Chemical crosslinking analysis of β-dystroglycan in dystrophin-deficient skeletal muscle

Materials

General analytical grade reagents and materials for chemical crosslinking, immunoprecipitation, gel electrophoresis, mass spectrometry and immunoblotting were supplied by Sigma Chemical Company (Dorset, UK), Bio-Rad Laboratories (Hemel-Hempstead, Hertfordshire, UK), GE Healthcare (Little Chalfont, Buckinghamshire, UK) and National Diagnostics (Atlanta, GA, USA). The homo-bifunctional and amine-reactive crosslinker bis[sulfosuccinimidyl]suberate (BS³) of 11.4-Å spacer arm length (catalogue number 21580) and C18 spin columns (catalogue number 89870) were purchased from Thermo Fisher Scientific (Dublin, Ireland). Chemiluminescence substrate (catalogue number 11500694001) and protease inhibitor cocktails (catalogue number 11836153001) were obtained from Roche Diagnostics (Mannheim, Germany). Biobasic C18 Picofrit columns (catalogue number 72105-254630) were from Dionex (Sunnyvale, CA, USA). Proteolytic digestion for the generation of peptides prior to mass spectrometric analysis was carried out with sequencing grade modified trypsin (catalogue number V5111) from Promega (Madison, WI, USA). Whatman nitrocellulose transfer membranes (catalogue number 88018) came from Invitrogen (Carlsbad, CA, USA). Primary antibodies were purchased from the following companies: Abcam, Cambridge, UK (polyclonal antibody ab43125 to β-dystroglycan, RRID: AB_955822, used at a dilution of 1:250; rabbit polyclonal antibody ab190264 to myoferlin (product code ab190264), used at a dilution of 1:250; monoclonal antibody ab124684 to dysferlin; RRID: AB_10976241, used at a dilution of 1:100), Novus Biologicals, Littleton, CO, USA (polyclonal antibody NBP1-80220 to cavin 1/PTRF (polymerase and transcript release factor); RRID: AB_11019160) and Santa Cruz Biotechnology, Santa Cruz, CA, USA (monoclonal antibody sc-33701 to β-dystroglycan; RRID: AB_627294). Merck-Millipore (MA, USA) provided peroxidase-conjugated secondary antibodies [catalogue numbers AQ132P (anti-rabbit) and 12-349 (anti-mouse)].

Isolation of the microsomal fraction from skeletal muscle

A crude microsomal membrane fraction was isolated from wild type and dystrophic mdx-4cv muscle samples, which were obtained from the Animal Facility of the University of Bonn²⁶. Transportation of quick-frozen muscle specimens to Maynooth University was carried out on dry ice in accordance with the Department of Agriculture animal by-product register number 2016/16 to the Department of Biology, Maynooth University. As outlined in the flow chart of Figure 1, combined skeletal muscles of the hind leg from 5-month old dystrophic mdx-4cv mice (n=4) and age-matched C57Bl/6 mice (n=4) were finely chopped and then homogenised on ice with the help of a hand-held IKA T10 Basic Homogeniser (IKA Labortechnik, Staufen, Germany). Homogenisation was performed in 10 volumes of buffer (20mM sodium pyrophosphate, 20mM sodium phosphate, 1mM MgCl₂, 0.303M sucrose, 0.5mM EDTA, pH 7.0) that was supplemented with a protease inhibitor cocktail from Roche Diagnostics (Mannheim, Germany). The suspension was incubated for 2h at 4°C and centrifuged at 14,000×g for 20 min at 4°C with a model 5417 R centrifuge from Eppendorf (Hamburg, Germany) to remove cellular debris. The membrane-containing supernatant was transferred to 4.9ml Optiseal ultracentrifugation tubes and centrifuged at 100,000×g for 1h at 4°C using an Optima L-100 XP ultracentrifuge from Beckman Coulter Incorporation (Fullerton, CA, USA). The resulting pellets with the crude microsomal membrane fraction were re-suspended in an appropriate volume of homogenisation buffer and then stored at -20°C until usage for mass spectrometric analysis, immunoprecipitation or chemical crosslinking. The protein concentration of individual samples was determined by the method of Bradford²⁷.

Figure 1. Bioanalytical workflow to identify β-dystroglycan and determine its oligomeric status in muscle membranes.

