Proceedings of the National Academy of SciencesVol. 66, No. 1, pp. 57-64, May 1970

Glycosphingolipids of Plasma Membranes of Cultured Cellsand an Enveloped Virus (SV5) Grown in These Cells

Hans-Dieter Klenk and Purnell W. ChoppinTHE ROCKEFELLER UNIVERSITY, NEW YORK, NEW YORK

Communicated by Maclyn McCarty, February 16, 1970

Abstract. Glycosphingolipids of rhesus monkey kidney (MK), bovine kidney(MDBK), and two lines of hamster kidney (BHK21-F and Hak) cells havebeen compared with those of parainfluenza (SV5) virions grow in these cells.There are qualitative and quantitative differences in the neutral glycolipids andgangliosides found in the various cells. Cells with a high neutral glycolipid con-tent (MK and MDBK) contain little or no gangliosides, and those with a rela-tively high ganglioside content (BHK21-F and HaK) contain little neutralglycolipid. Glycosphingolipids are found predominantly in the plasma mem-branes. Neutral glycolipids of the host cell membrane are incorporated intothe envelope of SV5 virions, but neither gangliosides nor protein-bound neura-minic acid are found in virions. The absence of neuraminic acid from the virionmay be due to the action of viral neuraminidase.

Introduction. SV5, like all myxoviruses and paramyxoviruses, is assembledand released at the cell surface.' In this budding process, the viral envelope isderived from a segment of the plasma membrane which has been altered by theincorporation of viral proteins. Analysis of such membrane-enclosed viruses andthe events in virus assembly is not only important to an understanding of virusstructure and replication, but may also contribute to the knowledge of membranestructure and biogenesis. All the proteins of the SV5 envelope appear to bevirus-specific.2 However, in studies with four cell types, their plasma mem-branes, and virus grown in each cell, we found that, with a few exceptions,lipids of the plasma membranes are incorporated quantitatively into the viron.3 4

This report describes the glycosphingolipids of cultured cells and their presencein SV5 virons.

Glycosphingolipids are composed of sphingosine, fatty acids, and carbohy-drates. They may be neutral glycolipids or neuraminic acid-containinggangliosides, and are present in wide variety in the membranes of animal cells.Their carbohydrate moieties may serve as antigenic determinants.5 Thus, ifincorporated in the viral envelope, a cellular glycolipid could act as a host antigenwhich is a constituent of the virion. In spite of the biological importance ofglycosphingolipids, little biochemical evidence is available concerning theircontent in cultured cells or the envelopes of membrane-enclosed viruses. Dif-ferences in ganglioside content between normal and virus-transformed cells havebeen described,6 7 and glycolipids have been detected in myxoviruses,8, 9 butnot analyzed in detail.

57

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021

58 MICROBIOLOGY: KLENK AND CHOPPIN PROC. N. A. S.

We have analyzed the glycolipids of four cell types, of isolated plasma mem-branes, and of SV5 virions grown in these cells. In an attempt to correlatebiological behavior with chemical composition in these and previous studies,3'rhesus monkey, bovine, and hamster kidney cells were selected. The plasmamembranes of these cells show striking biological differences, including differingsensitivities to virus-induced cell fusion and immune cytolysis, and large differ-ences in the amount of infective virus produced." 10, 11

Materials and Methods. Glycosphingolipids used as reference substances (Table1) were kindly provided by Drs. E. Klenk, W. Gielen, and G. Tschope of the Uni-versity of Cologne.

Cells: Primary rhesus monkey kidney (MK) cells and three cell lines, bovinekidney (MDBK), baby hamster kidney fibroblasts (BHK21-F), and adult hamsterkidney epithelioid cells (HaK), were grown as described previously.3' 10 For preparationof plasma membranes, BHK21-F and HaK cells were grown in suspension culture. Fordeterminations on whole cells, monolayers were washed with phosphate-buffered saline,scraped off, and pelleted. Cells in suspension culture were pelleted and washed twicewith the saline solution. Packed cells were suspended in 10 volumes of water and dis-rupted by sonication for about 3 min with a Branson sonifier. Protein was determinedon aliquots by the Lowry method, and preparations containing 0.5-3.0 gm protein werelyophilized and stored at -250C until used.Plasma membranes were isolated as described previously3 by a modification of the

fluorescein mercuric acetate method of Warren.12 A plasma membrane preparation fromBHK21-F cells containing 42 mg protein, and three preparations from MDBK cells con-taining 23, 18, and 11 mg protein were analyzed.

