Ganglioside inhibition of attachment and differentiation of cultured rat

granulosa cells: interactions with fibronectin

PAUL MORLEY, DAVID T. ARMSTRONG and ROBERT E. GORE-LANGTON*

Medical Research Council Gmup in Reproductive Biology and Departments of Physiology' and Obstetncs and Gynaecology: University ofWestern Ontario, 339 W'indermere Road, London, Ontario, Canada X6A 5A5

•Author for correspondence at MRC Group in Reproductive Biology, Department of Obstetrics and Gynaecology', The University ofWestern Ontario, 339 Windermcre Road, London, Ontario, Canada N6A 5A5

Summary

The involvement of fibronectin in the attachmentand differentiation of rat granulosa cells, cul-tured in a chemically defined medium, wasinvestigated using the inhibitory properties ofmixed brain gangliosides (MBGs) and highlypurified disialoganglioside, GDia. MBGs in-hibited cell attachment to plastic culture surfacesin a concentration-dependent manner, with0-lmmoll"1 causing significantly decreasedattachment between 0-5 and 24 h of incubation.Inhibition of attachment to a fibronectin-coatedsubstratum was even greater. The inhibitory ef-fect of MBGs was not caused by binding to thecell surface, but instead the inhibitory factor(s)were adsorbed on a surface of immobilized hu-man plasma fibronectin, thereby preventing cellattachment to this surface. The inhibitory actionof MBGs was also neutralized by the addition ofsoluble fibronectin. Furthermore, at least onecomponent of MBGs, detected chemically follow-ing thin-layer chromatography, was directlyshown to bind to human fibronectin.

MBGs inhibited to varying degrees the follicle-stimulating hormone(FSH)-dependent responses:

augmentation of cellular protein content, pro-duction of adenosine 3',5'-cyclic monophosphate(cyclic AMP) and progestins (progesterone+20o/-hydroxypregn-4-en-3-one+pregnenolone), andinduction of aromatase activity. These inhibitoryactivities of MBGs could not be eliminated byadsorption on immobilized fibronectin orreversed by addition of soluble fibronectin, thusdistinguishing these actions from the early in-hibition of cell attachment. FSH-dependentresponses were also inhibited by GDia, whileresponses to stimulation by dibutyryl cyclic AMPplus 3-isobutyl-l-methyl xanthine were less affec-ted by this ganglioside.

These results suggest that gangliosides inhibitattachment of granulosa cells in culture by bind-ing to fibronectin, whereas the inhibition of FSH-dependent differentiation occurs by other modesof action that are unrelated to the effects on celladhesion.

Key words: adenosine 3',5'-cyclic monophosphate,fibronectin, follicle-stimulating hormone, gangliosides,granulosa cells, steroidogenesis.

Introduction

Various cell types that produce their own adhesionfactor(s) attach to plastic culture surfaces in the ab-sence of serum, and tend to spread on the culturesubstratum (Grinnell, 1978). Undifferentiated granu-losa cells, isolated from ovarian follicles of immaturerats and cultured in a chemically defined medium,attach similarly to plastic substrata, but without exten-Journal of Cell Science 88, 205-217 (1987)Printed in Great Britain © The Company of Biologists Limited 1987

sive cell spreading (Morley & Gore-Langton, 1985;Morley et al. 1987). Recent studies have demonstratedthe synthesis and secretion of fibronectin by rat granu-losa cells (Skinner & Dorrington, 1984; Skinner et al.1985), suggesting that this molecule mediates theadhesion in culture. Moreover, since fibronectin is amultifunctional dimeric adhesive glycoprotein withbinding domains for a cell surface receptor, collagenand several other components normally present in

205

complex extracellular matrices (Hynes & Yamada,1982), its production by granulosa cells suggests apossible role in mediating cell-cell and cell-matrixinteractions within the follicle. We previously foundthat a culture substratum coated with exogenous fibro-nectin caused greatly increased spreading behaviour ofrat granulosa cells, thereby enhancing growth, but didnot influence responsiveness to FSH (Morley et al.1987).

Gangliosides, particularly complex di- and tri-sialo-gangliosides, have been shown to have an affinity forfibronectin, resulting in inhibition of various fibronec-tin-mediated cell-adhesion processes in vitrv (Klein-man et al. 1979; Yamada et al. 1981). Furthermore, anaffinity for fibronectin may reflect a physiologicalfunction of certain complex gangliosides, which sup-port formation of a fibronectin matrix at the cell surfaceof NCTC 2071A cells (Yamada et al. 1983), and have arole in cellular attachment to extracellular matrixproteins (Cheresh et al. 1986). Although cell attach-ment to fibronectin is thought to involve a membraneglycoprotein complex, the putative fibronectin receptor(Brown & Juliano, 1985; Hasegawa et al. 1985; Knud-son et al. 1985; Pytela et al. 1985), it has also beensuggested that terminal sialic acid residues of cellsurface gangliosides may have a secondary role inpromoting optimal receptor— ligand interactions in-volved in cell adhesion (Cheresh et al. 1986).

To investigate the nature of granulosa cell adhesionvia endogenous adhesive glycoprotein(s), specificallyfibronectin, and the possible influence of this adhesionon differentiation, in the present studies we haveutilized the affinity of exogenous gangliosides forfibronectin in order to inhibit cell attachment in vitrvand assess effects on FSH-responsiveness.

