cluster (arrowhead) length in the different conditions. Scale bar, 50 μm. b, c Quantification of the percentage of AChR clusters belonging to different size classes (b) and of the average AChR cluster length (c) in C2C12 myotubes cultured in the different conditions as in a. ColVI promotes AChR clustering, and this effect is abolished in the presence of anti-ColVI antibodies. Data are shown as mean ± s.e.m. of three independent replicates (**P < 0.01 vs untreated condition; *P < 0.05 vs untreated condition; §P < 0.05 vs ColVI treatment; unpaired two-tailed Student’s t test; n = 15-25 AChR clusters per condition). d Western blot for MyoG in protein lysates of C2C12 myotubes cultured in the different conditions as in a. Densitometric quantification of the MyoG/GAPDH ratio, as determined by three independent experiments, is shown in the right panel. The MyoG/GAPDH ratio is expressed as fold-change relative to the untreated control conditions (**P < 0.01 vs untreated condition; §P < 0.05 vs ColVI treatment; unpaired two-tailed Student’s t test). Error bars indicate s.e.m. Ctrl, control (no treatment). e qRT-PCR analysis for genes involved in postsynaptic AChR cluster stabilization, performed in C2C12 myotubes cultured in the different conditions as in a. The mRNA levels of Lrp4, Musk, Chrna, Chrng, Utrn and Rapsn genes (coding, respectively, for LRP4, MuSK, AChRα, AChRγ, utrophin and rapsyn) were determined by three independent experiments (**P < 0.01 vs untreated condition; *P < 0.05 vs untreated condition; §§P < 0.01 vs ColVI treatment; §P < 0.05 vs ColVI treatment; unpaired two-tailed Student’s t test). Error bars indicate s.e.m

Full size image

Since it was previously shown that AChR clustering depends on MyoG [34], we analyzed MyoG protein levels and found that they were significantly increased upon ColVI treatment, a response that was prevented by the addition of anti-ColVI antibodies (Fig. 6d). Agrin was shown to induce the expression of synaptic genes in vitro [32, 36]. We therefore investigated whether ColVI could, as well, modulate the expression of genes critical for postsynaptic AChR cluster stabilization. Treatment of C2C12 myotubes with native ColVI led to significant upregulation of the mRNA levels of Musk, Chrna, Chrng, Utrn and Rapsn genes, but not of the Lrp4 gene (Fig. 6e), as previously reported for agrin [25, 32, 36]. Since the presence of both N- and E-box regulatory elements was described for all the above ColVI responsive genes [16, 25, 26, 36, 43, 58, 62, 64], but not for Lpr4, these data point at the engagement of either ETS and/or bHLH transcription factor family-dependent pathways in the effects elicited by ColVI on AChR clustering.

Discussion

Previous work on ColVI in skeletal muscle, pointed at a crucial role for this protein in preserving myofiber homeostasis [7, 27], as well as in finely tuning the mechanical properties of muscle and in contributing to adult muscle satellite cell niche [67]. Here, we unravelled that ColVI is a component of the synaptic ECM and demonstrated that ColVI deficiency affects NMJ structure and function in vivo, thus highlighting an absolute novel role for ColVI in skeletal muscle. Our data indicate a major function for ColVI in promoting AChR clustering both in vivo and in vitro. Indeed, the tight contact of ColVI with AChR clusters, together with the enlargement and fragmentation of NMJs when the protein is missing, points at a role for ColVI in cluster stabilization.

