Biochemical Characterization of the Epithelial Dystroglycan Complex*

Dystroglycan is a widely expressed extracellular matrix receptor that plays a critical role in basement membrane formation, epithelial development, and synaptogenesis. Dystroglycan was originally characterized in skeletal muscle as an integral component of the dystrophin glycoprotein complex, which is critical for muscle cell viability. Although the dystroglycan complex has been well characterized in skeletal muscle, there is little information on the structural composition of the dystroglycan complex outside skeletal muscle. Here we have biochemically characterized the dystroglycan complex in lung and kidney. We demonstrate that the presence of sarcoglycans and sarcospan in lung reflects association with dystroglycan in the smooth muscle. The smooth muscle dystroglycan complex in lung, composed of dystroglycan, dystrophin/utrophin, β-, δ-, ε-sarcoglycan, and sarcospan, can be biochemically separated from epithelial dystroglycan, which is not associated with any of the known sarcoglycans or sarcospan. Similarly, dystroglycan in kidney epithelial cells is not associated with any of the sarcoglycans or sarcospan. Thus, our data demonstrate that there are distinct dystroglycan complexes in non-skeletal muscle organs as follows: one from smooth muscle, which is associated with sarcoglycans forming a similar complex as in skeletal muscle, and one from epithelial cells.

␣-Dystroglycan is a highly glycosylated peripheral membrane protein associated with the membrane-spanning ␤-dystroglycan. These two proteins were originally isolated from skeletal muscle as components of a large oligomeric complex further comprised of dystrophin, the syntrophin, and sarcoglycan complexes and the recently identified protein sarcospan (1)(2)(3)(4)(5). In skeletal muscle, ␣-dystroglycan binds to the extracellular matrix component laminin ␣2-chain (6), whereas the intracellular domain of ␤-dystroglycan binds to the cytoskeletal protein dystrophin (7). Thus, dystroglycan is thought to act as a transmembrane link between the extracellular matrix and the cytoskeleton, and this linkage seems to be crucial for maintaining normal function (4,8,9). In addition, a defect in any of the sarcoglycans results in specific loss of the sarcoglycansarcospan subcomplex, destabilization of ␣-dystroglycan, and eventually muscle cell death (4).
Many of the dystroglycan-associated proteins found in skeletal and cardiac muscle are also expressed in other tissues. For example, ␤and ␦-sarcoglycan, sarcospan, and the syntrophins are all at the RNA level expressed in tissues other than skeletal and cardiac muscle (3, 5, 10 -13). In addition, several dystrophin isoforms are ubiquitously expressed, including the autosomal dystrophin homologue utrophin (14 -17). By far, dystroglycan is the most widely expressed component. Dystroglycan is expressed at high levels in many cell types and is particularly prominent on the basal side of epithelial cells facing basement membranes (18 -23). Moreover, ␣-dystroglycan binds laminin-1, agrin, and perlecan (24 -28). ␤-Dystroglycan, in turn, also binds to the dystrophin isoforms Dp71, Dp116, and Dp260 (7,29) and has been shown to be associated with utrophin (30). Thus, different dystroglycan complexes may form in different tissues implying that dystroglycan may have important roles outside skeletal muscle. Indeed, dystroglycan has been implicated as a laminin/agrin receptor involved in epithelial cell development, basement membrane formation, and synaptogenesis (19,21,(31)(32)(33). Recently, it was also demonstrated that dystroglycan serves as a receptor for both lymphocytic choriomeningitis virus (34) and Mycobacterium leprae (35). The roles of the dystroglycan-associated proteins outside skeletal muscle, however, have remained elusive. To elucidate potential functional roles, a first important step includes the determination of the cellular localization of these proteins. Specifically, are the dystroglycan-associated proteins that are present in non-skeletal muscle organs derived from smooth muscle, epithelial cells, or both?
