T H E J O U R N A L OF

BIOLOGICAL CHEMISTRY

byT h e American Society for Biochemistry and Molecular Biology, Inc
Vol. 263, No. 14, Issue of May 15, pp. 6847-6853,1988 Printed in U.S.A.
Sialidase Activitiesof Cultured Human Fibroblasts and theMetabolism of G M 3 Ganglioside*
(Received for publication, September 15, 1987)
Seigou Usuki, Shu-Chen Lyu, and Charles C. SweeleyS
From the Department of Biochemistry, Michigan State University, East Laming, Michigan 48824
with NaB3H, according to the method of Schwarzmann (12). [3H] Sialyllactitol was prepared by the method of Bhavanandan et al. (13). from the National Institutes of Health. The costs of publication of The following materials were purchased from commercial sources: N this article were defrayed in part by the payment of page charges. acetyl-a-neuraminic acid from Koch-Light Laboratories (Colnbrook, This article must therefore be hereby marked aduertisement in U.K.), Sep-Pak Cle from Waters Associates (Milford, MA), [3,3-3H,] serine (25 Ci/mmol) and [l-4C]N-acetylmannosamine mCi/ (18 accordance with 18 U.S.C. Section 1734 solelyto indicate this fact. mmol) from DuPont-New England Nuclear (Boston, MA), precoated $ To whom correspondence should be addressed. The abbreviations used are: GSLs, glycosphingolipids;GLC, gas- high performance thin-layer plates (Silica Gel 60) from E. Merck Co. liquid chromatography; GM, NeuAca2-3Galpl-4Glc-Cer;GD3, (Darmstadt, F.R.G.), DEAE-Sephadex A-25from Pharmacia Fine NeuAccu2-8NeuAccu2-3Gal(3l-4Glc-Cer; LacCer, Gal(3l-4Glc-Cer;Gbs, Chemicals (Uppsala, Sweden), Unisil from Clarkson Chemical Co. Galal-4Galj31-4Glc-Cer; Gb,, GalNAc@1-3Galal-4Galpl-4Glc-Cer;(Williamport, PA), Iatrobeads (6RS 8060) from Iatron Co. (Tokyo, sialyllactitol, N-acetyl-a-neuraminosyl-(cu2-3)-lactitol; EGF, epider- Japan), sialyl-(cu2-3)-lactoseand sialidase (Vibrio cholerae) from Sigma. mal growth factor. Cell Culture-Human foreskin fibroblasts (FS-4) were obtained * The glycosphingolipidnomenclature is used according to IUPACIUB recommendations (Eur. J. Biochem. (1977) 79, 11-21) and from The Massachusetts Institute of Technology Cell Culture Center, and were cultured in Dulbeccos minimum essential medium suppleSvennerholms system (J. Neurochem. (1963) 10,612-623).
reduced growth factorFree sialic acid has been found in the cell-conditioned 3T3, KB, and A431 cells in culture and medium of human foreskin fibroblasts. It is proposed stimulated tyrosine kinase activity of growth factor receptors that the accumulation of extracellular sialic acid may (3,4). The inhibitory effect of G Mon EGF receptor phospho~ ~ result from the hydrolysis of G M ganglioside on the rylation was also demonstrated with isolated membranes and cell surface of these fibroblasts. Sialidase activities a partially purified receptor preparation, andit was postulated with GM3ganglioside and sidyllactitol as substrates that GM3 might bind to EGF receptor in a manner that inhibits were demonstrated in cell-conditioned medium, and receptor-receptor interactions (4). More recently, it was rethe levelsof their activities correlated positively with ported that IYSO-GM~ inhibits tyrosine phosphorylation also cell density. The GMa sialidase activity at pH 4.5 was of EGF receptor, while an analog of GM3, containing neura4.1 and 38 pmol/h/ml of medium at sparse and conminic acid rather than N-acetylneuraminic acid, strongly fluent densities, respectively; the corresponding activas 1 ities with sialyllactitol the substrate were2 and 75 stimulates EGF receptor phosphorylation in A431 cells ( 5 ). If endogenous G M ~ , a major ganglioside many mammalian of pmol/h/ml of medium (pH 4.5). The pH uersus activity profiles with GM3as the substrate suggested the pres- cells, has the same inhibitory effect as was observed with cells ence of a second sialidase with an optimal activity at exogenous G M ~ , would be unable to progress through the pH 6.6 in the conditioned medium of preconfluent cells. cell cycleafter growth factor stimulation.We have speculated This activity was virtually absent in the medium of that GM3on the plasma membrane may therefore undergo contact-inhibited cells and could not be assayed with growth-regulated turnover or, alternatively, may be physically sialyllactitol as the substrate. separated from the receptor to prevent the inhibition. Our The turnover of cell surface GM3 was assessed by initial efforts have been directed toward the possibility that pulse labeling human foreskin fibroblasts with a radio- on the outer surface of cultured human fibroblasts may GM3 active precursorof sialic acid ([ l-4C]N-acetylmanno- be metabolized to lactosylceramide and free sialic acid by an samine) and radioactive precursorofceramide ([3,3- extracellular sialidase in the plasma membrane or in the a Hz]serine). During a chase period of 24 h turnover of culture medium. Plasma membrane-bound sialidase activity the doubly labeled cellular GM3 was observed; there toward gangliosides (6) has been reported to be increased in was a loss of about 35% of the C-labeled sialic acid several transformed cell lines (7) and exogenously added without any measureable loss of 3H-labeled ceramide from GMs. We have speculated that the enzyme-cata- bacterial sialidases have been shown to stimulate the growth lyzed removal of sialic acid from the GM3ganglioside of contact-inhibited chick embryo cells (8). In preliminary studies, we have shown that the sialic acid residue of G 3 is M on the extracellular aspect of the plasma membrane may be a necessary event involved in the modulation metabolically active in growing fibroblasts and that there are several different sialidases that might account for the turnover of cell growth. In of GM3 these cells (9,101. thispaper we present evidence in for the correlation of cell density with G 3 metabolism and M two sialidase activities in the conditioned medium of human Gangliosides are sialic acid-containing glycosphingolipids fibroblasts. (GSLs)] which are generally believed to be localized predomEXPERIMENTALPROCEDURES inantly on the outer leaflet of the plasma membrane of mamMaterials-G,, containing N-acetylneuraminic acid was isolated malian cells (1,2). A potentially specific cell surface functional role of gangliosides is suggested by the findings that exoge- from human liver according to themethod of Seyfried et al. (11).[3H] nously administered GM~ and G 3 inhibited the growth of GMs was prepared by catalytic reduction of the ceramide double bond M

