Increased ion permeability of planar lipid bilayer membranes after treatment with the C5b-9 cytolytic attack mechanism of complement.

The ion permeability of planar lipid bilayers, as measured electrically, was found to increase modestly upon treatment with purified complement complex C5b,6 and complement components C7 and C8. The subsequent addition C9 greatly amplified this change. No permeability changes occurred when components were added individually to the membrane, or when they were used in paired combinations, or when C5b, C7, C8, and C9 were admixed prior to addition. Thus, there is a significant parallel between the permeability changes induced in the model membrane and damage produced in biological membranes by the C5b-9 complement attack sequence. The efficiency of membrane action by C5b-9 was critically dependent on the order in whcih components were added to the membrane. There were also differences in the electrical properties of membranes treated with C5b-8 and C5b-9, though in both cases the enhanced bilayer permeability is best attributed to the formation of trans-membrane channels. Collectively, the data are consistent with the hypothesis that the mechanism of membrane action by complement involves the production of a stable channel across the lipid bilayer, resulting in cell death by colloid-osmotic lysis.     

Transient changes in erythrocyte membrane permeability are induced by sublytic amounts of the complement membrane attack complex (C5b-9)

We have previously shown that sublytic heterologous complement induces large but transient increases in erythrocyte membrane permeability. We now report that when erythrocytes are bystanders in zymosan-activated autologous serum, they increase their Na+ permeability 10-fold, indicating that autologous complement can also induce transient membrane lesions. When we isolated the effect of the C5b-9 membrane attack complex of complement by using human C5b-9 assembled from purified components, we found there was minimal lysis but efficient Na+ uptake. Suspension of the sublytically damaged erythrocytes in K+ medium caused the cells to lyse, which is consistent with the cells recruiting a compensatory K+ efflux similar to that observed when human erythrocytes were exposed to heterologous complement. Sublytic C5b-9 exposure also became lytic when extracellular Ca2+ was limited and when the cells were exposed to charybdotoxin, an inhibitor of the Ca(2+)- activated K+ channel. This indicates that Ca2+ is required for the functional termination of the C5b-9 lesion. We also show that the membrane hyperpolarization resulting from activation of the Ca(2+)- dependent K+ efflux does not influence the termination of the C5b-9 lesion. Thus, the influx of Ca2+ through the complement lesion initiates at least two apparently independent adaptive responses: (1) a process that terminates the leak; and (2) a K+ efflux that has a volume regulatory function. Our data support the potential of the sublytic C5b- 9 lesion to act as a physiologic mediator for autologous erythrocytes.     

Interaction between terminal complement proteins C5b-7 and anionic phospholipids

We have recently shown that C5b-6 binds to the erythrocyte membrane via an ionic interaction with sialic acid before the addition of C7 and subsequent membrane insertion. In this study we assessed the role of anionic lipids in the binding of the terminal complement proteins to the membrane and the efficiency of subsequent hemolysis. Human erythrocytes were modified by insertion of dipalmitoyl phosphatidylcholine (DPPC), dipalmitoyl phosphatidylserine (DPPS), dipalmitoyl phosphatidylethanolamine (DPPE), or dipalmitoyl phosphatidic acid (DPPA). Lipid incorporation and the hemolytic assays were done in the presence of 100 micromol/L sodium orthovanadate to prevent enzymatic redistribution of lipid. We found that the neutral lipids, DPPC and DPPE, did not affect C5b-7 uptake or hemolysis by C5b-9. In contrast, the two acidic phospholipids, DPPS and DPPA, caused a dose-dependent increase in both lysis and C5b-7 uptake. We conclude that the presence of anionic lipids on the exterior face of the membrane increases C5b-7 uptake and subsequent hemolysis. It is known that sickle cell erythrocytes have increased exposure of phosphatidylserine on their external face and are abnormally sensitive to lysis by C5b-9. The data presented here provide a plausible mechanism for this increased sensitivity.     

Homologous species restriction in lysis of erythrocytes by terminal complement proteins.

The cytolytic efficiency of the terminal complement protein complex, C5b-9, varies with the species of origin of C8 and C9. In the present study, we explored the susceptibility of erythrocytes from various species to lysis by C5b6,7 plus C8 and C9 from different species. EC5b6,7 intermediates were prepared on human, guinea pig, rabbit, mouse, and rat erythrocytes with human C5b6 and guinea pig C7. The degree of lysis of these intermediates by C8 and C9 was found to vary widely depending on the species of the proteins and the target cells. In all cases, lysis was least efficient when C8 and C9 were homologous with respect to the target cell species. This effect was mostly attributable to C9. The inefficient lysis in a homologous system is not due to a failure of C9 binding. Rather, the poor lysis in the homologous system may be attributable to inefficient insertion or channel formation.     