Virus: Cells were inoculated with the W3 strain of SV513 at multiplicities of 5-20plaque-forming units/cell, and released virus only was harvested and purified as de-scribed previously.'4 Three samples of MDBK-grown virus containing 27, 22, and 10mg protein, and two samples of BHK21-F-grown virus containing 6 and 10 mg proteinwere analyzed.

TABLE 1. Glycosphingolipids used as reference substances.Structure Source Name

Gal-Cer Ox brain CerebrosideGal-Glu-Cer Ox spleen

Gal-Gal-Glu-Cer Ox spleenGalNAc-Gal-Glu-Cer Brain, Tay-Sachs disease

GalNAc-Gal-Gal-Glu-Cer Human erythrocytes GlobosideGal-GalNAc-Gal-Glu-Cer Cleavage product from

gangliosides-SO3-Gal-Cer Ox brain Sulfatide

NeuNAc-Gal-Glu-Cer Dog erythrocytes N-acetylneuraminosyl-lactoseganglioside (hematoside)

NeuNGly-Gal-Glu-Cer Horse erythrocytes N-glycolylneuraminosyl-lactoseganglioside (hematoside)

Gal-GaiNAc-Gal-Glu-Cer Human brain N-acetvlneuraminosyl-N-tetraoseganghioside

NeuNAcGal-GalNAc-Gal-Glu-Cer Human brain di(N-acetylneuraminosyl)-N-tetra-

ose gangliosideNeuNAc

NeuNAcGal-GalNAc-Gal-Glu-Cer Human brain tri(N-acetylneuraminosyl)-N-

tetraose gangliosideNeuNAc NeuNAc

NeuNAcAbbreviations: Cer = ceramide; Glu = glucosyl; Gal = galactosyl; GalNAc = N-acetyl-

galactosaminyl; NeuNAc = N-acetylneuraminosyl; and NeuNGly = N-glycolylneuraminosyl.

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021

VOL. 66, 1970 MICROBIOLOGY: KLENK AND CHOPPIN 59

Isolation and fractionation of glycolipids: Lyophilized samples were extractedwith chloroform-methanol-water and partitioned into an aqueous phase containingthe more polar glycolipids, and an organic phase containing the less polar glycolipids andother lipids.3 Lipids were extracted from the dialyzed, lyophilized aqueous phase withchloroform-methanol (1:1).6 Neuraminic acid content was determined on aliquots as aparameter of total ganglioside content. The glycolipid mixture was either submitted toanalytical thin-layer chromatography, or further purified and separated by preparativethin-layer chromatography.The organic phase was separated on a silicic acid column' into a fraction containing

neutral lipids and another containing phospholipids and less polar glycolipids. The latterwas dried in a flash evaporator, and glycerophospholipids were removed by alkalinesaponification.'5 These samples were analyzed by thin-layer chromatography. Theglycolipids were further fractionated and separated from sphingomyelin by preparativethin-layer chromatography either immediately or after chromatography on a Florisilcolumn (7.5 gm, prepared in chloroform). On elution with 30 ml of chloroform-methanolmixtures of 10:0, 8:2, 3:7, 2:8, 1:9, and 0:10, fraction 2 contained glucosylceramide,and fractions 4 and 5, N-acetylgalactosaminyl-galactosyl-galactosyl-glucosylceramide.Preparative thin-layer chromatography was done on 20 X 20 cm plates coated with 0.5mm Silica Gel H (Merck, Germany). Two solvent mixtures were employed: chloroform-methanol-water, 65: 25:4 and 60: 35: 8. Glycolipids were localized with iodine, and spotswere removed and extracted with methanol.

