We propose the use of membrane splitting by freezefracture for differential phospholipid analysis of protoplasmic and exoplasmic membrane leaflets (halves). Unfixed cells or tissues are quick-frozen, freeze-fractured, and platinum-carbon (Pt/C) shadowed. The Pt/C replicas are then treated with 2.5% sodium dodecyl sulfate (SDS) to solubilize unfractured membranes and to release cytoplasm or contents. While the detergent dissolves unfractured membranes, it would not extract lipids from split membranes, as their apolar domains are stabilized by their Pt/C replicas. After washing, the Pt/C replicas, along with attached protoplasmic and exoplasmic membrane halves, are processed for immunocytochemical labeling of phospholipids with antibody, followed by electron microscopic observation. Here, we present the application of the SDS-digested freeze-fracture replica labeling (SDS-FRL) technique to the transmembrane distribution of a major membrane phospholipid, phosphatidylcholine (PC), in various cell and intracellular membranes. Immunogold labeling revealed that PC is exclusively localized on the exoplasmic membrane halves of the plasma membranes, and the intracellular membranes of various organelles, e.g. nuclei, mitochondria, endoplasmic reticulum, secretory granules, and disc membranes of photoreceptor cells. One exception to this general scheme was the plasma membrane forming the myelin sheath of neurons and the Ca2+-treated erythrocyte membranes. In these cell membranes, roughly equal amounts of immunogold particles for PC were seen on each outer and inner membrane half, implying a symmetrical transmembrane distribution of PC. Initial screening suggests that the SDS-FRL technique allows in situ analysis of the transmembrane distribution of membrane lipids, and at the same time opens up the possibility of labeling membranes such as intracellular membranes not normally accessible to cytochemical labels without the distortion potentially associated with membrane isolation procedures.

The phospholipid distribution in membranes has been studied mainly by biochemical and physicochemical analyses conducted on artificial membranes, isolated cell homogenates, or non-complex blood elements, i.e. erythocytes and platelets. However, these approaches have significant limitations and pitfalls. For instance, evaluation of lipid distribution in an intracellular membrane requires cell lysis and isolation of the organelles. In some cases, ATP plays a pivotal role in membrane asymmetry (Zachowski, 1993). During the isolation procedure, the membrane is separated from cytoplasmic ATP, and a new lipid repartition may occur, different from that in situ (Schrier et al., 1992).

A particularly interesting property of membrane lipids is that many eukaryotic cells exhibit an asymmetric distribution of lipids across the bilayer membrane (Zachowski, 1993): in the erythrocyte membrane, 75% of the phosphatidylcholine and >85% of the sphingomyelin are in the exoplasmic membrane half, while the protoplasmic membrane half contains 80-85% of the phosphatidylethanolamine and >96% of the phosphatidylserine. However, at present, this transmembrane asymmetrical phospholipid distribution cannot be independently verified visually by more direct methods. Our objective here is to add a new electron microscopic approach to study phospholipid transmembrane distribution in various cell and intracellular membranes: the sodium dodecyl sulfate-digested freeze-fracture replica labeling (SDS-FRL) technique is a method enabling us to look directly at the two-dimensional distribution of phospholipids within biological membranes.

Fig. 1 shows schematically the rationale of our approach: after quick-freezing (Fig. 1a), freeze-fracture, and Pt/C shadowing (Fig. 1b), unfixed cells or intracellular organelles are treated with 2.5% SDS to dissolve unfractured membranes and contents (Fig. 1c). The cytoplasmic and exoplasmic halves of the membrane remain attached to the Pt/C cast. Although SDS dissolves unfractured portions of the membrane, it would not reach, micellize, and extract split membrane halves, as their apolar domains are positioned against, and stabilized by, their Pt/C casts. We reasoned that it should therefore be possible to label the cytoplasmic and exoplasmic leaflets of membranes in both isolated cells and tissue samples (Fig. 1d).

Fig. 1.

SDS-digested freeze-fracture replica labeling. (a) Cells are frozen. Only phospholipids are depicted. (b) Freeze-fracture splits the plasma membrane into inner protoplasmic (left side, PF) and outer exoplasmic (right side, EF) halves, and the fractured faces are stabilized by deposition of platinum/carbon (Pt/C). (c) Pt/C-casts are thawed and treated with SDS. The detergent dissolves unfractured areas of the plasma membrane, with release of the cytoplasm. (d) The replicas are then processed for immunogold labeling with antibody against a phospholipid. Note that in intracellular organelles, e.g. secretory granules and endoplasmic reticulum, the outer membrane half closest to the cytoplasm is designated the protoplasmic half, and the inner membrane half closest to the interior is designated the exoplasmic half. Thus, the fractured and replicated hydrophobic surface of the outer membrane half (labeled EF in Fig. 1b) is the P-face, and that of the inner membrane half (labeled PF in Fig. 1b) is the E-face.

Fig. 1.