Materials and methods

Animals

Immature female Wistar Crl:(WI)BR rats were obtainedfrom Canadian Breeding Laboratories (Montreal, Canada)and maintained with their mothers under standard tempera-ture (22°C) and lighting conditions (light 0500-1900h).Beginning at 19 days of age, they were treated for 3 days with1 mg diethylstilboesterol in 0 1 ml sesame oil by subcu-taneous injection, to increase cell yield and enhance respon-siveness to FSH (Goldenberg et al. 1972).

Cell cultures

Animals were killed at 22 days of age and ovaries removed.The ovarian bursa was peeled from each ovary and granulosacells were isolated into sterile culture medium by individuallypuncturing follicles with a 26 gauge needle. Cells wererecovered by centrifugation at 270g for 3 min and resus-pended in fresh culture medium. To ensure consistency inattachment assays cells were repeatedly pipetted with aPasteur pipette to obtain a uniformly disperse suspension.

The same method was used in preparing cell cultures foreither attachment or differentiation studies. Cultures nor-mally consisted of 500000 cells in 1 ml medium plated intoeach well in 24-well tissue culture plates (Falcon Plastics, LosAngeles, USA). In order to conserve material in those studiesusing highly purified ganglioside, these cultures consisted of80000 cells in 0 2 ml medium in 96-microwell culture plates.When fibronectin-coated substrata were used these wereprepared by incubating culture wells with medium contain-ing 5/ig human plasma fibronectin (Bethesda ResearchLaboratories, Gaithersburg, USA) per cm2 of culture surfacefor 45 min at room temperature, and then removing thismedium before plating cells. This procedure provides asurface that supports maximal granulosa cell spreading.Culture medium was a 1:1 (v/v) mixture of Dulbecco'smodified Eagle's medium and Ham's F12 medium containingl-2gl~' NaHCO3 and 15mM-Hepes (Gibco Canada, Inc.),plus antibiotics (50 units ml"' penicillin, 50jigml~' strepto-mycin and 0-625 fig ml" Fungizone). Incubations were car-ried out at 37°C in a humidified atmosphere of 5 % CO2 and95 % air.

FSH-dependent differentiation was studied in culturesincubated for 48 h in the presence or absence of ovine follicle-stimulating hormone (FSH), provided by the PituitaryHormone Distribution Program, N1AMDD (NIAMDD-oFSH-14; potency was 9x NIH-FSH-S1 by the hCGaugmentation bioassay of Steelman-Pohley). A concentrationof 500ngml"' FSH has been shown to give maximal steroid-ogenic responses. Other cultures were stimulated withlmmoll"1 dibutyryl adenosine 3',5'-cyclic monophosphate(dbcAMP) in the presence of 0'4mmoll" ' 3-isobutyl-l-methyl xanthine (IBMX), each purchased from SigmaChemical Co. (St Louis, USA). Testosterone (5xlO~7

mol 1" ) provided the substrate for oestrogen biosynthesis.Medium was sampled (0-2ml) after 2h for cyclic AMP assayand the remaining medium collected at 48 h for steroidassays. These incubation times were optimal for the respect-ive responses.

Cell attachment assay

After an incubation of normally 3 h, culture medium wasremoved and the cell monolayer washed manually with 0 5 mlof warmed CCvequilibrated medium. Firmly attached cellswere dispersed by incubating for 20 min at 37°C in calcium-free, magnesium-free Hanks' balanced salt solution contain-ing 0-25% trypsin (Sigma Chemical Co.) and 0-lmmoll" 'ethylenediamine tetraacetic acid (EDTA). A sample of dis-persed cells was counted in a model ZBI Coulter Counter(Coulter Electronics, Hialeah, USA). In separate studies, thetotal recovery of cells (attached + unattached) by this methodwas determined to be 97 ± 4 % (mean ± S.E.M., »/ = 24).Because of this consistently high cell recovery, corrections forcell losses were not made. Also, we did not find a largevariation in the total recovery of cells given different treat-ments. Inhibition of cell attachment was studied by additionto the culture medium of either bovine mixed brain ganglio-sides (MBGs) (Sigma Chemical Co.), which contain anundefined mixture of ganglioside types, or highly purifieddisialoganglioside, GD) a (Supelco, Bellafonte, USA). Gang-lioside-treated medium was prepared from stock solutions inchloroform:methanol (2:1, v/v), by drying samples under

206 P. Morlev et al.

nitrogen and then vortexing them in medium to bring theminto solution. Concentrations of MBGs are approximate sincethey were calculated on the basis of the average molecularweight of the four major gangliosides: GM], GM2, GDl a andGT, .

Thin-layer chmmatography of gangliosidesGangliosides dissolved in chloroform:methanol (2:1, v/v)were applied to LK6DF Linear-K silica gel thin-layerchromatography plates (Whatman Inc., Clifton, USA) anddeveloped as described (Griffiths et al. 1986) in chloroform:methanol:water (60:35:8, by vol.). Sialic acids of ganglio-sides were detected by spraying with the resorcinol reagent(Svennerholm, 1957) and heating plates until spots appeared.

Radioimmnnoassays of secreted productsA previously described radioimmunoassay (Reddoch & Arm-strong, 1984) was used to measure adenosine 3',5'-cyclicmonophosphate (cyclic AMP) in culture medium. For thispurpose, medium was collected and immediately heated in aboiling water bath for 20min to destroy phosphodiesteraseactivity, then stored at — 20°C. Steroids were measured inmedium stored at —20°C until assay. Radioimmunoassayswere previously validated for direct measurements: preg-nenolone (InabaeJ al. 1979; Lischinsky & Armstrong, 1983),progesterone (Leung & Armstrong, 1979), 20a/-hydroxy-pregn-4-en-3-one (Morley et al. 1987) and oestradiol-17/3(Daniel & Armstrong, 1984). The sum of the production ofpregnenolone, progesterone and 20a'-hydroxypregn-4-en-3-one (20a'-P) is referred to as progestins in this study, andcollectively reflects the activity of the rate-limiting enzyme,the cholesterol side-chain cleavage system.

