determination of plasma membrane lipid mass and composition in

THE JOURNAL OF BIOLOGICAL CHEMISTRY 8 1993 by The American Society for Biochemistry and Molecular Biology. Inc.

Vol. 268, No. 14, Issue of May 15, pp. 10145-10153,1993 Printed in U.S.A.

Determination of Plasma Membrane Lipid Mass and Composition in Cultured Chinese Hamster Ovary Cells Using High Gradient Magnetic Affinity Chromatography*

(Received for publication, November 16, 1992, and in revised form, December 29, 1992)

Dale E. Warnock$, Charlotte Roberts$, Mallory S. Lutz$, Wendy A. Blackburn$, William W. Young, Jr.6, and Jacques U. BaenzigerST From the $Department of Pathology, Washington University School of Medicine, St. Louis, Missouri 63110 and the §Departments of Biological and Biophysical Sciences and Biochemistry, Health Sciences Center, University of Louisuille, Louisuille, Kentucky 40292

We have utilized wheat germ agglutinin conjugated to ironldextran particles in conjunction with high gradient magnetic affinity chromatography (HIMAC) to prepare plasma membranes from cultured cells. Mem- brane-impermeable succinimidyl esters inactivate alkaline phosphodiesterase l (APDE-l) and were used to establish the proportion of APDE-1 expressed at the cell surface. The yield of inhibitable APDE- 1 provides an accurate indication of plasma membrane yield, which was >90% for Chinese hamster ovary (CHO) cells. Plasma membranes prepared by HIMAC contained <5-13% of endoplasmic reticulum, Golgi, mitochondria, lysosomes, or endosomes. Pulse-chase experiments performed with the a5@l integrin receptor confirmed the high yield of plasma membrane and demonstrated the utility of this procedure for examining trafficking of proteins to and from the plasma membrane. We determined the lipid content of plasma membranes prepared by HIMAC. CHO plasma membranes contain 49% of total cellular phospholipid, 69% of sphingomyelin, and 64% of cholesterol. Phosphati- dylserine was the only glycerophospholipid highly enriched (71%) in the retained fraction. The glycosphingolipids lactosylceramide and ganglioside GM, were enriched in the plasma membrane fraction to the same extent as sphingomyelin. The major fraction of the glycosphlngolipid precursors glucosylceramide and ceramide was localized to intracellular membranes. These findings indicate that the plasma membrane of CHO cells contains approximately half of the total cellular phospholipids and an even higher percentage of sphingomyelin and cholesterol. The high efficiency and rapidity of this isolation procedure should aid the analysis of plasma membrane components significantly.

The plasma membrane is a dynamic and complex organelle that has been the subject of extensive studies enhancing our

* This work was supported by National Institutes of Health Grants R37-CA21923 (to J. U. B.) and GM42698 (to W. W. Y.) and by American Cancer Society Grant FRA-358 (to W. W. Y.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduer- tisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

ll T o whom correspondence should be addressed Dept. of Pathol- ogy, Washington University School of Medicine, 660 S. Euclid Ave., BOX 8118, St. Louis, Missouri 63110. Tel.: 314-362-7440; Fax: 314- 362-8888.

understanding of the trafficking of proteins to and from the plasma membrane. Less is known about the lipid components of the plasma membrane and their trafficking. This largely reflects the difficulty of preparing sufficiently pure membranes in high yield, especially from cells in culture. Since lipid components may modulate the activity of receptors and influence trafficking of the plasma membrane, detailed information about the lipid composition of the plasma membrane is essential.

Analyses of the composition and mass of total lipid that comprises the plasma membrane of cultured cells have been controversial (1,2). Lange and co-workers (3) estimated that 90% of cholesterol and sphingomyelin are in the plasma membrane of cultured fibroblasts and that roughly 50% of the total phospholipid is also present in the plasma membrane (3). Others have suggested (4) that nearly all of the glycosphingolipids are also located in the plasma membrane. Van Meer has challenged these estimates (I, 2). Based on morpho- metric determinations of the surface areas of various intracellular compartments and the ratios of their lipid components, he suggested that only 30-35% of sphingomyelin and 25-40% of cholesterol are in the plasma membrane and that the glycosphingolipids are also predominantly intracellular (1, 2). The discrepancies between these estimates have been suggested to reflect unaccounted contributions of endosomal and trans-Golgi membranes to the plasma membrane preparations.

We have now characterized the phospholipid, glycolipid, and cholesterol content of plasma membranes prepared using a ligand-specific isolation procedure that we developed recently (5, 6). Iron/dextran particles (FeDex)’ bearing cova- lently linked wheat germ agglutinin (WGA/FeDex) are bound to cells, and following disruption of the cells, the plasma membranes are isolated by high gradient magnetic affinity chromatography (HIMAC). Membranes bearing sufficient numbers of FeDex particles are selectively retained by the magnetic field produced by placing steel wool between the

The abbreviations used are: FeDex, iron/dextran particles; WGA, wheat germ agglutinin; WGA/FeDex, WGA conjugated to iron/ dextran particles; HIMAC, high gradient magnetic affinity chromatography; APDE-1, alkaline phosphodiesterase 1; CHO, Chinese hamster ovary; LacCer, lactosylceramide; GM3, NeuAca2-3Gal@l- 4GlcCer; GlcCer, glucosylceramide; sulfo-SHPP, sulfosuccinimidyl- 3-(4-hydroxyphenyl)propionate; PBS, phosphate-buffered saline; Me,SO, dimethyl sulfoxide; sulfo-NHS-biotin, sulfosuccinimidobi- otin; NHS-disulfobiotin, sulfosuccinimidyl 2-(biotinamido)ethyl- 1,3-dithiopropionate; NBD-ceramide, N-[7-(4-nitrobenzo-2-oxa-1,3- diazole)]-6 aminocaproyl D-erythro-sphingosine; mAb, monoclonal antibody; PAGE, polyacrylamide gel electrophoresis.

10145

10146 Plasma Membrane Lipid Mass and Composition in CHO Cells

poles of a strong (4,000 G) magnet. Membranes that do not bear FeDex particles, individual FeDex particles, and soluble proteins are not retained by the column. The bound membranes are subsequently eluted by removing the column from the magnetic field. We have determined that the yield of plasma membrane can be accurately and conveniently estab- lished by selective inactivation of surface APDE-1 with the membrane-impermeant agent sulfosuccinimidyl-3-(4-hydrox- ypheny1)propionate (sulfo-SHPP). This procedure for membrane preparation allows multiple plasma membrane components to be measured simultaneously without dependence on the physical-chemical properties of the plasma membrane.

We have documented an 80-90% yield of plasma membrane largely free of endosomes, Golgi, endoplasmic reticulum, and lysosomes as measured by resident enzyme markers and re- porter molecules known to pass through these compartments. Our analyses indicate that half of the total cellular phospholipid is present in the plasma membrane of CHO cells. We find that 64-70% of sphingomyelin, cholesterol, and Gbf3 are also located in the plasma membrane fraction. These values represent some of the first direct measurements of plasma membrane lipid components from cultured cells which have not required large correction factors for yields.

