Differential Heparin Sensitivity of α-Dystroglycan Binding to Laminins Expressed in Normal and dy/dy Mouse Skeletal Muscle*

The α-dystroglycan binding properties of laminins extracted from fully differentiated skeletal muscle were characterized. We observed that the laminins expressed predominantly in normal adult rat or mouse skeletal muscle bound α-dystroglycan in a Ca2+-dependent, ionic strength-sensitive, but heparin-insensitive manner as we had observed previously with purified placental merosin (Pall, E. A., Bolton, K. M., and Ervasti, J. M. 1996 J. Biol. Chem. 271, 3817–3821). Rat skeletal muscle laminins partially purified by heparin-agarose affinity chromatography also bound α-dystroglycan without sensitivity to heparin. We also confirm previous studies of dystrophicdy/dy mouse skeletal muscle showing that the α2 chain of merosin is reduced markedly and that the laminin α1 chain is not up-regulated detectably. However, we further observed a quantitative decrease in the expression of laminin β/γ chain immunoreactivity in α2 chain-deficient dy/dy skeletal muscle and reduced α-dystroglycan binding activity in laminin extracts fromdy/dy muscle. Most interestingly, the α-dystroglycan binding activity of residual laminins expressed in merosin-deficientdy/dy skeletal muscle was inhibited dramatically (69 ± 19%) by heparin. These results identify a potentially important biochemical difference between the laminins expressed in normal anddy/dy skeletal muscle which may provide a molecular basis for the inability of other laminin variants to compensate fully for the deficiency of merosin in some forms of muscular dystrophy.

The laminins are a family of abundant and essential basement membrane proteins implicated in numerous processes important to normal tissue development and homeostasis (1,2). The prototypical laminin is a large, disulfide-linked heterotrimer consisting of ␣, ␤, and ␥ chains (1,2). Multiple isoforms for each chain have been identified, giving rise to numerous heterotrimer combinations with great potential for functional diversity/specificity (1,2). In normal adult skeletal muscle, for example, laminin-2 (␣2␤1␥1), colloquially known as merosin, is expressed predominantly throughout the basement membrane (3,4). In contrast, ␤2-containing laminin variants are localized specifically to neuromuscular and myotendinous junctions (3)(4)(5), suggesting that they play unique roles at these important membrane specializations. In fact, neuromuscular synaptogenesis is impaired severely in mice with a targeted disruption of the ␤2 chain gene (6). Evidence for functional specificity of laminin variants has also emerged from the characterization of muscular dystrophies in which the ␣2 chain of merosin is deficient (7)(8)(9)(10)(11). Although classical laminin-1 (␣1␤1␥1) supports many processes in myoblast differentiation (for review, see Refs. 12 and 13), the up-regulation of a laminin species originally presumed to be laminin-1 (see below) fails to correct the pathologies observed with merosin deficiency (8 -11), suggesting that the ␣2 chain of merosin possesses a specific functional activity. In support of this hypothesis, Engvall and coworkers demonstrated that merosin, but not laminin-1, could specifically stabilize and prolong the survival of differentiated myotubes (13).
Regarding the cell surface receptor(s) responsible for communicating a specific survival signal from merosin to muscle cells, the laminin-binding integrin ␣7␤1 has been examined the most extensively. In its favor, ␣7␤1 integrin displays musclespecific and differentiation-dependent alternative splicing, which enables at least one ␣7 isoform to discriminate biochemically between laminin-1 and merosin (14,15). Furthermore, ␣7␤1 integrin isoforms exhibit abnormal expression and cellular distribution in merosin-deficient muscle (16,17), and treatment of normal merosin-expressing myotubes with ␤1 integrin function-blocking antibodies impaired myotube survival significantly (16). However, targeted deletion of the ␣7 integrin gene results in a form of muscular dystrophy which is much less severe than the phenotype observed in merosin-deficient forms of muscular dystrophy (18). These data argue for the existence of at least one additional merosin-cell surface receptor interaction, distinct from ␣7 integrin, which plays an important role in supporting muscle cell survival.