Protein assay

Protein contents of those cultures used to study differen-tiation were determined at the end of the 48-h incubation.Attached cells were solubilized by sonication in 10 ml of 1 %sodium dodecyl sulphate and l m m o l P ' EDTA, and totalprotein was measured by the method of Lowry et al. (1951).The protein content of unattached cells in culture medium,which was stored frozen, was determined similarly afteraddition of one third volume 40% trichloroacetic acid andovernight incubation on ice, followed by centrifugation torecover the precipitate.

StatisticsWhere required, data were transformed logarithmically toremove heterogeneity of variance as determined by Bartlett'stest. Analysis of variance was performed and Duncan's newmultiple range test was subsequently used for multiplecomparisons, as described by Steel &Torrie (1980).

Results

Time course of attachment and inhibition by mixedgangliosides

Attachment of granulosa cells to plastic substratumincreased in a linear fashion between 30min and 4h,with only a slight non-significant further increase in

1001

'60-

0-05 mmol P 1 ; 28-3 ± 0-9% attachment occurred at0-1 mmol P 1 gangliosides compared with 64-6 ± 7 - 3 %attachment in the absence of gangliosides. These valuesindicate that 44% of the cells that showed attachmentin control incubations still attached at supramaximalconcentrations of gangliosides. Thus, a portion of thegranulosa cells attach to plastic surfaces by a mechan-ism that is insensitive to gangliosides.

Ganglioside effects on granulosa cells 207

To test whether this ganglioside-insensitive attach-ment might be due to non-specific adsorption to theplastic, a parallel attachment study was performedusing a fibronectin-coated substratum. The coatingprocedure was shown in previous investigations toexceed that required for optimal cell spreading activity,hence it was considered to provide a substratum onwhich fibronectin-mediated cell adhesion primarilyoccurred. Cell attachment to this fibronectin-coatedsubstratum was the same as on plastic in the absence ofgangliosides (Fig. 2). However, ganglioside-mediatedinhibition of attachment to fibronectin-coated sub-stratum, in comparison with plastic substratum, wasmore effective at all ganglioside concentrations >0-05mmoir ' . Cell attachment at 0-1 mmol P 1 ofgangliosides was 14-1 ±0-4%, which is significantlyless than on the uncoated plastic surface.

Inhibition of attachment by mixed gangliosides isrelated to fibronectin binding

Adsorption of inhibitory activity on immobilized fibro-nectin. To determine whether the inhibitory com-ponents present in the mixed gangliosides could beremoved by binding to fibronectin, culture medium

80-

60-

40-

20'

0-025 0-05 0-075 01Gangliosides (mmoir1)

0-2

Fig. 2. Concentration dependence of inhibition by mixedbrain gangliosides of attachment of rat granulosa cells.Cells were cultured on plastic ( • • ) or fibronectin-coated ( • • ) substrata. Cells were plated into mediumcontaining the indicated concentrations of mixedgangliosides. Cell attachment after 3 h was determined asdescribed in Materials and methods. Data are themean ± S.E.M. of triplicate cultures. Where error bars arenot shown the S.E.M. was between 0-2 and 1-3%. Asterisksindicate where there is a significant difference ( P < 0 0 5 ) inattachment on the two substrata. Significant (P

Table 1. Effects of unadsorbed and fibronectin-adsorbed mixed brain gangliosides on attachment of

rat granulosa cells to a plastic substratum

Culture medium % Attached cells

M, PIM, FnMBGs, PIMBGs, Fn

62-3 ± 2 - 1 '59-8 ±3-4 '19-0 ± 0-6b

57-0+1-8"

Granulosa cells were plated in plastic culture wells in mediumalone (M) or medium treated with O' lmmolP1 mixed brainganghosidcs (MBGs). These media were previously adsorbed byincubating for 1 h in either a standard plastic dish (PI) or in a dishcoated with human plasma fibronectin (Fn). Cell attachment wasassessed after 3 h as described in Materials and methods. Data arethe mean ± S.E.M. of quadruplicate cultures. Values with differentsuperscripts (a,b) are significantly different (P

acid-containing component of MBGs was preferen-tially and tightly bound to immobilized human fibro-nectin. A standard 60 mm culture dish and a fibronec-tin-coated dish, prepared as previously described, wereeach incubated for l h with a total of l-36mg MBGsin culture medium (0-15 mmolP1) . Medium wasremoved, the surfaces washed twice with phosphate-

GM,—

GD l a -

Or

D

Fig. 4. Thin-layer chromatography of mixed braingangliosides (MBGs) and of those components with anaffinity for immobilized fibronectin. A. 100^g MBGs;B, 10/Jg MBGs; C, material eluted with methanol from afibronectin-coated layer after incubation with MBGs inculture medium and subsequent washing of the surfacewith phosphate-buffered saline; and D, material elutedwith methanol from an uncoated plastic surface after asimilar incubation with MBGs. Following chromatographythe sialic acid-containing materials were visualized with theresorcinal reagent. The position of the origin (O) and ofstandard gangliosides are shown at the left. The arrow onthe right indicates the position of one component with anaffinity for fibronectin.

