Interaction between the C8 alpha-gamma and C8 beta subunits of human complement C8: role of the C8 beta N-terminal thrombospondin type 1 module and membrane attack complex/perforin domain

Human C8 is one of five complement components (C5b, C6, C7, C8, and C9) that interact to form the cytolytic membrane attack complex (MAC). It is an oligomeric protein composed of a disulfide-linked C8alpha-gamma heterodimer and a noncovalently associated C8beta chain. C8alpha and C8beta are homologous; both contain an N-terminal thrombospondin type 1 (TSP1) module, a low-density lipoprotein receptor class A (LDLRA) module, an extended central segment referred to as the membrane attack/perforin (MACPF) domain, an epidermal growth factor (EGF) module, and a second TSP1 module at the C-terminus. In this study, the segment of C8beta that confers binding specificity toward C8alpha-gamma was identified using recombinant C8beta constructs in which the N- and/or C-terminal modules were deleted or exchanged with those from C8alpha. Constructs were tested for their ability to bind C8alpha-gamma in solution and express C8 hemolytic activity. Binding to C8alpha-gamma was found to be dependent on the TSP1 + LDLRA + MACPF segment of C8beta. Within this segment, the TSP1 module and MACPF domain are principally involved and act cooperatively to mediate binding. Results from activity assays suggest that residues within this segment also mediate binding and incorporation of C8 into the MAC.     

Functional studies of the MACPF domain of human complement protein C8alpha reveal sites for simultaneous binding of C8beta, C8gamma, and C9

Human C8 is one of five components of the membrane attack complex of complement (MAC). It contains three subunits (C8alpha, C8beta, C8gamma) arranged as a disulfide-linked C8alpha-gamma dimer that is noncovalently associated with C8beta. C8alpha, C8beta, and complement components C6, C7, and C9 form the MAC family of proteins. All contain N- and C-terminal modules and an intervening 40-kDa segment referred to as the membrane attack complex/perforin (MACPF) domain. During MAC formation, C8alpha binds and mediates the self-polymerization of C9 to form a pore-like structure on target cells. The C9 binding site was previously shown to reside within a 52-kDa segment composed of the C8alpha N-terminal modules and MACPF domain (alphaMACPF). In the present study, we examined the role of the MACPF domain in binding C9. Recombinant alphaMACPF and a disulfide-linked alphaMACPF-gamma dimer were successfully produced in Escherichia coli and purified. alphaMACPF was shown to simultaneously bind C8beta, C8gamma, and C9 and form a noncovalent alphaMACPF.C8beta.C8gamma.C9 complex. Similar results were obtained for the recombinant alphaMACPF-gamma dimer. This dimer bound C8beta and C9 to form a hemolytically active (alphaMACPF-gamma).C8beta.C9 complex. These results indicate that the principal binding site for C9 lies within the MACPF domain of C8alpha. They also suggest this site and the binding sites for C8beta and C8gamma are distinct. alphaMACPF is the first human MACPF domain to be produced recombinantly and in a functional form. Such a result suggests that this segment of C8alpha and corresponding segments of the other MAC family members are independently folded domains.     

The heparin binding domain of S-protein/vitronectin binds to complement components C7, C8, and C9 and perforin from cytolytic T-cells and inhibits their lytic activities

S-Protein/vitronectin is a serum glycoprotein that inhibits the lytic activity of the membrane attack complex of complement, i.e., of the complex including the proteins C5b, C6, C7, C8, and C9n. We show that intact S-protein/vitronectin or its cyanogen bromide generated fragments also inhibit the hemolysis mediated by perforin from cytotoxic T-cells at 45 and 11 microM, respectively. The glycosaminoglycan binding site of S-protein/vitronectin is responsible for the inhibition, since a synthetic peptide corresponding to a part of this highly basic domain (amino acid residues 348-360) inhibits complement- as well as perforin-mediated cytolysis. In the case of C9, the synthetic peptide binds to the acidic residues occurring in its N-terminal cysteine-rich domain (residues 101-111). Antibodies raised against this particular segment react 25-fold better with the polymerized form of C9 as compared with its monomeric form, indicating that this site becomes exposed only upon the hydrophilic-amphiphilic transition of C9. Since the cysteine-rich domain of C9 has been shown to be highly conserved in C6, C7, and C8 as well as in perforin, the inhibition of the lytic activities of these molecules by S-protein/vitronectin or by peptides corresponding to its heparin binding site may be explained by a similar mechanism.     