Analytical thin-layer chromatography: Glycolipids were separated with the abovesolvents on plates coated with 0.3 mm Silica Gel H. For detection of hexose-con-taining lipids, plates were sprayed with H2S04 (50% by vol), and slowly charred on a hotplate. Before turning black, glycolipids showed a purple color, whereas phospholipidsappeared as brown spots. For detection of gangliosides, plates were sprayed with Bial'sreagent,'6 covered with a glass plate, and heated at 110C for 15 min. Phospholipids werelocated with Zinzadze's reagent.' For identification of sugars, purified glycolipids werehydrolyzed in 4 N HCl for 4 hr at 100'C. Acid was removed by lyophilization, and theresidue was submitted to thin-layer chromatography on coated cellulose sheets (MNPolygram Cel 300, Macherey-Nagel, Germany); solvent, ethyl acetate-pyridine-aceticacid-water (5:5:1:3); stain, alkaline silver nitrate reagent;"8 fixation, 5% sodium thio-sulfate in methanol-water (1:1).

Quantitative analyses: Hexose was determined by the anthrone-H2SO4 method1with glucose as standard, hexosamine by a modification of Elson-Morgan's reac-tion20 with glucosamine-HCl as standard, neuraminic acid with thiobarbituric acid2'after hydrolysis of glycolipids or lipid-extracted virus in 0.1 N H2S04 for 1 hr at 800C,or with resorcinol,22 using N-acetylneuraminic acid as standard.Enzyme degradation of gangliosides :23 About 100 gg ganglioside and 25 units

Vibrio cholerae neuraminidase (Behringwerke, Marburg, Germany) were dissolvedin 0.5 ml sterile distilled water, and incubated in a dialysis bag in 100 ml water overnightat 370C. After concentration of the dialysate in a rotary evaporator, liberated neuraminicacid was examined by thin-layer chromatography.24 The residue in the bag was lyo-philized, extracted with chloroform-methanol (1: 1), and assayed for glycolipids by thin-layer chromatography.

Results. The major glycosphingolipids of cells, plasma membranes, andvirions were characterized by: (1) comparison of migration rates with referencesubstances on thin-layer chromatography, (2) determination of constituentsugars after acid hydrolysis, and (3) partial degradation of gangliosides byneuraminidase.Neutral glycosphingolipids: Most of the neutral glycolipids were found in

the organic phase of the chloroform-methanol-water partition. Figure 1 showsa chromatogram of glycolipids obtained from MK and MDBK cells and SV5virions. The glycolipid identified as glucosylceramide had a slightly higher Rf

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021

60 MICROBIOLOGY: KLENK AND CHOPPIN PROC. N. A. S.

12 3:54 67 8 12 3 4 6 :6 78 1 32 45 6

FIG. 1.-Thin-layer chromatogram of neutral glycolipids of MK and MDBK cells, plasmamembranes, and SYS virions. (1) galactosylceramide, (2) galactosyl-galactosyl-glucosyl-ceramide, (3) N-acetylgalactosaminyl-galactosyl-galactosyl-glucosylceramide, (4) MK cells,(5) SV5 grown in MK cells, (6) MDBK cells, (7) MDBK cell plasma membranes, (8) SV5grown in MDBK cells. The glycolipids of the organic phase were applied. They were ob-tained from samples containing the following amounts of protein: MK and MDBK cells,10 mg; MK-grown SYS and MDBK plasma membranes, 2.5 mg; MDBK-grown SV5, 1 mg.Solvent: chloroform-methanol-water (65:25:4). H2S04 spray. The glycolipids may appearas double spots.33

FIG. 2.-Thin-layer chromatogram of gangliosides of BHK21-F, HaK, MDBK, and MKcells. (1) N-acetylneuraminosyl-lactose ganglioside, (2) BHK21-F cells, (3) N-glycolylneur-aminosyl-lactose ganglioside, (4) HaK cells, (5) N-acetylneuraminosyl-N-tetraose ganglioside,(6) di(N-acetylneuraminosyl)-N-tetraose ganglioside (upper band) and tri(N-acetylneur-aminosyl)-N-tetraose ganglioside (lower band), (7) MDBK cells, (8) MK cells. The ganglio-sides of the aqueous phase were applied. Each was derived from a cell sample containing 10 mgprotein. Solvent: chloroform-methanol-water (60:35:8). Bial's reagent.

