Surface location and high affinity for calcium of a 500-kd liver membrane protein closely related to the LDL-receptor suggest a physiological role as lipoprotein receptor

We describe a cell surface protein that is abundant in liver and has close structural and biochemical similarities to the low density lipoprotein (LDL) receptor. The complete sequence of the protein containing 4544 amino acids is presented. From the sequence a remarkable resemblance to the LDL-receptor and epidermal growth factor (EGF) precursor is apparent. Three types of repeating sequence motifs entirely account for the extracellular domain of the molecule. These are arranged in a manner resembling four copies of the ligand binding and the EGF-precursor homologous region of the LDL-receptor. Following a proline-rich segment of 17 amino acids are found six consecutive repeats with close homology to EGF. A single membrane-spanning segment precedes a carboxy-terminal 'tail' of 100 amino acids. This contains two seven-amino acid sequences with striking homology to the cytoplasmic tail of the LDL-receptor in the region that contains the signal for clustering into coated pits. The mRNA for this protein is most abundant in liver, brain and lung. By using an antibody raised against a 13-amino acid peptide corresponding to the deduced amino acid sequence of the carboxy-terminus of the protein we have demonstrated its existence on the cell surface and its abundance in liver. Like the LDL-receptor this protein also strongly binds calcium, a cation absolutely required for binding of apolipoproteins B and E to their receptors. We propose that this LDL-receptor related protein (LRP) is a recycling lipoprotein receptor with possible growth-modulating effects.

The LDL-receptor-related protein, LRP, is an apolipoprotein E-binding protein

The low-density-lipoprotein (LDL) receptor is a cell-surface protein that plays an important part in the metabolism of cholesterol by mediating the uptake of LDL from plasma into cells. Although LDL particles bind to the LDL receptor through their apolipoprotein B (apo B) and apolipoprotein E (apo E) moieties, other apo E-containing particles, like chylomicron remnants, are not dependent on the LDL receptor for uptake into cells. Chylomicrons formed in the intestinal mucosa during the absorption of the products of digestion, are processed by the peripheral circulation by lipoprotein lipase, which catalyses the breakdown of triglycerides in chylomicrons to free fatty acids and glycerol. The resulting chylomicron remnants, which are cholesterol-rich lipoproteins, are subsequently taken up in the liver. A second distinct protein that binds to apo E-containing lipoproteins, but not to LDL, has been proposed to be the receptor mediating the clearance of chylomicron remnants from the plasma. This protein has a relative molecular mass (Mr) of 56,000 (56K). More recent studies have failed, however, to establish whether this protein is a cell-surface receptor. Here we describe crosslinking experiments in which apo E liposomes were found to bind specifically to the cell surface of hepG2 cells and to human liver membranes. The size and immunological cross-reactivity of the protein to which the liposomes bound was indistinguishable from that of the recently cloned and sequenced LDL-receptor-related protein, LRP. We therefore conclude that the LRP might function as an apo E receptor.

Construction of a new family of high efficiency bacterial expression vectors: identification of cDNA clones coding for human liver proteins

Construction of a family of bacterial expression vectors, pEX1-3, is described. These vectors are derived from a cro-lacZ gene fusion plasmid which expresses large quantities of fusion protein under the control of the PR promoter of bacteriophage lambda. A polylinker has been engineered into the 3' end of the lacZ gene in all three translational reading frames, and stop signals for transcription and translation inserted, so that any open reading frame DNA may be expressed as a hybrid beta-galactosidase protein. cDNA fragments cloned in these vectors can be detected with an efficiency of greater than 1 in 3, thus enabling the detection of rare cDNA molecules. In addition, the low solubility of hybrid proteins leads to a rapid isolation procedure allowing antibodies of pre-determined specificity to be made against expressed regions of cloned DNA. We describe the cloning of albumin and complement C9 genes from a human cDNA library using polyclonal and monoclonal antibodies.

