The membrane attack complex of complement: C5b-8 complex as accelerator of C9 polymerization

Polymerization of C9 occurs spontaneously or can be induced by the tetramolecular complex C5b-8. Spontaneous C9 (0.15 mg/ml) polymerization required more than 3 days at 37 degrees C. In the presence of C5b-8, C9 polymerization was complete within 10 min. The molar C9:C5b-8 ratio determined the extent of tubular poly C9 formation by C5b-8-bearing phospholipid vesicles. When this ratio was 9:1 or 12:1, 72% of complex-bound C9 was present as SDS resistant tubular poly C9 (Mr = 1.1 X 10(6]. At lower C9:C5b-8 ratios, poly C9 was bound primarily in nontubular form. Tubular poly C9, as part of C5b-9, could also be generated on rabbit erythrocytes by using whole human serum as a complement source. At limiting serum concentration (molar C9 to C8 ratio approximately 2), no SDS-resistant tubular poly C9 was detected. At high serum concentration or when using serum that was supplemented with C9, up to 40% of the C9 was SDS-resistant tubular poly C9, and the rest was poly C9, which was incompletely polymerized. It is suggested that the C5b-8 complex acts as an accelerator of C9 polymerization, and that its relative concentration to C9 determines the ultrastructure of the C5b-9 complex.     

The relationship between channel size and the number of C9 molecules in the C5b-9 complex

We have recently shown by dose-response analyses with resealed erythrocyte ghosts that the channel formed by complement is a monomer of C5b-9 of the composition C5b61C71C81C9n, in which n = 1 for channels permitting passage of sucrose (0.9 nm molecular diameter) and n = 2 for channels allowing transit of inulin (3 nm molecular diameter) (1). We have now continued these experiments and expanded them by including ribonuclease A (molecular diameter, 3.8 nm) as a marker to assess whether additional C9 molecules enlarge the functional C5b-9 channel. Our results show that formation of C5b-9 channels displays one-hit characteristics with respect to C5b6 when tested by transmembrane passage of inulin or ribonuclease A. By contrast, analysis of dose-response curves of C9 indicate that n = 2-3 for channels allowing transit of inulin and n = 4 for channels allowing transit of ribonuclease A. We have also performed sieving experiments with ghosts carrying C5b-7 and containing two small markers, inositol and sucrose. Dose-response curves for C8 were performed in the presence of excess C9 to ensure conversion of all C5b-8 to C5b-9 channels. The results indicate that small channels (approximately 0.8 nm effective diameter) are not formed at high C9 multiplicity, thus confirming the results obtained with the larger markers, i.e., increase of C9 input leads to formation of larger channels.     

Molecular composition of the terminal membrane and fluid-phase C5b-9 complexes of rabbit complement. Absence of disulphide-bonded C9 dimers in the membrane complex

The terminal membrane C5b-9(m) and fluid-phase SC5b-9 complexes of rabbit complement were isolated from target sheep erythrocyte membranes and from inulin-activated rabbit serum respectively. In the electron microscope, rabbit C5b-9(m) was observed as a hollow protein cylinder, a structure identical with that of human C5b-9(m). Monodispersed rabbit C5b-9(m) exhibited an apparent sedimentation coefficient of 29 S in deoxycholate-containing sucrose density gradients, corresponding to a composite protein-detergent molecular-weight of approx. 1.4 X 10(6). Protein subunits corresponding to human C5b-C9 were found on sodium dodecyl sulphate/polyacrylamide-gel electrophoresis. By densitometry, there were consistently six molecules of monomeric C9 present for each monomeric C5b-8 complex. Fluid-phase rabbit SC5b-9 was a hydrophilic 23 S ma macromolecule that differed in subunit composition from its membrane counterpart in that it contained S-protein and only two to three molecules of C9 per monomer complex. The data are in accord with the previous report on human C5b-9 that C5b-9(m) contains more C9 molecules than SC5b-9 [Ware & Kolb (1981) Proc. Natl. Acad. Sci. U.S.A. 78, 6426-6430]. They corroborate the previous molecular-weight estimate of approx. 10(6) for C5b-9(m) and thus support the concept that the fully assembled, unit lesion of complement is a C5b-9 monomer [Bhakdi & Tranum-Jensen (1981) Proc. Natl. Acad. Sci. U.S.A. 78, 1818-1822]. They also show that C9 dimer formation is not required for assembly of the rabbit C5b-9(m) protein cylinder, or for expression of its membrane-damaging function.     