EXPERIMENTAL PROCEDURES

Materials-Unless otherwise specified, chemicals and chromatography reagents were purchased from Sigma. Sulfo-NHS-biotin and sulfo-SHPP were from Pierce Chemical Co. The luminol Western reagents were from Amersham Corp. The cholesterol oxidase assay kit was obtained from Boehringer Mannheim. Fetal bovine serum was purchased from HyClone Laboratories (Logan, UT) and cultu- reware from Corning and Costar. Tissue culture reagents were from the Tissue Culture Support Center, Washington University, St. Louis Mo. T r a ~ ~ ~ ~ S - l a b e l ([35S]methionine, up to 15% [35S]cysteine; 1,000 Ci/mmol) was from ICN Biomedicals, Inc.; NaiZ5I was from Amer- sham Corp.; [3H)palmitate was from Du Pont-New England Nuclear; and [3H]galactose was from American Radio Chemicals (St. Louis). IgSorb was obtained from Enzyme Center, Inc. (Maiden, MA). Sep- Pak C-18 cartridges were from Waters. Budget-Solv scintillation fluid was purchased from Research Products International Corp. (Mt. Prospect, IL). Autoradiographic image enhancers were from National Diagnostics, Inc. (Somerville, NJ) and Amersham Corp. XAR-5 di- agnostic film was purchased from Eastman-Kodak. Fetal-Tek 200 TLC plates were from Helena Laboratories (Beaumont, TX) and high performance thin layer chromatography plates from Merck. Monoclonal antibody H68 to the human transferrin receptor was a gift of Dr. Ian Trowbridge, Salk Institute, San Diego, CA. Antibodies to integrin a5 and Pl were provided by Dr. John McDonald, Mayo Clinic, Scottsdale, AZ. Determinations of APDE-1 in MOPC 315P cells by quantitative Western blotting (7) were performed in the laboratory of Dr. Scot Hickman, Washington University Medical Service, Veterans Affairs Medical Center, St. Louis.

Cell Culture-CHO cells were grown in a-minimal essential medium supplemented with 10% (v/v) heat-inactivated fetal bovine serum. Two days prior to experimentation, cells were split by trypsin/ EDTA dispersion and plated into 100-mm Corning culture dishes. CHO K1 cells were grown as the CHO cells except that they were maintained in Dulbecco's modified Eagle's medium containing 10% iron-supplemented, heat-inactivated fetal bovine serum, 2 mM glu- tamine, 1 X nonessential amino acids, 1 mM proline, and 1 X penicil- lin/streptomycin.

Metabolic Labeling-For pulse-chase experiments, cells were rinsed three times in Dulbecco's phosphate-buffered saline (PBS) and incubated for 1 h at 37 "C in cysteine/methionine-free medium. The medium was replaced with fresh Cys/Met-free medium containing 150 mCi/ml Tran3'S-labe1, and the cells were allowed to incorporate label for 20 min. Pulses were terminated by washing the cells twice with PBS, and the chase was initiated by the addition of complete medium. For metabolic labeling of lipids, 50 pCi of [3H]palmitate in 1.4 ml of 0.04% (w/v) defatted bovine serum albumin that had been dialyzed against a-minimal essential medium was added to 3.6 ml of culture medium for each 100 mm culture dish. After incubation in L3H]palmitate for periods ranging from 1 to 16 h, cells were rinsed

with ice-cold binding buffer (0.14 M NaCI, 5.4 mM KCl, 0.33 mM NaZHPO4.7H,O, 0.34 mM KH2P04, 0.8 mM MgS04, 2.7 mM CaCl. 2Hz0, 20 mM HEPES, pH 7.4) and membranes prepared for isolation by HIMAC. Under these conditions the incorporation of [3H]palmitate into phospholipids and glycosphingolipids increased linearly for at least 16 h. Since there was no change in the distribution of individual labeled species between the retained and nonretained fractions after 3 h of [3H]palmitate incorporation, the distribution of phospholipids and glycosphingolipids was examined after 3 h of labeling unless otherwise noted.

Incorporation of Succinimidyl Esters into Surface Proteins- Plasma membrane proteins expressed at the cell surface were selectively labeled with sulfo-NHS-biotin and NHS-disulfobiotin as described by Le Bivic et al. (8) Cells were washed twice with ice-cold PBS containing 1 mM MgClz and 0.1 mM CaC12 (PBS/C-M) and then incubated in PBS/C-M containing 0.5% Me,SO and 1 mg/ml sulfo- NHS-biotin or NHS-disulfobiotin for 30 min on ice. Cells were washed once and incubated in PBS/C-M, 50 mM NH,Cl for 15 min and then rinsed twice with PBS/C-M. Hydroxyphenylpropionate was

sulfo-SHPP as described by Thompson et al. (9). CHO cells were incorporated into surface proteins using the membrane-impermeable

washed twice with ice-cold PBS/C-M and then incubated on ice for 30 min in PBS/C-M, 0.5% Me,SO, and 0-2.5 mg/ml sulfo-SHPP. The reaction was stopped by incubating in PBS/C-M with 1 mg/ml lysine for 15 min at 4 "C. Cells were then rinsed once with PBS/C- M and membranes prepared for HIMAC isolation.

HZMAC-WGA was coupled to FeDex as described previously (5). For isolation of plasma membrane, subconfluent 100-mm diameter culture plates of CHO cells were washed twice in ice-cold binding buffer. Cells were incubated in binding buffer with 1 mg/ml bovine serum albumin and WGA/FeDex (1 pg/ml) for 1 h at 4 "C. The cells were then washed twice in binding buffer and taken up in 4 ml of sucrose buffer (0.25 M sucrose, 10 mM Tris, pH 7.4, protease inhibitor mixture: 0.5% Me2S0, 20 pg/ml pepstatin, 20 pg/ml antipain, 20 pg/ ml leupeptin, 20 pg/ml chymostatin, 25 IU/ml aprotinin A (final concentration)) by scraping with a rubber policeman. Cells were disrupted after 10 min on ice by 20 passes through a 22-gauge needle followed by 20 strokes in a glass Dounce homogenizer with a tight pestle (Kontes). A postnuclear supernatant was prepared from the homogenate by sedimentation for 10 min at 400 X gave at 4 "C to remove nuclei and unbroken cells. The postnuclear supernatant was loaded into a 3-ml syringe packed with stainless steel wool (Brunsnet, 25-pm airlayed web). The syringe was placed between the poles of a 4000-gauss permanent magnet for 2 h at 4 "C. The column was sequentially eluted with 20 ml of sucrose buffer (nonretained fraction), 10 ml of 300 mM KC1 in sucrose buffer (KI fraction) with the magnet in place, and then with 10 ml of sucrose buffer in the absence of the magnet and with agitation of the syringe column (retained fraction).

Enzyme and Protein Assays-Protein determinations and enzyme assays were adapted for performance in 96-well assay plates and analyzed using a Vmax Kinetic Microplate Reader (Molecular Devices, Palo Alto, CA). Unless indicated, the reactions were performed at room temperature. Protein was determined by the bicinchoninic acid assay (10) using bovine y-globulin as the standard. Proteins in 50 pl of sucrose buffer were mixed with 200 p1 of bicinchoninic reaction mixture, incubated at 37 "C for > 3 h, and the absorbance determined at 550 A. APDE-1 (11) was assayed by the addition of 100 pl of 0.2 M Tris, pH 8.9, 0.2% Nonidet P-40, 5 mM p-nitrophenylthymidine 5'-phosphate, to 100 pl of membrane fractions in sucrose buffer in a 26-well enzyme assay plate. The rate of increase in absorbance at 405 A was monitored using the kinetic program of the microplate reader. Cytochrome c oxidoreductase (12) was assayed by mixing 100 pl of sucrose buffer containing membrane fractions with 100 p1 of 1 mM NADH, 0.001% saponin and adding 100 pl of PBS, pH 8.0, containing 5 mg/ml cytochrome c (oxidized) to start the reaction. The change in absorbance at 560 8, was monitored in the microplate reader.

P-Hexosaminidase activity (13) was determined by the addition of 100 pl of 0.2 M acetate, 10 mM p-nitrophenyl N-acetyl-@-D-glucosa- minide; 0.2% Nonidet P-40 to 100 p1 of membrane fractions in sucrose buffer. a-Glucosidase I1 (14, 15) was determined by the addition of 100 pl of 0.2 M sodium phosphate, pH 6.9; 5 mM p-nitrophenyl a - D - glucopyranoside, 0.2% Nonidet P-40 to 100 pl of sucrose buffer containing membrane fractions. Both assays were terminated at four different time points by the addition of 50 pl of 1% SDS, 1 M Tris, pH 11. The amount ofop-nitrophenyl released was determined from the absorbance at 405 A using the microplate reader.