FIG. 3.-Thin-layer chromatogram of the three major gangliosides of HaK cells and theirproducts after neuraminidase treatment. (1) N-acetyl and N-glycolylneuraminosyl-lactoseganglioside before, and (2) after neuraminidase treatment, (3) N-acylneuram-inosyl-N-tetraoseganglioside before, and (4) after neuraminidase treatment, (5) di(N-acylneuraminosyl)-N-tetraose ganglioside before, and (6) after neurarninidase treatment. Solvent: chloroform-methanol-water (60:35:8). Bial's reagent. The spot in (2) did not show the typical color.

value than the reference galactosylceramide (Fig. 1, lane 1), and acid hydrolysisof this glycolipid released only glucose. Galactosyl-galactosyl-glucosylceramidewas identified by migration identical to the reference substance from ox spleen(lane 2), and by release of galactose and glucose. (Lactose sulfatide has a similar1?, value. It has been found in small amounts in human kidneys' and its presenceinMK cells cannot be excluded.) N-acetylgalactosaminyl-galactosyl-galactosyl-glucosylceramide had a value identical to the reference substance from humanerythrocytes (lane 3) and yielded galactosamine, galactose, and glucose onhydrolysis, with a molar ratio of hexosamine :hexose of 0.9:3. A small amount(10%) of this glycolipid was found in the aqueous phase.Neutral glycolipids are found in relatively large amounts in MK and MDBK

cells (Fig. 1 and Table 2), but are absent (or present in only trace amounts) inBHK21-F and HaK cells. MDBK cells contain glucosylceramide and N-acetylgalactosaminyl-galactosyl-galactosyl-glucosylceramjde. MK cells containthese and galactosyl-galactosyl-glucosylceramide. Plasma membranes showthe same patterns as whole cells, but the ceramides are present in concentrations4-8 times as high (Fig. 1, Table 3). The neutral glycolipid content of virusgrown in MDBK or MK cells clearly reflects that of the plasma membranes

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021

VOL. 66, 1970 MICROBIOLOGY: KLENK AND CHOPPIN 61

TABLE 2. Neutral glycolipid content of TABLE 3. Neutral glycolipid content ofMDBK and MK cells. MDBK cells, plasma membranes,

MDBK MK and SV5 grown in these cells.(Mug per 100 mg Plasma SV5

Glycolipid protein) Whole mem- vir-Glu-Cer 160 135 cells branes ions

Gal-Gal-Glu-Cer 0 128 (Mig per 100 mgGalNAc-Gal-Gal-Glu-Cer 510 112 Glycolipid protein)

Glu-Cer 160 1300 1550GalNAc-Gal-Gal-Glu-Cer 510 2200 3000

(Fig. 1 and Table 3). However, virions from MK cells contain additional,slowly migrating neutral glycolipids (Fig. 1, lane 5). These have not been com-pletely identified, but one component has the same Rf value as galactosyl-N-acetylgalactosaminyl-galactosyl-glucosylceramide.

