Shedding of the major plasma membrane sialoglycoprotein from the surface of 13762 rat ascites mammary adenocarcinoma cells

ARCHIVES OF BIOCHEMISTRY AND BIOPHYSICS Vol. 20’7, No. 1, March, pp. 40-50, 1981

Shedding of the Major Plasma Membrane the Surface of 13762 Rat Ascites Mammary SUSAN CORALIE Department

Sialoglycoprotein Adenocarcinoma

C. HOWARD, ANNE P. SHERBLOM,2 JOHN A. CAROTHERS CARRAWAY, AND KERMIT of Biochemistry,

Oklahoma

State

University, Stillwater,

from Cells1

W. HUGGINS, L. CARRAWAY Oklahoma

74078

Received August 6, 1980 ASGP-1 (ascites sialoglycoprotein 1) the major sialoglycoprotein of 13762 rat ascites mammary adenocarcinoma cells, is shed from MAT-B1 (nonxenotransplantable) and MAT-Cl (xenotransplantable) sublines when incubated in vitro aRer labeling in wivo with [Wlglucosamine. The rates of shedding of label in both particulate and soluble form are similar for the two sublines, but the turnover of label in the cells is 80% greater for MAT-Cl cells (tliz 2.4 days) than for MAT-B1 cells (tl12 4.1 days). Shed soluble ASGP-1 was smaller than ASGP-1 in the particulate fraction by gel filtration in dodecyl sulfate. By CsCl density gradient centrifugation, gel filtration, and sucrose density gradient centrifugation, all in 4 M guanidine hydrochloride, the shed soluble ASGP-1 was found to be slightly more dense and smaller than ASGP-1 purified from membranes. No differences in sialic acid or oligosaccharides released by alkaline borohydride treatment were found between the shed soluble ASGP-1 and purified ASGP-1. These results suggest that the shed soluble ASGP-1 is released from the membrane by a proteolytic cleavage. This mechanism is supported by the inhibition of the release of soluble shed ASGP-1 by aprotinin, a protease inhibitor. Soluble ASGP-1 in ascites fluid is also smaller by gel filtration, but is more heterogeneous, suggesting a similar release mechanism in tivo followed by more extensive degradation in the ascites fluid.

Cell surface glycoproteins are believed to play important roles in permitting the escape of tumor cells from destruction by the immune system of the host (1). It has been postulated that the shedding of cell surface antigens and other glycoproteins from the tumor may block destruction of the tumor cells by cells of the immune system (l-4), yet there is little information concerning the mechanisms by which cell surface material is shed. Miller et al. (5, 6) have investigated the kinetics of

release of glucosamine-labeled, perchloric acid-soluble components from the cell surface of the allo- and xenotransplantable mouse ascites TA3-Ha mammary adenocarcinoma. Vicia graminea receptors derived from the major cell surface sialoglycoprotein(s) epiglycanin were released in a biphasic manner (5). However, several questions remain unanswered concerning the mechanism(s) of release of cell surface components. Since the native form(s) of epiglycanin was not characterized, it is not possible to compare the released and cellassociated forms of the glycoprotein. The fact that the released material is soluble suggests an alteration in the glycoprotein during release. Moreover, such analyses ignore possible contributions of shed particulate fractions. Shedding of membrane fragments has been observed. with mammary carcinomas (7) and other cells (8) and may be an intermediate in the release of soluble cell surface material.

1 Journal Article 3749 of the Agricultural Experiment Station, Oklahoma State University, Stillwater, OK. This research was conducted in cooperation with the USDA, Agricultural Research Services, Southern Region, and supported by the National Institutes of Health (CA 19985 and CA 25173), a Presidential Challenge Grant from Oklahoma State University and the Oklahoma Agricultural Experiment Station. * Recipient of a postdoctoral fellowship from the National Institutes of Health (F32 CA 06273). 0003-9861/81/030040-11$02.00/O Copyright 0 1981 by Academic Press, Inc. All rights of reproduction in any form reserved.

