Selective extraction of membrane-bound proteins by phospholipid vesicles.

Extraction of membrane proteins from erythrocytes into sonicated phosphatidylcholine vesicles is described. In a process involving phospholipid and neutral lipid exchange, cell membrane proteins associate with the vesicles and can be separated from the cells by centrifugation. The protein transfer appears to be reversible; phospholipid vesicles mediate the delivery of small amounts of previously extracted protein into cell membranes. Prior to extraction, all but one of the proteins are accessible to lactoperoxidase iodination, and lipid analysis indicates that primarily the outer monolayer of the cell is involved in phospholipid exchange. Among the extracted proteins is acetylcholinesterase which is removed much more efficiently by this procedure than by concentrated salt solutions. The most abundant proteins of the erythrocyte membrane are not represented in the vesicle extract.

Extraction of membrane proteins from erythrocytes into sonicated phosphatidylcholine vesicles is described. In a process involving phospholipid and neutral lipid exchange, cell membrane proteins associate with the vesicles and can be separated from the cells by centrifugation.
The protein transfer appears to be reversible; phospholipid vesicles mediate the delivery of small amounts of previously extracted protein into cell membranes. Prior to extraction, all but one of the proteins are accessible to lactoperoxidase iodination, and lipid analysis indicates that primarily the outer monolayer of the cell is involved in phospholipid exchange. Among the extracted proteins is acetylcholinesterase which is removed much more efficiently by this procedure than by concentrated salt solutions. The most abundant proteins of the erythrocyte membrane are not represented in the vesicle extract.
Interactions between phospholipid vesicles and intact cells give rise to several processes.

Cholesterol
(1, 2) and phospholipids (3) exchange between membranes of cells and vesicles, phospholipid vesicles "fuse" with some types of cells (4-6) and with one another (71, and still other cells may incorporate vesicles by endocytosis (8). The extent to which such events occur depends on the type of cell, the lipid composition of the vesicles, exposure time and temperature, and a number of other variables.
In investigating cell-vesicle association between phosphatidylcholine vesicles and mammalian erythrocytes, we observed yet another phenomenon; the extraction of certain cell membrane proteins into the vesicle lipid fraction. The extraction process and its reverse, delivery of membrane proteins into cell membranes from vesicles, are described in this communication.
The accompanying paper (9) discusses the effects of these processes on cell function. Cholesterol Exchange-As was reported by Grunze and Deuticke (21, erythrocytes incubated with phosphatidylcholine vesicles lost cholesterol. In the 60-min time interval of these experiments, 15 to 20% of cell cholesterol was lost to the supernatant (Fig. 2). The reverse process, incorporation of [14C1cholesterol from labeled vesicles into cells, also is shown in Fig. 2.
Delivery of Vesicle Contents -Erythrocytes were incubated with dimyristoylphosphatidylcholine vesicles in the presence of WJsucrose, which was either entirely internal or entirely external to the vesicles. In Table II, cell incorporation of the internal and external sucrose is compared. Incorporation was not enhanced when the sucrose was inside the vesicles; indeed, uptake of internal sucrose was negligible when correction was made for sucrose released from vesicles during the experiment. Thus "fusion" of vesicles and erythrocytes, as described for other types of cells, did not occur.
Protein Extraction -Phosphatidylcholine vesicles contained associated proteins after incubation with erythrocytes. Shown in Fig. 3u are vesicles which had been exposed to cells for various time intervals, then separated from the cells and subjected to sucrose density gradient centrifugation at an average force of 248,000 x g. This procedure separated vesicles from soluble protein and, in samples incubated for longer times, separated dense, protein-containing vesicles from pure Selective  Fig. 3b. Although the fine structure of the denser bands was lost in fractionation, it could be shown that protein was associated with those bands and not with the less dense lipid band. Fig. 4 shows the result of an experiment designed to distinguish between membrane protein bound in some manner to vesicle membranes and soluble protein simply trapped in vesicle interstices. One of the extracted proteins is acetylcho-line&erase, and its enzymatic activity was found almost exclusively in the dense vesicle fractions. In contrast, when a soluble preparation of the same enzyme (beef erythrocyte acetylcholinesterase) was sonicated and incubated with vesicles under the same condition8 less than 2% of the esterase activity was carried to the high density fractions. In addition, polyacrylamide gel electrophores of protein8 from the vesicle and soluble fraction8 of supernatant showed different polypeptide distributions (Fig. 5). Thus, the proteins associated with vesicles were not representative of the proteins available in .he supernatant.  3. a, dimyristoylphosphatidylcholine vesicle suspensions centrifuged at 250,000 x g on a 0 to 30% sucrose density gradient, before (A) and after incubation with erythrocytes for (B) 0, CC') 5, (D) 10, (E) 15, and (F) 20 min. b, distribution of protein and phospholipid on sucrose density gradients after incubation times of 0 (protein, A; lipid, 0) and 20 (protein, 0; lipid, A) min. Protein distribution was determined by Lowry assay; phospholipid distribution was determined using ['~Clphosphatidylcholine, BSA, bovine serum albumin. gel electrophoresis. The high density vesicle fractions contained five major polypeptide species, of apparent molecular weights 91,000, 81,000, 68,000, 30,000, and 15,000. Minor bands with molecular weights of 42,000 and 20,000 appeared after long incubation times. Species with very high molecular weights (>200,000) appeared in the gels only after cell lysis3 became extensive. When lactoperoxidase-iodinated erythrocytes were incubated with vesicles, the iz51 distribution in the gels paralleled the Coomassie blue staining pattern except for the l&000-dalton species (Fig. 6). Thus, the four higher molecular weight proteins evidently originated on the cell surface.  Transfer of Extracted Proteins from Vesicles to Erythrocytes -When cells were incubated with vesicles in the presence of extracted proteins labeled with lZ51, the cells incorporated a small fraction of the label (Table III). In the absence of vesicles, uptake of the labeled protein could not be detected. Vesicles did not mediate uptake of soluble protein (diisopropyl fluorophosphate-inhibited trypsin) into red cell membranes. Interaction of Liposomes with Erythrocytes -Incubation of cells with suspension of large, multilamellar vesicles (liposomes) for 1 h at 37" resulted in no detectable extraction of acetylcholinesterase and no cell lysis.

