- Email: [email protected]
Microfilament association of asgp-2, the concanavalin A-binding glycoprotein of the cell-surface sialomucin complex of 13762 rat mammary ascites tumor cells
Experimental
Cell Research 178 (1988) 211-223
Microfilament Association of ASGP-2, the Concanavalin A-Binding Glycoprotein of the Cell-Surface Sialomucin Complex of 13762 Rat Mammary Ascites Tumor Cells OLUSEYI Departments
A. VANDERPUYE, CORALIE A. CAROTHERS and KERMIT L. CARRAWAY’ of Anatomy and Cell Biology, and Biochemistry University of Miami School of Medicine, Miami,
and Molecular Fioridu 33101
CARRAWAY, Biology,
Microfilament-associated proteins and membrane-microfilament interactions are being investigated in microvilli isolated from 13762 rat mammary ascites tumor cells. “Phalloidin shift” analyses on velocity sedimentation gradients of Triton X-100 extracts of [‘HIglucosamine-labeled microvilli identified a 120-kDa cell-surface glycoprotein associated with the microvillar microfilament core. The identification was verified by concanavalin A (Con A) blots of one- and two-dimensional (2D) electrophoresis gels of sedimented microfilament cores. By 2D-electrophoresis and lectin analyses the I20-kDa protein appeared to be a fraction of ASGP-2, the major Con A-binding glycoprotein of the sialomucin complex of the 13762 cells. This identity was confhmed by immunoblot analyses using immunoblot-purified anti-ASGP-2 from anti-membrane serum prepared against microvillar membranes. Proteolysis of the microvilli with subtilisin or trypsin resulted in an increase in the amount of ASGP-2 associated with the microfilament cores. An increase was also observed with siahdase treatment of the microvilli, suggesting that negative charges, probably present on the highly sialated sialomucin ASGP-1 of the ASGP-IIASGP-2 sialomucin complex, reduce ASGP-2 association with the microfilament core. Proteolysis of isolated microvillar membranes, which contain actin but not microfilaments, also increased the association of ASGP-2 with a T&on-insoluble, actin-containing membrane fraction. Purified ASGP-2 does not bind to microfilaments in sedimentation assays. Since the T&on-insoluble membrane residue is enriched in an actin-containing transmembrane complex, which contains a different glycoprotein, we suggest that the ASGP-2 is binding indirectly via this complex to the microfilament core in the intact microvilli. @ 1988 Academic Press. Inc.
Plasma membrane-microfilament interactions are important in determining cell morphology and the organization of cell-surface components and may also play a role in transmembrane signaling [l-3]. Two types of interactions have been described: indirect and direct. For indirect interactions microfilaments are linked to the membrane via one or more peripheral components which are associated with integral membrane proteins. An example of an indirect linkage is the association of actin protofilaments with band 3 in the erythrocyte membrane via spectrin and ankyrin [4]. A number of other proteins have been implicated in such linkages, including band 4.1 of erythrocytes [5], a 1 IO-kDa protein of intestinal brush border microvilli [6], actin-binding protein in platelets [71, and vinculin and talin of adherens junctions [8]. Linkages involving direct interactions of microfila-
’ To whom reprint requests should be addressed. 211
Copyright ($3 1988 by Acadcmrc Press, Inc. All rights of reproduction in any form resewed col4-4827/88 $03 Ml
212
Vunderpuye,
Carraway,
and Carraway
ments with integral plasma membrane components have been less thoroughly studied, but evidence for such linkages has been presented for ascites tumor cell microvilli [9, lo], Dictyostelium plasma membranes [ 11, 121, basal lamina-cell interactions [13], and platelets [14, 151. One method for examining such interactions has been to extract cells or cellsurface fractions with nonionic detergent and examine the “cytoskeletal” residues for cell-surface glycoproteins. Using such procedures, several putative cytoskeleton-associated glycoproteins have been identified [ 16-191. One difliculty with this method is that glycoproteins may be nonspecifically sedimented with the cytoskeleton residue. To improve the specificity of these analyses, we have recently developed a method for identifying microfilament-associated proteins of isolated ascites tumor cell microvilli [20, 211. This procedure involves velocity sedimentation of microvilli extracted with nonionic detergent in the presence or absence of the microfilament-stabilizing toxin phalloidin. Phalloidin reduces breakdown of the microfilaments, causing them to migrate further into the velocity sedimentation gradients. Because of the specificity of phalloidin, proteins which comigrate with the microfilaments in both the presence and absence of the toxin must be associated with the microfilaments. Such associations may be either direct or indirect interactions with actin. To study associations of cell-surface glycoproteins with microfilaments, we have used sublines of the 13762 rat mammary adenocarcinoma [22]. Microvilli isolated from these cells provide an excellent system for investigating membranemicrofilament interactions, since they are a pure plasma membrane fraction, uncontaminated by intracellular organelles, and containing an intact microtilament core [23-251. Moreover, the microvilli have a relatively simple complement of cell-surface glycoproteins. since the surface of these cells is dominated by the sialomucin complex ASGP-l/ASGP-2 [26], which comprises about 1 o/o of the total cell protein. The components of the complex are the highly glycosylated (70 5%carbohydrate) sialomucin ASGP-1 and the concanavalin A (Con A)-binding glycoprotein ASGP-2. A third important cell-surface glycoprotein of the microvilli is CAG (cytoskeleton-associated glycoprotein) [16]. We have obtained CAG from microvillar membranes as a component of a stable transmembrane complex with actin [9]. ASGP-l/ASGP-2 can be linked to the transmembrane complex in the presence of Con A by a Con A bridge between ASGP-2 and CAG [27]. However, the question remains whether ASGP-2 can be linked to the microtilaments without the lectin bridge. In the present studies we have used the “phalloidin shift” procedure to identify glycoproteins associated with the microfilament core of the isolated ascites microvilli. By glucosamine labeling the major microfilament-associated glycoprotein appears by lectin and immunological analyses to be a fraction of the microvillar ASGP-2. Interestingly, the amount of the ASGP-2 associated with the microfilament core is increased by protease or sialidase treatment of the microvilli. Moreover, ASGP-2 associates with Triton residues of microvillar membranes containing transmembrane complex. Although the microvillar membranes have been depleted of microfilaments, the binding of ASGP-2 to the membrane residue
Microfilament
associution
oj’ ASGP-2
? 13
is similar to that found with the microfilament cores. Therefore we suggest that binding of ASGP-2 to the cores is indirect, possibly through the transmembrane complex. MATERIALS
AND
METHODS
:Mureriu/s. Ampholytes, 4-chloronaphthol, and horseradish peroxidase-goat anti-rabbit immunoglobulin G were obtained from Bio-Rad. Nitrocellulose membranes were from Schleicher & Schull. Peanut agglutinin (PNA), Con A, horseradish peroxidase, proteases, and horseradish peroxidase conjugates of Loffrs fetragonoiobus lectin were obtained from Sigma. Other lectins were from E-Y Laboratories. All other chemicals were reagent grade from Sigma. Anulyri.v of microJi/ament-ussociuted glycoproteins by phalloidin shift uelociry sedimentation cenfrijugarion. Microvilli were isolated from 13762 MAT-Cl ascites tumor cells labeled with glucosand absence of amine and extracted with T&on-Pipes-KC1 (TPK) buffer [20, 281 in the presence phalloidin as described previously 1201. Extracts were fractionated by velocity sedimentation sucrose density gradient centrifugation, and the fractions were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) [20] followed by fluorography to detect glucosamine-labeled glycoproteins 1261. Preparation of microfilumenr cores und Triton-soluble fractions of microuilli. Microfilament core and Triton-soluble fractions were isolated from microvilli (2-5 mgiml protein) extracted with 2% Triton X-100 in 20 mM Pipes, pH 6.8, 150 mM KC], 2 mM MgClz (PK buffer) by centrifugation at lOO.OOOg for I h. The pellet was washed four times with the same buffer containing 0.1 % Triton to yield the “microfilament core.” Cytoskeletons with a qualitatively similar composition could be obtained hy extracting microvilli with Triton