Carbohydrate modifications are clearly important to the function of α-dystroglycan but their composition and structure remain poorly understood. In the present study, we describe experiments aimed at identifying the α-dystroglycan oligosaccharides important for its binding to laminin-1 and carbohydrate-dependent mAbs (monoclonal antibodies) IIH6 and VIA41. We digested highly purified skeletal muscle α-dystroglycan with an array of linkage-specific endo- and exoglycosidases, which were verified for action on α-dystroglycan by loss/gain of reactivity for lectins with defined glyco-epitopes. Notably, digestion with a combination of Arthrobacter ureafaciens sialidase, β(1-4)galactosidase and β-N-acetylglucosaminidase substantially degraded SiaAα2-3Galβ1-4GlcNAcβ1-2Man glycans on highly purified α-dystroglycan that nonetheless exhibited enhanced IIH6, VIA41 and laminin-1 binding activity. Additional results indicate that α-dystroglycan is probably modified with other anionic sugars besides sialic acid and suggest that rare α-linked GlcNAc moieties may block its complete deglycosylation with currently available enzymes.
The dystroglycan complex is a component of the dystrophin–glycoprotein complex, which spans the sarcolemma of striated muscle cells and physically couples the actin cytoskeleton with the extracellular matrix [1–3]. The dystroglycan complex consists of α-dystroglycan, a highly glycosylated extracellular protein that binds to several extracellular ligands, and β-dystroglycan, a single-pass transmembrane protein that links cytoplasmic dystrophin with α-dystroglycan . Both dystroglycan subunits are encoded by a single highly conserved propeptide that is proteolytically processed into α- and β-dystroglycan, which remain stably associated through non-covalent interactions . Like muscles deficient in dystrophin or other core components of the dystrophin–glycoprotein complex [1,2], deficiency of the dystroglycan complex in skeletal muscle results in compromised sarcolemmal integrity [4,5]. Thus it is generally thought that one important function of the dystroglycan complex in skeletal muscle is to mechanically protect the sarcolemma against shear stresses imposed during muscle contraction.
O-mannosyl-linked oligosaccharides are clearly important for α-dystroglycan binding to its extracellular ligands as well as its function in vivo because mutations in glycosyltransferases that O-mannosylate α-dystroglycan result in loss of extracellular ligand binding activity and muscular dystrophy [3,6–9]. α-Dystroglycan is also modified by more generic N-linked glycans and sialylated core 1 oligosaccharides with the structure SiaAα(2-3)Galβ(1-3)GalNAc . However, most studies show that enzymatic removal of these glycans had no effect on α-dystroglycan binding to laminins , agrin  or neurexins . At least two different O-mannosyl-linked oligosaccharides modify α-dystroglycan [14–16] but their specific role in extracellular matrix ligand binding is controversial . In the present study, we attempted to identify α-dystroglycan oligosaccharides that are important for laminin-1 binding through digestion with various combinations of linkage-specific endo- and exoglycosidases and reactivity with a panel of lectins of known carbohydrate specificities. The sialyl O-mannosyl glycan previously implicated in α-dystroglycan binding to laminin  was substantially degraded with a cocktail of three glycosidases, yet the laminin-1, monoclonal IIH6 and VIA41 binding activities of α-dystroglycan were all markedly increased after enzymatic deglycosylation. Other results suggest that α-dystroglycan oligosaccharides probably contain anionic sugars in addition to sialic acid and possibly rare α-linked GlcNAc moieties that could block complete enzymatic deglycosylation with currently available reagents.