␣-Dystroglycan is a ubiquitous, membrane-associated, extracellular glycoprotein (19) originally identified as a subunit of the skeletal muscle dystrophin-glycoprotein complex (20 -22). Biochemical evidence suggests that the dystrophin-glycoprotein complex serves to span the sarcolemmal membrane and link the cortical cytoskeleton with the extracellular matrix through the interaction of dystrophin with F-actin (23,24) and the binding of ␣-dystroglycan to merosin (7,25). Analysis of dystrophin-deficient mdx mouse muscle by a variety of approaches (26 -28) further suggests that one function of the transmembrane linkage formed by the dystrophin-glycoprotein complex is to provide mechanical reinforcement to the sarcolemmal membrane during muscle contraction or stretch. However, the dystrophin-glycoprotein complex appears to be unaffected in merosin-deficient skeletal muscle (16,29). Furthermore, a recent study demonstrated that merosin deficiency does not compromise sarcolemmal membrane integrity as is observed in dystrophin-deficient muscle (30). Therefore, it remains to be determined what cellular function(s) is supported by the interaction between ␣-dystroglycan and merosin.
In characterizing the laminin binding properties of purified skeletal muscle ␣-dystroglycan, we have identified a potentially important biochemical difference between purified laminin-1 and merosin (31,32). Previous studies demonstrated that the ␣-dystroglycan binding site in laminin-1 resides in the fourth and fifth repeats of the globular G domain (33,34) and that mouse laminin-1 binding to ␣-dystroglycan was inhibited dramatically by heparin (25,33). Recently, we observed that heparin only marginally inhibited ␣2 chain-containing human placental merosin binding to ␣-dystroglycan (31). Because heparin mimics many of the biological activities of abundantly expressed basement membrane heparan sulfate proteoglycans (35), we hypothesized that the differential heparin sensitivity of ␣-dystroglycan binding to laminin variants may identify a mechanism for modulating the binding of ␣-dystroglycan to different extracellular ligands specifically (31). Moreover, our results argue that like ␣7␤1 integrin isoforms (14,15), ␣-dystroglycan is capable of discriminating between laminin-1 and merosin, thereby suggesting that ␣-dystroglycan may be a second candidate receptor for mediating a merosin-specific effect on muscle cell stability. Although our data could also help to explain why the apparent up-regulation of laminin-1 fails to compensate for the absence of merosin in some forms of muscular dystrophy (7)(8)(9)(10)(11), the specificity of the monoclonal antibody used to detect ␣1 chain expression in merosin-deficient muscle (9 -11) has been reconsidered (36,37). In the most recent studies, the ␣1 chain of laminin-1 was neither detected in normal adult skeletal muscle (4,37,38) nor found to be up-regulated in response to merosin deficiency (4,38). One of these studies further noted an up-regulation of ␣4 chain in dy/dy mouse skeletal muscle (4). However, it remains unclear whether the laminin species expressed in dy/dy muscle can bind ␣-dystroglycan with properties that might explain its inability to replace one of the functions normally served by merosin.
To address this issue, we have characterized the ␣-dystroglycan binding properties of total laminins extracted from adult rat and mouse skeletal muscle using a blot overlay assay. We find that the laminin species expressed predominantly in normal adult rat or mouse skeletal muscle bind ␣-dystroglycan in a heparin-insensitive manner as we had observed previously with purified placental merosin (31,32). Rat skeletal muscle laminin partially purified by heparin affinity chromatography also bound ␣-dystroglycan without sensitivity to heparin. We demonstrate further that in addition to the previously observed reduction in ␣2 chain immunoreactivity (4,7,8,38), laminin extracts prepared from dy/dy skeletal muscle exhibit a significant reduction in laminin ␤/␥ chain immunoreactivity. Most interestingly, we observe that the residual laminins expressed in ␣2 chain-deficient dy/dy mouse skeletal muscle bind ␣-dystroglycan, but the interaction is inhibited dramatically by heparin or heparan sulfate. Our results suggest that basement membrane heparan sulfate proteoglycans may catastrophically perturb ␣-dystroglycan binding to the laminin variants present in merosin-deficient muscle and support our hypothesis for a specific and functionally important interaction between ␣-dystroglycan and merosin in normal adult muscle.