The membrane attack complex in complement-mediated glomerular epithelial cell injury: formation and stability of C5b-9 and C5b-7 in rat membranous nephropathy

Using a model of rat membranous nephropathy (MN), we examined the relationship between the development of glomerular epithelial cell injury and the formation and stability of the membrane attack complex (MAC) of complement. Isolated rat kidneys were perfused with buffered bovine albumin (BSA) or various plasmas (complement source). Kidneys containing nephritogenic amounts of complement-fixing sheep antibody to glomerular epithelial antigens (aFx1A) perfused with BSA (n = 5), and normal kidneys perfused with normal human plasma in BSA (50% v/v, n = 6) excreted 0.30 +/- 0.02 mg protein/min/g during 90 min perfusion (control groups). When normal plasma was added to the perfusate of aFx1A kidneys at concentrations of 12.5, 25, and 50% v/v, protein excretion rose in a time- and concentration-dependent manner. Perfusions with 25% plasma resulted in baseline proteinuria from 0 to 20 min that increased to 2.8 +/- 0.9 mg/min/g at 20 to 40 min and 8.6 +/- 2.1 at 40 to 60 min (n = 4, p less than 0.01 vs control groups). Removal of plasma at 20 min did not prevent this rise in protein excretion (3.9 +/- 2.4 and 5.8 +/- 2.6 mg/min/g at 30 to 40 and 55 to 65 min respectively, p less than 0.01, n = 4). Perfusion of aFx1A kidneys with C8-deficient (C8D) human plasma (25% v/v, n = 4) or C6D rabbit serum (25% v/v, n = 2) independently produced low levels of proteinuria comparable with BSA, but in combination, the two reagents restored enhanced protein excretion (n = 2). In aFx1A kidneys containing C5b-7, addition of C8 and C9 (C6D serum) after intervals of 20, 60, or 90 min immediately reconstituted heavy proteinuria. Thus, the magnitude of MAC-induced glomerular epithelial injury in rat MN is related to the complement dose. Altered glomerular permeability is delayed with respect to the onset of complement activation. Once sufficient C5b-9 is formed, proteinuria can develop despite cessation of new MAC assembly, implying that C5b-9 persists after formation. Moreover, the C5b-7 MAC intermediate is not eliminated rapidly in this model.     

Penetration of C8 and C9 in the C5b-9 complex across the erythrocyte membrane into the cytoplasmic space

We have developed a technique in which transglutaminase is used to measure the penetration of terminal complement proteins across the erythrocyte membrane into the cytoplasmic space. Penetration of a given terminal complement protein into the cytoplasmic space was assessed by labeling the protein of interest with radioactive iodine, forming the complement channel using the labeled protein, adding transglutaminase to only one side of the membrane, and allowing the enzyme to cross-link the susceptible proteins on that side of the membrane. Cross-linking was assessed by measuring the increase in molecular weight of the appropriate molecule on sodium dodecyl sulfate gels under reducing conditions. The results of these experiments indicate that C8 and C9 are rapidly cross-linked to high molecular weight from either the interior or the exterior of the membrane. In order to determine whether the cross-linking mediated by enzyme on the interior was occurring from within the ghosts and not via enzyme that had leaked into the extracellular medium, experiments were performed with dimethylcasein in the extracellular medium. In the presence of this protein, cross-linking of C8 and C9 from outside was negligible. Hence, if cross-linking occurs when transglutaminase is trapped inside the ghosts, it cannot be due to leakage of enzyme, but must be attributable to cross-linking from the inside. The results show that C9 definitely penetrated across the membrane into the intracellular space. With respect to C8, statistical evaluation indicates that C8 probably penetrated into the intracellular space.     

Monoclonal antibodies against neoantigens of the terminal C5b-9 complex of human complement

