Molecular aspects of complement-mediated bacterial killing. Periplasmic conversion of C9 from a protoxin to a toxin.

As part of the membrane attack complex complement protein C9 is responsible for direct killing of bacteria. Here we show that in the periplasmic space of an Escherichia coli cell C9 is converted from a protoxin to a toxin by periplasmic conditions missing in spheroplasts. This conversion is independent of the pathway by which C9 enters the periplasm. Both, C9 shocked into the periplasm and plasmid-expressed C9 targeted to the periplasm via a signal sequence are toxic. Toxicity requires disulfide-linked C9 because export into the periplasm of cells defective in disulfide bond synthesis (dsbA and dsbB mutants) is not toxic unless N-acetylcysteine is added externally to promote cystines. A N-terminal fragment, C9[1-144], is not toxic nor is cytoplasmically expressed C9, even in trxB mutants that are able to form disulfide bonds in the cytoplasm. Importantly, expression of full-length C9 in complement-resistant cells has no effect on their viability. Expression and translocation into the periplasm may provide a novel model to identify molecular mechanisms of other bactericidal disulfide-linked proteins and to investigate the nature of bacterial complement resistance.

Association of terminal complement proteins in solution and modulation by suramin.

The association of terminal complement proteins was investigated by analytical ultracentrifugation and multi-angle laser light scattering. Native C8 and C9 formed a heterodimer in solution of physiological ionic strength with a free-energy change DeltaG degrees of -8.3 kcal/mol and a dissociation constant Kd of 0.6 microM (at 20 degrees C) that was ionic strength- and temperature-dependent. A van't Hoff plot of the change in Kd was linear between 10 and 37 degrees C and yielded values of DeltaH degrees = -12.9 kcal/mol and DeltaS degrees = -15.9 cal mol-1 deg-1, suggesting that electrostatic forces play a prominent role in the interaction of C8 with C9. Native C8 also formed a heterodimer with C5, and low concentrations of polyionic ligands such as protamine and suramin inhibited the interaction. Suramin induced high-affinity trimerization of C8 (Kd = 0.10 microM at 20 degrees C) and dimerization of C9 (Kd = 0.86 microM at 20 degrees C). Suramin-induced C8 oligomerization may be the primary reason for the drug's ability to prevent complement-mediated hemolysis. Analysis of sedimentation equilibria and also of the fluorescence enhancement of suramin when bound to protein provided evidence for two suramin-binding sites on each C9 and three on each C8 in the oligomers. Oligomerization could be reversed by high suramin concentrations, but 8-aminonaphthalene-1,3,6- trisulfonate (ANTS2- ), which mimics half a suramin molecule, could not compete with suramin binding and oligomerization suggesting that the drug also binds nonionically to the proteins.

The N-terminal CUB-epidermal growth factor module pair of human complement protease C1r binds Ca2+ with high affinity and mediates Ca2+-dependent interaction with C1s.

The Ca2+-dependent interaction between complement serine proteases C1r and C1s is mediated by their alpha regions, encompassing the major part of their N-terminal CUB-EGF-CUB (where EGF is epidermal growth factor) module array. In order to define the boundaries of the C1r domain(s) responsible for Ca2+ binding and Ca2+-dependent interaction with C1s and to assess the contribution of individual modules to these functions, the CUB, EGF, and CUB-EGF fragments were expressed in eucaryotic systems or synthesized chemically. Gel filtration studies, as well as measurements of intrinsic Tyr fluorescence, provided evidence that the CUB-EGF pair adopts a more compact conformation in the presence of Ca2+. Ca2+-dependent interaction of intact C1r with C1s was studied using surface plasmon resonance spectroscopy, yielding KD values of 10.9-29.7 nM. The C1r CUB-EGF pair bound immobilized C1s with a higher KD (1.5-1.8 microM), which decreased to 31.4 nM when CUB-EGF was used as the immobilized ligand and C1s was free. Half-maximal binding was obtained at comparable Ca2+ concentrations ranging from 5 microM with intact C1r to 10-16 microM for C1ralpha and CUB-EGF. The isolated CUB and EGF fragments or a CUB + EGF mixture did not bind C1s. These data demonstrate that the C1r CUB-EGF module pair (residues 1-175) is the minimal segment required for high affinity Ca2+ binding and Ca2+-dependent interaction with C1s and indicate that Ca2+ binding induces a more compact folding of the CUB-EGF pair.