SDS-digested freeze-fracture replica labeling. (a) Cells are frozen. Only phospholipids are depicted. (b) Freeze-fracture splits the plasma membrane into inner protoplasmic (left side, PF) and outer exoplasmic (right side, EF) halves, and the fractured faces are stabilized by deposition of platinum/carbon (Pt/C). (c) Pt/C-casts are thawed and treated with SDS. The detergent dissolves unfractured areas of the plasma membrane, with release of the cytoplasm. (d) The replicas are then processed for immunogold labeling with antibody against a phospholipid. Note that in intracellular organelles, e.g. secretory granules and endoplasmic reticulum, the outer membrane half closest to the cytoplasm is designated the protoplasmic half, and the inner membrane half closest to the interior is designated the exoplasmic half. Thus, the fractured and replicated hydrophobic surface of the outer membrane half (labeled EF in Fig. 1b) is the P-face, and that of the inner membrane half (labeled PF in Fig. 1b) is the E-face.

In this paper, we first tested the feasibility of SDS-FRL for phospholipid using artificial membranes prepared from various phospholipids with an antibody, and chemically analyzed the replicas to estimate how much of the total phospholipids bound to Pt/C replicas remains after SDS-treatment. Then, we applied SDS-FRL to the immunolabeling of various cell and intracellular membranes.

Artificial membranes

Artificial membranes (multilamellar vesicles, MLVs) were prepared from mixtures of L-α-phosphatidylcholine (PC; isolated from bovine brain; Avanti Polar Lipids, Inc., Albaster, AL) and L-α-phosphatidylL-serine (PS; isolated from bovine brain; Sigma Chemical Co., St Louis, MO) in a 1:1 molar ratio; from mixtures of PS and sphingomyelin (SM; isolated from bovine brain; Avanti Polar Lipids, Inc., Albaster, AL) in a 1:1 molar ratio; from pure PC; and from pure PS according to the method of Kinsky et al. (1969). Small unilamellar vesicles (SUVs) (150-200 nm in diameter) were prepared by using ultrasonic irradiation to break up suspensions of the MLV preparations. Aqueous dispersions of phospholipid vesicles were sandwiched between two well polished gold double replica carriers (Balzers Union, Liechtenstein) at 37°C, and then quickly frozen by plunging into a slush of partially solidified liquid nitrogen.

Cells and tissues

Erythrocytes were obtained from freshly drawn, heparinized human blood (blood type O or AB, Rh+). The cells were washed four times, and suspended in cold Hepes buffer (145 mM NaCl, 5 mM KCl, 1 mM MgSO4, 10 mM glucose, 0.1 mM EGTA, and Na-Hepes, pH 7.5). Some erythrocytes were incubated with Hepes buffer containing 1 mM CaCl2 and 20 µM A23187 at 37°C for various times. Tissues (liver, pancreas, retina, and optic nerve from rats or mice) were cut into sections less than 100 µm thick by a Microslicer (Dosaka EM Co., Kyoto, Japan) or minced in small pieces, and collected in 0.01 M phosphate buffered saline (PBS). Erythrocytes and tissue slices were quick-frozen by contact with a copper block cooled with liquid helium (Heuser et al., 1979).

SDS-digested freeze-fracture replica labeling

The frozen samples were fractured in a Balzers BAF 400T freeze-etch unit (Balzers Union, Liechtenstein) at -110°C, replicated by deposition of Pt/C from an electron beam gun positioned at a 45° angle followed by carbon applied from overhead. To release the replicas from the specimen carrier, the carrier was immersed gently in 0.1% bovine serum albumin (fraction V, Miles Inc., Kankakee, IL) in PBS. After floating off, the pieces of replica were transferred to 5 ml of 2.5% SDS (Sigma Chemical Co., St Louis, MO) containing 10 mM Tris and 30 mM sucrose, pH 8.3. SDS-digestion was carried out for 10-60 minutes at room temperature, with vigorous stirring. The replicas obtained from collagen-rich tissues such as pancreas and retinas were first treated with 0.1% collagenase (type I or type II; Sigma) containing 130 mM NaCl, 12 mM NaHCO3, 3 mM NaH2PO4, 3 mM Na2HPO4, 2 mM MgSO4 and 1 mM CaCl2, for 1 hour at 37°C with stirring, and then treated with 2.5% SDS for 3 hours at room temperature, with vigorous stirring. After treatment with SDS, replicas (actually carbon-stabilized membrane halves) were rinsed for at least 1 hour, with four or more changes of PBS, and labeled with anti-PC monoclonal antibody (JE-1; Nam et al., 1990) diluted 1:40 in PBS for 1 hour at 37°C. After labeling, the replicas were washed three times with PBS and incubated for 1 hour at 37°C with secondary goat anti-mouse IgM antibody-conjugated colloidal gold (Janssen Pharmaceuticals, Piscataway, NJ) diluted 1:40 in PBS. After immunogold labeling, the replicas were rinsed several times in PBS, fixed with 0.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4, for 10 minutes at room temperature, washed twice with distilled water, and picked onto Formvar-coated grids. Electron microscopy was done using a JEOL 1200EX.