40

RELEASE

OF CELL SURFACE

MATERIAL

For our investigations of these questions we have used the MAT-B1 and MAT-Cl ascites sublines of the 13’76‘2 rat mammary adenocarcinoma. These are similar to the St and Ha sublines of the TA3 adenocarcinoma in many respects, but offer several advantages over these sublines. Although the MAT-Cl subline is xenotransplantable and MAT-B1 is not, both have large amounts of a single major cell surface sialoglycoprotein termed ASGP-1 (9-11). This contrasts with the TA3 sublines in which the allo- and xenotransplantable Ha subline has epiglycanin, but the nontransplantable St subline does not (12, 13). Moreover, the ASGP-1 from both MATBl and MAT-Cl sublines has been purified to homogeneity and shown to be the same size as the glycoprotein solubilized directly in SDS3 from the corresponding cells (11). Amino acid compositions of MATBl and MAT-Cl ASGP-1 are very similar, but their carbohydrate compositions, molecular weights, and oligosaccharides are different (11). During metabolic labeling in viva or in vitro ~70% of the glucosamine incorporated into protein of either subline is found in ASGP-1 (10, 11). Finally, previous studies have demonstrated the release of particulate cell surface fractions, including microvilli (14), from these cells, and procedures have been developed for fractionating these materials (14). The current studies set out to answer the following questions. What are the differences between the shed soluble and native membrane-bound forms of ASGP-1 that might relate to the mechanism of shedding? What is the nature of particulate material shed in vitro and in vivo? What are the rates of shedding of soluble and particulate fractions for the two sublines? How do the amounts and types of materials shed from the two sublines compare? Using the metabolic labeling and fractionation of released cellular components on Percoll gradients or by differential centrifugation we have found released ASGP-1 in particulate form and in a soluble form that 3 Abbreviations used: PBS, phosphate-buffered saline, see Ref. (16); SDS, sodium dodecyl sulfate; PPO, 2,5diphenyloxazole; PMSF, phenylmethyl sulfonyl fluoride; Gdn HCl, guanidine hydrochloride.

FROM ASCITES

CELLS

41

appears smaller than ASGP-1 from cells or shed particulate matter. Similar rates of release of cell surface material in vitro were found for the two sublines but MATCl cells have a 2.5-fold greater amount of soluble ASGP-1 in ascites fluid. Although this increased amount of soluble ascites fluid ASGP-1 for MAT-Cl cells correlates with the xenotransplantability, it is doubtful whether this quantitative difference in shedding can be responsible for the transplantation differences. EXPERIMENTAL

PROCEDURES

Materials McCoy’s 5a (modified), nonessential amino acids, fetal calf serum, EDTA, and penicillin/streptomycin solution were from Gibco, sodium pyruvate and gentamicin were from Microbiological Associates, PPO (2,5-diphenyloxazole), soybean trypsin inhibitor, trypsin inhibitor, trypsin, actinomycin D, cycloheximide, puromycin, colchicine, cytochalasin D, theophylline, sodium azide, 2-deoxyglucose, PMSF (phenylmethyl sulfonyl fluoride), Aprotinin, and EGTA were from Sigma, D-[l-3H]glucosamine HCl (206 Wmmol), carrier-free Nalz51, and D-[l-‘4C]ghcoSamine HCl (50-60 mCi/mmol) were from AmershanVSearle. Instage1 was from Packard, Percoll from Pharmacia, polycarbonate filters from Bio-Rad, peanut lectin from Miles-Yeda, Ltd., and calcium ionophore A23187, a gift from Eli Lilly. PBS was prepared from reagent grade salts and contained Mg2+, Ca*+, and glucose (16). The 13762 ascites tumors (15) were passaged as described previously (10). Characterization of shedding in vitro. MAT-B1 or MAT-Cl tumor-bearing rats were injected ip with 0.1 mCi [3H]glucosamine 16-17 h before sacrifice (10, 11). Tumor cells were removed aseptically and washed three times with McCoy’s 5a modified culture medium to which 25% fetal calf serum and 10 pg/ ml gentamicin had been added. Cells (3-8 x 1OVml) were resuspended in fresh media with fetal calf serum and incubated in spinner culture at 37°C in a 5% CO, atmosphere for various times. There was no apparent effect of cell density on rate of shedding over the range of cell concentrations used. For pulse-chase analysis of the kinetics of shedding the incubations contained 3.8 x 10e3 pmol/ml cold glucosamine. Particulate fractions were separated from whole cells using a 30% Percoll self-forming gradient (14). Soluble ASGP-1 was obtained from the supernatant after pelleting cells at 121g for 15 min by filtering through 0. l-pm-pore size polycarbonate membranes followed by dialysis against water or, alternatively, by centrifuging at 100,OOOgfor 1.5 h.

HOWARD ET AL.