Galactosyltransferase activity was assayed on membranes that had

Plasma Membrane Lipid Mass and Composition in CHO Cells 10147

been pelleted at 190,000 X geve for 2 h in a Beckman SW 55 T i rotor. Pellets were resuspended in 20% glycerol, 10 mM HEPES, pH 7.0, 0.1% Triton X-100, 5 mM EDTA (16), protease inhibitor mixture. Each assay of 50 p1 contained final concentrations of 1 mM ATP, 40 mM MnCI, 0.03 mM UDP-galactose (specific activity 0.2 mCi/WmOl), 100 mM Tris, pH 8.0, 200 pmol of dabsylated asialoagalactoglycopep- tide, and 20 p1 of enzyme buffer. Dabsylated glycopeptide was gener- ated by coupling dabsylchloride to glycopeptide fragments from fetuin as described (17). Sialic acid was removed from the glycopeptide by boiling in 2 N acetic acid for 15 min. Galactose was removed by digestion with diplococcal 8-galactosidase in 50 mM cacodylate buffer, pH 6.0. 8-Galactosidase activity was destroyed by heating to 65 "C for 10 min. Incorporation of [3H]galactose into the acceptor substrate was determined by isolation of the dabsylated glycopeptide product on (2-18 Sep-Pak cartridges. Immediately after stopping the galactosyl transferase reaction by dilution with 0.5 ml of ice-cold dH,O, the sample was loaded onto a Sep-Pak cartridge and unincorporated ['HI galactose eluted with 10 ml dH,O. Dabsylated glycopeptide was eluted in 5 ml of 50% methanol, and the amount of [3H]galactose incorporated was determined by scintillation counting.

Immunoprecipitations and Western Blots-Immunoprecipitations of fibronectin subunits were performed as described (18) except IgSorb was used instead of protein A-Sepharose to isolate antibodies, and 1 mg/ml heat denatured bovine serum albumin was present during all immunoprecipitations and rinses. Homogenates were sequentially precipitated with mAb 16, specific for human u5; mAb 33, specific for human and hamster a5; and mAb 66, specific for the 81 subunit (18). The 05 that was precipitated by mAb 16 and the 81 that was brought down by mAb 66 ( i e . not associated with 015) were analyzed separately by SDS-PAGE (18) using a 6% resolving gel and a 3% stacking gel.

Proteins for Western blot analysis were resolved by SDS-PAGE using a 10% resolving, 5% stacking polyacrylamide gel. Proteins were transferred to Immobilon-P by standard methods (19). All further incubations and washing were done a t room temperature in Tris- buffered saline with 0.2% Tween 20 (TBS-T20). Nonspecific binding was blocked by incubation in 3% bovine serum albumin in TBS-T20, for > 2 h. Incubations with primary antibodies were for 1 h and were followed by three 5-min washes with TBS-T2O. Specifically bound antibodies were visualized using horseradish peroxidase-conjugated goat anti-mouse antibody and enhanced chemiluminescence (Amer- sham Corp.).

Lipid Analysis-Plasma membranes from 10 100-mm dishes were prepared by HIMAC for chemical determination of membrane lipids. Membrane pellets were extracted twice on ice with at least 30 volumes of CHCI3:MeOH, 21, v/v, followed by CHC13:MeOH, 1:2 twice using an Omni Mixer (Omni International, Norwalk, CT). Total phospholipids were determined by the methods of Gerlach and Deuticke (20) and Marshall-Stewart (21). Cholesterol was converted to cholesten- one with cholesterol oxidase and assayed spectrophotometrically using a cholesterol oxidase assay kit. All volumes were reduced 10- fold so that absorbances could be read in 96-well plates with a Molecular Devices microplate reader. Sphingomyelin content was determined by chromatography in the solvent CHC13:MeOH:2-pro- panol:triethylamine:H,O, 30:9:25:25:7, v/v, using Fetal-Tek 200 TLC plates. Plates were sprayed with 3% cupric acetate in 8% phosphoric acid, heated a t 160 "C for 10 min, and the sphingomyelin band quantitated by scanning at 580 nm with a Shimadzu CS-9000 densi- tometer.

Analysis of [3H]palmitate-labeled lipids was performed as follows. The phospholipids of crude lipid extracts were analyzed by two- dimensional thin layer chromatography as described previously (22). High performance TLC plates were activated a t 110 "C for 1 h. Labeled samples were applied with cold carrier phospholipids, and development was performed in the first dimension solvent, CHC1,:MeOH:acetic acid, 65:25:10, v/v. After drying the plate for 20 min under a room temperature air stream, the plate was developed in the second dimension solvent, CHCI3:MeOH:88% formic acid, 65:25:10, V/V. The glycosphingolipids from a portion of ['Hlpalmitate- labeled lipid extracts were purified by the acetylation procedure (23), which consisted of acetylation overnight at room temperature in pyridine and acetic anhydride, Florisil chromatography, and deace- tylation of the glycospbingolipids in 0.15 M sodium methoxide a t room temperature for 30 min. Samples were analyzed on high performance TLC plates in the solvent CHC13:MeOH:H20, 62:30:6, V/V. The solvent diethyl ether:MeOH, 99:1, v/v, was used for ceramide quantitation of crude lipid extracts (24). Radiolabeled TLC spots were located using a TLC linear analyzer (EG&G Berthold) and

quantitated by scraping into scintillation vials and assaying in the presence of 10 ml of Budget-Soh scintillation fluid.

RESULTS

Isolation of Plasma Membranes from Cultured Cells by HI- MAC-Plasma membranes were prepared from cultured cells as described previously (5) by incubating cells with WGA/ FeDex at 4 "C, removing unbound WGA/FeDex, disrupting cells, and passing the membranes from the postnuclear supernatant over a steel wool column placed between the poles of a 4,000-gauss permanent magnet. The yields of enzyme markers specific for plasma membranes, lysosomes, Golgi, endoplasmic reticulum, and mitochondria in the postnuclear supernatant were similar, falling the range of 65-80% for individual experiments. After washing the columns in the presence of the magnetic field to remove free WGA/FeDex and membranes not bearing sufficient WGA/FeDex to be retained, the magnetic field was removed, and membranes bearing WGA/FeDex were eluted. The enzyme activities used as markers were quantitatively recovered in the nonretained and/or retained fractions as was observed previously with HepG2 cells (5, 6).

The retained fraction prepared from cultured CHO cells contained 70% of the APDE-1 and 2.7% of the cellular protein (Fig. 1) resulting in a 26-fold enrichment of APDE-1 with respect to total protein. Since APDE-1 is predominantly in the plasma membrane (see below), at least 70% of the plasma membrane from CHO cells was present in the retained fraction. Enzyme markers for intracellular organelles such as lysosomes, mitochondria, and endoplasmic reticulum were recovered at levels of 5-10% of the total loaded onto the HIMAC column (Fig. 1). Thirteen percent of the Golgi marker galactosyltransferase was present in the retained fraction and represented the highest level of a marker not typically associated with the plasma membrane.