High density lipoprotein is the major carrier of lipid hydroperoxides in human blood plasma from fasting donors

Analysis of untreated fresh blood plasma from healthy, fasting donors revealed that high density lipoprotein (HDL) particles carry most (approximately 85%) of the detectable oxidized core lipoprotein lipids. Low density lipoprotein (LDL) lipids are relatively peroxide-free. In vitro the mild oxidation of gel-filtered plasma from fasting donors with a low, steady flux of aqueous peroxyl radicals initially caused preferential oxidation of HDL rather than LDL lipids until most ubiquinol-10 present in LDL was consumed. Thereafter, LDL core lipids were oxidized more rapidly. Isolated lipoproteins behaved similarly. Preferential accumulation of lipid hydroperoxides in HDL reflects the lack of antioxidants in most HDL particles compared to LDL, which contained 8-12 alpha-tocopherol and 0.5-1.0 ubiquinol-10 molecules per particle. Cholesteryl ester hydroperoxides (CEOOHs) in HDL and LDL were stable when added to fresh plasma at 37 degrees C for up to 20 hr. Transfer of CEOOHs from HDL to LDL was too slow to have influenced the in vitro plasma oxidation data. Incubation of mildly oxidized LDL and HDL with cultured hepatocytes afforded a linear removal of CEOOHs from LDL (40% loss over 1 hr), whereas a fast-then-slow biphasic removal was observed for HDL. Our data show that HDL is the principal vehicle for circulating plasma lipid hydroperoxides and suggest that HDL lipids may be more rapidly oxidized than those in LDL in vivo. The rapid hepatic clearance of CEOOHs in HDL could imply a possible beneficial role of HDL by attenuating the build-up of oxidized lipids in LDL.

Identification, sequencing and expression of an integral membrane protein of the trans-Golgi network (TGN38)

Organelle-specific integral membrane proteins were identified by a novel strategy which gives rise to monospecific antibodies to these proteins as well as to the cDNA clones encoding them. A cDNA expression library was screened with a polyclonal antiserum raised against Triton X-114-extracted organelle proteins and clones were then grouped using antibodies affinity-purified on individual fusion proteins. The identification, molecular cloning and sequencing are described of a type 1 membrane protein (TGN38) which is located specifically in the trans-Golgi network.

Molecular cloning of S-protein, a link between complement, coagulation and cell-substrate adhesion

cDNA clones coding for human S-protein have been isolated using monoclonal antibodies to screen a cDNA library in pEX. These clones are shown to be authentic S-protein clones on the basis of sequence, composition and immunological criteria. The complete open reading frame sequence for S-protein has been determined and shows it to be a single polypeptide chain of 459 amino acids preceded by a cleaved leader peptide of 19 residues. No evidence was found for polymorphism of S-protein suggesting that different molecular weight forms arise by proteolytic degradation. Of the first 44 amino-terminal residues 42 are identical with the so-called somatomedin B peptide suggesting that S-protein is the somatomedin B precursor. Striking homology is found in the rest of the sequence with the serum spreading factor, vitronectin, which has also been shown to contain somatomedin B sequences at its amino terminus. We conclude that S-protein and vitronectin are identical and discuss the relevance of this finding to the coagulation and complement pathways.

TGN38 is maintained in the trans-Golgi network by a tyrosine-containing motif in the cytoplasmic domain

Sorting of proteins destined for different plasma membrane domains, lysosomes and secretory pathways takes place in the trans-Golgi network (TGN). TGN38 is an integral membrane protein found in this intracellular compartment. We show that TGN38 contains an autonomous targeting signal within its cytoplasmic domain which determines its intracellular location. Deletion analysis and site-directed mutagenesis of this domain demonstrate that a tyrosine motif homologous to the internalization signal of surface receptors is necessary and sufficient for correct localization. These findings suggest that TGN38 is maintained in the TGN by retrieval from the plasma membrane and employs a different mechanism for retention from that of the transferase enzymes of the trans-Golgi.

Structural/functional similarity between proteins involved in complement- and cytotoxic T-lymphocyte-mediated cytolysis