* This work was supported in part by Research Grant DK 12434
Fibroblast Sialidase Activities and G M 3 Metabolism
at 37C for 4h with 2 units of V. cholerae sialidase. After the incubation, the reaction mixture was further diluted with 2 ml of distilled water and passed through a Sep-Pak C,, cartridge followed by 5 ml of chloroform/methanol(2:1,v/v) elution. The aqueous eluate was concentrated and examined by TLC with n-propyl alcohol/water/ NH3 (70101, v/v/v) as the developing solvent, and resorcinol-HC1 reagent was used for detection of free sialic acid. The chloroform/ methanol eluate was further passed through a column (0.5-cm inner diameter X 5 cm) of DEAE-Sephadex A-25 following 4 volumes of chloroform/methanol (2:1, v/v). The eluate was concentrated under nitrogen and subjected to methanolysis with 3% methanolic HCI at 110C for 3 h. The distribution ofI4C and 3H radioactivity in the sphingosine, fatty acid, and carbohydrates was determined as described previously (15). Determination of Radiolabeled Sialic Acid in the Medium-The conditioned medium obtained after incubation with radiolabeled cells for different time periods was saved as described above. The radioactivity in the medium wasanalyzed by descending paper chromatography on Whatman 3MM paper (48 cm, 24 h) using the following solvent system: n-propyl alcohol/acetic acid/water ( 2 , v/v/v). For quantitation, paper chromatograms were cut into 1-cm strips and radioactivity was determined by liquid scintillation spectrometry. Free Sialit Acid in the Medium-FS-4 cells were cultured for 24 h from preconfluency (5 X io3 cells/cm2) to confluency (1.1 X io' cells/ cm') in a 100-mm dish with 10 mlof culture medium, and thesamples of conditioned medium were passed through Sep-Pak CI8cartridges, followed by an aqueous wash of 10 ml of distilled water. The eluate was concentrated and applied directly onto a column (1.2-cm inner diameter X 10 cm) of DEAE-Sephadex A-25 to separate free sialic acid. Elution was performed successively with 1 volume of 2 m M pyridinium acetate buffer (pH 5.4), 2 volumes of 100 m pyridinium M acetate buffer (pH 5.4), and 2 volumes of 500 m pyridinium acetate M buffer (pH 5.4). Free sialic acid was eluted with 100 m buffer and M was determined by thiobarbituric acid assay (16) and GLC (17). Assay of Sialidase Actiuity-Cells were seeded a t the following cell densities: 1 X lo3, 6 X lo3, and 6 X lo4 cells/cm2 in 100-mm dishes with 10 ml of medium for sparse, preconfluent, and confluent cultures, respectively. After 24 h incubation at 37 "C, the medium was centrifuged a t 3000 rpm for 10 min, the supernatant fraction was saved, and the cells were harvested as described above. Plasma membrane fractions were prepared by a previously described procedure (18). Ganglioside sialidase was assayed using [%]GM~ (13.7 nmol, specific radioactivity 818 cpm/pmol). Ten microliters of Triton CF-54 ( l % , v/v), 40 pl of 0.1 M sodium acetate/acetic acid buffer (pH 4.5),and enzyme source (50 pl for medium, 100 pg of protein for plasma membrane) were added to the tube. The incubation was carried out for 1 h at 37 "C with gentle shaking and was stopped by placing the sample tubes in an ice bath. Distilled water (5 ml) was then added, and the reaction mixture was applied to Sep-Pak CIScolumn, followed by an aqueous wash of 10 ml of distilled water. After eluting with 10 ml of methanol, the eluate was applied to a column (0.5-cm inner diameter X 5 cm) of DEAE-Sephadex A-25 as described previously (19) to separate released sialic acid and remaining [3H]G~3. Nonconditioned medium (for sialidase in the medium) and boiled plasma membrane (for sialidase in the plasma membrane) were used as controls. Sialyllactitol sialidase activity was assayed using [3H]sialyllactitol (1115 cpm/pmol) by the same procedure as the G Msialidase ~ assay, except that the incubations were without detergent and there was no Sep-Pak C,, column chromatography. Other Enzyme Assays and Protein Assay--5'-Nucleotidase (20) and acid phosphatase (21) were assayed using adenosine 5"mOnOphosphate (AMP) and p-nitrophenyl phosphate, respectively, as described previously. Protein was determined according to Lowry et al. (22) using bovine serum albumin as thestandard.

