A monoclonal antibody identifies a glycoprotein complex involved in cell-substratum adhesion

Experimental Cell Research 1.57(1985) 218-226

A Monoclonal Antibody Identifies a Glycoprotein Complex Involved in Cell-Substratum Adhesion KAREN A. KNUDSEN,’

ALAN F. HORWITZ* and CLAYTON A. BUCK’

‘The Wistar Institute of Anatomy and Biology, Philadelphia, and the ‘BiochemistrylBiophysics Department, University of Pennsylvania, Philadelphia, PA 19104, USA

The monoclonal antibody CSAT has been reported to perturb the adhesion of chick embryo cells to their substratum (Neff et al. [19]). Evidence is presented here that the antigen recognized by this monoclonal antibody is comprised of three membrane glycoproteins. The antigen is released from cells with non-ionic detergent and purified by monoclonal antibody atIinity chromatography. When analysed by SDS-PAGE under nonreducing conditions, the antigen resolves into three components of apparent molecular weights 160000 (band l), 135000 (band 2), and 110000 (band 3). Following reduction of each component, bands 1 and 2 migrate at slightly lower apparent molecular weights, while band 3 migrates at a higher apparent molecular weight, suggesting that band 3 has an internal disulfide bond. All three bands differ from one another as determined by peptide mapping and by immunologic cross-reactivity. It is postulated that the three glycoproteins function as a complex that plays a central role in cell-substratum adhesion. @ 1985 Academic Press, Inc.

Specialized molecules on the surface of cells serve to receive and transmit information to the cell’s interior. Some of these molecules are receptors that bind soluble stimuli such as hormones and growth factors, whereas others are molecules that interact with components fixed either to other cells or to extracellular matrices and serve to mediate cell adhesion. Together, these various types of interactions are critical to such processes as growth, differentiation, and organogenesis. Therefore, the identification of hormone receptors and surface membrane molecules involved in cell-cell and cell-substratum adhesion has become an important goal of cell biology. One approach to identify the membrane components involved in cellular adhesion has used antibodies that disrupt an adhesive process by binding to adhesion-related molecules on the cell surface. Initially, polyclonal antibodies were used to perturb cell-cell adhesion ([l-7], reviewed in [S]) and cell-substratum adhesion ([9-111, reviewed in [S]), and to isolate membrane-associated molecules postulated to be relevant to cellular adhesion [12-161. More recently, monoclonal antibodies have been employed [ 17-241. We describe here three adhesion-related membrane glycoproteins identitied using a previously described monoclonal antibody that disturbs cell-substratum adhesion in chick myoblasts and fibroblasts [19]. Copyright 0 1985 by Academic FTCSS, Inc. All rights of reproduction in any form reserved 0014-4827/85 503.00

Membrane

glycoproteins

MATERIALS Monoclonal

AND

in cell-substratum

adhesion

219

METHODS

Antibody

The monoclonal antibody, CSAT [19], was purified from mouse ascites fluid by ammonium sulfate precipitation [25] and protein A affinity chromatography as described previously [19]. The purified antibody was conjugated to cyanogen bromide (CNBr)-activated Sepharose 4B (Pharmacia Fine Chemicals, Piscataway, N.J.) according to the manufacturer’s instructions such that the antibody concentration was approx. 3 mg/ml CNBr-Sepharose 4B.

Cell Culture Chick embryo myoblasts and fibroblasts were prepared as described [26] and labelled with [3SS]methionine as described [19] or with [‘4C]n-glucosamine by adding R&200 uCi to the culture medium for 48-72 h. Under these conditions, the radioactivity is found predominantly in the pronaseresistant oligosaccharide portion of glycosylated cellular molecules [27].

Antigen Purification Antigen was extracted from cultured tibroblasts as described [ 191or from decapitated and eviscerated 11-day-old chick embryos. Embryos were homogenized for approx. 20-30 set at 4°C using a Sorvall Omni-mixer (DuPont Instruments) in 100 ml of 10 mM T&acetate, pH 8.0, and 0.5 mM Ca2+ containing 2 mM phenylmethylsulfonyl fluoride (PMSF). In one series of experiments, fibroblasts were exposed to 1 M urea, 3 M KCI, or 10 mM Tris-acetate, pH 8.0, in the presence of 2 mM PMSF for 15 min at 4°C prior to extraction with detergent. In all cases the cells were agitated in the detergent-containing buffer at 4°C for 15 min and then spun for 30 min at 15000 rpm in an SS34 head in the Sorvall RC2-B centrifuge. The supemate was precipitated with acetic acid and then with acetone as described previously [14], and applied to the immobilized antibody in buffer containing 10 mM Tris acetate, pH 8.0,0.5 % NP40, and 0.5 mM Ca” at 4°C and a flow rate of 0.3-0.5 ml/min. The column was washed with 15-20 column volumes of buffer and eluted with 50 mM diethylamine, pH 11.5. The antigen was detected by radioactive label or by protein determination [28], and, when required, was concentrated by precipitation overnight with a 6-fold excess of acetone at -20°C. Anion exchange chromatography and lectin aBinity chromatography were performed as previously described [14].

