Effect of glycosylation inhibitors on Akjlycosylpeptides and on invasion of

malignant mouse MO4 cells in vitro

ERIK A. BRUYNEEL1, MARC DE METS1, CHRISTIAN H. DRAGONETTI1, ROBERT J. HOOGHE2,

SERGIO DI VIRGILIO3 and MARC M. MAREEL1

'Laboratory of Experimental Cancerology, Department of Radiotherapy and Nuclear Medicine, University Hospital, De Pintelaan 1S5, B-9000Ghent, Belgiumdepartment of Radioprotection, SCK-CEN, B-2400 Mol, Belgium^Laboratoire de Chimie Gene'rale I, Universite Libre de Bruxelles, B-1050 Brussels, Belgium

Summary

Cell surface glycans are believed to play a role intumour invasion and metastasis. Yet, we havepreviously shown that the inhibitors of N-linkedglycan processing swainsonine (SW) and 1-deoxy-nojirimycin (dNM) did not prevent invasion ofchick heart fragments by MO4 murine fibrosar-coma cells in organ culture. We now present bio-chemical evidence that these and other inhibitors ofprocessing were indeed effective in remodelingglycans, including those expressed at the cell sur-face. After metabolic labeling with tritiated man-nose or fucose, glycosylpeptides were obtained byPronase treatment of material released from intactcells by trypsin. Glycosylpeptides were separatedby Biogel P-10 chromatography. With all drugstested, there was a shift towards lower molecularweight of the glycan chains. There were, however,major quantitative differences between the differ-ent drugs and also, for monensin (MON;0.1/igml"1), between fucose-labeled and mannose-labeled chains. The shift in apparent molecularweight affected mainly fucose-labeled peptidesafter treatment of MO4 cells with SW (0.4/igmr1).The shift induced by dNM (10mM)+SW(0.4/igml"1) in both fucosylated and mannosylatedchains was much larger than that induced by SWgiven alone. 1-Deoxymannojirimycin (dMM; lmM)had major effects on both mannose and fucose-

labeled structures and so did JV-methyl-1-deoxyno-jirimycin (MdNM; 2mM) and castanospermine (CS;100/igml"1). With the latter drugs, incorporation offucose in complex-type glycosylpeptides was dra-matically reduced. The effect of SW on fucose-labeled glycosylpeptides of embryonic chick heartwas similar to that observed on MO4 cells. Afterremoval of sialic acid, control and SW-treatedglycosylpeptides from both MO4 and embryonicchick heart cells had similar gel-chromatographicprofiles, suggesting that a decrease in cell surfacesialic acid accounts to a large extent for the differ-ence between glycans from control and SW-treatedcells. Additional biological experiments were donewith dMM (1 mM), MdNM (2mM), CS (100 fig ml"1),2,5-dihydroxymethyl-3,4-dihydroxypyrrolidine(DMDP; 250 fig ml"1) and SW (O^/igmr'j + dNM(10 mM). All these compounds or combinationsfailed to inhibit invasion. The observation thatinhibitors of N-linked glycan processing did notinterfere with invasion, although they clearly modi-fied the glycosylation of cell proteins, indicated thatthe integrity of glycans including those of the cellsurface might not be a prerequisite for invasion ofMO4 cells into living embryonic tissue in vitro.

Key words: glycosylation, invasion, MO4 cells, organ culture.

Introduction

Changes in cell surface glycosylation have been correlatedwith malignant transformation, on the one hand, andwith tumour progression towards invasiveness and meta-static capability, on the other. The biochemical evidencerested on structural analyses of glycans and glycosylatingenzymes (Smets and Van Beek, 1984; Dennis andLaferte, 1987; Bolscher et al. 1988a; Rademacher et al.

Journal of Cell Science 95, 279-286 (1990)Printed in Great Britain © The Company of Biologists Limited 1990

1988). Along the same lines, biological studies have reliedmainly on the use of glycosylation-defective mutant celllines and of inhibitors of protein glycosylation (Irimurae/al. 1981; Humphries et al. 1986a,fc; Dennis, 1986; Finneet al. 1989). Using the latter approach, Irimura et al.(1981) showed that melanoma cells precultured in thepresence of tunicamycin (TM) formed fewer pulmonarynodules than control cells after intravenous injection intosyngeneic mice. Humphries et al. (1986a,b) have made a

279

similar observation with swainsonine (SW) or castano-spermine (CS). Using an assay in vitro, we have shownthat inhibitors of N-linked glycosylation such as TMinhibited invasion of mouse malignant MO4 cells. Incontrast, swainsonine (SW) and 1-deoxynojirimycin(dNM) had no effect on the invasion of these cells(Mareel et al. 1985). Yet, both compounds have beenreported to interfere with the processing of N-linkedglycans and to prevent the synthesis of malignancy-related carbohydrate structures (Elbein, 1987; Mareeland De Metsj 1989).