⑀-Sarcoglycan, a homologue of ␣-sarcoglycan, was recently identified and shown to be expressed in a wide variety of tissues including epithelial cells (36,37). ⑀-Sarcoglycan has also been identified as an integral component of the smooth muscle sarcoglycan complex 1 and could also be part of an epithelial dystroglycan complex, possibly along with ␤and ␦-sarcoglycan and sarcospan.
Although dystroglycan has been shown to play important roles outside skeletal muscle, the dystroglycan complex from other cell types is only partially characterized. In peripheral nerve, the molecular mass of ␣-dystroglycan is 120 kDa in contrast to 156 kDa in skeletal muscle (38,39). The molecular weight of ␣-dystroglycan in epithelial cells has remained elusive. A 120-kDa band has been identified in lung (18), but it is unclear whether the 120-kDa dystroglycan is derived from epithelial cells or smooth muscle. Moreover, ␣-dystroglycan is believed to act as a laminin-1 receptor involved in the development of kidney epithelial cells (19). However, a direct association between kidney ␣-dystroglycan and laminin-1 has not been demonstrated.
Here, we demonstrate that the presence of the sarcoglycans and sarcospan in lung reflects association with dystroglycan in the smooth muscle within lung, and we show that epithelial dystroglycan is not associated with any of the known sarcoglycans and can be separated from the smooth muscle dystroglycan complex. Dystroglycan, ␤-sarcoglycan, ␦-sarcoglycan, ⑀-sarcoglycan, and sarcospan are all part of a smooth muscle complex, whereas neither ␣-, ␤-, ␥-, and ␦-sarcoglycan nor sarcospan are expressed in epithelial cells from lung and kidney. ⑀-Sarcoglycan, on the other hand, is expressed in epithelial cells (36) but is not complexed with dystroglycan, as revealed by sucrose gradient fractionation. We have also partially characterized dystroglycan from kidney, an organ mainly composed of epithelial cells. We found that the major part of ␣-dystroglycan from adult kidney has a molecular mass of 156 kDa, whereas the major part of ␣-dystroglycan from fetal kidney has a molecular mass of 120 kDa. Both forms were shown to bind laminin-1. Taken together, our results demonstrate a distinct smooth muscle dystroglycan complex and a distinct epithelial dystroglycan complex, and this information will be useful for further investigation of dystroglycan function.

EXPERIMENTAL PROCEDURES
Animals-New Zealand White rabbits were from Knapp Creek Farms (Amana, IA). Mice (C57BL/10) were bred at the University of Iowa from stocks originally obtained from Jackson Laboratories (Jackson Laboratories, Bar Harbor, ME). F1B control and BIO 14.6 cardiomyopathic hamsters were obtained from BioBreeders (Fitchburg, MA). All animals were kept in the animal care unit of the University of Iowa College of Medicine according to the animal care guidelines.
Northern Blot Analysis-Total RNA from rabbit kidney, lung, and skeletal muscle was extracted using RNAzol B (Tel-Test) according to the manufacturer's specifications. 20 g of total RNA was electrophoresed on a 1.25% agarose gel containing 5% formaldehyde. Equal loading was verified by ethidium bromide visualization of the RNA, which was subsequently transferred to Hybond N membrane (Amersham Pharmacia Biotech). RNA was cross-linked to the membrane using a Stratagene UV cross-linker. Membranes were then prehybridized and hybridized using standard methods (40). Hybridization was performed with the following cDNA probes: a 1.1-kb 2 cDNA probe representing mouse ␣-sarcoglycan (41); a 1-kb cDNA probe representing hamster ␤-sarcoglycan (40); a 1-kb cDNA probe representing hamster ␥-sarcoglycan (40); a 1-kb probe representing hamster ␦-sarcoglycan (40); a 0.7-kb cDNA probe representing human sarcospan (5); a rabbit full-length clone of dystroglycan (18); a 1-kb cDNA clone corresponding to exon 70 through the beginning of 79 of mouse utrophin; an 850-base pair EcoRI/ XhoI fragment derived from EST clone 1149778 corresponding to mouse ⑀-sarcoglycan (36). cDNA inserts were labeled with [␣-32 P]dCTP to a specific activity of 2 ϫ 10 8 cpm/g DNA using Ready Prime kit (Amersham Pharmacia Biotech). Washes were carried out at 65°C in 1ϫ SSC, 1% SDS initially and then in 0.1ϫ SSC, 0.1% SDS. Membranes were exposed for autoradiography.