Assembly of the terminal C5b-C9 complement components into the cytolytic C5b-9 complex is accompanied by exposure of characteristic neoantigens on the macromolecule. We report the production and characterization of mouse monoclonal antibodies to C9-dependent neoantigens of human C5b-9. Binding-inhibition assays with EDTA-human plasma and micro-ELISA assays with purified C9 showed that the antibodies did not react with native complement components and thus confirmed the specificity of the antibodies for the neoantigens. The monoclonal antibodies did, however, cross-react with cytolyticaIly inactive, fluid-phase C5b-9 complexes, Thus, expression of the neoantigenic determinants was not dependent on the formation of high molecular weight C9 polymers with the complex, since these are absent in fluid-phase C5b-9. Radioiodinated antibodies could be utilized in immunoradiometric assays for the detection and quantitation of C5b-9 on cell membranes. Cross-reactivities of the antibodies with C9-dependent neoantigens of several other animal species were examined and antibody clones cross-reacting with rabbit (clones 3BI, 3Dg, and 2F3), sheep (clones 3Dg and 2F3) and guinea-pig (clone 3D8) neoantigens were identified . Three of four tested clones (3D8, 2F3, IA12) precipitated C5b-9 complexes in double-diffusion assays, probably due to their interaction with multiple and repeating C9-epitopes on the terminal complexes. The monoclonal antibodies will be of value for definitive identification and quantitation of C5b-9 on cell membranes and in tissues, and for establishing immunoassays for detection and quantitation of terminal fluid-phase C5b-9 complexes in plasma.     

C5b-9 assembly: average binding of one C9 molecule to C5b-8 without poly-C9 formation generates a stable transmembrane pore

Membrane attack by serum complement normally results in the formation of C5b-9 complexes that are heterogeneous with respect to their C9 content. We here report that an apparently homogeneous population of C5b-9 complexes can be generated through treatment of C5b-7-laden sheep erythrocytes with C8 and C9 for 60 min at 0 degree C. Experiments performed by using radioiodinated C8 and C9 components have indicated that binding of C8 to these target cells is essentially temperature independent. In contrast, when a surplus of C9 molecules is offered to C5b-8 cells, an approximately fourfold to 4.5-fold higher number of C9 molecules become cell bound at 37 degrees C as opposed to 0 degree C. C5b-9 complexes isolated from target membranes treated with C9 at 0 degree C contain no polymerized C9 and do not exhibit the ring structure characteristic of the classical complement lesion. Nevertheless, these complexes generate stable transmembrane channels and cause hemolysis at 37 degrees C. The pores have been sized to 1 to 3 nm effective diameter by osmotic protection experiments. SDS-PAGE of the isolated complexes indicates an average stoichiometry of only one molecule C9 bound per C5b-8 complex. The results show that oligomerization of C9 with formation of ring lesions is not a basic requirement for the generation of stable transmembrane complement pores in sheep erythrocytes. They indirectly support the contention that terminal complement components other than C9 contribute to the intramembrane domains of C5b-9 pores.     

The structure of human complement component C7 and the C5b-7 complex

The molecular architecture of human complement component C7 was elucidated at several structural levels. The complete primary structure of C7 was derived from the cDNA sequence of clones isolated from a human liver library. C7 is a mosaic protein that consists of 821 amino acids. The amino-terminal two-thirds of C7 has 23-30% homology with complement components C8 and C9. In addition, the carboxyl-terminal third contains four cysteine-rich segments that have overlapping internal homology. The protein is a single polypeptide chain with 28 disulfide bonds and is glycosylated at two sites. Virtually all the cysteines are found in small units of 35-77 amino acids that exhibit homology with those of various proteins including the low density lipoprotein receptor, epidermal growth factor precursor, thrombospondin, and blood coagulation factors IX and X. The secondary structural analysis, estimated by circular dichroism, suggested a high content of beta-sheet (38%) and beta-turns (24%). The tertiary structure, visualized by transmission electron microscopy, indicated a flexible elongated molecule with dimensions of 151 X 59 X 43 A. The quaternary structure of the C5b-7 complex bound to lipid vesicles was observed to be in the form of monomers or dimers. The monomer C5b-7 consists of a leaflet and a long flexible stalk, and the dimer has two leaflets linked through a supercoiled stalk. Membrane binding is mediated by the stalk part of the complexes. Using a radioiodinated photoreactive cross-linking reagent bound to the polar head group of phosphatidylethanolamine, the stalk part of the C5b-7 complex could be labeled preferentially, and it was found to consist mainly of C6 and C7. Thus, C7 plays a major role in bringing about the hydrophilic-amphiphilic transition during the formation of the membrane attack complex, and it serves as a membrane anchor for the C5b-7 complex.     