Baculovirus-mediated expression of truncated modular fragments from the catalytic region of human complement serine protease C1s. Evidence for the involvement of both complement control protein modules in the recognition of the C4 protein substrate.

C1s is the modular serine protease responsible for cleavage of C4 and C2, the protein substrates of the first component of complement. Its catalytic region (gamma-B) comprises two complement control protein (CCP) modules, a short activation peptide (ap), and a serine protease domain (SP). A baculovirus-mediated expression system was used to produce recombinant truncated fragments from this region, deleted either from the first CCP module (CCP2-ap-SP) or from both CCP modules (ap-SP). The aglycosylated fragment CCP2-ap-SPag was also expressed by using tunicamycin. The fragments were produced at yields of 0.6-3 mg/liter of culture, isolated, and characterized chemically and then tested functionally by comparison with intact C1s and its proteolytic gamma-B fragment. All recombinant fragments were expressed in a proenzyme form and cleaved by C1r to generate active enzymes expressing esterolytic activity and reactivity toward C1 inhibitor comparable to those of intact C1s. Likewise, the activated fragments gamma-B, CCP2-ap-SP, and ap-SP retained C1s ability to cleave C2 in the fluid phase. In contrast, whereas fragment gamma-B cleaved C4 as efficiently as C1s, the C4-cleaving activity of CCP2-ap-SP was greatly reduced (about 70-fold) and that of ap-SP was abolished. It is concluded that C4 cleavage involves substrate recognition sites located in both CCP modules of C1s, whereas C2 cleavage is affected mainly by the serine protease domain. Evidence is also provided that the carbohydrate moiety linked to the second CCP module of C1s has no significant effect on catalytic activity.

Horse complement protein C9: primary structure and cytotoxic activity.

Lack of hemolytic activity of horse serum is an inherent property of horse C9. To understand the molecular reasons for this deficiency we have cloned C9 cDNA from a horse liver cDNA library and have sequenced the cDNA yielding the complete coding sequence for horse C9. Purification of C9 from horse plasma and microsequencing established the N-terminus of the mature protein and verified that the correct horse C9 cDNA clone had been isolated. The deduced amino acid sequence corresponds to a mature protein of 526 amino acids that is 77% identical to human C9. It has the same domain structure as human C9 and contains 22 cysteines and four invariant tryptophans. The few differences include the N-terminus, which is an unblocked glycine in horse C9 but pyroglutamine in human C9, and three potential N-glycosylation sites compared to two in human C9. The N-terminal difference is unimportant since microsequencing of bovine C9, which is strongly hemolytic, established that it also has an unblocked glycine identical to horse C9. There are no obvious structural differences apparent that could resolve the differences in hemolytic potency between the two molecules. Aside from a few conservative replacements, both C9 sequences are identical between positions 250 and 360. This region includes the membrane interaction domain in C9 and the postulated transmembrane segment that is thought to constitute the wall of a putative transmembrane pore and, therefore, should be required for cytotoxicity. In agreement with this prediction we have observed that, in contrast to the marked decrease in hemolytic activity, horse C9 is very efficient in killing a variety of Gram-negative bacteria. These results demonstrate that horse C9 is a structurally competent molecule with efficient cytotoxic activity. Its inability to lyse erythrocytes may be related to the action of control proteins on target cell membranes.

Chimeric horse/human recombinant C9 proteins identify the amino acid sequence in horse C9 responsible for restriction of hemolysis.

Equine C9, in contrast to human C9, has extremely low hemolytic activity against most mammalian erythrocytes, although the amino acid sequences of both proteins show 77% identity. In an attempt to define the region of human C9 responsible for conferring its lytic activity, or conversely, the region of equine C9 responsible for its restriction, recombinant human and equine C9 and four chimeric human/equine C9 proteins were constructed and expressed in COS-7 cells. Recombinant human and equine C9 displayed hemolytic profiles similar to those of the purified native proteins. Exchange of a fragment extending from residues 145 to 290 in horse C9 with the corresponding one from human C9 created a fully hemolytic protein. This region contains the putative hinge region but not the membrane-interacting domain. Nonlytic chimeric C9 proteins inhibited hemolysis and binding of human C9 to EAC1-8 cells, indicating that they bind to their receptor, but subsequent unfolding or insertion into the membrane is impaired. These results suggest that restriction factors, such as glycophorin, CD59, or homologous restriction factor, on erythrocytes may limit the activity of horse C9 by interacting with its hinge region. In support of this conclusion direct binding of CD59 to immobilized horse C9 was detected by ligand blotting, and it was observed that a polyclonal anti-CD59 Ab enhanced human and horse C9-mediated hemolysis of human EAC1-7, but the increase in hemolytic activity of horse C9 by inhibition of CD59 was less than what could be achieved by insertion of the human C9 hinge region into horse C9.