Since it could be argued that the drugs were not activein our cell system, we undertook the biochemical charac-terization of the cell-surface glycans synthesized in thepresence of these and several other inhibitors of glycan-processing. The latter included vV-methyl- 1-deoxynojiri-mycin (MdNM), castanospermine (CS), 2,5-dihydroxy-methyl-3,4-dihydroxypyrrolidine (DMDP), 1-deoxy-mannojirimycin (dMM), and dNM + SW, all of whichalso failed to inhibit invasion in vitro (this paper). Withall inhibitors tested, the pattern of newly synthesized cellglycans recovered by trypsinization was clearly modified.Taken together, our data indicated that major changes incell-surface and possibly also secreted carbohydrates werecompatible with invasion in vitro.

Materials and methods

MO4 cellsMO4 cells (obtained from A. Billiau, KUL, Leuven, Belgiumthrough M. De Brabander, Janssen Life Sciences, Beerse,Belgium) are C3H mouse cells transformed by Kirsten murinesarcoma virus (Billiau et al. 1973); they were cultured in Eagle'smodified Minimal Essential Medium with Earle's salts and non-essential amino acids (EMEM, Flow Laboratories Ltd, Irvine,Scotland) supplemented with 10% (v/v) fetal bovine serum,0.05% (w/v) L-glutamine, 250 i.u. penicillin per ml and 100 Hgstreptomycin per ml (hereafter called culture medium). MO4cells were invasive in vittv (Mareel et al. 1979) and producedinvasive and metastasizing sarcoma-like tumours in syngeneicmice (Meyvisch and Mareel, 1982).

DrugsMonensin (MON) (Calbiochem-Behring, La Jolla, CA) wasdissolved in ethanol (1 mgml~ ) and further diluted in culturemedium to a final concentration of 0.1 ng ml~'; control cultureswith solvent alone are included in these experiments. The otherdrugs were directly dissolved in culture medium at the givenconcentrations: TM (NSC 177382, provided by the NaturalProducts Branch, Division of Cancer Treatment, NCI, Beth-esda, MD) at ljUgml"1; 2-deoxy-D-glucose (dGlc) (Merck,Darmstadt, FRG) at 5 and 20 mM; dNM (a gift from E.Trusheit and D. Schmidt, Bayer-AG, Wuppertal, FRG) atlOmM; MdNM (a gift from R. Schwarz, The RockefellerUniversity, NY, and U. Klein, Institut fur Virologie,Universitat Giessen, Giessen, FRG) at 2mM; CS (a gift fromL. Fellows, Royal Botanic Gardens, Kew, UK) at 100 and4 0 0 ^ g m r ' ; DMDP (a gift from L. Fellows) at 2S0figm\~l;dMM (a gift from Dr Schiiller, Bayer-AG) at 1 mM; SW (a giftfrom Dr P. Dorling, School of Veterinary Studies, MurdochUniversity, Western Australia) at 0.4 and 3.0jUgml~'. Concen-trations were chosen according to published data (reviewed byMareel and De Mets, 1989) and to personal communicationsfrom the donators.

These drugs were chosen because of their well-documented(Elbein, 1987) effects on different steps of glycoprotein biosyn-thesis and processing, namely: (1) the biosynthesis of thedolichol-sugar precursors (TM; dGLc); (2) the binding of theoligosaccharide G (Glc3Man9GlcNAc2) from dolichol-P-P tothe polypeptide backbone (HNV); (3) the trimming of oligo-mannosidic type structures in the Golgi apparatus by the use ofglucosidase (dNM; MdNM; CS; DMDP) and mannosidaseinhibitors (dMM; SW). MON is a sodium ionophore thatinterferes with processing and further decoration of the N-linked glycans with terminal fucose and sialic acid moieties.

Isolation and analysis of glycosylpeptid.esMO4 cells were seeded at an initial density of 1.5 X 105 cells per75 cm2 plastic tissue culture flasks (Flow, Irvine, Scotland, cat.no. 61-450B5).

Chick embryo heart (CEH) cells were prepared from 9-day-old embryos. After dissection, the ventricles were fragmented,rinsed in a Ca2+- and Mg2+-free salt solution (137mM-NaCl,8mM-KCl, 0.7mM-NaH2PO4.H2O, 12niM-NaHCO3, llmM-glucose, O.41T1M-KH2PO4; adjusted to pH7.4 with HC1) con-taining 0.03% EDTA (w/v) and trypsinized (0.25% (w/v)trypsin (Difco Laboratories, 1:250, Detroit, MA) in 137 mM-NaCl, 5mM-KCl, 0.7mM-Na2HPO4.2H2O, 11 mM-glucose,25mM-Tris-HCl; adjusted to pH7.7 with HC1) for 15min at37°C. Fetal bovine serum was added to neutralize the trypsin,and after centrifugation (lOmin at 150|>f) the pellet wasresuspended in culture medium to obtain 0.7x10 cells per20 ml.