Deposition of the terminal C5b-9 complement complex in infarcted areas of human myocardium

Poly- and monoclonal antibodies to neoantigens of the human C5b-9 complement complex, as well as polyclonal antibodies to C5, C8, and C9, were used to detect and identify C5b-9 deposits in human myocardial tissue. Immunocytochemical studies were performed on fresh-frozen autopsy material derived from patients with myocardial infarctions; in addition, in 17 of these patients, paraffin sections of formalin-fixed tissue were investigated. Sixteen autopsies from patients with noncardiac diseases were analyzed as controls. Without exception, C5b-9 positivity was registered selectively and exclusively on and in myocardial cells located within the zones of infarction. The selectivity of staining was confirmed by control reactions for succinic dehydrogenase activity performed in adjacent, respective double-stained sections. Most intensive staining with anti-neoantigen antibodies was observed in the peripheral areas of the infarctions. Weak staining for C3d, rather strong staining for C5 and C9, and intermediate staining with anti-C8 antibodies were observed in the same localizations. Stainings for C4 and IgA were negative, whereas immunocytochemical reactions for IgG and IgM revealed an irregular and very weak staining. Only very weak staining was also observed with a monoclonal antibody to complement S-protein, indicating that the terminal complement components were deposited mainly in the form of membrane-damaging C5b-9 complexes. Immunocytochemical staining for C5b-9 was found to represent a most sensitive tool for detection of ischemic myocardial lesions, permitting easy detection even of single cell necroses. As a working hypothesis, we suggest that initial ischemia may cause loss of the ability of the heart muscle cells to regulate complement turnover at the membrane level. The resulting deposition of C5b-9 on the cell membranes may contribute to functional disturbance and irreversible damage of myocardial cells during the infarction process.     

Assembly of complement components C5b-8 and C5b-9 on lipid bilayer membranes: visualization by freeze-etch electron microscopy

We have visualized by freeze-etch electron microscopy the macromolecular complexes of complement, C5b-8 and C5b-9, respectively, assembled on synthetic phospholipid bilayers. These complexes were formed sequentially by using purified human complement components C5b-6 followed by C7, C8, and C9. Complexes of C5b-8 were observed on the external surface (ES) of vesicles as 12-nm particles that tended to form polydisperse aggregates. The aggregates were sometimes of a regular chainlike structure containing varying numbers of paired subunits. Etching of vesicles containing C5b-9 complexes revealed on the ES large rings of approximately 27-nm outer diameter. One or two knobs usually were attached to the perimeter of the rings. Splitting of the membrane resulted in partitioning of the C5b-9 with the outer leaflet. Thus, round holes of approximately 17-nm diameter were present in the protoplasmic face (PF), and raised circular stumps of a matching size were present on the exoplasmic face (EF) of C5b-9 vesicles. C5b-9 complexes were frequently localized in regions of the lowest lipid order. That is, in micrographs of the EF and ES, single C5b-9 complexes were located where the ripples of the P beta' phase bend or reach a dead end, and linear arrays of C5b-9 complexes outlined disclination-like structures in the lattice; the holes in the PF mirrored this distribution. The membrane immediately surrounding C5b-9 rings was often sunk inwardly over an area much larger than that of the ring itself.(ABSTRACT TRUNCATED AT 250 WORDS)     

The complement SC5b-9 complex mediates cell adhesion through a vitronectin receptor.

Adhesion of cells to the terminal complement complex of C5b through C9 containing the serum S-protein (SC5b-9) was investigated using a microtiter plate attachment assay with L8 myoblast indicator cells. The skeletal muscle-derived L8 myoblasts bound and spread on substratum coated with SC5b-9, and with the vitronectin/S-protein component of SC5b-9. The myoblasts did not adhere to substratum coated with collagen, laminin, or fibronectin. The cell attachment was blocked by antibody to vitronectin/S-protein, whereas antibody to the other components C5, C6, C7, C8, or C9 had minimal effect. The cells were not bound to free vitronectin because attachment activity was removed by adsorption with an anti-C6 antibody column. The L8 cell attachment was dependent on divalent cations, was blocked by synthetic peptides containing the amino acid sequence Arg-Gly-Asp, and was inhibited by antivitronectin receptor antibody. These results indicate that cells adhere to the SC5b-9 complex through interaction of the vitronectin component with an integrin vitronectin receptor. Cell attachment to terminal C complexes could be used for leukocyte adherence and migration during inflammation, and also for attachment of tissue cells during regeneration after disease or traumatic injury.     