buffered saline and the remaining tightly bound gang-liosides were eluted with absolute methanol. Followingresuspension in chloroform:methanol (2:1, v/v) theeluted materials were applied to a thin-layer plate, thenchromatographed and detected by the resorcinol re-agent. The preparation of MBGs was used as areference and when 100 fig was applied four major spotswere detected with blue-violet reactions typical ofsialic acids (Fig. 4); the approximate RF values were0-31, 0-29, 0-25 and 0-19. Only the spot at RF 0-25 wasstill distinct when just 10 fig MBGs was subsequentlychromatographed. The bound material that elutedfrom the immobilized fibronectin showed a faint spotwith Rp values similar to this latter MBG componentand to the GDia standard, while elutions from thestandard plastic dish resulted in no detectable spots.Thus, it is evident that only one of the detectablecomponents of the MBG preparation was tightly boundto fibronectin and that the amount bound was only asmall fraction of the total gangliosides incubated. Theidentity of the fibronectin-bound component is notestablished from this study but the results suggest thepossibility of a disialoganglioside.

Ganglioside effects on differentiationSince gangliosides inhibited attachment of rat granu-losa cells in culture and this inhibitory activity could beremoved by adsorption on fibronectin, we used thismodel to investigate effects of mixed gangliosides ondifferentiation. Effects of gangliosides were investi-gated by comparing FSH-dependent differentiatedresponses obtained without the addition of gangliosideswith those obtained by addition of O-lmmoll"1

MBGs, either adsorbed in a fibronectin-coated dish toremove factors inhibitory to cell attachment, oradsorbed in a standard plastic dish to control for anynon-specific removal of components.

Measurements of protein content of attached cells atthe end of the culture period were used to assess boththe overall effects on cell adhesion and the additionaltrophic effects of FSH. As shown in Fig. 5A, FSHtreatment significantly increased the protein content ofcultures, while ganglioside preparations inhibited boththe basal and FSH-stimulated contents to a similarextent (i.e. 52 and 57% of the control level, respect-ively, after 48 h). However, the adsorption of ganglio-sides on fibronectin did not eliminate their inhibitoryeffect on protein content as measured after 48 h, incontrast with the previously demonstrated removal ofinhibitory activity affecting cell attachment at 3h. Weinterpret this as indicating that certain components ofthe MBG preparation that are not quantitativelyremoved by adsorption on immobilized fibronectincause reduced cell attachment at longer incubationtimes.

210 P. Morlev et al.

100

80-

S 60.

a 40.o

20'

Control

60'

a 40.3

"5Q.

20.

Control

nd nd nd

FSH

M +G +PG M +G +PG M +G +PG M +G +PG

40'

aoD.

.2 20

UO4

FSH -r

Control

nd nd nd

2 2-Q.

FSH

Control

nd nd nd

M +G +PG M +G +PG M +G +PG M +G +PG

Fig. 5. Effects of mixed brain gangliosides on FSH-dependent responses of rat granulosa cells. Protein content of cellmonolayer (A), and production of cyclic AMP (B), 17/5-oestradiol (C) and total progestins (D) were determined. Cells werecultured in the presence of 5xlO~7moll~' testosterone, either without gonadotropin stimulation (control) or plus500ngml~' FSH. Culture media were prepared as follows: without gangliosides (M), plus 0-1 mmoll"' bovine mixed braingangliosides ( + G), or plus the same gangliosides after preadsorption for 1 h in a dish coated with human plasma fibronectin( + PG). A sample of medium was removed for cyclic AMP assay 2h after plating cells and the remaining medium collectedafter 48 h for determination of secreted steroids. Protein content of monolayers was determined only at the end of 48 h,hence 2h cyclic AMP results are expressed on a per-culture basis. Progestins represent the total production ofpregnenolone, progesterone and 20a"-P. Data are the mean ± s.E.M. of quadruplicate cultures. Bars with different letters aresignificantly different (P

oestradiol was substantially reduced by gangliosidetreatments (not shown), the results when normalizedby protein content of the cell monolayers were notsignificantly different when compared with culturesnot given gangliosides (Fig. 5C). Oestrogen pro-duction was similarly affected by gangliosides treatedby adsorption on a fibronectin-coated surface, indi-cating that the inhibitory effects of certain gangliosideson early cell attachment were of no consequence toFSH-stimulated oestrogen production.

Similar effects of MBGs were observed on theproduction of progestins, except that even the datanormalized by monolayer protein content showed sig-nificant inhibitory effects of ganglioside treatments(Fig. 5D). When total production of progestins wasdecreased in the presence of MBGs, this was due tosimilar decreases in progesterone and 20a-P (resultsnot shown). Thus, these results for oestrogen andprogestin secretion indicate that mixed gangliosidesinhibit FSH-stimulated steroidogenic differentiation,but that the inhibitory factor(s) cannot be quantitat-ively removed on the basis of their affinity for fibronec-tin.

Two further experiments, identical to that describedabove, were carried out in the presence of 2 orlOjigml"' human plasma fibronectin in order to testwhether the inhibitory effects of gangliosides on differ-entiated responses could be reversed by fibronectin, aspreviously shown for the inhibitory effects on early cellattachment. The results (not shown) were similar tothose already presented in Fig. 5 and established thateffects of mixed gangliosides on FSH-stimulated cellu-lar protein and steroidogenic responses were notreversed by addition of exogenous fibronectin.