Shown is an overview of the differential centrifugation scheme to isolate the crude microsomal fraction from wild type versus dystrophic mdx-4cv skeletal muscle homogenates. To evaluate potential protein interactions, both immunoprecipitation and chemical crosslinking in combination with gel-shift and immunoblot analysis were employed. The presence of dystroglycans in microsomes was established by mass spectrometry, which identified a considerable number of unique peptide sequences that are characteristic for both the α-subunit and the β-subunit of the dystroglycan subcomplex. The lower panel shows the peptide sequence of the pre-pro-protein of 893 amino acids including the signalling peptide and both dystroglycan subunits, which approximate portions of the entire protein sequence are marked on the right. The sequences of unique peptides are in bold and underlined.

Immunoprecipitation analysis of β-dystroglycan in normal skeletal muscle

The evaluation of the immunoprecipitation technique²⁸ as a potential bioanalytical approach for the identification of dystroglycan-associated proteins¹⁷ was carried out with crude microsomes that had been isolated from C57Bl/6 mice. Two distinct antibodies to β-dystroglycan were used, i.e. ab43125 from Abcam (Cambridge, UK) and sc-33701 from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Magnetic beads coupled to Protein A (for antibody ab43125) or Protein G (for antibody sc-33701) were used for the pre-clearance of 1mg of membrane protein using an incubation step for 30 min at 4°C with end-over-end rotation. The supernatant was isolated and incubated with 5µg antibody overnight at 4°C with gentle end-over-end rotation. The following day, the suspension of protein and antibody was added to 25µl of magnetic beads and incubated for 1h at room temperature with end-over-end rotation. The supernatant was removed using a magnetic rack, and the beads were washed five times, thrice with Tris-buffered saline with 0.05% Tween-20 and twice with distilled water. Proteins were subsequently eluted in one of two ways. Proteins were eluted first with 2M urea, 50mM Tris-HCl, pH 7.5 and 5µg/ml trypsin and incubated at 27°C for 30 min with gentle shaking. This supernatant was collected, and the magnetic beads were then washed twice with 2M urea, 50mM Tris-HCl, pH 7.5, 1mM dithiothreitol. The supernatant from this step was combined with the supernatant from the first elution. The samples were left to digest overnight at room temperature. The following day, 20µl of 5mg/ml iodoacetamide was added and incubated at room temperature in the dark for 30 min. 2% trifluoroacetic acid (TFA) in 20% acetonitrile (ACN) (3:1 (v/v) dilution) was then added and peptides were purified using C18 spin columns before mass spectrometry. Alternatively, after the five washes of the magnetic beads, proteins were eluted with Laemmli-type buffer²⁹ and incubated at 97°C for 7 min. These protein samples were digested by the FASP (filter aided sample preparation) method³⁰ and purified by C18 spin columns. In both cases, peptides were re-suspended in loading buffer (2% ACN, 0.05% TFA in LC-MS grade water), and were analysed by a Q-Exactive mass spectrometer over a 65 min gradient. Raw files obtained from mass spectrometry were analysed by Proteome Discoverer 1.4 and filtered for high confidence peptides.

Comparative chemical crosslinking analysis using bis(sulfosuccinimidyl)suberate

The water-soluble cross-linker bis(sulfosuccinimidyl)suberate (BS³) was dissolved at a concentration of 1mg/ml in 50mM citrate buffer, pH 5.0³¹. Microsomal samples were diluted to a concentration of 2 mg protein/ml with 50mM HEPES, pH 8.0. A concentration range of 10, 25, 50, 75, 100 and 150μg BS³ per mg protein was used to evaluate the crosslinking protocol employed in this study²⁴. For the comparative crosslinking and immunoblot analysis of β-dystroglycan, cavin-1, myoferlin and dysferlin oligomerisation, membrane samples were incubated for 30 min at 25°C with 0 versus 10µg BS³ per mg protein. The addition of 50μl 1M ammonium acetate per ml reaction mixture was used to quench chemical crosslinking and then an equal volume of reducing sample buffer was added to the samples. Following a heating step for 10 min at 50°C, crosslinked proteins were separated by one-dimensional gel electrophoresis alongside their non-crosslinked counterparts²⁴.

