Association of the Mr 58,000 Postsynaptic Protein of Electric Tissue with Torpedo Dystrophin and the M;. 87,000 Postsynaptic Protein*

Dystrophin was purified by immunoaffinity chro-matography from detergent-solubilized Torpedo electric organ postsynaptic membranes using monoclonal antibodies. A major doublet of proteins at M, 58,000 and minor proteins at M, 87,000, M, 45,000, and M, 30,000 reproducibly copurified with dystrophin. The M, 68,000 and M, 87,000 proteins were identical to previously described peripheral membrane proteins (M, 58,000 protein and 87,000 protein) whose muscle homologs are associated with the sarcolemma (Froeh-ner, S. C., Murnane, A. A,, Tobler, M., Peng, H. B., and Sealock, R. (1987) J. Cell Biol. 104, 1633-1646; Carr, C., Fischbach, G. D., and Cohen, J. B. (1989) J. Cell Biol. 109, 1753-1764). The copurification of dystrophin and M, 58,000 protein was shown to be specific, since dystrophin was also captured with a monoclonal

Duchenne muscular dystrophy (DMD)' is a fatal genetic disease characterized by skeletal muscle necrosis and wasting (Engel, 1986). The protein product encoded by the DMD gene is a 427-kDa protein called dystrophin which has major homologies to b-spectrin and a-actinin (Hoffman et d., 1987;Koenig et al., 1988). Dystrophin occurs as part of a complex of proteins (Ervasti et al., 1991) associated with the cytoplasmic surface of the plasma membrane in normal skeletal, cardiac, and smooth muscle (Bonilla et al., 1988). It is absent or severely reduced in amount in muscle from Duchenne patients and the rndx mouse (Bonilla et al., 1988;Hoffman et al., 1987;Hoffman and Kunkel, 1989), a dystrophin-minus animal model for DMD (Bullfield et al., 1984). The function of dystrophin is unknown for certain, although it is believed to provide mechanical stability to the membrane by analogy to erythrocyte spectrin (Mandel 1989) and/or to regulate the activities of calcium leak channels (Fong et aL, 1990) or calcium permeable mechano-transducing channels (Franco and Lansman, 1990). Evaluation of these hypotheses would be greatly aided by identification of the activities of proteins with which dystrophin is associated. In addition, efficient gene replacement therapy for DMD may require shortened forms of the dystrophin gene (Acsadi et al., 1991). Construction of optimal short forms will require knowledge of dystrophin binding proteins and their binding sites on dystrophin.
We have approached the question of dystrophin-associated muscle proteins through the use of acetylcholine receptor-rich postsynaptic membranes isolated from electric tissue of electric rays (Torpedo sp.). The electrogenic cells, or electroplaques, which make up the tissue are derived embryologically from immature striated muscle cells (Fox and Richardson, 1979) and retain many morphological and biochemical similarities to mammalian skeletal muscle. In particular, they contain a very large protein which is homologous to human muscle dystrophin by immunological criteria (Chang et al., 1989;Jasmin et al., 1990;Sealock et al., 1991) and deduced partial amino acid sequence (Yeadon et al., 1991). This Torpedo dystrophin appears to be confined to the innervated face of the electroplaque (Jasmin et al., 1990;Sealock et al., 1991) along with several other cytoplasmic, peripheral membrane proteins that also have homologs in muscle (reviewed in Froehner, 1991). These include a protein of M, 43,000 which is directly involved in AChR clustering (Froehner et al., 1990), and proteins of M , 58,000, 87,000, and 270,000 or 300,000 (Froehner et al., 1987;Carr et al., 1989;Woodruff et al., 1987). The M, 270,000/300,000 protein has been shown to be iden-' The abbreviations used are: DMD, Duchenne muscular dystrophy; AChR, nicotinic acetylcholine receptor; EGTA, [ethylenebis(oxyethylenenitri1o)ltetraacetic acid FITC, fluorescein isothiocyanate; mAb, monoclonal antibody; SDS, sodium dodecyl sulfate.