Chemical measurement of phospholipids bound to Pt/C replicas

The amount of phospholipids bound to SDS-treated Pt/C replicas was measured by the method of Zhou and Arthur (1992). Aqueous suspensions (2 µl) of small unilamellar vesicles (SUVs) of PC (10 mg/ml in Hepes buffer, pH 7.4) were sandwiched between two well polished gold double replica specimen carriers, quickly frozen by plunging into partially solidified liquid nitrogen, freeze-fractured, and Pt/C shadowed. Freeze-fracture with double replica specimen carriers gives two sets of replicas. Thus, one obtained from SUVs of PC was washed in distilled water, and the other was treated with 2.5% SDS for 30 minutes and then washed with distilled water. These two sets of replicas were picked on pre-cleaned coverglasses separately, and dried with nitrogen. To release inorganic phosphorus from the Pt/C shadowed membrane halves, the glass-bound replicas were treated with 70% perchloric acid at 200°C for 2 hours. The released phosphorus was measured by using malachite green. Some replicas of SUVs of PC were examined with a JEOL 1200EX electron microscope.

Our approach is based on a hypothesis that the split membrane halves are physically stabilized (fixed) by carbon-shadowing onto the apolar fracture face, thus the carbon-fixed membrane halves would not be extracted with organic solvents or detergents (Andersson-Forsman and Pinto da Silva, 1988; Fujimoto, 1995; Fujimoto and Ogawa, 1991; Fujimoto and Pinto da Silva, 1988, 1989, 1992). We applied the SDS-FRL technique to the immunogold labeling of the intercellular junction proteins, e.g. gap junction protein (connexins), tight junction protein (occludin), and the findings suggested the reliability and the potential significance of SDS-FRL in the immunocytochemical labeling of integral membrane proteins (Fujimoto, 1995). Even if the carbon film acts to fix or retain the integral membrane proteins, it seems surprising that the membrane lipids really are retained after the SDS-treatment. To confirm this hypothesis, we first tested the feasibility of SDS-FRL for phospholipid using artificial membranes (MLVs) prepared from mixtures of phosphatidylcholine and phosphatidylserine (PC/PS) (molar ratio, 1:1), mixtures of phosphatidylserine and sphingomyelin (PS/SM) (molar ratio, 1:1), and from pure PS and pure SM with an anti-PC monoclonal antibody, JE-1. Fig. 2a and b show the labeling of PC/PS and pure PS membranes with JE-1, respectively. Although PC/PS membranes were intensely labeled by JE-1, the labeling of pure PS, PS/SM and pure SM membranes (data not shown) resulted in an absence of labeling. This suggests the reliability of phospholipid labeling by the SDS-FRL and the specificity of the antibody.

Fig. 2.

SDS-FRL of artificial membranes (multilamellar vesicles) prepared from mixtures of L-α-phosphatidylcholine (PC) and L-α-phosphatidyl-L-serine (PS) (PC/PS molar ratio, 1:1) (a) and from PS (b) with anti-PC monoclonal antibody, JE-1. Mixtures composed of PC and PS are intensely labeled by JE-1 (a), while labeling of pure PS membranes results in an absence of labeling (b). Bars, 100 nm.

Fig. 2.

SDS-FRL of artificial membranes (multilamellar vesicles) prepared from mixtures of L-α-phosphatidylcholine (PC) and L-α-phosphatidyl-L-serine (PS) (PC/PS molar ratio, 1:1) (a) and from PS (b) with anti-PC monoclonal antibody, JE-1. Mixtures composed of PC and PS are intensely labeled by JE-1 (a), while labeling of pure PS membranes results in an absence of labeling (b). Bars, 100 nm.

An important question has arisen as to how much of the total phospholipid bound to Pt/C replica remains after SDS treatment. Our quantitative chemical analysis using SUVs of PC showed that SDS-untreated and SDS-treated replicas (7 mm2) contain 10.3×10-12 and 2.5×10-12 moles of PC (average of two experiments), respectively. Although the ratio of PC present on the SDS-treated replica to that on the SDS-untreated is about 0.25, we should notice that in the control the Pt/C replica with carbon-stabilized membrane halves was only washed with distilled water. Therefore, a considerable amount of fractured or cross-fractured PC vesicles remain attached to the carbon cast, either because they are attached to the carbon stabilized inner monolayer of their fractured membrane (Fig. 3, Fracture plane B), or because they were cross-fractured and attached to the replica (Fig. 3, Fracture plane C). The ratio can be evaluated using the following assumptions: outer monolayer a and inner monolayer b are stabilized (fixed) by carbon-shadowing onto the apolar fracture face, and bilayer c+d is attached to the carbon stabilized inner monolayer b, or bilayer e+f was cross-fractured and remains attached to the SDS-untreated replica (Fig. 3-1). Although SDS dissolves unfractured portions of the PC bilayers (Fig. 3-2, c′+d′ and e′+f), it would not extract the Pt-C shadowed monolayers (Fig. 3-2, outer monolayer a′and inner monolayer b′). Quantitative electron microscopy with random photography of ∼200 freeze-fractured PC SUVs demonstrated that about 35% of the SUVs were fractured in Fracture plane A, 35% in Fracture plane B, and about 30% in Fracture plane C. With these assumptions, the ratio (X) of PC present on the SDS-treated replica to that on the SDS-untreated replica can be calculated by using the following equations:

Fig. 3.