42

Radioactive fractions were counted in 10 ml Instage1 in a Tri-Carb liquid scintillation counter. Electrophoresis in SDS on 4.5-15% gradient polyacrylamide gels, Sepharose 2B chromatography, and scanning electron microscopy were performed as previously described (9-11, 17). Trypsin treatment of MAT-B1 cells. Labeled, washed MAT-B1 cells, obtained as described above, were resuspended in complete medium with 1 mg/ ml glucose, then treated with 50 pg/ml trypsin for 30 min in PBS. Soybean trypsin inhibitor (100 pg/ml) was added to stop the proteolysis; after 30 min the cells were washed and used for shedding experiments. Effect of perturbants on in vitro shedding. Experiments were performed as described above, using a 2-h incubation, except that the effecters were included at the concentrations noted. Effecters used were: soybean trypsin inhibitor (1 mM), phenylmethyl sulfonyl fluoride (1 mM), colchicine (1 mM), EGTA (1 mM) ionophore A23187 (0.5 mM) with and without 1 mM CaCl,, theophylline (1 mM), cytochalasin D (1 mM), sodium azide (1 mM), actinomycin D (0.1 mM), puromycin (0.1 mM), cycloheximide (1 mM), aprotinin (20 kIU/ml), EDTA (1 mM), deoxyglucose (1 mM), and dibucaine (1 mM). Characterization of material shed in vivo. The labeled cell suspension was aspirated from the rat peritoneal cavity without dilution and cells were pelleted at 480s for 5 min. Particulate and soluble fractions were obtained by Percoll gradient and differential centrifugation, respectively, as described above. Protein analysis was performed by the method of Lowry et al. (18). Other characterization procedures were as described above. Isopycnic CsCl centrifugation was performed as described previously (11). Quantitation of soluble ASGP-1 in as&es jluid. ASGP-1 was purified from MAT-B1 and MAT-Cl cells as previously described (11). Various amounts of the glycoproteins (0.02-1.5 pg) were applied to 4.5-15% gradient polyacrylamide gels and electrophoresed in SDS. The gels were treated with ‘Y-labeled peanut agglutinin, which reacts specifically with ASGP-1 (19). Labeled glycoprotein-lectin bands were cut from the gels and counted to prepare a standard curve, which was linear for each glycoprotein over the range of protein concentrations used. Samples of the soluble portion of ascites fluid were subjected to the same procedure, permitting determination of the concentrations of ASGP-1 in each (19). RESULTS

Characterization

of Shedding

in Vitro

For characterization of the shed material glycoproteins of MAT-B1 and MAT-Cl cells were labeled by injection of [3H]glucosa-

FIG. 1. Polyacrylamide gel electrophoresis in SDS of [3H]glucosamine-labeled MAT-B1 cells (A,D), particulate shed fraction (B,E), and soluble shed fraction (C,F). Gels A-C were stained with Coomassie blue and D-F are fluorograms showing the radioactive components. Note the predominance of ASGP-1 in all labeled fractions.

mine into the peritoneal cavity. As previously shown (g-11), the protein-bound label is found predominantly in one component, ASGP-1, illustrated in Figs. 1 and 2 for MAT-B1 and MAT-Cl cells, respectively. Several experiments indicate that this component is at the cell surface of both sublines. (i) It is labeled by lz51 and lactoperoxidase (10). (ii) Glucosamine label is released from the cells by trypsin or chymotrypsin (9). When cells are treated with Pronase most of the ASGP-1, assayed by SDS-polyacrylamide gel electrophoresis and autoradiography, is degraded before the permeability barrier of the cells is broken (14). (iii) ASGP-1 co-isolates with 5’-nucleotidase on putative plasma mem-

RELEASE

ABC9

OF CELL SURFACE

MATERIAL

EFGH

FIG. 2. SDS-polyacrylamide electrophoresis of [3H]glucosamine-labeled MAT-Cl cells (A,E), shed microvilli fraction (B,F), shed membrane fragment and vesicle fraction (C,G), and shed soluble material (D,H). Conditions are as in Fig. 1.

brane fractions obtained from a microsome preparation or from a Zn2+ stabilized membrane preparation4 (11). When cells are incubated in vitro, released cell surface material can be observed by dark-field microscopy or by fluorescence microscopy after staining with fluorescent Con A. MAT-Cl microvilli were identified by the unusual morphology of the branched microvilli (14, 16). The presence of ASGP-1 in these shed fractions was ascertained in prelabeled cells by low-speed centrifugation to remove the cells and quantitation of the radioactivity in the supernatant. SDSpolyacrylamide gel electrophoresis of the supernatants confirmed the presence of ASGP-1. However, differential centrifugation did not adequately separate cells from the shed particulate material. Therefore we used Percoll gradient centrifugation (14) which gave three fractions for MAT-Cl cells and two for MAT-B1 cells. By dark-field microscopy these were identified as (i) microvilli, (ii) membrane fragments or vesicles, and (iii) cells. The middle fraction is absent in preparations from MAT-B1 cells. Although all three fractions can be 4 A. P. Sherblom, unpublished observations.