MO4 or CEH cells were incubated on tissue-culture plastic at37°C for 4days; the medium was changed after 2 days,supplemented or not with one of the drugs. Cells, treated ornot, were metabolically labeled with either L-[6-3H]fucose(1/ iCimr1 ; 30Cimmor ' ) or D-[2-3H]mannose ( l ^ C i m P 1 ;19Cimmol~') during the last 20h of incubation. Radiochemi-cals were obtained from New England Nuclear (Boston, MA) orfrom the Radiochemical Center (Amersham, UK).

Glycosylpeptides were isolated in accordance with the tech-nique of Gilfix and Sanwal (1984). Briefly, cells were washedwith phosphate-buffered saline (PBS; 137mM-NaCl, lniM-CaCl2, 0.5mM-MgCl2.6H2O, 6.5 mM-Na2HPO4.2H2O,1.5mM-KH2PO4; adjusted to pH 7.4 with HC1) and trypsinizedfor lOmin at 37°C. After centrifugation (lOmin at 150 £) thesupernatant (called trypsinate) was exhaustively digested withPronase (from Streptomyces griseus, cat. no. 6174-01, Koch &Light, Genzyme Ltd, Suffolk, UK), and loaded on a90cmXl.6cm Bio-Gel P10, 200-400 mesh (Bio-Rad, Rich-mond, CA, cat. no. 150-150) column. Dextran Blue and PhenolRed were added to each sample in order to determine theexclusion and inner volumes of the column; fractions of 1.5 mlwere collected. Radioactivity was measured in the digestedsupernatant and in the cell pellet. Percentages of total radioac-tivity in the supernatant and radioactivity per mg protein of thecell pellet, as measured by the method of Lowry et al. (1951),were calculated.

Terminal sialic acids were removed from glycosylpeptides byincubation at 80°C in 0.1 M-H2SO4 during 90min. To liberateO-linked glycans from remnant peptides a reductive /3-elimin-ation reaction was done according to Carlson (1968). Extractionof glycolipids was performed according to Oliver and Hemming(1975).

These methods permitted us to analyse: (1) the effect of thedifferent glycosylation inhibitors on the apparent molecularweight of radiolabeled A'-glycosylpeptides from living celltrypsinates; (2) the amount of radioactivity released by trypsin;(3) the change of radiolabeled sugar incorporation per mgprotein in cultures on solid tissue-culture substrata.

280 E. A. Bruyneel et al.

Assay for invasionThe assay for invasion in vitro has been described previously(Mareel et al. 1979). Spheroids of MO4 cells (diameter=0.2 mm) were allowed to attach to precultured embryonic chickheart fragments (PHF, diameter=0.4mm), which consisted ofa core of myoblasts surrounded by a capsule of fibroblast-likecells, on top of a semi-solid agar medium for at least 2 h. Culturemedium, supplemented or not with drugs, was then added andindividual confronting pairs were transferred into a 5 ml Erlen-meyer flask containing 1.5 ml of the same medium. They werefurther incubated for 4 days at 37 °C on a GyrotoryR shaker(New Brunswick Scientific Co., New Brunswick, NJ) at120 revs min"1, fixed in Bouin Hollande's solution, embeddedin paraffin, and completely sectioned into 8fftn thick sections.Consecutive sections were stained with haematoxylin—eosin orwith an antiserum against chick heart (Mareel et al. 1981).Histological examination of invasion was done by at least twoobservers, based on an evaluation of occupation and degener-ation of the PHF. The score was as follows: grade I, when MO4cells were at the periphery of the outer fibroblast-like cells of thePHF; grade II, when MO4 cells had occupied only the outerfibroblast-like cell layers, but not the myoblasts; grades III andIV, when MO4 cells had occupied, respectively, less or morethan half of the PHF. According to Bracke et al. (1984), gradesIII and IV meet the criteria of invasion. We stress, here, thatinvasion in this assay correlates well with invasion but notnecessarily with metastasis in vivo (Mareel et al. 1987).

Results

GlycosylpeptidesThe radioactivity found in trypsinates of MO4 or of CEHcells varied between 19 % and 35 % of the total cellularradioactivity. The total radioactivity mg"1 cellular pro-tein of most treated cultures varied within 20 % of that ofthe control cells, showing a substantial overall incorpor-ation of fucose and of mannose. Exceptions were: CS(400 jig ml""1) doubled the incorporation of mannosewhen compared with untreated cells; MON (0.1 fig ml~ )reduced the incorporation of fucose to about half thevalue of control cells, whereas no quantitative differenceswere found after mannose labeling.

The gel chromatography patterns of the Pronase-digested glycosylpeptides from the radiolabeled MO4(Figs 1-3) and CEH cell surfaces (Fig. 4) could beinterpreted as follows (Warren et al. 1978): fractions35-40, eluting with or near to Dextran Blue, corre-sponded to proteoglycan-derived material; fractions41-65 contained the complex type JV-glycosylpeptides;fractions 66-90 represented the hybrid, oligomannosidicand immature ^-glycosylpeptides; finally, fractions91-110 and fractions 111-130 eluting close to Phenol Redwould contain, respectively, small precursors and freefucose or mannose. This interpretation was supported bypreliminary results from applications of the differentgroups to different lectins (data not shown). The distri-bution of the radioactivity found for fucose-labeledglycosylpeptides of untreated MO4 cell surfaces(Fig. 2A) was different from that of the correspondingmannose-labeled sample (Fig. 1 A), indicating that fucosewas preferentially incorporated in larger N-linked glycanswhile mannose was distributed over all types of struc-

tures. Even after extensive washing precursors or freelabel were found.

