Characterization of human platelet surface and intracellular membranes isolated by free flow electrophoresis.

High voltage free flow electrophoresis has been applied to the separation of human platelet membranes. After short treatment with neuraminidase at the whole cell level, three membrane vesicle subpopulations have been isolated. Using a surface label (125I-labeled Lens culinaris lectin), the marker enzyme NADH-cytochrome c reductase, and lipid analysis, two of the fractions have been identified as of surface origin and the other consists of intracellular membrane elements. The distribution of adenylate cyclase, leucyl aminopeptidase, 5'-nucleotidase and Ca2+-ATPase has also been investigated, and their usefulness as markers for the different membrane fractions has been evaluated. All three fractions are vesicular but differ in size and character. Their phospholipid and cholesterol contents have been determined, and the cholesterol/phospholipid ratios of the two surface fractions are over twice that of the intracellular membrane, which also has a significantly lower microviscosity as determined by fluorescence polarization using diphenyl hexatriene. The polypeptide profiles from sodium dodecyl sulfate-polyacrylamide gel electrophoresis are particularly distinctive, with actin present in the two surface membrane fractions and absent from the intracellular membranes. Myosin, confirmed by its ATPase characteristics, is almost exclusively localized in one of the surface membrane fractions, and actin-binding protein is a prominent feature of the other.

High voltage free flow electrophoresis has been applied to the separation of human platelet membranes. After short t r e a t m e n t with neuraminidase at the whole cell level, three membrane vesicle subpopulations have been isolated.
Using a surface label ('2SI-labeled Lens culinaris lectin), the marker enzyme NADH-cytochrome c reductase, and lipid analysis, two of the fractions have been identified as of surface origin and the other consists of intracellular membrane elements. The distribution of adenylate cyclase, leucyl aminopeptidase, 5'"nucleotidase and Ca2+-ATPase has also been investigated, and their usefulness as markers f o r the different membrane fractions has been evaluated. All three fractions are vesicular but differ in size and character. Their phospholipid and cholesterol contents have been determined, and the cholesterol/phospholipid ratios of the two s u r f a c e fractions are over twice that of the intracellular membrane, which also has a significantly lower microviscosity as determined by fluorescence polarization using diphenyl hexatriene. The polypeptide profiles from sodium dodecyl sulfate-polyacrylamide gel electrophoresis are particularly distinctive, with actin present in the two surface membrane fractions and absent from the intracellular membranes. Myosin, confirmed by its A T P a s e characteristics, is almost exclusively localized in one of the surface membrane fractions, and actin-binding protein is a p r o m i n e n t f e a t u r e of the other.
The hemostatic properties of blood platelets depend upon the responsiveness of the plasma membrane to surface signals, which are then transduced in some way to the interior of the cell to activate a sequence of cytoplasmic events which result in shape changes, aggregation, and the secretion of granulestored procoagulants and other constituents.
In order to investigate at the molecular level the various surface membrane constituents involved in these signal-response mechanisms, the isolation of plasma membranes free from intracellular membrane contamination is a desirable prerequisite. Despite considerable research efforts, however, by many groups using a variety of different technical approaches, such separations have proved extremely difficult to achieve, and the many procedures and problems have been well reviewed by Sixma and Lips (1).
Some of the technical difficulties certainly lie in the small size of the platelet and its unusual resistance to mechanical forces for disruption. At their best, most present procedures result in a single mixed membrane fraction containing surface and intracellular elements or a number of subfractions varyingly enriched with surface-oriented components but also containing membrane material of intracellular origin. An additional problem, too, is that unlike other cells, normal platelets have a considerable proportion of their plasma membrane invaginated as a surface-connected canalicular membrane system. Such domains may differ significantly in composition from the rest of the boundary membrane which is contiguous with them.
In this paper, we describe a new approach to the differential isolation of platelet surface and intracellular membranes using high voltage free flow electrophoresis. The technique was first introduced by Hannig and Heidrich (2, 3). It has been used successfully for the isolation of membrane subfractions from erythrocytes (4) and mitochondria (5), lysosomes ( 6 ) , and renal tubules (7), but to our knowledge, it has not been previously applied to platelet membrane fractionation.
With this procedure, we have been able to isolate, on a preparative scale, platelet membrane subfractions from homogenates of fresh human platelets in which the surface charge had been modified by pretreatment with neuraminidase at the whole cell level. Three membrane vesicle subpopulations are produced of clearly different electrophoretic mobilities and which also differ significantly from each other analytically, enzymatically, in polypeptide and lipid composition, and in microviscosity. One subpopulation is believed to consist of membrane material of intracellular origin, and the other two are believed to represent different surface membrane domains.