MATERIALS AND METHODS
Peroxidase-conjugated ConA (concanavalin A), ECA (Erythrina cristagalli agglutinin), WGA (wheat germ agglutinin) and ExtrAvidin were purchased from Sigma–Aldrich (St Louis, MO, U.S.A.). Peroxidase-conjugated LFA (Limax flavus agglutinin) and PNA (peanut agglutinin) were purchased from EY Laboratories (San Mateo, CA, U.S.A.). Biotin-conjugated GNL (Galanthus nivalis), GsII (Griffonia simplicifolia II) and Hippeastrum hybrid lectins were purchased from Vector Laboratories (Burlingame, CA, U.S.A.). α-Dystroglycan-specific mAbs (monoclonal antibodies) IIH6 and VIA41  were obtained from the University of Iowa Hybridoma Facility (IA, U.S.A), and Upstate Biotechnology (Lake Placid, NY, U.S.A.) respectively. Affinity-purified chicken polyclonal antibodies to α-dystroglycan were prepared as described previously . The DIG glycan detection kit and peroxidase-conjugated secondary antibodies were purchased from Roche Applied Science (Indianapolis, IN, U.S.A.). Purified mouse EHS laminin-1 was kindly provided by Dr H. Kleinman (Cell Biology Section, National Institute of Dental Research NIH, Bethesda, MD, U.S.A.).
α-Dystroglycan was purified from rabbit skeletal muscle by the method of Brancaccio and co-workers  with several modifications. Freshly frozen rabbit back and leg muscle (300 g) was homogenized in 5 ml/g of 50 mM Tris/HCl (pH 7.4), 200 mM NaCl, 0.02% NaN3 and 0.2 mM PMSF, centrifuged for 20 min at 30000 g, and the resulting supernatant loaded on to a 5 cm×14 cm DEAE-Sephacel column. After extensive washing, bound proteins were eluted in bulk with 230 ml of 50 mM Tris/HCl (pH 7.4), 500 mM NaCl and 0.02% NaN3. The DEAE eluate was circulated overnight at 4 °C through a 2.5 cm×12 cm WGA-agarose column, which was washed extensively and bound proteins eluted in bulk with 100 ml of 0.3 M N-acetyl-D-glucosamine in 50 mM Tris/HCl (pH 7.4), 500 mM NaCl and 0.02% NaN3. The WGA eluate was diluted with 233 ml of 50 mM Tris/HCl (pH 7.4), 1.43 mM CaCl2 and 1.43 mM MgCl2, and circulated overnight through a 2.5 cm×5 cm laminin-1-Sepharose column. After extensive washing with 50 mM Tris/HCl (pH 7.4), 150 mM NaCl, 1 mM CaCl2 and 1 mM MgCl2, bound proteins were eluted in 5 ml fractions with 100 ml of 50 mM Tris/HCl (pH 7.4), 1 M NaCl and 10 mM EDTA. The fractions containing α-dystroglycan were identified by Western-blot analysis, pooled and further purified by CsCl gradient centrifugation as described previously . Purified α-dystroglycan was exhaustively dialysed against water and quantified by A280 using E280=0.83 cm2/mg . Over ten trials, the method yielded on average 30 μg of purified α-dystroglycan starting with 300 g of skeletal muscle.
Unless noted otherwise, 5 μg of purified rabbit skeletal α-dystroglycan was incubated with individual enzymes or various enzyme combinations for 16–20 h at 37 °C in 100 μl of 50 mM sodium phosphate at pH 5, pH 6 or pH 7. Enzyme digestions were terminated by the addition of 25 μl of 5× Laemmli sample buffer (15%, w/v, SDS, 0.575 M sucrose, 0.325 M Tris/HCl, pH 6.8 and 5%, w/v, 2-mercaptoethanol), and 25 μl aliquots (containing 1 μg α-dystroglycan) were analysed by SDS/PAGE, lectin and Western blotting or laminin overlay as described below. The GLYCOPRO deglycosylation kit and PRO-LINK extender kit were purchased from Prozyme (San Leandro, CA, U.S.A.) and included the following enzymes with the concentrations used in parentheses: N-glycosidase F (50 units/ml), Arthrobacter ureafaciens sialidase (50 mU/ml), endo-O-glycosidase (12.5 mU/ml), β(1-4)galactosidase (30 mU/ml) and β-N-acetylglucosaminidase (0.4 unit/ml). The original source of enzymes provided in the GLYCOPRO deglycosylation kit and PRO-LINK extender kit used in the present study are no longer available directly from Prozyme but can be purchased in kits available from Calbiochem (San Diego, CA, U.S.A.), Sigma–Aldrich or QA-Bio (San Mateo, CA, U.S.A.).