against ␤-dystroglycan was also generated by Louise V. B. Anderson. Polyclonal antibodies against dystroglycan fusion protein B (18) were affinity purified from sheep OR12. Sheep OR12 was injected with fusion protein B and boosted with fusion protein D (18).
Immunoblot Analysis of Membrane-enriched Preparations-Preparation of KCl-washed membranes from adult rabbit kidney, lung, and skeletal muscle were described previously (45). Membranes were resolved by SDS-PAGE on 3-12% linear gradients and transferred to nitrocellulose membranes (46). Immunoblot staining was performed as described previously (45).
Laminin Preparation and Biotinylation-Mouse EHS laminin (laminin-1) was generously provided by Dr. Hynda K. Kleinman at the National Institutes of Health. Laminin-1 was biotinylated with NHSbiotin (Vector Laboratories) in a molar ratio of 1:750 -1000 in reaction buffer 0.2 M NaHCO 3, pH 8.5, containing 0.5 M NaCl. The reaction was quenched, and unbound biotin was removed by dialysis against Trisbuffered saline (50 mM Tris, pH 7.5, 150 mM NaCl).
Laminin Overlay Assay-Dystroglycan containing sucrose gradient fractions were separated on 3-12% gradient gels and transferred to nitrocellulose membranes as described (45). The blots were blocked in laminin binding buffer (140 mM NaCl, 1 mM MgCl 2 , 10 mM triethanolamine, pH 7.6) containing 5% non-fat dry milk and subsequently washed for 90 min in laminin binding buffer containing 0.05% Tween 20 (TLBB). Blots were incubated in TLBB containing 3% bovine serum albumin and biotinylated laminin-1 for 8 h and washed in TLBB for 20 min. After incubation with ABC reagents (Vector Laboratories), the blots were washed in TLBB for 20 min and developed in 4-chloro-1-naphthol and H 2 O 2 .
Immunofluorescence Analysis-For immunofluorescence analysis, 8-m cryosections were prepared from C57BL/10 wild type mouse lungs, kidneys, and skeletal muscle, and F1B and BIO 14.6 hamster lungs, kidneys, and skeletal muscle. All procedures were performed at room temperature. Sections were blocked with 3% bovine serum albumin in phosphate-buffered saline for 30 min and then incubated with the primary antibodies for 90 min or overnight. After washing in phosphate-buffered saline, sections were incubated with Cy3-conjugated secondary antibodies (Vector Laboratories) for 90 min and then mounted with Vectashield mounting medium and observed under a Bio-Rad MRC-600 laser scanning confocal microscope.