6213
This is an Open Access article under the CC BY license.

Torpedo Dystrophin Complex
tical to Torpedo dystrophin, with the discrepancy in reported molecular weights of the muscle and Torpedo proteins resulting from difficulties in gel calibration for very large proteins (Sealock et al., 1991). In previous work, we showed that the detailed distributions of dystrophin and the muscle homolog of the M , 58,000 protein are very similar in cultured Xenopus muscle (Kramarcy and Sealock, 1990), as in adult muscle (Carr et al., 1989;Froehner et al., 1987), and raised the possibility that dystrophin and the M , 58,000 protein may be associated. In this study, we demonstrate their association in detergent extracts of AChR-rich membranes from Torpedo and show that the association of the M , 58,000 protein2 with muscle sarcolemma is altered in dystrophin-minus muscle. We also provide evidence for a distinct complex containing the M , 58,000 and 87,000 proteins.
Immunoaffinity Purification of the Dystrophin Complex-Torpedo acetylcholine receptor-enriched membranes were purified according to Porter and Froehner (1983) or with additional proteinase inhibitors in all solutions (5 pg/ml each of leupeptin, aprotinin, and antipain, and 0.5 pg/ml of pepstatin; Chapel Hill laboratory). Membranes (2 mg/ml) were solubilized in buffer A (150 mM NaCl, 5 mM EDTA, 5 mM EGTA, 5 pg/ml leupeptin, 2.5 pg/ml phenylmethylsulfonyl fluoride, 5 pg/ml pepstatin, 5 mM iodoacetamide, 10 mM sodium phosphate, pH 7.4 or 400 mM NaCI, 5 mM EDTA, 10 mM sodium phosphate, pH 7.4, plus the proteinase inhibitor mix listed above) containing 1% Triton X-100, incubatedon ice for 20 min, and clarified by centrifugation. MAb IgG-Affigel columns were prepared as previously described (LaRochelle and Froehner, 1987). MAb IgG-Sepharose columns were prepared according to recommendations of Pharmacia LKB Biotechnology Inc. The extracts were incubated in batch with immunoaffinity resins (1-2 ml) for 1-2 h. The resins were transferred to columns and washed extensively with buffer A containing 1% Triton or buffer A/Triton followed by buffer A alone. Columns were eluted with elution buffer (0.1 M NHIAc or with 0.1 M triethylamine plus the proteinase inhibitor mix) at pH 11.5. Neutralized fractions were precipitated with trichloroacetic acid prior to gel electrophoresis. The different solution compositions used gave very similar results, but the yields of dystrophin complex appears to be consistently higher in the Chapel Hill laboratory (second set of conditions in each pair above).
SDS-Gel Electrophoresis and Zmmunoblotting-Gel electrophoresis in a mini-gel system was performed as previously described (La-Rochelle and Froehner, 1987). For immunoblotting of dystrophin, proteins were transferred to nitrocellulose using the Tris glycine system (Sealock et al., 1991) in most cases. For all other proteins, transfer was carried out in 25 mM sodium phosphate, pH 6.5, at 250 mA for 2 h.
For analysis of M, 58,000 protein in normal and mdx mouse skeletal and cardiac muscle, lower leg muscles and the heart were dissected, frozen in liquid nitrogen, and pulverized with a cold mortar and We refer to all mAb 1351-reactive proteins in this paper as M, 58,000 proteins. The proteins in mouse skeletal and cardiac muscle typically migrated slightly more slowly on SDS gels than the Torpedo protein (see Fig. 6), and apparent molecular weights of 51,000 (Torpedo electric organ) and 48,000 (Xenopus skeletal muscle) have also been reported (Carr et al., 1989;Chen et al., 1990). Since the reasons for these different values are unknown, a single name for these proteins is used for simplicity.
pestle. Samples were then dissolved in 1% SDS, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 5 mM EGTA, 2 units/ml trasolyl, at a concentration of 250 mg of tissue/ml, boiled for 4 min, and centrifuged 2 min in a microcentrifuge to remove insoluble material. The protein concentration of the supernatant was determined by the Lowry method. Aliquots (75-150 pg) were subjected to gel electrophoresis and then transferred to nitrocellulose in 25 mM sodium phosphate, pH 6.5.