Effects of drug-binding on the thermal denaturation of human serum albumin.

Asymmetric thermograms of defatted albumin, alone and in the presence of two model drugs, have been obtained in phosphate buffers at three pH values. The albumin is less thermally stable in the N form, but is protected by both drugs. The nonsteroidal antiinflammatory benoxaprofen offers more protection than warfarin against thermal denaturation.

The membrane attack complex of complement. Assembly, structure and cytotoxic activity.

The membrane attack complex of complement is formed by the molecular fusion of the five terminal complement proteins, C5, C6, C7, C8, and C9. While the assembly process on a target membrane and its modulation by restriction factors present on host cells is now quite well understood the molecular details of the architecture of the complex still need much further clarification. This is especially true for the interaction of the last acting protein C9, which provides the cytotoxic action of the complex, with the precursor C5b-8 complex. Because of this lack of structural details the molecular mechanisms that lead to complement-mediated cell death remain cryptic, however, it is hoped that recent advances in controlling the assembly process and in site-specific modification of the terminal complement proteins by recombinant DNA techniques should change this predicament quickly.

The expression of hemolytically active human complement protein C9 in mammalian, insect, and yeast cells.

The cDNA sequence encoding mature human C9 protein and its signal peptide was cloned into three expression vectors for expression in COS-7 (mammalian), Spodoptera frugiperda IPLB-SF-21AE (insect), and Saccharomyces cerevisiae (yeast) cells. In addition, C9 cDNA encoding only the mature protein was fused to the yeast invertase leader sequence (SUC2) and cloned for expression in yeast. Under optimal conditions COS-7 and IPLB-SF-21AE cells secreted recombinant C9 (rC9) at concentrations of about 111 and 700 ng C9/ml culture supernatant, respectively. By comparison S. cerevisiae, whether transformed with C9 cDNA containing its native or yeast invertase leader sequence, secreted only very small amounts of rC9 (5-10 ng/ml). However, upon lysis concentrations of up to 500 ng/mg dry wt were found in yeast cells transformed with C9 cDNA. SDS-PAGE followed by Western blot analysis revealed COS-7 cell and S. cerevisiae expressed rC9 to have a MW similar to that of native C9 purified from human serum, while rC9 from IPLB-SF-21AE cells was about 4 kDa smaller. No hemolytic activity of S. cerevisiae secreted rC9 could be detected and the specific hemolytic activity of S. cerevisiae intracellular rC9 was also very low. However, the specific hemolytic activities of COS-7 and IPLB-SF-21AE secreted rC9 were indistinguishable from that of purified native human C9. Thus, for future studies on the structure and function of C9 where the production of large quantities of mutant protein would be desirable, the baculovirus-insect cell expression system appears to offer considerable advantages.

Small angle neutron scattering studies of C8 and C9 and their interactions in solution.

Small angle neutron scattering (SANS) results revealed that contrary to most reports C9 is not a globular protein. Its radius of gyration (Rg) at pH 8 and an ionic strength of 0.5 is 32.2 +/- 1.4 A increasing to 35 A at physiologic ionic strength. In contrast, C8, which has a 2.2-fold larger mass, has a similar Rg value [34.6 +/- 1.6 A]. Calibration plots of Rg vs. M(r) indicate that native C8 is a spherical protein whereas native C9 is elongated. From previous reports it was known that native C8 and C9 associate in solutions of low ionic strength. SANS results confirmed this observation but also demonstrated that C8-C9 heterodimers are already formed at physiologic ionic strength. The dimeric complex is globular [Rg = 40 +/- 0.8 A] indicating that the proteins associate side-by-side rather than end-to-end. In contrast, in presence of the drug Suramin, a potent inhibitor of the assembly of the C5b-9 complex, C9 forms a complex with twice the molecular mass that is still elongated (Rg = 48.8 +/- 0.8 A), suggesting that in this case the protein dimerizes end-to-end via a bridging Suramin molecule.