Expression of the Sarcoglycans, Sarcospan, Dystroglycan, and Utrophin in Kidney and
Lung-In order to analyze which proteins are associated with dystroglycan in kidney and lung, we first screened for the presence of sarcoglycan mRNAs in these organs. As expected, the 1.6-kb ␣-sarcoglycan transcript was only detected in skeletal muscle (Fig. 1). The previously characterized 4.4-and 3.0-kb transcripts representing ␤sarcoglycan (10,11) were present in both lung and kidney as well as the previously characterized ␦-sarcoglycan mRNAs (40) of 9.5, 4.3, 2.3, and 1.4 kb (Fig. 1). In kidney and lung the 9.5-kb ␦-sarcoglycan transcript appeared to be the most prominent. ␥-Sarcoglycan mRNAs of 2.5 and 1.7 kb were faintly expressed in kidney, moderately in lung, and strongly in skeletal muscle (Fig. 1). Sarcospan mRNA has been shown to be expressed in several organs (5), and the 4.5-kb transcript could also be detected in kidney and lung (Fig. 1). In contrast to the sarcoglycan and sarcospan mRNAs, which are all more highly expressed in skeletal muscle than kidney and lung, an equal intensity of the 5.8-kb dystroglycan transcript was detected in the three organs (Fig. 1). The 1.8-kb ⑀-sarcoglycan mRNA was highly expressed in kidney and lung (Fig. 1). ⑀-Sarcoglycan was also detected in skeletal muscle but at a lower level. Similarly, The rabbit polyclonal antibody against human ⑀-sarcoglycan has a higher affinity for hamster ⑀-sarcoglycan than mouse ⑀-sarcoglycan. We therefore used hamster tissues to analyze ⑀-sarcoglycan expression (P, T, and X). A-D and M-P show stainings of the smooth muscle of arteries, whereas E-H and Q-T show stainings of smooth muscle and respiratory epithelium. e, epithelium; sm, smooth muscle. Arrows in R and S denote epithelial basement membranes. Bar, 50 m. the 13-kb utrophin transcript was also detected in kidney and lung and at a lower level in skeletal muscle (Fig. 1).
Because it is possible that transcripts are present in cells without being translated, we next analyzed whether the sarcoglycan protein products were expressed in the kidney and lung. KCl-washed membranes from kidney, lung, and skeletal muscle were screened for protein expression. Although the ␤-, ␥-, and ␦-sarcoglycan transcripts were readily expressed in the kidney, we were not able to detect their corresponding polypeptides (Fig. 1). In the lung, however, the 43-kDa ␤-sarcoglycan, the 35-kDa ␥-sarcoglycan, and the 35-kDa ␦-sarcoglycans were clearly detectable (Fig. 1). Likewise, the 25-kDa sarcospan was not present in the kidney but was present in the lung (Fig. 1). In contrast, dystroglycan with a molecular mass of 156 kDa, the 46-kDa ⑀-sarcoglycan, and the 395-kDa utrophin were all readily detected in kidney and lung (Fig. 1).
Previous reports have shown that full-length dystrophin is present in the smooth muscle of kidney but not epithelial cells (16,47). In the lung, dystrophin is exclusively expressed in smooth muscle (48). The adult lung is mainly composed of epithelial cells and smooth muscle, whereas the kidney is largely composed of epithelial cells. Thus, the expression data suggest that the presence of dystroglycan, ␤-sarcoglycan, ␥/␦sarcoglycan, ⑀-sarcoglycan, sarcospan, and utrophin/dystrophin in lung could represent the dystroglycan complex from smooth muscle. Likewise, the presence of dystroglycan, ⑀-sarcoglycan, and utrophin in the kidney could represent an epithelial complex. To test that hypothesis we used immunofluorescence analysis and sucrose gradient fractionation to determine whether proteins observed by Western blot were derived from epithelial cells, smooth muscle, or both.
Distribution of Dystroglycan-associated Components in Lung-By immunofluorescence analysis we found that ␣and ␥-sarcoglycan were absent in smooth muscle surrounding respiratory epithelium, whereas ␤-, ␦and ⑀-sarcoglycan and sarcospan were readily detected, along with utrophin and ␤-dystroglycan (Fig. 2). Similar results were found in the smooth muscle of arteries within the lung (Fig. 2). In addition, in other smooth muscle containing tissues similar observations have been made. 1 Although ␥-sarcoglycan was readily detected in lung by Western blot analysis (see Fig. 1), we were never able to detect it by immunofluorescence in lung or in any other organ, and currently we do not have any explanation for this discrepancy. As we did not detect ␣-sarcoglycan in the lung, we can rule out contamination of striated muscle in the lung preparations. Dystroglycan is expressed in most epithelial cells underlying the basement membrane (23). This is also true for lung, as exemplified in Fig. 2R, where ␤-dystroglycan was detected in the basement membrane area of the respiratory epithelium. As expected, ␤-dystroglycan was also expressed in the surrounding smooth muscle. Also, utrophin was co-expressed with ␤-dystroglycan in the epithelial basement membrane region and smooth muscle (Fig. 2S). However, none of the sarcoglycans could be detected in the epithelial cells. These data indicate that there might be two dystroglycan complexes in lung, one associated with epithelial cells and one associated with smooth muscle.
