Fat Cell Plasma Membranes I. PREPARATION, CHARACTERIZATION, AND CHEMICAL COMPOSITION*

Perirenal adipose tissue from rabbit, rat, and calf was disrupted without the use of proteases by a sieving procedure to yield fat cells from which plasma membranes were prepared. These membranes were isolated from the homogenate of the cells by differential and sucrose density gradient centrifugation. The major plasma membrane fraction, which represented the lightest band resolved from the microsomal pellet, was obtained from rabbit in a yield of about 4 mg/lOO g of tissue and was purified about l’l-fold in respect to 5’-nucleotidase activity. The membranes were rich in cholesterol and phospholipids (total lipid, 57% of membrane weight) and had high alkaline phosphatase activity. Only low levels of succinic dehydrogenase, NADPH-cytochrome c reductase, and nucleic acids were observed, indicating absence of significant contamination with intracellular components. The amino acid compositions of the plasma membranes from the three species were quite similar, with glutamic acid, leucine, and aspartic acid occurring as the major constituents. The carbohydrate of the membranes, however, showed some species variation. The total saccharide content of the rabbit membrane was 8.7 mg/lOO mg of protein and it was present as galactose, mannose, glucose, fucose, ribose, N-acetylneuraminic acid, glucosamine, and galactosamine in the molar ratios of 6.0:4.6:4.0:1.5:1.2:2.4:7.2:1.0, respectively. An assessment of the protein and glycoprotein subunit composition of the rabbit, rat, and calf plasma membranes was made by polyacrylamide gel electrophoresis in sodium dodecyl sulfate with differential staining. While substantial species differences were apparent in the polypeptide pattern, the three membranes were alike in containing a major glycoprotein component (periodic acid-Schiff-reactive) with an apparent molecular weight in the range of 74,000 to 79,000. The rat membrane contained an additional prominent glycoprotein band with an apparent molecular weight of 88,000. The most striking observation was the presence of the same major glycoprotein band(s) in all six membrane fractions obtained by density gradient centrifugation of the microsomal pellet. This finding may have a bearing on plasma membrane biogenesis in the fat cell.


Perirenal
adipose tissue from rabbit, rat, and calf was disrupted without the use of proteases by a sieving procedure to yield fat cells from which plasma membranes were prepared. These membranes were isolated from the homogenate of the cells by differential and sucrose density gradient centrifugation.
The major plasma membrane fraction, which represented the lightest band resolved from the microsomal pellet, was obtained from rabbit in a yield of about 4 mg/lOO g of tissue and was purified about l'l-fold in respect to 5'-nucleotidase activity. The membranes were rich in cholesterol and phospholipids (total lipid, 57% of membrane weight) and had high alkaline phosphatase activity. Only low levels of succinic dehydrogenase, NADPH-cytochrome c reductase, and nucleic acids were observed, indicating absence of significant contamination with intracellular components.
The amino acid compositions of the plasma membranes from the three species were quite similar, with glutamic acid, leucine, and aspartic acid occurring as the major constituents.
The carbohydrate of the membranes, however, showed some species variation. The total saccharide content of the rabbit membrane was 8.7 mg/lOO mg of protein and it was present as galactose, mannose, glucose, fucose, ribose, N-acetylneuraminic acid, glucosamine, and galactosamine in the molar ratios of 6.0:4.6:4.0:1.5:1.2:2.4:7.2:1.0, respectively. An assessment of the protein and glycoprotein subunit composition of the rabbit, rat, and calf plasma membranes was made by polyacrylamide gel electrophoresis in sodium dodecyl sulfate with differential staining. While substantial species differences were apparent in the polypeptide pattern, the three membranes were alike in containing a major glycoprotein component (periodic acid-Schiff-reactive) with an apparent molecular weight in the range of 74,000 to 79,000. The rat membrane contained an additional prominent glycoprotein band with an apparent molecular weight of 88,000. The most striking observation was the presence of the same major glycoprotein band(s) in all six membrane fractions obtained by density gradient centrifugation of the microsomal pellet. This finding may have a bearing on plasma membrane biogenesis in the fat cell. years that a number of biologically active molecules including hormones, lectins, and toxins can influence the metabolism of the fat cell by interaction with components in its plasma membrane (1,2). An understanding of these intermolecular events depends to a large measure on a description of the chemical architecture of the cell surface, including the identification and structural analysis of the membrane components directly or indirectly involved. Indeed, the work of Cuatrecasas has already led to the solubilization and isolation of a specific insulin-binding protein from membranes of the adipocyte (3).