DISCUSSION
Incubation of human erythrocytes with sonicated phosphatidylcholine vesicles results in exchange of cholesterol, phospholipids, and certain proteins between their membranes. The molecular details of these processes are not understood at present, although it is known that small vesicles (-250 A) rather than liposomes are required. The protein transfer is selective to the extent that many abundant proteins of the cell surface are not extracted, but the basis for this selectivity is not evident.
With one exception, the extracted species are cell surface proteins, since they are accessible to lactoperoxidase-catalyzed iodination in intact cells. Moreover, sphingomyelin and phosphatidylcholine are the only cell phosphatides found in appreciable quantity in cell-treated vesicles. Van Deenen and coworkers (20) showed that most phosphatidylserine and phosphatidylethanolamine of the erythrocyte exists in the inner membrane monolayer, whereas phosphatidylcholine and sphingomyelin are localized mostly in the outer monolayer. These observations suggest a superficial interaction between cell and vesicle in which neither loses its integrity. Present evidence suggests that the process is reversible; phospholipids, cholesterol, and small amounts of protein originating in the vesicles associated strongly with the cells during incubation.
It is conceivable that a highly specific budding process occurring only in particular membrane regions could yield vesicles of the observed composition. However, to account for the reversibility of lipid and protein transfer, such budding would have to be coupled to the reverse process-fusion of vesicles with the cells. Attempts to deliver contents of the vesicle lumen into cells were unsuccessful, and detectable amounts of intracellular proteins (hemoglobin and NADPreducing enzymes) did not appear in the supernatant until well after protein extraction was underway (see accompanying paper (9)). Since lipid and protein transfer occur without mixing of cell and vesicle contents, a true fusion and budding mechanism appears unlikely.
Vesicle extraction of membrane proteins differs from other extraction processes in several respects. Extraction of erythrocytes with detergents, low or high ionic strengths, urea, and Ca2+-free buffers all yield classes of proteins presumed to associate with the cells by specific electrostatic, hydrophobic, or cation-dependent forces (21,22). The proteins extracted by these methods bear little resemblance to the vesicle extract, except for one or two common bands. Moreover, vesicle-extracted proteins are not technically "solubilized." They are associated with the vesicle membranes to the extent that physiological concentrations of salts do not dissociate them, but they are solubilized by 1 M saline or detergents which disrupt the vesicles.3 This association is not simply trapping in collapsed vesicle interstices since (a) protein species found in lipid fractions separated from soluble protein by density centrifugation are not representative of those in the soluble fractions, as would be the case for solubilized proteins partially trapped, and (b) the extracted acetylcholinesterase is equally active in intact vesicles and in vesicles disrupted by concentrated salt solutions, indicating full exposure to the solvent. There is no conclusive evidence that extracted proteins associate with the vesicle membrane in a manner analogous with their native state, but their behavior is consistent with such an inference.
Continuing investigations are aimed at determining the roles of lipid charge, acyl chain length, cell surface charge and composition, cation concentrations, and other variables in the selectivity of protein extraction. Any mechanistic insight gained from these studies is of interest, not least because alteration of the selectivity could expand the field of proteins which could be partially purified by this approach. Other investigations are underway to improve the efficiency of protein transfer from vesicles back to cells since this delivery could provide a useful avenue to selective modification of membranes.