Diagram for chemical analysis of total phospholipid bound to Pt/C replica after SDS-digestion. The fracture plane passes through the hydrophobic middle of the lipid bilayers (Fracture plane A and Fracture plane B), or crosses the interior of the vesicle (Fracture plane C). Thus, bilayers (c+d) and (e+f) are attached to the SDS-untreated replica (1). Although SDS dissolves unfractured bilayers (2, c′+d′ and e′+f), it does not extract Pt/C-shadowed monolayers (2, a′+b′). (3) Conventional freeze-fracture electron micrographs of the hydrophobic portion of outer (left) and inner monolayer (center) of the PC unilamellar vesicles, and the cross-fractured PC unilamellar vesicle (right). Bar, 200 nm.

Fig. 3.

Diagram for chemical analysis of total phospholipid bound to Pt/C replica after SDS-digestion. The fracture plane passes through the hydrophobic middle of the lipid bilayers (Fracture plane A and Fracture plane B), or crosses the interior of the vesicle (Fracture plane C). Thus, bilayers (c+d) and (e+f) are attached to the SDS-untreated replica (1). Although SDS dissolves unfractured bilayers (2, c′+d′ and e′+f), it does not extract Pt/C-shadowed monolayers (2, a′+b′). (3) Conventional freeze-fracture electron micrographs of the hydrophobic portion of outer (left) and inner monolayer (center) of the PC unilamellar vesicles, and the cross-fractured PC unilamellar vesicle (right). Bar, 200 nm.

formula
rearranging:
formula
thus:
formula

Comparing the calculated value (0.35) with the experimental data (0.25), we conclude that at least 70% of PC bound to the Pt/C replica remains after SDS treatment.

We next applied the SDS-FRL technique to the immunolabeling of the human erythrocytes with JE-1. SDS (2.5%; 10 minutes at room temperature) dissolved most unfractured plasma membranes and released haemoglobin. The ultrastructure of the exoplasmic fracture face (E-face) and protoplasmic fracture face (P-face) of the erythrocyte membrane as revealed by SDS-FRL was equivalent to that of conventional freezefracture replicas cleaned by treatment with household bleach. A homogeneous distribution of intramembrane particles (IMPs) on the P- and E-faces was observed. More IMPs were visible on the P-face than on the E-face. The replicas of erythrocytes, which were labeled with JE-1, showed that the Efaces were densely labeled (Fig. 4a, EF), while the P-faces were virtually unlabeled (Fig. 4a, PF). The immunogold particles bound to the exoplasmic membrane half, which remains associated with the replica after the SDS digestion and labeling procedures, could be readily recognized through the replica (see Fig. 1d). Thus, the cytochemical labeling was seen superimposed on the image of an E-face.

Fig. 4.

SDS-FRL of human erythrocytes with JE-1. (a) Labeling of intact erythrocytes. The replicas of intact erythrocytes show that the Efaces (EF) are densely labeled with immunogold particles, while the P-faces (PF) are virtually unlabeled. (b) Labeling of 1 mM calcium/20 µM ionophore treated erythrocytes (10 minutes, 37°C). In contrast to intact erythrocytes, the labeling is observed on both the P-face (PF) and E-face (EF) revealed as smooth areas denuded of intramembrane particles. Cell attachment is frequently observed (arrowheads). (c) SDS-FRL of intact erythrocytes treated with phospholipase A2. The immunogold labeling is completely diminished. Bars, 100 nm.

Fig. 4.

SDS-FRL of human erythrocytes with JE-1. (a) Labeling of intact erythrocytes. The replicas of intact erythrocytes show that the Efaces (EF) are densely labeled with immunogold particles, while the P-faces (PF) are virtually unlabeled. (b) Labeling of 1 mM calcium/20 µM ionophore treated erythrocytes (10 minutes, 37°C). In contrast to intact erythrocytes, the labeling is observed on both the P-face (PF) and E-face (EF) revealed as smooth areas denuded of intramembrane particles. Cell attachment is frequently observed (arrowheads). (c) SDS-FRL of intact erythrocytes treated with phospholipase A2. The immunogold labeling is completely diminished. Bars, 100 nm.

We conclude that the absence of PC immunolabeling on the P-faces implies a nearly complete asymmetrical transmembrane distribution of PC in the erythrocyte membranes. However, there is an alternative explanation for the absence of PC immunolabeling on P-faces; a peripheral membrane protein and/or a cytosolic protein matrix, e.g. membrane cytoskeleton, remain attached to the protoplasmic membrane half after SDS-treatment, masks PC and inhibits antibody binding. To exclude this possibility we incubated the SDS-treated replicas of erythrocytes with protease (0.1 mg/ml trypsin or 1 µg/ml proteinase K) for 10 minutes at 37°C prior to cytochemical labeling. No difference in PC immunolabeling was observed (data not shown). Although this may support our conclusion, the influence of integral membrane proteins on PC immunolabeling is still debatable.