The yield of APDE-1 indicated that isolation of plasma membranes from CHO cells using WGA/FeDex and HIMAC

c "1 T

FIG. 1. High yield isolation of plasma membrane from cultured cells by HIMAC. 100-mm diameter culture plates of CHO cells were incubated at 4 "C for 1 h with WGA/FeDex (1 Fg/ml) in 5 ml of binding buffer, 1 mg/ml bovine serum albumin. The cells were then washed twice in ice-cold binding buffer to remove unbound WGA/FeDex. Cells from two plates were disrupted, and plasma membrane was isolated by HIMAC as described under "Experimental Procedures." HIMAC nonretained and retained fractions were assayed for enzymatic marker activity specific for plasma membrane (APDE-I), lysosomes ((Lhexosaminidase), endoplasmic reticulum (cytochrome c reductase and a-glucosidase II), Golgi (galactosyl transferase), and for protein. The results are presented as the amount of enzyme or protein isolated in the retained fraction as a percent of total recovered from the HIMAC column. The error burs are equal to the standard deviation of 10 separate experiments to determine the average.


was highly efficient. To assure that this was generally true for cell surface proteins, we performed the experiments described in Fig. 2. Cell surface proteins were selectively biotinylated using the membrane-impermeant agent sulfo-NHS- biotin (8). Plasma membranes were then prepared using WGA/FeDex and HIMAC as described above. The membranes in the retained and nonretained fractions were disrupted by freeze-thawing (Fig. 2 A ) or washing with Na2CO:{, pH 11 ( 2 5 ) (Fig. ZB) , and pelleted a t 150,000 x g to remove nonintegral proteins, such as extracellular matrix proteins, which had incorporated biotin. Following SDS-PAGE and transfer to Immobilon, the biotinylated proteins were identi- fied and quantitated by probing with '"I-streptavidin. The retained fractions shown in Fig. 2 , A and B, contained 64 and 63%, respectively, of the biotinylated membrane proteins recovered. This was true for the total population of proteins and for individual proteins such as the protein migrating with a molecular mass of 140 kDa. The yield of biotinylated proteins from the cell surface was therefore similar to the yield

Selective Inactivation of APDE-1 with Succinimidyl Es- ters-In the course of biotinylating surface proteins with sulfo-NHS-biotin we noted that the majority of APDE-1 had been inactivated. This suggested that APDE-1 expressed at the cell surface could be selectively inactivated using membrane impermeant agents such as sulfo-NHS-biotin, thus providing a means to determine readily and accurately the proportion of APDE-1 expressed at the surface of cultured cells. APDE-1 is inhibited by a number of agents that react

of APDE-1.

A N.R. Ret. "

200

0 N.R. Ret. "

1 I6

46

FIG. 2. Surface-labeled integral plasma membrane proteins are highly enriched in the HIMAC retained fraction. Proteins at the cell surface were exogenously labeled with the membrane- impermeant reagent sulfo-NHS-biotin as described under "Experi- mental Procedures." The addition of WGA/FeDex to the plasma membrane and isolation by HIMAC were performed as described in Fig. 1. Equal volumes of the nonretained (N.R.) and retained (Ret . ) fractions were treated with one freeze-thaw cycle (panel A ) or Na2COn, pH 11 (panel B ) , on ice for 5 min and then pelleted for 15 min a t 150,000 X g in an air centrifuge. Membrane pellets were solubilized in Laemmli buffer (47) and subjected to SDS-PAGE. Following electrophoretic transfer to Immobilon by standard procedures, biotinylated plasma membrane proteins were visualized by Western blotting with "'1-streptavidin. The amount of bound streptavidin was quantitated either by counting the individual lanes in a y-counter or by laser densitometry of autoradiographs. Molecular mass markers are: myosin, 200 kDa; &galactosidase, 116 kDa; ovalbumin, 46 kDa.

6 0 ,001 .01 .1 1 10

SHPP (rnghl)

FIG. 3. Concentration-dependent inactivation of APDE-1 by sulfo-SHPP. Subconfluent CHO cells were treated for 30 min on ice with varying concentrations of sulfo-SHPP in either the presence or the absence of detergent. Reactions carried out in the absence of Triton X-100 (0) were terminated by washing the cells 15 min in PBS/C-M, 1 mg/ml lysine. The cells were then washed once and taken up in PBS/C-M, 0.2% Triton X-100, protease inhibitor mixture. Reactions carried out in the presence of PBS/C-M, 0.2% Triton X-100, protease inhibitor mixture (0) were terminated by adding a final concentration of 1 mg/ml lysine. Cells were scraped into the Triton solution, disrupted by 10 passes through a 22-gauge needle, and immediately assayed for APDE-1 enzymatic activity. Experimental values were determined in triplicate, and the mean of the activity per mg of protein & S.E. is shown.

with sulfhydryl groups such as dithiothreitol, N-succinimidyl 3-(2-pyridyldithio)propionate, N-ethylmaleimide, and dimethyl 3,3'-dithiobispropionimidate (l l) , ' suggesting that APDE-1 is inactivated by forming a thioester with sulfo- NHS-biotin. The membrane impermeant agent sulfo-SHPP also inhibited APDE-1 in a concentration-dependent manner, with maximum inhibition being attained a t 1 mg/ml sulfo- SHPP (Fig. 3). In the absence of detergent, a maximum of 77% of the cell-associated APDE-1 was sensitive to inhibition by sulfo-SHPP, whereas in the presence of Triton X-100, APDE-1 was completely inactivated. Thus, selective inactivation of APDE-1 at the cell surface using membrane-impermeant agents such as sulfo-SHPP can be used to determine the proportion of cellular APDE-1 expressed at the cell surface. We have found that the proportion of APDE-1 expressed at the cell surface varies considerably among cultured cell lines' making an accurate determination of plasma membrane APDE-1 essential for establishing the yield of plasma membrane.

In a typical experiment with CHO cells, 77% of the cell- associated APDE-1 was sensitive to inactivation with sulfo- SHPP, i.e. accessible at the cell surface. Seventy-four percent of the cell-associated APDE-1 was present in the retained fraction obtained by HIMAC, indicating that as much as 97% of the accessible APDE-1 ((74%/77%) x 100) was present in the retained fraction. When plasma membranes were prepared following sulfo-SHPP inactivation of surface APDE-1, only one-fifth of the remaining intracellular APDE-1 activity (equal to 5% of the total cell-associated APDE-1 prior to inactivation) was present in the HIMAC retained fraction. Thus, APDE-1 originating from intracellular membranes ac- counted for 5 5% of APDE-1 in the retained fraction and indicated that the yield of plasma membrane was 92%.

Plasma membranes were isolated from a number of other cell lines to establish the generality of this method for membrane isolation. The plasma cell line MOPC 315P, selected for resistance to prednisone, expresses levels of APDE-1 which are 100-fold greater than the parent cell line (26), making it possible to examine the distribution of both APDE-

' D. Warnock and J. U. Baenziger, unpublished observation.

Plasma Membrane Lipid Mars and Composition in CHO Cells 10149

1 activity and APDE-1 protein. Using WGA/FeDex and HI- MAC 70% of the total cellular APDE-1 activity was recovered in the retained fraction from MOPC 315P. Quantitative West- ern blot analysis of the retained and nonretained fractions using an APDE-1-specific antibody (7) (Fig. 4) indicated that 76% of the APDE-1 was in the retained fraction. Sulfo-SHPP inactivated 87% of the APDE-1 enzymatic activity but did not alter the distribution of APDE-1 between the retained and nonretained fractions (Fig. 4). Based on the yield of sulfo- SHPP-sensitive APDE-1 activity, 82% of the plasma membrane was present in the retained fraction. The isolation of plasma membranes from other cells lines such as HepG2 and 3T3 was not as efficient as from CHO and MOPC 315P cells.' This may reflect more extensive cell-cell and/or cell-substrate contacts that may limit the accessibility of selective portions of the plasma membrane to WGA/FeDex. Treatment with sulfo-SHPP does not alter either the distribution or yield of APDE-1 in the membrane fraction obtained by HIMAC. The yield of sulfo-SHPP-sensitive APDE-1 activity provides an accurate estimate of plasma membrane.