in 0.15 M NaCI. 30 mM sodium phosphate, pH 7.4 (PBS). Anribndies to microuilkr membrane proteins and ASGP-2. Microvillar membranes were prepared from microvilli as previously reported [24]. ASCP-IIASGP-2 complex. isolated as previously described 1261. and ASGP-2, isolated by a modification (Hull. unpublished results) of a previously described method 1261, were prepared and kindly provided by Dr Steven Hull. Antibodies to microvillar membranes were obtained by injecting membranes subcutaneously in Freund’s complete adjuvant into female New Zealand white rabbits with booster injections in incomplete adjuvant at 2week intervals. The rabbits were bled weekly after the second injection. Specific antibodies to ASGP2 were “blot-purified” by adsorption of anti-membrane antiserum or anti-membrane immunoglobulin preparations by the method of Smith and Fisher [29]. Elecfrophoresis and blotting procedures. SDS-PAGE and two-dimensional isoelectric focusingSDS-PAGE were performed as previously described 19, 241. For lectin binding analyses on gels or immunoblotting with antisera, proteins were transferred from one- or two-dimensional gels using the Bio-Rad Transblot system according to the manufacturer’s recommendations. Briefly, SDS gels were blotted in 5 mM ethanolamine, 25 mM glycine, 20% methanol, pH 9.5, for 16 h at 30 V (0. I2 mamp) and I h at 60 V (0.33 mamp). Two-dimensional gels were equilibrated in transfer buffer by three lomin washes and blotted as above. Nitrocellulose replicas were “blocked” with 20 mM Tris, 500 mM NaCl containing 0.05% Tween 20 (TBS-Tween) for 2 h and then processed for antibody or lectin staining. For Con A blots the lectin (50 kg/ml) in TBS containing 10 mg/ml bovine serum albumin (BSA) was incubated with the replica for 2 h at room temperature and washed four times with TBS-Tween (IO min each) and twice with TBS. The replicas were incubated for 2 h with horseradish peroxidase (50 kg/ml) in TBS with 10 mg/ml BSA, washed as described above, and treated with 4chloronaphthol to reveal the lectin-binding bands. Other lectins were used as direct conjugates with hor\eradish peroxidase at concentrations of 48 ~giml. For all lectins binding was abolished in the presence of the appropriate inhibitory saccharide. For immunoblots the replicas were incubated with 100 to 250-fold dilutions of antiserum or immunoglobulin preparations in TBS-BSA for 2 h, washed as described above. and incubated with horseradish peroxidase-conjugated antibodies to rabbit immunoglobulin for 2 h. Blots were washed and reactive bands revealed with 2-chloronaphthol as above. Proteolysis ofmicrouilli. Microvilli (5 mgiml protein), which are right-side out and sealed [24], were incubated with l-100 ugiml protease in PK or PBS buffer for I6 h at 20°C. Appropriate protease inhibitors were added to stop the reaction: phenylmethylsulfonyl fluoride (PMSF). 0.3 mM; leupeptin, 12.5 ug/ml; aprotinin, IO &ml; antipain, IO kg/ml; benzamidine, IO ug/ml. The proteolyzed microvilli were sedimented at 100,000~ for I h, washed twice in the same buffer with protease inhibitors. and used for preparation of microfilament cores and supernates as described above.
214
Vanderpuye,
Carraway,
and Carraway
Sialidase digestion of microvilli. Microvilli (1.8 mg protein) were incubated with Vibrio cholerae sialidase (2 units) in I ml of PK buffer containing 1 mM PMSF and 40 ug/ml each of bacitracin, benzamidine, and aprotinin for 20 min at 37°C. Microvilli were then sedimented at 10,OOOgfor 20 min, washed once with PK buffer, and extracted with Triton X-100 to prepare microfilament core and Triton-soluble fractions. Subtilisin digestion of microvillar membranes. Microvillar membranes (2 mg/ml protein) were treated with subtilisin as described above for microvilli. Triton residues and soluble fractions were prepared by centrifugation at 100,OOOgfor 1 h.
RESULTS Phalloidin
Shift Analysis
of Microfilament-Associated
Glycoproteins
Our previous studies have shown that microvillar microfilament-associated proteins could be identified by velocity sedimentation analysis of Triton extracts of microvilli prepared in the presence and absence of phalloidin, the microtilament-stabilizing toxin [20, 211. To identify microvillar microfilament-associated glycoproteins, microvilli were isolated from [3H]glucosamine-labeled cells, as described previously [24], and analyzed by this phalloidin shift procedure (Fig. 1). Fluorography of the SDS-PAGE gels from the velocity sedimentation gradients showed a major glucosamine-labeled protein with an apparent molecular weight of 120 kDA (Fig. IB) in both untreated and phalloidin-treated samples. For simplicity only the phalloidin-treated samples are shown. The 120-kDa glycoprotein comigrated with ASGP-2 of the sialomucin complex [26]. Lectin Analysis of Microfilament-Associated
Glycoproteins
Having observed a glucosamine-labeled glycoprotein which comigrated with ASGP-2, we sought a simpler, more sensitive method for analysis. For this purpose Tiiton extracts of the microvilli were sedimented at 100,OOOg to obtain microfilament “cores” [28]. Since ASGP-2 has been shown to bind Con A [26, 27, 301, SDS-PAGE gels of the cores were transferred to nitrocellulose and stained with Con A-peroxidase. As observed for the phalloidin shift analyses, the 120-kDa glycoprotein was the predominant microfilament core-associated species detected (Figs. 2 A, B). A second glycoprotein at 80 kDa was also observed in the core fraction. The Con A blotting procedure was used to analyze two-dimensional isoelectric focusing-SDS-PAGE gels of the microfilament core (Fig. 20 and showed that the 120-kDa glycoprotein comigrated with ASGP-2, the major Con A-binding glycoprotein of the microvilli and a component of the cell-surface sialomucin complex of the ascites cells. The results described above suggest that the 120-kDa glycoprotein may be a fraction of the ASGP-2 of the microvilli which is associated with the microfilament core. To determine whether the microfilament-associated 120-kDa component might be a form of ASGP-2, we first performed comparative lectin analyses of the microfilament cores. By lectin blotting identical reactivities were shown for ASGP-2 and the 120-kDa glycoprotein using a series of eight lectins with different specificities. Both the 120-kDa glycoprotein and ASGP-2 reacted with Con A, Pisum sativum lectin, Bandeiraea simplicifolia-I lectin, soybean agglutinin, and
Microfilament
association
of ASGP-2
215
4120K
-Actin
-120K
1 2 3 4 5 6 7 8 9 10 1112 13 Fig. 1. Phalloidin shift analysis of Triton extracts of microvilli. Microvilli from glucosamine-labeled 13762 MAT-Cl cells were lysed in TPK buffer and the extracts were fractionated by velocity sedimentation sucrose density gradient centrifugation in the presence of phalloidin [20, 211. (A) Coomassie blue-stained gel of fractions from velocity sedimentation gradients from top to bottom (left to right) of gradients. (B) Fluorography of gel showing the glucosamine-labeled glycoproteins in the corresponding fractions.
wheat germ agglutinin, but failed to react with peanut agglutinin, Lotus tetragonolobus lectin, and Limulus polyphenus lectin. Analyses with Antisera
Prepared
against Microvillar
Membranes
To address the question of the identity of the 120-kDa glycoprotein and ASGP2, we used antiserum prepared against microvillar membranes. Immunoblots of microvilli using anti-membrane antibodies (Fig. 3B, lane I) showed a strong reaction in the ASGP-2 region of the gel and in the region of components of XL55 kDa. No reaction was detected with ASGP-1, which is the major membrane glycoprotein but which is highly sialylated and weakly immunogenic. Microvillar Triton extracts (Fig. 3 B, lane 2) and microfilament cores (Fig. 3 B, lane 3) were also examined by immunoblot analysis with the anti-membrane serum. The staining pattern for the microvillar extracts was essentially the same as that for U-888340
216
Vanderpuye,
Carraway,
and Carraway
ASGP-2-
42
Fig. 2. Con A-binding proteins of Triton-soluble (lane I) and microfilament core (lane 2) fractions of MAT-Cl microvilli. (A) Con A blot; (B) Coomassie blue stain. The position of ASGP-2 is noted by the arrow in (A) and by the top connecting line between the panels. Arrowhead notes IO-kDa component. Molecular weight markers (kDa) are shown to the right of (B). (C) A Con A blot of the twodimensional isoelectric focusing-SDS-PAGE gel of the microfilament cores. The migration position in the second dimension is shown for ASGP-2.
the microvilli. Interestingly, the major stained band in the microfilament cores corresponded in migration to the I20-kDa component or ASGP-2 (Fig. 3B, lane 3).