Gangliosides: The aqueous phase contained most of the gangliosides. Figure2 shows a chromatogram of gangliosides from the four cell types. The threemajor gangliosides from HaK cells (Fig. 2, lane 4) were isolated by preparativethin-layer chromatography. The fastest-migrating material was found to bepredominantly N-acetylneuraminosyl-galactosyl-glucosylceramide, with a smallamount of N-glycolylneuraminosyl-galactosyl-glucosylceramide. These com-pounds formed a double spot with Rf values identical to those for the referencesubstances from dog and horse erythrocytes (lanes 1 and 3). Acid hydrolysisreleased glucose and galactose, and the products of neuraminidase degradationwere galactosyl-glucosylceramide (Fig. 3), N-acetylneuraminic acid, and N-glycolylneuraminic acid. A smaller amount of this glycolipid was recoveredfrom the organic phase. The second ganglioside was identified as galactosyl-N-acetylgalactosaminyl-(N-acylneuraminosyl-) galactosyl-glucosylceramide be-cause: (a) it had an Rf value identical to the reference substance from humanbrain (Fig. 2, lane 5); (b) hydrolysis released glucose, galactose, and galactos-amine; and (c) it was unaffected by neuraminidase (Fig. 3), which showed thatneuraminic acid is not linked to the terminal galactose. This second gangliosidefrom HaK cells and N-acetylneuraminic acid were the products of neuramini-dase action on the third ganglioside (Fig. 3); this indicates the presence of anadditional neuraminic acid residue in the latter. Since its Rf value was dif-ferent from galactosyl-N-acetylgalactosaminyl-di(N-acetylneuraminosyl-)galac-tosyl-glucosylceramide (Fig. 2, lane 6), this third ganglioside appears to be N-acetylneuraminosyl-galactosyl-N-acetylgalactosaminyl - (N- acylneuraminosyl-) -galactosyl-glucosylceramide.

There are distinctive qualitative (Fig. 2) and quantitative (Table 4) differencesbetween the gangliosides of the various cells. BHK21-F cells contain the mosttotal ganglioside, which is predominantly N-acetylneuraminosyl-lactose gangli-oside, and very small amounts of unidentified gangliosides. HaK cells have alower content, but contain the three major gangliosides described above.MK and MDBK cells have patterns similar to HaK cells, but contain only smallamounts of gangliosides.As with neutral glycolipids, plasma membranes of BHK21-F and MDBK cells

contain the same gangliosides as whole cells, but in much higher concentrations

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021

62 MICROBIOLOGY: KLENK AND CHOPPIN PROC. N. A. S.

TABLE 4. Ganglioside- content of BHK21-F, HaK, MDBK, and MK cells, plasma mem-branes, and SV5 grown in these cells.

PlasmaCells Whole cells membranes SV5 virions

(Glycolipid-bound neuraminic acid, pxg per 100 mg protein)BHK21-F 83 320 None detectedHaK 20 ...

*...

MDBK 9 30 None detectedMK 8 ... None detected

* Not determined.

(Table 4). Gangliosides were not found in most preparations of SV5 virions,but were detected in trace amounts in one sample of BHK-grown virus. Thelack of neuraminic acid-containing glycolipids in virions grown in cells that con-tain large amounts in their plasma membranes provides the first example of amembrane lipid that is not incorporated into the viral envelope. To determineif protein-bound neuraminic acid was present in SV5 virions, which contain about6% carbohydrate,'4 we analyzed virus samples containing 0.8-3.4 mg protein.No neuraminic acid was detected (neuraminic acid content therefore <0.03%).As discussed below, the absence of this substance from virions could be related tothe presence of the viral neuraminidase.

Discussion. Table 5 summarizes results for the glycosphingolipids found inthe four cell types. The various cells show distinctive patterns. BHK21-F cellscontain the highest amount of gangliosides, almost all in the form of N-acetyl-neuraminosyl lactose ganglioside, a finding similar to that of Hakomori andMurakami in another line of BHK21 cells.6 The other cells contain mono(N-acylneuraminosyl)-and di(N-acylneuraminosyl)-N-tetraose gangliosides, a pat-tern found in cultured mouse cells.7 A striking finding is that BHK21-F andHaK cells, which contain large amounts of gangliosides, contain almost noneutral glycolipids, whereas the reverse is true with MDBK and MK cells.The relative concentrations of glycolipids in whole cells and plasma membranesindicate that these substances are located predominantly in plasma membranes,but their presence in other cellular structures cannot be excluded. It is clearthat glycolipids comprise a significant portion of the total polar lipids of theplasma membranes, 30% in MDBK cells and 10% in BHK21-F cells.