mented with 10% fetal calf serum in 5% carbon dioxide/air incubator. The cells were routinely subcultured by washing monolayers with a solution of 0.05% trypsin and 0.025% EDTA. After removal of the rinse, dishes were placed in an incubator a t 37 "C for 10 min. Trypsinization was stopped by the addition of medium containing serum, and cell viability was evaluated by trypan blue exclusion. The cells were easily contact-inhibited, and cultures of different cell population densities were obtained by seeding, respectively, 1X 103, 6 X I@, and 6 X lo' cells/cm2 on the bottom of a 100-mm dish containing 10 ml of medium for sparse, preconfluent, and confluent cultures, respectively. Cultures showing more than 90% viability and their conditioned medium were used for enzyme assays. Quuntitatiue Isolation of GSLs-FS-4 cells were grown in monolayers to preconfluency, and about 3 X 10' cells were then harvested in 0.025% EDTA solution in phosphate-buffered saline. After washing three times with phosphate-buffered saline, cell pellets were prepared by centrifugation (2500 rpm, 10 min) and were freeze-dried. Total lipids were extracted from the residue with 5 ml of chloroform/ methanol (2:1, 1:1, 1:2, v/v) at 45 "C.The combined crude total lipid extract was passed through a column (0.5-cm inner diameter X 5 cm) of DEAE-Sephadex A-25 and eluted successively with 5 volumes of chloroform/methanol(21, v/v) (Fraction A) and 5 volumes of 0.3 M sodium acetate in methanol (Fraction B). Fraction A containedtotal neutral GSLs; it was evaporated, and the residue in chloroform was applied to a column (0.5-cm inner diameter X 4 cm) of Unisil (activated silicic acid) and eluted successively with 5 volumes of chloroform and 3 volumes of acetone/methanol (9:1, v/v). The acetone/methanol fraction containing neutral GSLs was dried under nitrogen. Fraction B, which contained total acidic GSLs, was added to the neutral GSLs. This recombined fraction of total GSLs was subjected to mild alkali-catalyzed methanolysis in methanolic 0.5 M NaOH overnight at ambient temperature. The solution was neutralized with 1 M acetic acid, dialyzed against distilled water, and then freeze-dried. The mixture of gangliosides and neutral GSLs thus obtained was examined by TLC with chloroform/methanol/0.5% CaCI, (55:45:10, v/v/v), and thespots were visualized by 0.2% orcinol, 2 N H,S04 reagent or resorcinol-HC1 reagent. Purification and Carbohydrate Composition of GSLs-The total GSLs were applied to TLCplates as a 10-cm band, and theplate was developed with chloroform/methanol/water(60:35:8, v/v/v). One-cm zones a t each edge of the plate were cut off and visualized with orcinol reagent. Bands corresponding to each orcinol-positive spot were scraped from the remaining part of the plate and extracted with chloroform/methanol/water (10:101, v/v/v). The extracts were evaporated undernitrogen and concentrates in chloroform passed through a column (0.5-cm inner diameter x 2 cm) of Iatrobeads, eluting with 3 volumes of chloroform/methanol/water (603523, v/v/v). The eluate was dried under nitrogen. The molar ratios of carbohydrates in TLC-purified GSLs were determined by gas-liquid chromatography (GLC) of trimethylsilylated methyl glycosides (14). Labeling Celk and Qwntitation of GSLs-When cells reached a preconfluent density (6 X lo3cells/cm3), 20 pCi of [3,3-3Hz]serine(25 Ci/mmol) and 400pCi of [l-'4C]N-acetylmannosamine(18 mCi/ mmol) wereadded to 10 ml of culture medium. The cells were allowed to become confluent by incubation for an additional 24 h. The cells were then harvested by trypsinization and were seeded at a preconfluent cell density (7 X lo3cells/cm2)in 10 ml of fresh culture medium. The incubation time was varied from 0 to 24 h. After appropriate incubation periods the medium was removed and saved, and thecells were washed three times with PBS. Cell pellets were prepared by centrifugation (2500 rpm, 10 min) and were freeze-dried. The radioactive total GSLs were prepared as described above and were examined by TLC in chloroform/methanol/0.5% CaCI, (50:45:10, v/v/v) as the developing solvent. Autoradiograms of the plates were obtained on Cronex x-ray film after spraying with an enhancer. The radioactive spots corresponding to G Mand LacCer were directly scraped from ~ the TLC plates, andthe radioactivity of each scraped spot was determined by liquid scintillation spectrometry (Packard TRI-CARB 460C liquid scintillation spectrometer), which was programmed for dual label analysis of 3H and I4C. Distribution of 14Cand 3H Radioactivity in the Labeled GSLs-The labeled GSLs obtained as described above were applied to a column (1.2-cm inner diameter X 20 cm) of Iatrobeads and eluted with a linear gradient of 60 ml of chloroform/methanol/water(75:23:2 and 50:47:3, v/v/v); fractions corresponding to labeled G S and LacCer M were combined into two fractions and dried under nitrogen. The labeled G 3 was dissolved in 0.1 ml of distilled water and incubated M