These results suggest that the 120-kDa glycoprotein associated with the microfilament core may be a form of ASGP-2. Therefore we purified ASGP-2 from the sialomucin complex [26] for analysis of its reactivity with anti-membrane serum and purification of anti-ASGP-2 antibody. Immunoblots with anti-membrane serum showed a strong reaction against ASGP-2 (Fig. 3 c). These results clearly show that the antiserum contains antibodies against ASGP-2, defined by its
Microfilament B
A
C
of ASGP-2
association D
217
E
ASGP-2-
1
42
‘30
123
1 2 3
2
3
Fig. 3. Immunoblots of microvilli (lane I), T&on-soluble microvillar proteins ament cores (lane 3) with preimmune serum (A), anti-membrane antibodies purified anti-ASGP-2 (D). (0 An immunoblot of purified ASGP-2 stained with and used for immunoblot purification of anti-ASGP-2. (E) Amido black-stained
(lane 2), and microfil(B and C), and blotanti-membrane serum replica of microvilli.
purification from the sialomucin complex [26], rather than against some comigrating component. These immunoblots were then used to prepare ASGP-2-specific antibodies by immunoaffinity purification, eluting the antibodies from the immunoblots. The purified antibody was used to stain immunoblots of the microfilament cores (Fig. 3 D, lane 3) and supernates (Fig. 3 D, lane 2) from Triton extracts of microvilli. Bands were observed in the 120-kDa region of the gels for both Triton supernates and microfilament cores. These studies strongly suggest that the microfilament core-associated 120-kDa glycoprotein is a fraction of the microvillar ASGP-2 and that ASGP-2 must be one of the more immunogenic proteins in the microvillar membranes. Proteolytic
Digestion
of Microvilli
To investigate relationship between the 120-kDa glycoprotein and ASGP-2, microvilli, which are right-side out and sealed [24], were treated with proteases. Preliminary experiments were performed to establish the appropriate proteases, time, and concentrations for digestion. Figure 4 shows the effect of subtilisin treatment. For these experiments the protease-treated microvilli were extracted with Triton and fractionated by centrifugation. The Triton supernates and microfilament cores were analyzed by SDS-PAGE followed by transfer to nitrocellulose for blotting with anti-ASGP-2. Surprisingly, immunoblot analysis of these cores with specific anti-ASGP-2 indicates that the amount of ASGP-2 associated with the microfilament cores is increased with subtilisin proteolysis (Fig. 4). The effects of proteolysis on the association of ASGP-2 with the cores were verified using trypsin (data not shown). To confirm that the pelleted ASGP-2 is associated with microfilaments after proteolysis, we fractionated Triton extracts of subtilisin-treated microvilli by
218 Vanderpuye,
Carraway,
and Carraway
A
6
Mr
x lo3
120
1
2
3
123
Fig. 4. Effect of subtilisin treatment on ASGP-2 association with the microfdament core. ASGP-2 was detected by immunoblotting with blot-purified anti-ASGP-2. (A) Biton-soluble microvillar components; (B) microfilament cores. Lanes l-3 contain samples from treatments with 0, 1, and 10 &ml of subtilisin, respectively.
velocity sedimentation and analyzed the gradient fractions by immunoblotting with anti-ASGP-2. ASGP-2 was readily detected in the microfilament fractions from the gradient of the subtilisin-treated (Fig. 5B), but not the untreated (Fig. 5A), microvilli. These results clearly show that proteolysis induces an increased association of ASGP-2 with the microfilament core. Effect of Sialidase Digestion of Microvilli on the Association of ASGP-2 with the Microfilament Core One possible reason for the increased association of ASGP-2 with the microfilament core after proteolysis is that the association of ASGP-1 with ASGP-2 inhibits association of ASGP-2 with the core. We have shown previously that proteolysis removes most of the ASGP-1 [31, 321. This inhibition might be caused by steric hindrance, conformational changes in ASGP-2, or charge effects due to the high negative charge of highly sialylated ASGP-1 [33]. Preliminary experiments indicated that a number of proteases cleaved ASGP-1 from microvilli, with concomitant increases in ASGP-2 association with the microfilament core (data not shown). To investigate specifically the possibility that charge effects contribute to the association, we treated microvilli with sialidase and analyzed microtilament core and Tiiton-soluble fractions by blotting SDS-PAGE gel transfers with anti-membrane serum. As shown in Fig. 6, sialidase treatment resulted in a clear increase in ASGP-2 associated with the microtilament core. The results of the sialidase treatments support a model in which charge effects play a role in the association of ASGP-2 with the microfilament core and suggest that ASGP-1 may be inhibiting this association through its interaction with ASGP-2.
Microfilament
association
of ASGP-2
219
A
1234567891011 B
1234567891011 Fig. 5. Velocity sedimentation analysis of ASGP-2 association with the microfilament core of subtilisin-treated (B) and untreated (A) microvilli. Note association of ASGP-2, detected with antiASGP-2, with microfilament core fractions from gradient in (B).
Effect of Subtilisin Digestion of Membranes of ASGP-2 with Triton Residues
on the Association
An important question concerning the association of cell-surface glycoproteins with microfilaments is whether the association is direct or indirect. To try to answer this question, we prepared microvillar membranes, which are depleted in microfilaments, but still contain membrane-associated actin [24, 2.51. The membranes were treated with subtilisin, extracted with Triton X-100, and centrifuged. We have previously shown that these extraction and centrifugation conditions sediment from MAT-Cl microvillar membranes primarily a transmembrane complex of actin, CAG, and a 5%kDa protein [9]. In membranes treated with Con A, these conditions also pellet the ASGP-UASGP-2 complex, which is linked to the transmembrane complex by a Con A bridge [27]. When microvillar membranes were treated with increasing concentrations of subtilisin, extracted with Triton, and sedimented, immunoblots of the pellets showed increased amounts of ASGP2 with increasing subtilisin concentrations (Fig. 7). These results indicate that the presence of microlilaments is not necessary for association of ASGP-2 with a Triton-insoluble residue fraction.