TABLE 5. Glycolipid content offour types of cultured cells.Cells-

BHK21-F HaK MDBK MKNeutral glycolipids:

Glu-Cer i i + +Gal-Glu-Cer - - +

Gal-Gal-Glu-Cer - - - +GaJNAc-Gal-Gal-Glu-Cer - - + +

Gangliosides:NeuNAc-Gal-Glu-Cer

+ + + + +NeuNGly-Gal-Glu-Cer

Gal-GalNAc-Gal-Glu-Cer - + + +

NeuNAcGal-GalNAc-Gal-Glu-Cer - + + +

NeuNAc NeuAAc

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021

VOL. 66, 1970 MICROBIOLOGY: KLENK AND CHOPPIN 63

Previous studies3 4indicated that, in general, quantitative differences in thelipid patterns of the plasma membranes were clearly reflected in the virionsgrown in different cells. This has now been found with neutral glycolipids, buthere the qualitative differences between various cell membranes are also reflectedin the virions, e.g., galactosyl-galactosyl-glucosylceramide is found in MK cellsand virions grown in them, but not in MDBK cells or virions. Such differencesin the lipids of virions which, though grown in different cells, contain the samevirus-specific proteins again emphasize the importance of the host cell membranein determining the lipid composition of the virion.In contrast to the incorporation of cholesterol, phospholipid, and neutral

glycolipids of the plasma membrane into SV5 virions is the lack of incorporationof gangliosides into virions. A possible explanation is that neuraminidase issynthesized in infected cells and incorporated into the viral envelope. Thisenzyme might either cleave neuraminic acid from previously formed gangliosides,or interfere with the synthesis of gangliosides by preventing the addition ofneuraminic acid residues to the terminal sugar of a growing carbohydrate chain,thus leading to the synthesis of other, neutral glycolipids. That the latter maybe the case is suggested by the presence of unidentified neutral glycolipids withlong carbohydrate chains in virus grown in MK cells (Fig. 1), and by the absenceof neuraminidase-insensitive gangliosides, i.e., those with neuraminic acid in anonterminal position. Whatever the precise explanation, the neuraminidaseaction apparently occurs only at regions of cell membrane where virus proteinsare incorporated, because infected cells still contain membrane-bound neuraminicacid and because colloidal iron, which stains neuraminic acid residues, stains thesurface of infected cells everywhere except where virus is budding.25

Gangliosides thus represent the only class of plasma membrane lipids that areeliminated, or drastically decreased, when the membrane is transformed into theviral envelope. It is therefore interesting that the ganglioside content of thefour cell types investigated is directly proportional to the extent of fusion in-duced in these cells by SV5, and inversely proportional to the yield of infectivevirus.', 10 BHK21-F cells, with the highest content, fuse rapidly and extensivelyand then disintegrate, and produce little virus because of a block in virus matura-tion at the cell surface; whereas MK cells, which contain a small amount ofgangliosides, are resistant to cell fusion and produce a high yield of infectivevirus with little cell damage. These and previous results,3' 4in which there wasa correlation between high phosphatidylethanolamine content, and high virusyield, and little cell fusion, suggest that the lipid composition of the plasmamembranes and changes in lipids that occur in virus-altered regions of thesemembranes may play an important role in virus yield, cell fusion, and cell death.

Glycolipids may be antigenic5 and contribute to the immunological specificityof cell surfaces. When isolated from different sources, N-acetylgalactosaminyl-galactosyl-galactosyl-glucosylceramide (which is a major glycolipid in MK andMDBK cells) exhibits different immunological specificities, including Forssmanantigenicity, depending on the anomeric configuration of the terminal disac-charide.26' 27 Blood group and Forssman antigenic activities were found as-sociated with myxovirus and paramyxovirus particles grown in cells whichpossess these antigens,^' 29 although in some cases the virus was purified only