The structure of human thrombospondin, an adhesive glycoprotein with multiple calcium-binding sites and homologies with several different proteins

Thrombospondin is one of a class of adhesive glycoproteins that mediate cell-to-cell and cell-to-matrix interactions. We have used two monoclonal antibodies to isolate cDNA clones of thrombospondin from a human endothelial cell cDNA library and have determined the complete nucleotide sequence of the coding region. Three regions of known amino acid sequence of human platelet thrombospondin confirm that the clones are authentic. Three types of repeating amino acid sequence are present in thrombospondin. The first is 57 amino acids long and shows homology with circumsporozoite protein from Plasmodium falciparum. The second is 50-60 amino acids long and shows homology with epidermal growth factor precursor. The third occurs as a continuous eightfold repeat of a 38-residue sequence; structural homology with parvalbumin and calmodulin indicates that these repeats constitute the multiple calcium-binding sites of thrombospondin. The amino acid sequence arg-gly-asp-ala is included in the last type 3 repeat. This sequence is probably the site for the association of thrombospondin with cells. In addition, localized homologies with procollagen, fibronectin, and von Willebrand factor are present in one region of the thrombospondin molecule.     

The plasma cell membrane glycoprotein, PC-1, is a threonine-specific protein kinase stimulated by acidic fibroblast growth factor.

A 32P-labeled protein that co-purified with acidic fibroblast growth factor (aFGF) receptor from bovine liver proved to be a distinct membrane protein, which itself has kinase activity that is stimulated by aFGF. The protein was designated MAFP for major aFGF-stimulated phosphoprotein. MAFP was purified from bovine liver using immunoaffinity chromatography with monoclonal antibody to MAFP following Triton X-100 extraction of plasma membranes and wheat germ lectin-Sepharose 4B column chromatography. The purified MAFP showed molecular masses of 130 kDa and 260 kDa on sodium dodecyl sulfate-polyacrylamide gel electrophoresis under reducing and nonreducing conditions, respectively. Purified MAFP elicited aFGF-stimulated Thr-specific autophosphorylation activity and phosphorylation activity toward protein substrates (myelin basic protein and histone). Amino acid sequence analyses of 16 peptide fragments of MAFP, produced by endoproteinase Lys-C digestion followed by reduction and S-pyridylethylation, showed approximately 80-100% homology with the cDNA-deduced amino acid sequences of human and mouse plasma cell membrane glycoprotein, PC-1 (Buckley, M. F., Loveland, K. A., McKinstry, W. J., Garson, O. M., and Goding, J. W. (1990) J. Biol. Chem. 265, 17506-17511), suggesting that MAFP is the bovine version of PC-1. The amino acid sequences of bovine MAFP, human and mouse PC-1 reveal a putative ATP binding site in their extracellular domains. These results suggest that MAFP(PC-1) is an ectoprotein kinase. In addition to the kinase activity, MAFP(PC-1) was also found to possess alkaline nucleotide phosphodiesterase activity. It is now clear that several of the unique properties previously attributed to the aFGF receptor kinase are actually properties of this novel Thr-specific ectoprotein kinase, which co-purifies with the aFGF receptor and is responsive to stimulation by aFGF.     

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.     

Chimeric and truncated forms of human complement protein C8 alpha reveal binding sites for C8 beta and C8 gamma within the membrane attack complex/perforin region.