Purified skeletal muscle α-dystroglycan (5 μg) was incubated in 50 μl of 10 mM sodium metaperiodate, 50 mM sodium acetate (pH 5.5) for 0–6 h and the reaction terminated by the addition of 50 μl of 20 mM sodium disulphite and 25 μl of 5× Laemmli sample buffer. Aliquots (25 μl) were analysed by SDS/PAGE, lectin and Western blotting or laminin overlay as described below.
SDS/PAGE, Western and lectin blotting
Samples were resolved on 3–12% SDS/polyacrylamide gels and either stained with Coomassie Blue, Stains-All, or transferred on to nitrocellulose as described previously . Western blotting was performed as described previously . Lectin blotting was performed by blocking nitrocellulose transfers with PBS containing 0.05% Tween 20 followed by incubation with 1 μg/ml of biotinylated or peroxidase-conjugated lectin in blocking buffer for 1 h at room temperature (25 °C). Blots incubated with biotinylated lectins were washed twice with blocking buffer and incubated for 1 h at room temperature with a 1:1000 dilution of peroxidase-conjugated ExtrAvidin in blocking buffer. Lectin blots were washed extensively with PBS before staining was developed in 20 mM Tris/HCl (pH 7.4) using 4-Cl-1-napthol as substrate, or by enhanced chemiluminescence using Pierce SuperSignal Pico West substrate. Oxidation and digoxigenin labelling of blotted α-dystroglycan followed method B of the manufacturer's instructions provided with the DIG glycan detection kit. Images of gels, blots and chemiluminescence films were acquired using a UVP GelDoc imaging system and LabWorks image analysis software before importation into CorelDraw 10 for figure preparation.
Laminin-1 binding to blotted control and glycosidase-digested α-dystroglycan was assessed using a previously described laminin overlay assay . Solid-phase radioligand and ELISA binding assays were performed as described previously [17,19]. Briefly, break-apart polystyrene microtitre wells coated overnight at 4 °C with 0.1 μg of purified laminin-1 or α-dystroglycan were blocked with 3% (w/v) BSA in 50 mM Tris/HCl (pH 7.4), 150 mM NaCl, 1 mM CaCl2 and 1 mM MgCl2, and incubated with 1–80 nM 125I-α-dystroglycan or 4 nM laminin-1 in blocking buffer. After two brief washes with blocking buffer, binding of 125I-α-dystroglycan to immobilized laminin-1 was determined by breaking apart individual microtitre wells, which were counted in a Packard 5650 γ-counter . The binding of unlabelled laminin-1 to immobilized α-dystroglycan was detected with affinity-purified rabbit polyclonal antibodies against laminin-1, followed by horseradish peroxidase-conjugated anti-rabbit secondary antibody and colour development with 3,3′,5,5′-tetramethylbenzidine (Bio-Rad Laboratories, Hercules, CA, U.S.A.). Colour development was quantified kinetically at 655 nm using a Molecular Dynamics Spectramax 340 UV–visible microplate reader.
Characterization of purified skeletal muscle α-dystroglycan
Although α-dystroglycan is typically purified from tissue homogenates or membrane-enriched fractions through various combinations of lectin, ion exchange and laminin-1-affinity chromatography, the method of Brancaccio and co-workers [18,20] is widely used because it is the least labour-intensive, requires only commercially available reagents, and can be applied to a variety of frozen tissues. The method involves a combination of DEAE-Sephacel and WGA-Sepharose chromatography of Tris-buffered saline extracts sometimes followed by laminin-1 affinity chromatography. When starting with rabbit skeletal muscle, we find that the method consistently yields fractions greatly enriched in α-dystroglycan immunoreactivity and laminin-binding activity. However, the presence of several residual glycoprotein contaminants potentially complicates the characterization of the carbohydrate