Domain structure, functional activity, and polymerization of trout complement protein C9.

The 3' region of trout C9 has been resequenced and found to differ from the previously published sequence (Stanley and Herz, EMBO J. 6:1951; 1987). In contrast to other sequenced C9 molecules, but in common with the other terminal complement components, trout C9 was found to contain an additional carboxy terminal thrombospondin domain. This domain does not restrict polymerization, as has been previously suggested (Stanley and Luzio, Nature 334:475; 1988), since alternative pathway activation of trout complement by rabbit erythrocytes lead to the formation of circular membrane attack complement lesions on the erythrocyte membrane. Although the trout C9 molecule is larger than human C9, the diameters of circular trout membrane attack complexes were approximately 30% smaller than their human counterparts. No lysis of erythrocytes bearing human C5b-7 or C5b-8 complexes was detected following incubation with trout serum containing EDTA, which suggests that trout C8 and C9 are unable to bind to human C7 and C8, respectively. Finally, trout and human serum were equally effective at killing the human serum-sensitive strain Salmonella minnesota Re595.

Formation of ion-conducting channels by the membrane attack complex proteins of complement.

The effects of sequential additions of purified human complement proteins C5b-6, C7, C8, and C9 to assemble the C5b-9 membrane attack complex (MAC) of complement on electrical properties of planar lipid bilayers have been analyzed. The high resistance state of such membranes was impaired after assembly of large numbers of C5b-8 complexes as indicated by the appearance of rapidly fluctuating membrane currents. The C5b-8 induced conductance was voltage dependent and rectifying at higher voltages. Addition of C9 to membranes with very few C5b-8 complexes caused appearance of few discrete single channels of low conductance (5-25 pS) but after some time very large (greater than 0.5 nS) jumps in conductance could be monitored. This high macroscopic conductance state was dominated by 125-pS channels having a lifetime of approximately 1 s. The high conductance state was not stable and declined again after a period of 1-3 h. Incorporation of MAC extracted from complement-lysed erythrocytes into liposomes and subsequent transformation of such complexes into planar bilayers via an intermediate monolayer state resulted in channels with characteristics similar to the ones produced by sequential assembly of C5b-9. Comparison of the high-conductance C5b-9 channel characteristics (lifetime, ion preference, ionic-strength dependence) with those produced by poly(C9) (the circular or tubular aggregation product of C9) as published by Young, J.D.-E., Z.A. Cohn, and E.R. Podack. (1986. Science [Wash. DC]. 233:184-190.) indicates that the two are significantly different.

Thermal unfolding and aggregation of human complement protein C9: a differential scanning calorimetry study.

The thermotropic behavior of purified human complement protein C9 was investigated by high-sensitivity differential scanning calorimetry. When dissolved in physiological buffers (pH 7.2, 150 mM NaCl), C9 underwent three endothermic transitions with transition temperatures (Tm) centered at about 32, 48, and 53 degrees C, respectively, and one exothermic transition above 64 degrees C that correlated with protein aggregation. The associated calorimetric enthalpies of the three endothermic transitions were about 45, 60, and 161 kcal/mol with cooperative ratios (delta Hcal/delta HvH) close to unity. The total calorimetric enthalphy for the unfolding process was in the range of 260-280 kcal/mol under all conditions. The exothermic aggregation temperature was strongly pH dependent, changing from 60 degrees C at pH 6.6 to 81.4 degrees C at pH 8.0, whereas none of the three endothermic transitions was significantly affected by pH changes. They were, however, sensitive to addition of calcium ions; most affected was Tm1 which shifted from 32 to 35.8 degrees C in the presence of 3 mM calcium, i.e., the normal blood concentration. Kosmotropic ions stabilized the protein by shifting the endothermic transitions to slightly higher temperatures whereas inclusion of chaotropic ions (such as choline), removal of bound calcium by addition of EDTA, or proteolysis with thrombin lowered the transition temperatures. Previous studies had indicated the formation of at least three different forms of C9 during membrane insertion or during heat polymerization, and it is suggested that the three endothermic transitions reflect the formation of such C9 conformers. Choline, which is present at high concentrations on the surface of biological membranes, and calcium ions have the ability to shift the transition temperatures of the first two transitions to be either close to or below body temperature. Thus, it is very likely that C9 is present in vivo in a partially unfolded state when bound to a membrane surface, and we propose that this facilitates membrane insertion and refolding of the protein into an amphiphilic conformation.