MATERIALS AND METHODS
Neuraminidase (Clostridium perfringens) was the affinity-purified Sigma type IX preparation, diluted 1:50 with 0.1 M acetate buffer, pH 5.5, and frozen in small aliquots. Lens culinaris lectin was a gift from Dr. M. Crumpton (Imperial Cancer Research Fund Laboratories, Lincoln's Inn Fields, London). The lectin was trace-labeled with ' 9 using the chloramine-T procedure (8). The specific activity at the time of preparation was 5 X lo7 cpm/mg of lectin. Carrier-free NalZ5I was obtained from The Radiochemical Centre, Amersham, U. K. All other chemicals were commercial preparations of analytical grade.
Isolation of PZateZets-Platelets were isolated from fresh buffy coats and processing was started within 3-4 h of collection. The isolation and preparation of washed platelets was carried out at room temperature. The buffy coats were centrifuged at 200 X g for 15 min and the resulting platelet-rich plasma was diluted 1:1 with 0.15 M NaCl, 4 m M EDTA, and the platelets were then sedimented by centrifugation at 2000 X g for 20 min.

4095
treated Washed Platelets-The platelets were resuspended in 10 mM Hepes' buffer, pH 6.2, containing 0.150 M NaCI, 4 mM KC1, 3 mM EDTA, and 0.1% bovine serum albumin and incubated at 37 O C for 10 min. The platelet suspension (usually 20 ml containing approximately 10" platelets/ml) was divided into two equal parts: 100 pl of neuraminidase (in 0.1 M acetate buffer, pH 5.5) was added to one half, giving a concentration of 0.03-0.05 units/ml (neuraminidase-treated platelets). An equal volume of acetate buffer, pH 5.5, was added to the other half as a control (control platelets). Both platelet suspensions were incubated for 20 min at 37 "C. A t the end of the incubation, the suspensions were rapidly diluted 1:3 with 10 mM Hepes buffer, pH 7.2, containing 0.150 M NaCI, 4 mM KCI, 3 mM EDTA, and 0.1% bovine serum albumin (washing buffer), and immediately centrifuged at 1800 X g for 15 min. The control and treated platelets were washed three times by resuspension in the washing buffer and finally resuspended in 10 mM Hepes buffer, pH 7.2, containing 0.3 M sorbitol (homogenizing buffer) to a ratio of 4 ml of buffer/g wet weight of cells. Each suspension was sonicated at setting 6 for 10 s at 4 'C using a Dawe sonicator with a 4-mm diameter probe. The sonicate was centrifuged at 2000 X g for 20 min at 4 "C, the supernatant was retained, and the pellet of unbroken cells and debris was resuspended in the homogenizing buffer and resonicated. After centrifugation, the two supernatants were combined and constituted the starting homogenate for all subsequent subfractionation. Preparation of Platelet Membranes-Platelet homogenate (-6 m l ) was layered onto a 15-ml linear sorbitol gradient prepared from 1 M and 3.5 M sorbitol solutions containing 1 mM EDTA and buffered to pH 7.2 with 5 mM Hepes. Sorbitol was chosen because L. culinaris lectin which is used for the membrane labeling is eluted by sucrose. The tubes were centrifuged in a swing-out rotor (3 X 25 m l ) at 42,000 X g for 90 min at 4 "C. Platelet membranes which located in the upper of the two discrete particulate zones were removed with a Pasteur pipette and were centrifuged at 90,OOO X g for 60 min.
Free Flow Electrophoresis of Platelet Membranes-Free flow electrophoresis was carried out in an Elphor VaP5 apparatus (Bender & Hobein, Munich) according to Hannig and Heidrich (3). The electrode buffer was that recommended by Heidrich and Leutner (4) and consisted of 100 mM triethanolamine, 100 mM acetic acid, adjusted to pH 7.2 with NaOH. The buffer for the separating chamber was IO mM triethanolamine, 10 mM acetic acid, pH 7.2, made iso-osmolar with 0.27 M sorbitol. The membrane pellet was resuspended in the chamber buffer to a concentration of 3-4 mg of membrane protein/ ml and injected into the chamber at a rate of 1 ml/h under the following electrophoresis conditions: 1200 V, 130 mA, and 6 "C. The chamber buffer flow rate was 2 ml/h/fraction.
Labeling of Platelets with L. culinaris '251~Lectin-Labeling of both control and neuraminidase-treated platelets with the iodinated lectin was carried out after Greenberg and Jamieson (9). Dependent upon the specific activity, between 5-10 yg of the labeled lectin per ml of platelet suspension was added, and the mixture was incubated at room temperature (20-22 "C) for 20 min.