Extraction of Laminins from Skeletal Muscle-Laminins from skeletal muscle of 2-3-month-old female Sprague-Dawley rats (Harlan Sprague-Dawley, Madison, WI) were extracted with EDTA essentially as described by Paulsson and Saladin (40). Rat skeletal muscle was homogenized in 20 ml/g Tris-buffered saline (50 mM Tris-HCl, pH 7.4, 150 mM NaCl) containing 100 g/ml benzamidine and 40 g/ml phenylmethylsulfonyl fluoride and centrifuged at 10,000 ϫ g for 20 min. The resulting pellet was rehomogenized in 70% original buffer volume and the centrifugation step repeated. Washed pellets were collected and resuspended to 2.5 ml/g Tris-buffered saline containing 100 g/ml benzamidine, 40 g/ml phenylmethylsulfonyl fluoride, and 10 mM EDTA. After incubation at 4°C for 2 h with stirring, the suspension was centrifuged at 10,000 ϫ g, the supernatant retained, and the pellet reextracted with EDTA. For gel and Western blot analysis, the EDTA extracts were pooled, brought to 40% saturation with solid ammonium sulfate, and centrifuged at 10,000 ϫ g. The ammonium sulfate pellet was resuspended in Tris-buffered saline and dialyzed exhaustively against Tris-buffered saline. When used in overlay binding experiments, pooled EDTA extracts were centrifuged at 100,000 ϫ g, and the resulting supernatant was concentrated and dialyzed exhaustively against Tris-buffered saline. EDTA extracts were prepared from skeletal muscle of 4 -7-week-old C57BL/6J-Lama2 dy (dy/dy) and control littermate mice (Jackson Laboratory, Bar Harbor, ME) in the same manner. The protein concentrations of the EDTA extracts were determined with the Bio-Rad DC protein assay using bovine serum albumin as standard.
SDS-Polyacrylamide Gel Electrophoresis and Blotting Assays-All samples were separated electrophoretically on 3-12% SDS-polyacrylamide gels in the presence of 1% ␤-mercaptoethanol and either stained with Coomassie Blue or transferred to nitrocellulose membranes as described previously (22). Laminin-reactive antibodies used in this study were an affinity-purified rabbit polyclonal antibody raised against purified mouse laminin-1 (Sigma) and rabbit antiserum (41) raised against a recombinant protein corresponding to amino acids 1575-1974 of the mouse laminin ␣2 chain (the kind gift of Dr. Yoshihiko Yamada). Laminin antibody staining was detected with a peroxidaseconjugated anti-rabbit secondary antibody (Boehringer Mannheim) and chemiluminescence using SuperSignal CL-HRP (Pierce) as substrate (31) or with 125 I-protein A (29). Coomassie Blue-stained gels, chemiluminescence films, and autoradiograms were analyzed densitometrically with a Bio-Rad model GS-670 imaging densitometer. The intensities of scanned bands were quantitated by volume integration after background subtraction (31).
To monitor specifically ␣-dystroglycan binding by rat and mouse skeletal muscle laminins in EDTA extracts, nitrocellulose transfers containing purified rabbit skeletal muscle ␣-dystroglycan were blocked in phosphate-buffered saline (8 mM sodium phosphate monobasic, 42 mM sodium phosphate dibasic, pH 7.4, 150 mM NaCl) containing 5% nonfat dry milk for 1 h at room temperature. Blocked transfers were rinsed briefly with Tris-buffered saline and incubated for 2 h at room temperature in Tris-buffered saline containing 3% bovine serum albumin, 1 mM CaCl 2 , 1 mM MgCl 2 , and either rat, control mouse (both typically used at a 1:10 dilution), or dy/dy mouse (used at a 1:5 dilution) skeletal muscle laminin EDTA extracts. Heparin and chondroitin sulfates A and C were purchased from Sigma; heparan sulfate was obtained from Seikagaku America (Ijamsville, MD). Laminin binding to ␣-dystroglycan was monitored with the affinity-purified polyclonal antibodies raised against laminin-1 (Sigma) as described previously (31,32).