RESULTS

Examination of GSLs of Human Fibroblusts-TLC of the GSLs of human fibroblasts is shown in Fig. 1. By gas chromatographic analysis, the molar ratios of g1ucose:galactose: sialic acid were 1.00:0.92:0.89 (GI-1) and 1.00:0.900.88 (G12) for GM3 ganglioside, and the molar ratios of glucose:galactose were 1.00:0.92 (G2-1) and 1.00:0.95 (G2-2) for LacCer. As reported by Dawson et al. (23), when the total GSLs were examined by co-chromatography with their authentic standards, human fibroblasts were found t o contain
Fibroblast Sialidase Activities and GM3 Metabolism

CMH 1 , LacCer \

GD3 GDl a
FIG. 1 Autoradiographic examination of radioactive GSLs incorporated from [Hlserine and [C]. N-acetylmannosamine. The labeled GSLs were obtained from double isotope labeled human fibroblasts as described under Experimental Procedures. TLC was developed with chloroform/methanol/0.5%CaC12 (55:45:10, v/v/v). Standard GSLs ( l a n e I ) and total GSLs of the cells ( l a n e 2) were visualized by spraying with orcinol reagent and heating a t 110C for 10 min. A portion ( l a n e 3-8) of the TLCplate shows autoradiograms of GSLs of the double isotope-labeled fibroblasts after different incubationtimes following removal of labeled precursors ( l a n e 3 , O h; lane 4,2h; lane 5,4 h; lane 6,6h; lane 7,12 h; lane 8,24 h). CMH, Glc-Cer and Gal-Cer.
one major ganglioside,G M(Gl-1 and ~ G1-2), and four neutral of incorporated 3H radioactivity in the ceramide residues of GSLs, Glc-Cer and Gal-Cer, LacCer (G2-1and G2-2), globo- each GSL. The percentage distribution of 3H radioactivity was as follows: Glc-Cer and Gal-Cer (2.5/0.5%), LacCer (8.5/ triaosylceramide (Gb3),and globotetraosylceramide (Gb,). Gb, (60/60%), and Intermolecular Distribution of 14C and 3H in the Labeled GM3 8.0%),Gb3 (19.2/20.1%), (9.8/11.4%), GM3 (trace) for chase periodsof 0/24 h. The only GSLto show and LacCer-As describedpreviously (9), the labeled G M ~GD3 reached asteady state level of radioactivity after 24 h of pulse a substantial loss of 3H from the ceramide moiety was Glclabeling with N-acetylmannosamine or serine. GM3 and Cer and Gal-Cer. Fig. 1 shows a representative TLC autoraLacCerwere purified fromdoubleisotope-labeledcells as diogram of total GSLs from the cells. Several labeled bands, were describedunder Experimental Procedures. When the double corresponding to LacCer, Gb3, Gb,, and G M ~ , detected. are isotope-labeled G Mwas treated with sialidase (V. cholerae), The counts recovered from scraped bands of GM3 shown ~ in Table 11. As cell density increased, there was a 34.6% 87% of the G Mwas degraded to free sialic acid and LacCer. ~ The distribution of radioactivity in the fatty acid, sphingoid TABLE I1 base, and sugar moietiesis summarized in Table I. We found Time course of 14C/3H ratio of double isotope-labeled , in G 98.9% of the 14Cand 97.5% of the 3H label in the sialic acid human fibroblasts and sphingoid base portions of G M ~ , respectively. Purified Results are averages for three dishes of cells. The times are subLacCer fromthe labeled cellscontained no 14C,and 98.7% of sequent to removal of labeled precursors from the medium. the 3H label was in the sphingoid base moiety (Table I). C/H Time 3H C ratio Time Course of the change of 14C/H Ratio of Celhhr GM3 h wm and Free Sialic Acid Accumulation in Medium-At different 0 13,601 f 44190,103 f 5140.151 f 0.005 times of incubation after pulse labeling for 24 h with [14C]N2 13,352 f 310 90,815 f 7150.147 f 0.006 acetylmannosamine and [3H]serine,the radioactive metabof 8200.134 f 0.12,012 f 11589,721 lites were analyzed as described under Experimental Proce6 11,90,578 4780.124 f 0.001 f 49091,041 f 811 0.113 0.004 1210,264 dures. During a chase period of 24 h, there was little change