structures present specifically on α-dystroglycan. We previously observed that α-dystroglycan sedimented to the densest regions of isopycnic CsCl gradients , presumably because the high negative charge density of α-dystroglycan [21,22] causes it to form a complex with heavy cesium ions. Therefore the α-dystroglycan-enriched fractions obtained after serial DEAE-, WGA- and laminin-1 chromatography were subjected to CsCl gradient centrifugation. The recovered protein appeared as a single broad band with an apparent Mr of 156000 when stained with Coomassie Blue, radiolabelled on tyrosine residues with 125I, or when transferred on to nitrocellulose and stained with polyclonal antibodies to α-dystroglycan core protein or mAb IIH6 to its oligosaccharides (Figure 1A). The 156000 Mr protein also stained weakly with ConA (Figure 1A), Galanthus nivalis and Hippeastrum hybrid lectins (results not shown), which is consistent with the presence of high mannose N-linked oligosaccharides [23,24]. Blotted α-dystroglycan stained strongly when overlaid with purified laminin-1 (Figure 1A). When incubated at various concentrations in microtitre wells bearing immobilized laminin-1, the 125I-labelled α-dystroglycan displayed saturable binding with a half-maximal concentration of 7 nM (Figure 1B), which is in excellent agreement with the half-maximal concentration measured for 125I-laminin-1 binding to immobilized skeletal muscle α-dystroglycan (8 nM) previously reported by Pall and co-workers . Using an ELISA-based assay, we also confirmed that laminin-1 binding to purified skeletal muscle α-dystroglycan was inhibited by N-acetylneuraminic acid and heparin with measured IC50 values of 7 mM and 14 μM respectively (Figure 1C). α-Dystroglycan prepared by our modified method was also recently shown to bind recombinant laminin-2 in a calcium-dependent, but heparin-insensitive manner . We conclude that the α-dystroglycan prepared with the additional CsCl gradient centrifugation step was devoid of any detectable glycoprotein contaminants and retained well-established laminin binding properties.
Characterization of purified skeletal muscle α-dystroglycan
Heterogeneity in α-dystroglycan sialylation
We previously demonstrated that α-dystroglycan staining by WGA and Maackia amurensis agglutinin was completely ablated after digestion with Vibrio cholerae sialidase [10,11], but we later demonstrated that a subpopulation of skeletal muscle α-dystroglycan became reactive with the GalNAc-specific lectin Vicia villosa agglutinin after digestion with sialidase preparations from Clostridium perfringens or Aureafaciens ureafaciens, but not V. cholerae sialidase [17,26]. These results led us to conclude that skeletal muscle α-dystroglycan displays heterogeneity in its sialoglycosylation [17,26]. To further assess this possibility, we digested purified skeletal muscle α-dystroglycan with sialidases prepared from A. ureafaciens, C. perfringens or V. cholerae and performed blot overlays using peroxidase-conjugated WGA and LFA. WGA specifically binds N-acetylneuraminic acid, whereas LFA is more broadly reactive with both N-acetylneuraminic acid and N-glycolylneuraminic acid . As previously reported , WGA binding to α-dystroglycan was completely abrogated after digestion with A. ureafaciens, C. perfringens or V. cholerae sialidase (Figure 2). In contrast, LFA binding to α-dystroglycan was lost after digestion with A. ureafaciens or C. perfringens sialidases, but retained after digestion with V. cholerae sialidase (Figure 2). These results indicate that V. cholerae sialidase removes N-acetylneuraminic acid but not N-glycolylneuraminic acid moieties from α-dystroglycan and provides new evidence that skeletal muscle α-dystroglycan is heterogeneous with respect to its sialoglycosylation.