Gel Electrophoresis Samples were analysed by SDS-PAGE using 7% acrylamide gels according to the method of Laemmli [29]; when non-reduced gels were used, the reducing agent was omitted during sample preparation. Gels were stained with Coomassie Blue, impregnated with En3Hance (New England Nuclear) and exposed to Kodak XR-5 X-ray film at -60°C. Alternatively, the proteins were transferred to nitrocelhtlose paper overnight at 10°C and 400 mA in a phosphate buffer containing 10% methanol.

Peptide Maps AfBnity-purified antigen was labelled with “‘1 by the chloramine-T method. Individual bands were cut from preparative gels and peptides generated with 0.1% trypsin (Worthington Enzymes, Freehold, N.J.) or 0.01% a- chymotrypsin (Miles Laboratories) by the procedure of Cleveland et al. [30]. Peptides were separated in a 12% acrylamide resolving gel.

Characterization

by Enzyme Sensitivity

The mixture of endo- and exoglycosidases purified from Sfreptococcus pneumoniae and containing neuraminidase, #I-galactosidase, /I-N-acetylglucosaminidase, a-fucosidase, endo-/I-galactosidase, endo-/I-N-acetylglucosaminidase D, and endo-a-N-acetylgalactosaminidase was a gift of G. Ashwell, NIH. The preparation was reported to be protease-free (personal communication) and was inactive when assayed against the proteolytic substrate Azocoll (Sigma) using the manufacturer’s instructions. In addition, the same glycosidase preparation caused no proteolytic degradation of laminin when it was used by Howe [31] to remove 73 % of the carbohydrate residues from this molecule. Routinely, the preparation releases 60-80% of the r4C from [‘4C]ghtcosamine-labelled glycoproteins in an overnight incubation. Enzyme activity was added in 500-fold excess to the antigen and the mixture Exp Cell Res 157(1985)

220 Knudsen, Horwitz and Buck band 123

band l-

&

i c-J&

-116

2-

-116

3-

-94

- 94

-68

-68

-

43

-43

‘i, “-

Non-reduced

Reduced

Fig. 1. Autoradiogram of the SDS-PAGE analysis of monoclonal antibody affinity-purified antigen labelled metabolically with [35S]methionine and isolated from chick skeletal tibroblasts. Each of the three bands resolved under non-reducing conditions was excised and re-analysed by SDS-PAGE under reducing conditions. was incubated overnight at pH 6 and 37°C. PMSF (2 mM) was added as a precaution against possible protease activity contaminating the antigen preparation. Purified collagenase with no protease activity was purchased from Advance Biofactures Corp. (Lynbrook, N.Y .) and incubated with the antigen in lUO-fold excess for 3 h at 37°C. The products were analysed by SDS-PAGE under non-reducing conditions.

Amino Acid Analysis Antigen was isolated by immunoatfinity chromatography from decapitated eviscerated embryos. The SDS-PAGE profile of a typical preparation used for amino acid analysis is shown in fig. 3, lane B. Amino acids were analysed using a Beckman Model 121 amino acid analyser.

Preparation of Antisera to Monoclonal Antibody Affinity-Puri’ed

Antigen

Rabbits were injected with each of the proteins resolved by SDS-PAGE under non-reducing conditions and transferred to nitrocellulose paper. The proteins were detected by autoradiography and the individual bands excised, dissolved in DMSO, and mixed with Freund’s adjuvant prior to subcutaneous injection. Approx. 50 ug of antigen was injected at 2-week intervals. Antiserum to all three components was raised in a mouse by immunizing with the three bands eluted from gels.

Immunoblots Immunoblot analysis was accomplished following published procedures for “Western” blotting [32, 331. ‘251-Protein A was used to detect antibody binding to protein bands on the nitrocellulose paper.