248.620

f 61390,682

4420.095

f 0.004
TABLE I Intermolecular distribution of 14Cand 3H radioactivities in double isotope-lubeledG , and LacCer

GM~ (Gl-1 and GI-2)

C LacCer (G2-1 and G2-2) C
Glucose 0.21 0.16 0.11 Galactose 0.30 0.15 0.16 Sialic acid 98.9 00.8 0.01 0.45 97.5 98.7 Long-chain base Fatty acid 0.14 2.11 1.02 Bands G1-1and G1-2and bands G2-1and G2-2corresponding to GM3 and LacCer (Fig. 1) were purified from the GSLs labeled with ["GIN-acetylmannosamine and [3H]serine. 14Cand 3H radioactivities were determined by a liquid scintillation spectrometer, which was programmed with dual label regions for C and 3H.

TIME (HOURS)

FIG. 2 Time course of free [C]sialic acid accumulation in. the medium from cultures of fibroblasts. [14C]Sialic acid released into the medium from the labeled cells was measured a t different times asdescribed under Experimental Procedures. 14Cradioactivities were determined by liquid scintillation spectrometry,which was programmed with dual label regions for 3H and 14C.
decrease in the recovered 14C (sialic acid residue) in GM3 with little or no change in the 3H label. The same result was observed where the isotope ratio of [%]sialic acid to [3H] ceramide (4C/3H) in the GM3 was calculated (Table II); this ratio changed as cell density increased toward confluency. We also observed a significant accumulation of %-labeled free sialic acid in the medium as cell density increased (Fig. 2). The total decrease of [%]sialic acid from the double isotopelabeled GM3 was approximately the same as the amount of accumulated %-labeled free sialic acid in the medium. Free Sialic Acid Identification and Quuntitation in the Cellconditioned Medium-A typical gas-liquid chromatogram was obtained with the trimethylsilyl derivative of standard free sialic acid, as reported by Yu et al. (17) (data not shown). Although GLC chromatograms of samples from the cellconditioned medium showed complicated patterns, with large amounts of several unrelated components, a peak corresponding to the retention time of standard free sialic acid was

SE WP o3

05 = a20-
c 5 F -3ovi <E 3 ,-* ur

I TIIIAII I

345678
FIG. 3. Mass spectrum of the free sialic acid isolated from the conditioned medium of fibroblast cultures. The human foreskin fibroblasts were cultured for 24 h from preconfluency (5 x lo3 cells/cm) to confluency (1.1 X lo4 cells/cm) in a loo-mm diameter dish containing 10 ml of culture medium, and the conditioned medium was saved. The medium was treated as described under Experimental Procedures, and the sialic acid was analyzed as the methyl ester methyl ketal trimethylsilyl derivative with an LKB-2091 gas chromatograph-mass spectrometer using a glass column (2 mm x 3.5 m) packed with 5% OV-17 on 100/120 mesh Supelcoport. The calculated molecular weight of the trimethylsilylated methyl ester methyl ketal is 625; ions at m/z 610 (M - 15)+, 566 (M - 59)+, 420 (M - 205)+, 298 (M - 32 - 90 - 205)+, and m/z 259 (M - 102 - 205 - 59)+ are definitive for this structure. TABLE III

FIG. 4. Effect of pH on GMs sialidase activity in the conditioned medium and in the plasma membrane. The cells were incubated at 37 C for 24 h at the following cell densities; sparse (1 x lo3 cells/cm*), preconfluent (6 X IO3 cells/cm), and confluent (6 X lo4 cells/cm*). The number of cells reached 2.1 x 1Oa cells/cm*, 1.4 x lo4 cells/cm*, and 6 X lo4 cells/cm*, and the cell-conditioned medium and the plasma membrane fractions were prepared and assayed for sialidase activity as described under Experimental Procedures. Further treatment was as described under Experimental Procedures, except that the assay buffers were as follows: 0.1 M sodium acetate/ acetic acid buffer (pH 3.0-5.0), 0.05 M maleic acid buffer (pH 5.57.0), and 0.05 M Tris/HCl buffer (pH 7.5-8.0). a, G,, sialidase activities in the medium were expressed as picomoles of sialic acid released in 1 h at 37 C/ml of cell-conditioned medium obtained after 24 h of culture at different cell densities: sparse culture (O), preconfluent culture (0), confluent culture (A). b, GM3 sialidase activities in the plasma membrane were expressed as picomoles of sialic acid released per total membrane derived from 1 X lo6 confluent cells. Specific activities in the plasma membrane from confluent cultures were about 4-fold higher than those of total cell homogenates. 5Nucleotidase and acid phosphatase activities in the prepared plasma membranes of confluent cultures were 781.4 and 0.8 nmol. h-l. cell-. lo, and corresponded to 31- and 0.14-fold higher than those of total cell homogenates, respectively.
of the cell-conditioned medium of human foreskin fibroblasts Results are averages of more than four dishes of cells/experiment. The human foreskin fibroblasts were cultured for 24 h from preconfluency (5 X lo3 cells/cm) to confluency (1.1 X lo4 cells/cm*) in lOOmm diameter culture dishes containing 10 ml of culture medium, and the samples of conditioned medium were saved. These samples were treated as described under Experimental Procedures. The free sialic acid content was quantitated by thiobarbituric acid assay and GLC. Free sialic acid Thiobarbituric Cell-conditioned Control medium a Free sialic metric method *Free sialic medium 1.01 +- 0.011 0.59 * 0.021 determined by gas-liquid by thiobarbituric chromatography. acid