FROM ASCITES

43

CELLS

obtained from MAT-Cl cells on a carefully controlled gradient, the microvilli and fragment fractions tend to overlap, as judged by dark-field microscopy. Figure 3 shows a typical Percoll gradient profile for shed material from MAT-Cl cells in which fraetions numbered 5 and 6 contain predominantly fragments and those numbered 2-4 contain predominantly microvilli. For the results presented here, we have considered the particulate shed fractions together for quantitative analysis. The soluble fraction of shed material was obtained by centrifugation at 100,OOOg for 1 h. For characterization of the particulate shed materials a sample of MAT-Cl cells was incubated in medium for 3 h at 37°C then fractionated by differential centrifugation. Each pellet was further separated into cells and particulate material by Percoll gradient centrifugation. From these results it was found in a typical experiment that 66% of the glucosamine label was still cell associated, 4.3% was in a particulate fraction sedimentable at 1OOOg for 5 min (chiefly microvilli), 3.6% was in a particulate fraction sedimentable between 1000 and lO,OOOg, and 3.2% was sedimentable only at 100,OOOg. When combined with the 11% of the glucosamine found in the soluble fraction, the total shed material repre-

I I

I I 5 IO Elutlon Volume (ml)

1 15

FIG. 3. Per-co11 gradient profile of MAT-Cl cells (major peak) and particulate fraction (smaller peak) shed in vitro. Gradient was run as previously described (14).

44

HOWARD

sented 22% of the glucosamine incorporated into the cells with a total recovery of 88% of the label. These results show that the shed particulate material is quite heterogeneous in size. Dark-field microscopy indicated that microvilli are predominant in the heavier fractions from the differential centrifugation. Shed material from MAT-B1 cells showed a similar distribution on eactionation. Examination of the labeled shed material by SDS-polyacrylamide gel electrophoresis and fluorography showed ASGP-1 to be the only significant labeled component in the soluble fraction (Figs. 1 and 2). Microvilli and the fragment (vesicle) fraction also contained predominantly ASGP-1, although a small amount of a second labeled component (ASGP-2) was present but not readily discernible in the photographs of Figs. 1 and 2. To show that the soluble ASGP-1 was being released from the cell surface rather than secreted from within the cell, labeled MAT-B1 cells were treated with trypsin, which cleaves cell surface ASGP-1 (9). After incubation with trypsin inhibitor the cell suspension was washed to remove protease and incubated in medium for 2 h. Released, soluble ASGP-1 was obtained after centrifugation. Electrophoresis and fluorography indicated that essentially all of the ASGP-1 shed from the trypsintreated cells had been cleaved (Fig. 4), indicating that it had come from the cell surface rather than from within the cell. Gel Filtration

of Shed Materials

Since ASGP-1 is the predominant glycoprotein of the surface of 13762 ascites cells, the mechanism by which it is shed is of interest. Cellular ASGP-1 is membrane-bound, although it can be released from the membranes with guanidine hydrochloride (11) or nonionic detergent (9). The fact that part of the shed ASGP-1 is soluble indicates that it was altered in the shedding process. No change in apparent molecular size is observed by SDS-polyacrylamide gel electrophoresis. However, the negative charge and anomalous detergent binding characteristics of sialoglycoproteins make SDS-polyacrylamide gel electrophoresis

ET AL.

FIG. 4. Release of soluble ASGP-1 from [3H]glucosamine labeled MAT-B1 cells with (D) and without (C) trypsin pretreatment. MAT-B1 whole cells are shown in (A) and isolated microvilli in (B). The corresponding fluorogram is shown in wells E-H in the same order.

procedures unreliable for evaluating glycoprotein molecular sizes and size changes (20). To determine if alterations in the size of ASGP-1 occur during shedding, we have used gel filtration in SDS to compare ASGP-1 from shed fractions with ASGP-1 purified from plasma membranes by density gradient centrifugation in CsCl containing 4 M guanidine hydrochloride (11). The shed particulate fraction from labeled MAT-Cl cells, solubilized in SDS and chromatographed on Sepharose 2B, gave a large peak (ASGP-1) and a smaller peak (ASGP-2). The ASGP-1 elution was almost coincident with that of purified MAT-Cl ASGP-1 (Fig. 5A). ASGP-1 from the particulate shed fraction appeared slightly larger than ASGP-1 purified from MAT-Cl membranes, possibly reflecting some differences in carbohydrate composition between the microvillus ASGP-1 represented by the shed particulate material, and ASGP-1 isolated from the MAT-Cl microsomal fraction. In contrast the elution profile of soluble, shed ASGP-1 indicated that it was significantly smaller than the purified ASGP-1 (Fig. 5B). These results suggest

RELEASE

OF CELL SURFACE

Fractlan

MATERIAL

No.