Heterogeneous sialoglycosylation of skeletal muscle α-dystroglycan
Enhanced laminin-1, mAb IIH6 and VIA41 binding to α-dystroglycan after enzymatic deglycosylation
We previously reported that enzymatic digestion with a combination of N-glycosidase F, V. cholerae sialidase and endo-O-glycosidase significantly altered the apparent Mr and lectin reactivity of skeletal muscle α-dystroglycan but had no effect on its laminin-1 binding activity . Given the identification of SiaAα2-3Galβ1-4GlcNAcβ1-2Man glycans on α-dystroglycan [14–16] and the results shown in Figure 2, it remained possible that our previous experiments  lacked the full complement of enzymes necessary to effect complete deglycosylation. Therefore we digested purified α-dystroglycan with a cocktail of five enzymes that together were expected to substantially degrade all N-linked and O-linked oligosaccharides known to modify skeletal muscle α-dystroglycan. The five enzymes were N-glycosidase F, A. ureafaciens sialidase, endo-O-glycosidase, β(1-4)galactosidase and β-N-acetylglucosaminidase, which are commercially available in kit form from several vendors (see the Materials and methods section). To simplify communication of the experiments, simultaneous digestion of α-dystroglycan with all five enzymes is operationally defined as ‘GLYCOPRO’ after the name of the commercial enzymatic deglycosylation kit used. Digestion with the GLYCOPRO kit effected an apparent Mr decrease of 40000 in skeletal muscle α-dystroglycan (Figure 3A, CB, α-DG pAb), which is the greatest shift that we have ever observed with any combination of glycosidases tested. However, GLYCOPRO-treated α-dystroglycan retained covalently bound carbohydrate based on digoxigenin-hydrazide incorporation after mild periodate oxidation (Figure 3A, Dig-Hz). Although GLYCOPRO digestion resulted in the complete loss of α-dystroglycan staining by sialic acid-reactive lectin LFA (Figure 3A), Stains-All binding to α-dystroglycan was only reduced 30% (Figure 3A). Because Stains-All reacts with anionic constituents of the protein, carbohydrate and nucleic acid, these results suggest that sialic acid accounted for only one-third of the acidic constituents present on α-dystroglycan. α-Dystroglycan also became more strongly reactive with ConA after GLYCOPRO digestion (Figure 3A), possibly due to exposure of O-linked mannosyl residues within SiaAα2-3Galβ1-4GlcNAcβ1-2Man glycans of α-dystroglycan.
Enzymatic deglycosylation of skeletal muscle α-dystroglycan
Most surprisingly, GLYCOPRO-digested α-dystroglycan consistently exhibited markedly enhanced binding to laminin-1 and mAbs VIA41 and IIH6 (Figure 3B), all of which depend on the presence of unidentified oligosaccharides on α-dystroglycan . The observed increase in α-dystroglycan binding to laminin-1, and mAbs VIA41 and IIH6 may be partly due to a more effective transfer of deglycosylated versus native α-dystroglycan because core protein-reactive polyclonal antibodies exhibited slightly increased reactivity for GLYCOPRO-digested α-dystroglycan (Figure 3A, α-DG pAb). However, the greatly enhanced laminin-1 binding activity after GLYCOPRO digestion strikingly conflicts with a previous study reporting the complete loss of laminin-1 binding to peripheral nerve α-dystroglycan after digestion with sialidase alone . The results of Figure 3 suggest that digestion of α-dystroglycan with the five GLYCOPRO enzymes led to removal of oligosaccharides that normally repress laminin and carbohydrate-specific mAb binding.
Validation of glycosidase specificity
To more specifically test whether the GLYCOPRO enzyme cocktail was active on the known oligosaccharides of α-dystroglycan, we digested α-dystroglycan with different combinations of GLYCOPRO enzymes and performed blot overlays with a panel of lectins. A. ureafaciens sialidase digestion rendered α-dystroglycan strongly reactive with PNA, which was lost after digestion with endo-O-glycosidase but not by β(1-4)galactosidase and β-N-acetylglucosaminidase (Figure 4A). These results confirmed the presence and specific removal of sialylated core 1 oligosaccharides (Figure 4B) previously identified on skeletal muscle α-dystroglycan [10,28]. Based on its demonstrated modification with SiaAα2-3Galβ1-4GlcNAcβ1-2Man glycans (Figure 4B), A. ureafaciens sialidase-digested α-dystroglycan was predicted to react with the lectin ECA, which specifically binds Galβ(1-4)GlcNAc but not SiaAα(2-3)Galβ(1-4)GlcNAc . A. ureafaciens sialidase digestion rendered α-dystroglycan strongly reactive with ECA (Figure 4A), confirming the exposure of Galβ(1-4)GlcNAc glycans. ECA staining was completely ablated by digestion with β(1-4)galactosidase and glucosaminidase but was unaffected by endo-O-glycosidase (Figure 4A). Interestingly, simultaneous digestion with A. ureafaciens sialidase, β(1-4)galactosidase and β-N-acetylglucosaminidase made α-dystroglycan strongly reactive with GsII lectin (Figure 4A), which is specific for terminal α- and β-linked GlcNAc . We further observed that digestion of α-dystroglycan with A. ureafaciens sialidase alone increased ConA reactivity, which was more dramatically enhanced by digestion with β(1-4)galactosidase and glucosaminidase (Figure 4A). Taken together, the experiments depicted in Figures 3 and 4 demonstrate that digestion of α-dystroglycan with the GLYCOPRO regime substantially degraded the terminal dissacharide of the sialyl O-mannosyl glycan thought to be important for laminin binding, yet α-dystroglycan binding to laminin-1 and mAbs IIH6 and VIA41 was paradoxically enhanced.