Biochemical Characterization of Dystroglycan Complexes from Lung-We demonstrated the presence of two dystroglycan complexes in lung by partially purifying dystroglycan from lung. Purification of dystroglycan was achieved by WGA affinity chromatography followed by ion exchange chromatography and high speed centrifugation through sucrose gradients. Only proteins that bind with high affinity and specificity will be retained with dystroglycan during sucrose gradient fractionation. This purification method has been successfully used to purify the dystroglycan complex from skeletal and cardiac mus-cle (49). Dystroglycan was extracted from adult rabbit lung, and proteins from the sucrose gradient fractions were separated by SDS-PAGE. The resulting polyacrylamide gels were immunoblotted with various antibodies. Staining with monoclonal antibodies against ␤-dystroglycan revealed one peak of ␤-dystroglycan in fractions 6 -8 and one peak in the last three fractions (Fig. 3). ␤-Dystroglycan in the last fractions was accompanied with dystrophin, utrophin, ␤-sarcoglycan, ␥/␦-sarcoglycan, ⑀-sarcoglycan, and sarcospan ( Fig. 3 and data not shown). A 156-kDa ␣-dystroglycan band was detected in fractions 3-8, suggesting that some ␣-dystroglycan is associated with ␤-dystroglycan seen in fractions 6 -8 but also that some ␣-dystroglycan is not associated with ␤-dystroglycan (Fig. 3). Antibody IIH6 which shows a carbohydrate-dependent staining did not detect any ␣-dystroglycan in the last fractions. Interestingly, IIH6 does not detect smooth muscle ␣-dystroglycan. 1 However, it has been shown that smooth muscle ␣-dystroglycan has a molecular mass of 100 and 156 kDa. 1 In none of the earlier fractions (fractions 6 -8) did we observe a co-migration of dystroglycan with other dystroglycan-associated proteins. Together, these data indicate that ␣and ␤-dystroglycan seen in the earlier fractions may represent dystroglycan derived from epithelial cells, and dystroglycan seen in fractions 11-14 accompanied by dystroglycan-associated proteins may represent the smooth muscle dystroglycan complex. This was further shown by sucrose gradient fractionation analysis of purified dystroglycan from the BIO 14.6 hamster lung. Recently, a mutation in the ␦-sarcoglycan gene of the BIO 14.6 cardiomyopathic hamster was identified to cause the disease in this animal strain (50,51). As previously shown, the null mutation in the ␦-sarcoglycan gene affects the distribution of ␣and ␤-dystroglycan and the sarcoglycans in skeletal muscle of the BIO 14.6 hamster (40,43,52). As ␦-sarcoglycan is absent from epithelial cells, the null mutation in the BIO 14.6 hamster should not affect dystroglycan in epithelial cells of the lung but would presumably affect the expression in lung smooth muscle. We demonstrate that this is indeed the case. The control hamster sucrose gradient profile showed that ␦and ⑀-sarcoglycan co-sedimented in fractions 9 -14 (Fig. 4A). ␤-Dystroglycan co-sedimented with the smooth muscle sarcoglycans in fractions 9 -14 (Fig. 4A). In addition, there were earlier fractions (fractions 6 -8), with high levels of ␤-dystroglycan without association of the sarcoglycans (Fig. 4A). The epithelial ␤-dystroglycan signals in these fractions remained in the BIO 14.6 lung fractions, whereas smooth muscle ␤-dystroglycan in the later fractions was absent (Fig. 4A). As expected, no sarcoglycan staining could be detected in BIO 14.6 lung fractions (Fig. 4A). The presence of ␤-dystroglycan in lung epithelial cells of the BIO 14.6 hamster was further verified by immunofluorescence analysis. ␤-Dystroglycan remained in the basement membrane region of lung epithelial cells but was absent in lung smooth muscle of the BIO 14.6 hamster (Fig.  4B). Taken together these results suggest that within one tissue there are different dystroglycan complexes. To determine further the composition of the dystroglycan complex from epithelial cells, we next isolated the dystroglycan-containing com-plex from adult rabbit kidney.