Plasma Membranes Isolated by HIMAC Are Largely Free of Intracellular Organelles-The use of WGA/FeDex and HI- MAC to follow the movement of proteins and other membrane components into and out of the plasma membrane of cultured cells requires minimal contamination with intracellular organelles and endosomes. The data summarized in Fig. 1 show that this condition was met for lysosomes, mitochondria, and endoplasmic reticulum. The amount of galactosyltransferase activity in the retained fraction suggested some Golgi contamination. The extent of contamination with endoplasmic reticulum, Golgi, and endosomal membranes was further addressed by characterizing the synthesis and distribution of the fibronectin receptor ( ~ ~ 5 0 1 integrin) and the transferrin receptor.

Plasma membranes were prepared from CHO K1 cells expressing recombinant human (u5 integrin subunit (Fig. 5B). Following a 20-min labeling period with [:%]Met, the major portion of the integrin a5 subunit is present in the nonre-

Whole, N.R. , , Ret. , - + - + Sdfo SHPP

205

1 16 97 66 45 29

FIG. 4. Inactivation of APDE-1 does not alter its cellular distribution. The effect of sulfo-SHPP treatment on the isolation of plasma membranes by HIMAC was examined using MOPC 315P cells, which express high levels of APDE-1. Cells were harvested from two TlOO flasks, pelleted, and washed two times in PBS/C-M. Equal aliquots of cells were incubated in the presence (1 mg/ml) or absence of sulfo-SHPP a t 4 "C in PBS/C-M, 0.5% Me2S0. Plasma membranes were isolated by HIMAC as described under "Experimental Proce- dures." Aliquots of HIMAC retained (Ret.) and nonretained (N.R.) membranes were analyzed by SDS-PAGE. The amount of APDE-1 in the retained and nonretained fractions was determined by Western blot analysis using a rabbit antibody to APDE-1, as described by Rebbe et al. (7). Molecular mass standards are: rabbit myosin, 205 kDa; @-galactosidase, 116 kDa; phosphorylase b, 97 kDa; bovine albumin, 66 kDa; ovalbumin, 45 kDa; and carbonic anhydrase, 29 kDa.

NON RETAINED RETAINED

A 1 2 3 4 5 6 7 8 91011 12

-alphas

-me

92.5 -

r 9,-

20c - a 5

-pre a5

- PI 0 20406080100 0 20406080100

FIG. 5. Intracellular and surface distribution of endogenous integrin @ subunits and newly synthesized recombinant integrin a5 expressed in CHO cells. Subconfluent CHO K1 cells, wild type or expressing recombinant human integrin (u5, were pulse-labeled with Tran:'"S-label-labeled methionine and chased in complete medium for the times indicated below the figure. Plasma membranes were then isolated by HIMAC. Aliquots of nonretained and retained fractions from each time point were pelleted for 2 h a t 190,000 X gave. Pellets were solubilized and subjected to immunoprecipitation of the fibronectin subunits as described in Roberts et al. (18) except that sequential immunoprecipitations were performed on each sample as described under "Experimental Procedures." Panel A, preintegrin (31 immunoprecipitated with antibody 66 from HIMAC membrane fractions of untransfected CHO K1 cells. Panel R, integrin a5 immunoprecipitated with antibody 33 from nonretained and retained membrane fractions of CHO K1 cells overexpressing human integrin a5.

tained fraction during the first 40 min of chase. By 60 min of chase both the recombinant a5 and the endogenous D l subunits are found predominantly in the retained fraction. The high levels of recombinant a 5 produced in these cells allows the endogenous D l subunits to rapidly form a561 dimers and reach the plasma membrane. The amount of a 5 found in the retained fraction a t 0 and 20 min of chase is indicative of the low level of contamination with intracellular membranes.

Quantitation of the pulse-chase experiments shown in Fig. 5 by laser densitometry indicated that the plasma membrane fraction (retained fraction) contained no more than 13 f 3% of precursor D l at any of the times examined (Fig. 5 A ) . In addition, the half-time for transport of newly synthesized integrin (u5 subunit to the plasma membrane was estimated by linear regression analysis to be 45 min for both endogenous a5 and recombinant human a5. The level of contamination of the plasma membrane preparations with intracellular membranes is therefore low, and the yield of plasma membranes is sufficiently high to characterize readily the transport of newly synthesized proteins to the cell surface.

The steady-state distributions of the a5 and D l integrin subunits were determined by quantitative immunoprecipitation" from the retained and nonretained fractions after labeling CHO K1 cells overnight with [""S]methionine. The plasma membrane fraction obtained by HIMAC contained 69 * 8% of the a5 subunit and 74 * 8% of the D l subunit (mean of four separate experiments). The yields of (u5 and D l were similar to that obtained for surface APDE-1, indicating that virtually all of the a561 integrin resides at the cell surface. This agrees well with previous estimates that 80% of the (u5pl integrin resides at the plasma membrane of CHO cells (27, 28).

The levels of contamination with endosomal membranes

D. Warnock and T. Roberts, unpublished observation.


were addressed by examining the distribution of a fluid phase marker, horseradish peroxidase. After CHO cells were allowed to take up horseradish peroxidase by fluid phase pinocytosis, < 10% of the internalized horseradish peroxidase was present in the retained fraction (not shown). Since damage to endosomes could result in release of soluble horseradish peroxidase and an artificially low estimate of the amount of endosomes present in the retained fraction, we also examined the distribution of the transferrin receptor.

The transferrin receptor is constitutively internalized from the plasma membrane to early endosomes through the coated pits and recycled to the plasma membrane (29). As a result 30-40% of the transferrin receptor is found at the cell surface while the remainder is present in intracellular membranes (30). Quantitative Western blot analysis was used to determine the amount of transferrin receptor in the HIMAC retained and nonretained fractions (Fig. 6). Thirty percent of the transferrin receptor was present in the retained fraction under conditions in which 75% of the plasma membrane was isolated as calculated from the recovery of the sulfo-SHPP- sensitive APDE-1. At most 40% of the transferrin receptor can be localized to the cell surface based on isolation of the surface membrane using WGA/FeDex and HIMAC, in good agreement with previous estimates using different methods (30). The proportion of transferrin receptor in the plasma membrane fraction stands in marked contrast to that of the fibronectin receptor above. The distribution of the transferrin receptor, like the yield of horseradish peroxidase, indicates that plasma membranes prepared from CHO cells using WGA/FeDex and HIMAC do not contain significant amounts of endosomal membranes.

This conclusion was further substantiated by determining the amount of endosomal transferrin receptor in the retained fraction prepared by HIMAC using a modification of the approach developed by Rodriguez-Boulan and co-workers (8). Transferrin receptor at the cell surface was labeled with NHS- disulfobiotin. After allowing the disulfobiotin-labeled transferrin receptor to be internalized, biotin present on transferrin receptor remaining at the cell surface was removed by treatment with glutathione and plasma membranes prepared by HIMAC. The amount of biotinylated transferrin receptor in the retained and nonretained fractions was determined by affinity chromatography on avidin-agarose and quantitative Western blot analysis using a monoclonal antibody specific for the transferrin receptor (Fig. 7). The plasma membrane fraction (retained fraction) contained 10.4 k 4.9% of endosomal, i.e. internalized, transferrin receptor (average of four

N.R. Ret. "

Sulfo SHPP - -k - -k

cl)mmclr FIG. 6. Transferrin receptor distribution as determined by

HIMAC. Plated CHO cells were washed two times with PBS/C-M and then incubated for 30 min on ice with PBS/C-M, 0.5% Me2S0 (sulfo-SHPP -) or with PBS/C-M, 0.5% Me2S0, 1 mg/ml sulfo- S H P P (sulfo-SHPP +). Cells were then incubated for 15 min in PBS/ C-M, 1 mg/ml lysine. After washing once with PBS/C-M, plasma membranes were isolated using WGA/FeDex and HIMAC, as described. Equal aliquots of HIMAC retained (Ret . ) and nonretained (N.R.) fractions were pelleted as described in Fig. 5, and the amount of transferrin receptor was determined by Western blot analysis using the mAb H68 (30) to the human transferrin receptor. Bound H68 was visualized using goat anti-mouse antibody conjugated to horseradish peroxidase and reaction through the luminol detection system.

