Membrane attack complex of complement: Generation of highaffinity phospholipid binding sites by fusion

Transcription

1 Proc. Natl. Acad. Sci. USA Vol. 76, No. 2, pp , February 1979 Immunology Membrane attack complex of complement: Generation of highaffinity phospholipid binding sites by fusion of five hydrophilic plasma proteins (transmembrane channel/protein-phospholipid micelles/mechanism of complement-dependent cytolysis) ECKHARD R. PODACK, GREGORY BIESECKER, AND HANS J. MULLER-EBERHARD Department of Molecular Immunology, Research Institute of Scripps Clinic, La Jolla, California Contributed by Hans J. Muller-Eberhard, November 13, 1978 ABSTRACT The molecular basis of the membranolytic activity of the membrane attack complex (MAC) of complement was investigated. By using density gradient equilibrium ultracentrifugation, the binding of egg yolk lecithin to the isolated MAC and to its intermediate complexes and precursor proteins was measured. No stable phospholipid-protein complexes were formed with the MAC precursor components C5b-6, C7, C8, and C9. Stable complexes of phospholipid and protein were formed by CMb-7, C5b-8, CAb-9, and the MAC (C5b-9 dimer) and they exhibited densities of , 1.184, , and g/ml, respectively. The molar phospholipid/protein ratios for the four complexes were determined to be: CMb-7, 399:1; C5b-8, 841:1; CMb-9, 9181; and C5b-9 dimer, 1460:1. Electron microscopy of the isolated phospholipid-protein complexes revealed no lipid bilayer structures. The magnitude of the phospholipid binding capacity of the MAC is consistent with the interpretation that the MAC forms phospholipid-protein mixed in micelles in lipid bilayers and biological membranes and thus causes formation of hydrophilic lipid channels. Complement-dependent cytolysis is caused by the membrane attack complex (MAC) which is a fusion product of five plasma proteins: C5b, C6, C7, C8, and C9 (1-3). The MAC isolated from the membranes of complement-lysed cells has a sedimentation coefficient of 33.5 S and a molecular weight of 1,700,000, and constitutes the dimeric form of C5b-9 with the probable structural formula (C5b, C6, C7, C8, C93)2 (4). It is this dimer that, upon electron microscopy, creates the image of the characteristic complement-dependent membrane lesions (4-7). The mode of action of the MAC on biological membranes and lipid bilayers is unknown, except that it causes formation of functional transmembrane channels for small ions and water (8). To explain the formation of these channels, insertion of the forming MAC into the target membrane has been postulated (9, 10). Because the five precursors of the MAC are hydrophilic proteins, their insertion into the hydrophobic interior of membranes has been difficult to envisage. This communication reports that high-affinity phospholipid binding sites appear de novo and increase in number as the MAC assembles from its precursors to the dimeric C5b-9 complex. The newly expressed phospholipid binding capacity may constitute the physical basis of MAC membranolytic activity. MATERIALS AND METHODS Complement Proteins. C5b-6 (11), C7 (12), C8 (13), C9 (14), and (C5b-9)2 (4) were purified as described. The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C solely to indicate this fact. 897 Radiolabeling. Proteins were radiolabeled with 125I by the method of McConahey and Dixon (15). Free iodine was removed by gel filtration of the labeled mixture on 10-ml columns of Sephadex G-25 equilibrated with barbital-buffered (ph 7.4) saline. The radiolabeled proteins were mixed with the respective unlabeled proteins and the specific radioactivity was determined by measuring the protein concentration (16) and