Enzyme Assays-The antimycin-insensitive NADH-cytochrome c reductase (EC 1.6.99.3) was assayed essentially as described by Tolbert (10). The final reaction mixture contained 80 mM potassium phosphate buffer, pH 7.0, 80 pM cytochrome c, 0.2 M KCN, 10 p1 of antimycin (2 m g / d in ethanol), and 0.85 mM NADH. The reaction was initiated by the addition of NADH and was followed for 5-10 min at 550 nm. Leucyl aminopeptidase activity was measured by the procedure of Tuppy et al. (ll), and 5'-nucleotidase was measured by a modification of the method of Hawks and Wallach (12) using a reaction mixture containing 50 mM Tris-HCI, pH 7.0, 1 mM MgCL, 2 m~ p-nitrophenylphosphate, 0.5% Triton X-100, and 0.2 mM [3H]-AMP (5 X lo5 cpm; Amersham). Adenylate cyclase activity was determined with a reaction mixture containing 10 mM Tris-HC1 buffer, pH 7.5, 10 mM NaF, 5 mM MgC12, 10 mM theophylline, 1 mM ATP, with 4 mM phosphoenolpyruvate and 0.04 mg/ml of pyruvate kinase. After 20 min of incubation, the reaction was terminated by addition of an equal volume of 7.5 mM EDTA and boiling for 3 min. After centrifugation, the cAMP in the supernatant was determined with an Amersham radioassay cAMP kit (TRK 432). For determination of the myosin-like Ca2'-ATPase measured under high ionic strength maleate buffer, pH 6.0, 0.6 M KC1, 3 mM CaC12, and 3 mM ATP. The conditions, the following reaction mixture was used: 30 mM Tris/ reaction was allowed to proceed for 30 min at 37 "C and was stopped ' The abbreviations used are: Hepes, 4(2-hydroxyethyl)-l-piperazineethanesulfonic acid SDS, sodium dodecyl sulfate. by the addition of 0.5 ml of 1.2 M HC104. Inorganic phosphate was determined by the Martin and Doty procedure (13).
Sialic Acid-Sialic acid was assayed using the thiobarbituric acid assay of Warren (14) after hydrolysis in 0.1 N HzS04 at 80 "C for 1 h.
Lipid Estimations-Lipids were extracted with chloroformmethanol (1:2, v/v) according to the procedure of Bligh and Dyer (15). Cholesterol was determined by the method of Crawford (16) and phospholipid phosphorus was determined according to King (17). Phospholipid was calculated by assuming 25 pg of phospholipid/pg of lipid phosphorus (18).
Microuiscosity-The microviscosity of the membrane vesicle subpopulations was determined essentially by the procedure of Inbar et al. (19). A dispersion of tetrahydrofuran-solubilized diphenyl hexatriene in phosphate-buffered saline was mixed with the vesicle suspensions to a final concentration of approximately 3 X 10" M diphenyl hexatriene and incubated for 15 min at 37 "C in the dark. The vesicles were washed and resuspended in buffered saline, and the fluorescence polarization and intensity were measured in an ELCINT fluorescence polarimeter over a temperature range 12-40 OC. The fluorescence polarization (P) and the apparent microviscosity, q, were calculated according to Shinitzky et al. (20).
Protein Determination-The microtannin turbidimetric method of Katzenellenbogen and Dobryszycka (21) was used. This method is unaffected by the presence of sorbitol. Standard curves were prepared for each batch of assays using bovine serum albumin.
SDS-Polyacrylamide Gel Electrophoresis-Membranes were solubilized in 0.15 M Tris-HC1 buffer, pH 6.5, 5% mercaptoethanol, 3% SDS by heating for 2 min at 100 "C. Approximately 50 pl of the solubilized membrane samples containing 10% glycerol and 0.001% bromphenol blue were electrophoresed in a Pharmacia slab gel apparatus on a 12% resolving gel overlaid with 6% stacking gel according to the method of Laemmli (22). Gels were run at 220 V until the tracking dye reached the bottom of the gel and then stained for protein with 0.25% Coomassie brilliant blue, 40% methanol, 10% acetic acid, and destained in 40% methanol, 10% acetic acid.