While in the past only a limited number of proteins having a role within the synaptic basal lamina were described, the network of players in the last few years has become increasingly crowded. In the absence of ColVI, two of its known direct interactors [31, 66], collagen IV and perlecan, do increase at the NMJ level. Collagen IV is involved in the highly tuned organization of presynaptic terminals during development, and specific collagen IV chains are expressed at the synaptic cleft and are involved in the maintenance of NMJs [21, 39]. Perlecan was described as being involved in the stabilization of acetylcholinesterase clusters, and perlecan null mice are unable to express acetylcholinesterase at the NMJ [4]. Similarly, another interactor of ColVI, biglycan, was shown to be involved in agrin-dependent MuSK phosphorylation and in vitro AChR clustering [2]. Our data show that in the absence of ColVI both MuSK gene expression and LRP4 protein levels are altered. Altogether, these findings indicate a previously unexpected role for ColVI at the neuromuscular synaptic basal lamina, where its known binding partners are affected by its absence. Interestingly, a recent work demonstrated that Col6a3 gene expression is strongly affected in a model of congenital myasthenic syndrome, the ColQ null mouse, suggesting ColVI as a candidate partner for collagen Q [61] and opening interesting perspectives on ColVI involvement in other NMJ-specific diseases.

Lack of ColVI affects not only the expression of known interactors present in or facing the synaptic basal lamina, but also resulted in increased expression of NMJ-enriched proteins such as NOS1 and utrophin (as well as dystrophin). Of note, both proteins were reported to be modulated by electrical activity, one being overexpressed upon denervation [5], the other upon reinnervation [65], again pointing at defects in neuromuscular transmission in Col6a1−/− mice. Although nerve terminals do not seem to fail in reaching the postsynaptic boutons in Col6a1−/− mice, both NMJ fragmentation and AChRγ and AChRα mRNA upregulation were previously reported as clear signs of either defective NMJ contacts and transmission, or re-innervation defect, as it occurs in aging [9, 19, 37, 64, 71]. Indeed, our data show that ColVI deficiency affects neuromuscular transmission, as both nerve-evoked EPPs and nerve-independent mEPPs are significantly reduced in the soleus and diaphragm muscles of Col6a1−/− mice, together with a reduction of the safety factor.

Remarkably, CHRNG and CHRNA genes were also differentially upregulated in the more severe UCMD patients, compared to BM. Although limited to the analysis of muscle samples already available in a repository, due to constraints and ethical concerns in obtaining fresh nerve-guided muscle biopsies, the performed immunofluorescence data on patients’ biopsies do show alterations in ColVI deposition at the NMJ and dysregulation of NMJ-enriched proteins, such as collagen IV and utrophin, in agreement with what displayed by Col6a1−/− mice. These data suggest the presence of NMJ abnormalities not only in Col6a1−/− mice, but also in patients affected by ColVI-related diseases. In particular, the more overt NMJ defects in UCMD patients, compared to BM patients, suggest that NMJ alterations may substantially contribute to the worsening of ColVI-dependent myopathic features. In a more affected muscle, the latent safety factor defects, seen in Col6a1−/− mice, may enter into play, thus compromising neuromuscular transmission and contributing to a faster decline of muscle function in UCMD patients. On the other hand, most neuromuscular diseases ultimately impact the neuromuscular unit, leading to common clinical features including daily fluctuations of stamina, mostly related to muscle weakness [18]. Therefore, the potential implication of the nervous counterpart and of NMJ in BM and UCMD need further studies to be fully characterized. Detailed electromyographic studies of the NMJ and the demonstration of an impairment of the safety margin of neuromuscular transmission would be the proof-of-principle of the pathogenic role of NMJ in collagen VI-related myopathies [40]. However, such in-depth studies are not part of the routine assessment in either BM or UCMD and they have not been performed in our cohort of patients. Despite this, the observed defects at the NMJ in patients affected by ColVI-related myopathies, described in this work, open the field for deeper prospective studies, also in the light of recent data reporting a number of patients with both histological (fiber type grouping and fiber splitting) and electromyographic neurogenic features, suggesting the occurrence of denervation [29, 68].

Of interest, the most recent approaches for ColVI-related myopathies targeted the modulation of autophagic activity, which was found to be altered both in patients and in Col6a1−/− mice [10, 15]. Since autophagy was demonstrated to have a role in maintaining NMJ functionality [9, 50], NMJ amelioration should be taken into account as novel clinical endpoint, also prompting the search for noninvasive methods to assess it [57].