Isolation of a human erythrocyte membrane protein capable of inhibiting expression of homologous complement transmembrane channels.

Erythrocytes are poorly lysed by homologous complement, whereas they are readily lysed by heterologous complement. This phenomenon had been attributed to an interference by the cell surface with the action of complement components C8 and C9. To isolate the responsible membrane constituent, detergent-solubilized human erythrocyte (EH) membranes were subjected to affinity chromatography by using human C9-Sepharose. The isolated protein had a mass of 38 kDa and, incorporated into liposomes, was highly effective in inhibiting complement-mediated channel expression, including the C5b-8, membrane attack complex, and tubular polymer of C9 channels. Antibody produced to the 38-kDa protein caused a 20-fold increase in reactive lysis of EH by isolated C5b6, C7, C8, and C9. The antibody did not enhance C5b-7 uptake, but it affected C9 binding to the target cell membrane. Antibody to human decay-accelerating factor, used as a control, had no effect on reactive lysis of EH. Anti-38-kDa protein did not enhance the action on EH of C8 and C9 from other species, indicating that the action of this regulatory protein is species specific. It was therefore termed homologous restriction factor (HRF). Blood cells other than erythrocytes, such as polymorphonuclear leukocytes, also exhibited cell-surface HRF activity. In immunoblots of freshly isolated EH membranes, anti-38-kDa HRF detected primarily a 65-kDa protein, suggesting that the 38-kDa protein constitutes an active fragment of membrane HRF. Because of the specific binding reaction observed between HRF and C8 or C9, HRF was tested with anti-human C8 and anti-human C9. A limited immunochemical relationship of HRF to C8 and C9 could be established and solid-phase anti-C9 proved an efficient tool for the isolation of HRF from solubilized EH membranes.     

Defective regulation of complement by the sickle erythrocyte: evidence for a defect in control of membrane attack complex formation

A prominent clinical manifestation of sickle cell disease (SCD) is hemolytic anemia. Although complement activation can lead to intravascular hemolysis, its role in the hemolysis of SCD is not known. Because normal red blood cells induced to vesiculate by treatment with calcium and ionophore become sensitive to damage by activated complement and because sickle cells release microvesicles as they circulate, we postulated that sickle cells might also be unusually sensitive to complement-dependent hemolysis. Complement activation is tightly regulated on the membrane of the normal erythrocyte; therefore, defective complement regulation by the sickle cell would be necessary for complement-dependent hemolysis to occur. These studies show a defect in the regulation of membrane attack complex (C5b-9) formation in sickle erythrocytes, particularly in the most dense cells. The defect is characterized by increased binding of C5b-7 and of C9 to denser sickle cells and results in increased susceptibility of sickle cells to C5b-9-mediated (reactive) lysis initiated by either C5b6 or activated cobra venom factor. Among the densest sickle cells, irreversibly sickled cells are especially sensitive to reactive lysis. The similarity of this defect to that previously described in a patient with paroxysmal nocturnal hemoglobinuria suggests that complement- mediated hemolysis could play a role in the anemia of SCD.     

Electron spin resonance studies on interaction of complement proteins with erythrocyte membranes.

Sheep erythrocytes have been spin labeled with 5-, 12-, and 16-nitroxystearic acid in order to investigate complement-induced changes in the physical state of the lipid bilayer. Formation of osmotic lesions in the membrane causes an increase in the fluidity of the membrane which overcomes the decrease in membrane fluidity caused by the interaction of the complement proteins. A decrease in membrane fluidity is observed only when complement-lysed membranes are resealed or when complement proteins react with isosmolar ghosts that do not undergo osmotic lysis. The decrease in bulk fluidity of the membrane is first observed when C8 binds to the membranes bearing C5b67 and is enhanced upon the subsequent binding of C9. The decrease in membrane fluidity shown by the electron spin resonance spectra of spin-labeled fatty acids suggests that certain of the complement proteins penetrate the membrane and interact with hydrophobic regions of the lipid bilayer.     