The following controls have been performed: (1) thegel chromatography profile of untreated MO4 cell glyco-sylpeptides was influenced neither by extraction of glyco-

• lipids nor by a reductive ^-elimination reaction (the latterprocedure liberates O-linked glycans and destroys rem-nants of the peptide moieties, if still present); (2)glycosidase contamination of the Pronase by a a-D-glucosidase, (X-D-mannosidase, <r-D-galactosidase, a-L-fucosidase, /3-D-glucuronidase, /3-D-galactosidase and iV-acetyl-/3-D-hexosaminidase could not be detected in asolution of 2mgml~' Pronase (J. Kint, personal com-munication).

The effects of drug treatments on gel chromatographyprofiles of trypsinates from cultures radiolabeled withfucose and/or mannose are illustrated in Figs 1-4. Ageneral feature of all treatments was the decrease oflabeled complex type Af-glycosylpeptides or a shifttowards a lower molecular weight with an increase insmall precursors and/or free fucose and mannose.

Since mannose was present in the core of all N-linkedglycans, labeling with [3H]mannose provided a con-venient approach for investigating the effect of inhibitors.In the absence of any treatment (Fig. 1A), a largefraction of the label was recovered in complex, and inhybrid plus oligomannosidic types of structure. Smallprecursors and free label were also found. The labeling ofcomplex type glycans with [3H] mannose was dramati-cally reduced with all inhibitors tested (Fig. 1), inparticular with dMM (Fig. 1C), when SW and dNMwere used together (Fig. IE), and with MdNM (2mM)and CS (100 and 400jigmr1) (data not shown). Therewas a compensatory increase of hybrid and oligomannosi-dic type structures after treatment with MON (Fig. IB),and with MdNM and CS (data not shown). Smallprecursors accumulated in cells treated with SW(Fig. ID), with dMM (Fig. 1C), with MdNM and withCS (data not shown).

Labeling of MO4 cells with fucose almost exclusivelyoccurred in complex type glycosylpeptides (Fig. 2A).After treatment with SW, fucose-labeled glycosylpep-tides were smaller than in control cells (Fig. 2D). Thiswas even more the case after treatment with SW plusdNM (Fig. 2E) and with TM (profile not shown).Fucose labeling of complex type glycosylpeptides wasvery low after treatment with MON or dMM(Fig. 2B,2C). At 5mM, dGlc produced gel profiles (notshown) similar to these obtained after SW treatment (seeFig. 2D), while 20mM dGlc gave rise to a profile (notshown) comparable to that of dMM (see Fig. 2C).

Mild acid-treatment of trypsinates (removing terminalsialic acids) caused a shift of the profile towards lowermolecular weights in fucose-labeled MO4 cells (Fig. 3).Such a shift was observed in fucose-labeled MO4 cellsthat were untreated (compare Fig. 3A and B) or treatedwith SW (compare Fig. 3C and D). Gel chromatographyprofiles from MON-treated MO4 cells labeled with man-nose or with fucose (see Figs IB and 2B) remainedunaltered after removal of sialic acids (profiles notshown).

Glycosylation and invasion in vitro 281

2A

030 50 70

t t90t

1101

1301Fraction

30

no.t50 70

t90t

1101

1301

Figs 1,2. Gel chromatography profiles of MO4cell glycosylpeptides on Bio-Gel P10. MO4 cellswere grown for 48 h at 37 °C in culture mediumalone (Figs 1A and 2A), supplemented withmonensin (0.1;Ugml~ ; Figs IB and 2B), 1-deoxymannojirimycin (lrain; Figs 1C and 2C),swainsonine (0.4^gml~'; Figs ID and 2D), orswainsonine+ 1-deoxynojirimycin(0.4^gmr ; Figs IE and 2E) and

3metabolically labeled with D-[2-3H]mannose(Fig. 1) or L-[6-3H]fucose (Fig. 2) during thelast 20 h of incubation. Ordinate: percentage oftotal radioactivity. Abscissa: fraction number;vertical bars in the abscissa indicate the elutionmaxima of Dextran Blue (left) and of PhenolRed (right); each experiment was repeated atleast once, with similar results (compare, forexample, Fig. 2A with Fig. 3A, and Fig. 2Dwith Fig. 3C). Arrows delineate zonescontaining presumably, from left to right, thecomplex type glycosylpeptides (fractions41—65), the hybrid, oligomannosidic orimmature vV-glycosylpeptides (fractions 65-90),the small precursors (fractions 91-110) and freelabel (fractions 111-130).