Characterization of the Dystroglycan Complex from Adult Rabbit Kidney-The dystroglycan complex was extracted from adult rabbit kidney by WGA affinity chromatography followed by ion exchange on a DEAE column and sucrose gradient fractionation. Proteins from the sucrose gradient fractions were separated by SDS-PAGE. The resulting polyacrylamide gels were immunoblotted with antibodies to the components of the dystroglycan complex. Staining with monoclonal antibody IIH6 revealed a 156-kDa ␣-dystroglycan band, peaking in fractions 7 and 8 (Fig. 5). Immunoblotting with antiserum against purified skeletal muscle dystroglycan complex (goat 20) revealed ␣-dystroglycan as a broad smear between 100 and 156 kDa suggesting the existence of smaller molecular weight isoforms of ␣-dystroglycan (Fig. 5). Goat 20 antiserum was raised against the skeletal muscle dystroglycan complex (5) and recognizes dystrophin, ␣-dystroglycan, ␤-dystroglycan, the sarcoglycan and syntrophin complexes (Fig. 5). Staining with goat 20 antiserum also revealed a peak of ␤-dystroglycan in fractions 7 and 8 (Fig. 5). This was further verified using monoclonal antibody 43DAG/8D5 against ␤-dystroglycan (data not shown).
As previously indicated we did not detect ␣-, ␤-, ␥-, ␦-sarcoglycan or sarcospan in the kidney as shown by immunostaining using goat 20 antiserum and monoclonal antibodies against the sarcoglycans (Fig. 5 and data not shown). Furthermore, sucrose gradient fractionation separated ⑀-sarcoglycan and utrophin from dystroglycan. Western blotting with ⑀-sarcoglycan antibodies demonstrated that ⑀-sarcoglycan began to sediment in fraction 7 and peaked in fractions 10 and 11 (Fig. 5). As described, ␣and ␤-dystroglycan peaked in fractions 7 and 8 (Fig.  5). Utrophin began to sediment in fraction 10 and peaked in fractions 12-14 (Fig. 5). Taken together, it appears that ⑀-sarcoglycan and utrophin are not integral components of a kidney dystroglycan complex, at least not in the adult kidney. In the fetal kidney, however, dystroglycan and ⑀-sarcoglycan are colocalized in epithelial cells (19,36). Utrophin mRNA is also highly expressed in fetal kidney (53). Hence, we next isolated fetal kidney dystroglycan complex to determine whether dystroglycan, ⑀-sarcoglycan, and utrophin could be complexed at early stages of development.
Characterization of the Dystroglycan Complex from Fetal Rabbit Kidney-The dystroglycan complex was extracted from fetal rabbit kidneys. Centrifugation of the dystroglycan complex through sucrose density gradients followed by SDS-PAGE and immunoblotting with monoclonal antibody IIH6 revealed a major dystroglycan band of 120 kDa (Fig. 6). Also, dystroglycan antibodies from sheep OR 12 reacted with a 120-kDa ␣-dystroglycan band in addition to ␤-dystroglycan bands of 43 kDa (Fig. 6). At present it is unclear why two bands of ␤-dystroglycan around 43 kDa are detected. They could represent post-translational modification and/or processing events of ␤-dystroglycan. Stainings with antiserum from goat 20 demonstrated the presence of a possible 100-kDa form of ␣-dystroglycan (Fig. 6). In summary, it appears that ␣-dystroglycan in fetal rabbit kidney has a molecular mass of 120 kDa, and smaller isoforms may also exist.