Skeletal muscle α-dystroglycan contains latent glyco-epitopes reactive with conA and GsII
To assess whether other known carbohydrate modifications may block enzymatic removal of the established O-linked core structures present on α-dystroglycan, we evaluated the effect of 11 additional enzymes on the laminin-1, antibody and lectin binding activities of α-dystroglycan either alone or in combination with the five GLYCOPRO enzymes (see Supplementary Table 1 at http://www.BiochemJ.org/bj/390/bj3900303add.htm). The enzymes targeted α-linked Gal and GalNAc, α-linked fucose, β-linked glucuronic acid, α- and β-linked Man, the naturally occurring sialic acid analogue 2-keto-3-deoxy-D-glycero-D-galactonononic acid as well as sulphation and phosphorylation. However, none of the enzymes tested had any effect on the laminin-1, antibody or lectin binding activities of α-dystroglycan.
Evidence for modification of α-dystroglycan by α-linked GlcNAc
Gastric mucins are uniquely modified by ConA-reactive glyco-epitopes that resist, or are even enhanced by periodate oxidation . This so-called paradoxical ConA staining has subsequently been shown to be due to terminal α-linked GlcNAc [31,32], a carbohydrate linkage recognized by both ConA and GsII  and for which no specific exoglycosidase is currently available. Since skeletal muscle α-dystroglycan exhibited latent reactivity with both ConA and GsII that was resistant to digestion with a wide variety of enzymes (Figures 3 and 4, Supplementary Table 1), we examined the effect of periodate oxidation on α-dystroglycan ConA reactivity (Figure 5). We observed that prolonged (2–16 h) exposure of α-dystroglycan to 10 mM periodate at room temperature caused a shift in apparent Mr from 156000 to 59000 with complete abrogation of IIH6 and laminin-1 binding activity (Figure 5A). We also observed that prolonged periodate oxidation resulted in the complete loss of Stains-All binding to α-dystroglycan (Figure 5B), which suggests that anionic carbohydrate is required for Stains-All binding to α-dystroglycan. After 30 min of periodate oxidation, we detected an apparent intermediate of 88000 Mr that retained strong IIH6 and laminin-1 binding activity. Most interestingly, ConA reactivity was retained in both the 59000 and 88000 Mr species up to 60 min of periodate oxidation, reminiscent of the time course for paradoxical ConA staining in gastric mucin .
Periodate oxidation of α-dystroglycan
The presence of two novel O-mannosyl glycan structures on α-dystroglycan has been well documented using established analytical methods [14–16]. O-Mannosyl oligosaccharides of unknown structure are also clearly important for α-dystroglycan function in vivo because mutations in genes encoding demonstrated or putative glycosyltransferases are associated with hypoglycosylated α-dystroglycan, loss of its extracellular ligand binding activities and muscular dystrophy [3,6]. However, the specific importance of SiaAα2-3Galβ1-4GlcNAcβ1-2Man in the extracellular ligand binding activity of α-dystroglycan rests tenuously on two dubious experiments. In the first experiment, sialic acid  and α2-3 sialyl N-acetyllactosamine  were shown to inhibit laminin binding to peripheral nerve α-dystroglycan with IC50 values of approx. 6 mM and >10 mM respectively. Using a similar assay, we confirmed that sialic acid inhibited laminin binding to α-dystroglycan with a similar IC50 value of 7 mM. However, the potential significance of this result is tempered when compared directly with heparin (Figure 1), which inhibited binding with a 500-fold lower IC50. It was also reported that digestion of α-dystroglycan with sialidase resulted in the complete loss of laminin binding activity . Unfortunately, this second experiment lacked controls proving that the loss of laminin binding was not due to α-dystroglycan core protein degradation caused by a protease contaminant within the sialidase preparation. Moreover, neither we [11,17] nor several other groups [12,22,33] observe a loss of extracellular matrix ligand binding activity when α-dystroglycan from several different tissues was exhaustively digested with a wide variety of sialidase preparations.