FIG. 5. Gel filtration on Sepharose 2B in SDS comparing shed and “native” labeled material. (A) Particulate fraction shed from MAT-Cl cells labeled with [3H]glucosamine (-) was isolated as described under Experimental Procedures, solubilized in SDS and cochromatographed with [‘4C]glucosamine-labeled ASGP-1 (- - -) isolated from MAT-Cl membranes as previously described. (B) Soluble shed fraction (-) from MAT-Cl cells was treated with SDS and cochromatographed with purified ASGP-1 (- - -).

that ASGP-1 is being released from the cells in soluble form by a proteolytic cleavage mechanism. The release of a soluble form of ASGP-1 without a change in mobility on SDSpolyacrylamide gels suggests that the protein is being cleaved to leave a small fragment of the polypeptide with the cells, releasing the larger fragment which contains most of the carbohydrate. This mechanism predicts that the released fragment should be slightly more dense and smaller than ASGP-1 from membranes and should contain its full complement of carbohydrate. Because of the limited amount of shed soluble material available we made these comparisons directly by a dual labeling procedure. MAT-Cl cells were labeled in, viva in separate animals with r3H]- and [14C]glucosamine. Cells labeled with 14C were incubated in vitro for the isolation of shed soluble ASGP-1, while cells labeled

FROM ASCITES

45

CELLS

with 3H were homogenized for membrane preparation. The shed material in 4 M guanidine hydrochloride and 2.37 M CsCl was mixed with membranes and centrifuged at 55,000 rpm for 40 h in a ‘75 Ti rotor. As shown in Table I the shed material was slightly more dense. Fractions between density 1.38 and 1.44 g/ml were pooled for further analysis. Both gel filtration and sucrose density gradient centrifugation in 4 M guanidine hydrochloride indicated that the shed soluble ASGP-1 was smaller than membrane ASGP-1. From these we estimate that the molecular weight in 4 M guanidine hydrochloride of the shed soluble ASGP-1 is about 5-10% less than that of the membrane ASGP-1. Two further experimentss were performed on the labeled mixture to determine if loss of carbohydrate, particularly sialic acid, could account for the molecular weight difference. First, the amount of label present as sialic acid was determined by Bio-Gel P-4 gel filtration of the mixture following treatment with 0.05 M H,SO, at 80°C for 1 h (11). For both shed and membrane ASGP-135% of the radioactivity was released and the 14CPH ratios of the void volume and released material were the same. Second, the O-linked oligosaccharides were released by treatment with alkaline borohydride and the mixture was chromatographed on a Bio-Gel P-4 column. The TABLE

I

COMPARISONOFMEMBRANEANDSHEDGLYCOPROTEIN FROMMAT-Cl CELLS

Membrane Density (g/ml) Gel filtration (K,,)” Apparent sedimentation coefficient (SY Percentage of glucosamine label as sialic acid

Shed

1.407 0.39

1.413 0.45

3.93

3.72

35%

35%

n Gel filtration was performed on a Sepharose CLZB column equilibrated with 4 M GdnHCl, 10 mM Tris, pH 7.4, at 4°C.

b The sedimentation coefficient was estimated by sedimentation on a 5-30% sucrose gradient containing 4 M GdnHCl at 4°C; corrections for density and viscosity have not been made.

46

HOWARD ET AL.

normalized 14C and 3H profiles were identical within experimental uncertainty. The net recovery in each peak differed by less than lo%, and no consistent shift in the size or distribution of the released oligosaccharides was observed. Thus both the sialic acid content and distribution of oligosaccharides for shed ASGP-1 appear unchanged from that of the membrane ASGP-1. Kinetics of Shedding Processes

To investigate the release process, MATBl and MAT-Cl cells were labeled in VizIo, transferred to culture, and incubated in the presence of cold glucosamine. Aliquots were removed at intervals and fractionated to give cells, particulate shed material, and soluble shed material (Fig. 6). In three separate experiments similar amounts of both particulate and soluble material were observed in the shed fractions when MATBl and MAT-Cl cells were compared. Loss of label from the cells follows a first-order relationship (Fig. 7) with half-lives of 4.1 ? 0.5 and 2.4 2 0.2 days (three experiments each) for MAT-B1 and MAT-Cl cells, respectively. Since the rate of turnover of label is different between MAT-B1 and MAT-Cl cells, while the rate of shedding is approximately the same, turnover must involve other processes in addition to shedding, at least for MAT-Cl cells. It is likely that plasma membrane

FIG. 6. Release of soluble MAT-B1 (A), soluble MAT-Cl (.A), particulate MAT-B1 (O), and particulate MAT-Cl (0), fractions during in vitro incubations. A typical experiment of three trials is shown. The failure of MAT-Cl cells to show zero shedding at the zero time point occurs because the centrifugation technique to obtain cells for the shedding experiment does not remove all of the particulate shed material from the cells. Percoll gradient centrifugation is necessary for complete separation.