nmd/ml

sialic

acid content

GLC? 0.74 + 0.026 0.25 f 0.013 acid-colori-
acid content (16). acid content
detected. This peak was identified as the trimethylsilyl derivative of the methyl ,B-ketoside methyl ester of N-acetylneuraminic acid by gas chromatography-mass spectrometry analysis (Fig. 3). The amounts of free sialic acid in the medium were determined by two different methods (GLC and thiobarbituric acid assay) for comparison (Table III). The fact that thiobarbituric acid assay values were higher than GLC values can probably be attributed to other chromogens in the medium that interfered with the calorimetric assay. However, the results (difference between cell-conditioned and control medium values) were similar using the two methods: 0.42 and 0.49 nmol/ml medium for thiobarbituric acid assay and GLC, respectively. Sialidase Activity in the Cell-conditioned Medium and the Plasma Membrane-Sialidase activities in the cell-conditioned medium and the plasma membrane were examined from pH 3.0 to 8.0, with [H]GM3 and [3H]sialyllactitol as substrates (Figs. 4 and 5). Plasma membrane sialidase activities for both substrates exhibited a sharp pH curve with the optimum around pH 4.5; the sialidase activity with sialyllactitol was about 2-fold higher than the GM3 sialidase activity at the optimal pH. The GM13 sialidase and the sialyllactitol sialidase activities of cell-conditioned medium were determined with sparse, preconfluent, and confluent cultures, and the results are shown in Figs. 4a and 5a, respectively. The sialidase activities measured with GW3 and sialyllactitol at pH
Fibroblast Sialidase Activities and M 3 Metabolism G
FIG.5. Effect of pH on sialyllactitol sialidaseactivity inthe conditioned medium and in the plasma membrane. The cells were incubated at 37 C for 24 h at the following cell densities; sparse (1 X lo3 cells/cm2), preconfluent (6 X lo3 cells/cm2), and confluent (6 X lo cells/cm2). The number of cells reached 2.1 X lo3, 1.4 X lo4, and 6 X 10 cells/cmz, and the cell-conditioned medium and the plasma membrane fractions were prepared and assayed using [3H] ~ sialyllactitol by the same procedure as G M sialidase assay, except that no detergent was used and the Sep-Pak CIScolumn procedure was omitted. Further treatments were as described under Experimental Proceduresexcept that theassay buffers were as follows: 0.1 M sodium acetate/acetic acid buffer (pH 3.0-5.0), 0.05 M maleic acid buffer (pH 5.5-7.0), and 0.05 M Tris/HCl buffer (pH 7.5-8.0). a, sialyllactitol sialidase activities in the mediumwere expressed as picomoles of sialic acid released in 1h a t 37 C/ml of cell-conditioned medium obtained after 24 h of culture at different cell densities of cells: sparse culture (O), preconfluent culture (O),confluent culture (A). sialyllactitol sialidase activities in the plasma membrane were b, expressed as picomoles of sialic acid released per total membrane derived from 1 X 10 confluent cells.

TRITON CF-54 (%)

FIG. 7. Effect of Triton CF-54 concentration on GMa sialidase activity in cell-conditioned medium. The cells were cultured for 24 h from preconfluency (5 X lo3 cells/cm2) to confluency (1.1 X lo4cells/cm*) in 100-mm diameter culture dishes containing 10 ml of culture medium. GM3sialidase activities in the cell-conditioned medium were determined at different Triton CF-54 concentrations using 0.1 M sodium acetate/acetic acid buffer at pH 4.5 ( 0 )and 0.05 M Tris/HCl buffer at pH 6.5 (0).
pearing when the cells became contact-inhibited (Fig. 4a). However, there was no difference in either the GM3 sialidase activity or the sialyllactitol sialidase activity of plasma membrane between preconfluent and confluent cultures. The GM3sialidase activities of cell-conditioned medium were then studied in more detail as afunction of cell density. As shown in Fig. 6, sialidase activity at pH 4.5 increased with cell density and reached a plateau level at confluency. In contrast, the sialidase activity at pH6.5 reached a maximum level in sparse cultures and decreased to very low levels at confluency. The optimal concentration of Triton CF-54 detergent required for GM3metabolism by the sialidases in cell-conditioned medium was pH dependent. As shown in Fig. 7, detergent was not required when the assays were carried out at pH 4.5, but at pH detergent was required and optimal activity 6.5 was observed at 1%Triton CF-54.