assaying an aliquot for 125I radioactivity. The functional activity of 125I-labeled C5b-6 (1251-C5b-6) was assessed by adding 125I-C5b-6 to normal human serum and determination of the conversion of 1251-C5b-6 to the 23S SC5b-9 by sucrose density gradient ultracentrifugation (17). Functional activity of 125I-C9 was also determined by its conversion to a 23S form after zymosan activation of normal serum (18). Both radiolabeled proteins were found to be fully active (>85%). Lipid Binding Studies. Egg lecithin (Avanti Biochemical Inc., Birmingham, AL) was mixed with phosphatidyl[14c]- choline (New England Nuclear, Boston, MA) to give a specific radioactivity of 1.7 MiCi/pidtol. The specific radioactivity was ascertained by phosphorus determination (19) and measurement of the radioactivity in a scintillation counter. The organic solvent was removed in vacuo, and the dry egg lecithin was dissolved to give a 1.3 mm concentration in 0.09 M NaCl/0.02 M Tris acetate/2 mm EDTA/0.02% sodium azide/2.5 mm sodium deoxycholate, ph 8.1. Samples (0.1 ml) of this mixture containing phospholipid-deoxycholate mixed micelles were incubated with the following complement proteins: (i) 35,ug of 1251-C5b-6; (ii) 35,ug of 1251-C5b-6 and 22,ug of C7; (lii) 35,ug of 1251 C5b-6, 22,Ag of C7, and 20,ug of C8; (iv) 35,ug of C5b-6, 22 jig of C7, 20 tig of C8, and 22,Ag of 125I-C9; (v) 120 jig 125I-(C5b-9)2 extracted from membranes; (vi) 20,ug of C7; (vii) 20,ug of C8; (viii) 30,g of C9; (ix) 35 gg of 1251-C5b-6. All samples were incubated for 10 min at 300C and then dialyzed at 4 C in Spectrapor membranes (molecular weight cut-off, 50,000; Spectrum Medical Industries Inc., Los Angeles, CA) for 16 hr against three changes of 500 ml of 0.02 M Tris acetate, ph 8.1/0.09 M NaCI/2 mm EDTA/0.02% NaN3. In some experiments the detergent was separated by gel filtration on Sephadex G-50 (20). After dialysis the samples were applied to 5-ml CsCI density gradients gradients ranging from 8 to 40% (wt/wt), the CsCI being dissolved in the dialysis buffer. The samples were centrifuged to equilibrium for 20 hr at 45,000 rpm at 40C in an SW 50.1 rotor (Beckman). Fractions (0.2 ml) were collected from the top of the centrifuge tube by means of a Densi Flow 11c Collector (Searle, Buchler, Fort Lee, NJ). The fractions were analyzed for 125I and 14C radioactivity. Care was taken that 14C radioactivity exceeded 125I radioactivity about 10-fold in order Abbreviations: PL, phospholipids; DOC, deoxycholate; MAC, membrane attack complex, C5b-9 dimer.

2 898 Immunology: Podack et al. to facilitate correction of 14C radioactivity for 125I radioactivity which is also detected in the scintillation counter. Further analysis included measurement of the refractive index with a Zeiss (West Germany) refractometer. The refractive index was converted to densities by using the data from International Critical Tables, as provided by E. Merck, Darmstadt, Germany. Molar ratios of lipid to protein were obtained directly from the ratio of phospholipid radioactivity of protein radioactivity in the peak fractions after CsCl density gradient ultracentrifugation. Lipid/protein ratios were also determined from the density of the lipid-protein complex according to the following formula: x = (ippl - i5p)/(i5 -VPPL) in which x is the amount of phospholipid (in g) bound per 1 g of protein. i5ppl, up, and VPL are the partial specific volumes of the protein-phospholipid complex, of free protein (iv = ml/g), and of free phospholipids (iv = ml/g), respectively. These calculations assume that i-s of free and complexed lipid and of free and complexed protein are identical. Electron Microscopy. Samples (0.1 mg/ml) were placed on 