Human C8 is one of five components of the membrane attack complex of complement. It is an oligomeric protein composed of three subunits (C8 alpha, C8 beta, and C8 gamma) that are derived from different genes. C8 alpha and C8 beta are homologous and both contain a pair of tandemly arranged N-terminal modules [thrombospondin type 1 (TSP1) + low-density lipoprotein receptor class A (LDLRA)], an extended middle segment referred to as the membrane attack complex/perforin region (MACPF), and a pair of C-terminal modules [epidermal growth factor (EGF) + TSP1]. During biosynthetic processing, C8 alpha and C8 gamma associate to form a disulfide-linked dimer (C8 alpha-gamma) that binds to C8 beta through a site located on C8 alpha. In this study, the location of binding sites for C8 beta and C8 gamma and the importance of the modules in these interactions were investigated by use of chimeric and truncated forms of C8 alpha in which module pairs were either exchanged for those in C8 beta or deleted. Results show that exchange or deletion of one or both pairs of modules does not abrogate the ability of C8 alpha to form a disulfide-linked dimer when coexpressed with C8 gamma in COS cells. Furthermore, each chimeric and truncated form of C8 alpha-gamma retains the ability to bind C8 beta; however, only those containing the TSP1 + LDLRA modules from C8 alpha are hemolytically active. These results indicate that binding sites for C8 beta and C8 gamma reside within the MACPF region of C8 alpha and that interaction with either subunit is not dependent on the modules. They also suggest that the N-terminal modules in C8 alpha are important for C9 binding and/or expression of C8 activity.     

Plasma cell membrane glycoprotein PC-1. cDNA cloning of the human molecule, amino acid sequence, and chromosomal location

The murine cell membrane glycoprotein PC-1 is a homodimer with restricted tissue distribution, being first characterized in plasma cells. We now describe the isolation of cDNA clones encoding the human homolog of the murine PC-1 protein, its complete amino acid sequence, and its chromosomal location. Overall, the amino acid sequence of the human protein is about 80% identical to the murine protein, although the extent of homology varies in different domains. It had not been possible to assign a definitive amino terminus to the murine protein. Comparison of the murine and human sequence necessitates reassignment of the amino terminus, resulting in a cytoplasmic tail of 24 amino acids rather than 58 amino acids as previously published for the mouse. The sequence of several independently obtained cDNA clones indicates that the 3' end of the mRNA is subject to alternative splicing. Southern blots suggest a single copy gene. In situ chromosomal hybridization localizes the gene for human PC-1 to chromosome 6q22-q23, a common site for deletions in human lymphoid neoplasia.     

The binding of human complement proteins C5, factor B, beta 1H and properdin to complement fragment C3b on zymosan.

The covalent binding of complement fragment C3b to zymosan by the action of the alternative-pathway C3 convertase and the reversible binding of several complement proteins (component C5, factor B, beta 1H and properdin) to C3b on zymosan have been investigated. When C3b is deposited on zymosan after activation by a surface-bound C3 convertase, the C3b molecules are deposited in foci around the C3 convertase site, with an average of 30 C3b molecules per site. The association constants of C5, factor B, beta 1H, and properdin for C3b bound to zymosan have been determined. The association constants ranged from 6.5 x 10(-5) M-1 for factor B to 2.9 x 10(7) M-1 for properdin. An approximate stoichiometry of 1 : 1 for C5, factor B, and properdin binding to C3b has been observed. Curvilinear Scatchard plots were observed for beta 1H binding to C3b, with the maximal extrapolated ratio of beta 1H to C3b of 0.32. Physiological amounts of properdin increase by 7-fold the affinity constant for factor B binding to C3b with no alteration in the stoichiometry. Similarly, physiological amounts of factor B increase the affinity constant of properdin to C3b about 4-fold with only a small measured difference in stoichiometry. Competition binding studies and protein modification suggest that C5, factor B, beta 1H, and properdin each bind to a distinct region on C3b.     