220
Vunderpuye,
Carraway,
and Carraway
1 Fig. 6. Effect of sialidase digestion blot with anti-membrane antiserum. position of ASGP-2.
2
on ASGP-2 association Lane I, untreated;
with the microfilament lane 2, sialidase-treated.
core. ImmunoArrow shows
Direct Binding Assays for ASGP-2 with MicrojXuments To determine whether ASGP-2 could be binding directly to microtilaments, we incubated purified ASGP-2 with actin under actin polymerizing conditions and sedimented the microfilaments formed. Analyses of the pellets by SDS-PAGE and either staining or lectin blotting showed no evidence of ASGP-2 associated with the microfilaments (data not shown), providing further evidence that the interaction of ASGP-2 with the microfilament core is indirect.
t
ASGP-2
1234 Fig. 7. Effect of subtilisin digestion of microvillar residues of membranes. Membrane residues were serum. Lanes 14, 0, I. IO and 100 ugiml subtilisin.
membranes on ASGP-2 analyzed by immunoblots respectively.
association with Triton with anti-membrane
Microflament
association
of ASGP-2
221
DISCUSSION In many cell types and cell-surface domains regulation of the topography of cell-surface molecules appears to be determined by microtilament-membrane interactions [3]. Thus it is important to identify cell-surface molecules which are able to associate with submembrane microfilaments. The 13762 microvilli are an excellent system for investigating this problem. The microtilament core of these microvilli is more loosely organized than that of intestinal brush border microvilli 13, 24, 281 and also has microtilament-associated proteins more typical of the terminal web region of brush border than of the brush border microvilli [23, 24, 281. Thus the 13762 microvilli are more similar in composition and organization to the cortical region of the cell than to highly structured microvilli. We have been using these microvilli as a model system for studying how microfilaments associate with membranes. In previous studies we identified a major microfilament-membrane interaction site to be the transmembrane complex [9] of CAG (cytoskeleton-associated glycoprotein) [16] and actin. In the present studies, using the phalloidin shift method of identifying microfilamentassociated proteins, we found a major cell-surface glycoprotein associated with the microvillar microfilament core. Surprisingly, this glycoprotein was identified as a fraction of ASGP-2, a component of the sialomucin complex, which dominates the surface of the 13762 cells. Most of the ASGP-2 of the microvilli is solubilized by the Triton extraction procedure used to prepare microfilament cores. Interestingly, the distribution of ASGP-2 between the soluble and core fractions can be altered by protease or sialidase treatment of the microvilli before extraction. Only a small fraction of the ASGP-2 is normally associated with the microfilament core. This fraction can be increased by proteolysis or sialidase treatments, but it is still a relatively minor fraction of the total ASGP-2. Several observations suggest that association of ASGP-2 with the microfilaments is indirect. (a) Proteolysis of microvillar membranes, which are depleted in microfilaments, still leads to an increase in the association of ASGP-2 with a sedimentable residue. This residue contains primarily transmembrane complex [9]. (b) Sialidase treatment of microvilli leads to increase of the association of ASGP-2 with the microtilament core. This result indicates an inhibitory effect of charge on the interaction. Since sialidase removes only residues on the external surface, this result also suggests that the interaction occurs at the external surface. (c) Purified ASGP-2 does not bind to microtilaments. Based on the arguments above, the most likely component for the interaction with ASGP-2 is CAG, which is the only component of the transmembrane complex present at the external membrane surface 116, 241. What is the significance of the binding of ASGP-2 to the microfilament core? ASGP-2, which is present at the cell surface as a complex with the sialomucin ASGP-I [26, 301, is the major Con A-binding protein of the 13762 ascites cell surface [26]. When 13762 ascites cells are capped with fluorescent Con A, the sialomucin complex is the major species contributing to the redistribution of the fluorescence [30]. We have previously shown that capping in these cells is initiated by formation of a Con A bridge between ASGP-2 and CAG of the
222 Vanderpuye,
Carraway,
and Carrawq
transmembrane complex [27]. This observation explains how Con A capping can be directed by linkage to microfilaments. However, this mechanism does not explain capping of molecules by crosslinking agents, such as antibodies, which do not bind directly to a transmembrane complex. We suggest that the binding of a fraction of ASGP-2 to the core may offer an explanation for this second type of capping. Perhaps glycoproteins of transmembrane complexes contain a site which can bind to many other cell-surface molecules. The binding would not have to be of high affinity because the clustering of the cell-surface component caused by crosslinking with antibodies or lectins should stabilize the aggregates. What type of binding may be involved is not known, but the mechanism of interaction of ASGP-2 with the microfilament core, probably through CAG of the transmembrane complex, is of great interest in this regard. We thank Maria Carvajal for excellent technical assistance and Dr S. Hull for providing samples of sialomucin complex and purified ASGP-2. This work was supported by grants from the National Institutes of health (GM 33795 and CA 3 1695) and by the Papanicolaou Comprehensive Cancer Center of the University of Miami (CA 14395).
REFERENCES 1. 2. 3. 4. 5. 6.
Geiger, B. (1983) Biochim. Biophys. Acta 737, 305. Carraway, K. L., and Carraway, C. A. C. (1985) Bioessays 1, 55. Carraway, K. L., and Carraway, C. A. C., unpublished. Bennett, V. (1985) Annu. Rev. Biochem. 54, 273. Anderson, R. A., and Lovrien, R. E. (1984) Nature (London) 307, 655. Matsudaira, P., and Burgess, D. R. (1979) J. Cell Biol. 83, 667. 7. Fox, J. (1985) J. Biol. Chem. 260, 11,970. 8. Horwitz, A., Duggan, K., Buck, C., Beckerle, M. C., and Burridge, K. (1986) Nature (London) 320, 531.
Carraway, C. A. C., Jung, G., and Carraway, K. L. (1983) Proc. N&l. Acad. Sci. USA 80, 430. 10. Carraway, C. A. C., Jung, G., and Carraway, K. L. (1983) Cell. Motil. 3, 491. 11. Luna, E. J., Goodloe-Holland, C. M., and Ingalls, H. M. (1984) J. Cell Biol. 99, 58. 12. Wuestehube, L. J., and Luna, E. J. (1987) J. Cell Biol. 105, 1741. 13. Brown, S. S., Malinoff, H. L., and Wicha, M. S. (1983) Proc. Natl. Acad. Sci. USA 80, 5927. 14. Painter, R. G., Gaarde, W., and Ginsberg, M. H. (1985) J. Cell. Biochem. 27, 277. 15. Painter, R. G., Prodouz, K. N., and Gaarde, W. (1985) J. Cell Biol. 100, 652. 16. Carraway, C. A. C., Jung, G., Craik, J. R., Rubin, R. W., and Carraway, K. L. (1983) Exp. Cell 9.
Res. 143, 303. 17. MOSS, D. J. (1983) Eur.
18. 19. 20. 21. 22.
J. Biochem.
135, 291.
Painter, R. G., and Ginsberg, M. (1982) J. Cell Biol. 92, 565. Sheterline, P., and Hopkins, C. R. (1981) J. Cell Biol. 90, 743. Carraway, C. A. C., and Weiss, M. (1985) Exp. Cell Res. 161, 150. Carraway, C. A. C., Sindler, C., and Weiss, M. (1986) Biochim. Biophys. Acta 885, 68. Carraway, K. L., Doss, R. C., Huggins, J. W., Chesnut, R. W., and Carraway, C. A. C. (1979) J. Cell Biol.
83, 529.
23. Carraway, K. L., Huggins, J. W., Cerra, R. F., Yeltman, D. R., and Carraway, C. A. C. (1980) Nature
(London)
285,
508.
24. Carraway, C. A. C., Cerra, R. F., Bell, P. B., and Carraway, K. L. (1982) Biochim. Acra
25. 26. 27. 28. 29.
719,
Biophys.
126.
Carraway, K. L., Cerra, R. F., Jung, G., and Carraway, C. A. C. (1982) J. Cell Biol. 94, 624. Sherblom, A. P., and Carraway, K. L. (1980) J. Biol. Chem. 255, 12,051. Jung, G., Helm, R. M., Carraway, K. L., and Carraway, C. A. C. (1984) J. Cell Biol. 98, 179. Carraway, C. A. C., Jung, G., Hinkley, R. E., and Carraway, K. L. (1985) Exp. Cell Res. 115,71. Smith, D. E., and Fisher, P. A. (1984) J. Cell Biol. 99, 20.
Microfilament 30. 31. 32. 33.
Helm, R. Buck, R. Spielman, Sherblom,
association
of ASGP-2
M., and Carraway, K. L. (1981) Exp. Cell Res. 135, 418. L., Sherblom, A. P., and Carraway, K. L. (1979) Arch. Biochem. Biophys. J., Rockley, N. L., and Carraway, K. L. (1987) J. Biol. Chem. 262, 269. A. P., Buck, R. L., and Carraway, K. L. (1980) J. Biol. Chem. 255, 783.