400-mesh copper grids coated with Parlodion and carbon and made hydrophilic with bacitracin (1 mg/ml). After excess sample was removed with a filter paper, the samples were negatively stained with 1% Na phosphotungstate (ph 7.2), 12 mm uranyl oxalate (ph 6.8), or 0.5% uranyl formate (ph 4.5) (21). Grids were examined in a Hitachi 12A electron microscope operated at 75 kv and a direct magnification of X100,000 with a 500-Aim second aperture and a 20-Am objective aperture. The bright-field image was recorded on Kodak 4463 electron image film. Calculation of the Surface Area of Protein and Phospholipid. The surface area of a spherical protein with a molecular weight of 1.7 X 106 is 8.04 X 104 A2, assuming a volume of 1.25 A3 per molecular weight unit for a protein of partial specific volume = 0.73 ml/g. From the fact that the 1.7 X 106 dalton C5b-9 dimer is not a spherical complex (4, 7) and the surface roughness index (22), a total accessible surface area for (C5b-9)2 of approximately 1.6 X 105A2 is derived. Above the transition temperature, the area per phospholipid molecule in a lipid bilayer is A2 (23). Assuming the formation of protein-lipid micelle, approximately 60-70% of the total surface area of the C5b-9 dimer is covered with lipid. A B Proc. Natl. Acad. Sci. USA 76 (1979) RESULTS The isolated MAC and its various intermediate complexes were tested for the ability to bind phospholipid by using the experimental approach outlined in Fig. 1. C5b-7, C5b-8, and C5b-9 were formed in the presence of deoxychalate (DOC)-phospholipid mixed micelles by using isolated C5b-6, C7, C8, and C9. C5b-9 was also obtained by extraction from complement-lysed cells; after purification by gel filtration, it was added to DOC-phospholipid mixed micelles. All preparations were dialyzed to remove DOC. Fig. 2 shows that both C5b-9 assemblages result in the formation of typical ultrastructural complement lesions in liposomes: the upper panel depicts the complex assembled from isolated precursors; the lower panel shows the complex extracted from complement-treated cell membranes after reassociation with phospholipid vesicles. For quantitative lipid binding studies the protein/phospholipid mixtures were subjected to isopycnic CsCl density gradient ultracentrifugation. Fig. 3 shows the results of one experiment which indicate that a portion of the phospholipid sedimented in firm association with the four complexes examined, whereas unbound phospholipid remained at the top of the CsCl gradient. Controls included C5b-6, C7, C8, and C9 individually treated with DOC-phospholipid mixed micelles, dialyzed, and analyzed by CsCl density gradient ultracentrifugation. DOC-PL + C5b-6 +C7 DOC-PL-C5b-7 +C8 +C9 DOC-PL-C5b-9 EC5b-9 I DOC-C5b-9 Extraction with 10% DOC Gel Filtration in 2% DOC +PL DOC-PL-C5b-9 DOC Di( Glysis PL-C5b-9 DOC Isopyknic CsCI Density Gradient Ultracentrifugation FIG. 1. Schematic representation of the experimental design to obtain phospholipid-protein complexes of the MAC and its intermediate complexes. DOC, deoxycholate; PL, phospholipid. FIG. 2. Production of typical complement lesions on phospholipid vesicles by the forming MAC and the previously formed, isolated MAC. (X500,000.) (Upper) C5b-6, C7, C8, and C9 were mixed with DOC-phospholipid micelles and the detergent was removed by dialysis. (Na phosphotungstate stain.) (Lower) MAC isolated from complement-treated cell membranes was incubated with DOCphospholipid micelles and DOC was removed by gel filtration, (Uranyl formate stain.) The size of the complement lesion in both electron micrographs is identical. In Upper, some MACs are seen in profile and some are detached from the vesicle membrane.