Cytolysis mediated by complement or cytolytic lymphocytes results in the formation of morphology similar lesions in the target membrane. These lesions, formed by the polymerization of C9 or perforin respectively, contribute the major killing action by causing osmotic lysis of the target cell. Following the suggestion of Mayer that the mechanisms of humoral and cell-mediated cytotoxicity might be related, studies into the morphology of the membrane lesions formed, and the proteins responsible for causing the lesions, have shown several similarities. While the lesion caused by natural and T-killer cells is a little larger than that caused by complement, its overall shape is similar and in both cases the cylindrical pore is formed by polymerization of a monomeric subunit, C9 (relative molecular mass, Mr = 71,000) for complement, and perforin (Mr = 66,000) for cell-mediated cytotoxicity. C9 has an absolute requirement for a receptor in the target membrane formed by the earlier membrane attack complex components, C5b, C6, C7 and C8 (ref. 8). For perforin, polymerization in a target membrane requires no receptor, specificity being derived from the specific recognition between killer and target cell. Both proteins can be made to polymerize in vitro by the addition of divalent cations (Zn2+ for C9 (ref. 16) and Ca2+ for perforin) and the resultant complexes closely resemble their physiological counterparts. Antibodies raised against lymphocyte-killed targets have also been shown to cross-react with complement proteins, but the antigenically related proteins were not determined in these studies. We show here using purified proteins that perforin, C9 and complexes involving C7 and C8 share a common antigenic determinant which is probably involved in polymerization.

Enhanced expression of cro-beta-galactosidase fusion proteins under the control of the PR promoter of bacteriophage lambda

Hybrid plasmids carrying cro-lacZ gene fusions have been constructed by joining DNA segments carrying the PR promoter and the start of the cro gene of bacteriophage lambda to the lacZ gene fragment carried by plasmid pLG400 . Plasmids in which the translational reading frames of the cro and lacZ genes are joined in-register (type I) direct the synthesis of elevated levels of cro-beta-galactosidase fusion protein amounting to 30% of the total cellular protein, while plasmids in which the genes are fused out-of-register (type II) produce a low level of beta-galactosidase protein. Sequence rearrangements downstream of the cro initiator AUG were found to influence the efficiency of translation, and have been correlated with alterations in the RNA secondary structure of the ribosome-binding site. Plasmids which direct the synthesis of high levels of beta-galactosidase are conditionally lethal and can only be propagated when the PR promoter is repressed. Deletion of sequences downstream of the lacZ gene restored viability, indicating that this region of the plasmid encodes a function which inhibits the growth of the cells. The different applications of these plasmids for expression of cloned genes are discussed.

Subcellular distribution and movement of 5'-nucleotidase in rat cells

1. Cell-surface 5'-nucleotidase was assayed by incubating whole-cell suspensions with 5'[3H]-AMP in iso-osmotic buffer and measuring [3H]adenosine production. The activity of cell-surface 5'-nucleotidase in hepatocytes, adipocytes and lymphocytes isolated from the rat was 15.0, 0.5 and 0.8pmol/min per cell at 37 degrees C respectively. 2. Disruption of the cells by vigorous mechanical homogenization or detergent treatment exposed additional 5'-nucleotidase activity, which represented 52%, 25% and 21% of the total activity in the three cell types respectively. This increase in 5'-nucleotidase activity which occurred when the cells were homogenized was due to a second pool of 5'-nucleotidase within the cell, rather than activation of the cell-surface enzyme. 3. In hepatocytes the intracellular 5'-nucleotidase activity was membrane-bound, indistinguishable from cell-surface 5'-nucleotidase in its inhibition by rabbit anti-(rat liver 5'-nucleotidase) serum and its kinetics with AMP, and was located on the extracytoplasmic face of vesicles within the cell. 4. The cell-surface 5'-nucleotidase of rat hepatocytes was rapidly inhibited when rabbit anti-(rat liver 5'-nucleotidase) serum or concanavalin A was added to the medium at 37 degrees C. Incubation with antiserum for 5 min at 37 degrees C inhibited 83 +/- 3% of the cell-surface enzyme. 5. Incubation of hepatocytes with exogenous antiserum or concanavalin A for 30 min at 37 degrees C resulted in over 50% inhibition of the intracellular enzyme. This inhibition was not prevented by disruption of the cytoskeleton or by ATP depletion. 6. Incubation of hepatocytes with exogenous antiserum or concanavalin A for up to 2h at 0 degrees C caused little or no inhibition of the intracellular enzyme, but over 75% inhibition of the cell-surface enzyme. 7. When surface-inhibited hepatocytes were washed and resuspended in buffer at 37 degrees C, 5'-nucleotidase was observed to redistribute from the intracellular pool to the cell surface.