1 0

I 8

I Hours

16 Chose

24

FIG. 7. Loss of [3H]glucosamine labeled material from MAT-B1 (A) and MAT-Cl (0) cells.

internalization also contributes to turnover rate of the cell surface material. Perturbation of the Shedding Processes by Effecters

As a further characterization of the mechanisms of shedding processes, a number of effecters were tested for their ability to inhibit or accelerate release of soluble ASGP-1 or particulate surface material. Only aprotinin and EDTA, of the effecters tested, were able to inhibit release of soluble ASGP-1 and each caused a 30-50% inhibition. None of the effecters gave a consistent, significant inhibition of release of the particulate fraction. Shedding of both soluble and particulate fractions was enhanced by 1 InM dibucaine (soluble, 1.7-fold; insoluble, 2.4-fold) and 0.5 mM ionophore A23187 in the presence of 1 mM calcium (soluble, 2.7-fold; particulate, 3.5fold). The effects on the release of soluble ASGP-1 further support a proteolytic mechanism, possibly involving a cation-dependent protease. The effects of dibucaine and ionophore suggest a role for Caz+. These experiments also demonstrate that serum has little effect on shedding, since the rate of shedding over 3 h in the absence of serum was >90% of that in its presence. Shed Materials in Ascites Fluid of Animals Bearing MAT-B1 an,d MAT-Cl Tumors

To determine whether similar shedding processes occur in vivo, ascites cells and

RELEASE

OF CELL

SURFACE

MATERIAL

FROM ASCITES

47

CELLS

ASGP-1 (Fig. 9). ASGP-1 from the insoluble fraction isolated from ascites fluid was approximately the same size as purified ASGP-1 from MAT-Cl cells. Similar results were obtained for MAT-B1 cells. For additional characterization the soluble fraction from ascites fluid was examined by density gradient centrifugation in CsCl in 4 M guanidine hydrochloride, the technique used for purifying ASGP-1. A peak corresponding to ASGP-1 is observed at a density of 1.4 for both MAT-B1 and MAT-Cl (Fig. 10). Both peaks of material from ascites fluid are broader than those for purified ASGP-1. The MAT-B1 peak shows evidence of considerable heterogeneity, suggesting that substantial degradation of ASGP-1 is occurring in the ascites fluid of MAT-B1 cells. Some ASGP-1 degradation mav occur on released particulate material “in the fluid. The greater

FIG. 8. Fluorogram of polyacrylamide slab gel of ascites fluid components showing MAT-B1 soluble fraction (A), MAT-Cl soluble fraction (B), MAT-B1 particulate fraction (C,D), and MAT-Cl particulate fraction (E,F). A second labeled component can be observed in the more heavily loaded insoluble fractions (C,F).

fluid were removed from animals injected with [3H]glucosamine. Immediate fractionation of the material on Percoll gradients gave results similar to those found for in vitro shedding with a broad peak at the density of the insoluble material. Highspeed centrifugation of labeled ascites fluid gave a soluble fraction which by SDSpolyacrylamide gel electrophoresis contained ASGP-1 as the only detectable labeled component (Fig. 8). ASGP-1 was also the predominant labeled component of the insoluble fraction isolated by Percoll gradient centrifugation. Consistent with results from in vitro shedding, gel filtration in SDS of fractions from the ascites fluid indicated that soluble shed ASGP-1 from MAT-Cl cells was smaller than the corresponding purified

6(1 Froctlon

No

FIG. 9. Gel filtration on Sepharose 2B in SDS comparing ascites fluid and “native” labeled material. (A) Particulate fraction from MAT-Cl ascites fluid labeled with [3H]glucosamine (-) was isolated as described under Experimental Procedures, solubilized in SDS and cochromatographed with [W]glucosamine-labeled ASGP-1 (- - -) isolated from MATCl membranes as previously described. (B) Soluble fraction (-) from MAT-Cl ascites fluid was treated with SDS and cochromatographed with purified ASGP-1 (- - -).

48

HOWARD

apparent degradation of MAT-B1 ASGP-1 is consistent with previous observations on the trypsinization and release of ASGP-1 fragments from MAT-B1 and MAT-Cl cells (9). The differences in susceptibilities between the glycoproteins from the two sublines may be due to the greater carbohydrate content, particularly sialic acid, of MAT-Cl (11). Dark-field microscopy showed only membrane fragments and vesicles from the ascites fluid of both MAT-B1 and MAT-Cl cells, indicating that any intact microvilli shed from the cell surfaces are degraded in the animal. Scanning electron microscopy of cell-free, glutaraldehyde-fixed samples of ascites fluid showed stringy aggregates of particulate material (data not shown). No significant differences in structure were observed between samples from MAT-B1 and MAT-Cl tumors. Amounts of particulate shed material and soluble shed ASGP-1 in ascites fluid were quantitated by protein analysis and radiochemical lectin assay (19), respectively. The insoluble fraction represented 0.8 and 0.6% of the total ascites fluid protein for MAT-B1 and MAT-Cl cells, respectively. Soluble ASGP-1 protein was 0.12% of total ascites protein for MAT-B1 cells and 0.29% for MAT-Cl cells. Thus there is about 2.5 times as much soluble ASGP-1 in MAT-Cl ascites fluid as in MAT-Bl, but about equal amounts of particulate fraction in the two sublines. DISCUSSION

Numerous studies of enzymically or metabolically labeled cells have demonstrated shedding of cell surface components (5, 6, 21-28). Most of these have concentrated on soluble shed material even though shedding of particulate material can be demonstrated for some cells (7, 5, 29). In such cases the release of the soluble components by degradation of shed, particulate material must be considered. In the 13762 ascites tumors we have demonstrated shedding of both soluble and particulate fractions. Most previous studies have not examined shedding of discrete, well-characterized

ET AL.