Methods hitherto reported for isolating plasma membranes from fat cells (4,5) have utilized as the initial step the procedure of Rodbell (6) in which the adipose tissue is digested with crude collagenase. Since this enzyme preparation is known to contain a complex mixture of proteases (7,8) and may also contain glycosidases (9) it would appear to be unsuitable for use in preparing membranes for analytical and structural investigations.
This view is emphasized by reports that treatment of isolated fat cells with proteases or neuraminidase leads to cell surface changes which are translated into altered metabolic activities and hormonal responsiveness of the cell (10, 11).
Since it was the aim of the present study to undertake a detailed examination of the chemistry of the surface components of fat cells, a procedure for preparing plasma membranes was employed which, by relying entirely on mechanical disruption of the tissue, minimized the possibility of enzymatic degradation.
The composition and enzymatic properties of membranes thus obtained from perirenal fat of rabbit, rat, and calf were determined and their protein and glycoprotein subunits were identified by polyacrylamide gel electrophoresis in sodium dodecyl sulfate.  (20). The RNA content of the supernatant was determined by the orcinol reaction for ribose (21) and correction for any hexose contribution to this reaction was made on the basis of measurements with the anthrone procedure (22)  were resolved and separately determined on a Technicon NC-2 amino acid analyzer, with a program which utilized pH 5 buffers (261, after hydrolysis of the membranes for 15 h in constant boiling HCl in sealed tubes under nitrogen at 100". Sialic acid was released from the membrane by hydrolysis with 0.1 N H,SO, at 80" for 1 h and was determined by the thiobarbituric acid assay of Warren (27) after its separation on Dowex 1 (formate columns (22). Hexuronic acids were measured by the carbazole reaction of Dische (28). In order to detect small amounts of these sugars without interference from hexoses, samples were hydrolyzed in 1 N HCl at 100" for 6 h. The hydrolysates were passed through coupled columns of Dowex 50 (H+) and Dowex 1 (formate) and after washing with water, hexuronic acid was eluted with 0.5 N formic acid from the Dowex 1 column. The acid was removed from the eluate by lyophilization prior to assay by the carbazole reaction and paper chromatography.
Sulfate was determined by the benzidine method of Dodgson and Spencer (29)  The apparent molecular weights of the protein bands were determined from the migration of standard proteins, which included myosin, phosphorylase a, bovine serum albumin, ovalbumin, chymotrypsinogen, and myoglobin (36).

Isolation of Membranes from Rabbit Adipose
Tissue -Examination under the phase microscope of the material obtained by flotation of the sieved adipose tissue indicated that it consisted of intact adipocytes, some broken fat cells and vesicles probably derived from the broken cell fragments. Connective tissue, blood vessels, and tissue fragments were entirely left on the sieves and stromal cells were found in the sediment obtained during centrifugation of the sieved tissue. The Polytron homogenizer was effective in disrupting the fat cells, but the power setting of the instrument and the temperature of the medium substantially influenced the final yield and purity of the plasma membranes. The conditions of homogenization adopted included Polytron treatment at a low power setting for a brief interval in a medium sufficiently warm to soften the fat inside the cells. Different temperatures had to be selected for homogenization of fat cells from the three species studied, presumably reflecting differences in the fatty acid composition of the triglycerides.
Marker enzymes were assayed in the four subcellular fractions (N, M, P, and S) into which the homogenate was separated by differential centrifugation.
5'-Nucleotidase, which is believed to be a plasma membrane-bound enzyme, and NADPH-cytochrome c reductase, which is primarily a marker for endoplasmic reticulum, showed the highest specific activities in the P fraction, while the mitochondrial enzyme succinic dehydrogenase had the greatest specific activity in the M pellet.