3 Immunology: Podack et al. Fraction FIG. 3. Demonstration of the stable phospholipid-protein complexes formed by the MAC and its intermediate complexes. C5b-7, C5b-8, and C5b-9 were built up from C5b-6, C7, C8, and C9 in the presence of DOC-phospholipid micelles. Membrane C5b-9 is the C5b-9 dimer extracted from complement-treated cells and purified (4). It was also mixed with DOC-phospholipid micelles. All samples were dialysed to remove DOC and then applied to CsCl density gradients. Table 1 presents the quantitative binding data (means of three separate determinations). The lipid content of the proteins and their complexes was assessed by density determination and by measurement of '4C-labeled phospholipid bound to 1251_ labeled protein. No phospholipid binding was observed for C7, C8, or C9, and questionable or minimal binding was detectable in the case of C5b-6. The C5b-7 complex resulting from the addition of C7 to C5b-6 acquired a phospholipid binding capacity of 400 molecules per 450,000 daltons and a partial specific volume of The C5b-7 complex formed in the absence of phospholipid had a partial specific volume of 0.763, Proc. Natl. Acad. Sci. USA 76 (1979) 899 corresponding to that of a lipid-free protein (not listed in Table 1). Binding of CS to C5b-7 approximately doubled phospholipid binding, and formation of C5b-9 raised lipid binding to 920 molecules per 800,000 daltons which is the calculated molecular weight of the C5b-9 monomer. The C5b-9 dimer that was extracted from the membranes of complement-lysed cells bound 1450 phospholipid molecules per 1,700,000 daltons. In view of the large number of phospholipid molecules bound per C5b-9 complex, the question arose as to whether the lipid was arranged around the protein in the form of a bilayer. Such an arrangement could not be detected. In Fig. 4 are two electron micrographs that show the ultrastructural difference between protein-free lipid obtained from the top of a CsCl density gradient and protein-lipid complexes that had sedimented during the same ultracentrifugation experiment to the position of their density. Although bilayer structure can be detected in the protein-free lipid particles (Fig. 4 upper), it cannot be detected in the protein-lipid complexes (Fig. 4 lower). DISCUSSION Interaction of the MAC of human complement with phospholipid bilayers has been shown to cause conductance changes (24, 25), release of trapped markers from liposomes (26, 27), and appearance of the typical complement-dependent ultrastructural lesions (4-7). It was shown in this laboratory that the C5b-9 complex and its intermediate complexes are capable of binding ionic detergents and phospholipids (17). We have now been able to quantitate the phospholipid binding capacity of these complexes, and the data at hand argue strongly in favor of one of the various possible models of membrane damage by complement: the model of proteinphospholipid micelles. Our experimental approach is based on the phenomenon that the MAC inserted into lipid bilayers of liposomes tends to detach itself from the liposomes (28). This tendency, which is shared by the intermediate complexes of the MAC, is greatly enhanced under the conditions of CsCl density gradient ultracentrifugation. In the experiments reported above, the phospholipidprotein complexes sedimented to their respective density within the density gradient, whereas the protein-free phospholipid particles accumulated at the top of the gradient. Upon electron microscopic analysis, the phospholipid-protein complexes contained no lipid bilayer structures, indicating that most of Table 1. Phospholipid binding to terminal complement components and their complexes* Phospholipid bound, Molecular Refractivet Density, Partial specific mol/mol protein Protein weight index g/ml volume Calculated from UI Measured C5b-6 328, <30 <30 C7 121, C8 153, C9 74, C5b-7 450, C5b-8 600, C5b , (C5b-9)2$ 1,700, * Data are shown as mean of three separate experiments. t Refractive index of CsCl gradient at the protein peak. Calculated according to x = (UppL - Up)/I(pL -VppL) in which x is the amount of phospholipid (in g) bound per 1 g of protein. VppL, Up, and UpL are partial specific volumes of the phospholipid-protein complex, of protein, and of phospholipids, respectively. Up = 0.763,!PL = , and UPPL was obtained from the CsCl density gradient. Values measured directly from the ratio of 1251-labeled protein and 14C-labeled phospholipid and the known specific radioactivities. II C5b-9 built up from purified C5b-6, C7, C8, and C9, assuming three C9 molecules per complex. C5b-9 dimer extracted from membranes with DOC.