The sequence and topology of human complement component C9

A partial nucleotide sequence of human complement component C9 cDNA representing 94% of the coding region of the mature protein is presented. The amino acid sequence predicted from the open reading frame of this cDNA concurs with the amino acid sequence at the amino-terminal end of three proteolytic fragments of purified C9 protein. No long stretches of hydrophobic residues are present, even in the carboxy-terminal half of the molecule which reacts with lipid-soluble photoaffinity probes. Monoclonal antibody epitopes have been mapped by comparing overlapping fragments of C9 molecule to which the antibodies bind on Western blots. Several of these epitopes map to small regions containing other surface features (e.g., proteolytic cleavage sites and N-linked oligosaccharide). The amino-terminal half of C9 is rich in cysteine residues and contains a region with a high level of homology to the LDL receptor cysteine-rich domains. A model for C9 topology based on these findings is proposed.

Androgen receptor expression is greater in macrophages from male than from female donors. A sex difference with implications for atherogenesis

BACKGROUND

Male sex is an independent risk factor for the extent and severity of atherosclerosis. The influence of androgens on foam cell formation, a key event in atherogenesis, has not yet been investigated.

METHODS AND RESULTS

Primary human monocytes were allowed to differentiate into macrophages. RNA was then extracted from healthy male-donor (n=8) and premenopausal female-donor (n=8) macrophages, and message for the androgen receptor (AR) was examined by RT-PCR. There was a significantly higher level of AR mRNA in macrophages isolated from men than in those from women (0.64+/-0.06 versus 0.15+/-0.02 amol/microgram total RNA; P<0.001). AR mRNA levels were similar in macrophages from postmenopausal and premenopausal women (P=0.16). The functional consequence of this sex difference was then explored. Lipid-loading studies were performed on male (n=9) macrophages treated with the androgen dihydrotestosterone (DHT) and/or the AR antagonist hydroxyflutamide. These showed that DHT caused a dose-dependent and receptor-mediated increase in macrophage cholesteryl ester content (109+/-10%, 117+/-3%, and 120+/-4% for 4, 40, and 400 nmol/L DHT, respectively, as a percentage of control, P=0.002; 95+/-8% for DHT with hydroxyflutamide, P=0.58 versus controls). By contrast, there was no significant effect of androgen on lipid loading in female-donor macrophages (P>0.2 versus controls).

CONCLUSIONS

Sex differences in androgen-mediated macrophage lipid loading may contribute to the greater prevalence and severity of atherosclerosis in men.

Repeating structure of chick tropoelastin revealed by complementary DNA cloning

A cDNA library was constructed from chick aorta poly(adenylic acid)-containing RNA in the expression vector pEX1. Several clones were identified by screening the library with a polyclonal antiserum raised against chick tropoelastin and confirmed by DNA sequencing. Analysis of the deduced amino acid sequence, corresponding to the mature tropoelastin and most of the signal peptide, revealed that the molecule is composed of at least 8, and possibly 13, repeating units. The common features of each unit include an N-terminal region composed largely of alanines and lysines and ending with an aromatic amino acid, followed by a GAG span and then a C-terminal region consisting mostly of valines, prolines, and glycines often present in several copies of the sequence (VPGV). This structure is discussed in terms of the functional properties of the molecule.

The propeptide of macrophage inhibitory cytokine (MIC-1), a TGF-beta superfamily member, acts as a quality control determinant for correctly folded MIC-1

Macrophage inhibitory cytokine (MIC-1), a divergent member of the transforming growth factor-beta (TGF-beta) superfamily and activation associated cytokine, is secreted as a 28 kDa dimer. To understand its secretion, we examined its processing in MIC-1-transfected Chinese hamster ovary cells. Mature MIC-1 dimer arises post-endoplasmic reticulum (ER) by proteolytic cleavage of dimeric pro-MIC-1 precursor at a furin-like site. Unlike previously characterized TGF-beta superfamily members, MIC-1 dimers are also secreted in constructs lacking the propeptide. A clue to the function of the propeptide came from the observation that a range of proteasome inhibitors, including lactacystin and MG132, cause major increases in levels of undimerized pro-MIC-1 precursor. There was no effect of proteasome inhibitors on cells expressing mature MIC-1 without the propeptide, suggesting that the propeptide can signal misfolding of MIC-1, leading to proteasomal degradation. Deletion mutagenesis showed the N-terminal 28 amino acids of the propeptide are necessary for proteasomal degradation. This is the first demonstration, to our knowledge, of a quality control function in a propeptide domain of a secretory protein and represents an additional mechanism to ensure correct folding of proteins leaving the ER.