Complement-induced vesiculation and exposure of membrane prothrombinase sites in platelets of paroxysmal nocturnal hemoglobinuria

Paroxysmal nocturnal hemoglobinuria (PNH) is an acquired stem-cell disorder in which the glycolipid-anchored membrane proteins, including the cell-surface complement inhibitors, CD55 and CD59, are partially or completely deleted from the plasma membranes of mature blood cells. To gain insight into the pathogenesis of thrombosis that is frequently observed in this disorder, the procoagulant responses of PNH platelets exposed to the human terminal complement proteins C5b-9 were investigated. C5b-9 complexes were assembled on gel-filtered platelets by incubation with purified C5b6, C7, C9, and limiting amounts of C8. Platelet microparticle formation and exposure of plasma membrane- binding sites for coagulation factor Va were then analyzed by flow cytometry. PNH platelets exhibiting undetectable levels of surface CD59 antigen showed an approximately 10-fold increase in sensitivity to C5b- 9-stimulated expression of membrane-binding sites for factor Va when compared with platelets from normal controls. Expression of catalytic surface for the prothrombinase complex (VaXa) paralleled the exposure of factor Va-binding sites; the rate of prothrombin conversion by C5b-9- treated PNH platelets exceeded that of C5b-9-treated normal controls by approximately 10-fold at the maximal input of C8 tested (500 ng/mL). These data indicate that PNH platelets deficient in plasma membrane CD59 antigen are exquisitely sensitive to C5b-9-induced expression of prothrombinase activity, and suggest that the tendency toward thrombosis in these patients may be due, at least in part, to the deletion of this complement inhibitor from the platelet plasma membrane.     

Membrane attack complex of complement: distribution of subunits between the hydrocarbon phase of target membranes and water

Membrane destruction by complement is effected by the membrane attack complex (MAC) which is the dimer of a fusion product of the complement proteins C5b, C6, C7, C8, and C9. Phospholipid bilayer vesicles were used as target membranes for the MAC and its intermediate complexes. The subunits of these membrane-bound complexes were explored as to their relative exposure to the hydrocarbon phase of the lipid bilayer and to water surrounding the lipid vesicles. Protein exposed to the aqueous phase was labeled with 125I; protein exposed to the hydrocarbon phase was labeled by using tritiated azido phospholipids and irradiation. Analysis of the membrane-bound MAC showed that subunits C5b, C8 beta, and C9 were exposed to the aqueous phase. The subunits C8 alpha-gamma and C9 were primarily in contact with the hydrocarbon phase. C6 and C7 were little exposed to either phase, suggesting that these proteins are inaccessible within the MAC. Analysis of the intermediate complexes showed that C5b was the subunit most exposed to water in membrane-bound C5b-7, and C5b and C8 beta were the water-exposed subunits in C5b-8. Subunit exposure to the hydrocarbon phase of the lipid bilayer changed during MAC assembly. Whereas all three subunits of C5b-7 carried the phospholipid photolabel; most of the label was bound to the C8 subunit in C5b-8 and to C9 in the MAC. It is proposed that contact with the hydrocarbon core of membranes is established by C5b-7 through each of its subunits, by C5b-8 through C8, and by the MAC through C8 and, particularly, C9.     

On the lysis of paroxysmal nocturnal hemoglobinuria erythrocytes by complement: Dual role of C3b

Summary The efficiency of cytolysis by the terminal complement proteins C5b-9 can be markedly enhanced by C3b molecules bound on the target cell membrane (Hammer et al. 1976). This enhancement was shown to be proportional to the number of C3b molecules on the cell membrane. The present experiments have shown that the hemolytic efficiency of the complement membrane attack system is two to five times greater on paroxysmal nocturnal hemoglobulinuria erythrocytes (PNHE) than on normal human E. This difference is attribut to a derivative of C3, probably C3b, on PNHE since it was abolished by anti-C3 but not by anti-C2. The efficiency of C5b-9 to lyse PNHE was only partially decreased by C3b inactivator and 1H, indicating that the C3b on PNHE is not readily inactivated by its regulatory proteins. Furthermore, cells from a single severely affected patient consumed 3-fold more C5b6 than normal human E yet concommitantly measured membrane fluidity was normal. From these observations we conclude that cell-bound C3b on PNHE serves two functions: (a) it increases the hemolytic efficiency of membrane attack components of the complement system; and (b) it provides sites for assembly of the alternative pathway convertases.     