The elevation of the cytoplasmic Ca2+ level is currently assumed to activate a pathway for transbilayer diffusion of phospholipids, and to cause the disruption of lipid asymmetry of the plasma membrane of erythrocytes (Williamson et al., 1992). To re-evaluate the effect of Ca2+ on the transmembrane phospholipid distribution by the SDS-FRL with JE-1, erythrocytes were incubated with Hepes buffer containing 1 mM CaCl2 and 20 µM A23187 (Ca2+/ionophore) at 37°C for various times, frozen, and subjected to SDS-FRL for PC. Incubation with Ca2+/ionophore altered the fracture faces markedly. In contrast to intact erythrocytes, the IMPs of the P- and E-faces of Ca2+/ionophore-treated erythrocytes were not distributed randomly but were grouped in clusters. In addition to this morphological alteration, roughly equal amounts of immunogold particles for PC were seen on each membrane half in less than 10 minutes (Fig. 4b). These gold particles were confined to the intervening smooth areas between the clusters of IMPs. Cell attachment was frequently observed, and occurred only between the smooth areas denuded of IMP (Fig. 4b, hollow arrows). Incubation with either Ca2+ or ionophore alone did not affect the asymmetrical distribution of PC and did not change the freeze-fracture morphology (data not shown). The specificity of the PC labeling was confirmed through several control experiments. In replicas treated with phospholipase A2 for 1 hour at 37°C prior to cytochemical labeling, the labeling of the E-face was completely diminished (Fig. 4c). A similar result was obtained when the primary antibody was adsorbed with excess amounts of pure PC multilammellar liposomes prior to immunocytochemical labeling, or the primary antibody was omitted in the staining procedure (data not shown). These initial experiments conducted on erythrocyte membranes show the asymmetrical transmembrane distribution of PC and its redistribution under a certain condition, and provide compelling evidence for our working hypothesis, that membrane phospholipids of split membrane halves remain attached to the Pt/C replicas even after SDS-treatment.

Next, we examined the distribution of PC in various cells and intracellular organelles. The fine structure of the fracture face observed in the Pt/C replicas of various cells and tissues subjected to SDS-FRL was the same as that of the conventional freeze-fractured faces. Therefore, although the immunogold particles were actually localized on the surface of carbon-stabilized membrane halves attached to the replicas, we used here the conventional nomenclature of freeze-fractured faces (Branton et al., 1975). In this nomenclature, the membrane leaflet (half) closer to the cytoplasm, nucleoplasm, or mitochondrial plasm (mitochondrial matrix), is designated as the protoplasmic membrane leaflet or P leaflet, and the membrane leaflet closer to the exoplasmic, endoplasmic, or perinuclear space (cisterna) is designated as the exoplasmic leaflet or E leaflet. Thus the fractured and replicated membrane surfaces of P leaflet and E leaflet are P- and E-faces, respectively.

SDS-FRL using JE-1 revealed that the E-faces of most intracellular membranes of various cell organelles, e.g. nuclei (Fig. 5b), mitochondria (Fig. 5c), secretory granules (Fig. 5d), endoplasmic reticulum (Fig. 5e), and discs of the outer segment of photoreceptor cells (Fig. 5f), and the E-faces of plasma membranes (Figs 5a,g and 6a) except for apical plasma membranes were densely immunolabeled, while the P-faces of these membranes generally displayed weaker labeling (Fig. 5a and h). All the results on the transmembrane distribution of PC obtained from the SDS-FRL clearly showed the preferential localization of PC to the exoplasmic membrane half (leaflet) of the cellular membrane. One exception to this general scheme was the apical (luminal) plasma membrane of glandular epithelial cells such as pancreatic acinar cells and the plasma membrane forming the myelin of neurons.

Fig. 5.