Increased plasma concentrations of C9, C1-inhibitor and alpha 1-protease inhibitor associated with cigarette smoking

The association between various parameters of acute and chronic smoking status and plasma levels of three proteins, C9, C1-inhibitor (C1-INH) and alpha 1-protease inhibitor (alpha 1-PI) were determined for 49 male cigarette smokers and 49 age-matched nonsmokers (mean age = 32.2 years). The mean number of cigarettes smoked was 28.7 per day while the cumulative consumption was only 18.1 pack-years. Plasma levels of all three proteins were significantly higher in the smokers than nonsmokers. Plasma C9 and alpha 1-PI concentrations correlated with cumulative cigarette consumption and plasma nicotine concentrations. While C1-INH concentration did not correlate with either cumulative cigarette consumption or plasma nicotine concentration, it correlated significantly with serum thiocyanate concentration. No consistent correlation was found between plasma concentration of these proteins and parameters of pulmonary function.     

Integrin alpha 8 beta 1 promotes attachment, cell spreading, and neurite outgrowth on fibronectin.

The integrin alpha 8 subunit, isolated by low stringency hybridization, is a novel integrin subunit that associates with beta 1. To identify ligands, we have prepared a function-blocking antiserum to the extracellular domain of alpha 8, and we have established by transfection K562 cell lines that stably express alpha 8 beta 1 heterodimers on the cell surface. We demonstrate here by cell adhesion and neurite outgrowth assays that alpha 8 beta 1 is a fibronectin receptor. Studies on fibronectin fragments using RGD peptides as inhibitors show that alpha 8 beta 1 binds to the RGD site of fibronectin. In contrast to the endogenous alpha 5 beta 1 fibronectin receptor in K562 cells, alpha 8 beta 1 not only promotes cell attachment but also extensive cell spreading, suggesting functional differences between the two receptors. In chick embryo fibroblasts, alpha 8 beta 1 is localized to focal adhesions. We conclude that alpha 8 beta 1 is a receptor for fibronectin and can promote attachment, cell spreading, and neurite outgrowth on fibronectin.     

Chromosomal assignment of genes encoding the alpha, beta, and gamma subunits of human complement protein C8: identification of a close physical linkage between the alpha and the beta loci

The eighth component of human complement (C8) is a serum protein containing three nonidentical subunits (alpha, beta, gamma) that are arranged as a disulfide-linked alpha-gamma dimer and a noncovalently associated beta chain. In earlier genetic studies, electrophoretic analysis of C8 protein polymorphisms revealed several allelic variants of alpha-gamma and beta. These were governed by separate loci designated C8A and C8B for alpha-gamma and beta, respectively. Genetic linkage analyses indicated that these loci were linked to each other and to chromosome 1 marker loci PGM1 and Rh, but it was unclear at the time if C8A was a single locus coding for a single-chain precursor form of alpha-gamma or if separate loci existed for alpha and gamma. Since evidence now indicates that alpha, beta, and gamma are encoded by separate genes, cDNA probes corresponding to each subunit were used to make direct assignments of the individual loci. Analysis of somatic cell hybrids revealed that only the alpha and beta loci are located on chromosome 1. Parallel analysis of genomic DNA digests using 5' and 3'-specific cDNA probes showed they are physically linked (less than 2.5 kb) and oriented 5' alpha-beta 3'. Further probing of the hybrid panel revealed that gamma is located on chromosome 9q. Thus, the observed genetic linkage of alpha-gamma to beta must be determined solely by alpha. In accordance with these findings, the C8 loci should now be designated C8A, C8B, and C8G for alpha, beta and gamma, respectively.     