To examine whether inhibitors of glycoprotein pro-cessing would affect also the N-linked surface glycosyl-ation of the heart cells, CEH cultures were treated with0.4 ;Ug SW ml~ . The gel chromatography profile of fuco-sylated glycosylpeptides of untreated CEH cells shows abroader distribution of apparent molecular weights(Fig. 4A) as compared with that of untreated MO4 cells(see Fig. 2A). As observed for MO4 cells (see Fig. 2D),SW caused a shift of CEH fucosylated complex typeglycosylpeptides towards lower molecular weight frac-tions (Fig. 4B). As for MO4 cells, elimination of terminalsialic acids by acid treatment shifted the gel chromatogra-phy profile of control and SW-treated CEH cells to aregion of lower molecular weights (Fig. 4C and D).

InvasionPrevious experiments have demonstrated that TM, dGlcand MON inhibited invasion of MO4 cells in vitro in areversible way, whereas dNM and SW permitted in-vasion (Mareel et al. 1985). Here, we have extended theseobservations to the effects of other glucosidase and

mannosidase inhibitors on invasion. Treatment with SW,dNM or a combination of SW plus dNM in a matchedexperiment did not inhibit invasion of MO4 cells into thePHF (Table 1). The other trimming inhibitors dMM(Fig. 5), MdNM, CS, DMDP and dNM+SW alsopermitted invasion (Table 1). As an example, in thepresence of 1 mM-dMM MO4 cells have occupied andreplaced the heart tissue (Fig. 5) very much like controlswithout addition of drug. In contrast, with lOOmM-dGlcMO4 cells have surrounded the intact core of myoblasts(Fig. 6), as was also observed with many other kinds ofanti-invasive agents (reviewed by Mareel and De Mets,1989).

Discussion

The main finding of the present work is that importantchanges in many, if not most, of the glycosylpeptides didnot affect the invasion of MO4 cells in an assay in vitro.

Acquisition of the invasive phenotype by cultured cellpopulations after genetic or epigenetic manipulation has

282 E. A. Bruyneel et al.

3A 4A

30t

50 70tFraction

90tno.

not130t

Fig. 3. Gel chromatography profiles of MO4 cellglycosylpeptides on Bio-Gel P10. MO4 cells were untreated(A and B) or treated with swainsonine (0.4^gml~ ) (C andD) and labeled with L-[6-3H]fucose; trypsinates were acidtreated (B and D) ore not (A and C). Ordinate: percentage oftotal radioactivity. Abscissa: fraction number; vertical bars inthe abscissa indicate the elution maxima of Dextran Blue(left) and of Phenol Red (right). Arrows are as for Figs 1and 2.

been correlated with a shift of cell surface-exposed N-glycosylpeptides towards a higher molecular weight (Col-lard et al. 1986; Bolscher et al. 1986, 1988a,6; Bruyneelet al. 19896). Here, we were wondering whether drug-mediated reduction of the complexity of these glycosyl-peptides would interfere with the expression of theinvasive phenotype by MO4 cells in vitro. Our resultsindicated that a glycosylpeptide profile 'normalized' bytreatment with inhibitors of Golgi glucosidases or man-nosidases was compatible with the expression of theinvasive phenotype, at least in vitro.

Although glucosidase and mannosidase inhibitors havebeen extensively studied for their effects on a variety ofcellular functions (reviewed by Elbein, 1987; Rade-macher et al. 1988; Mareel and De Mets, 1989), theyhave rarely been used in assays for invasion. The organculture assay used in our experiments has been shown toscore for the confronting cells' capability of invading

o'•5

90 110t t IFraction no.

Fig. 4. Gel chromatography profiles of the cellglycosylpeptides of chick embryo heart cells (CEH) on Bio-Gel P10. CEH cells were grown on tissue-culture plastic inculture medium alone (A and C) or in the presence of0.4jUgml~ swainsonine (B and D) and labeled withL-[6- HJfucose. Sialic acid was removed from trypsinates ofuntreated (C) or swainsonine-treated (D) CEH cells.Ordinate: percentage of total radioactivity. Abscissa: fractionnumber; vertical bars in the abscissa indicate the elutionmaxima of Dextran Blue (left) and of Phenol Red (right).Arrows are as for Figs 1 and 2.

living tissue. This need not necessarily coincide withthese cells' metastatic capability, as was demonstrated forcell subtypes with high and low metastatic capability thatwere equally invasive (Mareel et al. 1987). CS and SWwere reported to reduce the capability of mouse B16melanoma cells to form lung colonies after intravenousinjection (Irimura et al. 1981; Humphries et al. 1986a,b\Dennis, 1986). These observations are not necessarily incontradiction with our present results, since reducedcolony formation might be ascribed to interference withsteps that are different from invasion per se, such assurvival in the circulation and site-specific adhesion tovascular endothelium.