Immunofluorescence Microscopy-Frozen sections (6 pm) of diaphragm or gastrocnemius muscle were incubated sequentially with mAb I g G (25-50 nM), biotinylated goat anti-mouse IgG, and FITCavidin, according to established procedures. For analysis of cardiac muscle, 6-pm sections were incubated sequentially with FITC-derivatized anti-58,000 mAb 1351, rabbit anti-FITC (East Acres; 1:150), and FITC-goat anti-rabbit IgG (Jackson Immunoresearch; 1:250). Substitution of primary antibodies with nonspecific mouse IgG or with monoclonals of the IgG, subclass directed against other antigens were used as control for nonspecific labeling.

RESULTS
Identification of dystrophin-associated proteins in AChRrich membranes was accomplished by Triton X-100 solubilization of the membranes followed by immunoaffinity purifications using monoclonal antibodies. After incubation with the Triton extracts, antibody columns were washed extensively, and proteins were eluted at pH 11.5. In the experiment shown in Fig. 1, dystrophin was purified from one portion of an extract using the anti-dystrophin mAb 1808, and the M , 58,000 protein was purified from a second portion using the anti-58,000 mAb 1351. When subjected to SDS-polyacrylamide gel electrophoresis, the eluate of the anti-dystrophin column contained a major protein corresponding to dystrophin, a doublet of proteins having mobility very similar to the M , 58,000 protein (Froehner et al., 1987), and weaker bands

Torpedo Dystrophin Complex
a t mobilities corresponding to M , 140,000,87,000,45,000, and 30,000 (Fig. 1, lane 2). The two bands near M , 200,000 were presumably degradation products of dystrophin, since similar bands reacted with anti-dystrophin antibodies on immunoblots in previous experiments (Sealock et al., 1991). The material purified on an anti-58,000 column from the second portion of the extract contained proteins of M, 58,000 and M , 87,000 (both of which ran as doublets) and presumptive dystrophin (Fig. 1, lane 3 ) . None of these proteins was detectable in eluates from control mouse IgG columns incubated with detergent-solubilized membranes and then washed and eluted as described above (data not shown). Immunoblotting and additional purifications on columns containing antibodies against other Torpedo postsynaptic proteins were used to identify eluted proteins and to test the specificity of our methods. In a series of control experiments, a Triton X-100 extract of postsynaptic membranes was passed sequentially through three monoclonal antibody columns derivatized with an anti-AChR, an anti-43,000, and an antidystrophin mAb, respectively. Each monoclonal used (see "Materials and Methods") was of the IgGl subclass. After extensive wash, the columns were eluted at pH 11.5, and the eluates were subjected to SDS-polyacrylamide gel electrophoresis and immunoblotting. Dystrophin and the M , 58,000 protein were specifically recognized by mAbs 1808 and 1351, respectively, on blots of the eluate from the third, antidystrophin column ( Fig. 2A, lanes 2 and 5 ) . The M , 58,000 protein copurified with dystrophin even when the column was washed thoroughly with 1 M NaCl prior to elution at pH 11.5 (data not shown). In contrast, no AChR ( Fig. 2A, lane 4 ) and  immunoblotting (A, lanes 2-6 or B and C, lanes 2-4). were detected in this fraction. The latter may not be significant, since similar amounts of the M , 43,000 protein were detected on immunoblots of the eluate from the anti-AChR column (Fig. 2B, lane 3 ) , and the M , 43,000 protein and AChR are known not to be associated in detergent extracts. In contrast, the M , 58,000 protein was not visible in the eluates from the anti-AChR or anti-43,000 columns (Fig. 2, B and C, respectively), either by Coomassie Blue staining (lane 1 in each panel) or by immunoblotting (lane 2). Both of these columns gave efficient purification of their respective antigens, however (Figs. 2B, lane 4 and 2C, lane 3 ) . These results establish that one or both of the proteins in the copurifying doublet at M, 58,000 was identical to the M , 58,000 protein described previously (Froehner et al., 1987).