RanBPM, Muskelin, p48EMLP, p44CTLH, and the armadillo-repeat proteins ARMC8alpha and ARMC8beta are components of the CTLH complex

Ran-binding protein in microtubule organising centre (RanBPM) was originally isolated as a protein that binds to the small GTPase Ran. Recently our group and other groups reported that RanBPM was associated with several proteins and composed a large protein complex. Here, we used tandem MS with an antibody against RanBPM to purify this complex from a soluble extract of HEK293 cells: we identified Muskelin, p48EMLP, p44CTLH, and the novel armadillo-repeat proteins ARMC8alpha and ARMC8beta as components. In RanBPM, Muskelin, p48EMLP, and p44CTLH we found LisH/CTLH motifs, which are present in proteins involved in microtubule dynamics, cell migration, nucleokinesis, and chromosome segregation. We renamed the 20S large protein complex the CTLH complex. The N-terminal 364 amino acids of ARMC8alpha and ARMC8beta were completely conserved, suggesting that these proteins are probably alternatively spliced products from the same gene. We confirmed the in vivo association of each component by co-immunoprecipitation assays with Cos-7 cells in which these components were exogenously overexpressed. A pull-down assay with bacterially-expressed Twa1 revealed binding of each in vitro-translated component to Twa1. Finally, we confirmed the cellular localization of these proteins. Taken together, our results reveal that RanBPM, ARMC8alpha, ARMC8beta, Muskelin, p48EMLP, and p44CTLH form complexes in cells.     

Mapping the intermedilysin-human CD59 receptor interface reveals a deep correspondence with the binding site on CD59 for complement binding proteins C8alpha and C9

CD59 is a glycosylphosphatidylinositol-anchored protein that inhibits the assembly of the terminal complement membrane attack complex (MAC) pore, whereas Streptococcus intermedius intermedilysin (ILY), a pore forming cholesterol-dependent cytolysin (CDC), specifically binds to human CD59 (hCD59) to initiate the formation of its pore. The identification of the residues of ILY and hCD59 that form their binding interface revealed a remarkably deep correspondence between the hCD59 binding site for ILY and that for the MAC proteins C8α and C9. ILY disengages from hCD59 during the prepore to pore transition, suggesting that loss of this interaction is necessary to accommodate specific structural changes associated with this transition. Consistent with this scenario, mutants of hCD59 or ILY that increased the affinity of this interaction decreased the cytolytic activity by slowing the transition of the prepore to pore but not the assembly of the prepore oligomer. A signature motif was also identified in the hCD59 binding CDCs that revealed a new hCD59-binding member of the CDC family. Although the binding site on hCD59 for ILY, C8α, and C9 exhibits significant homology, no similarity exists in their binding sites for hCD59. Hence, ILY and the MAC proteins interact with common amino acids of hCD59 but lack detectable conservation in their binding sites for hCD59.     

Polarized expression of integrin receptors (alpha 6 beta 4, alpha 2 beta 1, alpha 3 beta 1, and alpha v beta 5) and their relationship with the cytoskeleton and basement membrane matrix in cultured human keratinocytes

In human keratinocytes cultured in conditions which allow differentiation and stratification and are suitable to reconstitute a fully functional epidermis, alpha 6 beta 4 and two members of the beta 1 integrin family (alpha 2 beta 1 and alpha 3 beta 1) were respectively polarized to the basal and lateral domains of the plasmamembrane both in growing colonies and in the reconstituted epidermis. Conversely, the alpha v integrin subunit, presumably in association with beta 5, was expressed at the basal surface in growing and migrating but not in stationary keratinocytes. The integrin alpha 6 beta 4: (a) was organized in typical patches which often showed a "leopard skin" pattern where spots corresponded to microfilament-free areas; (b) was not associated with focal contacts containing vinculin and talin but rather corresponded to relatively removed contact areas of the basal membrane as shown by interference reflection microscopy; and (c) was coherent to patches of laminin secreted and deposited underneath the ventral membrane of individual cells. The two beta 1 integrins (alpha 2 beta 1 and alpha 3 beta 1), both endowed with laminin receptor properties, were not associated with focal adhesions under experimental conditions allowing full epidermal maturation but matched the lateral position of vinculin (but not talin), cingulin, and desmoplakin, all makers of intercellular junctions. Often thin strips of laminin were observed in between the lateral aspects of individual basal keratinocytes. The integrin complex alpha v beta 5 had a topography similar to that of talin- and vinculin-containing focal adhesions mostly in the peripheral cells of expanding keratinocyte colonies and in coincidence with fibronectin strands. The discrete topography of beta 1 and beta 4 integrins has a functional role in the maintenance of the state of aggregation of cultured keratinocytes since lateral aggregation was impaired by antibodies to beta 1 whereas antibodies to beta 4 prevented cell-matrix adhesion (De Luca, M., R. N. Tamura, S. Kajiji, S. Bondanza, P. Rossino, R. Cancedda, P. C. Marchisio, and V. Quaranta. Proc. Natl. Acad. Sci. USA. 87:6888-6892). Moreover, the surface polarization of integrins followed attachment and depended both on the presence of Ca2+ in the medium and on the integrity of the cytoskeleton. We conclude that our in vitro functional tests and structural data suggest a correlation between the pattern of integrin expression on defined plasmamembrane domains and the mechanism of epidermal assembly.     