As in the adult kidney, we did not detect ␣-, ␤-, ␥-, ␦-sarcoglycans or sarcospan in fetal kidney (Fig. 6). Also, sucrose gradient fractionation separated fetal kidney ⑀-sarcoglycan from fetal kidney dystroglycan, suggesting that ⑀-sarcoglycan is not an integral component of a fetal kidney dystroglycan complex (Fig. 6). Utrophin began to sediment in fractions 7 and 8. Then the expression declined in fractions 9 and 10 and peaked in fractions 12-14 (Fig. 6). Unlike ⑀-sarcoglycan, utrophin expression still overlaps with dystroglycan expression, raising the possibility that in the fetal kidney, utrophin could be part of a dystroglycan complex.
Kidney Dystroglycan Binds Laminin-1-The role of laminin-1 and dystroglycan in embryonic kidney development has been well established (19,54), but it has never been formally shown that they are directly associated. The molecular mass of ␣-dystroglycan in the adult kidney is 156 kDa and in the fetal kidney ␣-dystroglycan is 120 kDa (Figs. 5-7). To address the question whether the two dystroglycan isoforms bind laminin-1, we performed a laminin-overlay assay using biotinylated laminin-1. As shown in Fig. 7, biotinylated laminin-1 bound the 156-kDa ␣-dystroglycan from adult kidney and the 120-kDa ␣-dystroglycan from fetal kidney. DISCUSSION The overall goal of this study was to biochemically characterize the dystroglycan complex in epithelial cells. Our results demonstrated the following: 1) dystroglycan present in epithelial cells is not associated with any of the known sarcoglycans; 2) the presence of ␤-sarcoglycan, ␦-sarcoglycan, and sarcospan in non-skeletal muscle organs reflects association with dystroglycan in smooth muscle within these organs; and 3) ␣-dystroglycan from fetal and adult kidney may be differentially glycosylated, but both isoforms bind laminin-1.
Our statements are based on the following observations. We did not detect ␣-, ␤-, ␥-, ␦-sarcoglycan or sarcospan in kidney or lung epithelial cells. Similar results were obtained in esophagus, bladder, and small intestine (data not shown). Although we identified sarcoglycan and sarcospan transcripts in the kidney, we were never able to detect corresponding polypeptides. By immunofluorescence, we found expression of these proteins in smooth muscle of large blood vessels only (data not shown). ␤-Sarcoglycan, ␦-sarcoglycan, and sarcospan present in lung were shown to be part of a smooth muscle dystroglycan complex as revealed by sucrose gradient fractionation. Two other dystroglycan-associated proteins are ⑀-sarcoglycan and utrophin. The former protein has been shown to be an integral component of smooth muscle sarcoglycan complex 1 and the latter an integral component of the dystroglycan complex found in neuromuscular junctions (55). Dystroglycan and ⑀-sarcoglycan are co-localized in embryonic kidney epithelium, and utrophin mRNA is also present in the embryonic kidney, hence these molecules could be associated with each other (19,36,53). Likewise, dystroglycan and utrophin are partially co-expressed in adult kidney epithelium (17,19). However, in adult kidney, dystroglycan, ⑀-sarcoglycan, and utrophin dissociate from each other during sucrose gradient fractionation, indicating that they are not tightly associated in adult kidney epithelial cells. Moreover, dystroglycan and utrophin are co-expressed in epithelial cells of the lung (see Ref. 17, see also Fig. 2), but we were also unable to demonstrate an association of these two proteins in lung epithelial cells. Nevertheless, we were able to demonstrate an association of ␤-dystroglycan, ␤-, ␦-, and ⑀-sarcoglycan, and utrophin in smooth muscle indicating that the extraction method applied here does not disrupt the interactions between the components of the smooth muscle dystroglycan complex. Weak interactions between ␤-dystroglycan, ⑀-sarcoglycan, and utrophin in epithelial cells could possibly become disrupted in our preparations, but also genetic evidence supports our interpretation: ␤-dystroglycan is retained in lung epithelial cells of utrophin null mice (17).
