Characterization of an antiserum to guinea pig antithrombin III (AT III). I. Reactivity with guinea pig T lymphocytes

In the course of studies investigating the effects of antisera prepared against a variety of guinea pig proteins on lymphocyte function, a goat antiserum prepared against a guinea pig gamma-globulin preparation was found to react with guinea pig T lymphocytes. This antiserum, serum 592, contained a significant titer of antibodies that were cytotoxic for a subpopulation of lymph node cells and thymocytes, and mitogenic for lymph node T cells. Immunoelectrophoretic analysis and selective absorptions of the antiserum demonstrated that the antigen recognized on thymocytes was also present on an alpha 2 globulin in guinea pig plasma, which, on the basis of physiochemical characteristics and heparin-binding affinity, appeared to be guinea pig antithrombin III (AT III). Although the antiserum was shown to contain antibodies to both protein and carbohydrate determinants on the AT III molecule, studies comparing the effects of 7 M guanidine and periodate oxidation on the antigenicity of the AT III determinant also recognized on the thymocytes indicated that this shared antigenic determinant was carbohydrate in nature. The thymocyte membrane molecule bearing this determinant was also isolated and was found to be a 210,000-dalton macromolecule that was very sensitive to proteolytic and/or autolytic degradation. These data raise the interesting possibility that guinea pig lymphoid cells may have a membrane-associated protease inhibitor related to plasma AT III.     

Carbohydrate composition of lymphocyte plasma membrane from pig mesenteric lymph node.

Pig lymphocyte plasma membrane isolated from mesenteric lymph node contained 69 mug of carbohydrate/mg dry wt., which was made up of neutral sugar, amino sugar and sialic acid in the molar proportions 5:1.7:1. The neutral sugar comprised fucose, ribose, mannose, glucose, galactose and inositol (molar proportions 2:9:11:15:26:1), and the amino sugar glucosamine and galactosamine (molar ratio 2:1). The ribose was most probably derived from RNA. All of the fucose and mannose and almost all of the glucosamine were associated with the membrane protein whereas the membrane lipid contained all of the inositol. The remaining sugars were distributed in various ratios between the protein and lipid fractions.     

Production and in vivo effect of antibodies against guinea pig lymphokines

Supernatants from guinea pig lymph node lymphocytes stimulated with insoluble Concanavalin A in serum-free medium were fractionated by Sephadex chromatography and preparative electrophoresis. The isolated fraction possessed migration inhibition, mitogenic, and skin reactive activities. Associated with these were apparently two newly synthesized haemoproteins of unknown function. Antibodies were prepared against this partially purified lymphokine fraction. MIF produced by sensitized lymphocytes activated with an antigen (PPD tuberculin) could be completely absorbed from whole supernatants by immunoadsorbent columns prepared with that antibody whereas mitogenic factor and skin reactive factor were not retained. The anti-lymphokine antiserum totally inhibited the delayed skin response of sensitized guinea pigs challenged with PPD.     

Characterization of an antiserum to guinea pig antithrombin III (AT III). II. Stimulation of T lymphocyte proliferation

A goat antibody specific for an antigenic determinant shared between guinea pig antithrombin III (AT III) and thymocytes was shown to be mitogenic for lymph node T lymphocytes in the presence of macrophages. Although the antiserum was not mitogenic for purified populations of B lymphocytes, B lymphocytes were as efficient as T lymphocytes in absorbing the mitogenic activity of the serum. The shared antigenic determinant appeared to be carbohydrate in nature in that native and guanidine-treated AT III, but not periodate oxidized AT III, were capable of inhibiting the mitogenic activity of the serum when added continuously to the cultures. The possibility that the plasma protease inhibitor AT III or an antigenically related membrane protein are involved in the regulation of T cell activation is discussed.     

Unmasking of membrane enzyme activities and the problem of subcellular localization of adenylate cyclase in pig lymph node lymphocytes

A large-scale purification of plasma membranes from pig lymph node lymphocytes is described. Centrifugation on a discontinuous sucrose density gradient was performed in a zonal rotor. Adenylate cyclase activity of untreated fractions displayed a profile different from that of plasma membrane enzymatic markers and was maximal at higher density. However, when latent adenylate cyclase was unmasked by Lubrol PX treatment, its maximum was shifted to lower density and was no longer significantly different from that of plasma membrane markers. These results are discussed in terms of cell surface topography.     