Since the P pellet contained the largest amount of membrane-bound 5'-nucleotide activity, it was further submitted to a sucrose density gradient centrifugation which yielded six ' The abbreviation used is: SDS, sodium dodecyl sulfate.
of Fat Cell Plasma Membranes 6231 fractions (P, through P,) (Table I). P, was obtained as a distinct white band floating upon a light brown layer (P,) which was present at the interface between layers of densities 1.032 and 1.142. P3 and P, were faint bands which formed at the upper and lower interface, respectively, of the layer with a density of 1.161, while P, consisted of dark brown material which accumulated at the interface between layers with densities of 1.182 and 1.219, respectively. In addition, a small brownish pellet (P,J was formed at the bottom of the centrifuge tube while some congealed free fat floated on the uppermost layer of the gradient. Characterization of the fractions from sucrose density gradient centrifugation was performed by assays of marker enzymes and chemical components (Figs. 1 and 2). 5'-Nucleotidase activity was highest in the P, membranes but was low in the heavy fractions. Alkaline phosphatase, which is considered to be another enzyme associated with the plasma membrane, showed a similar distribution with most of the activity being localized in the lighter fractions from the density gradient. In contrast, succinic dehydrogenase and NADPH-cytochrome c reductase showed the highest specific activity in the heavy P, fraction. RNA, a marker for rough endoplasmic reticulum and free ribosomes was also recovered primarily in the heavy membranes (P, and P,) while cholesterol, which is usually associated with plasma membranes, was concentrated in the light membrane bands (Fig. 2). The distribution of lipid phosphorus paralleled closely that of cholesterol so that this component was concentrated in the P, fraction.
On the basis of these analyses it became apparent that the P, fraction consists primarily of plasma membranes. P, represented 28% of the total protein in the P pellet from the rabbit adipose tissue (Table I) and had 17-fold greater 5'-nucleotidase activity than the unfractionated fat cell homogenate. Electron microscopic examination of the P, fraction from Height of bars indicates specific activity of the enzyme in fraction, while width represents the per cent of the total recovered protein.
The values represent the average of three preparations. The recovery of protein from the P pellet in the density gradient was 107%, while that of 5'-nucleotidase activity was 120% (Table I). The specific activities of the unfractionated P pellet in nanomoles/min/ mg of protein for the other enzymes were: succinic dehydrogenase, 4.3; NADPH/cytochrome c reductase, 12.7; and alkaline phosphatase, 7.2, The recoveries of activity from the P pellet for these enzymes were 66, 79, and 86% respectively. The designations of the fractions are described in text and Table I. rabbit adipose tissue indicated that it consisted primarily of membranous vesicles and membrane sheets. Flask-shaped invaginations which have been shown to be characteristic of fat cell plasma membranes (39) were observed in some portions of the larger vesicles. In some fields occasional ribosomes attached to membranes were noted, which could account for the low RNA content of the P, fraction. Components originating from mitochondria were very rarely seen. Preparation of Membranes from Rat and Calf Adipose Tissue -For comparative purposes the procedure for isolating plasma membranes from the perirenal fat of rabbits was employed to prepare membranes from rat and calf. The procedure developed with rabbit tissue could be applied to these other two species without difficulty, and qualitatively similar patterns of membrane bands were obtained upon sucrose density gradient centrifugation of their P pellets. In rat and calf, as in rabbit, the P, fraction had the highest 5'-nucleotidase activity. This enzyme in P, was enriched over the homogenate &fold in the rat and 28-fold in the calf. The specific activity of the 5'nucleotidase in the P, fraction differed among the species, being higher in rat than in rabbit (567 nmol/mg/min) and lower in calf than in rabbit (37 nmol/mg/min).  D Values for the rabbit membrane are given as the mean -c standard deviation of the mean of three preparations; the rat and calf analyses are for a single membrane preparation each. ' -, galactosamine in the rat membrane was not determined but this sugar was identified by paper chromatography. ' Paper chromatography revealed that the sialic acid was present solely as N-acetylneuraminic acid.
Chemical Composition of Plasma Membranes -Analytical data for the chemical components of the plasma membrane fraction (P,) from rabbit are shown in Table II. The membranes contained a large amount of phospholipid, which was present in greater quantities than the cholesterol, so that the molar ratios of cholesterol to phospholipid was 0.33. The total phosphorus content of the P, membranes was 176 pmol/lOO mg of protein, of which 163 ~mol/lOO mg (93%) was accounted for by the phospholipid.