In the present study, we observed substantial retention of laminin binding activity in an 88000 Mr α-dystroglycan intermediate detected after incubation with 10 mM periodate for 30 min at room temperature (Figure 5). In contrast, sialic acid is readily oxidized within 20 min by 1 mM periodate at 0 °C. Most importantly, we enzymatically removed all detectable sialic acid and galactose moieties from SiaAα2-3Galβ1-4GlcNAcβ1-2Man glycans on α-dystroglycan, but observed increased laminin-1, IIH6 and VIA41 binding activities (Figures 3 and 4). Finally, Matsumura and co-workers  recently identified a distal myopathy patient with mutations in the UDP-N-acetylglucosamine 2-epimerase/N-acetylmannosamine kinase gene. Although this patient exhibited gross hyposialylation of skeletal muscle glycoproteins, the laminin binding activity of α-dystroglycan was not impaired. Taken together, these results suggest that the terminal SiaAα2-3Gal moieties on O-mannosyl glycans of skeletal muscle α-dystroglycan are not essential determinants for laminin binding activity and that their removal leads to enhanced binding activity for laminin-1, and carbohydrate-specific mAbs IIH6 and VIA41.
Our results further indicate that α-dystroglycan is modified with anionic sugars distinct from sialic acid because its Stains-All reactivity was only reduced by 30% after sialidase digestion (Figure 3) but completely abolished after periodate oxidation (Figure 5). Consistent with our results, sialidase-resistant acidic oligosaccharides were previously isolated from peripheral nerve α-dystroglycan, but have not been further characterized (see Figure 6 of ). These anionic sugars may also provide key acidic functional groups necessary for calcium co-ordinated binding of α-dystroglycan to laminin . Finally, our results suggest that oligosaccharides on α-dystroglycan may contain α-linked GlcNAc substituents, which could explain its resistance to complete enzymatic deglycosylation. Digestion of α-dystroglycan with sialidase and β(1-4)galactosidase revealed a very robust enhancement of ConA staining and latent GsII reactivity (Figure 4B). Certainly, a portion of GsII reactivity in sialidase and β(1-4)galactosidase digested α-dystroglycan may be due to undigested GlcNAcβ1-2Man cores within SiaAα2-3Galβ1-4GlcNAcβ1-2Man glycans. However, the strong ConA and GsII staining remaining after GLYCOPRO digestion could not be eliminated by a wide array of enzymes tested (Supplementary Table 1). To our knowledge, the only carbohydrate structure recognized by both ConA and GsII is terminal α-linked GlcNAc . Furthermore, the ConA reactivity of α-dystroglycan was also observed to withstand prolonged periodate oxidation (Figure 5), which is a notable feature of the α-linked GlcNAc terminated ‘paradoxical’ ConA epitope so far only detected on gastric mucins [30–32]. Interestingly, mutations in the LARGE gene lead to hypoglycosylated α-dystroglycan, loss of its extracellular ligand binding activities and muscular dystrophy [3,6], whereas overexpression of LARGE rescued these defects . Although the biochemical function of the LARGE gene product is not certain, sequence analysis identified two domains homologous with bacterial α-glycosyltransferase and mammalian N-acetylglucosaminyltransferase respectively [36,37]. Unfortunately, currently there exists no purified glycosidase that can specifically remove α-linked GlcNAc [31,32] so its confirmation in α-dystroglycan awaits further investigation using biosynthetic approaches [7–9] and/or MS. With respect to the latter method, our highly purified and extensively characterized preparation of skeletal muscle α-dystroglycan will provide an ideal substrate for analysis.
We thank Dr H. Kleinman for purified laminin-1 and Dr H. Miller (Selectin Biosciences, Pleasant Hill, CA, U.S.A.) for helpful advice. This work was supported by the Muscular Dystrophy Association and National Institutes of Health grant ARO1985 to J.M.E.