223
198, 12.
Received December 12, 1987 Revised version received April 15, 1988
Printed in Sweden
Cell Research 178 (1988) 211-223
Microfilament Association of ASGP-2, the Concanavalin A-Binding Glycoprotein of the Cell-Surface Sialomucin Complex of 13762 Rat Mammary Ascites Tumor Cells OLUSEYI Departments
A. VANDERPUYE, CORALIE A. CAROTHERS and KERMIT L. CARRAWAY’ of Anatomy and Cell Biology, and Biochemistry University of Miami School of Medicine, Miami,
and Molecular Fioridu 33101
CARRAWAY, Biology,
Microfilament-associated proteins and membrane-microfilament interactions are being investigated in microvilli isolated from 13762 rat mammary ascites tumor cells. “Phalloidin shift” analyses on velocity sedimentation gradients of Triton X-100 extracts of [‘HIglucosamine-labeled microvilli identified a 120-kDa cell-surface glycoprotein associated with the microvillar microfilament core. The identification was verified by concanavalin A (Con A) blots of one- and two-dimensional (2D) electrophoresis gels of sedimented microfilament cores. By 2D-electrophoresis and lectin analyses the I20-kDa protein appeared to be a fraction of ASGP-2, the major Con A-binding glycoprotein of the sialomucin complex of the 13762 cells. This identity was confhmed by immunoblot analyses using immunoblot-purified anti-ASGP-2 from anti-membrane serum prepared against microvillar membranes. Proteolysis of the microvilli with subtilisin or trypsin resulted in an increase in the amount of ASGP-2 associated with the microfilament cores. An increase was also observed with siahdase treatment of the microvilli, suggesting that negative charges, probably present on the highly sialated sialomucin ASGP-1 of the ASGP-IIASGP-2 sialomucin complex, reduce ASGP-2 association with the microfilament core. Proteolysis of isolated microvillar membranes, which contain actin but not microfilaments, also increased the association of ASGP-2 with a T&on-insoluble, actin-containing membrane fraction. Purified ASGP-2 does not bind to microfilaments in sedimentation assays. Since the T&on-insoluble membrane residue is enriched in an actin-containing transmembrane complex, which contains a different glycoprotein, we suggest that the ASGP-2 is binding indirectly via this complex to the microfilament core in the intact microvilli. @ 1988 Academic Press. Inc.
Plasma membrane-microfilament interactions are important in determining cell morphology and the organization of cell-surface components and may also play a role in transmembrane signaling [l-3]. Two types of interactions have been described: indirect and direct. For indirect interactions microfilaments are linked to the membrane via one or more peripheral components which are associated with integral membrane proteins. An example of an indirect linkage is the association of actin protofilaments with band 3 in the erythrocyte membrane via spectrin and ankyrin [4]. A number of other proteins have been implicated in such linkages, including band 4.1 of erythrocytes [5], a 1 IO-kDa protein of intestinal brush border microvilli [6], actin-binding protein in platelets [71, and vinculin and talin of adherens junctions [8]. Linkages involving direct interactions of microfila-
’ To whom reprint requests should be addressed. 211
Copyright ($3 1988 by Acadcmrc Press, Inc. All rights of reproduction in any form resewed col4-4827/88 $03 Ml
212
Vunderpuye,
Carraway,
and Carraway
ments with integral plasma membrane components have been less thoroughly studied, but evidence for such linkages has been presented for ascites tumor cell microvilli [9, lo], Dictyostelium plasma membranes [ 11, 121, basal lamina-cell interactions [13], and platelets [14, 151. One method for examining such interactions has been to extract cells or cellsurface fractions with nonionic detergent and examine the “cytoskeletal” residues for cell-surface glycoproteins. Using such procedures, several putative cytoskeleton-associated glycoproteins have been identified [ 16-191. One difliculty with this method is that glycoproteins may be nonspecifically sedimented with the cytoskeleton residue. To improve the specificity of these analyses, we have recently developed a method for identifying microfilament-associated proteins of isolated ascites tumor cell microvilli [20, 211. This procedure involves velocity sedimentation of microvilli extracted with nonionic detergent in the presence or absence of the microfilament-stabilizing toxin phalloidin. Phalloidin reduces breakdown of the microfilaments, causing them to migrate further into the velocity sedimentation gradients. Because of the specificity of phalloidin, proteins which comigrate with the microfilaments in both the presence and absence of the toxin must be associated with the microfilaments. Such associations may be either direct or indirect interactions with actin. To study associations of cell-surface glycoproteins with microfilaments, we have used sublines of the 13762 rat mammary adenocarcinoma [22]. Microvilli isolated from these cells provide an excellent system for investigating membranemicrofilament interactions, since they are a pure plasma membrane fraction, uncontaminated by intracellular organelles, and containing an intact microtilament core [23-251. Moreover, the microvilli have a relatively simple complement of cell-surface glycoproteins. since the surface of these cells is dominated by the sialomucin complex ASGP-l/ASGP-2 [26], which comprises about 1 o/o of the total cell protein. The components of the complex are the highly glycosylated (70 5%carbohydrate) sialomucin ASGP-1 and the concanavalin A (Con A)-binding glycoprotein ASGP-2. A third important cell-surface glycoprotein of the microvilli is CAG (cytoskeleton-associated glycoprotein) [16]. We have obtained CAG from microvillar membranes as a component of a stable transmembrane complex with actin [9]. ASGP-l/ASGP-2 can be linked to the transmembrane complex in the presence of Con A by a Con A bridge between ASGP-2 and CAG [27]. However, the question remains whether ASGP-2 can be linked to the microtilaments without the lectin bridge. In the present studies we have used the “phalloidin shift” procedure to identify glycoproteins associated with the microfilament core of the isolated ascites microvilli. By glucosamine labeling the major microfilament-associated glycoprotein appears by lectin and immunological analyses to be a fraction of the microvillar ASGP-2. Interestingly, the amount of the ASGP-2 associated with the microfilament core is increased by protease or sialidase treatment of the microvilli. Moreover, ASGP-2 associates with Triton residues of microvillar membranes containing transmembrane complex. Although the microvillar membranes have been depleted of microfilaments, the binding of ASGP-2 to the membrane residue
Microfilament
associution
oj’ ASGP-2
? 13
is similar to that found with the microfilament cores. Therefore we suggest that binding of ASGP-2 to the cores is indirect, possibly through the transmembrane complex. MATERIALS
AND
METHODS
:Mureriu/s. Ampholytes, 4-chloronaphthol, and horseradish peroxidase-goat anti-rabbit immunoglobulin G were obtained from Bio-Rad. Nitrocellulose membranes were from Schleicher & Schull. Peanut agglutinin (PNA), Con A, horseradish peroxidase, proteases, and horseradish peroxidase conjugates of Loffrs fetragonoiobus lectin were obtained from Sigma. Other lectins were from E-Y Laboratories. All other chemicals were reagent grade from Sigma. Anulyri.v of microJi/ament-ussociuted glycoproteins by phalloidin shift uelociry sedimentation cenfrijugarion. Microvilli were isolated from 13762 MAT-Cl ascites tumor cells labeled with glucosand absence of amine and extracted with T&on-Pipes-KC1 (TPK) buffer [20, 281 in the presence phalloidin as described previously 1201. Extracts were fractionated by velocity sedimentation sucrose density gradient centrifugation, and the fractions were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) [20] followed by fluorography to detect glucosamine-labeled glycoproteins 1261. Preparation of microfilumenr cores und Triton-soluble fractions of microuilli. Microfilament core and Triton-soluble fractions were isolated from microvilli (2-5 mgiml protein) extracted with 2% Triton X-100 in 20 mM Pipes, pH 6.8, 150 mM KC], 2 mM MgClz (PK buffer) by centrifugation at lOO.OOOg for I h. The pellet was washed four times with the same buffer containing 0.1 % Triton to yield the “microfilament core.” Cytoskeletons with a qualitatively similar composition could be obtained hy extracting microvilli with Triton in 0.15 M NaCI. 30 mM sodium phosphate, pH 7.4 (PBS). Anribndies to microuilkr membrane proteins and ASGP-2. Microvillar membranes were prepared from microvilli as previously reported [24]. ASCP-IIASGP-2 complex. isolated as previously described 1261. and ASGP-2, isolated by a modification (Hull. unpublished results) of a previously described method 1261, were prepared and kindly provided by Dr Steven Hull. Antibodies to microvillar membranes were obtained by injecting membranes subcutaneously in Freund’s complete adjuvant into female New Zealand white rabbits with booster injections in incomplete adjuvant at 2week intervals. The rabbits were bled weekly after the second injection. Specific antibodies to ASGP2 were “blot-purified” by adsorption of anti-membrane antiserum or anti-membrane immunoglobulin preparations by the method of Smith and Fisher [29]. Elecfrophoresis and blotting procedures. SDS-PAGE and two-dimensional isoelectric focusingSDS-PAGE were performed as previously described 19, 241. For lectin binding analyses on gels or immunoblotting with antisera, proteins were transferred from one- or two-dimensional gels using the Bio-Rad Transblot system according to the manufacturer’s recommendations. Briefly, SDS gels were blotted in 5 mM ethanolamine, 25 mM glycine, 20% methanol, pH 9.5, for 16 h at 30 V (0. I2 mamp) and I h at 60 V (0.33 mamp). Two-dimensional gels were equilibrated in transfer buffer by three lomin washes and blotted as above. Nitrocellulose replicas were “blocked” with 20 mM Tris, 500 mM NaCl containing 0.05% Tween 20 (TBS-Tween) for 2 h and then processed for antibody or lectin staining. For Con A blots the lectin (50 kg/ml) in TBS containing 10 mg/ml bovine serum albumin (BSA) was incubated with the replica for 2 h at room temperature and washed four times with TBS-Tween (IO min each) and twice with TBS. The replicas were incubated for 2 h with horseradish peroxidase (50 kg/ml) in TBS with 10 mg/ml BSA, washed as described above, and treated with 4chloronaphthol to reveal the lectin-binding bands. Other lectins were used as direct conjugates with hor\eradish peroxidase at concentrations of 48 ~giml. For all lectins binding was abolished in the presence of the appropriate inhibitory saccharide. For immunoblots the replicas were incubated with 100 to 250-fold dilutions of antiserum or immunoglobulin preparations in TBS-BSA for 2 h, washed as described above. and incubated with horseradish peroxidase-conjugated antibodies to rabbit immunoglobulin for 2 h. Blots were washed and reactive bands revealed with 2-chloronaphthol as above. Proteolysis ofmicrouilli. Microvilli (5 mgiml protein), which are right-side out and sealed [24], were incubated with l-100 ugiml protease in PK or PBS buffer for I6 h at 20°C. Appropriate protease inhibitors were added to stop the reaction: phenylmethylsulfonyl fluoride (PMSF). 0.3 mM; leupeptin, 12.5 ug/ml; aprotinin, IO &ml; antipain, IO kg/ml; benzamidine, IO ug/ml. The proteolyzed microvilli were sedimented at 100,000~ for I h, washed twice in the same buffer with protease inhibitors. and used for preparation of microfilament cores and supernates as described above.
214
Vanderpuye,
Carraway,
and Carraway
Sialidase digestion of microvilli. Microvilli (1.8 mg protein) were incubated with Vibrio cholerae sialidase (2 units) in I ml of PK buffer containing 1 mM PMSF and 40 ug/ml each of bacitracin, benzamidine, and aprotinin for 20 min at 37°C. Microvilli were then sedimented at 10,OOOgfor 20 min, washed once with PK buffer, and extracted with Triton X-100 to prepare microfilament core and Triton-soluble fractions. Subtilisin digestion of microvillar membranes. Microvillar membranes (2 mg/ml protein) were treated with subtilisin as described above for microvilli. Triton residues and soluble fractions were prepared by centrifugation at 100,OOOgfor 1 h.
RESULTS Phalloidin
Shift Analysis
of Microfilament-Associated
Glycoproteins
Our previous studies have shown that microvillar microfilament-associated proteins could be identified by velocity sedimentation analysis of Triton extracts of microvilli prepared in the presence and absence of phalloidin, the microtilament-stabilizing toxin [20, 211. To identify microvillar microfilament-associated glycoproteins, microvilli were isolated from [3H]glucosamine-labeled cells, as described previously [24], and analyzed by this phalloidin shift procedure (Fig. 1). Fluorography of the SDS-PAGE gels from the velocity sedimentation gradients showed a major glucosamine-labeled protein with an apparent molecular weight of 120 kDA (Fig. IB) in both untreated and phalloidin-treated samples. For simplicity only the phalloidin-treated samples are shown. The 120-kDa glycoprotein comigrated with ASGP-2 of the sialomucin complex [26]. Lectin Analysis of Microfilament-Associated
Glycoproteins
Having observed a glucosamine-labeled glycoprotein which comigrated with ASGP-2, we sought a simpler, more sensitive method for analysis. For this purpose Tiiton extracts of the microvilli were sedimented at 100,OOOg to obtain microfilament “cores” [28]. Since ASGP-2 has been shown to bind Con A [26, 27, 301, SDS-PAGE gels of the cores were transferred to nitrocellulose and stained with Con A-peroxidase. As observed for the phalloidin shift analyses, the 120-kDa glycoprotein was the predominant microfilament core-associated species detected (Figs. 2 A, B). A second glycoprotein at 80 kDa was also observed in the core fraction. The Con A blotting procedure was used to analyze two-dimensional isoelectric focusing-SDS-PAGE gels of the microfilament core (Fig. 20 and showed that the 120-kDa glycoprotein comigrated with ASGP-2, the major Con A-binding glycoprotein of the microvilli and a component of the cell-surface sialomucin complex of the ascites cells. The results described above suggest that the 120-kDa glycoprotein may be a fraction of the ASGP-2 of the microvilli which is associated with the microfilament core. To determine whether the microfilament-associated 120-kDa component might be a form of ASGP-2, we first performed comparative lectin analyses of the microfilament cores. By lectin blotting identical reactivities were shown for ASGP-2 and the 120-kDa glycoprotein using a series of eight lectins with different specificities. Both the 120-kDa glycoprotein and ASGP-2 reacted with Con A, Pisum sativum lectin, Bandeiraea simplicifolia-I lectin, soybean agglutinin, and
Microfilament
association
of ASGP-2
215
4120K
-Actin
-120K
1 2 3 4 5 6 7 8 9 10 1112 13 Fig. 1. Phalloidin shift analysis of Triton extracts of microvilli. Microvilli from glucosamine-labeled 13762 MAT-Cl cells were lysed in TPK buffer and the extracts were fractionated by velocity sedimentation sucrose density gradient centrifugation in the presence of phalloidin [20, 211. (A) Coomassie blue-stained gel of fractions from velocity sedimentation gradients from top to bottom (left to right) of gradients. (B) Fluorography of gel showing the glucosamine-labeled glycoproteins in the corresponding fractions.
wheat germ agglutinin, but failed to react with peanut agglutinin, Lotus tetragonolobus lectin, and Limulus polyphenus lectin. Analyses with Antisera
Prepared
against Microvillar
Membranes
To address the question of the identity of the 120-kDa glycoprotein and ASGP2, we used antiserum prepared against microvillar membranes. Immunoblots of microvilli using anti-membrane antibodies (Fig. 3B, lane I) showed a strong reaction in the ASGP-2 region of the gel and in the region of components of XL55 kDa. No reaction was detected with ASGP-1, which is the major membrane glycoprotein but which is highly sialylated and weakly immunogenic. Microvillar Triton extracts (Fig. 3 B, lane 2) and microfilament cores (Fig. 3 B, lane 3) were also examined by immunoblot analysis with the anti-membrane serum. The staining pattern for the microvillar extracts was essentially the same as that for U-888340
216
Vanderpuye,
Carraway,
and Carraway
ASGP-2-
42
Fig. 2. Con A-binding proteins of Triton-soluble (lane I) and microfilament core (lane 2) fractions of MAT-Cl microvilli. (A) Con A blot; (B) Coomassie blue stain. The position of ASGP-2 is noted by the arrow in (A) and by the top connecting line between the panels. Arrowhead notes IO-kDa component. Molecular weight markers (kDa) are shown to the right of (B). (C) A Con A blot of the twodimensional isoelectric focusing-SDS-PAGE gel of the microfilament cores. The migration position in the second dimension is shown for ASGP-2.
the microvilli. Interestingly, the major stained band in the microfilament cores corresponded in migration to the I20-kDa component or ASGP-2 (Fig. 3B, lane 3).