Electron micrographs of rat hepatocytes (a-c), rat pancreatic acinar cells (d,e), and mouse retinal photoreceptor cells (f-h) processed for SDS-FRL with anti-PC monoclonal antibody, JE-1. The E-faces (EF) of most cellular membranes of intracellular organelles, e.g. nuclei (b), mitochondria (c), secretory granules (zymogen granules) (d), endoplasmic reticulum (e), and discs of the outer segment of photoreceptor cells (light-adapted; f) are densely labeled. The E-faces (EF) of most plasma membranes (a and g) are exclusively labeled. (a) This micrograph depicts the plasma membranes of adjacent hepatocytes. The fracture plane has jumped from one membrane to an adjacent membrane, thus both the E- (EF) and P-faces (PF) are present side by side. (b) The nuclear envelope consists of inner and outer nuclear membranes that are continuous at the margin of nuclear pores (arrowheads). The cavity between the inner and outer nuclear membranes is called the perinuclear space or cisterna (arrows). The fracture plane jumps from the less particulate E-face (EF) of the outer nuclear membrane to the more particulate P-face (PF) of the inner nuclear membrane. (c) Mitochondria are composed of two membranes, and in this freeze-fracture both the E-face (EF) of the outer mitochondrial membrane and the P-face (PF) of the inner mitochondrial membrane are present. More IMPs remain associated with the P-face than with the E-face. (d) The gently curving, convex E-face (EF) of the secretory granule (zymogen granule) membrane is easily distinguished from the concave P-face of the secretory granule membrane. (e) Oblique freeze-fracture through long narrow cisternae (arrowheads) of rough endoplasmic reticulum of rat pancreatic acinar cell. The cytoplasmic leaflet (PF) retains the majority of the IMPs, whereas the exoplasmic leaflet (EF) is largely devoid of particles. Cytoplasm (cyt) is present around the cisternae. (f) This micrograph illustrates the flattened membranous discs that are stacked in the rod outer segment of the photoreceptor cell (light-adapted). The E-faces (EF) of the disc membranes are densely labeled. (g and h) These micrographs show the labeling of JE-1 on the plasma membranes of mouse photoreceptor cells. There are two types of fracture faces associated with the outer segment plasma membrane in the retina, one concave E-face (g) and one convex P-face (h). As previously reported (Sjöstrand and Kreman, 1979), the P-faces show a particulate structure with dense arrangement of the particles in the range 8-10 nm in diameter. The particulate surface is interrupted by a smooth surface structure (h, arrows). The concave E-faces show some structure with particularly smooth areas (g, square brackets) surrounded by areas exposing a very fine particulate structure. The labeling is revealed as aggregates of immunogold particles on the particularly smooth areas (g, square brackets), while no labeling is observed on the P-face (h). There is no difference in the labeling of PC on the disc membranes and the plasma membranes between light- and dark-adapted photoreceptor cells (data not shown). Arrowheads indicate the cross-fractured disc membranes. Bars, 100 nm.

Fig. 5.

Electron micrographs of rat hepatocytes (a-c), rat pancreatic acinar cells (d,e), and mouse retinal photoreceptor cells (f-h) processed for SDS-FRL with anti-PC monoclonal antibody, JE-1. The E-faces (EF) of most cellular membranes of intracellular organelles, e.g. nuclei (b), mitochondria (c), secretory granules (zymogen granules) (d), endoplasmic reticulum (e), and discs of the outer segment of photoreceptor cells (light-adapted; f) are densely labeled. The E-faces (EF) of most plasma membranes (a and g) are exclusively labeled. (a) This micrograph depicts the plasma membranes of adjacent hepatocytes. The fracture plane has jumped from one membrane to an adjacent membrane, thus both the E- (EF) and P-faces (PF) are present side by side. (b) The nuclear envelope consists of inner and outer nuclear membranes that are continuous at the margin of nuclear pores (arrowheads). The cavity between the inner and outer nuclear membranes is called the perinuclear space or cisterna (arrows). The fracture plane jumps from the less particulate E-face (EF) of the outer nuclear membrane to the more particulate P-face (PF) of the inner nuclear membrane. (c) Mitochondria are composed of two membranes, and in this freeze-fracture both the E-face (EF) of the outer mitochondrial membrane and the P-face (PF) of the inner mitochondrial membrane are present. More IMPs remain associated with the P-face than with the E-face. (d) The gently curving, convex E-face (EF) of the secretory granule (zymogen granule) membrane is easily distinguished from the concave P-face of the secretory granule membrane. (e) Oblique freeze-fracture through long narrow cisternae (arrowheads) of rough endoplasmic reticulum of rat pancreatic acinar cell. The cytoplasmic leaflet (PF) retains the majority of the IMPs, whereas the exoplasmic leaflet (EF) is largely devoid of particles. Cytoplasm (cyt) is present around the cisternae. (f) This micrograph illustrates the flattened membranous discs that are stacked in the rod outer segment of the photoreceptor cell (light-adapted). The E-faces (EF) of the disc membranes are densely labeled. (g and h) These micrographs show the labeling of JE-1 on the plasma membranes of mouse photoreceptor cells. There are two types of fracture faces associated with the outer segment plasma membrane in the retina, one concave E-face (g) and one convex P-face (h). As previously reported (Sjöstrand and Kreman, 1979), the P-faces show a particulate structure with dense arrangement of the particles in the range 8-10 nm in diameter. The particulate surface is interrupted by a smooth surface structure (h, arrows). The concave E-faces show some structure with particularly smooth areas (g, square brackets) surrounded by areas exposing a very fine particulate structure. The labeling is revealed as aggregates of immunogold particles on the particularly smooth areas (g, square brackets), while no labeling is observed on the P-face (h). There is no difference in the labeling of PC on the disc membranes and the plasma membranes between light- and dark-adapted photoreceptor cells (data not shown). Arrowheads indicate the cross-fractured disc membranes. Bars, 100 nm.