Heparin-Agarose Chromatography-EDTA extract prepared from 20 g of rat skeletal muscle was loaded onto a 25-ml heparin-agarose column (Sigma) that had been preequilibrated with Tris-buffered saline containing 100 g/ml benzamidine and 40 g/ml phenylmethylsulfonyl fluoride. After extensive washing with Tris-buffered saline containing 100 g/ml benzamidine and 40 g/ml phenylmethylsulfonyl fluoride, bound proteins were eluted with 50 mM Tris-HCl, pH 7.4, 250 mM NaCl (42). The 250 mM NaCl eluate was concentrated in a Centriplus 100 (Amicon), dialyzed exhaustively against Tris-buffered saline, and examined for ␣-dystroglycan binding activity as described above.

RESULTS AND DISCUSSION
Homogenization in Tris-buffered saline containing EDTA was shown previously to extract laminins from a variety of adult mouse tissues (40). Therefore, we followed a similar protocol to extract the laminins expressed in adult rat skeletal muscle. Densitometric analysis of the resulting rat skeletal muscle EDTA extract resolved on Coomassie Blue-stained SDS-polyacrylamide gels indicated that it consisted of greater than 20 discrete protein bands (Fig. 1). An identical nitrocellulose transfer stained with polyclonal antiserum raised against a recombinant protein corresponding to a portion of the mouse ␣2 chain revealed the presence of a 300,000 M r band in the skeletal muscle homogenate and EDTA extracts (Fig. 1). The ␣2 antiserum also stained bands with apparent M r of 540,000 and 700,000 (Fig. 1). A 220,000 M r band was predominantly detected in the skeletal muscle homogenate and EDTA extracts when another nitrocellulose transfer was stained with affinity-purified rabbit polyclonal antibodies raised against laminin-1 (Fig. 1). These antibodies also detected additional bands with an apparent M r of 400,000, 540,000, and 700,000 ( Fig. 1). Because the laminin-1 polyclonal antibodies reacted equally well with ␣1, ␤, and ␥ chains of purified mouse laminin-1 (not shown), the prominent 220,000 M r band is likely the ␤/␥ chains expressed in rat skeletal muscle, and the 400,000 M r band is presumably an ␣ chain. The 540,000 and 700,000 M r bands detected with both ␣2 antiserum and laminin-1 polyclonal antibodies were likely laminin heterodimers/trimers because their staining intensity decreased substantially when samples were preincubated for 5 min at 100°C in the presence of 5% ␤-mercaptoethanol before electrophoresis (not shown). These results indicated that the previously described EDTA extraction protocol (40) efficiently solubilized the laminins expressed in rat skeletal muscle.
To characterize the ␣-dystroglycan binding properties of laminins present in the rat skeletal muscle EDTA extracts, we performed a modified blot overlay assay (31,43) in which nitrocellulose transfers containing purified rabbit skeletal muscle ␣-dystroglycan were incubated with various dilutions of skeletal muscle extract, washed extensively, and laminin binding specifically detected with affinity-purified rabbit polyclonal antibodies raised against intact laminin-1 ( Fig. 2A). Laminins in the rat skeletal muscle EDTA extract bound strongly to purified ␣-dystroglycan at a dilution of 1:10 ( Fig. 2A), and binding was clearly detectable at dilutions as great as 1:100 (not shown). As observed previously with purified laminin-1 and merosin (25,31), ␣-dystroglycan binding by laminins in the rat skeletal muscle EDTA extract was dependent on Ca 2ϩ and sensitive to ionic strength ( Fig. 2A). However, inclusion of 1 mg/ml heparin in the binding medium had no effect on the rat laminin extract binding to ␣-dystroglycan ( Fig. 2A). These data suggest that the predominant laminin expressed in adult rat skeletal muscle binds ␣-dystroglycan in a heparin-insensitive manner as was shown previously for purified merosin (31,32). Our data are also consistent with previous studies indicating that merosin is the predominant laminin expressed in normal adult striated muscle (3)(4)(5).