GPI-linked proteins do not transfer spontaneously from erythrocytes to liposomes. New aspects of reorganization of the cell membrane

Exposure of cells to liposomes results in the release of integral membrane proteins. However, it is still controversial whether the release is due to spontaneous protein transfer from cells to liposomes or shed vesicles released from cells. We investigated this issue in an erythrocyte-liposome system by examining the location of acetylcholinesterase (AChE, an integral membrane protein marker), cholesterol (erythrocyte membrane lipid marker), hemoglobin (cytosolic protein marker), and a nonexchangeable lipid marker in liposomes in a sucrose density gradient at high resolution. The density distribution showed that AChE is not transferred to the liposomes but is located on small (about 50 nm) light (10-20 wt % sucrose) or large (about 200 nm) heavy shed vesicles (more than 30 wt % sucrose). AChE in the light shed-vesicle fraction markedly increased even after its level in the heavy fraction reached a plateau. AChE was also released from isolated heavy shed vesicles and accumulated in the small light shed-vesicle fraction in the presence of liposomes. After incubation of spherical erythrocytes (morphological index, 5.0) with liposomes, AChE hardly appeared in the heavy shed-vesicle fraction, and the majority (>99%) appeared in the light shed-vesicle fraction, indicating that AChE is released from both the erythrocytes and heavy shed vesicles to the light shed-vesicle fraction, which becomes rich in AChE. Our results demonstrated for the first time that GPI-linked proteins do not spontaneously transfer from erythrocytes to liposomes. Our study also suggests that in vivo GPI-linked membrane proteins do not spontaneously transfer between cell membranes but that some catalyst is needed.     

Membrane retrieval in the guinea pig neurohypophysis: biochemical characterization of a retrieval structure

[3H]Choline and [35S]methionine injected into the guinea pig hypothalamus in vivo were incorporated into the lipids and proteins, respectively, of secretory vesicles transported to the neural lobe. Prolonged in vivo stimulation of hormone secretion by dehydration decreased the [3H]choline content of secretory vesicles, with a concomitant increase in the [3H]choline content of a membrane fraction isolated on sucrose gradients. After stimulation of neural lobes in vitro in the presence of horseradish peroxidase, this extracellular fluid marker was found in the same membrane fraction. SDS electrophoresis of membrane proteins radiolabelled by [35S]methionine in vivo demonstrated that this fraction contained at least one major protein also present in the secretory vesicle membrane. These results suggest that we have isolated a membrane fraction containing the structure(s) involve in membrane retrieval in the neurohypophysis.     

Subcellular fractionation of pig coronary artery smooth muscle

A detailed procedure for subcellular fractionation of the smooth muscle from pig coronary arteries based on dissection of the proper tissue, homogenization, differential centrifugation and sucrose density gradient centrifugation is described. A number of marker enzymes and Ca2+ uptake in presence or absence of oxalate, ruthenium red and azide were studied. The ATP-dependent oxalate-independent azide- or ruthenium red-insensitive Ca2+ uptake, and the plasma membrane markers K+-activated ouabain-sensitive p-nitrophenylphosphatase, 5'-nucleotidase and Mg2+-ATPase showed maximum enrichment in the F2 fraction (15-28% sucrose) which was also contaminated with the endoplasmic reticulum marker NADPH: cytochrome c reductase, and to a small extent with the inner mitochondrial marker cytochrome c reductase, and also showed a small degree of oxalate stimulation of the Ca2+ uptake. F3 fraction (28-40% sucrose) was maximally enriched in the ATP- and oxalate-dependent azide-insensitive Ca2+ uptake and the endoplasmic reticulum marker NADPH: cytochrome c reductase but was heavily contaminated with the plasma membrane and the inner mitochondrial markers. The mitochondrial fraction was enriched in cytochrome c oxidase and azide- or ruthenium red-sensitive ATP-dependent Ca2+ uptake but was heavily contaminated with other membranes. Electron microscopy showed that F2 contained predominantly smooth surface vesicles and F3 contained smooth surface vesicles, rough endoplasmic reticulum and mitochondria. The ATP-dependent azide-insensitive oxalate-independent and oxalate-stimulated Ca2+ uptake comigrated with the plasma membrane and the endoplasmic reticulum markers, respectively, and were preferentially inhibited by digitonin and phosphatidylserine, respectively. This study establishes a basis for studies on receptor distribution and further Ca2+ uptake studies to understand the physiology of coronary artery vasodilation.     