In vitro reaction of radioactive 7beta, 8alpha, --dihydroxy--9alpha, 10alpha-epoxy--7,8,9,10--tetrahydrobenzo (A) pyrene and 7beta,8alpha--dihydroxy--9beta, 10beta--epoxy--7,8,9,10--tetrahydrobenzo (A) pyrene with DNA.

The $\text{in vitro}$ reaction of bacteriophage T7-DNA with the radioactive diastereomeric benzo(a)pyrene-diol-epoxides, (±) [${}^3\text{H}{}_9$, ${}^3\text{H}{}_{10}$]-7β,8α-dihydroxy-9α,10β-epoxy-7,8,9,10-tetrahydrobenzo(a)pyrene, and (±) [${}^3\text{H}{}_9$, ${}^3\text{H}{}_{10}$]-7β,8α-dihydroxy-9β,19β-epoxy-7,8,9,10-tetrahydrobenzo(1)pyrene, was investigated. Chromatographic analysis of digests of the DNA allowed the distinction of characteristic deoxynucleoside adduct peaks for the two benzo(a)pyrene-diol-epoxides. Our results, together with data from the literature, allow the identification of these adducts as mostly N₂-(10-7β,8α,9α-trihydroxy-7,8,9,10-tetrahydrobenzo(a)pyreney1)deoxyguanosine and N₂-(10-7β,8α,9β-trihydroxy-7,8,9,10-tetrahydrobenzo(a)pyreney1)deoxyguanosine, respectively. DNA-benzo(a)pyrene adducts with the same chromatographic properties were formed in mouse embryo fibroblasts upon treatment with benzo(a)pyrene.     

Complementary DNA and derived amino acid sequence of the beta subunit of human complement protein C8: identification of a close structural and ancestral relationship to the alpha subunit and C9

A cDNA clone encoding the beta subunit (Mr 64,000) of the eighth component of complement (C8) has been isolated from a human liver cDNA library. This clone has a cDNA insert of 1.95 kilobases (kb) and contains the entire beta sequence [1608 base pairs (bp)]. Analysis of total cellular RNA isolated from the hepatoma cell line HepG2 revealed the mRNA for beta to be approximately 2.5 kb. This is similar to the message size for the alpha subunit of C8 and confirms the existence of different mRNAs for alpha and beta. This finding supports genetic evidence that alpha and beta are encoded at different loci. Analysis of the derived amino acid sequence revealed several membrane surface seeking segments that may facilitate beta interaction with target membranes during complement-mediated cytolysis. Determination of the carbohydrate composition indicated 1 or 2 asparagine-linked but no O-linked oligosaccharide chains. Comparison of the beta sequence to that reported for alpha in the preceding paper [Rao, A. G., Howard, O. M. Z., Ng, S. C., Whitehead, A. S., Colten, H. R. & Sodetz, J. M. (1987) Biochemistry (preceding paper in this issue)] and to that of human C9 revealed a striking homology between all three proteins. For beta and alpha, the overall homology is 33% on the basis of identity and 53% when conserved substitutions are allowed. For beta and C9, the values are 26% and 47%, respectively.(ABSTRACT TRUNCATED AT 250 WORDS)