Increased plasma interleukin-7 level correlates with decreased CD127 and Increased CD132 extracellular expression on T cell subsets in patients with HIV-1 infection

BACKGROUND

Interleukin (IL)-7 levels are increased in patients with human immunodeficiency virus type 1 (HIV-1)-associated lymphopenia; however, the effects of this on IL-7 receptor (IL-7R) expression, disease progression, and immune reconstitution remain unclear.

METHODS

Plasma IL-7 levels were measured, by enzyme-linked immunoassay, in patients with primary, chronic, or long-term nonprogressive HIV-1 infection (PHI, CHI, and LTNP, respectively) before and after 40-48 weeks of antiretroviral therapy (ART). Cell-surface expression and intracellular expression of the IL-7R components CD127 and CD132 were measured by flow cytometry. The effects of IL-7 and cycloheximide on IL-7R expression by peripheral blood mononuclear cells were examined in vitro.

RESULTS

Plasma IL-7 levels were increased in both patients with PHI and those with CHI; administration of ART resulted in normalized plasma IL-7 levels in patients with PHI but not in those with CHI. Plasma IL-7 levels positively correlated with CD4(+) T cell immune reconstitution in patients with PHI. In vitro, exogenous IL-7 rapidly down-regulated cell-surface CD127 expression, but not CD132 expression, whereas subsequent reexpression required active protein synthesis. HIV-1 infection resulted in progressive decreases in the CD127(+)132(-) subset and increases in the CD127(-)132(+) subset of CD4(+) and CD8(+) T cells. Changes in CD4(+) T cell expression of IL-7R components were evident in patients with LTNP who lost viral control, and these changes preceded increases in plasma IL-7 levels.

CONCLUSIONS

Perturbations in the IL-7/IL-7R system were clearly associated with disease progression but did not reliably predict immune reconstitution.

The role of intermediates in mitochondrial fatty acid oxidation

1. Rat liver mitochondria oxidizing [16-14C]palmitoylcarnitine accumulate saturated long-chain thiester intermediates which may be detected by radio-g.1.c.2. Time-courses of intermediate accumulation display no product-precursor relationships and the end product, measured as [14C]citrate, is produced without a detectable initial lag. 3. A short pulse of [16-14C]palmitoylcarnitine followed by unlabelled palmitoylcarnitine showed that the observed intermediates(at least in the greater part)were not the direct precursors of [14C]citrate. 4. The quantity of saturated intermediates depended on the total accumulated flux of acyl units through the pathway provided that some mitochondrial CoA and unused substrate remained. 5. In the presence of rotenone and carnitine, 2-unsaturated, 3-unsaturated and 3-hydroxy intermediates were formed as well as saturated intermediates...

Efficient construction of cDNA libraries in plasmid expression vectors using an adaptor strategy

We describe a method for the construction of large DNA fragment libraries in plasmid vectors, in which complementary, single-stranded extensions are ligated onto both vector and insert DNA using un-phosphorylated adaptor oligonucleotides. Special consideration has been taken of the requirements of expression screening as follows: cDNA synthesis using random oligonucleotide primers is described which maximises the probability of obtaining open reading frame fragments from large mRNA molecules, the adaptors use codons found in high abundance E. coli proteins to minimise problems of premature termination when using strong promoters, and the sequence encoded by the adaptors, when cloned into the bacterial expression vector pEX1, promotes a surface location for the foreign antigenic determinant where it is accessible to antibodies used for screening.

Multiplexed tandem PCR: gene profiling from small amounts of RNA using SYBR Green detection

Multiplexed tandem PCR (MT-PCR) is a process for highly multiplexed gene expression profiling. In the first step, multiple primer pairs are added to the RNA to be analysed together with reverse transcriptase and Taq DNA polymerase. Following reverse transcription, the multiplexed amplicons are simultaneously amplified for a small number of cycles so as to avoid competition between amplicons. The reaction product is then diluted and analysed in multiple individual PCRs using primers nested inside the primers used for the multiplexed amplification. As the second PCR uses a template enriched in the amplicons of interest, the conditions can be optimized to significantly reduce ‘primer dimer’ formation allowing SYBR Green chemistry to be used for quantification. MT-PCR can be configured for as little as 10 pg RNA (equivalent to a single mammalian cell) and works well with RNA extracted from archival formalin-fixed paraffin-embedded sections. We illustrate MT-PCR with gene expression profiles of breast cancer cell lines.