RESULTS The monoclonal antibody, CSAT, was reported previously to round and detach chick embryo myoblasts from the substratum and to localize along portions of the Exp Cell Res 157 (1985)

Membrane

glycoproteins

in cell-substratum

adhesion

221

Fig. 2. Autoradiogram

of peptide maps generated from ‘251-labelled antigen purified from chick skeletal fibroblast. Bands 1, 2, and 3 (lanes 1, 2 and 3, respectively) were excised from gels and exposed to (a) trypsin, or (b) a-chymotrypsin, as described in Materials and Methods. The resulting fragments were separated in a 12% acrylamide resolving gel.

stress fiber on fibroblasts [193. Antigen purified from cultured fibroblasts by monoclonal antibody affinity chromatography blocked the effect of CSAT on myoblasts and after SD!%PAGE appeared as a heterogeneous population of peptides with an apparent molecular weight in the region of 140000 [19]. This antigen has been more extensively characterized and the results reported here. Monoclonal antibody affinity-purified antigen was prepared from [35S]methionine-labelled chick embryo fibroblasts and analysed by SDS-PAGE under nonreducing conditions. Under these conditions, the sample was resolved into three distinct bands having apparent molecular weights of 160000 (band l), 135000 (band 2), and 110000 (band 3) (fig. 1). After separating the bands under nonreducing conditions, each individual band was cut from the gel and analysed under reducing conditions (fig. 1). The apparent molecular weights of bands 1 and 2 decreased to 140000 and 115000, respectively, whereas the apparent molecular weight of band 3 increased to 124000. The origin of the lower molecular weight component observed when band 3 is reduced is not known. It may be a degradation product or a component released following treatment with 2-mercaptoethanol. These results not only explain the poor resolution of the antigen seen previously [19], but also demonstrate that band 3 contains internal disulfide bonds, Iodination and Coomassie Blue staining of non-reduced gels do not reveal additional bands. To determine whether the monoclonal antibody recognizes a similar antigenic site common to all three components, or whether the antibody binds to one band ExpCell

Res 157 (1985)

222 Knudsen,

Horwitz

and Buck

Table 1. Summary of the amino acid composition of the affinity-purified antigen (SDS-PAGE profile of typical antigen preparation shown in fig. 3, lane B) Amino acid Cysteic acid 3-OH pro1 Methionine sulfoxide 4-OH pro1 Aspartic Threonine Serine Glutamic Proline Glycine Alanine

Residues/1000

tr 76.53 62.28 55.28 111.08 72.24 93.90 87.15

Amino acid

Residues/1000

l/2 cystine Valine Methionine Isoleu Leucine Tyrosine Phenylala OH lys Lysine Histidine Arginine

2.93 12.78 44.42 48.77 85.92 35.44 45.80 87.98 25.62 51.85

with which the other components co-purify, an immunoblot was performed using the monoclonal antibody. The antibody did not bind to any of the three bands following their separation in this manner (not shown), indicating that the conditions of SDS-PAGE destroys the antigenic determinant and suggesting that the antigenic determinant is protein in nature and not carbohydrate. All attempts to separate the three components under non-denaturing conditions by molecular sieving, anion exchange, or lectin aflinity chromatography have been unsuccessful. Thus, it has not been possible to test directly the binding of the monoclonal antibody to the individual components. The three bands making up the antigenic complex are largely distinct and not related by proteolytic cleavage. This was demonstrated by the combined results of two independent experimental approaches, one biochemical and the second immunological. Following labelling of affinity-purified antigen with 1251,peptide maps were generated according to the method of Cleveland et al. [30]. Individual bands (1, 2 or 3) were separated by SDS-PAGE and cut from gels in the manner demonstrated in fig. 1. The bands were electrophoresed into gels containing either trypsin (fig. 2A) or chymotrypsin (fig. 2 B). Following incubation at room temperature (see Materials and Methods), electrophoresis was continued and the resulting peptides autoradiographed. The peptides generated from bands 1 and 3 by both enzymes are clearly distinct from one another. In both cases, the peptide map of band 3 contains two to three major peptides of higher molecular weight than those seen in digests of band 1. The possibility of some homology with band 2 still exists. However, the map of tryptic peptides of band 2 is missing the first major peptide seen in band 1 and does not contain two of the three major peptides seen with band 3 (fig. 2A). In addition, the peptides generated by chymotrypsin from band 2, although clustered in the same general region of the gel as those generated from band 1, appear to differ slightly in molecular weight (fig. 2 B). It should be noted that the analysis of band 2 is complicated by its position between Exp Cell Res 157 (198s)