I5

14 13 12 15 0 Elut~on

Welght

5

IO

15

Cgl

FIG. 10. Isopycnic density gradient profiles in 4 M guanidine hydrochloride and 2.37 M CsCl of MAT-Cl purified ASGP-1 (A), MAT-Cl ascites fluid soluble ASGP-1 (B), MAT-B1 purified ASGP-1 (C), and MAT-B1 ascites fluid soluble ASGP-1 (D). Samples were subjected to density gradient centrifugation as previously described (11).

cell surface glycoproteins, making it difficult to specify a molecular mechanism for the shedding process. From comparisons of shed soluble and membrane ASGP-1 based on our previous characterization of the glycoprotein we postulate a proteolytic cleavage mechanism for the release of soluble ASGP-1. A similar mechanism has been suggested for shedding of glycoprotein from vesicular stomatitis virus-infected Chinese hamster ovary cells (28). The major points favoring this hypothesis for ASGP-1 are as follows. (i) The shed glycoprotein is soluble and apparently not aggregated. In contrast ASGP-1 isolated from membranes aggregates when not in the presence of detergents or protein denaturants. (ii) The soluble ASGP-1 is slightly more dense than the membrane form, suggesting the loss of polypeptide rather than carbohydrate. (iii) Soluble ASGP-1 is smaller by gel filtration in SDS or guanidine hydrochloride or sucrose density centrifugation in guanidine hydrochloride. (iv) Sialic acid and oligosaccharide comparisons do not show significant differences between the soluble and membrane forms. (v) The

RELEASE

OF CELL

SURFACE

MATERIAL

protease inhibitor aprotinin decreases release of soluble ASGP-1 when incubated with cells. From these results we suggest that ASGP-1 binds to the membrane via a hydrophobic portion of the polypeptide, which remains with the membrane on release. Exactly how this association occurs is unknown. One possibility is that a hydrophobic segment of ASGP-1 binds to the membrane bilayer as postulated for cytochrome b, (30) and erythrocyte glycophorin (31). Release could occur by proteolytic cleavage at a point between the membrane attachment and the bulk of the carbohydrate. However, since the sialoglycoprotein of MAT-Cl cells appears to be present in fibrillar structures on the cell surface (lo), it may not interact directly with the membrane. An alternate possibility is that ASGP-1 exists in polymerized form on the cell surface, interacting with the membrane via a coupling molecule. This mechanism of binding can explain the release of ASGP-1 from membranes by protein denaturants (urea, guanidine hydrochloride), and by nonionic detergents. The question of the importance of shedding to the survival of the tumor cells is still unsettled. Although morphologically identifiable cell surface structures (e.g., microvilli) can be observed in the medium after shedding in vitro, no discrete structural elements are observed in ascites fluid. Apparently shed structural elements break down in the ascites fluid. The structures and amounts of insoluble surface products are quite similar for MATBl and MAT-Cl cells in vivo, suggesting that shedding of insoluble material does not account for biological differences between these sublines. The rates of shedding of both particulate and soluble labeled components are quite similar for the two sublines in vitro, but the MAT-Cl subline has 2.5 times as much soluble ASGP-1 in the ascites fluid than does MAT-Bl. This result indicates that the relative rates of shedding of soluble material must be different in wiwo or, more likely, that other processes are important in determining the amount of soluble ASGP-1 in ascites fluid. Regardless, it seems questionable whether

FROM

ASCITES

49

CELLS

these quantitative differences in ascites fluid glycoprotein are sufficient to account for transplantation differences between the sublines. Thus our results do not provide positive evidence that shedding of sialoglycoproteins is a major factor for the 13762 tumor in permitting transplantation across histocompatibility barriers, as suggested for the TA3 mammary adenocarcinoma (3, 5, S), unless the shedding processes differ significantly for the sublines in the foreign host animal. It is more likely that other processes instead of or in addition to shedding are involved in escaping transplantation rejection. Further studies of this and similar systems are necessary to try to define these factors. ACKNOWLEDGMENT We would like to thank technical assistance.