Isolation of plasma membranes and Golgi apparatus from a single chicken liver homogenate

Chicken liver plasma membranes, minimally contaminated with Golgi apparatus-derived vesicles, were prepared from a low-speed (400 g) pellet by means of flotation in isotonic Percoll solution, followed by a hypotonic wash and flotation in a discontinuous sucrose gradient. Based on the analysis of suitable marker enzymes, alkaline phosphatase and alkaline phosphodiesterase, two plasma membrane fractions were isolated with enrichments, depending on the equilibrium density and marker of 28-97 and with a total yield of 4-5%. Golgi apparatus fractions were prepared by flotation of microsomes, obtained from the same homogenate as the low-speed pellet, in a discontinuous sucrose gradient. The trans-Golgi marker galactosyltransferase was 27-fold enriched in a fraction of intermediate density (d=1.077-1.116 g/ml). Approximately 12% of galactosyltransferase was recovered in the membranes equilibrating d=1.031-1.148 g/ml. Contamination with plasma membrane fragments was low in the light (d=1.031-1.077 g/ml) and intermediate density Golgi vesicles. The isolation of purified plasma membranes and Golgi vesicles from one liver homogenate will enable future studies on receptor cycling between these cell organelles.     

Albumin associated with purified pig lymphocyte plasma membrane.

Plasma-membrane preparations purified from pig lymphocytes contained a major polypeptide component of mol.wt. about 68 000. This component was identified as pig albumin by the following comparisons with authentic pig serum albumin: (a) co-migration when analysed by sodium dodecyl sulphate/polyacrylamide-gel electrophoresis under reducing and non-reducing conditions; (b) identical isoelectric points; (c) similar "fingerprints" of arginine-containing tryptic peptides; (d) reactivity with anti-(pig albumin) serum. The albumin was bound tightly to the plasma membrane. Biosynthetic labelling of pig lymphocytes under a variety of conditions failed to provide evidence that albumin was synthesized by lymphocytes, suggesting that the plasma-membrane-associated albumin was of extraneous origin. Radiolabelled pig serum albumin, however, failed to bind to the plasma-membrane fraction when added before cell disruption. Although lymphocyte plasma membrane preparations from other species possessed a polypeptide of about 68 000 mol.wt., this was judged not to be albumin on the basis of electrophoretic mobility under non-reducing conditions; also, no polypeptide was precipitated by anti-albumin sera. It is concluded that pig lymphocyte plasma-membrane preparations possess albumin which, although firmly attached, was probably of extraneous origin. This association appeared not to be common to lymphocytes from other species.     

Differences in membrane lipid composition and fluidity of transplanted GRSL lymphoma cells, depending on their site of growth in the mouse

GRSL lymphoma cells were isolated from various growth sites in the host. The relative membrane lipid fluidities of these cells and of normal lymphoid cells were estimated by fluorescence polarization, using the probe diphenylhexatriene and by measuring the (free) cholesterol/phospholipid molar ratio in whole cells. The results indicate that the membrane fluidity (reciprocal of the lipid structural order) of the lymphoma cells increases in the order of their location: peripheral blood less than spleen less than mesenterial lymph node less than ascites fluid. The membrane fluidities of normal lymphocytes from thymus, mesenterial lymph node and spleen were about the same, but higher than of peripheral blood lymphocytes, and between those of the lymphoma cells from lymph node and spleen. These results are confirmed by more extensive analysis on purified plasma membranes from the splenic and ascitic GRSL lymphoma cells and from normal splenocytes and thymocytes. The significantly higher lipid order parameter found in the GRSL plasma membrane isolated from the spleen as compared to those from the ascites cells could be fully explained by the differences measured in the major chemical determinants of the fluidity, i.e., the cholesterol/phospholipid ratio, the sphingomyelin content and the degree of saturation of the fatty acyl groups of the phospholipids. It was also found that the cholesterol/phospholipid ratio in erythrocyte membranes isolated from the peripheral blood of the tumor bearers was higher than in those from normal control mice. The observed differences in membrane fluidity between distinct subsets of tumor cells may be relevant to the sensitivity of these cells to immune attack or to drugs.     