Gel chromatography profiles, as performed in thepresent study, were not meant to reveal the nature of themolecular alterations responsible for shifts towards lower

Glycosylation and invasion in vitro 283

Table 1. Effect of inhibitors of glycoprotein processing on invasion of M04 cells into embryonic heart fragments invitro

Drug

1-Deoxynojirimycin')'jV-methyl-1-deoxynojirimycinCastanospermine2,5-Dihydroxymethyl-3,4-dihydroxypyrrolidine1-DeoxymannojirimycinSwainsoninefSwainsonine+1-deoxynojirimycin'fNonet

Concentration

10 mM2mM

100 ng ml"1

1 mM0.4/jgmP1

I

oo

oo

oo

oo

Grading*

II

00001000

of invasion after 4 days

III

02202012

IV

4*334I43§

*As described in Materials and methods: grades III and IV meet the criteria of invasion. For comparison: all confronting cultures treated' 1with monensin (0.1 ;<gml~'), 2-deoxy-D-glucose (100mM) or tunicamycin (1 jUgmF

•f Matched experiments.X Number of cultures examined histologically.

1) were scored as grade II (Mareel et al. 1985).

m

— 5B

<?*

i h h * m

6B

Figs 5,6. Photomicrographs of consecutive sections from confronting cultures between MO4 cell aggregates (m) and embryonicchick heart fragments (h) cultured for 4 days in presence of 1 mM-1-deoxymannojirimycin (Fig. S); inhibition of invasion bylOOmM-2-deoxy-D-glucose as also published previously (Mareel et al. 1985) is shown for comparison (Fig. 6); staining withhaematoxylin-eosin (Figs 5A and 6A) or with an antiserum against chick heart (Figs 5B and 6B). Bars, 50jUm.

284 E. A. Bruyneel et al.

molecular weights. Comparison between profiles fromSW-treated MO4 cells and profiles from untreated cellswith removal of terminal sialic acids (Fig. 3D and B)suggested a reduced sialylation due to SW treatment. Onthe other hand, analysis on concanavalin A-Sepharosecolumns has shown that in untreated MO4 cells tri- andtetra-antennary structures (85 %) prevailed over bi-antennary structures (15%), whereas after treatmentwith SW the former (14%) were inferior to the latter(86%) (unpublished results).

Dramatic reduction in fucosylation of jV-glycosylpep-tides was observed after treatment of MO4 cells with theanti-invasive compounds TM (not shown) and MON(Fig. 2B), but also with dMM (Fig. 2C), which permit-ted invasion. Inhibition of fucosylation has been relatedto the lack of invasion by MO4 cells in vitro at 28°C(Bruyneel et al. 1989a), and to reduced metastaticcapability of Eb as compared to ESb cells (Schwartz et al.1984) or of Wa4 as compared to B16-F1 melanoma cells(Finne et al. 1982, 1989). In the present experiments,dMM-treated MO4 cells were an example of low fucosyl-ation together with invasion. Thus, there was no simplecorrelation between the type of glycans expressed andinvasion or metastasis.

The anti-invasive agents (TM, dGlc and HNV), whichinterfered with the dolichol-P cycle in the endoplasmicreticulum, also inhibited growth (Mareel et al. 1985).Concentrations that differentially affected growth andinvasion, as described for other agents (Mareel et al.1982) were not found with these inhibitors of glycosyl-ation, making them less attractive for the study ofinvasion mechanisms.

Taken together, our present experiments indicated thatimportant changes in most, if not all, of the glycosylpep-tides were compatible with the invasion of MO4 cells invitro. This did not, however, rule out the possibleimplication of glycosylpeptides in invasion for a numberof reasons. (1) Some inhibitors of glycosylpeptide biosyn-thesis were shown to be anti-invasive (Mareel et al.1985); (2) the present gel chromatography profiles frommetabolically labeled cells demonstrated a modified gly-cosylation pattern between 28 and 48 h after adminis-tration of the drugs. Extrapolating to the invasion assay,we could state that MO4 cells had fully processedglycoproteins at the moment of initial contact with thePHF, and that this was sufficient for invasion to occur;(3) Glycosylation inhibitors affected both MO4 cells andCEH cells, in contrast with the aforementioned exper-iments in vivo where only the tumour cells were treated(Irimura et al. 1981; Humphries et al. 1986a,b\ Dennis,1986). For l-0-octadecyl-2-0-methylglycero-3-phospho-choline, an alkyllysophospholipid affecting N-glyco-sylation, it was shown that the antithetic effects oninvasion crictically depended upon treatment of eitherone or both of the confronting partners (Schallier et al.1988; Bolscher et al. 1988a; Bruyneel et al. 19896); (4)MO4 cells and CEH cells were treated separately toobtain glycosylpeptides for chromatography, whereas inthe invasion assay both partners were treated together.Since metabolic collaboration between MO4 cells andheart tissue has been suggested from other experiments

(Bruyneel et al. 1989a), we could not exclude thepossibility that cell surface glycosylation in confrontingorgan cultures responded to the glucosidase and mannosi-dase inhibitors in a different way as compared to separatecell cultures; (5) it could be that modification of glycosyl-ation by the drugs, which permitted invasion, was nevercomplete. Metabolic labeling allowed only the character-ization of newly synthesized glycans and glycoproteinswith a low turnover, perhaps influenced by the presenceof the drugs and unaltered glycans at the cell surface,could remain present during the invasion assay. There-fore, it might be that the molecules critical for invasionremained normally glycosylated.