4 900 Immunology: Podaick et A Proc. Natl. Acad. Sci. USA 76 (1979) the phospholipid molecules were bound directly to complement protein through hydrophobic interaction. Our quantitations showed that the phospholipid binding capacity increased markedly with assembly of the MAC from the precursor stage to the dimeric C5b-9 complex. The precursor proteins exhibited no detectable phospholipid binding capacity, whereas C5b-7 bound 400 molecules per 450,000 daltons, C5b-8 bound 800 molecules per 600,000 daltons, and (C5b-9)2 bound 1400 molecules per 1,700,000 daltons. Whether these complexes bind similar amounts of membrane lipids when inserted into the lipid compartment of biological membranes is not known. The correlation between the extent of lipid binding by these protein complexes and their ability to cause increased ion permeability of membranes is good. C5b-7 causes no leakiness of cells and no impairment of synthetic lipid bilayers unless they consist entirely of highly unsaturated phospholipids (29). C5b-8 causes slow lysis and C5b-9 causes rapid lysis of cells. The enhanced lysis of cells due to binding of C9 may be due to the fact that the C5b-9 dimer exerts a greater effect on the lipids at a localized membrane site than does any of its preceding intermediate complexes. Considering the known dimensions of phospholipid molecules (23) and the ultrastructural measurements of the C5b-9 dimer (4), it appears that 1460 phospholipid molecules may cover more than 50% of the surface of the dimer. The same number of phospholipid molecules within a bilayer occupies a circular area of 120 A radius. This value is similar to the radius of the complement lesions. It appears therefore that the phosa C b*9 IN + b {0C5b-9 C K+ Kx+ c C~~b-9 FIG. 4. Lack of lipid bilayer structures in the phospholipidprotein complexes formed by the MAC and its intermediate complexes. Samples were obtained by isopycnic CsCl density gradient centrifugation of the MAC-vesicle recombinants shown in Fig. 2. (Upper) Bilayer lipid free of protein recovered from the top of the CsCl density gradient. (Lower) Ultrastructure of MAC-phospholipid complexes after CsCl density gradient centrifugation; no bilayer lipid can be detected. The diameter of the MAC-phospholipid complexes varies from 250 to 400 A. Due to the bound phospholipid, no details in the subunit structure are visible. (Uranyl oxalate stain; X250,000.) FIG. 5. Schematic representation of three models of membrane damage by the MAC. (A) Protein channel model; (b) lipid channel model; and (c) mixed micelle model. pholipid molecules displaced by insertion of the C5b-9 dimer become instrumental in partially covering the inserted complex. This process may also involve some lateral expansion of the lipid bilayer. The resulting reorientation of the phospholipid molecules may eventuate in decreased stability of the lipid bilayer and formation of hydrophilic channels. Fig. 5 illustrates three models of transmembrane channel formation by complement. Models a and b propose that the MAC traverses the entire membrane, thereby forming a hydrophilic protein (a) or lipid (b) channel. In addition to lipid channels, model b allows a protein channel (not indicated in the drawing). Both models take into consideration that the interaction of the MAC with the lipid bilayer may be both hydrophobic and hydrophilic. Model c proposes that the cytolytic activity of the MAC is due to its ability to bind phospholipids and to form mixed protein-phospholipid micelles and thus hydrophilic lipid channels in its immediate vicinity. It is emphasized that model c proposes a reorientation of boundary lipid, which may be envisioned as depicted schematically in the figure or in a different manner provided it results in a perturbation of the normal bilayer structure of the membrane. Model c is consistent with the body of available data: (i) the initial complement produced channel is small (25,30), (ii) the cytoplasmic leaflet of membranes attacked by the MAC exhibits no morphological alterations (31), and (iii) the MAC (28) and lipid (32) tend to detach from liposomes (33-35) and bacterial cell walls (36), probably in the form of protein-phospholipid complexes. The authors thank Manuel Puentes and Kerry L. Wadey Pangburn for excellent technical assistance. This is Publication No from the Research Institute of Scripps Clinic. This investigation was supported by U.S. Public Health Service Grants Al and HL G.B. is supported by U.S. Public Health Service Training Grant HL E.R.P. is an Established Investigator of the American Heart Association. H.J.M-E. is the Cecil H. and Ida M. Green Investigator in Medical Research, Research Institute of Scripps Clinic.