Also, dystroglycan and ⑀-sarcoglycan in the fetal kidney are not associated as revealed by sucrose gradient fractionation. Utrophin, however, may be associated with dystroglycan in fetal kidney. Although the major part of utrophin does not co-migrate with dystroglycan, as revealed by sucrose gradient fractionation, there is a weaker peak of utrophin in fractions 7 and 8, co-migrating with dystroglycan, preceding the peak of utrophin in the later fractions (see Fig. 2). It is interesting to note that as skeletal muscle matures utrophin is gradually replaced by dystrophin (56), and one possibility is that dystroglycan in mature non-muscle organs interacts with yet unidentified cytoskeletal proteins or some of the shorter dystrophin isoforms. For example, Dp140, a shorter dystrophin isoform, is expressed in a subset of both fetal and mature kidney epithelial cells (16,47). However, we have not been able to demonstrate an interaction between Dp140 and ␤-dystroglycan. 3 Still, it would be interesting to generate mice lacking all possible dystrophin and utrophin isoforms to study the roles of these proteins in epithelial cells.
␤-Dystrobrevin is another dystrophin-related protein that has been shown to be highly expressed in kidney, lung, and liver, but the subcellular localization of this protein is not known, and it has yet to be determined whether ␤-dystrobrevin Dystroglycan-rich sucrose gradient fractions from adult and fetal dystroglycan preparations were separated on 3-12% gradient gels and transferred to nitrocellulose and subsequently probed with IIH6 for a side-by-side comparison of the molecular weight of ␣-dystroglycan from adult and fetal kidney. Nitrocellulose transfers were also incubated with biotinylated laminin-1. Both the 156-and 120-kDa isoforms of kidney ␣-dystroglycan bind laminin-1. Molecular masses in kDa are indicated at left. is directly or indirectly associated with ␤-dystroglycan (57)(58)(59). The syntrophins are dystrophin-binding proteins that also have been shown to be ubiquitously expressed (60), and very recently the ␤2-syntrophin was found to be associated with dystrobrevin in Madin-Darby canine kidney cells (61). Whether this is also seen in epithelial cells in vivo remains to be determined.
In this report, we have also characterized epithelial dystroglycan by analyzing dystroglycan from kidney. ␣-Dystroglycan from fetal kidney has a molecular mass of 120 kDa, whereas the adult kidney ␣-dystroglycan has 156 kDa. These data suggest that ␣-dystroglycan is differentially glycosylated in fetal versus mature tissue. The molecular weights of ␣-dystroglycan in other tissues also vary. For example, ␣-dystroglycan in mammalian skeletal muscle is 156 kDa (39), 140 kDa in cardiac muscle (62), and 120 kDa in brain and peripheral nerves (24,38). There is evidence that ␣-dystroglycan is a sialylated mucin-type glycoprotein, and the structures of the sialylated Olinked oligosaccharides have been analyzed to some extent (63,64), but the biological relevance of these differentially glycosylated isoforms is presently unclear. All dystroglycan isoforms appear to bind at least laminin-1 and laminin-2 (6,18,38), but the different glycosylation patterns in different tissues could reflect plasticity of the receptor function.
In summary, we have shown the presence of two different dystroglycan complexes in non-skeletal muscle organs: one from epithelial cells, which is not associated with any of the known sarcoglycans, and one from smooth muscle cells which is associated with ␤-, ␦-, ⑀-sarcoglycan, sarcospan, and dystrophin/utrophin. There is yet the possibility that dystroglycan in epithelial cells is associated with hitherto unidentified molecules forming a similar complex as in skeletal and cardiac muscle, and this is currently being investigated.