The above experiments indicate that gangliosidecomponents in addition to those that bind to fibronec-tin are inhibitory to steroid production. However, sincethe extent to which unattached cells may continue tosecrete 17/3-oestradiol and progestins is unknown, thedegree of inhibition of steroid production by MBGsmay be underestimated as a consequence of the methodof normalizing steroid values by the amount of proteinin the attached monolayer. To examine this possibilityan additional experiment was performed to comparesteroid production by granulosa cells that were at-tached or unattached after 24 h incubation in either theabsence or presence of O-lmmoll" MBGs. Thus,after 24 h in culture, 'unattached' cells were obtainedby transferring the medium containing unattached cellsto a new culture well. 'Attached' cells were obtainedfrom other identical cultures by first centrifuging theunattached cells from the culture medium and thenreturning this original medium to the monolayer. Thislatter manipulation required that the attached cells bebriefly incubated with fresh medium while the originalmedium was being centrifuged. A third 'recombined'

group, which acted as an additional control, consistedof attached and unattached cells recombined together,after first subjecting cells to the same manipulationsdescribed for the 'attached' group except that unat-tached cells in the culture medium were not removedby centrifugation. By keeping cells of all groups in theoriginal plating medium, we have tried to avoid poss-ible inadvertent effects arising from changes in theculture medium or treatments. Also, the steroid con-tent in the medium after the first 24 h was measured inidentical parallel cultures, and the relatively low valuesobtained were then subtracted from the values obtainedfrom the cultures incubated for 24—48 h, thus giving atrue measure of steroid production occurring between24 and 48 h. The results given in Table 3 show thattreatment with MBGs inhibits both the percentage cellattachment at 24 h, and the subsequent extent of cellattachment assessed at 48 h in each of the aboveexperimental groups, as based on final protein contentin the monolayer and in the medium. First, in theabsence of MBGs the cells from each of the 'attached','unattached' and 'recombined' groups at 24 h sub-sequently resulted in a similar high degree of cellattachment at 48 h as judged from the negligibleamount of protein found in the medium. In contrast,each of the groups treated with MBGs showed substan-tial increases in the amount of protein in the medium at48 h. In the 'recombined' groups, the total proteinrecovered in the monolayer plus medium was similarin either the absence or the presence of MBGs, andin comparison with the total protein recovered in'attached' + 'unattached' groups was within the vari-ation expected from the manipulations. In the absenceof MBGs, steroid secretion over 24— 48 h when nor-malized by the total protein recovered, was not signifi-cantly different for each of the experimental groups,thus suggesting that those cells that were attached orunattached at 24 h subsequently have similar steroid-ogenic capabilities. With MBGs present, both 17/3-oestradiol and progestin production were markedlydecreased in each of the experimental groups. Theresults show, however, that the cells that were unat-tached after 24 h continue to secrete both classes ofsteroids, and that the normalized steroid production issignificantly greater than for the attached cells alone.These findings suggest that while gangliosides decreaseboth cell attachment and overall steroid production,unattached cells continue to contribute significantly tosteroid production. Thus, the overall inhibitors' effectsof MBGs on steroid production, as shown in theprevious experiment, may in fact be underestimated asa result of normalizing steroid values by the proteincontent of the cell monolayer alone.

212 P. Morley et al.

Table 3. Steroid production by attached and unattached granulosa cells cultured in the absence or presence ofmixed brain gangliosides

MBGs

-

+

cells at 24 h

88-0 ±4-8

20-0 ±2-7

Cell groups

AttachedUnattachedRecombined

AttachedUnattachedRecombined

Monolayer

62-9 ±5-831-6±2-l82-0 + 2-8

24-1+5-440-0 ±4-546-9± 1-4

Protein contentat 48 h (fig)

Medium

1-7 ± 1-3ndnd

26-7 ±4-011-5 ± 1-229-5 ±2-2

Total

64-6 ±431-6±282-0 ±2

50-8 ±751-5 ±576-4 ±3

•6•1

•8

•6•8•4

Steroidover

(ngmg-'

17^-Oestradiol

19-7 ±0-7'18-5 ± 1-9*22-1 ±0-8"

0-9 ±0-5"ll-2±0-6b

6-5±0-9b

production24-48 htotal protein)

Progestins

2756 ±1041"3202 ±616*2011 ±324'

394 ±158"1235 ± 127h

671 ±63l b

Granulosa cells were cultured in medium containing FSH (500ngml ') and testosterone (5 X 10 7moll '), and either in the absence orpresence of 01 mmoll"' of MBGs. After 24 h culture, cells were separated as three groups according to whether they were 'attached','unattached' or 'recombined' from both attached and unattached cells, as described in Results. After 48 h in culture, the protein contents ofmonolayer and medium were determined, and the steroids in the medium assayed. Steroid values shown were adjusted by subtracting themean values determined for medium at 24h, thus giving the steroid production over 24—48h. Data are the mean ± S.E.M. for quadruplicatecultures, nd, not detectable. MBGs significantly decreased 17/3-oestradiol and progestin production. Within the —MBGs or +MBGstreatment groups, different superscripts (a,b) signify that steroid production values between cell groups are significantly different(P

ganglioside on granulosa cell differentiation are inde-pendent of the substratum used in the cultures.