Our present data confirmed previous observationsindicating that inhibitors of A'*-glycan processing do notprevent invasion and that this was the case for concen-trations of inhibitor that do alter glycosylation. There-fore, we supposed that the integrity of Ar-glycan chainswas not a prerequisite for invasion in vitro.

This work was supported by grants from the Nationaal Fondsvoor Wetenschappelijk Onderzoek (3.0032.87), and the Kan-kerfonds van de Algemene Spaar-en Lijfrentekas, Brussels,Belgium.

The authors thank L. Baeke, R. Colman and A. Verspeelt fortechnical assistance, J. Roels van Kerckvoorde for preparing theillustrations and G. Matthys-De Smet for typing the manu-script.

References

BILLIAU, A., SOBIS, H., EYSSEN, H. AND VAN DEN BERGHE, H.(1973). Non-infectious intracisternal A-type particles in a sarcoma-positive, leukemia-negative mouse cell line transformed by murinesarcoma virus (MSV). Arch. ges. Yirusforsch. 43, 345-351.

BOLSCHER, J. G. M., SCHALLIER, D. C. C , SMETS, L. A., VANROOY, H., COLLARD, J. G., BRUYNEEL, E. A. AND MAREEL, M.M. K. (1986). Effects of cancer-related and drug-inducedalterations in surface carbohydrates on the invasive capacity ofmouse and rat cells. Cancer Res. 46, 4080-4086.

BOLSCHER, J. G. M., SCHALLIER, D. C. C , VAN ROOY, H., STORME,G. A. AND SMETS, L. A. (1988a). Modification of cell surfacecarbohydrates and invasive behaviour by an alkyllysophospholipid.Cancer Res. 48, 977-982.

BOLSCHER, J. G. M., VAN DER BIJL, M. M. W., NEEFJES, J. J.,HALL, A., SMETS, L. A. AND PLOEGH, H. L. (19886). /tos(proto)oncogene induced alteration of cell surface carbohydrates and thesubsequent acquisition of invasive potential preceeds morphologicaltransformation. EMBOJ. 7, 3361-3368.

BRACKE, M. E., VAN CAUWENBERGHE, R. M.-L. AND MAREEL, M.M. (1984). (+)-Catechin inhibits the invasion of malignantfibrosarcoma cells into chick heart in vitro. Clin. exp. Metast. 2,161-170.

BRUYNEEL, E. A., BOLSCHER, J. G., SMETS, L. A., D E METS, M.AND MAREEL, M. M. (1989a). Restored invasion of mouse MO4

cells into chick heart in vitro through mutual conditioning atreduced temperature. Clin. exp. Metast. 7, 361-371.

BRUYNEEL, E. A., SCHALLIER, D. C , BOLSCHER, J. G., STORME, G.A., D E METS, M. AND MAREEL, M. M. (19896). Effect of racemic-l-0-octadecyl-2-0-methylglycero-3-phosphocholine (ET-I8-OCH3)on invasion in vitro and on N-linked surface glycosylation. InPharmacological Effects of Lipids (J. J. Kabarra, ed.), chap. 33,pp. 303-310, JAOCS, Champaign, IL.

CARLSON, D. M. (1968). Structures and immunochemical propertiesof oligosaccharides isolated from pig submaxillary mucins. J. biol.Chen,. 243, 616-626.

COLLARD, J. G., SCHIJVEN, J. F., BIKKER, A., LA RIVIERE, G.,BOLSCHER, J. G. M. AND ROOS, E. (1986). Cell surface sialic acid

Glycosylation and invasion in vitro 285

and the invasive and metastatic potential of T-cell hybridomas.Cancer Res. 46, 3521-3527.

DENNIS, J. W. (1986). Effects of swainsonine and polyinosinic:polycytidylic acid on murine tumour cell growth and metastasis.Cancer Res. 46, 5131-5136.

DENNIS, J. W. "AND LAFERTE, S. (1987). Tumor cell surfacecarbohydrate and the metastatic phenotype. Cancer Metast. Rev.5, 185-204.

ELBEIN, D. A. (1987). Inhibitors of the biosynthesis and processingof N-linked oligosaccharide chains. A. Rev. Biochem. 56, 497-534.

FINNE, J., BURGER, M. M. AND PRIEELS, J.-P. (1982). Enzymatic

basis for a lectin-resistant phenotype: increase in afucosyltransferase in mouse melanoma cells. J. Cell Biol. 92,277-282.

FINNE, J., CASTORI, S., FEIZI, T. AND BURGER, M. M. (1989).