The remaining phosphorus of the membranes can largely be attributed to the small amounts of nucleic acids which are present.
The amino acid composition of the plasma membranes from rabbit as well as rat and calf adipose tissue was determined and found to be quite similar. The major components were glutamic acid, leucine, and aspartic acid, which were present as 97. 8, 90.7, and 86.7 wmol/lOO mg of membrane protein, respectively, in the rabbit. No hydroxyproline was detectable in the preparations, indicating the absence of basement membranes and collagens.
Ethanolamine, which originated from the acid hydrolysis of phosphatidylethanolamine, was measured on a Technicon amino acid analyzer where it emerged just prior to the ammonia peak and gave a molar color yield of 0.51 that of norleutine. From the value of this component (52. 6 4 (center). Polyacrylamide gel electrophoresis in SDS of membrane fractions obtained from rat fat cells by sucrose density gradient centrifugation.
The protein applied to the gels stained with Coomassie blue was 25 pg of P, and 50 pg of each of the other fractions (P, through P3. Twice as much protein was applied to the gels when the periodic acid-Schiff reagent was employed. The conditions of electrophoresis were the same as in Fig. 3. rabbit membrane protein) it was calculated that in the rabbit plasma membranes phosphatidylethanolamine constitutes 32% of the total phospholipids on a molar basis, which is comparable to the value of 29% reported for liver membrane (40).
The sugar composition of the fat cell plasma membranes from rabbit, rat, and calf are shown in Table III. The neutral sugars were identified by paper chromatography in Solvent System A as well as by borate complex anion exchange chromatography, while the amino sugars were distinguished by paper chromatography (Solvent System A) and on the amino acid analyzer. Paper chromatography of the sialic acids from the rabbit membrane was performed in Solvent System B and indicated that N-acetylneuraminic acid, but no detectable amount of N-glycolylneuraminic acid, was present. No significant amount of ester sulfate or uranic acid (less than 0.01%) was found in the rabbit P, fraction, indicating that proteoglycans are not components of these plasma membranes.

Polyacrylamide
Gel Electrophoresis of Membranes -In order to examine the protein and glycoprotein components of the fat cell membranes, polyacrylamide gel electrophoresis in SDS of various fractions obtained from the sucrose density gradient of the P pellet was performed. Electrophoresis of the rabbit membranes after solubilization with 2-mercaptoethanol revealed, upon staining with Coomassie blue, that the light fractions (PI, P2, and P3) had a similar and relatively simple peptide pattern (Fig. 3). Five major components ranging in apparent molecular weight from 35,000 to 85,000 were evident in addition to several minor bands and some material which failed to penetrate the 7.5% gel. A considerably more complex electrophoretic pattern was observed in the heavier membrane fractions which are enriched in mitochondria and endoplasmic reticulum; in P5 and Ps particularly, a large number of bands in the low molecular weight range were seen.
Upon staining duplicate gels with the periodic acid-Schiff reagent all fractions contained one major discrete glycoprotein band which moved to a position corresponding to a molecular weight of 79,000 and appeared to coincide with a prominent Coomassie blue-positive component. In the light membrane fractions, P, and P2, a broad band migrating just behind the tracking dye was observed after staining with the Schiff reagent (Fig. 3). This material which did not significantly react with Coomassie blue has been shown to be completely extractable with organic solvents (411 and probably represents cellsurface glycolipids which would be expected to be most prominent in the plasma membrane (P,) fraction.