Canine traceealis membrane fractionation and characterization

Canine trachealis was homogenized and the various membrane fractions isolated by differential centrifugation and discontinuous sucrose gradient centrifugation. A membrane fraction enriched in the plasma membrane marker enzymes 5′-nucleotidase (5-fold) and K⁺-activated ouabain sensitive p-nitrophenylphosphatase (3-fold) was obtained. The fraction contained very low levels of the inner mitochondrial marker succinate: cytochrome c oxidoreductase. These plasma membrane vesicles showed higher ATP-dependent Ca-uptake (20 μmoles/g protein) than any other submicrosomal fraction. The active Ca-uptake was enhanced by oxalate. The Ca taken up by the plasma membrane vesicles was released instantaneously by dilution in 5m$\text{M}$ EGTA and 10μ$\text{M}$ A23187 and more slowly by dilution only in 5m$\text{M}$ EGTA.     

Biochemical studies of the excitable membrane of Paramecium tetraurelia. III. Proteins of cilia and ciliary membranes

As a first step in the biochemical analysis of membrane excitation in wild-type Paramecium and its behavioral mutants we have defined the protein composition of the ciliary membrane of wild-type cells. The techniques for the isolation of cilia and ciliary membrane vesicles were refined. Membranes of high purity and integrity were obtained without the use of detergents. The fractions were characterized by electron microscopy, and the proteins of whole cilia, axonemes, and ciliary membrane vesicles were resolved by SDS polyacrylamide gel electrophoresis and isoelectric focusing in one and two dimensions. Protein patterns and EM appearance of the fractions were highly reproducible. Over 200 polypeptides were present in isolated cilia, most of which were recovered in the axonemal fraction. Trichocysts, which were sometimes present as a minor contaminant in ciliary preparations, were composed of a very distinct set of over 30 polypeptides of mol wt 11,000--19,000. Membrane vesicles contained up to 70 polypeptides of mol wt 15,000--250,000. The major vesicle species were a high molecular weight protein (the "immobilization antigen") and a group of acidic proteins with mol wt similar to or approximately 40,000. These and several other membrane proteins were specifically decreased or totally absent in the axoneme fraction. Tubulin, the major axonemal species, occurred only in trace amounts in isolated vesicles; the same was true for Tetrahymena ciliary membranes prepared by the methods described in this paper. A protein of mol wt 31,000, pI 6.8, was virtually absent in vesicles prepared from cells in exponential growth phase, but became prominent early in stationary phase in good correlation with cellular mating reactivity. This detailed characterization will provide the basis for comparison of the ciliary proteins of wild-type and behavioral mutants and for analysis of topography and function of membrane proteins. It will also be useful in future studies of trichocysts and mating reactions.     

Purification and characterization of a cortical secretory vesicle membrane fraction

A membrane fraction has been prepared by sucrose density gradient fractionation of purified cortical secretory vesicles from the eggs of the sea urchin Strongylocentrotus purpuratus. The purified cortical vesicle membrane fraction has a phospholipid to protein ratio of 1.76 and exhibits a morphology typical of biological membranes as seen by electron microscopy. The protein composition of the purified membranes was analyzed by SDS-polyacrylamide gel electrophoresis and shown to be distinct from that of eggs, cell surface complex, cortical vesicles, fertilization product, and yolk platelets. Alkaline extraction (pH 11.0) of peripheral membrane proteins increased the phospholipid to protein ratio to 2.55 and removed several polypeptides. Immunoblot analysis of the isolated cortical vesicle membrane fraction revealed low levels of contamination with two major cortical vesicle content proteins. Fractions enriched in egg plasma membranes and yolk platelet membranes also have been isolated and compared with the cortical vesicle membranes by SDS-polyacrylamide gel electrophoresis. The protein compositions of the three membrane fractions were found to contain very little overlap, indicating that the cortical vesicle membrane preparation is relatively free of contamination from these likely noncortical vesicle sources of membrane. Both the plasma membrane and cortical vesicle membrane samples were found by immunoblotting to contain actin.     