Membrane

glycoproteins

in cell-substratum A

ABC

adhesion

223

B C

Mr

Mr

92.5 -

116-

9468-

: 68-

43-

j 3

4

Fig. 3. Western immunoblot analysis. Monoclonal antibody affinity-purified antigen isolated from decapitated, eviscerated chick embryos, was resolved into the three bands by SDS-PAGE under nonreducing conditions, transferred to nitrocellulose paper, and then exposed to dye or antiserum followed by ‘2SI-Protein A. A, Autoradiogram of a 1 : 50 dilution of rabbit antiserum raised against band 1; B, amido black stain; and C, autoradiogram of a 1 : 25 dilution of mouse antiserum raised against all three bands. Fig. 4. Autoradiograms of SDS-PAGE analysis under non-reducing conditions of monoclonal antibody afhnity-purified antigen isolated from chick skeletal libroblasts and A, metabolically labelled with [‘4C-n]glucosamine; B, metabolically labelled with [35S]methionine and incubated overnight in the absence of the mixed glycosidases but the presence of 2 mM PMSF; and C, labelled with [35S]methionine and incubated overnight at 37°C in the presence of the mixed glycosidase preparation plus 2 mM PMSF.

bands 1 and 3, making it difficult to excise from the gel without some crosscontamination. Further differences in bands 1, 2 and 3 were noted using antisera to bands eluted from non-reduced gels. A polyclonal antiserum was raised in a rabbit against band 1, which is the largest molecule and should therefore contain the greatest number of common antigenic determinants if bands 2 and 3 were to represent cleavage products of band 1. This antiserum stained only band 1 in an immunoblot (fig. 3, lane A). Furthermore, an antiserum raised in a mouse by injecting a mixture of gel bands 1, 2 and 3 reacted exclusively with band 3 in immunoblots (fig. 3, lane C). This latter result suggests not only that the three components are distinct but also that band 3 is the most immunogenic. A mixture of antisera raised in rabbits against the individual bands, on the other hand, reacts with all three bands and is able to round and detach myoblasts from their substratum (not shown). Thus, both biochemical and immunological data suggest bands 1, 2 and 3 are distinct molecules. Material recognized by the CSAT antibody was further characterized in several ways. Antigen purified from chick embryos (fig. 3, lane B) was subjected to 15-858333

Exp CeNRes 157(1985)

224 Knudsen,

Horwitz

and Buck

amino acid analysis (table 1). The absence of OH-lysine and OH-proline revealed in this analysis is consistent with the insensitivity of the antigen to collagenase (data not shown), suggesting that the polypeptides are not collagen-like. Glycosylation of all three proteins was demonstrated by metabolic labelling with [14C-~]glucosamine (fig. 4, lane A) and by a decrease in the apparent molecular weight of each component after incubation of the affinity-purified antigen with a proteasefree mixture of glycosidases (fig. 4, lane C). The additional lowest molecular weight band seen in the glycosidase-free control (fig. 4, lane B) likely resulted from a small amount of degradation during the overnight incubation at 37°C. Treatment of cells with 1 M urea, 3 M KC1 and hypotonic shock failed to release any of the three molecules from the surface. Therefore, the antigen appears to be integrally associated with the membrane.

DISCUSSION The monoclonal antibody, CSAT, and a Fab fragment of this antibody, rounds and detaches chick embryo myoblasts in vitro by binding to a cell surface antigen that can be released by detergent and purified by monoclonal antibody affinity chromatography. Analysis of the affinity-purified antigen by SDS-PAGE reveals three components, They appear to be cell surface molecules relevant to cell-substratum adhesion, since antisera taised against the bands eluted from gels perturb cellular adhesion and morphology. In addition, the antigen is comprised of molecules that are integrally associated with the membrane, since detergent. is required for their solubilization. The three bands seen following SDS-PAGE are all proteins, since they can be labelled metabolically with [35Slmethionine and are sensitive to trypsin and chymotrypsin. The absence of OH-lysine and OHproline implies that the proteins are not collagen-like and differ, therefore, from the 140000 MW adhesion-related glycoprotein, rich in OH-lysine and OH-proline, described by Carter & Hakomori [34, 351. All three peptides are labelled by [‘4C]glucosamine, which has been shown to be incorporated virtually exclusively into carbohydrate residues 1277,and are sensitive to mixed glycosidases, suggesting that these peptides are glycosylated. The presence of three glycoproteins following SDS-PAGE is not due to differences in glycosylation, since treatment with mixed glycosidases does not eliminate the differences in apparent molecular weight. This is in contrast to N-CAM, the neural cell-cell adhesion molecule, in which neuraminidase treatment eliminates heterogeneity in the immunoaffinity-purified antigen [17]. Several lines of evidence suggest that the three glycoproteins are distinctly different. First, while the behavior of all three components during SDS-PAGE is influenced by the reduction of disulfide bonds, only band 3 migrates more slowly after treatment with 2-mercaptoethanol, indicating that internal disulfide bonds are more critical to the conformation of band 3 than to that of bands f and 2. Second, both tryptic EXP Cell Ret 157 (1985)