When electrophoresis was performed on 5% gels rather than 7.5%, most of the material previously at the origin penetrated the gel slightly, consistent with molecular weights of greater than 250,000. Electrophoresis was, however, routinely performed in the 7.5% gels, as they gave the best separation of the components which were in the range of polyacrylamide gel resolution. No change in electrophoretic patterns was obtained when the membrane samples were heated in buffered SDS at 100" up to 30  The amount of protein applied to each gel was 50 pg when staining was accomplished with Coomassie blue and 100 pg when the periodic acid-Schiff reagent was used. The conditions of electrophoresis were the same as in Fig. 3.
of Fat Cell Plasma Membranes tissue membranes revealed a pattern distinct from that of the rabbit (Fig. 4). The light membrane fractions (P, through P3) when stained with Coomassie blue again presented only a few prominent bands. The two major Coomassie blue-reactive components had apparent molecular weights of 88,000 and 74,000 and corresponded in their migration to bands which stained with the periodic acid-Schiff reagent. These two glycoproteins appeared to be present in all the membrane fractions in a similar ratio to each other, although the fast migrating Schiff reagent-positive glycolipid material was essentially confined to P, and P,. As in the rabbit the heavier rat membrane fractions contained a large number of bands which stained with Coomassie blue and distributed over a wide range of molecular weights.
The electrophoretic patterns of the calf membranes were again different from the other two species (Fig. 5). Despite the presence of a large number of protein bands, particularly in the heavy fractions, only a single major glycoprotein band was observed which corresponded to an apparent molecular weight of 78,000.
The presence of the same major glycoprotein band(s) in all the membrane fractions obtained by density gradient centrifugation of the P peilet of a given species is a finding of some interest. Although the glycoprotein bands were most intense in the light plasma membrane fractions, their occurrence in the heavier membranes can not be attributed to cross-contamination in view of the enzymatic and chemical data (Figs. 1 and 2) and the observation that the periodic acid-Schiff-reactive glycolipid bands were not similarly distributed throughout the P fractions. Electrophoresis of the high speed supernatant (S) failed to reveal the presence of the same glycoprotein bands which were present in the membrane fractions, thereby excluding the possibility of nonselective binding of such cytosol components to the particles.

DISCUSSION
The present study has indicated that it is feasible to prepare plasma membranes from adipose tissue of several species without using a proteolytic digestion step. The procedure employed yielded a membrane preparation which on the basis of enzymatic, chemical, and morphological criteria appeared to be suitable for structural investigations.
Components of the adipose tissue, other than fat cells, were removed by the sieving technique as well as by the flotation step. The plasma membranes were obtained in the lightest fraction after density gradient centrifugation of the microsomal pellet from the fat cells. They were primarily vesicular in nature and were characterized by high 5'.nucleotidase and alkaline phosphatasespecific activities as well as by a large cholesterol content. Polyacrylamide gel electrophoresis, moreover, indicated that glycolipids which are believed to be primarily cell surface components (42) were concentrated in this membrane fraction. The low to negligible levels of enzymatic and chemical markers for mitochondria, endoplasmic reticulum, nuclei, and basement membranes in this light membrane fraction further indicated that it was primarily of plasma membrane origin. Since the adipocyte is not endowed with an elaborate endoplasmic reticulum (43), it might be expected that the possibility of contamination with membranes of this network would be less than in more complex cells.
While the lightest membrane band (P,) obtained by density gradient centrifugation in this study was considered to be the primary plasma membrane fraction, it should be noted that the underlying band (P,) which contained substantially less cholesterol and phospholipid had, nevertheless, fairly high specific activities for the plasma membrane enzyme markers. Moveover, P, and P, were essentially indistinguishable in terms of their polypeptide and glycoprotein components as seen by polyacrylamide gel electrophoresis and by amino acid and sugar analyses performed on the delipidated membranes (41). The occurrence of two plasma membrane fractions of distinct buoyant densities with different lipid contents has been previously observed in other cell types (44)(45)(46).
Although plasma membranes have been prepared by a number of investigators (4,5,47,48) from rat adipose tissue enzymatically dissociated by the procedure of Rodbell (61, no detailed analyses of the peptide and carbohydrate consitituents of these membranes have as yet been reported. The plasma membranes isolated by mechanical disruption of fat tissue in the present study were obtained in sufficient quantities from three species to permit complete amino acid and sugar determinations to be performed. The fat cell membranes appear to be quite similar in their amino acid composition to the plasma membranes from a number of other sources (40,44,45,49) in which glutamic acid, leucine, and aspartic acid are also the most abundant constituents. WhiIe the hexose, hexosamine, and siaiic acid components of the fat cell membranes belong to glycoproteins and glycolipids (41), the small amount of ribose can readily be accounted for by the RNA present. The membranes from rabbit, rat, and calf adipose tissue had similar hexosamine compositions, but showed distinct differences in their neutral sugars analyses. The high ratio of glucosamine to galactosamine (approximately 7:l) in the adipocyte membranes is similar to that observed in the plasma membranes of human platelets (45) and calf thymocytes (501, but differs from that reported from human erythrocyte (50) and pig lymphocyte membranes (51) in which these two hexosamines are present in more equivalent amounts. The membranes from the latter cells have also been found to be relatively rich in sialic acid and gala&se.