The cytolytic C5b-9 complement complex: feedback inhibition of complement activation

We describe a regulatory function of the terminal cytolytic C5b-9 complex [C5b-9(m)] of human complement. Purified C5b-9(m) complexes isolated from target membranes, whether in solution or bound to liposomes, inhibited lysis of sensitized sheep erythrocytes by whole human serum in a dose-dependent manner. C9 was not required for the inhibitory function since C5b-7 and C5b-8 complexes isolated from membranes were also effective. No effect was found with the cytolytically inactive, fluid-phase SC5b-9 complex. However, tryptic modification of SC5b-9 conferred an inhibitory capacity to the complex, due probably to partial removal of the S protein. Experiments using purified components demonstrated that C5b-9(m) exerts a regulatory effect on the formation of the classical- and alternative-pathway C3 convertases and on the utilization of C5 by cell-bound C5 convertases. C5b-9(m) complexes were unable to inhibit the lysis of cells bearing C5b-7(m) by C8 and C9. Addition of C5b-9(m) to whole human serum abolished its bactericidal effect on the serum-sensitive Escherichia coli K-12 strain W 3110 and suppressed its hemolytic function on antibody-sensitized, autologous erythrocytes. Feedback inhibition by C5b-9(m) represents a biologically relevant mechanism through which complement may autoregulate its effector functions.     

On the mechanism of cytolysis by complement: evidence on insertion of C5b and C7 subunits of the C5b,6,7 complex into phospholipid bilayers of erythrocyte membranes.

The doughnut hypothesis of cytolysis by complement [Mayer, M. M. (1972) Proc. Nat. Acad. Sci. USA 69, 2954-2958] describes an annular structure made up of C5b-9 (complement factors C5b, C6, C7, C8, and C9) which becomes inserted in the lipid bilayer of the cell membrane, thus creating a hole. We now present initial explorations of this hypothesis. EAC1-6 and EAC1-7 (sheep erythrocytes carrying rabbit antibody and complement factors C1 through C6 or C1 through C7, respectively), prepared with either 125I-C3 or 125I-C5 were incubated with trypsin and the release of bound 125I was measured. In the case of 125I-C3, all of the radioactivity was released by trypsin from both intermediates. With 125I-C5, trypsin released all of the 125I from EAC1-6, but only 40-55% from EAC1-7. Possible reasons for resistance of the C5b subunit in EAC1-7 to tryptic digestion are discussed; in terms of the doughnut hypothesis it would be due to shielding by lipid molecules as a consequence of insertion into the lipid bilayer. In accord with this interpretation we have also found that C5b in EAC1-7, but not in EAC1-6, resists elution by 0.3 M NaC1. Similarly, we have found that 125I-C7 in EAC1-7 resists stripping by trypsin. Hence, we now propose the hypothesis that hydrophobic polypeptide chains from the C5b and the C7 subunits of C5b,6,7 complex become inserted in the phospholipid bilayer and that subsequent reactions with C8 and C9 open a channel across the membrane.     

Restriction of cell lysis by homologous complement: II. Protection of erythrocytes against lysis by newly activated complement

Our previous work revealed that homologous complement (C) was ineffective in lysing antibody-sensitized erythrocytes (EA) even at high concentrations. It was also shown that activation of complement on homologous EA resulted in the binding of C9 and the formation of EA bearing complement proteins C1 through C9 (EAC1-9), yet few hemolytic sites were formed. Instead, as shown here, the formation of homologous EAC1-9 caused the cells to become resistant to lysis even by heterologous complement during a second incubation. In contrast, when homologous EAC1-8 were produced by incubating EA with C9-depleted serum, such intermediates were not protected against lysis by heterologous complement during a second incubation. Furthermore, homologous C9 on EAC1-9 was able to reduce the hemolytic efficiency of heterologous complement without blocking C activation and the formation of new C5b-9 complexes. Protection was not modified when homologous EAC1-9 were produced in one step, by incubation of EA with serum, or sequentially by adding C9 to EAC1-8. The minimum number of 9-sites required to confer a protective effect on EAC1-9 was less than 200 per cell. Thus, in addition to its known effect in heterologous cell killing, homologous C9 is capable of protecting homologous cells against inadvertent complement lysis.     