These results suggest that the 120-kDa glycoprotein associated with the microfilament core may be a form of ASGP-2. Therefore we purified ASGP-2 from the sialomucin complex [26] for analysis of its reactivity with anti-membrane serum and purification of anti-ASGP-2 antibody. Immunoblots with anti-membrane serum showed a strong reaction against ASGP-2 (Fig. 3 c). These results clearly show that the antiserum contains antibodies against ASGP-2, defined by its
Microfilament B
A
C
of ASGP-2
association D
217
E
ASGP-2-
1
42
‘30
123
1 2 3
2
3
Fig. 3. Immunoblots of microvilli (lane I), T&on-soluble microvillar proteins ament cores (lane 3) with preimmune serum (A), anti-membrane antibodies purified anti-ASGP-2 (D). (0 An immunoblot of purified ASGP-2 stained with and used for immunoblot purification of anti-ASGP-2. (E) Amido black-stained
(lane 2), and microfil(B and C), and blotanti-membrane serum replica of microvilli.
purification from the sialomucin complex [26], rather than against some comigrating component. These immunoblots were then used to prepare ASGP-2-specific antibodies by immunoaffinity purification, eluting the antibodies from the immunoblots. The purified antibody was used to stain immunoblots of the microfilament cores (Fig. 3 D, lane 3) and supernates (Fig. 3 D, lane 2) from Triton extracts of microvilli. Bands were observed in the 120-kDa region of the gels for both Triton supernates and microfilament cores. These studies strongly suggest that the microfilament core-associated 120-kDa glycoprotein is a fraction of the microvillar ASGP-2 and that ASGP-2 must be one of the more immunogenic proteins in the microvillar membranes. Proteolytic
Digestion
of Microvilli
To investigate relationship between the 120-kDa glycoprotein and ASGP-2, microvilli, which are right-side out and sealed [24], were treated with proteases. Preliminary experiments were performed to establish the appropriate proteases, time, and concentrations for digestion. Figure 4 shows the effect of subtilisin treatment. For these experiments the protease-treated microvilli were extracted with Triton and fractionated by centrifugation. The Triton supernates and microfilament cores were analyzed by SDS-PAGE followed by transfer to nitrocellulose for blotting with anti-ASGP-2. Surprisingly, immunoblot analysis of these cores with specific anti-ASGP-2 indicates that the amount of ASGP-2 associated with the microfilament cores is increased with subtilisin proteolysis (Fig. 4). The effects of proteolysis on the association of ASGP-2 with the cores were verified using trypsin (data not shown). To confirm that the pelleted ASGP-2 is associated with microfilaments after proteolysis, we fractionated Triton extracts of subtilisin-treated microvilli by
218 Vanderpuye,
Carraway,
and Carraway
A
6
Mr
x lo3
120
1
2
3
123
Fig. 4. Effect of subtilisin treatment on ASGP-2 association with the microfdament core. ASGP-2 was detected by immunoblotting with blot-purified anti-ASGP-2. (A) Biton-soluble microvillar components; (B) microfilament cores. Lanes l-3 contain samples from treatments with 0, 1, and 10 &ml of subtilisin, respectively.
velocity sedimentation and analyzed the gradient fractions by immunoblotting with anti-ASGP-2. ASGP-2 was readily detected in the microfilament fractions from the gradient of the subtilisin-treated (Fig. 5B), but not the untreated (Fig. 5A), microvilli. These results clearly show that proteolysis induces an increased association of ASGP-2 with the microfilament core. Effect of Sialidase Digestion of Microvilli on the Association of ASGP-2 with the Microfilament Core One possible reason for the increased association of ASGP-2 with the microfilament core after proteolysis is that the association of ASGP-1 with ASGP-2 inhibits association of ASGP-2 with the core. We have shown previously that proteolysis removes most of the ASGP-1 [31, 321. This inhibition might be caused by steric hindrance, conformational changes in ASGP-2, or charge effects due to the high negative charge of highly sialylated ASGP-1 [33]. Preliminary experiments indicated that a number of proteases cleaved ASGP-1 from microvilli, with concomitant increases in ASGP-2 association with the microfilament core (data not shown). To investigate specifically the possibility that charge effects contribute to the association, we treated microvilli with sialidase and analyzed microtilament core and Tiiton-soluble fractions by blotting SDS-PAGE gel transfers with anti-membrane serum. As shown in Fig. 6, sialidase treatment resulted in a clear increase in ASGP-2 associated with the microtilament core. The results of the sialidase treatments support a model in which charge effects play a role in the association of ASGP-2 with the microfilament core and suggest that ASGP-1 may be inhibiting this association through its interaction with ASGP-2.
Microfilament
association
of ASGP-2
219
A
1234567891011 B
1234567891011 Fig. 5. Velocity sedimentation analysis of ASGP-2 association with the microfilament core of subtilisin-treated (B) and untreated (A) microvilli. Note association of ASGP-2, detected with antiASGP-2, with microfilament core fractions from gradient in (B).
Effect of Subtilisin Digestion of Membranes of ASGP-2 with Triton Residues
on the Association
An important question concerning the association of cell-surface glycoproteins with microfilaments is whether the association is direct or indirect. To try to answer this question, we prepared microvillar membranes, which are depleted in microfilaments, but still contain membrane-associated actin [24, 2.51. The membranes were treated with subtilisin, extracted with Triton X-100, and centrifuged. We have previously shown that these extraction and centrifugation conditions sediment from MAT-Cl microvillar membranes primarily a transmembrane complex of actin, CAG, and a 5%kDa protein [9]. In membranes treated with Con A, these conditions also pellet the ASGP-UASGP-2 complex, which is linked to the transmembrane complex by a Con A bridge [27]. When microvillar membranes were treated with increasing concentrations of subtilisin, extracted with Triton, and sedimented, immunoblots of the pellets showed increased amounts of ASGP2 with increasing subtilisin concentrations (Fig. 7). These results indicate that the presence of microlilaments is not necessary for association of ASGP-2 with a Triton-insoluble residue fraction.