The heparin binding sites of mouse laminin-1 and human placental merosin (a mixture of predominantly laminin-2 and -4) have been characterized extensively (42, 44 -53). With relevance to muscle forms of laminin, a number of studies have shown that ␣2 chain-containing laminins can bind heparin (42, 49 -53). However, one study (50) observed that a significant fraction of placental merosin exhibited no apparent heparin binding activity. Therefore, it was necessary to determine whether the heparin insensitivity of rat skeletal muscle laminin binding to ␣-dystroglycan was caused by a lack of heparin binding activity. Rat skeletal muscle EDTA extracts were applied to a heparin-agarose column, the column was washed extensively with Tris-buffered saline (0.15 M NaCl), and the proteins eluted with 0.25 M NaCl. These heparin-binding laminins were evaluated subsequently for ␣-dystroglycan binding in the absence and presence of heparin (Fig. 2B). Consistent with our findings using the total rat skeletal muscle EDTA extract, the laminins eluted from heparin-agarose bound ␣-dystroglycan equally well either in the absence or presence of 1 mg/ml heparin (Fig. 2B). These results demonstrate that rat skeletal muscle laminins bind heparin but that their binding to ␣-dystroglycan is insensitive to heparin.
EDTA extracts were also prepared from skeletal muscle of 4 -6-week-old normal control and littermate dy/dy mice. Consistent with a previous study of dy/dy muscle using a similar extraction protocol (7), the total protein compositions of EDTA extracts from control and dy/dy mouse skeletal muscle were FIG. 1. Extraction of laminins from rat skeletal muscle. Shown in the left panel is a Coomassie Blue-stained SDS-polyacrylamide gel containing equal volumes of rat skeletal muscle homogenate (HOMOG), the supernatants obtained from homogenization in Tris-buffered saline (W1 and W2), the supernatants obtained from sequential EDTA extractions (EDTA1 and EDTA2), and the proteins recovered from the pooled EDTA extracts after precipitation with ammonium sulfate (AmSO 4 ). Shown in the middle and right panels are identical SDS-polyacrylamide gels transferred to nitrocellulose and stained with either an affinitypurified polyclonal antibody (pAb) against laminin-1 (LAM) or polyclonal antiserum raised against a recombinant protein corresponding to a portion of the mouse ␣2 chain (␣2). Identified are the apparent M r of bands detected with mouse ␣2 chain antibodies or laminin-1 polyclonal antibodies; the asterisk indicates a band also stained by preimmune serum. Molecular weight markers (ϫ10 Ϫ3 ) are shown on the left. similar as assessed on Coomassie Blue-stained SDS-polyacrylamide gels (Fig. 3). Furthermore, the predominant 300,000 M r band and minor 540,000 and 700,000 M r species detected in control mouse EDTA extracts with the ␣2 chain antiserum were reduced markedly in EDTA extracts prepared from dy/dy mouse skeletal muscle when detected with 125 I-protein A (Fig.  3) or by chemiluminescence (not shown). Densitometric analysis of three additional control and dy/dy EDTA extracts stained with ␣2 polyclonal antiserum and detected with 125 Iprotein A indicated that the average intensity of the 300,000 M r ␣2 band was 24 Ϯ 4% of that measured in the control EDTA extracts. These results confirm previous findings (4,7,8,38) that the ␣2 chain of merosin is reduced markedly in skeletal muscle from dy/dy mice.