Identification of the beta-endorphin-binding subunit of the SC5b-9 complement complex: S protein exhibits specific beta-endorphin-binding sites upon complex formation with complement proteins

Beta-Endorphin has been reported to specifically interact with SC5b-9 complement complexes via non-opioid binding sites. Covalent cross-linking of [125I]beta H-endorphin to SC5b-9 and analysis of the cross-linking products by gel electrophoresis and subsequent autoradiography revealed a single specifically labelled species which was identical with the S protein subunit of the complement complex. In contrast to SC5b-9, no cross-linking of labelled beta-endorphin to subunits of C5b-9(m) could be observed, indicating that beta-endorphin binding to SC5b-9 was mediated exclusively via S protein. Beta-Endorphin binding to SC5b-9 was compared with binding to purified S protein. Whereas beta-endorphin binding to purified S protein was only modest, complex formation of S protein with complement proteins led to a strong increase in beta-endorphin-binding site concentration, compatible with the exposure of primarily cryptic beta-endorphin-binding sites on S protein.     

Bacterial killing and inhibition of inner membrane activity by C5b-9 complexes as a function of the sequential addition of C9 to C5b-8 sites

The assembly of the C5b-9 complex on the outer membrane of C-sensitive cells of Escherichia coli results in a rapid inhibition of inner membrane function and ultimately a loss of cell viability. Cells bearing C5b-8 sites suffer no deleterious effects; however, the addition of C9 results in a rapid inhibition of inner membrane function and cell death. An attempt was made to examine the relationship between the toxic effects of the C5b-9 complex and the number of C9 molecules per C5b-8 site. Cells bearing C5b-8 sites were exposed to excess C9 at 0 degrees C and washed three times at 4 degrees C. The number of C9 molecules bound to each cell was equivalent to the number of C5b-8 sites present on each cell, and no additional C9 molecules could be bound when the cells were maintained at 4 degrees C. These cells were then incubated at 37 degrees C for 3 min and returned to 0 degrees C, a technique which exposed additional C9-binding sites equivalent to the number of C9 molecules previously bound to the cells. This technique was repeated and demonstrated that the sequential build-up of a C5b-9 site with two C9 molecules per C5b-8 site was capable of inhibiting both inner membrane function (respiration and amino acid transport) and cell viability. Three C9 molecules per complex had effects that approached the inhibitory effects of complexes formed in the presence of excess C9.     

Complement lysis: evidence for an amphiphilic nature of the terminal membrane C5b-9 complex of human complement

The terminal, membrane-derived C5b-9 complex of human complement (C) is an apparently hollow, cylindrical macromolecule vertically oriented on the target membrane. In the present study, an antiserum to the complex has been used to probe its immunobiochemical properties. "Neoantigenic" determinants characteristic of the complex have been detected, which are absent on native C5-C9 molecules. Evidence that the C5b-9 complex is an amphiphilic molecule that possesses apolar, detergent-binding surfaces has been obtained by using charge-shift crossed immunoelectrophoresis, and by direct demonstration of Triton X-100 binding to the complex in quantitative immunoelectrophoresis. By the same criteria, serum C5, C6, and C9 are hydrophilic molecules. The results indicate that assembly of C5-C9 into the terminal membrane C5b-9 complex is accompanied by conformational changes in the individual C components that lead to the exposure of apolar molecular regions in the complex. It is proposed that this constitutes the basis for the lipid-binding properties of the macromolecule, which enable it to become inserted into biologic and artificial lipid membranes with apparent generation of a transmembrane pore.     

Elimination of terminal complement intermediates from the plasma membrane of nucleated cells: the rate of disappearance differs for cells carrying C5b-7 or C5b-8 or a mixture of C5b-8 with a limited number of C5b-9

We have previously shown that multiple complement (C) channels are required for lysis of a nucleated cell in contrast to the single channel requirement for erythrocytes. To further investigate this multichannel requirement for nucleated cells, we examined the stability of terminal C complexes in the plasma membrane of Ehrlich ascites tumor cells. Ehrlich cells bearing C5b-7 or C5b-8 with or without C9 were incubated at 37 degrees C or 0 degree C for various time intervals before converting the remaining complexes to lytic C5b-9 channels. C5b-7, C5b-8, and C5b-8 in the presence of a limited number of C5b-9 complexes disappeared functionally from the plasma membrane at 37 degrees C, with initial half-lives of 31, 20, and 10 min, respectively. Disappearance of these complexes did not occur at 0 degree C, nor did disappearance occur at 37 degrees C when formed on sheep erythrocytes. The fate of C5b-8 complexes on the surface of Ehrlich cells was traced with colloidal gold particles bound to C5 determinants on C5b-8 with the use of immunoelectron microscopy. Colloidal gold could be seen on the cell surface after specific binding to cells carrying C5b-8 sites at 0 degree C. After incubating these cells at 37 degrees C, gold particles were internalized into the cell continuously via endocytic vesicles. It is postulated that terminal C complexes may stimulate or accelerate the removal of these complexes from the cell surface.     