Label-free liquid chromatography mass spectrometric analysis

Protein digestion and mass spectrometric analysis of peptide populations was carried out by an optimized method²⁶. Following pre-treatment of the microsomal fraction with the Ready Prep 2D clean up kit (catalogue number 163-2130) from Bio-Rad Laboratories (Hemel-Hempstead, Hertfordshire, UK), pellets obtained were re-suspended in label-free solubilisation buffer (6 M urea, 2 M thiourea, 10 mM Tris, pH 8.0 in LC-MS grade water) and protein suspension volumes were equalised with label-free solubilisation buffer. Samples were reduced with dithiotreitol and alkylated with iodoacetamide, followed by the proteolytic digestion using a combination of the enzymes Lys-C and trypsin, as previously described in detail³². The digestion step was stopped by acidification with 2% TFA in 20% ACN (3:1 (v/v) dilution). Peptide suspensions were purified using Pierce C18 Spin Columns from Thermo Fisher Scientific (Dublin, Ireland) and the resulting peptide samples were dried through vacuum centrifugation and suspended in loading buffer consisting of 2% ACN and 0.05% TFA in LC-MS grade water. An Ultimate 3000 NanoLC system (Dionex Corporation, Sunnyvale, CA, USA) coupled to a Q-Exactive mass spectrometer (Thermo Fisher Scientific), situated in the Proteomics Suite of Maynooth University, was used for the label-free liquid chromatography mass spectrometric (LC-MS/MS) identification of dystroglycan. Peptide mixtures were loaded by an autosampler onto a C18 trap column (C18 PepMap, 300 µm id × 5 mm, 5 µm particle size, 100-Å pore size; Thermo Fisher Scientific). The trap column was switched on-line with an analytical Biobasic C18 Picofrit column (C18 PepMap, 75 µm id × 50 cm, 2 µm particle size, 100-Å pore size; Dionex). Peptides generated from muscle microsomal proteins were eluted using the following binary gradient solvent A [2% (v/v) ACN and 0.1% (v/v) formic acid in LC-MS grade water] and 0–90% solvent B [80% (v/v) ACN and 0.1% (v/v) formic acid in LC-MS grade water]: 0–2% solvent B for 10.5 min, 2–40% solvent B for 110 min, 40–90% solvent B for 2.5 min, 90% solvent B for 9 min and 2% solvent B for 43 min. Peptides generated from immunoprecipitation were eluted over the same binary gradient, with 0–90% solvent B over a 65 min gradient. The column flow rate was set to 0.25 µl/min. Data were acquired with Xcalibur software version 2.0.7 (Thermo Fisher Scientific). The mass spectrometer was operated in positive mode and data-dependent mode and was externally calibrated. Survey MS scans were conducted in the Q-Exactive mass spectrometer in the 300–1700 m/z range with a resolution of 140,000 (m/z 200) and lock mass set to 445.12003. CID (collision-induced dissociation) fragmentation was carried out with the fifteen most intense ions per scan and at 17,500 resolution. Within 30s, a dynamic exclusion window was applied. An isolation window of 2 m/z and one microscan were used to collect suitable tandem mass spectra. Mass spectrometry files were analysed by Proteome Discoverer 1.4 (Thermo Fisher Scientific) software with Sequest HT as the search engine and the UniProtKB-SwissProt sequence database. The following search parameters were used for protein identification: (i) peptide mass tolerance set to 10 ppm, (ii) MS/MS mass tolerance set to 0.02 Da, (iii) up to two missed cleavages, (iv) carbamidomethylation set as a fixed modification, (v) methionine oxidation set as a variable modification and vi) peptide probability set to high confidence.

Immunoblot analysis of chemically crosslinked muscle proteins

A comparative immunoblotting approach was employed to evaluate the potential shift in electrophoretic mobility of muscle protein species following chemical crosslinking^33–35. For immunoblotting, microsomes were suspended in 2x standard Laemmli-type buffer for one-dimensional sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE)²⁹, heated for 10 min at 50°C and then loaded onto hand-cast 10% SDS-PAGE gels. Following gel electrophoretic separation, proteins were transferred to nitrocellulose membranes, blocked in a fat-free milk protein solution (2.5% milk powder in 10% phosphate-buffered saline) and incubated 1.5 h in an appropriately diluted solution with primary antibody³⁶. Immuno-decorated membranes were carefully washed and incubated overnight with peroxidase-conjugated secondary antibodies. Antibody-labelled protein bands were visualized with the enhanced chemiluminescence technique³⁷.