Discussion

We have demonstrated that the /;; vitm adhesion of ratgranulosa cells to either plastic- or fibronectin-coatedsubstrata is substantially inhibited in a concentration-dependent manner by bovine mixed brain gangliosides(MBGs). Our results suggest that MBGs inhibit theearly (3 h) attachment of granulosa cells as a directresult of binding to fibronectin. Evidence for interac-tions of the inhibitor}' component(s) in MBGs withexogenous fibronectin include the following: the in-hibitory component(s) were quantitatively adsorbed onimmobilized human plasma fibronectin, cell attach-ment to a fibronectin-coated surface was inhibitedwhen the surface was treated with MBGs then washedto remove unbound or weakly bound material, and theinhibition of attachment was completely reversed byaddition of soluble plasma fibronectin. Furthermore,the inhibitory effects of MBGs were not the result ofthe direct binding of components to the granulosa cellsurface. These findings are completely consistent withthe studies of Kleinman et al. (1979) demonstratingthat mixed or tri- and disialogangliosides interactwith fibronectin, thereby inhibiting the fibronectin-mediated binding of Chinese hamster ovary cells to acollagen substratum. Gangliosides have also beenreported to inhibit the stimulatory effects of fibronectinon spreading behaviour of baby hamster kidney cells(Perkinses al. 1982«; Yamadae/a/. 1981), haemagglu-tination of erythrocytes (Yamada et al. 1981), and therestoration of a normal phenotype in transformed cells(Yamada et al. 1981). In addition, gangliosides in-hibited the binding of fibronectin to human fibroblasts(Rennard et al. 1981), and fluorescent gangliosideswere shown to bind to the fibronectin matrix producedby human fibroblasts (Spiegel et al. 1984).

Our results also provide direct evidence that only onemajor chromatographically separated component of theMBGs binds tightly to immobilized human fibronec-tin, while none of the major components binds to aplastic culture surface. Thus, the ganglioside com-ponents that inhibit cell attachment and have anaffinity for fibronectin may include or be identical tothe component detected chemically following chroma-tography. Also, the ganglioside components that do notadsorb to fibronectin, which predominate in MBGs,may be responsible for the various non-specific inhibi-tory effects discussed below.

Although our results confirmed the interaction ofgangliosides with exogenous fibronectin, we wereunable to test directly whether gangliosides inhibitedgranulosa cell attachment by interactions with theendogenous fibronectin secreted by rat granulosa cells.

The specificity of ganglioside inhibition of fibronectin-mediated cell adhesion has been questioned by thefinding that cell attachment to immobilized lectinsubstrata was also inhibited (Perkins et al. 19826;Rauvala et al. 1981). Therefore, we cannot rule outpossible interactions between gangliosides and otheradhesion factors produced by granulosa cells. Never-theless, we presume fibronectin to be the principal ifnot the only adhesion molecule involved in granulosacell attachment, since fibronectin constitutes the mostabundant labelled protein secreted into the culturemedium of rat granulosa cells incubated with[35S]methionine (Skinner & Dorrington, 1984). Fur-thermore, interactions of gangliosides with granulosacell membrane components seem unlikely to be thecause of decreased cell attachment, since cells incu-bated for 1 h with gangliosides, then washed to removeunbound material, subsequently showed normal at-tachment to plastic.

Since maximally effective concentrations of ganglio-sides inhibited granulosa cell attachment to plasticsubstrata to only 44% of control levels, other ganglio-side-insensitive adhesion mechanisms could be in-volved in the observed attachment. A probable trivialexplanation of ganglioside-insensitive binding wouldbe the non-physiological adhesion to plastic via adsorp-tion sites, which remain unoccupied in low proteinsolutions, as described (Grinnell & Feld, 1979). Thisinterpretation is consistent with the more effectiveinhibition by gangliosides of attachment to a fibronec-tin-coated substratum.

In contrast with the inhibitory effects of MBGs onattachment, the inhibitory effects on FSH-stimulatedresponses do not appear to be due to components thatspecifically bind fibronectin. Thus, a similar degree ofinhibition of FSH-stimulated cyclic AMP productionoccurred in cultures treated with unadsorbed MBGs orMBGs previously adsorbed on immobilized fibronec-tin. Since cyclic AMP accumulation was measured 2hafter the start of culture, it is apparent that theinhibition of this response was not attributable to thedifferences in early cell attachment occurring betweencultures in the presence of unadsorbed and fibronectin-adsorbed gangliosides.

It is possible that the inhibitory action of ganglio-sides on cyclic AMP production is due to effects on theFSH receptor. Gangliosides have been reported toinhibit the binding of several glycoprotein hormones totheir receptors (Fishman & Brody, 1976). Further-more, gangliosides are associated with the glycoproteinreceptor for thyroid-stimulating hormone (TSH) (Alojet al. 1979), and gangliotriaosylceramide (GM3) formsa functional association with the transferrin receptor(Okada et al. 1985). Since the components of MBGsthat inhibit FSH-dependent responses are not readilyadsorbed on immobilized fibronectin, it suggests the

214 P. Morley et al.

possibility that monosialogangliosides, which areknown to be least potent in the inhibition of cellattachment (Kleinman et al. 1979), may be responsiblefor impairing FSH action. In any case, it is apparentfrom our results that a complex ganglioside, GDia, iseffective in inhibiting FSH action.

The inhibitory effects of MBGs on FSH-stimulatedsecretion of 17/S-oestradiol and the progestins, during a48 h incubation, were also apparently not caused bycomponents with an affinity for fibronectin, as indi-cated by similar actions of MBGs regardless of adsorp-tion on immobilized fibronectin, and by the failure ofexogenous fibronectin to reverse these effects. Furtherinterpretation of the inhibitors' effects of MBGs onthese differentiated responses is complicated by thefinding that cellular protein content of cultures after48 h, as a measure of adherent cells, was also similarlyinhibited in cultures treated with unadsorbed or fibro-nectin-adsorbed MBGs. Thus, certain components ofMBGs that do not have strong affinities for fibronectinare apparently responsible for decreased cell adhesionin longer incubations. From these observations alone itis therefore uncertain whether the inhibitory effects onsteroid production could be a consequence of thereduced cell adhesion or, as is more probable, the resultof other effects of gangliosides, including the inhibitionof FSH-stimulated cyclic AMP production. However,our demonstration that unattached cells from ganglio-side-treated cultures produce relatively more steroidsthan attached cells, strongly suggests that the inhibi-tory effects of gangliosides are not a consequence of thedecreased cell adhesion. Since purified GDin alsoinhibited steroid responses following stimulation withFSH, and to a lesser extent following stimulation withdbcAMP plus IBMX, it seems likely that the inhibitoryactions of the MBG preparation are attributable togangliosides and not contaminants.