Western blot analysis of control and dy/dy EDTA extracts with the laminin-1 polyclonal antibody yielded more complicated but interesting results. As observed with rat skeletal muscle EDTA extracts, 220,000, 400,000, 540,000, and 700,000 M r bands were observed in control mouse EDTA extracts when detected by chemiluminescence (Fig. 3) or on overexposed autoradiograms when detected with 125 I-protein A (not shown). Using chemiluminescence detection, the laminin-1 polyclonal antibody revealed a prominent 220,000 M r band in dy/dy extracts with an intensity similar to that observed in control muscle extracts (Fig. 3). However, the staining intensities of the higher M r species reactive with laminin-1 polyclonal antibody were decreased markedly in dy/dy extracts (Fig. 3). When detected by 125 I-protein A, the 220,000 and all higher M r bands reactive with the laminin-1 polyclonal antibody were all reduced markedly in dy/dy extracts compared with control extracts (Fig. 3). The difference in relative ␤/␥ immunoreactivity obtained for control and dy/dy extracts by chemiluminescence versus 125 I-protein A can be attributed to saturation of the film response caused by the extreme sensitivity of chemiluminescence reagents. Densitometric analysis of autoradiograms from three additional control and dy/dy EDTA extracts stained with laminin-1 polyclonal antibodies and detected with 125 I-protein A indicated that the average intensity of the predominant 220,000 M r ␤/␥ band was 52 Ϯ 18% of that measured in the control EDTA extracts. These results thus demonstrate a quantitative decrease in the expression of laminin ␤/␥ chains in ␣2 chain-deficient dy/dy skeletal muscle.
To characterize the ␣-dystroglycan binding properties of laminins expressed in merosin-deficient skeletal muscle, nitrocellulose transfers containing purified rabbit skeletal muscle ␣-dystroglycan were incubated with EDTA extracts prepared from control or dy/dy mouse skeletal muscle and then probed for laminin binding with affinity-purified polyclonal antibodies to laminin-1. The intensity of laminin binding observed with dy/dy EDTA extracts was consistently decreased compared with that observed with control muscle EDTA extracts (Fig. 4). For example, the average densitometric intensity for three independent experiments using different dy/dy EDTA extracts was only 62% of the signal measured with control extracts when dy/dy EDTA extracts were used at 2-fold higher concentrations (Fig. 4). These results are consistent with our immunoblot data (Fig. 3) indicating that the total laminin content of dy/dy skeletal muscle is decreased relative to normal muscle. ␣-Dystroglycan binding by laminins in both control and dy/dy EDTA extracts required Ca 2ϩ and was inhibited substantially by inclusion of 500 mM NaCl to the overlay medium (not shown). However, the ␣-dystroglycan binding activity of laminins present in dy/dy EDTA extracts exhibited a dramatic sensitivity to heparin which was not apparent with control muscle EDTA extracts (Fig. 4A). Heparin was also effective in displacing dy/dy muscle laminin that had been prebound to immobilized ␣-dystroglycan in the absence of heparin (not shown). As demonstrated previously for laminin-1 binding to ␣-dystroglycan (31), heparan sulfate also inhibited ␣-dystroglycan binding by dy/dy muscle laminins dramatically, whereas equivalent amounts of chondroitin sulfate A or chondroitin sulfate C were without effect (Fig. 4B). Densitometric analysis of three independent experiments performed with EDTA extracts prepared from different mice indicated that the average ␣-dystroglycan binding activity of dy/dy muscle laminins in the presence of 1 mg/ml heparin was only 31 Ϯ 19% of the   FIG. 3. Analysis of EDTA extracts prepared from normal control and dy/dy mouse skeletal muscle. Shown is a Coomassie Bluestained SDS-polyacrylamide gel (CB) containing equal amounts of EDTA extracts prepared from normal control and dy/dy skeletal muscle. Also shown are identical nitrocellulose transfers stained with affinity-purified polyclonal antibody (pAb) against laminin-1 (LAM) or polyclonal antiserum raised against a recombinant protein corresponding to a portion of the mouse ␣2 chain (␣2). Bands stained by the laminin-1 polyclonal antibody were detected either with peroxidase-conjugated secondary antibody and chemiluminescence (HRP/CL) or with 125 Iprotein A ( 125 I-PrA). The ␣2-reactive bands were detected with 125 Iprotein A. The apparent M r of bands detected with laminin-1 polyclonal antibodies or mouse ␣2 chain antibodies are indicated. Molecular weight markers (ϫ10 Ϫ3 ) are shown on the left. binding measured in the absence of heparin. In contrast, the average ␣-dystroglycan binding activity of control muscle laminins in the presence of 1 mg/ml heparin was 97 Ϯ 24% of the binding observed in the absence of heparin. ␣-Dystroglycan binding by dy/dy muscle laminins was inhibited over a wide range of heparin concentrations with an IC 50 near 0.5 mg/ml, whereas heparin concentrations as high as 50 mg/ml failed to inhibit ␣-dystroglycan binding by laminins in control EDTA extracts (Fig. 5). These data suggest that the differential heparin sensitivity of ␣-dystroglycan binding by laminins expressed in control and dy/dy skeletal muscle was not caused by a quantitative difference in their heparin binding affinities.