5 Immunology: Podack et al. 1. Kolb, W. P. & Muller-Eberhard, H. J. (1975) J. Exp. Med. 141, Podack, E. R., Kolb, W. P. & Muller-Eberhard, H. J. (1976) J. Immunol. 116, Bhakdi, S., Ey, P. & Bhakdi-Lehnen, B. (1976) Biochim. Biophys. Acta 419, Biesecker, G., Podack, E. R., Halverson, C. & Muller-Eberhard, H. J. (1978) J. Exp. Med. 149, in press. 5. Borsos, T., Dourmashkin, R. R. & Humphrey, J. H. (1964) Nature (London) 202, Humphrey, J. H. & Dourmashkin, R. R. (1969) Adv. Immunol. 11, Tranum-Jensen, J., Bhakdi, S., Bhakdi-Lehnen, B., Bjerrum, 0. J. & Speth, V. (1978) Scand. J. Immunol. 7, Green, M., Barrow, P. & Goldberg, B. (1959) J. Exp. Med. 110, Muller-Eberhard, H. J. (1975) Annu. Rev. of Biochem. 44, Mayer, M. M. (1972) Proc. Natl. Acad. Sci. USA 69, Podack, E. R., Kolb, W. P. & Muller-Eberhard, H. J. (1978) J. Immunol. 120, Podack, E. R., Kolb, W. P. & Muller-Eberhard, H. J. (1976) J. Immunol. 116, Kolb, W. P. & Muller-Eberhard, H. J. (1976) J. Exp. Med. 143, Hadding, U. & Muller-Eberhard, H. J. (1969) Immunology 16, McConahey, P. J. & Dixon, F. J. (1966) Int. Arch. Allergy Appl. Immunol. 29, Bradford, M. M. (1976) Anal. Biochem. 72, Podack, E. R. & Muller-Eberhard, H. J. (1978) J. Immunol. 121, Kolb, W. P. & Muller-Eberhard, H. J. (1973) J. Exp. Med. 138, Bartlett, G. R. (1958) J. Biol. Chem. 234, Brunner, J., Skrabal, P. & Hauser, H. (1976) Biochim. Biophys. Acta 455, 322. Proc. Nati. Acad. Sc. USA 76 (1979) Haschemeyer, R. H. & Myers, R. J. (1972) in Principles and Techniques of Electron Microscopy: Biological Applications, ed. Hayat, M. A. (Van Nostrand-Reinhold, New York), Vol. 2, pp Richards, F. M. (1977) Annu. Rev. Biophys. Bioeng. 6, Engelman, D. M. (1971) J. Mol. Biol. 58, Wobschall, D. & McKeon, C. (1975) Biochim. Biophys. Acta 413, Michaels, D. W., Abramowitz, A. S., Hammer, C. M. & Mayer, M. M. (1976) Proc. Natl. Acad. Sci. USA 73, Haxby, J. A., Kinsky, C. B. & Kinsky, S. C. (1968) Proc. Natl. Acad. Sci. USA 61, Lachmann, P. J., Munn, E. A. & Weismann, G. (1970) Immunology 19, Podack, E. R., Halverson, C., Esser, A. F., Kolb, W. P. & Muller-Eberhard, H. J. (1978) J. Immunol. 120, Esser, A. F., Podack, E. R., Biesecker, G. & Muller-Eberhard, H. J. (1978) in Sixth International Biophysics Congress (Kyoto, Japan), abstr. VIII-16-G, p Sims, P. J. & Lauf, P. K. (1978) Proc. Natl. Acad. Sci. USA 75, Iles, G. H., Seeman, P., Naylor, D. & Cinader, B. (1973) J. Cell Biol. 56, Inoue, K. (1972) in Research in Immunochemistry and Immunobiology, ed. Kwapinski, J. B. G. (University Park, Baltimore, MD), Vol. 1, pp Shin, M. L., Paznekas, W. A. & Mayer, M. M. (1978) J. Immunol. 120, Kinoshita, T., Inoue, K., Okada, M. & Akiyama, Y. (1977) J. Immunol. 119, Shin, M. L., Paznekas, W. A., Abramowitz, A. S. & Mayer, M. M. (1977) J. Immunol. 119, Inoue, K., Kinoshita, T., Okada, M. & Akiyama, Y. (1977) J. Immunol. 119,65-72.