Differences in membrane lipid composition and fluidity of transplanted grsl lymphoma cells, depending on their site of growth in the mouse

GRSL lymphoma cells were isolated from various growth sites in the host. The relative membrane lipid fluidities of these cells and of normal lymphoid cells were estimated by fluorescence polarization, using the probe diphenylhexatriene and by measuring the (free) cholesterol/phospholipid molar ratio in whole cells. The results indicate that the membrane fluidity (reciprocal of the lipid structural order) of the lymphoma cells increases in the order of their location: peripheral blood < spleen < mesenterial lymph node < ascites fluid. The membrane fluidities of normal lymphocytes from thymus, mesenterial lymph node and spleen were about the same, but higher than of peripheral blood lymphocytes, and between those of the lymphoma cells from lymph node and spleen. These results are confirmed by more extensive analysis on purified plasma membranes from the splenic and ascitic GRSL lymphoma cells and from normal splenocytes and thymocytes. The significantly higher lipid order parameter found in the GRSL plasma membrane isolated from the spleen as compared to those from the ascites cells could be fully explained by the differences measured in the major chemical determinants of the fluidity, i.e., the cholesterol/phospholipid ratio, the sphingomyelin content and the degree of saturation of the fatty acyl groups of the phospholipids. It was also found that the cholesterol/phospholipid ratio in erythrocyte membranes isolated from the peripheral blood of the tumor bearers was higher than in those from normal control mice. The observed differences in membrane fluidity between distinct subsets of tumor cells may be relevant to the sensitivity of these cells to immune attack or to drugs.     

Outer acrosomal membrane of guinea pig spermatozoa: Isolation and structural characterization

We describe a protocol to isolate a highly enriched fraction of outer acrosomal membrane from guinea pig spermatozoa and present new data on the ultrastructure of this membrane domain. Cauda epididymal spermatozoa were suspended into a low ionic strength buffer and subjected to brief homogenization; this stripped the plasma membrane from the spermatozoa and severed the acrosomal apical segment from the spermatozoon. The crescent-shaped apical segments retained the outer acrosomal membrane and specific components of the acrosomal matrix. Enriched fractions of apical segments were isolated on discontinuous sucrose gradients and the outer acrosomal membrane purified by subsequent centrifugation onto Percoll density gradients. The isolated outer acrosomal membrane did not form vesicles, but instead rolled up into spiral sheets. Both thin section and negatively stained specimens revealed a paracrystalline arrangement of filaments associated with the luminal surface of the membrane. The isolated outer acrosomal membrane revealed a limited number of polypeptides by SDS-PAGE, and the polypeptide pattern was distinct from the plasma membrane fraction. The isolated acrosomal membranes possessed no oubain sensitive Na+, K+-ATPase activity, whereas about 20% of the ATPase activity of the plasma membrane enriched fraction was inhibited by oubain. The potential function of the structural differentiations of the outer acrosomal membrane in the membrane fusion events of the acrosome reaction is discussed.     

Immunological properties of membrane fractions from wild type and dnaA mutants of Escherichia coli

Membrane vesicles from Escherichia coli wild type and an otherwise isogenic dnaA mutant were used to immunize rabbits. In addition, a membrane protein fraction, containing the material found deficient in dnaA mutants, was purified by preparative polyacrylamide gel electrophoresis in sodium dodecylsulfate, and used for immunization. The antisera produced were analyzed by immunoelectrophoresis and immunofluorescence microscopy. The antisera obtained by immunization with membrane vesicles from either wild type or dnaA mutant membrane preparations were qualitatively similar in the precipitin bands seen after immunoelectrophoresis. The antisera obtained by immunization with the purified protein fraction contained a subset of the antibodies seen when whole vesicles were used for immunization. In a semiquantitative precipitin assay, the antisera prepared against whole membrane vesicles or the isolated protein fraction both caused the precipitation of more protein from sodium dodecylsulfate-solubilized membranes of wild type than of dnaA mutants. No difference was seen by immunoelectrophoresis between the protein composition of wild type or dnaA membrane preparations. Thus, the dnaA mutant appears to differ from the wild type in the quantitative composition of its membrane proteins, whereas no qualitative differences were detected.Fluorescein-conjugated antiserum preparations were employed to assess the reactivity of intact cells, spheroplasts and membrane vesicles with the antisera studied above. Wild type cells of E. coli have a barrier to reaction with the antisera; this barrier is removed when the cells are converted to spheroplasts or to membrane vesicle. Similarly, a highly permeable mutant of E. coli permits reaction of the antisera with unaltered cells. Antisera to both whole membrane vesicles and to the isolated protein fraction react identically with the cellular and subcellular preparations. Thus, antisera prepared from membrane proteins isolated after sodium dodecylsulfate-polyacrylamide gel electrophoresis can still recognize some antigens present in membrane vesicle preparations.     