Material purified from an extract using a column derivatized with anti-58,000 mAb 1351 was subjected to immunoblotting analysis using mAb 1351, anti-dystrophin mAb 1808, and anti-87,000 mAb 20H2, which was one of the mabs used by Carr et al. (1989) to identify the M , 87,000 protein. The presumed 58,000 and 87,000 proteins and dystrophin were each specifically recognized by the appropriate antibody (Fig.  3, lanes 5-7). The two proteins recognized by mAb 1351 and 20H2, respectively, had identical mobilities to the proteins recognized by these mAbs in the starting membranes (Fig. 3,   lanes 2 and 3 ) 1351 (lanes 2 and 5), anti-87,000 mAb 20H2 (lanes 3 and 6) or anti-dystrophin mAb 1808 (lane 7). Lane 8 was incubated with control mouse IgG. Lanes 1 and 4 are Coomassie Blue stained; the lanes 2, 3, and 5-8 are immunoblots. Differences in mobility of the M, 58,000 and 87,000 proteins, when compared to the gels in Fig. 1, are due to different SDS gel conditions. The relatively small amount of dystrophin in the membranes could not be detected by antidystrophin mAb (data not shown) because the conditions used for electrophoretic transfer (25 mM NaP04, pH 6.5, for 2 h at 250 mA) were inefficient for large proteins.

Torpedo Dystrophin Complex
shown). Similarly, quantitative removal of dystrophin from extracts using mAb 1808 did not remove all the M , 58,000 protein. When the remaining M , 58,000 protein was captured using mAb 1351, it contained substantial amounts of M , 87,000 protein (data not shown). Hence, the M , 58,000 and 87,000 proteins can apparently exist in a complex independently of dystrophin. The presence of small amounts of M , 87,000 protein in preparations of dystrophin (Fig. 1, lane 2 ) would be compatible with a ternary M , 58,000/87,000/dystrophin complex. However, quantitative gel scanning showed that the ratio of M , 87,000 to 58,000 protein was 2-to 3-fold lower in dystrophin preparations than in M , 58,000 preparations (starting with initial extracts in both cases). Much of the dystrophin-bound M , 58,000 protein is therefore not accompanied by the M , 87,000 protein, and the origins of the small amounts of M , 87,000 protein in the dystrophin preparations are not clear.
If the M , 58,000 protein is associated with dystrophin in skeletal muscle, as has been suggested by indirect evidence (Kramarcy and Sealock, 1990;Sealock et al., 1991), it could be present in reduced quantities or even absent in dystrophinminus muscle compared to normal muscle. To test this, skeletal and cardiac muscle from mdx mice and age-matched normal mice were probed for the M , 58,000 protein by immunofluorescence and immunoblotting. As shown previously (Bonilla et al., 1988;Chang et al., 1989;Sealock et d., 1991), immunofluorescent staining for dystrophin was confined to the sarcolemma of normal skeletal muscle (Fig. a), with particularly strong staining at the neuromuscular junction, but was undetectable in muscle from mdx mice (Fig. 4R). The distribution of the M , 58,000 protein in normal muscle is indistinguishable from that of dystrophin (Fig. 4C). In mdx muscle, however, the intensity of anti-58,000 staining on the sarcolemma outside the neuromuscular junction was on average markedly reduced (Fig.  4 0 ) ; i t ranged from almost completely negative in many mdx fibers to moderately strong but noticeably less intense than in normal fibers. Similar results were obtained in comparing normal and mdx cardiac muscle (Fig. 4, E and F) and normal and Duchenne human muscle (not shown).
A consistent and notable exception to this pattern was the neuromuscular junction in skeletal muscle. The postsynaptic region of the junction (identified by staining with rhodaminea-bungarotoxin) was consistently brightly stained by anti-58,000 mAb in both normal and mdx samples (Fig. 4, C and   0).