Our results suggest that gangliosides may impairFSH responsiveness by effects on cyclic AMP pro-duction and perhaps by more distal actions. Variousactions of gangliosides are possible as a consequence oftheir spontaneous insertion into plasma membranes.Recent studies have shown that gangliosides causefibronectin retention in cell surface fibrillar networks(Yamada et al. 1983; Spiegel et al. 1984). Also,evidence for Ca2+ binding by terminal sialic acids of di-and trisialogangliosides (Abramson et al. 1972; Gol-denring et al. 1985), and additional evidence for theregulation by Ca2+-ganglioside complexes of specificprotein phosphorylation in a membrane fraction frombrain (Goldenring et al. 1985), suggests the furtherpossibility that gangliosides could become inserted intomembranes and influence Ca +-mediated events.

The reasons for the relatively greater steroid pro-duction by unattached, rounded cells cultured in thepresence of gangliosides, has not been determined. It

has been shown that treatment of cultured rat granu-losa cells with microtubule-depolymerizing agents,thus causing cell rounding, results in increased ster-oidogenic activity (Carnegie et al. 1987), but otherstudies have shown that the extent of granulosa cellspreading on a fibronectin substratum does not influ-ence FSH-dependent steroidogenic responses (Morleyet al. 1987). Further possible explanations for ourfindings might be that gangliosides cause preferentialinhibition of attachment of the most steroidogenicallyactive cells or that unattached cells are less sensitive tothe toxic actions of gangliosides.

In conclusion, these studies suggest that at least onecomponent of mixed brain gangliosides inhibits theearly attachment of cultured rat granulosa cells bybinding to fibronectin. Inhibitory effects on adhesionafter longer incubation periods, however, might in-volve other factors (probably gangliosides) that are notreadily adsorbed to immobilized fibronectin. Mixedgangliosides and purified GD|n were both inhibitory toFSH-stimulated production of cyclic AMP and ster-oids, and these effects of gangliosides appear to beunrelated to inhibition of fibronectin-mediated attach-ment. Inhibition of FSH-stimulated steroidogenesismight result from the observed inhibition of cyclicAMP production or from undetermined actions on theplasma membrane.

We are grateful to H. E. Ross for excellent technicalassistance. This work was supported by grants from theMedical Research Council of Canada (Group grant toR.E.G.-L. and D.T.A.). R.E.G.-L. is the recipient of aCareer Scientist Award from the Ontario Ministry of Health,Health Research Personnel Development Program, andD.T.A. is a Career Investigator of the MRC. We thank Dr F.Labrie (Laval University, Quebec City, Canada) for cAMPantiserum, Dr H. R. Behrman (Yale University School ofmedicine, New Haven, CT) for progesterone antiserum, DrG. D. Niswender (Colorado State University, Fort Collins,CO) for pregnenolone antiserum, and Dr R. F. Seamark(Queen Elizabeth Hospital, Adelaide, Australia) for 20

granulosa cell steroidogenesis. Endocrinology 120,819-828.

CHERESH, D. A., PIERSCHBACHER, M. D., HERZIG, M. A.

& MuJOO, K. (1986). Disialogangliosides GD2 and GD3are involved in the attachment of human melanoma andneuroblastoma cells to extracellular matrix proteins.J. Cell Biol. 102, 688-696.

DANIEL, S. A. J. & ARMSTRONG, D. T. (1984). Site of

action of androgens on follicle-stimulating hormone-induced aromatase activity in cultured rat granulosacells. Endocrinology 114, 1975-1982.

FISHMAN, P. H. & BRADY, R. 0 . (1976). Biosynthesis and

function of gangliosides. Science 194, 906—915.GOLDENBERG, R. L., VArTUKAITIS, J. L. & ROSS, G. T.

(1972). Estrogen and FSH interactions on follicle growthin rats. Endocrinology 90, 1492-1498.

GOLDENRING, J. R., OTIS, L. C , YU, R. K. &DELORENZO, R. J. (1985). Calcium/ganglioside-dependent protein kinase activity in rat brain membrane.J.Neurochem. 44, 1229-1234.

GRIFFITHS, S. L., PERKINS, R. M., STREULJ, C. H. &

CRTTCHLEY, D. R. (1986). Variants of BALB/c 3T3 cellslacking complex gangliosides retain a fibronectin matrixand spread normally on fibronectin-coated substrates.J. Cell Biol. 102, 469-476.

GRINNELL, F. (1978). Cellular adhesiveness andextracellular substrata. Int. Rev. Cytol. 53, 65-144.

GRINNELL, F. & FELD, M. K. (1979). Initial adhesion ofhuman fibroblasts in serum-free medium: possible role ofsecreted fibronectin. Cell 17, 117-129.

HASEGAWA, T., HASEGAWA, E., CHEN, W.-T. & YAMADA,

K. M. (1985). Characterization of a membrane-associated glycoprotein complex implicated in celladhesion to fibronectin. J. Cell Biochem. 28, 307-318.

HYNES, R. O. & YAMADA, K. M. (1982). Fibronectins:

multifunctional molecular glycoproteins. J. Cell Biol. 95,369-377.