DISCUSSION

Cells/cm2

15 10-3)

FIG. 6. Effect of cell density on GMs sialidase activity inthe cell-conditioned medium. The cells were cultured for 48 h from preconfluent cell density (5 X lo3 cell/cm2) to confluency (1.7 X lo4 cells/cm2) in a 100-mm diameter dish containing 10 ml of culture medium; the cell-conditioned medium was saved at appropriate incubation times. G M 3 sialidase activities in the cell-conditioned medium were determined as described under ExperimentalProcedures using 0.1 M sodium acetate/acetic acid buffer at pH4.5 (0)and 0.05 M Tris/HCI buffer at pH6.5 (0).
4.5 were about 9- and 6-fold higher, respectively, in the confluent culture of medium than those of sparse cultures. The sialidase activities with both substrates exhibited a similar sharp optimum near pH 4.5 with conditioned medium from confluent cultures. Another form of sialidase was appar~ ent when G M was the substrate; this activity was optimal near pH 6.5 and was highly dependent on cell density, disap-
This work was precipitated by the observation that exogenously added GM3ganglioside inhibits the tyrosine kinase activity of epidermal growth factor receptor, disabling the cascade of events leading to DNA synthesis and cell division following the stimulation of quiescent cells by epidermal growth factor (3,4). When this finding is considered together with the dramatic increase in GM3 levels in the plasma membrane after transformed cells are treated with a cellcycle blocker such as butyrate (24, 25), as well as adecrease of GM3 when cells lose their growth control mechanism after oncogenic transformation (26), it is attractive to speculate that GM3 may be intimately involved in an early step of growth control. The experiments described in this paper have been directed specifically toward the metabolism of endogenous GM3in actively growing cultures of human fibroblasts. Two interesting results of our study are the likely presence of acidic and neutral forms of sialidase activity, with GM3 as the substrate, in the cell-conditioned medium of cultured human fibroblasts and preferential turnover of the sialic acid moiety of GM3, relative to the ceramide group, in preconfluent cultures of fibroblasts. It was previously reported that sialidase activity for ganglioside metabolism is localized primarily in theplasma membrane of mammalian cells (6) andis much higher in an adenovirus-transformed line of baby hamster

turnover of the sialic acid residue only in actively growing cultures. Our data suggest a model in which GM3on the plasma membrane is metabolized to LacCer at some point in the cell cycle of growing cells, releasing the cells from inhibition of the tyrosine kinase activity of the epidermal growth factor receptor and enabling prereplicative mechanisms to occur. At confluency, when the extracellular sialidase activity at pH 6.5 is very low, the G 3 appears to be no longer metabolized and M inhibition of the mitogenic effect of epidermal growth factor wouldbe prolonged. If this were true, suppression of the sialidase activity might inhibit the growth of cultured cells. In fact, it was reported by Hakomori et al. (32)in 1980 that the addition of 125 pM N-trifluoroacetyl-GM3 (an inhibitor of membrane-bound sialidase activity) to kirsten murine sarcoma virus-transformed 3T3 cells caused modifications of cell growth behavior and morphology. This experiment is somewhat difficult to interpret, however, because exogenous G 3 M itself inhibits cell growth (3,4),and this activity cannot be distinguished from the sialidase inhibitory property of the GM3 analog. We have found that a different and more specific inhibitor of sialidase activity, 2-deoxy-2,3-dehydro-N-acetylneuraminic acid, is also an effective growth inhibitor at concentrations as low as 50 ~ L M the medium (33).It will be in interesting to determine whether this growth inhibition extends to other types of mammalian cells, including malignant cells, and whether its activity is related to a specific sialidase activity.
Acknowledgments-We are indebted to Dr. Kimihiro Kanemitsu who provided synthetic [3H]G,3, to Dr. Yosuke Shigematsu for assistance in mass spectrometry analyses, to Patricia Hoops for assistance in cell culture, and to Jill Crane for preparation of the manuscript.
REFERENCES 1. Hakomori, S., and Kanfer, J. N. (1983) in Handbook of Lipid Research, Vol. 3: Sphingolipid Biochemistry (Hanahan, D. J., ed) pp. 327-379, Plenum Publishing Cow., New York 2. Wiegandt, H. (1985) in New Comprehensive Biochemistry, Glycolipids (Neuberger, A., and Van Deenen, L. L. M., eds) Vol. 10, pp. 199-245, Elsevier Science, Amsterdam 3. Bremer, E. G., Schlessinger, J., and Hakomori, S. (1986) J. Biol. Chem. 261,2434-2440 4. Bremer, E. G., Hakomori, S., Bowen-Pope, D. F., Raines, E., and Ross, R. (1984) J. Biol. Chem. 259,6818-6825 5. Hanai, N., Nores, G., Torres-Mindez, C.-R., and Hakomori, S. (1987) Biochem. Biophys. Res. Commun. 147, 127-134 6. Zeigler, M., and Bach, G. (1981) Biochem. J. 198,505-508 7. Schengrund, C.-L., Lausch, R. N., and Rosenberg, A. (1973) J. Biol. Chem. 8 , 4424-4428 8. Vaheri, A., Ruoslahti, E., and Nordling, S. (1972) Nature New Biol. 238,211-212 9. Sweeley, C. C., and Usuki, S. (1988) in New Trends inGanglioside Research: Neurochemical and Neuroregenerative Aspects (Ledeen, R. W., ed) 307-315, Liviana Press, Padova, Italy 10. Usuki, S., and Sweeley, C. C. (1987) Zndian J. Biochem. Biophys., in press 11. Seyfried, T. N., Ando, S., and Yu, R. K. (1978) J. Lipid Res. , 538-543 12. Schwarzmann, G. (1978) Biochim. Biuphys. Acta 9 , 106-114 13. Bhavanandan, V. P., Yeh, A. K., and Carubelli, R. (1975) Anal. Biochem. 69, 385-394 14. Vance, D. E., and Sweeley, C. C. (1967) J. Lipid Res. 8 , 621-630 15. Usuki, S., and Nagai, Y. (1986) Anal. Biochem. 152, 172-177 16. Aminoff, D. (1961) Biochem. J. 81,384-392 17. Yu, R. K., and Ledeen, R. W. (1970) J. Lipid Res. 11, 506-516 18. Carey, D. J., and Hischberg, C. B. (1980) J. Biol. Chem. 5 , 4348-4354 19. Yohe, H. C., Saito, M., Ledeen, R. W., Kunishita, T., Sclafani, J. R., and Yu, R. K. (1986) J. Neurochem. 46,623-629 Jr., 20. Touster, O., Aronson, N. N., Dulaney, J. T., and Hendrickson,