Lectin resistant variants and revertants of mouse melanoma cells:differential expression of a fucosylated cell surface antigen andaltered metastizing capacity. Int. J. Cancer 43, 300-304.

GILFIX, B. M. AND SANWAL, B. D. (1984). Relationship between cellsurface asparagine-linked glycoproteins and myoblastdifferentiation. Analysis of wheat germ agglutinin-resistantmutants. Can. J. Biochem. Cell Biol. 62, 60-71.

HUMPHRIES, M. J., MATSUMOTO, K., WHITE, S. L. AND OLDEN, K.

(1986fl). Oligosaccharide modification by swainsonine treatmentinhibits pulmonary colonization by B16-F10 murine melanomacells. Proc. natn. Acad. Sci. U.SA. 83, 5131-5136.

HUMPHRIES, M. J., MATSUMOTO, K., WHITE, S. L. AND OLDEN, K.

(19866). Inhibition of experimental metastasis by castanosperminein mice: blockage of two distinct stages of tumor colonization byoligosaccharide processing inhibitors. Cancer Res. 46, 5215-5222.

IRIMURA, T., GONZALEZ, R. AND NICOLSON, G. L. (1981). Effects of

tunicamycin on B16 metastatic melanoma cell surface glycoproteinsand blood-borne arrest and survival properties. Cancer Res. 41,3411-3418.

LOWRY, B. H., ROSEBROUGH, N. J., FARR, L. A. AND RANDALL, R.J. (1951). Protein measurement with the folin phenol reagent. J.biol. Chem. 193, 265-275.

MAREEL, M. M., BRUYNEEL, E. A., D E BRUYNE, G. K.,

DRAGONETTI, C. H. AND VAN CAUWENBERGE, R. M.-L. (1982).

Growth and invasion: separate activities of malignant MO4 cellpopulations in vitro. In Membranes in Tumour Growth (T.Galeotti et al.} eds), pp. 223-232, Elsevier Biomedical Press,Amsterdam.

MAREEL, M. M., D E BRUYNE, G. K., VANDESANDE, F. AND

DRAGONETTI, C. (1981). Immunohistochemical study of embryonicchick heart invaded by malignant cells in three dimensionalculture. Invasion Metast. 1, 195-204.

MAREEL, M. M. AND D E METS, M. (1989). Anti-invasive activities ofexperimental chemotherapeutic agents. Crit. Rev. Oncol. JHematol.9, 263-303.

MAREEL, M. M., DRAGONETTI, C. H., HOOGHE, R. J. AND

BRUYNEEL, E. A. (1985). Effect of inhibitors of glycosylation andcarbohydrate processing on invasion of malignant mouse MO4 cellsin organ culture. Clin. exp. Metast. 3, 197-207.

MAREEL, M., KINT, J. AND MEYVISCH, C. (1979). Methods of studyof the invasion of malignant C3H-mouse fibroblasts into embryonicchick heart in vitro. Virchows Arch. B. Zellpath. 30, 95-111.

MAREEL, M., VAN ROY, F., MESSIAEN, L., BRACKE, M., BOGHEART,

E. AND COOPMAN, P. (1987). Investigation of tumour-invasionmechanisms. In Cells, Membranes, and Disease, Including Renal(E. Reid, G. M. W. Cook and J. P. Luzio, eds), pp. 359-370.Plenum Publ. Corporation, New York.

MEYVISCH, C. AND MAREEL, M. (1982). Influence of implantationsite of MO4 cell aggregates on the formation of metastases.Invasion Metast. 2, 51-60.

OLIVER, G. J. A. AND HEMMING, F. W. (1975). The transfer ofmannose to dolichol phosphate oligosaccharides in pig liverendoplasmic reticulum. Biochem. J. 152, 191-199.

RADEMACHER, T. W., PAREKH, R. B. AND DWEK, R. A. (1988).

Glycobiology. A. Rev. Biochem. 57, 785-838.SCHALLIER, D. C. C , BOLSCHER, J. G. M. AND SMETS, L. A.

(1988). Alterations in cell surface carbohydrates affecting invasionin vitro. In Lectins and Clycoconjugates in Oncology (H. J. Gabiusand L. A. Nigel, eds.), pp. 87-96. Springer-Verlag, Heidelberg.

SCHWARTZ, R., SCHIRRMACHER, V. AND MUHLRADT, P. F. (1984).Glycoconjugates of murine tumour lines with different metastaticcapacities. I. Differences in fucose utilization and in glycoproteinspattern. Int. J. Cancer 33, 503-509.

SMETS, L. A. AND VAN BEEK, W. P. (1984). Carbohydrates of thetumour cell surface. Biochim. biophys. Ada 738, 237-249.

WARREN, L., BUCK, C. A. AND TUSZYSKI, G. P. (1978).Glycopeptide changes and malignant transformation. A possiblerole for carbohydrate in malignant behaviour. Biochim. biophvs.Ada 516, 97-127.

(Received 31 July 1989 - Accepted 24 October 1989)

286 E. A. Bruyneel et al.