INABA, T., WlEST, W. G. & NlSWENDER, D. (1979).Radioimmunoassay of serum and ovarian pregnenolonein immature rats. Steroids 34, 663-675.

KLEINMAN, H. K.., MARTIN, G. R. & FISHMAN, P. H.

(1979). Ganglioside inhibition of fibronectin-mediatedcell adhesion to collagen. Proc. natn. Acad. Sci. U.SA.76, 3367-3371.

KNUDSEN, K., HORWITZ, A. & BUCK, C. (1985). A

monoclonal antibody identifies a glycoprotein complexinvolved in cell substratum adhesion. Expl Cell Res. 157,218-226.

LEUNG, P. C. K. & ARMSTRONG, D. T. (1979). Estrogen

treatment of immature rats inhibits ovarian androgenproduction in vitro. Endocrinology 104, 1411-1417.

LISCHINSKY, A. & ARMSTRONG, D. T. (1983). Granulosa

cell stimulation of thecal androgen synthesis. Can. J.Physiol. Phannac. 61, 472-477.

LOWRY, O. H., ROSEBROUGH, N. J., FARR, A. L .&

RANDALL, R. J. (1951). Protein measurement with theFolin phenol reagent. J . biol. Chan. 193, 265-275.

MORLEY, P., ARMSTRONG, D. T. & GORE-LANGTON, R. E.

(1987). Fibronectin stimulates growth but not follicle-stimulating hormone-dependent differentiation of ratgranulosa cells in vitro. J. cell. Physiol. (in press).

MORLEY, P. & GORE-LANGTON, R. E. (1985). Fibronectin

mediates increases in protein and DNA content withoutinfluencing early differentiation of rat granulosa cells inculture. Biol. Reprod. 32 (1), Abstract 324, 199.

OKADA, Y., MATSUURA, H. & HAKAMORI, S. (1985).

Inhibition of tumor cell growth by aggregation of atumor-associated glycolipid antigen: a close functionalassociation between gangliotriaosyl ceramide andtransferrin receptor in mouse lymphoma L-5178Y.Cancer Res. 45, 2793-2801.

PERKINS, R. M., KELLIE, S., PATEL, B. & CRJTCHLEY, D.

R. (1982a). Gangliosides as receptors for fibronectin?Comparison of cell spreading on a ganglioside specificligand with that on fibronectin. Expl Cell Res. 141,231-243.

PERKINS, R. M., KELLIE, S., PATEL, B. & CRITCHLEY, D.

R. (19826). The cell surface receptor for fibronectin: therole of gangliosides in membranes and tumor growth. InDevelopments in Cancer Research, vol. 7 (ed. T.Galeotti, A. Cittandini, G. Neri & S. Papa), pp. 29-38.Amsterdam: Elsevier/North-Holland Biomedical Press.

PYTELA, R., PIERSCHBACHER, M. D. & RUOSLAHTI, E.

(1985). Identification and isolation of a 140 Kd cellsurface glycoprotein with properties expected of afibronectin receptor. Cell 40, 191-198.

RAUVALA, H., CARTER, W. G. & HAKOMORI, S. (1981).

Studies on cell adhesion and recognition. I. Extent andspecificity of cell adhesion triggered by carbohydrate-reactive proteins (glycosidases and lectins) and byfibronectin. J . Cell Biol. 88, 127-137.

REDDOCH, R. B. & ARMSTRONG, D. T. (1984). Interactions

of a phosphodiesterase inhibitor, 3-isobutyl-l-methylxanthine, with prostaglandin E2, follicle-stimulatinghormone, luteinizing hormone, and dibutyryl cyclic3',5'-adenosine monophosphate (cAMP) and steroidproduction by neonatal rat ovaries in vitro.Endocrinology 115, 11-18.

RENNARD, S. I., WIND, M. L., HEWITT, A. T. &

KLEINMAN, H. K. (1981). Effect of collagen and cellshape on binding of fibronectin to cells. Archs Biochem.Biophys. 206, 205-212.

SKINNER, M. K. & DORRINGTON, J. H. (1984). Control of

fibronectin synthesis by rat granulosa cells in culture.Endocrinology 115, 2029-2031.

SKINNER, M. K., MCKERACHER, H. L. & DORRINGTON, J.

H. (1985). Fibronectin as a marker of granulosa cellcytodifferentiation. Endocrinology 117, 886—892.

SPEIGEL, S., SCHLESSINGER, J. & FISHMAN, P. H. (1984).

Incorporation of fluorescent gangliosides into humanfibroblasts: mobility, fate and interaction withfibronectin. J. Cell Biol. 99, 699-705.

STEEL, R. E. D. & TORRIE, J. H. (1980). Principles and

Procedures of Statistics. New York: McGraw-Hill.

216 P. Morley et al.

SVENNERHOLM, L. (1957). Quantitative estimation of sialic interaction of fibronectin with ganglioside-deficient cells.acids. II. A colorimetric resorcinol-hydrochlonc acid Expl Cell Res. 143, 295-302.method. Bwchim. biophys. Ada 24, 604-611. YAMADA, K. M., KENNEDY, D. W., GROTENDORST, G. R.

YAMADA, K. M., CRITCHLEY, D. R., FISHMAN, P. H. & & MOMOI, T. (1981). Glycolipids: receptors forMoss, J. (1983). Exogenous gangliosides enhance the fibronectin} J. cell. Physiol. 109, 343-351.

(Received 5 February 1987 - Accepted 26 May 19S7)

Ganglioside effects on granulosa cells 217