We have demonstrated that ␣-dystroglycan binding properties of laminins expressed predominantly in normal adult rat and mouse skeletal muscle are consistent with those observed previously using merosin purified from human placenta (31,32). In particular, laminins from normal rat and mouse skeletal muscle bound ␣-dystroglycan in a heparin-insensitive manner (Figs. 2 and 4). Our present results address potential concerns that the heparin insensitivity of ␣-dystroglycan binding to commercial preparations of merosin was caused by protein degradation/denaturation or that differential heparin sensitivity was simply a reflection of species variation between human merosin and mouse laminin-1.
We also demonstrated the existence of a heparin-binding population of rat skeletal muscle laminin which binds ␣-dystroglycan in a heparin-insensitive manner (Fig. 2B). These data suggest that either the heparin or ␣-dystroglycan binding sites located in G domain repeats G4 -G5 may differ between the ␣1 and ␣2 chains of laminin. Alternatively, the observed heparin inhibition of ␣-dystroglycan binding to laminin-1 (25, 31-33) may not be caused by simple competition between overlapping heparin and ␣-dystroglycan binding sites within the G4 -G5 domain of the ␣1 chain. It is possible that heparin binds to a remote site on laminin-1, perhaps the cryptic site in G1-G3 (46) or the site located in domain VI of the ␣ chain short arm (48), and induces a long distance conformational change in G4 -G5 to inactivate its ␣-dystroglycan binding site. This remote heparin binding site would presumably be absent from merosin or its communication with G4 -G5 somehow disrupted. In any event, the heparin insensitivity of rat skeletal muscle merosin binding to ␣-dystroglycan does not appear to be the result of a lack of heparin binding activity.
With regard to whether the laminin ␣1 chain is expressed in normal skeletal muscle or up-regulated in response to merosin deficiency, a previous study (8) detected the presence of a 400,000 M r band in both control and dy/dy muscle extracts using a polyspecific antiserum to human placental laminin. We also detected a minor 400,000 M r band in rat and mouse skel-etal muscle using a polyclonal antibody reactive with mouse ␣1 chain; however, the intensity of this band was actually decreased in dy/dy muscle extracts (Figs. 1 and 3). Although the polyclonal antibody used in the present study was affinity purified against mouse laminin-1, we cannot exclude the possibility that it may cross-react with other laminin ␣ chains. Recent studies using ␣1 chain-specific antibodies (4,37,38) were unable to detect any ␣1 chain immunoreactivity in either normal or dy/dy mouse muscle, suggesting the up-regulation of a different laminin variant in dy/dy skeletal muscle. Sanes and colleagues (4) further demonstrated an increased immunoreactivity for ␣4 laminin chain in merosin-deficient dy/dy mouse muscle, suggesting that laminin-8 and not laminin-1 is upregulated in dy/dy muscle. However, we have shown that like purified laminin-1, the laminins expressed in dy/dy muscle bind poorly to ␣-dystroglycan in the presence of heparin (Fig.  4). The implications of these novel results are discussed more fully below.