kidney cells than in untransformed cells (7).We have found a similar sialidase activity in the plasma membrane of human fibroblasts using G Mand a water-soluble substrate, sialyllac~ titol, inthe assays. This sialidase has asingle optimal activity at pH 4.5 and is unaffected by cell density. The pH profiles of activity with G 3 as the substrate suggest the possibility M that there are atleast two different forms of sialidase in the cell-conditioned medium, with optimal activities at pH 4.5 and pH The activity of the acidic form, which metabolized 6.5. sialyllactitol as well as GM3, increased with cell density, but on a percell basis the total activity in dense cultures was only about 30% of the level calculated from the per cell activity in the medium of sparse cultures. We did not observe an increase in the level of this sialidase activity when fibroblasts were cultured in the presence of mannose 6-phosphate (data not shown), indicating that the plasma membrane form is not bound to cell surface mannose 6-phosphate receptors and therefore may be different from the acidic sialidase in the conditioned medium. Sialidase activity at pH 6.5 in the cell-conditioned medium had adifferent substrate specificity, was negatively correlated with cell density, and exhibited a different detergent depend4.5 ence than the pH activity. The sialidase activity could be assayed with Gh.13 but not with sialyllactitol; it increased with time at pH 6.5 in sparse cultures but then decreased as the cells approached confluency. Interestingly, rat liver contains a cytosolic sialidase with optimal activity at pH 6.5 with ganglioside substrates (27,28). We are not aware of a mechanism for the secretion of a cytosolic protein but have not ruled out the possibility that the activity observed at pH 6.5 might be derived from the cytosol of the fibroblasts. The incorporation of 3H and 14Cinto G 3 from labeled NM acetylmannosamine and serine enabled us to study the turnover of endogenously synthesized total cellular GM3. We were surprised to find that thesphingoid base lost none of its label during a 24-h chase period while about 35% of the sialic acid residues lost their label. We have interpreted this to mean that a significant proportion of the cellular GM3is actively metabolized in growing cells, by a process that involves removal of at least the sialic acid group and perhaps the galactose and glucose residues as well. In contrast to exogenously added gangliosides, which are predominantly broken down in our lysosomes (29,30), results suggest that thesphingoid base may be recycled back to G 3 by a mechanism similar to that M described for other membrane constituents. Although the pulse-chase experiments do not provide evidence for a particular mechanism, it is tempting to speculate that plasma n membrane G 3 is degraded i situ to LacCer, which is then M cycled through the Golgi apparatus where GM, is synthesized by addition of sialic acid and the newly synthesized G 3 is M returned to theplasma membrane. The release of labeled free sialic acid into themedium during the chase of doubly labeled cells supports such a mechanism but does not rule out other possibilities. For example, an intracellular compartment of labeled LacCer might be converted to G Mand transported to ~ the plasma membrane during the chase period. In human polymorphonuclear neutrophils, more than 75% of total LacCer is in an intracellular granule (31). Not all of the G 3 molecules lost their labeled sialic acid M residues during a 24-h chase of the fibroblasts. After confluency was reached, continued incubation did not further M. change the ratio of 14C to 3H in the G S One explanation of this finding is that there is more than one pool of GM3 the in fibroblasts, not all of which are actively metabolized. Alternatively, G 3 metabolism may becell density-dependent, with M