Membrane

glycoproteins

in cell-substratum

adhesion

225

and chymotryptic peptide maps of the three bands differ, suggesting that each glycoprotein has a largely unique polypeptide backbone. Third, the three glycoproteins differ immunologically, i.e., a polyclonal antiserum raised against denatured band 1, the largest component, did not react with the two smaller bands in an immunoblot. Furthermore, an antibody that recognizes band 3 does not bind to bands 1 and 2. Bands 2 and 3 therefore do not appear to be proteolytic fragments of band 1. It is likely that the three glycoproteins exist as a complex in the membrane, since they co-purify, except under the denaturing conditions of SDS-PAGE. This possibility is further supported by the fact that all three bands migrate as a complex during sucrose density gradient centrifugation with a SZO,,+,value of 8.6 (Horwitz & Buck, unpublished observatoins). Assuming this interpretation, during antibody affinity chromatography the antibody would bind to a site on only one of the glycoproteins and the other two would co-purify due to their association with the bound glycoprotein. Alternatively, CSAT monoclonal antibody might bind to a site formed by the quaternary structure of the three polypeptides. The complex of glycoproteins recognized by CSAT monoclonal antibody plays an important role in cell-substratum adhesion in myoblasts and fibroblasts, since CSAT and polyclonal antisera against immunoaffinity-purified CSAT antigen perturb adhesion of these cells. Furthermore, localization studies show that CSAT staining co-aligns with portions of stress fibers and is particularly prominent just adjacent to vinculin-rich focal contact sites [19, 361. Thus, the antigen is in a location consistent with its proposed involvement in adhesion. Recently Chen & Singer [37] have proposed models for several types of chick fibroblast-substratum interactions based upon immunoelectron-microscopic observations. These contact sites can be distinguished on the basis of the distance between the membrane and the substratum, the composition of the substratum, and the organization of cytoskeletal elements at the point of cell-substratum interactions. They speculate that there exist at each contact site molecules which span the membrane and serve to bridge the extracellular matrix and the cytoskeleton. Oesch 8z Birchmeier [20], using a monoclonal antibody, FC-1, which delays the attachment of fibroblasts to a substratum, have identified a 60 000 MW membrane protein which may serve as such a bridge at focal contact sites. In their studies, Chen & Singer used a monoclonal antibody, JG-22, which perturbs the cell-substratum adhesion of myoblasts and immunoprecipitates a poorly resolved polypeptide or group of polypeptides of approx. 138000 MW [18]. Similar to CSAT, this antibody binds primarily adjacent to focal contact sites [37, 381at close contacts. Furthermore, we have found that the monoclonal antibodies JG22 and CSAT bind to the same antigen, which under the conditions reported here is resolved into three distinct glycoproteins. More recently, Hasegawa et al. [39] have demonstrated that the antigen reacting with IG22 also exists as a complex of three distinct glycoproteins with properties similar to those reported here for the CSAT antigen. This glycoprotein complex is therefore a strong Exp Cell Res 157 (1985)

226 Knudsen, Horwitz and Buck candidate for the transmembrane component that links the extracellular matrix and the cytoskeleton at the close contact adhesion sites. We gratefully acknowledge the excellent technical assistance of Elizabeth Hud-Broderick and Kathleen Kelly as well as the help of Helen Schorr and Marina Hoffman in preparing and editing the manuscript, respectively. Our special thanks to Dr Robert Alper of the Connective Tissue Research Institute (Philadelphia, Pa.) for performing the amino acid analysis and to Drs Tom O’Brien and Caroline Damsky for their valuable comments on the manuscript. This work was supported by grants CA-19144 , GM 23244, and CA-10815.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8.