We also observed a quantitative decrease in the expression of laminin ␤/␥ chain immunoreactivity in ␣2 chain-deficient dy/dy skeletal muscle (Fig. 3). Interestingly, Yurchenco and co-workers (54) recently demonstrated that laminin ␤/␥ chains are not secreted in the absence of ␣1 chain secretion. Assuming that a similar mechanism holds true for other laminin ␣ chains and that all functional laminins exist as ␣␤␥ trimers, our observed decrease in ␤/␥ immunoreactivity (Fig. 3) suggests that the total laminin content of dy/dy mouse skeletal muscle is reduced relative to normal muscle. This conclusion is supported further by the reduced ␣-dystroglycan binding activity observed in laminin extracts from dy/dy muscle (Fig. 4) and suggests that the amount of ␣4 chain expressed in dy/dy muscle may be insufficient to compensate for ␣2 chain deficiency in dy/dy muscle. It must be noted that our observed decrease in ␤/␥ immunoreactivity (Fig. 3) conflicts with some (7,8,38) but not all (4,55) previous studies that examined the expression of ␤/␥ chains in dy/dy mouse muscle. All but one (8) of the previous studies assessed ␤/␥ chain expression by indirect immunofluorecence, which may not detect the 50% decrease in ␤/␥ chain reactivity we measured by quantitative immunoblot analysis. In the only other study to utilize immunoblot analysis, chemiluminescence methods were used to detect similar levels of ␤/␥ immunoreactivity in control and dy/dy skeletal muscle (8). Although we also observed similar ␤/␥ immunoreactivity in control and dy/dy muscle extracts using chemiluminescence detection, we have concluded that its extreme sensitivity can result in saturation of the immune signal in control samples and obscure quantitative differences in immunoreactivity which are apparent by detection with 125 I-protein A (Fig.  3). Alternatively, this difference in results may be caused by the different dy mouse strains evaluated because Engvall and colleagues (8) analyzed 129B6F1/J-Lama2 dy and 129/ReJ-Lama2 dy mice, whereas we examined the C57BL/6J-Lama2 dy strain. In support of this possibility, there are several muscular dystrophies (56 -58) in which defects in the same gene yield different pleitropic effects.
Finally, our results have implications for the potential of other laminin species to replace fully the function(s) normally served by merosin, which has been considered as a therapeutic approach for merosin-deficient muscular dystrophies (4,37). Whereas laminin-8 (␣4␤1␥1) is up-regulated in merosin-deficient dy/dy mouse muscle (4), the mutant phenotype is severe nonetheless. We have demonstrated here that ␣-dystroglycan binding to the laminins expressed in dy/dy skeletal muscle is inhibited dramatically by heparin (Fig. 4). Our results suggest that abundantly expressed basement membrane heparan sulfate proteoglycans could potentially perturb ␣-dystroglycan FIG. 5. Heparin sensitivity of ␣-dystroglycan binding by laminin extracts from control and dy/dy mouse skeletal muscle. ␣-Dystroglycan binding by control (q) and dy/dy (f) mouse laminin extracts is plotted as a function of heparin concentration with data points expressed as the average percentage (n Ն 2 for all points except 50 mg/ml) of ␣-dystroglycan binding activity measured in the absence of added heparin.
binding to the laminin variants expressed in merosin-deficient muscle, thereby providing another explanation for the lack of functional compensation in the dy/dy mouse. Although it is not known presently which laminin is up-regulated in merosindeficient human muscle (9 -11), muscle from patients with Duchenne muscular dystrophy exhibits an increased expression of laminin ␣5 (37). It may be relevant that the ␣5 chain of laminin is most homologous with the ␣ chain of Drosophila laminin (59). In an interesting parallel with our results, heparin exquisitely inhibited both mouse laminin-1 and Drosophila laminin binding to perlecan (IC 50 Ͻ 1 g/ml), whereas heparin concentrations as high as 500 g/ml had no effect on perlecan binding to merosin (50,60). Elucidation of a function for the ␣-dystroglycan/merosin interaction awaits the results of experiments designed to perturb this interaction in muscle cells specifically. In the meantime, we have shown that ␣-dystroglycan can discriminate biochemically between merosin and other laminin variants, thus strengthening its candidacy as one of the cell surface receptors mediating a critical survival signal provided to muscle cells by merosin.