Topological mapping of complement component C9 by recombinant DNA techniques suggests a novel mechanism for its insertion into target membranes

cDNA molecules coding for mouse and trout C9 have been isolated and the derived amino acid sequences compared with that of human C9. Regions of high homology between the closely related species (mouse and human) correlate with putative domains in the protein structure supporting a model of C9 having five globular domains. Comparison between the more distant species (trout and human) suggests regions of particular importance to C9 structure and function. In addition the three related sequences allow the secondary structure to be predicted with more confidence and we have tested the prediction by mapping surface features of the protein. Reported here is a recombinant DNA approach to fine mapping of antibody epitopes. Two of the putative domains of C9 are connected by a stretch of about 40 amino acid residues in which features characteristic of individual conformational forms of C9 are concentrated. We suggest that this region might act as a hinge allowing the rearrangement of globular domains necessary for membrane insertion. In the membrane inserting domain one highly conserved sequence has the potential to form an amphipathic alpha-helix once it is buried in the lipid bilayer. These features suggest a novel mechanism for the irreversible, post-translational insertion of C9 into target membranes.

Solubilization and immune-detection of beta-galactosidase hybrid proteins carrying foreign antigenic determinants

Three DNA fragments representing almost the entire E1 gene of Semliki Forest virus (SFV) were inserted into a cro-lacZ expression vector by oligo dC.oligo dG tailing. Fragments inserted close to the 5' end of the lacZ gene gave rise to hybrid proteins which were rapidly degraded. Insertion of the same fragments at the 3' end, however, resulted in the synthesis of stable hybrid proteins which precipitated in an insoluble form within the bacteria. Insufficient hybrid protein was soluble to allow detection by immunoradiometric assay. Colonies grown on nitrocellulose filters, however, could be solubilized in SDS and subsequently renatured such that antibodies raised against the intact or SDS-denatured E1 protein cross-reacted with the hybrid proteins in a high percentage of colonies. This model system demonstrates a simple procedure for identifying DNA exon fragments by the immunological detection of expressed hybrid proteins.

Clathrin heavy chain, light chain interactions

Purified pig brain clathrin can be reversibly dissociated and separated into heavy chain trimers and light chains in the presence of non-denaturing concentrations of the chaotrope thiocyanate. The isolated heavy chain trimers reassemble into regular polygonal cage structures in the absence of light chains. The light chain fraction can be further resolved into its two components L alpha and L beta which give different one-dimensional peptide maps. Radiolabelled light chains bind with high affinity (KD < 10(-10) M) to heavy chain trimers, to heavy chain cages and to a 110,000 mol. wt. tryptic fragment of the heavy chain. Both light chains compete with each other and with light chains from other sources for the same binding sites on heavy chains and c.d. spectroscopy shows that the two pig brain light chains possess very similar structures. We conclude that light chains from different sources, despite some heterogeneity, have a highly conserved, high affinity binding site on the heavy chain but are not essential for the formation of regular cage structures.

Intracellular targetting signals of polymeric immunoglobulin receptors are highly conserved between species

A rat liver cDNA library, constructed in the plasmid expression vector pUEX, was immunoscreened using a rabbit polyclonal antiserum raised against rat liver Golgi membrane proteins. A sub-set of isolated clones were shown to encode the rat polymeric immunoglobulin receptor (pIgR). DNA sequence analysis of these clones provided the complete coding sequence of rat pIgR. Subsequent alignment of rat, rabbit and human predicted amino acid sequences demonstrated that the greatest degree of homology between the three pIgRs lies in their cytoplasmic tails; a region previously shown to be important for correct targetting and trancytosis of rabbit pIgR [(1984) Nature 308, 37-43].

Complementary DNA cloning of complement C8 beta and its sequence homology to C9

The complete amino acid sequence of mature C8 beta has been derived from the DNA sequence of a cDNA clone identified by expression screening of a human liver cDNA library. Comparison with the amino acid sequence of C9 shows an overall homology with few deletions and insertions. In particular, the cysteine-rich domains and membrane-inserting regions of C9 are well conserved. These findings are discussed in relation to a possible mechanism of membrane attack complex formation.