Mass spectrometric identification of β-dystroglycan in the microsomal fraction from skeletal muscle

Mass spectrometry-based proteomics was used to identify dystroglycan in the muscle membrane fraction. As a key component of the dystrophin-associated glycoprotein complex, the products of the DAG1 gene (that encodes dystroglycan with the sequence ID NP_034147.1) form a sub-complex of the core dystrophin network, which links the extracellular matrix protein laminin-211 of the basal lamina and the subsarcolemmal actin cytoskeleton in skeletal muscle. The encoded pre-pro-protein of 895 amino acids (aa), which includes the signalling peptide and both dystroglycan subunits³⁸, undergoes extensive O- and N-glycosylation, as well as proteolytic processing to generate extracellular α-dystroglycan and the integral sarcolemma protein β-dystroglycan in rabbit muscle³⁶. The equivalent pre-pro-protein from mice shows a high degree of sequence homology and consists of 893 aa. The mass spectrometric survey of mouse skeletal muscle microsomes identified unique peptide sequences that cover both the α-subunit (aa 28–651) and the β-subunit (aa 652–893) of the dystroglycan subcomplex (Figure 1). However, mass spectrometric analysis did not efficiently recognize this relatively low-abundance protein after immunoprecipitation or chemical crosslinking. As an alternative approach, comparative immunoblotting was used to evaluate the effects of chemical crosslinking.

Immunoprecipitation analysis of β-dystroglycan

Prior to chemical crosslinking analysis, we attempted to study potential interactions between β-dystroglycan and closely associated muscle proteins using one of the most frequently employed methods for the investigation of protein-protein interactions, i.e. co-immunoprecipitation. However, this approach resulted in a large number of precipitated protein species that appeared to lack a high degree of specificity, so this bioanalytical approach was therefore not further pursued. Antibody-based precipitation analysis led to the identification of 228 proteins with ≥ 2 unique peptides using the two-step urea-based elution method with the polyclonal antibody ab43125, and 211 proteins with ≥ 2 unique peptides with the monoclonal antibody sc-33701. Elution with Laemmli-buffer resulted in the identification of 273 proteins with ≥ 2 unique peptides with the ab43125 antibody and 245 proteins with ≥ 2 unique peptides with the sc-33701 antibody. Interesting co-immunoprecipitated proteins, which were previously shown to interact directly or indirectly with dystrophin⁵, included α-sarcoglycan, β-sarcoglycan, δ-sarcoglycan, γ-sarcoglycan, desmin, synemin, dysferlin, tubulin, cytokeratin, vimentin, tetraspanin and desmoplakin. However, dystroglycan (Q62165) was recognized by only 2 peptides, which equates to 2% coverage of the DAG1 sequence. This was deemed not to be efficient enough for a reliable oligomerization analysis of this dystrophin-associated glycoprotein.

Chemical crosslinking analysis of β-dystroglycan

An established non-cleavable and homo-bifunctional agent with amine-reactive and water-soluble properties, bis[sulfosuccinimidyl]suberate (BS³), was used to evaluate patterns of protein clustering in the microsomal membrane fraction from normal versus dystrophin-deficient skeletal muscles. Artefacts of random crosslinking are known challenges in chemical crosslinking experiments, although it has been suggested that only those proteins that are non-randomly closely associated will become crosslinked with sufficient frequency to be detected amongst the background of all other proteins. To minimize any potential artefacts, conditions regarding buffer composition, temperature, pH-value and length of incubation time, parameters which have been previously optimised by our laboratory for muscle membrane proteins^{24,31,33–35}, were carefully controlled. In particular, the protein-crosslinker conjugation reaction was carried out under experimental conditions, which are relatively close to physiological conditions and low levels of crosslinker were used (10µg BS³ per mg protein)²⁴.

The degree of chemical crosslinker-induced protein oligomerisation was determined by a combination of one-dimensional gel-shift analysis and immunoblotting. Building on the recently established and optimized protocol for the comparative chemical crosslinking analysis of muscular dystrophy²⁴, antibody decoration of β-dystroglycan was used to study the oligomeric status of this sarcolemmal glycoprotein. As illustrated in the silver-stained 10% SDS-PAGE gel in Figure 2, increasing amounts of BS³ had a considerable influence on the protein banding pattern with the appearance of a distinct band above the 150 kDa gel zone. Immunoblotting of crosslinked complexes was carried out with a 6% gel system in order to capture the potential appearance of proteins with a shift to very high molecular mass. The immunoblot in Figure 2 demonstrated a clear shift to reduced electrophoretic mobility of the monomeric 43 kDa band of β-dystroglycan to a region of above 250 kDa following incubation with BS³.

Figure 2. Chemical crosslinking, gel shift and immunoblot analysis of β-dystroglycan from skeletal muscle membranes.

Shown is a silver-stained 10% SDS-PAGE gel of normal mouse muscle microsomes that were incubated with 0, 10, 25, 50, 75, 100 and 150 μg of the homo-bifunctional, non-cleavable, water-soluble and amine-reactive 11,4-Å cross-linker bis(sulfosuccinimidyl)suberate (BS³) per mg membrane protein, as well as a corresponding immunoblot that has been transferred from a 6% gel and labelled with an antibody to the dystrophin-associated glycoprotein β-dystroglycan. Molecular mass standards (in kDa) are indicated on the left of panels. The position of monomeric versus crosslinker-stabilized oligomeric forms of β-dystroglycan (DG) are marked by arrowheads.