Caveolin-1 and dynamin-2 are essential for removal of the complement C5b-9 complex via endocytosis

The complement system, an important element of both innate and adaptive immunity, is executing complement-dependent cytotoxicity (CDC) with its C5b-9 protein complex that is assembled on cell surfaces and transmits to the cell death signals. In turn, cells, and in particular cancer cells, protect themselves from CDC in various ways. Thus, cells actively remove the C5b-9 complexes from their plasma membrane by endocytosis. Inhibition of clathrin by transfection with shRNA or of EPS-15 with a dominant negative plasmid had no effect on C5b-9 endocytosis and on cell death. In contrast, inhibition of caveolin-1 (Cav-1) by transfection with an shRNA or a dominant negative plasmid sensitized cells to CDC and inhibited C5b-9 endocytosis. Similarly, both inhibition of dynamin-2 by transfection with a dominant negative plasmid or by treatment with Dynasore reduced C5b-9 endocytosis and enhanced CDC. C5b-9 endocytosis was also disrupted by pretreatment of the cells with methyl-β-cyclodextrin or Filipin III, hence implicating membrane cholesterol in the process. Analyses by confocal microscopy demonstrated co-localization of Cav-1-EGFP with C5b-9 at the plasma membrane, in early endosomes, at the endocytic recycling compartment and in secreted vesicles. Further investigation of the process of C5b-9 removal by exo-vesiculation demonstrated that inhibition of Cav-1 and cholesterol depletion abrogated C5b-9 exo-vesiculation, whereas, over-expression of Cav-1 increased C5b-9 exo-vesiculation. Our results show that Cav-1 and dynamin-2 (but not clathrin) support cell resistance to CDC, probably by facilitating purging of the C5b-9 complexes by endocytosis and exo-vesiculation.     

Inhibition of homologous complement by CD59 is mediated by a species-selective recognition conferred through binding to C8 within C5b-8 or C9 within C5b-9.

The capacity of the human complement regulatory protein CD59 to interact with terminal complement proteins in a species-selective manner was examined. When incorporated into chicken E, CD59 (purified from human E membranes) inhibited the cytolytic activity of the C5b-9 complex in a manner dependent on the species of origin of C8 and C9. Inhibition of C5b-9-mediated hemolysis was maximal when C8 and C9 were derived from human (hu) or baboon serum. By contrast, CD59 showed reduced activity when C8 and C9 were derived from dog or sheep serum, and no activity when C8 and C9 were derived from either rabbit or guinea pig (gp) serum. Similar specificity on the basis of the species of origin of C8 and C9 was also observed for CD59 endogenous to the human E membrane, using functionally blocking antibody against this cell surface protein to selectively abrogate its C5b-9-inhibitory activity. When E bearing human CD59 were exposed to C5b-8hu, CD59 was found to inhibit C5b-9-mediated lysis, regardless of the species of origin of C9, suggesting that the inhibitory function of CD59 can be mediated through recognition of species-specific domains expressed by human C8. Consistent with this interpretation, CD59 was found to bind to C5b-8hu but not to C5b67hu or C5b67huC8gp. Although CD59 failed to inhibit hemolysis mediated by C5b67huC8gpC9gp, its inhibitory function was observed for C5b67huC8gpC9hu, suggesting that, in addition to its interaction with C5b-8hu, CD59 also interacts in a species-selective manner with C9hu incorporated into C5b-9. Consistent with this interpretation, CD59 was found to bind both C5b67huC8gpC9hu and C5b-8huC9gp, but not C5b67huC8gpC9gp. Taken together, these data suggest that the capacity of CD59 to restrict the hemolytic activity of human serum complement involves a species-selective interaction of CD59, which involves binding to both the C8 and C9 components of the membrane attack complex. Although CD59 expresses selectivity for C8 and C9 of human origin, this "homologous restriction" is not absolute, and this human complement regulatory protein retains functional activity toward C8 and C9 of some nonprimate species.