The decrease in anti-58,000 staining in mdx muscle was not a consequence of the general loss of cytoskeletal proteins associated with the membrane, or of the physical state of the membrane. Staining of mdx muscle with antispectrin was only slightly reduced and somewhat more diffuse than in normal muscle (Fig. 5 , A and R ) . Similar results were ohtained with anti-a-actinin antibodies (not shown). In douhle label immunofluorescence for the M , 58,000 protein and vinculin in mdx muscle, antivinculin staining was essentially unaffected (Fig. 5, C and D), even in regions that were almost completely negative for the M , 58,000 protein (Fig. 5, F: and F ) . These results indicate that the M , 58,000 protein is selectively lost in dystrophic muscle. On immunoblots of normal and mdx mouse muscle, mAb 1351 specifically recognized a single protein in both skeletal and cardiac muscle which migrated slightlv more slowly than the Torpedo M , 58,000 protein (Fig. 6). The amounts of this muscle M , 58,000 protein in normal uersus skeletal (hind leg) muscle (Fig. 6, upper panel) and in normal uprsus cardiac muscle (Fig. 6, lower panel) appeared in most experiments to be very similar or indistinguishable hv this method. In the experiment which showed the greatest divergence between normal and mdx muscle (shown in Fig. fi 58,000 protein in mdx muscle appeared to reduced a t most by about 50%. These data suggest that the abundance of the M , 58,000 protein does not depend to a large extent on the presence of dystrophin.

DISCUSSION
Using monoclonal antibodies, we have shown that Torpedo dystrophin purified by anti-dystrophin immunoaffinity chromatography is accompanied by the M , 58,000 protein. Similarly, dystrophin copurifies with the M , 58,000 protein isolated with an anti-58,000 antibody. Since neither protein was captured with control antibodies or with antibodies against other major proteins in these membranes ( M , 43,000 protein and AChR), these findings constitute strong evidence that dystrophin and the M , 58,000 protein exist in a complex. The two proteins also occurred independently in our Triton extracts, since each could be captured alone after complete removal of the other. These individual entities may reflect dissociation that occurs after membrane solubilization, or may represent physiologically important states of these proteins.
Our second finding was that immunoaffinity-purified M , 58,000 protein contained large quantities of the M , 87,000 postsynaptic protein described by Carr et al. (1989). Th' IS was true whether or not dystrophin was removed from extracts prior to capture of the M , 58,000 protein. Hence, the M , 58,000 protein also apparently forms a separate complex with the M , 87,000 protein, a possibility which has not been previously suspected. The M , 87,000 protein was present in most of our dystrophin preparations (i.e. dystrophin prepared using antidystrophin columns), but the amounts were low. In the absence of crossed immunoaffinity purification (the use of anti-87,000 columns to purify dystrophin), it is difficult to be certain that this copurification was due to a specific associa-tion. The M , 87,000 protein is unlikely to mediate the interaction between dystrophin and the M , 58,000 protein since the M , 87,000/58,000 protein ratio was much lower in dystrophin preparations than in M , 58,000 protein preparations. It remains possible that the M, 58,000/87,000 complex could be associated with dystrophin in situ, but that during memhrane preparation and solubilization, the M, 87,000 protein largely dissociates from this ternary complex. At the same time, the free M , 58,000/87,000 complex would remain intact. The simpler hypothesis is that of two binary complexes for the M , 58,000 protein, but further studies comparing the composition of the complexes solubilized under different conditions are needed to definitively answer these questions.
Other proteins of M , 140,000,45,000, and 30,000 also reproducibly copurified with dvstrophin and are potential components of a dystrophin multiprotein complex. However, the amounts of these proteins relative to dvstrophin or the M, 58,000 protein were low, and in the absence of specific antibodies to assess the specificitv of their appearance in these preparations, t,heir potential association with dystrophin is speculative.