Nonidet P-40 extraction of lymphocyte plasma membrane. Characterization of the insoluble residue.

Purified preparations of lymphocyte plasma membrane were extracted exhaustively with Nonidet P-40 in Dulbecco's phosphate-buffered saline medium. The insoluble fraction, as defined by sedimentation at 10(6) g-min, contained about 10% of the membrane protein as well as cholesterol and phospholipid. The lipid/protein ratio, cholesterol/phospholipid ratio and sphingomyelin content were increased in the residue. Density-gradient centrifugation suggested that the lipid and protein form a common entity. As judged by sodium dodecyl sulphate/polyacrylamide-gel electrophoresis, the Nonidet P-40-insoluble fractions of the plasma membranes of human B lymphoblastoid cells and pig mesenteric lymph-node lymphocytes possessed similar qualitative polypeptide compositions but differed quantitatively. Both residues comprised major polypeptides of Mr 28 000, 33 000, 45 000 and 68 000, together with a prominent band of Mr 120 000 in the human and of Mr 200 000 in the pig. The polypeptides of Mr 28 000, 33 000, 68 000 and 120 000 were probably located exclusively in the Nonidet P-40-insoluble residue, which also possessed a 4-fold increase in 5'-nucleotidase specific activity. The results indicate that a reproducible fraction of lymphocyte plasma membrane is insoluble in non-ionic detergents and that this fraction possesses a unique polypeptide composition. By analogy with similar studies with erythrocyte ghosts, it appears likely that the polypeptides are located on the plasma membrane's cytoplasmic face.     

Effects of thrombin on washed, human platelets: changes in the subcellular fractions.

Pressure homogenization and subcellular fractionation has been performed on washed, human platelets and platelets treated with thrombin to undergo the so-called release reaction. Electron microscopy revealed that the particulate zones obtained from the control sample corresponded to membrane vesicles (B), small storage granules (D) as well as mitochondria and larger storage granules (E). Only a few storage granules could be observed in the particulate zones isolated from thrombin-treated platelets. Visual comparison of the sucrose gradient patterns revealed that one granule fraction (D) had disappeared from the thrombin-treated sample. Sodium dodecysulfate-polyacrylamide gel electrophoresis showed a major protein band (mol. wt 145 500 plus or minus 1000) in the extracellular phase (supernatant after removal of the platelets) of the thrombin-treated sample and in the granule fractions (D and E) of the control (mol. wt 147 000 plus or minus 1000). Incubation of whole, washed platelets with thrombin for 5 min at 37 degrees C followed by sodium dodecylsulfate-polyacrylamide gel electrophoresis of the isolated membrane fraction revealed no reproducible differences in the protein band pattern compared to membranes isolated from control platelets. However, after treatment with thrombin for 30 min, a protein band (mol. wt 183 000 plus or minus 3500) had disappeared. The distribution of protein and beta-N-acetylglucosaminidase activity among the subcellular fractions were measured. Both were mainly recovered in the soluble fraction (greater than 77%). The granule fractions, D and E of the control contained 3.0% plus or minus 0.8% and 6.4% plus or minus 1.3% of the total amount of beta-N-acetylglucosaminidase in the gradient. Fraction E of the thrombin-treated cells contained 3.3% plus or minus 1.0% of total while fraction D was lacking.     

The isolation of endosome-derived vesicles from rat hepatocytes.

Intracellular 5'-nucleotidase involved in membrane circulation in rat hepatocytes is latent, and is protected from inhibition when whole cells are incubated with inhibiting antiserum at 2 degrees C [Stanley, Edwards & Luzio (1980) Biochem. J. 186, 59-69]. These two criteria were used to identify intracellular membrane vesicles containing 5'-nucleotidase on Ficoll density gradients. A sharply defined turbid band containing intracellular 5'-nucleotidase isolated on density gradients was further fractionated by immunoadsorption of plasma-membrane fragments derived from the cell surface of surface-inhibited cells on to an anti-(immunoglobulin G) immunoadsorbent. The resulting non-adsorbed membrane fraction consisted of vesicles of uniform size (approx. 65 nm diam.), but was not identifiable as any known organelle. This fraction could account for approx. 5% of the total cell 5'-nucleotidase activity, and the enzyme activity measured was 55% latent. The fraction had a restricted polypeptide composition but similar phospholipid composition compared with plasma membrane. We suggest that the vesicles observed in this fraction were derived from the endocytic pathway.