Gerisch, G, Curr topics in dev biol 14 (1977) 359. Brackenbury, R, Thiery, J-P, Rutishauser, U L Edelman, G M, J biol them 252 (1977) 6835. Kemler, R, Babinet, C, Eisen, H & Jacob, F, Proc natl acad sci US 74 (1977) 4449. Urushihara, H, Ozaki, H S & Takeichi, M, Dev biol70 (1979) 206. Bertolotti, R, Rutishauser, U & Edelman, G M, Proc natl acad sci US 77 (1980) 4831. Ducibella, T, Dev bio179 (1980) 356. Damsky, C H, Knudsen, K A, Dorio, R J & Buck, C A, J cell bio189 (1981) 173. Darnsky, C H, Knudsen, K A & Buck, C A, Biology of glycoproteins (ed R Ivatt) p. 1. Academic Press, New York (1983). 9. Wylie, D E, Damsky, C H & Buck, C A, J cell biol 89 (1979) 385. 10. Tarone, G, Galletto, G, Prat, M & Comoglio, P, J cell bio194 (1982) 179. 11. Hsieh, P & Sueoka, N, J cell biol 86 (1980) 866. 12. Thiery, J-P, Brackenbury, R, Rutishauser, U & Edelman, G M, J biol them 252 (1977) 6841. 13. Hyatil, F, Babinet, C & Jacob, F, Cell 26 (1980) 447. 14. Knudsen, K A, Rao, P E, Damsky, C H & Buck, C A, Proc natl acad sci US 78 (1981) 6071. 15. Yoshida, C & Takeichi, M, Cell 28 (1982) 217. 16. Damsky, C H, Richa, J, Solter, D, Knudsen, K A & Buck, C A, Cell 34 (1983) 455. 17. Hoffman, S, Sorkin, B C, White, P C, Brackenbury, R, Mailhammer, R, Rutishauser, U, Cunningham, B A & Edelman, G M, J biol them 257 (1982) 7720. 18. Greve, J M & Gottlieb, D I, J cell biochem 18 (1982) 221. 19. Neff, N T, Lowrey, C, Decker, C, Tovar, A, Damsky, C, Buck, C & Horwitz, A F, J cell biol95 (1982) 654. 20. Oesch, B & Birchmeier, W, Cell 31 (1982) 671. 21. Gallin, W J, Edelman, G M & Cunningham, B A, Proc natl acad sci US 80 (1983) 1038. 22. Vollmers, H P & Birchmeier, W, Proc natl acad sci US 80 (1983) 3729. 23. Shirayoshi, Y, Okada, T S & Takeichi, M, Cell 35 (1983) 631. 24. Grumet M, Hoffman, S & Edelman, G M, Proc natl acad sci US 81 (1984) 267. 25. Jonak, Z L, Monoclonal antibodies (ed R H Kennett, T J McKeam & K B Bechtol) p. 405. Plenum Press, New York (1980). 26. Knudsen, K & Horwitz, A F, Dev biol 58 (1977) 328. 27. Buck, C A, Glick, M C & Warren, L, Biochemistry 9 (1970) 4567. 28. Lowry, 0 H, Rosebrough, N J, Farr, A L & Randall, R J, J biol them 193 (1951) 265. 29. Laemmli, U, Nature 227 (1970) 680. 30. Cleveland, D W, Fisher, S G, Kirschner, M W & Laemmli, U K, J biol them 252 (1977) 1102. 31. Howe, C C, Mol cell biol4 (1984) 1. 32. Towbin, H, Staehelin, T & Gordon, J, Proc natl acad sci US 76 (1979) 4350. 33. Bumette, W N, Anal biochem 112 (1981) 195. 34. Carter, W G & Hakomori, S, J biol them 256 (1981) 6953. 35. Carter, W G, J biol them 257 (1982) 3247. 36. Damsky, C H, Buck, C & Horwitz, A F, J cell bio197 (1983) 99a. Abstr. 37. Chen, W-T & Singer, J, J cell biol95 (1982) 205. 38. Chen, W-T, Chapman, A, Singer S J, Greve, J & Gottlieb, D, J cell biol91 (1981) 258a. Abstr. 39. Hasegawa, T, Hasegawa, E, Chen, W-T & Yamada, K M, J cell bio199 (1984) 165a. Abstr. Received April 26, 1984 Revised version received October 1, 1984 Exp CeliRes 157(1985)

Printed

in Sweden