Electron Microscopy-Fractions were fixed as suspensions with 2% glutaraldehyde in cacodylate buffer (0.06 M, pH 7.4). They were left for 2 h at room temperature and centrifuged, and the pellet was washed twice with cacodylate buffer. The pellets were then fixed for 1 h in buffered 1% OsO,, dehydrated in alcohol including a step containing uranyl acetate in 100% ethanol, and through propylene oxide to Epon/Araldite mixture in which they were embedded. Thin sections were mounted on grids and stained with Reynolds lead citrate and 5% aqueous uranyl acetate. Specimens were viewed in a Philips 400T electron microscope. Fig. 1 (top) shows full distribution profiles for protein, NADHcytochrome c reductase, and 1251-lectin in the membrane fractions separated by high voltage electrophoresis from control platelets. Two subfractions are resolved (CI and CII) of clearly different electrophoretic mobilities. However, both fractions were labeled with the '"I-lectin with the least electronegative fraction (CII) being somewhat more enriched. The activity of NADH-cytochrome c reductase showed a broad peak predominantly associated with the most electronegative fraction (CI), but this activity overlapped considerably with the CII fraction.

Electrophoretic Profiles of the Membrane Preparutions-
Pretreatment of platelets at the whole cell level with neuraminidase, which under our conditions removes 30-40s of the total cell sialic acid as determined after acid hydrolysis, resulted in a reduction in the electrophoretic mobility of a large proportion of the membrane (Fig. 1, bottom). These membranes now separated into three discrete fractions, Nr, Nrr, and NIII, NI and Nl11 being respectively the least and most affected by neuraminidase when their mobilities were compared with fractions from platelets not treated with neuraminidase. T h e '"I-lectin label was now only associated with fractions NII and NIII, being substantially higher in NIII. Fraction NI carried virtually no lectin label but the NADH-cytochrome c reductase was located predominantly in this fraction with only a small proportion of activity associated with NII and none with fraction Nw. With respect to the distribution of membrane protein between the fractions, NI generally accounts for 25-35% of the material applied to the electrophoresis chamber, and NII and NIIl together, account for the remainder. In addition to the EDTA routinely included as a Ca2'-dependent protease inhibitor at various stages in the preparation, a study was made of the value of adding a wide range protease inhibitor mixture during the preparation. Fig. 2b shows the electrophoresis protein profile for the separated membrane fractions prepared from platelets processed in the presence of phenylmethylsulfonyl fluoride, leupeptin, and pepstatin. The membrane separation was essentially the same as those prepared in the absence of these inhibitors (Fig. 2a), and no differences were revealed in the polypeptide patterns prepared under the two experimental conditions. Enrichment of Markers in Membrane Subfractions-The membrane fractions from the neuraminidase-treated cells were pooled across the peaks according to the protein profiles as indicated in Fig. 1 (bottom). The specific activities of the homogenate and the electrophoretically separated membrane subfractions for the ""I-lectin label and NADH-cytochrome c reductase activity are shown in Fig. 3. It can be seen that the membrane subfraction pools NII and Nlrr were both significantly enriched in the lectin label ( N~I , 2.9-3.7-fold; and NIII, 4.8-5.7-fold) with respect to homogenate, and the NI fraction showed no lectin enrichment. In contrast, this latter fraction had the highest enrichment for NADH-cytochrome c reductase (12.8-13.4-fold) with no enrichment of this enzyme in fractions NII and NlI,. A number of other enzyme activities were investigated for their localization and possible suitability as markers in the characterization of the membrane fractions. These have been selected for inclusion in Fig. 3    that the enzyme enrichment value with respect to homogenate in a particular membrane subfraction was accompanied by depletion in the other two fractions. Leucyl aminopeptidase and 5"nucleotidase appear to be useful markers for the Nr fraction, though their lower enrichment values may suggest multisite location in the platelet. Adenylate cyclase is predominantly associated with the NII and NIII fractions, and the Ca2'-ATPase is almost exclusively located in fraction N I~ with 11-13-fold enrichment. This ATPase measured under high ionic strength conditions could be readily eluted from the membrane with 0.6 M KC1, and the activity showed a pH prof'iie with two peaks at pH 5.8 and above pH 9.0. In the presence of 3 mM Ca'+, it was half-maximally inhibited by 25 p~ Mg". These features are characteristic for the Ci2'-ATPase of myosin. Studies of the activity of alkalinep-nitrophenyl phosphatase and phosphodiesterase activity towards either bis( p-nitrophenyl phosphate) diester or the corresponding fluorescent substrate bis(methylumbellifery1) phosphate (23) revealed that they had no value as marker enzymes for these human platelet membrane fractions. This is in contrast to earlier studies with pig platelet surface and intracellular membrane fractions (24), where two phosphodiesterase activities showed differential localization with good enrichments.