1 2 3 4 5 6 7 8 9 1 0

. m 0 - m

NR Ret NR Ret NR Ret NR Ret NR Ret Biotin - + + + + Glutathione - - + - + Temperature 4'C 4'C 4'C 37'C 37'C

FIG. 7. Endosomal transferrin receptor does not co-isolate with plasma membrane. CHO cells were washed two times with PBS/C-M and then labeled three times (10 min) with PBS/C-M, 0.5% Me2S0, 1 mg/ml NHS-disulfobiotin a t 4 or 37 "C. Labeling was terminated by washing the cells a t 4 "C for 15 min in PBS/C-M, 50 mM NH,Cl. Cells were then incubated for 30 min at 4 "C in complete culture medium k 50 mM glutathione to strip plasma membrane biotin label. After washing once with PBS/C-M, plasma membranes were isolated using WGA/FeDex and HIMAC, as described. Equal aliquots of HIMAC retained (Ret ) and nonretained ( N R ) fractions were pelleted as described in Fig. 5. Membrane pellets were resuspended in immunoprecipitation buffer. Aliquots of nonretained and retained membrane protein were precipitated by avidin-agarose and washed three times with PBS, 1% Triton X-100 before 10% SDS- PAGE separation. Proteins were transferred to Immobilon-P, and Western blot analysis for the transferrin receptor was performed as described in Fig. 6. The amount of transferrin receptor was quantitated by laser densitometry. Odd-numbered lunes are nonretained, and even numbered lanes are retained fractions. Lanes I and 2, no biotin label; lunes 3 and 4 , 4 "C biotin label; lanes 5 and 6,4 "C biotin label, GSH strip; lunes 7 and 8, 37 "C biotin label; lunes 9 and 10, 37 "C biotin label, GSH strip.

separate determinations). We conclude that 510% of the endosomes present in CHO cells are found in the plasma membrane fraction prepared by HIMAC.

Distribution of Lipids in CHO Cells-The high yield and purity of plasma membranes prepared using WGA/FeDex and HIMAC suggested this would be an ideal method to prepare membranes for determination of their lipid composition. Chemical analysis indicated that the plasma membrane fraction contained 45% of total phospholipid, 64% of total sphingomyelin, and 58% of total cholesterol from CHO cells (Table I). When corrected for the yield of accessible APDE, these values increased to 49% for total phospholipid, 69% for total sphingomyelin, and 64% for total cholesterol. The ratio of cholesterol to phospholipid and sphingomyelin to phospholipid increased 1.6- and 2.2-fold, respectively, in the retained plasma membrane fraction as compared with whole cells (Table I). These data indicate that the plasma membrane of CHO cells represents a major fraction of total membrane bilayer and that the bulk of cholesterol and sphingomyelin resides in the plasma membrane.

Metabolic incorporation of ["Hlpalmitate makes it possible to examine simultaneously the distribution of sphingomyelin, phospholipids, and glycosphingolipids. After metabolic labeling of CHO cells with ["Hlpalmitate, 62% of ["Hlsphingo- myelin was present in the plasma membrane fraction (Fig. 8), a value nearly identical to that obtained by chemical analysis (Table I). Similarly, 43% of the "H-labeled total phospholipids were in the retained fraction as compared with 45% determined by chemical analysis (Table I). Phosphatidylserine was the only glycerophospholipid highly enriched (65%) in the retained fraction. Only 43% of phosphatidylethanolamine was present in the retained fraction, whereas even lower percentages of phosphatidylcholine, the major phospholipid, and phosphatidylinositol were in the retained fraction (Fig. 8).

CHO cells contain a simple pattern of glycosphingolipids (31, 32), namely GlcCer, LacCer, and GM:,. Following metabolic incorporation of ["Hlpalmitate, GM:, and LacCer were highly enriched in the retained fraction (Fig. 8). In striking


TABLE I Distribution of lipids

The abbreviations used are: SM, sphingomyelin; CHOL, cholesterol; PL, phospholipid. In CHO cells

SM CHOL PL Ratio CHOL/PL Ratio CHOL/SM Ratio SM/PL fmol/cell"

1.9 10.8 30.0 0.36 5.7 0.063 In nonretained (NR) and retained (Ret) membrane

APDE-1 SM CHOL P L Ratio CHOL/PL Ratio CHOL/SM Ratio SM/PL

NR Ret NR Ret NR Ret

% Retainedb Mean 70 64.3 57.8 44.9 0.33 0.59' 5.9 4.3 0.064 0.14d S.E.' (4.8) (4.5) (4.0) (5.0) (0.02) (0.09) (1.3) (0.72) (0.02) (0.006)

Mean of two separate determinations. * % retained = Ret/(NR + Ret) X 100, where NR is the nonretained fraction and RET is the retained fraction. Greater than 90% of all

E The ratio of cholesterol to total phospholipid in the retained fraction was significantly greater than that of the nonretained fraction; p <

The ratio of sphingomyelin to total phospholipid in the retained fraction was significantly greater than that of the nonretained fraction;

markers in the postnuclear supernatant were recovered in nonretained and retained fractions following HIMAC.

0.05.

p < 0.01. e Mean and standard error of five separate determinations.

100 I _ _ _ _ ~

I 1 I i

SM PC PS PI PE Total Cer Glc Lac GM, APDEl PL Cer Cer

FIG. 8. Distribution of [3H]palmitate-labeled lipids in the plasma membrane. Subconfluent CHO cells were labeled for 3 h with [3H]palmitate in complete medium. After washing twice with binding buffer, plasma membranes were isolated using WGA/FeDex and HIMAC, as described. Equal aliquots of HIMAC retained and nonretained fractions were pelleted as described in Fig. 5. The phospholipids and the glycolipids were analyzed by thin layer chromatography as described under "Experimental Procedures." Percent retained cpm divided by total recovered cpm values from four separate experiments are presented as the mean & S.E. Total phospholipid is the sum of the five major phospholipids (sphingomyelin ( S M ) , phosphatidylcholine (PC), phosphatidylserine ( P S ) , phosphatidylinositol ( P I ) , and phosphatidylethanolamine ( P E ) ) .

contrast, only 40% of GlcCer was present in the retained fraction (Fig. 8). Ceramide, the biosynthetic precursor for both sphingomyelin and the glycosphingolipids, was the least plentiful lipid in the plasma membrane fraction with only 23% in the retained fraction (Fig. 8).

DISCUSSION

Investigations of the plasma membrane using cultured cells have been hampered by the difficulty of obtaining highly enriched plasma membranes in good yield. This has been particularly troublesome for the analysis of lipids. Conclu- sions about the composition and distribution of lipid components in the plasma membrane have in some instances been questioned because of uncertainties about the levels of contamination with endosomal membranes and other intracellular organelles ( 2 ) . In this study we provide evidence that the combination of WGA/FeDex and HIMAC is able to overcome these difficulties, providing an effective method for the analysis of both protein and lipid components of the plasma membrane. We prepared plasma membrane fractions from cultured

cells such as CHO with yields in the range of 80-90% and with low levels of contamination from intracellular membranes including endosomes, endoplasmic reticulum, lysosomes, mitochondria, and Golgi.