In SDS-FRL of pancreatic acinar cells with JE-1, both the P- and E-faces of apical plasma membranes were virtually unlabeled (Fig. 6a). These findings are consistent with previous biochemical studies (Simons and van Meer, 1988) on the purified apical membrane fractions, suggesting that the apical plasma membrane is poor in PC, when compared to the basolateral plasma membrane. The plasma membrane of epithelial cells is divided into an apical and a basolateral domain by tight junctions (Fig. 6, TJ) that encircle the apex of each cell. Because it seemed likely that the tight junctions block membrane lipid diffusion, we next studied glycerol-impregnated material. Radical, qualitative changes in the ultrastructure of the tight junctions are observed as a result of differences in fixation and glycerol-impregnation (Pinto da Silva and Kachar, 1982). Thus, we expected that a barrier formed by the tight junctions to the diffusion of lipids in the exoplasmic (outer) membrane half might be affected by glycerol-impregnation. Interestingly, in chemically unfixed, glycerol-impregnated pancreatic acinar cells, the PC-immunogold complexes were detected on the E-faces of the apical plasma membranes as well as on the E-faces of the lateral plasma membranes (Fig. 6b). This implies that the tight-junctional barrier is broken by glycerol-impregnation, thus PC can freely diffuse between the lateral and apical plasma membranes.

Fig. 6.

SDS-FRL of rat pancreatic acinar cells (a,b) and myelin sheath (c) with JE-1. The labeling is not seen on the E-face of apical plasma membranes of the intact pancreatic acinar cells (a, ap), while the labeling can be observed on the E-face of that of unfixed, glycerol-impregnated (final concentration 30%; for 1 hour at 4°C) cells (b, ap). Arrow indicates desmosomes. (c) Labeling of myelin from rat optic nerve with JE-1. Although most of the fractured face of myelin exposed by freezefracturing are labeled, the P-face of the plasma membrane of the axon (AX) is not labeled with JE-1 (arrowheads). Bars, 100 nm.

Fig. 6.

SDS-FRL of rat pancreatic acinar cells (a,b) and myelin sheath (c) with JE-1. The labeling is not seen on the E-face of apical plasma membranes of the intact pancreatic acinar cells (a, ap), while the labeling can be observed on the E-face of that of unfixed, glycerol-impregnated (final concentration 30%; for 1 hour at 4°C) cells (b, ap). Arrow indicates desmosomes. (c) Labeling of myelin from rat optic nerve with JE-1. Although most of the fractured face of myelin exposed by freezefracturing are labeled, the P-face of the plasma membrane of the axon (AX) is not labeled with JE-1 (arrowheads). Bars, 100 nm.

The myelin sheath of myelinated nerves is a morphologically and biochemically specialized membranous structure, as it contains exceptionally high levels of lipid, and very low levels of protein (Guidotti, 1972). Thus, this could be a useful model for our study. Interestingly, the immunogold labeling of PC was observed on both membrane halves (Fig. 6c), suggesting that the myelin sheath, with the symmetrical distribution of phospholipids, at least PC, is a unique example.

Although Ca2+ is involved in many biological membrane fusion phenomena (Hoekstra et al., 1985), the mechanism of its action is not well understood. When erythrocytes were incubated with Ca2+ and ionophore, many cells came into sufficiently close contact, but the adjacent plasma membranes were not fused. In addition to the morphological event, at this stage, PC is distributed evenly in both membrane halves. Another example is the myelin sheath of the myelinated nerve with its indefinitely stable, extensive and extremely close contact of adjacent cell membranes. The SDS-FRL with JE-1 showed that in the myelin, the phospholipid distribution is almost symmetrical. These findings suggest, that the symmetrical distribution of phospholipids, at least PC, may be adapted to cell membrane contact. Further analysis on the distribution and movement of phospholipids using SDS-FRL, which occurs during membrane contact and fusion induced by various fusogens, will help elucidate the chemical property of membranes fusing to adjacent membranes.

Conventional freeze-fracture replica electron microscopy reveals the precise macromolecular architecture of the membranes, but offers little information on the biochemical composition of cell membrane components. The SDS-FRL technique overcomes this obstacle, and may be useful both for ultrastructural observation and for the membrane lipid and protein cytochemistry of the biological membranes. In this study, SDS-FRL revealed that PC is exclusively localized on the exoplasmic membrane halves of the plasma membranes, and the intracellular membranes of various cell organelles. These findings seem to imply a nearly complete asymmetrical transmembrane distribution of PC in the membranes. However, there is an alternative explanation for the absence of labeling on the protoplasmic membrane halves; a peripheral membrane protein, a cytosolic protein matrix (e.g. membrane cytoskeleton), or an integral membrane protein, remains attached to the protoplasmic membrane half after SDS-treatment, masks PC and inhibits antibody binding. We cannot exclude this possibility at the present time.

We gratefully acknowledge the helpful discussions with Drs Toru Noda and Masayuki Murata. We thank Drs Chizuka Ide and Kazuo Ogawa for their encouragement throughout this study. We thank Miss Sawako Nakamura and Kasumi Nomura for their secretarial assistance, and Mr Akira Uesugi for his help with the photographic work. We, and Dr K. Fujimoto in particular, want to express appreciation to Dr Masako Fujimoto for her excellent technical assistance. This work was supported by a research grant from the Ministry of Education, Science and Culture of Japan (to K. Fujimoto).