The cholesterol and phospholipid contents and cholesterol/ phospholipid ratios for the homogenate, and the pooled membrane fractions are presented in Table I. The three membrane subfractions differ significantly in a number of respects. The highest cholesterol/phospholipid ratio is found in fraction NIII being approximately 2.5 times greater than in fraction NI. This ratio difference is accounted for entirely by the high cholesterol content of NI11, since the phospholipid contents of NI and N I I 1 are substantiallly the same. Fig. 4 shows a typical fluorescence polarization study of the three membrane sub-  fractions using the probe diphenylhexatriene. The derived microviscosity ( q ) values have been plotted against temperature, and it is clear that the membrane vesicles of fraction NI have a very significantly lower microviscosity than the vesicles of the NII and NLII fractions. This finding is consistent with the much lower cholesterol content found in this fraction, a feature which is now considered to at least partially account for higher fluidity characteristics of certain cell membranes. Fig. 5 shows the SDS-polyacrylamide gel separations of the mixed membrane fraction and the three membrane subpopulations from a typical free flow electrophoretic separation. The mixed membrane fraction contains between 30 and 40 discrete polypeptides, the most prominent being two which have mobility characteristics corresponding to actin (Mr = 43,000) and the heavy chain of myosin ( M , = 200,000). These two proteins co-migrated with standard preparations of rabbit muscle actin and myosin. The most significant difference between the fractions was the complete absence of these two contractile proteins in fraction NI. Both proteins were well represented in the membrane fraction NII, but while actin featured most prominently in the N1l1 fraction, the myosin band was barely detectable. This almost exclusive association of myosin with the Nrl membrane fraction is substantiated by the specific localization of the Ca"-ATPase activity referred to earlier which displayed characteristic myosin-like properties. Above the position of myosin, there is a faint band in NII which is more prominent in Nllr and absent from NI. Although the identity of this polypeptide has not been f m l y established, it migrates with a mobility closely corresponding to that of actin-binding protein (Mr = 260,000) and this is being further characterized. The electron microscopy of the mixed membrane fraction showed vesicles of widely varying diameters (Fig. 6). The membrane subfraction NI contained a fairly homogeneous population of quite small vesicles, whereas NII and NlIl consisted of vesicles of considerably greater diameters, with very large vesicles predominating in fraction Nlr.

Isolation of Platelet Surface and Intracellular Membranes
It is our view that fraction NI consists of membrane vesicles of predominantly intracellular origin, since in addition to significant difference from NII and N11l in lipid composition, microviscosity, electron microscopic appearance, and a characteristic polypeptide profile, it represents membrane regions which at the whole cell level are inaccessible to both the lectin label and neuraminidase (Fig. 1). The enrichment and almost exclusive localization of antimycin-insensitive NADH-cytochrome c reductase in the NI fraction (Fig. 1, bottom, and Fig.   3) suggests that these vesicles have some of the characteristics of endoplasmic reticulum fractions prepared from other cells. The presence of, and enrichment of 5'-nucleotidase in the NI fraction is particularly interesting, since despite its wide use as a plasma membrane marker, it has been identified in endoplasmic reticulum (25) and in microsomal fractions (26) of other cells, and previously we have found that this enzyme, with AMP as substrate, is not well expressed in surface membrane-enriched fractions prepared from pig platelets (27).
Concerning fractions NII and Nlrl, the location of the lectin, the absence of reductase activity and vulnerability to neuraminidase attack suggests that both of these fractions are rich in surface-oriented components and may represent two different domains of the plasma membrane. Our results suggest, too, that N11 is somewhat less accessible to both the neuraminidase and lectin label than NIII, and may represent invaginated regions of the platelet surface. The microviscosity profiles for Ntl and Nlll are closely similar, as are also the polypeptide patterns, with the one major exception that NII is very much enriched in myosin. This myosin is readily eluted from the vesicles with 0.6 M KCl, and it is not known whether its presence depends upon association with actin or some other membrane component. The relatively low lipid content of this NI1 fraction, expressed in terms of membrane protein, may of course be accounted for by the rich content of cytoskeletal proteins.
In conclusion, we believe that high voltage free flow electrophoresis has some significant advantages over existing procedures for the differential isolation of platelet membranes. With this technique, we believe we have been able to separate surface and intracellular membrane vesicles from a human platelet mixed membrane fraction. Appropriate markers for the membrane subfractions show good enrichment and negligible cross-contamination, and the procedure is reproducible and can be carried out on a preparative scale. Additionally, the surface membrane separates into two discrete subpopulations which differ significantly in morphological and molecular criteria. We believe these two fractions may represent different plasma membrane domains, and the possible site of origin in the intact platelet of all three membrane subfractions is being investigated.  6. Electron micrographs of mixed membrane (a) and membrane subfractions NI, NII, and NIII (b, c, and d, respectively).
The bars equal 1 pm.