The comparative analysis of the crosslinked microsomal protein fraction from wild type versus the mdx-4cv mouse model of dystrophinopathy, using an optimized ratio of 10μg BS³ per mg microsomal protein, revealed oligomers that contain β-dystroglycan (Figure 3). The densitometric analysis of the silver-stained gel illustrates statistically significant increases in densitometry values corresponding to increased protein staining in high-molecular mass regions of the gel above 150 kDa for both wild type and dystrophic crosslinked samples (Figure 4). Hence, chemical crosslinking resulted in the stabilisation of protein complexes and reduced electrophoretic mobility of protein species, resulting in an increased number of proteins remaining at high molecular mass regions following gel electrophoretic separation. Interestingly this sarcolemma-spanning key component of the core dystrophin complex exhibited a comparable reduction in gel electrophoretic mobility in both normal and dystrophic samples.

Figure 3. Comparative chemical crosslinking analysis of β-dystroglycan, cavin-1, myoferlin and dysferlin from normal versus dystrophic muscle.

Shown is a silver-stained SDS-PAGE gel of muscle microsomes from wild type and dystrophic mdx-4cv skeletal muscles and corresponding immunoblots that were labelled with antibodies to β-dystroglycan, cavin-1, myoferlin and dysferlin. Lanes 1 to 4 represent microsomal samples from wild type (wt) and mdx-4cv muscles that were incubated with 0 versus 10 μg of the cross-linker bis(sulfosuccinimidyl)suberate (BS³) per mg protein, respectively. Molecular mass standards (in kDa) are indicated on the left of the gel panel. The position of monomeric versus crosslinker (XL) -stabilized oligomeric forms of β-dystroglycan (DG), cavin-1 (CAV1) myoferlin (MYF) and dysferlin (DYF) are marked by arrowheads.

Figure 4. Densitometric scanning of chemical crosslinking and gel-shift analysis of normal versus dystrophic muscle.

Shown are densitometric values, which reflect the extent of protein staining in wild type control, wild type crosslinked, mdx-4cv control and mdx-4cv crosslinked samples. The comparative densitometric evaluation was focused on protein species above 150 kDa. Statistical analysis was performed with a Student’s t-test (mean value ±SEM; n=4; **p≤0.01; ***p≤0.001).

Immunoblotting with antibodies to cavin-1 identified this caveolae-associated protein in both control and crosslinked samples from both wild type and mdx-4cv muscle preparations (Figure 3). In control lanes, a cavin-1 band was identified at approximately 80 kDa, which is higher than the previously observed electrophoresed protein band of the apparent monomer at 43 kDa, so this band might present a dimeric structure. The immuno-decoration at a higher molecular mass may also be due to post-translational modifications, abnormal electrophoretic mobility patterns and/or incomplete protein reduction during sample preparation for gel electrophoretic separation. In both wild type and mdx-4cv crosslinked preparations, cavin-1 was detected at lower amounts at approximately 80 kDa than in control lanes, and 3 additional bands were evident at approximately 100, 150 and 260 kDa (Figure 3). This indicates that cavin-1 may be involved in a number of protein interactions giving rise to these 3 distinct bands in crosslinked muscle samples. The fainter upper cavin-1 band of approximately 260 kDa overlaps with the position of crosslinked β-dystroglycan, which suggests that a certain proportion of cavin-1 molecules may exist in a protein complex with β-dystroglycan.

The membrane repair proteins dysferlin and myoferlin, which play a central role in fibre regeneration and counteract the dystrophic phenotype, were also shown to exist in high-molecular mass complexes in normal and dystrophic muscle (Figure 3). However, in contrast to β-dystroglycan, both sarcolemmal repair proteins only showed a portion of their monomeric protein species to shift to a higher molecular mass range. This limited reduction in electrophoretic mobility of myoferlin and dysferlin in both wild type and dystrophic skeletal muscle, upon the addition of the BS³ chemical cross-linker, suggests that only a sub-population of these proteins form tight protein complexes. The presence of additional bands representing these proteins in mdx-4cv muscle indicates that in dystrophic muscle these proteins probably form additional protein complexes in a dynamic fashion, and that some of these protein assemblies are not present in wild type skeletal muscle.