Andersson-Forsman
,
C.
and
Pinto da Silva
,
P.
(
1988
).
Fracture-flip: new high-resolution images of cell surfaces after carbon stabilization of freeze-fractured membranes
.
J. Cell Sci
.
90
,
531
541
.
Branton
,
D.
,
Bullivant
,
S.
,
Gilula
,
N. B.
,
Karnovsky
,
M. J.
,
Moor
,
H.
,
Muhlethaler
,
K.
,
Northcote
,
D. H.
,
Packer
,
L.
,
Satir
,
P.
,
Speth
,
V.
,
Staehelin
,
L. A.
,
Steere
,
R. L.
and
Weinstein
,
R. S.
(
1975
).
Freeze-etching nomenclature
.
Science
190
,
54
56
.
Fujimoto
,
K.
and
Pinto da Silva
,
P.
(
1988
).
Macromolecular dynamics of the cell surface during the formation of coated pits is revealed by fracture-flip
.
J. Cell Sci
.
91
,
161
173
.
Fujimoto
,
K.
and
Pinto da Silva
,
P.
(
1989
).
Surface views of nuclear pores in isolated rat liver nuclei as revealed by fracture-flip/Triton-X
.
Eur. J. Cell Biol
.
50
,
390
397
.
Fujimoto
,
K.
and
Ogawa
,
K.
(
1991
).
Fracture-flip and fracture-flip cytochemistry to study macromolecular architecture of membrane surfaces: practical procedures, interpretation and application
.
Acta Histochem. Cytochem
.
24
,
111
117
.
Fujimoto
,
K.
and
Pinto da Silva
,
P.
(
1992
).
Fracture-flip/Triton X-100 reveals the cytoplasmic surface of human erythrocyte membranes
.
Acta Histochem. Cytochem
.
25
,
255
263
.
Fujimoto
,
K.
(
1995
).
Freeze-fracture replica electron microscopy combined with SDS digestion for cytochemical labeling of integral membrane proteins. Application to the immunogold labeling of intercellular junctional complexes
.
J. Cell Sci
.
108
,
3443
3449
.
Guidotti
,
G.
(
1972
).
Membrane proteins
.
Annu. Rev. Biochem
.
41
,
731
752
.
Heuser
,
J. E.
,
Reese
,
T. S.
,
Jan
,
L. Y.
,
Dennis
,
M. J.
and
Evans
,
L.
(
1979
).
Synaptic vesicle exocytosis captured by quick-freezing and correlated with quantal transmitter release
.
J. Cell Biol
.
81
,
275
300
.
Hoekstra
,
D.
,
Wilschut
,
J.
and
Scherphof
,
G.
(
1985
).
Fusion of erythrocyte ghosts induced by calcium phosphate. Kinetic characteristics and the role of Ca2+, phosphate and calcium-phosphate complexes
.
Eur. J. Biochem
.
146
,
131
140
.
Kinsky
,
S. C.
,
Haxby
,
J. A.
,
Zopf
,
D. A.
,
Alving
,
C. R.
and
Kinsky
,
C. B.
(
1969
).
Complement-dependent damage to liposomes prepared from pure lipids and Forssman hapten
.
Biochemistry
8
,
4149
4158
.
Nam
,
K. S.
,
Igarashi
,
K.
,
Umeda
,
M.
and
Inoue
,
K.
(
1990
).
Production and characterization of monoclonal antibodies that specifically bind to phosphatidylcholine
.
Biochim. Biophys. Acta
1046
,
89
96
.
Pinto da Silva
,
P.
and
Kachar
,
B.
(
1982
).
On tight-junction structure
.
Cell
28
,
441
450
.
Schrier
,
S. L.
,
Zachowski
,
A.
,
Herve
,
P.
,
Kader
,
J.-P.
and
Devaux
,
P. F.
(
1992
).
Transmembrane redistribution of phospholipids of the human red cell membrane during hypotonic hemolysis
.
Biochim. Biophys. Acta
1105
,
170
176
.
Simons
,
K.
and
van Meer
,
G.
(
1988
).
Lipid sorting in epithelial cells
.
Biochemistry
27
,
6197
6202
.
Sjöstrand
,
F. S.
and
Kreman
,
M.
(
1979
).
Freeze-fracture analysis of structure of plasma membrane of photoreceptor cell outer segments
.
J. Ultrastruct. Res
.
66
,
254
275
.
Williamson
,
P.
,
Kulick
,
A.
,
Zachowski
,
A.
,
Schlegel
,
R. A.
and
Devaux
,
P. F.
(
1992
).
Ca2+ induces transbilayer redistribution of all major phospholipids in human erythrocytes
.
Biochemistry
31
,
6355
6360
.
Zachowski
,
A.
(
1993
).
Phospholipids in animal eukaryotic membranes: transverse asymmetry and movement
.
Biochem. J
.
294
,
1
14
.
Zhou
,
X.
and
Arthur
,
G.
(
1992
).
Improved procedures for the determination of lipid phosphorus by malachite green
.
J. Lipid Res
.
33
,
1233
1236
.