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021

64 MICROBIOLOGY: KLENK AND CHOPPIN PROC. N. A. S.

by differential centrifugation. The present studies provide biochemical evi-dence of such glycosphingolipids in highly purified parainfluenza virus particles.These results and the finding of a glycopeptide host cell antigen in influenzavirions30 31 suggest that if host cell antigens are present in enveloped viruses,carbohydrate moieties will be the antigenic determinants. The recent evidencesuggesting that all the proteins of myxovirus and paramyxovirus particles areprobably virus-coded2' 32supports this concept.We thank Mrs. Barbro Hammarstr6m and Miss Anna Wangel for excellent technical assis-

tance.* Supported by research grant AI-05600 from the National Institute of Allergy and Infectious

Diseases and contract AT(30-1)-3983 from the U.S. Atomic Energy Commission.1 Compans, R. W., K. V. Holmes, S. Dales, and P. W. Choppin, Virology, 30, 411 (1966).2 Caliguiri, L. A., H.-D. Klenk, and P. W. Choppin, Virology, 39, 460 (1969).3 Klenk, H.-D., and P. W. Choppin, Virology, 38, 255 (1969).4Klenk, H.-D., and P. W. Choppin, Virology, in press (1970).6 Martensson, E., Progr. Chem. Fats Lipids, 10, 367 (1969).6 Hakomori, S., and W. T. Murakami, these PROCEEDINGS, 59, 254 (1968).7 Mora, P. T., R. 0. Brady, R. M. Bradley, and V. W. McFarland, these PROCEEDINGS, 63,

1290 (1969).8 Armbruster, O., and U. Beiss, Z. Naturforsch., 13b, 75 (1958).9 Blough, H. A., and D. E. M. Lawson, Virology, 36, 286 (1968).10 Holmes, K. V., and P. W. Choppin, J. Exptl. Med., 124, 501 (1966)."1 Holmes, K. V., H.-D. Klenk, and P. W. Choppin, Proc. Soc. Exptl. Biol. Med., 131, 651

(1969).12 Warren, L., M. C. Glick, and M. K. Nass, J. Cell Physiol., 68, 269 (1966).13 Choppin, P. W., Virology, 23, 224 (1964).14 Klenk, H.-D., and P. W. Choppin, Virology, 37, 155 (1969).15 Eto, T., Y. Ichikawa, K. Nishimura, S. Ando, and T. Yamakawa, J. Biochem., 64, 205

(1968).16 Klenk, E., and H. Langerbeins, Z. Physiol. Chem., 270, 185 (1941).17 Dittmer, J. C., and R. L. Lester, J. Lipid Res., 5, 126 (1964).18 Stahl, E., Dfinnschichtchromatographie 2nd ed. (Berlin: Springer-Verlag, 1967).19 Scott, T. A., Jr., and E. H. Melvin, Anal. Chem., 25, 1656 (1953).20 Gatt, R., and E. R. Berman, Anal. Biochem., 15, 167 (1966).21 Aminoff, D., Biochem. J., 81, 384 (1961).22 Suzuki, K., Life Sci., 3, 1227 (1964).23 Klenk, E., and W. Gielen, Z. Physiol. Chem., 330, 218 (1963).24 Granzer, E., Z. Physiol. Chem., 328, 277 (1962).25 Klenk, H.-D., R. W. Compans, and P. W. Choppin, unpublished experiments.26 Makita, A., C. Suzuki, and Z. Yosizawa, J. Biochem., 60, 502 (1966).27 Koscielak, J., S. Hakomori, and R. W. Jeanloz, Immunochem., 5, 441 (1968).28 Rott, R., R. Drzeniek, M. S. Saber, and E. Reichert, Arch. Ges. Virusforsch., 19, 273

(1966).29 Isacson, P., and A. E. Koch, Virology, 27, 129 (1965).30 Laver, W. G., and R. G. Webster, Virology, 30, 104 (1966).31 Lee, L. T., C. Howe, K. Meyer, and H. U. Choi, J. Immunol., 102, 1144 (1969).32 Holland, J. J., and E. D. Kiehn, Science, 167, 202 (1970).33 Adams, E. P., and G. M. Gray, Chem. Phys. Lipids, 2, 147 (1968).

Dow

nloa

ded

by g

uest

on

Mar

ch 1

0, 2

021