C8 binding protein bears I antigenic determinants

Summary C8 binding protein (C8bp) is an integral membrane glycoprotein of peripheral blood cells, which inhibits the C5b-9-mediated lysis in a homologous system. In the present study, we analyzed the carbohydrate portion of the C8bp. We found that C8bp is associated with I antigenic determinant, a sugar sequence found on human erythrocytes of adults. To assess whether or not the sugar residues are essential for the C8bp function, I determinant was cleaved off from the isolated C8bp by endo--galactosidase (E.C. 3.2.2.103) that hydrolyses internal -galactosidic-linked-N-acetyllactosamine residues. Enzyme treatment removed I-antigen, the inhibitory function of C8bp, however, was not affected. When intact erythrocytes were treated with endo--galactosidase, I-antigen was lost and the lytic insusceptibility of human erythrocytes to homologous C5b-9 could not be abolished. Thus, I-antigen is associated with the C8bp, but its presence is not required for the homologous species restriction.     

Enhanced complement-mediated lysis of type III paroxysmal nocturnal hemoglobinuria erythrocytes involves increased C9 binding and polymerization.

The interaction of terminal complement proteins (C5-C9) with normal erythrocytes and type III paroxysmal nocturnal hemoglobinuria erythrocytes (PNH-E) has been compared in terms of binding of the C5-9 complex, C9 polymerization, and C9 insertion into membranes. Complement components C5, C7, and C8 bind equally well to both types of erythrocytes, whereas the binding of C9 to PNH-E is 5-6 times greater than that to normal erythrocytes. The kinetics of C9 binding was compared with the kinetics of lysis for both types of cells under conditions leading to 100% lysis. There was a noticeable lag time between C9 binding and lysis of normal erythrocytes, but the lysis of PNH-E proceeded without a lag and the kinetics of lysis more closely paralleled C9 binding. The efficiency of C9 insertion was similar for both types of cells, but C9 polymerization was significantly enhanced on PNH-E. These data indicate that the enhanced susceptibility of type III PNH-E toward lysis by C5-9 can be correlated with abnormally high C9 binding and increased formation of poly(C9).     

Presence of immunoglobulins, C3 and cytolytic C5b-9 complement components on the surface of erythrocytes from patients with β-thalassaemia/HbE disease

The occurrence of IgG, IgM, IgA, C3 and C5b-9 complement complexes on erythrocytes from 43 patients with β-thalassaemia HbE disease was investigated. Indirect immunoradiometric assays using radioiodinated protein A were employed to quantify the individual components. We confirmed that circulating erythrocytes from thalassaemic patients contained elevated amounts of IgG, and small but significant amounts of C3. In addition, small but significant amounts of C5b-9 were detected. Levels of cell-bound IgG, C3 and C5b-9 were higher in splenectomized versus non-splenectomized patients. The presence of C5b-9 on circulating cells from five splenectomized patients was confirmed by an ELISA employing a monoclonal antibody specific for a C5b-9 neoantigen. When C5b-9 positive cells from two patients were solubilized with detergent and subjected to sucrose density gradient centrifugation, the terminal complexes sedimented as 25–40S macromolecules, thus behaving as membrane C5b-9 complexes. The presence of C8 and C9 in these high molecular weight fractions was directly demonstrated by Western blotting. These results constitute the first demonstration that circulating diseased erythrocytes may carry low numbers of potentially cytolytic C5b-9 complement complexes which may be partly responsible for the known ionic disturbances found in thalassaemic cells. Both bound C3 and C5b-9 could promote removal of diseased cells in the reticuloendothelial system.     

Calcium-loaded erythrocytes have a defect in complement regulation distinct from that resulting from exposure to 2- aminoethylisothiouronium bromide

Calcium-loaded red blood cells (RBCs) previously have been shown to have an increased sensitivity to complement-mediated hemolysis and particularly to lysis mediated by the C5b-9 membrane attack complex (MAC) of complement. Because RBCs exposed to 2-aminoethylisothiouronium bromide (AET) also have been shown to be particularly sensitive to the MAC, a direct comparison of calcium-loaded and AET-treated RBCs was performed. Calcium-loaded and AET-treated RBCs shared a marked increase in sensitivity to lysis by the MAC in two different assays. However, measurements of C5b-7 and C9 binding suggested that different mechanisms were responsible. AET-treated RBCs showed an increase in C9 binding and an increased C9/C7 ratio consistent with functional loss of CD59/membrane inhibitor of reactive lysis (MIRL). In contrast, calcium- loaded RBCs had minimally increased C9 binding that resulted in C9/C7 ratios that were less than those for untreated RBCs, suggesting that CD59/MIRL inactivation had not occurred. When RBCs were incubated in acidified serum, AET-treated cells demonstrated a marked increase in C3b binding and hemolysis that was observed in neither control nor calcium-loaded RBCs. These results suggest that the underlying lesions responsible for an increase in susceptibility to complement-mediated hemolysis are different for calcium-loaded and AET-treated RBCs.     