The yield and purity of plasma membranes prepared by HIMAC were verified by independent criteria. Although APDE-1 is frequently used as a plasma membrane marker, the proportion of APDE-1 present at the plasma membrane can vary considerably among different cells and under different growth conditions.* The proportion of cell-associated APDE-1 expressed at the cell surface can be calculated accurately from its inactivation by membrane-impermeant agents such as sulfo-SHPP. Since only APDE-1 accessible at the cell surface is inactivated by SHPP, the yield of APDE-1 sensitive to inactivation by sulfo-SHPP is indicative of the yield of accessible surface membrane. Yields of plasma membrane from CHO, HepGP, and 3T3 cells were higher from subconfluent, as compared with confluent, cells. The more extensive cell-cell and cell-matrix contacts that develop at confluence may limit the accessibility of selective regions of the plasma membrane to WGA/FeDex but not to SHPP. Plasma membranes can be also isolated from cells grown in suspension. Variations in the yield of plasma membrane do not significantly alter the fold enrichment or the levels of contamination with intracellular membranes of these preparations.

The a5Dl integrin is largely confined to the plasma membrane (27, 28). During synthesis the a5pl integrin undergoes a number of post-translational modifications making it possible to identify those components that are in the endoplasmic reticulum and Golgi. The transport of newly synthesized a501 integrin to the cell surface was examined to determine if HIMAC could be used to establish the kinetics of transport to the plasma membrane and the extent to which endoplasmic reticulum and Golgi membranes were present in the HIMAC retained fraction. Even in the presence of large pools of intracellular precursors, it was possible to detect the arrival of the a5 and P l subunits at the cell surface readily. In addition, little of the synthetic precursors were detected in the retained fraction, indicating low levels of contamination with either endoplasmic reticulum or Golgi. Similarly, based on the yield of intracellular transferrin receptor (Fig. 7) plasma membranes prepared by HIMAC are free of endosomes. Based on both enzyme markers and the yield of en-


dogenous receptors, plasma membranes prepared by HIMAC are highly enriched and free of endoplasmic reticulum, Golgi, mitochondria, and endosomes.

Plasma membranes isolated using WGA/FeDex and HI- MAC are particularly suited to the characterization of lipid components and their trafficking to and from the plasma membrane. Earlier studies used isolation procedures that provided yields of 4-20% for the plasma membrane (3, 33- 35). The total plasma membrane lipid content is determined in these studies by multiplying the molar ratios of individual lipid components in the isolated plasma membrane fraction by factors ranging from 5 to 25 to correct for yield. The high yields of plasma membrane obtained with WGA/FeDex and HIMAC require the use of correction factors of c1.3. The ratios of cholesterol, sphingomyelin, and total phospholipid for the cells, retained, and nonretained fractions were calculated to permit comparison with these previous studies (Table 1).

We obtained a molar ratio of cholesterol to total phospholipid in whole CHO cells of 0.36 (Table I), a value identical to that reported for human fibroblasts (3). This ratio increased an average of 1.6-fold to 0.59 in the retained fractions, indicating enrichment of cholesterol in the retained fraction rel- ative to total phospholipid. In two experiments this ratio reached 0.8 in the retained fraction, a value equal to the maximum reported for plasma membranes purified by other techniques (3, 36). Lange et al. (3) found that the cholesterol to phospholipid ratio varies widely with cell density. Differ- ences in cell density may account for the variations in this ratio which we observed among different experiments (Table I). The ratio of sphingomyelin to total phospholipid increased an average of 2.2-fold, from 0.063 to 0.14, in the retained fraction as compared with the total homogenate of CHO cells (Table I), indicating that sphingomyelin is also enriched in the plasma membrane. The ratio of cholesterol to sphingomyelin was reduced consistently in the retained fraction when compared with whole cells or the nonretained fraction (Table I), suggesting that sphingomyelin is more highly enriched than cholesterol in the plasma membrane.

When the lipid values obtained by chemical analysis of plasma membranes purified using WGA/FeDex and HIMAC were corrected for the yield of accessible APDE, approximately 69% of sphingomyeIin, 64% of cholesterol, and 49% of total phospholipid were found in the plasma membrane. These results indicate that in CHO cells the plasma membrane represents a major fraction of' the total membrane bilayer and that sphingomyelin and cholesterol are primarily found in the plasma membrane. Lange et al. (3) reported that approximately 90% of sphingomyelin and cholesterol and 50% of total phospholipids are present in the plasma membrane of human fibroblasts (3). Based on susceptibility to cholesterol oxidase Lange and Ramos (37) estimated that 92% of the total cellular cholesterol of CHO cells resides in the plasma membrane. Our estimates of plasma membrane sphingomyelin and cholesterol content are 15-25% lower than those of Lange et al. (3, 37). The lower yield of sphingomyelin and cholesterol in plasma membranes prepared by HIMAC may reflect more efficient separation of endosomes from plasma membranes.

The results of metabolic labeling with [3H]palmitate were consistent with the data obtained by chemical assay. The retained fraction contained 62% of the ['H]sphingomyelin as compared with 64% determined by chemical analysis and 43% of the total 3H-phospholipids as compared with 45% by chemical analysis (compare Fig. 8 with Table I). Phosphatidylserine was the only glycerophospholipid highly enriched in the . ~~ ~ Y " . "

plasma membrane, in agreement with earlier reports (for review see Ref. 1). The results obtained by metabolic incorporation of [3H]palmitate can therefore be compared directly with those obtained by chemical analysis of lipid components.

The subcellular distribution of glycosphingolipids and the intracellular locations of the glycosyltransferases responsible for their synthesis have only recently been examined (for review see Ref. 1). Information about the intracellular distribution of intermediates involved in synthesis of glycosphingolipids such as G M ~ may provide insights about the function of these glycolipids as well as the regulation of their synthesis and transport through the cell. Recent evidence suggests that glucose is added to ceramide on the cytosolic surface of the Golgi (38-41), whereas galactose and sialic acid are added to GlcCer and LacCer, respectively, within the Golgi lumen. LacCer and GM3 are thought to move then to the plasma membrane via vesicular traffic using the same pathway and machinery as glycoproteins destined for the cell surface (for review see Ref. 42). The neutral glycosphingolipids LacCer and Forssman have been localized to intracellular membranes in human neutrophils (43) and Madin-Darby canine kidney epithelial cells (44), respectively, raising the possibility that the distribution of glycolipids may not be the same in all cell types.

We found that ceramide and its glycosylated derivatives were not represented equally in the plasma membrane fractions (Fig. 8). The same percentages of G M ~ and LacCer, 58 and 61%, respectively, were in the retained plasma membrane fraction as sphingomyelin and phosphatidylserine. When corrected for the yield of APDE-I, this indicated 64% of GMR and 68% of LacCer were at the plasma membrane. In contrast only 40% of GlcCer and 23% of ceramide, the synthetic precursors of LacCer and GM3, were in the retained fraction. Contamination with intracellular membranes could account for one-third to one-half of the ceramide found in the plasma membrane, indicating that as little as 10-15% of the total ceramide in the cell may be in the plasma membrane of CHO cells. The plasma membrane content of endogenous ceramide and GlcCer has not been addressed previously. The distribution of GlcCer in the plasma membrane differs from that of LacCer and GMs, possibly reflecting conversion of GlcCer to LacCer and GM3 as it passes through the Golgi. Alternatively, GlcCer, which is synthesized on the cytosolic surface of the Golgi, may reach the plasma membrane by a different mech- anism than LacCer and GM3, resulting in a different distribution.

The low levels of ceramide in the plasma membrane are particularly striking. Recently, ceramide has been proposed to be an intracellular mediator in the regulation of cell prolif- eration and differentiation (45, 46). For example, in human leukemia HL-60 cells a neutral sphingomyelinase is activated by extracellular tumor necrosis factor-a, producing an increase in ceramide. The ceramide activates a membrane- associated protein kinase that has been implicated in the mode of action of tumor necrosis factor-a. If HL-60 cells have levels of ceramide in their plasma membrane similar to those we have observed in CHO cells and the increases in ceramide stimulated by tumor necrosis factor-a are confined to the plasma membrane, the increase in ceramide at the plasma membrane may be much greater than appreciated previously and would be consistent with its role as an activator of protein kinase.