Detection of refolding conformers of complement protein C9 during insertion into membranes.

Human complement protein C9 is a hydrophilic serum glycoprotein responsible for efficient expression of the cytotoxic and cytolytic functions of complement. It assembles on the surface of a target cell together with C5, C6, C7 and C8 to form the membrane attack complex (MAC) and therefore has to change structure to become an integral membrane protein. As the protein assumes a stable structure in an aqueous environment, the question arises as to how it can enter the hydrophobic interior of a membrane. During MAC assembly C9 polymerizes into a circular structure, termed poly(C9) (ref. 8), which is responsible for the cylindrical electron microscopic appearance of the MAC. The suggestion has been made that C9 must at least partly unfold in order to enter a membrane and also that polymerization of the molecule is intimately linked to insertion and cytotoxicity. The extent of unfolding and the mechanism of polymerization are not understood, nor is it known precisely which parts of the molecule participate in the proposed structural changes. We have been able to capture refolding C9 conformers during membrane insertion with the help of sequence-specific anti-peptide antibodies. Some of these antibodies inhibit C9-mediated haemolysis but not C9 polymerization, while others have the opposite effect. This suggests that the two processes are independent.

Identification of the discontinuous epitope in human complement protein C9 recognized by anti-melittin antibodies.

Polyclonal rabbit antibodies against melittin recognize human C protein C9 and retard C9-mediated hemolysis. Human C9 contains a tetrameric and a pentameric sequence (amino acids 293-296 and 528-532, respectively) that together match a continuous segment in the melittin sequence, i.e., residues 8-16. It has been suggested that the tetrameric and the pentameric regions on C9 form a discontinuous epitope on folded C9 that mimics the structure of melittin. To further test this hypothesis, antibodies to C9-sequence-specific peptides were prepared. Peptides containing either the homologous tetrameric or the homologous pentameric sequence together with short stretches of the respective amino- and carboxyl-terminal flanking regions were synthesized, as well as a composite peptide predicted to resemble the discontinuous epitope as a linear, nine-amino acid sequence. Direct and competitive binding assays demonstrated that the tetrameric and the pentameric sequences are part of the epitope on human C9 that is recognized by anti-melittin IgG. However, only antibodies directed against the complete epitope are capable of inhibiting hemolysis. Because neither anti-tetramer nor anti-pentamer antibodies affect hemolysis whereas anti-melittin and anti-composite antibodies do, we propose that human C9 changes conformation around a hinge located between residues 296 and 528 and that the latter two antibodies inhibit unfolding required for membrane insertion and subsequent hemolysis.

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)

Comparison between complement and melittin hemolysis: anti-melittin antibodies inhibit complement lysis.

A comparison is made between the hemolytic actions of melittin and the ninth component of complement (C9). Melittin and C9 produce "pores" of similar effective radius in erythrocytes under standardized conditions, and their hemolytic action is suppressed by metal ions at similar concentrations, suggesting a common mechanism. Polyclonal anti-melittin immunoglobulin G (IgG) produced in rabbits retards hemolysis mediated by human C9 in a specific manner. Such antibodies react in several immunoassays with human and monkey C9 but not with C9 from lower animals, and no inhibition of lysis mediated by C9 molecules from these animals is observed. Thus, it is unlikely that anti-melittin IgG reacts with a structural element, such as an amphipathic helix, on human C9 since such structures are also predicted to exist in other C9 molecules. Human C9 and melittin block cross-reactivity in a dose-dependent manner, and anti-melittin IgG recognizes an epitope located between amino acid residues 245 and 390 of human C9 on "Western" blots. Comparison of the melittin and human C9 sequences indicates two regions of complete homology, a tetrapeptide at positions 292-295, and a pentapeptide at positions 527-531 in human C9, corresponding to residues 8-16 in melittin. Inhibition of hemolysis is not caused by blocking of C9 binding to the C5b-8 complex; rather the antibody must dissociate from the bound C9 before lysis ensues, indicating that it interferes with a postbinding event. It is proposed that anti-melittin binds to a conformational epitope on native, folded human C9 and thereby retards unfolding of the molecule, which is required for membrane insertion and hemolysis.