Charlene

Bymaster

for

REFERENCES 1. NICOLSON, G. L., AND POSTE, G. (1976) New Engl. J. Med. 295,253-258. 2. HELLSTROM, K. E., AND HELLSTROM, I. (19’74) Advan. Immunol. 18,209-277. 3. COOPER, A. G., CODINGTON, J. F., AND BROWN, M. C. (1974) Proc. Nat. Acad. Sci. USA 71, 1224-1228. 4. KIM, U., BAUMLER, A., CARROTHERS, C., AND BIELAT, K. (1975) Proc. Nat. Acad. Sci. USA 72, 1012-1016. 5. MILLER, D. K., AND COOPER, A. G. (1978) J. Bid. Chem. 253,8798-8803. 6. MILLER, D. K., COOPER, A. G., AND BROWN, M. C. (1978) J. Biol. Chem. 253, 8804-8811. 7. NORDQUIST, R. E., ANGLIN, J. H., AND LERNER, M. P. (1977) Science 197,366-367. 8. VAN BLIITERSWIJK, W. J., EMMELOT, P., HILKMANN, H. A. M., OOMEN-MEULEMANS, E. P. M., AND INBAR, M. (1977) Biochim. Biophys. Acta 467, 309-320. 9. BUCK, R. L., SHERBLOM, A. P., AND CARRAWAY, K. L. (1979) Arch. Biochem. Biophys. 198, 12-21. 10. SHERBLOM, A. P., HUGGINS, J. W., CHESNUT, R. L., BUCK, R. L., OWNBY, C. L., DERMER, G. B., AND CARRAWAY, K. L. (1980) Exp. Cell Res. 126, 417-426. 11. SHERBLOM, A. P., BUCK, R. L., AND CARRAWAY, K. L. (1980) J. Biol. Chem. 255,783-790. 12. CODINGTON, J. F., COOPER, A. G., BROWN, M. C., AND JEANLOZ, R. W. (1975) Biochemistry 14, 855-859.

50

HOWARD ET AL.

13. MILLER, S. C., HAY, E. D., AND CODINGTON, J. F. (19’7’7)J. Cell Biol. 72, 511-529. 14. HUGGINS, J. W., TRENBEATH, T. P., YELTMAN, D. R., AND CARRAWAY, K. L. (1980) Exp. Cell Res. 12’7, 37-46. 15. CARRAWAY, K. L., HUGGINS, J. W., SHERBLOM, A. P., CHESNUT, R. W., BUCK, R. L., HOWARD, S. P., OWNBY, C. L., AND CARRAWAY, C. A. C. (1978) in Glycoproteins and

Glycolipids in Disease Processes (Walborg, E. F., Jr., ed.), pp. 432-445, Amer. Chem. Sot. Symp. Ser. 80, Washington D. C. 16. DULBECCO, R. (1954) J. Exp. Med. 99, 167-199. 17. CARRAWAY, K. L., Doss, R. C., HUGGINS,

18.

19.

20.

21.

J. W., CHESNUT, R. W., AND CARRAWAY, C. A. C. (1979) J. Cell. Biol. 83, 529-543. LOWRY, 0. H., ROSEBROUGH, J. N., FARR, A. L., AND RANDALL, R. J. (1951) J. Biol. Chem. 193.265-275. HUGGINS, J. W., TRENBEATH, T. P., SHERBLOM, A. P., HOWARD, S. C., AND CARRAWAY, K. L. (1980) Cancer Res. 40, 1873-1878. SEGREST, J. P., JACKSON, R. L., ANDREWS, E. P., AND MARCHESI, V. T. (1971) Biochem. Biophys. Res. Commun. 44,390-395. KAPPELER, M., GAL-OZ, R., GROVER, N. B.,

AND

DOLJANSKI,

F.

(1973)

Exp.

Cell.

Res.

79, 152- 158. 22. TRUDING, R., SHELANSKI, M. L., AND MORRELL, P. (1975) J. Biol. Chem. 250.9348-9354. 23. RAHMAN,

A.

F.

R.,

LIAO,

S. K.,

AND

DENT,

P. B. (1977) In Vitro 13, 580-585. 24. TOKES, 2. A., AND DERMER, G. B. (1977) Supramol. Strut. 7, 515430. 25. BYSTRYN,

J.-C.

(1977)

J.

Nat.

Cancer

J.

Inst.

59, 325-328. 26. RITTENHOUSE,

H. G., RI~ENHOUSE, J. W., AND TAKEMOTO, L. (1978) Biochemistry 17, 829-837. 27. DAMSKY, C. A., LEVY-BENSHIMOL, A., BUCK, C. A., AND WARREN, L. (1979) Exp. Cell Res. 119, 1-13. 28. LITTLE, S. P., AND HUANG, A. S. (1978) J. Viral. 27, 330-339. 29. RAZ, A., GOLDMAN, R., YULI, I., AND INBAR, M.

(1978) Cancer Immunol. 53-59.

Immunother.

4,

30. STRITTMATTER, P., ROGERS, M. J., AND SPATZ, L. (1972) J. Biol. Chem. 247, 7188-7194. 31. FURTHMAYR, H., GALARDY, R. E., TOMITA, M., AND MARCHESI, V. T. (1978) Arch. B&hem. Biophys. 185,21-29.