The isolation of highly enriched plasma membranes in high yield using WGA/FeDex in conjunction with HIMAC has provided new information about the lipid constituents of the olasma membrane. Sohinnomvelin, cholesterol, phosphatidyl-

Plasma Membrane Lipid Mass and Composition in CHO Cells 10153

serine, LacCer, and GM3 are found predominantly in the plasma membrane. Furthermore, half of the total cellular phospholipid is present in the plasma membranes of CHO cells, indicating that a major fraction of the lipid bilayer is represented by the plasma membrane. Finally, even though synthetic precursors to G M ~ , i.e. ceramide and GlcCer, are found in the plasma membrane they are present predominantly on intracellular membranes. The efficiency, rapidity, and simplicity of plasma membrane isolation using HIMAC will make it possible to address such issues as the kinetics of transport of newly synthesized glycoproteins and glycolipids to the cell surface. Such studies will provide insights into the trafficking of lipid components to and from the plasma membrane.

Acknowledgment-We thank James Blomquist for suggestions.

REFERENCES 1. Van Meer, G. (1989) Annu. Reu. Cell Biol. 5 , 247-275

3. Lange, Y., Swaisgood, M. H., Ramos, B. V., and Steck, T. L. (1989) J. Biol. 2. Van Meer, G . (1987) Trends Biochem. Sci. 12 , 375-376

4. Thomuson. T. E.. and Tillack. T. W. (1985) Annu. Reu. Bioohvs. Bioohvs. Chem. 264 , 3786-3793

Chim. 14 , 361-386 I " L _

6. Becich, M. J., Mahklouf, S., and Baenziger, J. U. (1991) Eur. J . Cell Biol. 5. Becich, M. J., and Baenziger, J. U. (1991) Eur. J. Cell Biol. 5 5 , 71-82

65.83-93 7. Rebbe, N.~F. , Tong, B. D., Finley, E. M., and Hickman, S. (1991) Proc.

8. Le Bivic, A., Sambuy, Y., Mostov, K., and Rodriguez-Boulan, E. (1990) J .

9. Thompson, J. A., Lau, A. L., and Cunningham, D. D. (1987) Biochemistry Cell Biol. 1 1 0 , 1533-1539

2 6 , 743-750 10. Smith, P. K., Krohn, R. I., Hermanson, G. T., Mallic, A. K., Gartner, F.

H., Provenzano, M. D., Fujimoto, E. K., Goeke, N. M., Olson, B. J., and Klenk, D. C. (1985) Anal. Biochem. 1 5 0 , 76-85

Natl. A d . S C ~ . Li. S. A. 8 8 , 5192-5196

11. Razzell, W. E. (1963) Methods Enzymol. 6 , 236-258 12. Phillips, A. H., and Langdon, R. G. (1962) J. Bid. Chem. 2 3 7 , 2652-2660 13. Barrett, A. J., and Heath, M. F. (1977) in Lysosomes: A Laboratory Hand-

book (Dingle, J. T., ed) pp. 19-145, Elsevier/North Holland Inc., New York

14. Reitman, M. L., Trowbridge, I. S., and Kornfeld, S. (1982) J. Biol. Chem. 257 , 10357-10362

15. Burns D. M., and Touster, 0. (1982) J . Bid. Chem. 257,9991-10000 16. Lau, i. T. Y., and Carlson, D. M. (1981) J. Bid. Chem. 256 , 7142-7145 17. Plummer. T. H.. Elder. J. H.. Alexander. S.. Phelan. A. W.. and Tarentino. ~ ~~

A.LT~(l984j-J.Biol. Chem: 259 , 10700-10704 ~~'

Yamada. S. S.. Chen, W.-T., Yamada. K. M., and McDonall J. A'. (i988j 18. Roberts, C. J., Birkenmeier, T. M., McQuillan, J. J., Aki ama S K

J. Biol. Chem.'263,4586-4592 19. Burnette, W. N. (1981) Anal. Biochem. 112, 195-203 20. Gerlach, E., and Deuticke, B. (1963) Biochem. Z. 337,477-479 21. Marshall-Stewart, J. C. (1980) Anal. Biochem. 104,lO-14 22. Esko J. D. and Raetz, C. R. H. (1980) J. Biol. Chem. 255,4474-4480 23. Saitd, T., i nd Hakomori, S. (1971) J. Lipid Res. 1 2 , 257-259 24. Merrill, A. H., Jr., and Wan E. (1986) J. Biol. Chem. 261,3764-3769 25. Howell, K. E., and Palade, 8: 26. Hickman, S., Wong-Yip, Y. P., Rebbe, N. F., and Greco, J. M. (1985) J.

E. (1982) J. Cell Biol. 92,822-832

27. Bretscher M. S. (1989) EMBO J. 8 , 1341-1348 28. Raub, T. b., and Kuentzel, S. L. (1989) Exp. Cell Res. 184,407-426 29. May, W. S., Jr., and Cuatrecasas, P. (1985) J . Membr. Biol. 8 8 , 205-215 30. Collawn, J. F., Stangel, M., Kuhn, L. A,, Esokogwu V., Jing, S., Trowbridge,

31. Briles, E. B., Li, E., and Kornfeld, S. (1977) J . Biol. Chem. 252 , 1107-1116 32. Yo eeswaran, G., Murray, R. K., and Wright, J. A. (1974) Biochem. Biophys.

33. Renkonen, O., Gahmberg, C. G., Simons, K., and Kaarianen, L. (1972)

34. Kaplan, M. R., and Simoni, R. D. (1985) J. Cell Biol. 101,441-445 35. Morre D. J. and Morre, D. M. (1989) BioTechni ues 7,946-958 36. KobaGashi *. and Pagano R. E. (1989) J. Biol. 8hem. 264,5966-5973 37. Lange Y. 'and Ramos, B. \i. (1983) J . Bid. Chem. 2 5 8 , 15130-15134 38. TrincBera', M., Fabbri, M., and Ghidoni, R. (1991) J. Biol. Chem. 2 6 6 ,

39. Co_ste,-H., Martel, M. B., and Got, R. (1986) Biochim. Biophys. Acta 8 5 8 ,

Biol. Chem. 260,6098-6106

I. S., and Tainer, J. A. (1990) Cell 6 3 , 1061-1Oh

i e s . Commun. 5 6 , 1010-1016

Biochim. Biophys. Acta 255,66-78

20907-20912

40. Futerman A. H. and Pagano, R. E. (1991) Biochem. J . 280 , 295-302 41. Jeckel, D.: Karrdnbauer, A., Burger, K. N. J., Van Meer, G., and Wieland,

42. Young, W. W., Jr. (1993) Adu. Lipid Res., in rem 43. Symington, F. W., Murray, W. A,, Bearman, g. I., and Hakomori, S. (1987)

44. van Genderen, I . L., Van, Meer, G., Slot, J. W., Geuze, H. J., and Voorbout,

45. Bielawska, A,, Linardic, C. M., and Hannun, Y. A. (1992) J. Biol. Chem.

46. Dressler, K. A., Mathias, S., and Kolesnick, R. N. (1992) Science 255 ,

b-12

F. (1992) J . Cell B~ol. 117 , 259-267

J. Bcol. Chem. 2 6 2 , 11356-11363

W. F. (1991) J. CellBcol. 115 , 1009-1019

267 , 18493-18497 171K-171Q

47. Laemmli, U. K. (1970) Nature 227,720-729 L I L Y AIL"

determination of plasma membrane lipid mass and composition in

Documents