Cytolysis of nucleated cells by complement: cell death displays multi-hit characteristics.

Lysis of nucleated cells by complement was studied to determine whether the lytic process by C5b-9 conforms to a one-hit mechanism as in the case of erythrocytes. Two nucleated cell lines, Molt 4 and U937, derived from human T lymphocytes and histiocytes, respectively, were employed as targets. The antibody-sensitized cells were used to develop the titration curves, measuring cell death as a function of limiting quantities of human C6 or C5,6 complex in the presence of an excess of other complement components. The cytolysis curves generated in both experiments were sigmoidal, in sharp contrast to the monotonic curves observed in lysis of erythrocytes treated similarly. The sigmoidal curves of cytolysis indicate a cooperative action of several molecules of C6 or acid-activated C5,6 complex, C(56)a. In contrast to the multi-hit characteristics of cytolysis, dose-response measurements of the release of 86Rb indicated that only one effective molecule of C6 per cell is required for assembly of a 86Rb-releasing channel. This divergence indicates that lysis requires formation of several channels or, alternatively, assembly of large channels that are formed by several molecules of C6. Because prior studies with erythrocyte ghosts have shown that only a single effective molecule of C6 is required for assembly of a transmembrane channel, regardless of size, we prefer to interpret the multi-hit characteristics of nucleated cell lysis as an indication of a multi-channel requirement, rather than channel enlargement.     

Impaired expression of erythrocyte glycosyl-phosphatidylinositol-anchored membrane CD59 in patients with psoriatic arthritis. Relation to terminal complement pathway activation

OBJECTIVE: Complement-mediated injury is regulated by many factors; among these CD59 has been identified as a widely distributed glycoprotein that inhibits membrane C5b-9 (terminal complement component) formation. The aim of the study was to assess erythrocyte CD59 expression in patients with psoriatic arthritis in order to understand the role of CD59 in the pathogenesis. METHODS: Washed erythrocytes from 50 patients with psoriatic arthritis, 8 with cutaneous psoriasis and 24 healthy subjects were incubated with monoclonal anti-CD59 antibody followed by a second FITC conjugated antibody and fluorescence intensity analysed by FAC-Scan flow cytometer to assess their CD59 membrane expression. SC5b-9 levels were measured in the plasma by ELISA and results compared with CD59 values. Immune complexes, complement C3 and C4 and rheumatoid factor were also determined. RESULTS: Impaired expression of erythrocyte membrane-anchored CD59 was found in patients with psoriatic arthritis; the lowest levels were seen in active patients (p < 0.01). Increased SC5b-9 was seen in the plasma of patients with active disease. An inverse correlation was also found between plasma C5b-9 and the CD59 expression levels (r = -0.81, p < 0.001). CONCLUSION: The low CD59 expression on erythrocytes from patients with psoriatic arthritis may be an index of a low tissue CD59 expression. This impairment could facilitate the activation of complement pathway and increase the risk for arthritis. Membrane attack complex formation in deficient membrane bound CD59 may also exacerbate synovial cell injury and inflammation.     

Evidence for a two-domain structure of the terminal membrane C5b-9 complex of human complement.

Lipid vesicles carrying the purified membrane C5b-9 complex [C5b-9(m)] of complement were analyzed immunochemically and in the electron microscope after treatment with a combination of trypsin and alpha-chymotrypsin. Under reducing conditions, the externally oriented annulus was removed. The remaining part of the C5b-9(m), representing approximately half of the total mass of the macromolecular complex, was visualized in the electron microscope as a hollow cylindrical structure with walls of 1-nm thickness. This structure remained tenaciously attached to the lipid bilayer, projecting 8-9 nm from the external membrane surface into the aqueous environment. Cleavage of C5b-9(m) by proteolysis and reduction resulted in a sharp reduction of tis antigenic determinants. One hydrophilic protease-resistant C5 derivative was released from the membrane and recovered in the fluid phase. The membrane-bound residue almost totally lacked antigens precipitable with antisera to C5, C6, C9, and C5b-9(m).