Several lines of indirect evidence, in addition to the well known similarities of Torppdo and mammalian muscle proteins, suggest that a dystrophin", 58,000 complex also exists in muscle. In previous immunofluorescence experiments, we showed that, to a high degree of precision, dvstrophin-positive regions of sarcolemma in cultured X m o p u s muscle are also positive for the M , 58,000 protein (Kramarcv and Sealock 1990). Here we found that the anti-58,000 mAh gave suhstantially reduced immunofluorescent staining of dystrophin-minus muscle relative to normal muscle, with some fihers being nearly negative for the M , 58,000 protein. The average steady state levels of the M , .58,000 protein in dystrophic (rndx) mouse muscle, as assessed hy immunoblot. analvsis, were generally unaffected, although one experiment showed a decrease of approximately 505. We considered three possibilities to explain the decreased immunofluorescence staining.
One is that lower abundance of the M , 58,000 protein, although minimal, is directly responsihle for the reduced staining of some fibers. A second possibility is that the absence of dystrophin hinders the incorporation of the M , 58,000 protein onto the sarcolemma. Finally, an unlikely but formal alternative possibility is that access of mAh 1351 to its epitope is hindered in the absence of dystrophin. In the absence of antibodies against additional epitopes on the M , 58,000 protein, we cannot yet evaluate this possihilitv. Whatever the cause, the results indicate an altered mode of association of the M , 58,000 protein with the membrane in the absence of dystrophin.
Our findings bear strong resemhlance to some of the results obtained with the purified dystrophin complex from rahbit skeletal muscle by Campbell and coworkers (Ervasti ~t al., 1990). In that complex, they found two proteins of M , 59,000 and 88,000 and four glycoproteins of M, 156,000, 50,000, 43,000, and 35,000. Ry the criterion of molecular weight and staining intensities relative to dystrophin. the first two of these proteins are potent.ially the muscle homologs of the Torpedo M , 58,000 and 87,000 proteins. A marked difference between our results and those of Ervasti ~t a/. (1990) is that we have so far failed to find evidence for glycoproteins in the Torpedo dystrophin complex. In several attempts, we failed to isolate dystrophin by affinity procedures using wheat germ agglutinin or concanavalin A, even though the presumptive Na,K-ATPase of electric tissue (an M , 90.000 protein present in the same extracts) was readily captured.', It will be impor-' N. R. Kramarcy and R. Sealock, unpuhlished results. tant to determine whether these differences arise for technical reasons or because of species, tissue, or isoform differences.
Our finding that the M, 58,000 protein can occur in Triton extracts of postsynaptic membranes independently of dystrophin may be related to the fact that it is found in tissues such as kidney that express dystrophin at only very low level^.^ Clusters of AChR in cultured muscle are also rich in the M, 58,000 protein (Froehner et al., 1987) but, at least in Xenopus muscle, contain little dystrophin (Sealock et al., 1991). In dystrophic muscle, immunofluorescent staining of the neuromuscular junction with mAb 1351 remained bright while staining of the extrajunctional sarcolemma was weak. The M, 58,000 protein in mammalian tissues therefore appears to interact with additional proteins, possibly including the M , 87,000 protein and other members of the dystrophin family.
Candidates in the latter include the muscle isoform of pspectrin described by Bloch and Morrow (1989) and the dystrophin-related protein described by Fardeau et al. (1990), both of which are concentrated at neuromuscular junctions.
Dystrophin-independent activities for the M , 58,000 protein may explain the partial reduction of M, 58,000 protein content and sarcolemmal incorporation in dystrophic muscle, in contrast to the near total absence of the dystrophin-associated M, 156,000 glycoprotein studied by Ervasti et al. (1990). The function of the M, 58,000 protein is not known, and sequence analyses of Torpedo and mouse skeletal muscle cDNAs encoding the M , 58,000 protein have not revealed homology to known protein^.^ Its diminished association with the sarcolemma in dystrophic muscle may, however, contribute to the pathology of muscular dystrophy. Abnormalities in the function or expression of the M, 58,000 protein could also potentially be the primary defect in other muscle pathologies in which dystrophin is normal.