Because of their high phospholipid content the rabbit fat cell plasma membranes have a molar ratio of cholesterol to lipid phosphorus (0.33) which falls into the lower portion of the rather wide range (0.24 to 1.3) which has been reported for this value in the surface membranes of various cells (52, 53). The overall analyses of the rabbit fat cell plasma membrane indicated that it had a high total lipid content compared to other plasma membranes (49) with a protein to lipid ratio of 0.67. The total carbohydrate contents of the fat cell membranes fell into the range (2 to 10% of membrane weight) reported for other plasma membranes from mammalian cells (49). The presence of small amounts of RNA as noted in the rabbit membranes from the present study has been observed in plasma membranes from various other sources (521, including those isolated from rat fat cells by the procedure of McKee1 and Jarett (4).
Polyacrylamide gel electrophoresis of the membranes with differential staining permitted an assessment to be made of their protein and glycoprotein subunits. The plasma membrane fractions (P, as well as P,) of the fat cells from the three species examined had a rather simple pattern of polypeptide components when compared to the surface membranes of other cells (54) and to the heavier membrane of the fat cells themselves. Aside from the rapidly migrating glycolipid material, the periodic acid-Schiff reagent revealed only one major glycoprotein band in the rabbit and calf plasma membranes and two such components in the rat membranes, although additional glycoproteins which either do not enter the polyacrylamide gel or fail to react sufficiently with the Schiff reagent may be present. The small number of visible glycoprotein components in the fat cell membranes contrasts with the situation which prevails in the membranes of liver (54, 551, kidney (541, and placenta (56) as well as fibroblasts (57), in which a large number of Schiff-positive bands have been observed. The human erythrocyte membrane is more similar to the fat membranes in demonstrating one predominant periodic acid-schiffreactive band upon gel electrophoresis (38,54). A comparison of the electrophoretic patterns of the plasma membrane fractions obtained from rabbit, rat, and calf in this study indicates that substantial species differences exist among the polypeptide compents. The three species are alike, however, in containing a major Schiff-reactive glycoprotein subunit with an apparent molecular weight of 74,000 to 79,000. The rat membranes differed from the other two species in having an additional prominent glycoprotein band with a somewhat higher molecular weight (88,000). The occurrence of two such glycoprotein components has previously been observed in plasma membranes isolated from rat fat cells which were prepared by the enzymatic digestion procedure (47, 48). However, in the present study there was no evidence for a less prominent M, 62,000 glycoprotein band which has been observed in membranes from enzymatically treated rat fat cells (47). Furthermore, the larger number of Coomassie Bluestaining components, particularly in the Iower molecular weight range, observed upon electrophoresis of plasma membranes prepared from enzymatically dissociated cells (47, 48) suggests the possibility that some proteolytic degradation of cell surface proteins may occur during preparation of fat cells by the collagenase digestion procedure.
One of the more striking observations made on the basis of gel electrophoresis was that in each species all of the membrane fractions from the microsomal (P) pellet contained the same major glycoprotein band(s). Since it was shown that this finding could not simply be the result of membrane crosscontaminations or of nonspecific adsorption of soluble glycoproteins, it would appear likely that it may have a significant bearing on plasma membrane biogenesis. Inasmuch as the relationship between the components of the endoplasmic reticulum and the cell surface membrane are not clearly understood, it will be important to characterize the glycoproteins which are found in these membranes.
The fat cell appears to be particularly attractive for the study of plasma membrane glycoproteins as it has a relatively simple complement of such molecules and furthermore is believed not to be involved in the synthesis of soluble glycoproteins for export.
In an accompanying report information in regard to the saccharide units of the plasma membrane glycoproteins from rabbit adipose tissue will be presented (41).