220
Vunderpuye,
Carraway,
and Carraway
1 Fig. 6. Effect of sialidase digestion blot with anti-membrane antiserum. position of ASGP-2.
2
on ASGP-2 association Lane I, untreated;
with the microfilament lane 2, sialidase-treated.
core. ImmunoArrow shows
Direct Binding Assays for ASGP-2 with MicrojXuments To determine whether ASGP-2 could be binding directly to microtilaments, we incubated purified ASGP-2 with actin under actin polymerizing conditions and sedimented the microfilaments formed. Analyses of the pellets by SDS-PAGE and either staining or lectin blotting showed no evidence of ASGP-2 associated with the microfilaments (data not shown), providing further evidence that the interaction of ASGP-2 with the microfilament core is indirect.
t
ASGP-2
1234 Fig. 7. Effect of subtilisin digestion of microvillar residues of membranes. Membrane residues were serum. Lanes 14, 0, I. IO and 100 ugiml subtilisin.
membranes on ASGP-2 analyzed by immunoblots respectively.
association with Triton with anti-membrane
Microflament
association
of ASGP-2
221
DISCUSSION In many cell types and cell-surface domains regulation of the topography of cell-surface molecules appears to be determined by microtilament-membrane interactions [3]. Thus it is important to identify cell-surface molecules which are able to associate with submembrane microfilaments. The 13762 microvilli are an excellent system for investigating this problem. The microtilament core of these microvilli is more loosely organized than that of intestinal brush border microvilli 13, 24, 281 and also has microtilament-associated proteins more typical of the terminal web region of brush border than of the brush border microvilli [23, 24, 281. Thus the 13762 microvilli are more similar in composition and organization to the cortical region of the cell than to highly structured microvilli. We have been using these microvilli as a model system for studying how microfilaments associate with membranes. In previous studies we identified a major microfilament-membrane interaction site to be the transmembrane complex [9] of CAG (cytoskeleton-associated glycoprotein) [16] and actin. In the present studies, using the phalloidin shift method of identifying microfilamentassociated proteins, we found a major cell-surface glycoprotein associated with the microvillar microfilament core. Surprisingly, this glycoprotein was identified as a fraction of ASGP-2, a component of the sialomucin complex, which dominates the surface of the 13762 cells. Most of the ASGP-2 of the microvilli is solubilized by the Triton extraction procedure used to prepare microfilament cores. Interestingly, the distribution of ASGP-2 between the soluble and core fractions can be altered by protease or sialidase treatment of the microvilli before extraction. Only a small fraction of the ASGP-2 is normally associated with the microfilament core. This fraction can be increased by proteolysis or sialidase treatments, but it is still a relatively minor fraction of the total ASGP-2. Several observations suggest that association of ASGP-2 with the microfilaments is indirect. (a) Proteolysis of microvillar membranes, which are depleted in microfilaments, still leads to an increase in the association of ASGP-2 with a sedimentable residue. This residue contains primarily transmembrane complex [9]. (b) Sialidase treatment of microvilli leads to increase of the association of ASGP-2 with the microtilament core. This result indicates an inhibitory effect of charge on the interaction. Since sialidase removes only residues on the external surface, this result also suggests that the interaction occurs at the external surface. (c) Purified ASGP-2 does not bind to microtilaments. Based on the arguments above, the most likely component for the interaction with ASGP-2 is CAG, which is the only component of the transmembrane complex present at the external membrane surface 116, 241. What is the significance of the binding of ASGP-2 to the microfilament core? ASGP-2, which is present at the cell surface as a complex with the sialomucin ASGP-I [26, 301, is the major Con A-binding protein of the 13762 ascites cell surface [26]. When 13762 ascites cells are capped with fluorescent Con A, the sialomucin complex is the major species contributing to the redistribution of the fluorescence [30]. We have previously shown that capping in these cells is initiated by formation of a Con A bridge between ASGP-2 and CAG of the
222 Vanderpuye,
Carraway,
and Carrawq
transmembrane complex [27]. This observation explains how Con A capping can be directed by linkage to microfilaments. However, this mechanism does not explain capping of molecules by crosslinking agents, such as antibodies, which do not bind directly to a transmembrane complex. We suggest that the binding of a fraction of ASGP-2 to the core may offer an explanation for this second type of capping. Perhaps glycoproteins of transmembrane complexes contain a site which can bind to many other cell-surface molecules. The binding would not have to be of high affinity because the clustering of the cell-surface component caused by crosslinking with antibodies or lectins should stabilize the aggregates. What type of binding may be involved is not known, but the mechanism of interaction of ASGP-2 with the microfilament core, probably through CAG of the transmembrane complex, is of great interest in this regard. We thank Maria Carvajal for excellent technical assistance and Dr S. Hull for providing samples of sialomucin complex and purified ASGP-2. This work was supported by grants from the National Institutes of health (GM 33795 and CA 3 1695) and by the Papanicolaou Comprehensive Cancer Center of the University of Miami (CA 14395).
REFERENCES 1. 2. 3. 4. 5. 6.
Geiger, B. (1983) Biochim. Biophys. Acta 737, 305. Carraway, K. L., and Carraway, C. A. C. (1985) Bioessays 1, 55. Carraway, K. L., and Carraway, C. A. C., unpublished. Bennett, V. (1985) Annu. Rev. Biochem. 54, 273. Anderson, R. A., and Lovrien, R. E. (1984) Nature (London) 307, 655. Matsudaira, P., and Burgess, D. R. (1979) J. Cell Biol. 83, 667. 7. Fox, J. (1985) J. Biol. Chem. 260, 11,970. 8. Horwitz, A., Duggan, K., Buck, C., Beckerle, M. C., and Burridge, K. (1986) Nature (London) 320, 531.
Carraway, C. A. C., Jung, G., and Carraway, K. L. (1983) Proc. N&l. Acad. Sci. USA 80, 430. 10. Carraway, C. A. C., Jung, G., and Carraway, K. L. (1983) Cell. Motil. 3, 491. 11. Luna, E. J., Goodloe-Holland, C. M., and Ingalls, H. M. (1984) J. Cell Biol. 99, 58. 12. Wuestehube, L. J., and Luna, E. J. (1987) J. Cell Biol. 105, 1741. 13. Brown, S. S., Malinoff, H. L., and Wicha, M. S. (1983) Proc. Natl. Acad. Sci. USA 80, 5927. 14. Painter, R. G., Gaarde, W., and Ginsberg, M. H. (1985) J. Cell. Biochem. 27, 277. 15. Painter, R. G., Prodouz, K. N., and Gaarde, W. (1985) J. Cell Biol. 100, 652. 16. Carraway, C. A. C., Jung, G., Craik, J. R., Rubin, R. W., and Carraway, K. L. (1983) Exp. Cell 9.
Res. 143, 303. 17. MOSS, D. J. (1983) Eur.
18. 19. 20. 21. 22.
J. Biochem.
135, 291.
Painter, R. G., and Ginsberg, M. (1982) J. Cell Biol. 92, 565. Sheterline, P., and Hopkins, C. R. (1981) J. Cell Biol. 90, 743. Carraway, C. A. C., and Weiss, M. (1985) Exp. Cell Res. 161, 150. Carraway, C. A. C., Sindler, C., and Weiss, M. (1986) Biochim. Biophys. Acta 885, 68. Carraway, K. L., Doss, R. C., Huggins, J. W., Chesnut, R. W., and Carraway, C. A. C. (1979) J. Cell Biol.
83, 529.
23. Carraway, K. L., Huggins, J. W., Cerra, R. F., Yeltman, D. R., and Carraway, C. A. C. (1980) Nature
(London)
285,
508.
24. Carraway, C. A. C., Cerra, R. F., Bell, P. B., and Carraway, K. L. (1982) Biochim. Acra
25. 26. 27. 28. 29.
719,
Biophys.
126.
Carraway, K. L., Cerra, R. F., Jung, G., and Carraway, C. A. C. (1982) J. Cell Biol. 94, 624. Sherblom, A. P., and Carraway, K. L. (1980) J. Biol. Chem. 255, 12,051. Jung, G., Helm, R. M., Carraway, K. L., and Carraway, C. A. C. (1984) J. Cell Biol. 98, 179. Carraway, C. A. C., Jung, G., Hinkley, R. E., and Carraway, K. L. (1985) Exp. Cell Res. 115,71. Smith, D. E., and Fisher, P. A. (1984) J. Cell Biol. 99, 20.
Microfilament 30. 31. 32. 33.
Helm, R. Buck, R. Spielman, Sherblom,
association
of ASGP-2
M., and Carraway, K. L. (1981) Exp. Cell Res. 135, 418. L., Sherblom, A. P., and Carraway, K. L. (1979) Arch. Biochem. Biophys. J., Rockley, N. L., and Carraway, K. L. (1987) J. Biol. Chem. 262, 269. A. P., Buck, R. L., and Carraway, K. L. (1980) J. Biol. Chem. 255, 783.
223
198, 12.
Received December 12, 1987 Revised version received April 15, 1988
Printed in Sweden