Purification of C1q receptors and functional analysis.

The recognition subunit of C1, C1q, has emerged as an important player in various pathophysiologic conditions largely in part due to its ability to interact with pathogen-associated or cell surface expressed ligands and receptors. Identification and purification of these molecules is therefore of paramount importance if we are to procure valuable information with regards to the structure, function, and cell surface distribution. Since the interaction of C1q is better served when the receptors are purified from homologous species, we discuss here a simple guideline for the purification and characterization of the two C1q receptors, cC1qR (calreticulin) and gC1qR, from human cell lines.

Neuroprotective effects of the SCR1-3 functional domain of CR1 on acute cerebral ischemia and reperfusion injury in rats.

OBJECTIVE: Complement receptor type 1 (CR1), one of the most potent inhibitors in complement activation, shows a protective effect on cerebral ischemia/reperfusion (CI/R) injury due to its ability to bind C3b and C4b and to inactivate C3/C5 convertases. So far, no study assessed the effect of the first three short consensus repeats (SCR1-3) with low molecular weight, one of the most active functional domains of CR1, binding C4b with a powerful decay-acceleration effect on classical and alternative C3/C5 convertases pathways. Therefore, we aim to assess this effect on CI/R injury in the present study. METHODS: Seventy-five adult male Sprague-Dawley rats were randomly divided into three groups: sham operation group (n = 15), CI/R group (n = 30), and CI/R group treated with CR1-SCR1-3 protein (n = 30). After middle cerebral artery occlusion (MCAO) for 1 hour and reperfusion for 24 hours, neurological motor deficits, cerebral infarct size, and biochemical parameters including myeloperoxidase (MPO), malondialdehyde (MDA), and superoxide dismutase (SOD) were assessed. Meanwhile, tissues in cerebral cortex were collected and processed for western blotting, immunohistochemistry, and HE staining. RESULTS: CR1-SCR1-3 could improve neurological functions in brain with a 26.8% decrease in neurological motor deficit score and could lead to a 63.8% reduction in cerebral infarct size. Besides, pretreatment using CR1-SCR1-3 could prevent neutrophil infiltration and alleviate inflammation severity and subsequent tissue damage. Decreased C4b expression and action, as well as improved morphological changes, were also observed in cerebral tissues of CI/R+CR1-SCR1-3 rats. CONCLUSION: CR1-SCR1-3 protein could possess a neuroprotective effect on acute CI/R injury.

Immobilization of the soluble domain of human complement receptor 1 on agarose-encapsulated islets for the prevention of complement activation.

The transplantation of islets of Langerhans has been successfully applied to the treatment of insulin-dependent diabetes. However, a shortage of human donors is the hardest obstacle to overcome. We aimed to develop a bioartificial pancreas that can realize xeno-islet transplantation. The islets were encapsulated in agarose microbeads carrying the soluble domain of human complement receptor 1 (sCR1), which is an effective inhibitor of the classical and alternative complement activation pathways. When naked rat islets were cultured in rabbit serum, large amounts of insulin leaked from the damaged islets over the course of a few days incubation, but no damaged cells were observed among islets in sCR1-agarose microbeads cultured in rabbit serum for 4 days. Although low levels of insulin were detected in the rabbit serum, the insulin did not leak from damaged ß-cells, it was physiological insulin secreted by the ß-cells.

Construction of the plasmid, expression by Chinese hamster ovary cell, purification and characterization of the first three short consensus repeat modules of human complement receptor type 1.

Short consensus repeat (SCR1-3), the first three SCR modules from N-terminus of type 1 complement receptor (CR1), is expected to accelerate dissociation of complement components and suppress complement activity by binding the main component of complement C4b. In order to clarify the three-dimensional structure, which triggers the activity of SCR1-3 on complement, we constructed an over-expression system in CHO DG44 cells which facilitated mass production of SCR1-3. The mass production was achieved by a two-stage culture system and optimum culture conditions using ASF104N medium and MTX-, NaBu-containing alpha-MEM/10% FBS medium, respectively. The constructed gene of SCR1-3 was confirmed by restriction enzyme digestion and DNA sequence analysis, and the expressed protein by CHO DG44 cells was confirmed by western blotting. The expressed SCR1-3 was proved containing N-linked sugar chain, an important factor to the proper expression of protein, by the cleavage with glycosidase of N-linked oligosaccharide (PNGase F). The suppression effect of the yield protein on complement-mediated inflammation was investigated by haemolytic assay and necrosis assay of stromal cells. Both assays showed that SCR1-3 possessed complement control activity. However, residing sugar chain on SCR1-3 did not show significant difference in the complement control activity.

Allorecognition in urochordates: identification of a highly variable complement receptor-like protein expressed in follicle cells of Ciona.

The evolutionary origin of allorecognition in vertebrates is unknown. Urochordates, being the closest living relatives of vertebrates [Delsuc F, Brinkmann H, Chourrout D, Philippe H]. Tunicates and not cephalochordates are the closest living relatives of vertebrates. Nature 2006; 439: 965-8], have efficient mechanisms to prevent both allogeneic fusion and self fertilization. To shed light on allorecognition in urochordates and on the molecules involved in preventing self fertilization, we compared gonadal cDNAs of three genetically unrelated Ciona intestinalis individuals by suppression subtractive hybridisation (SSH). Here, we report the discovery and characterization of a highly polymorphic gene coding for a transmembrane protein with several short consensus repeat domains (SCR/CCP). The protein, termed variable complement receptor-like 1 (vCRL1), is structurally similar to vertebrate complement receptors. However, in contrast to vertebrate complement receptors, vCRL1 shows an unprecedented high degree of amino acid variations among Ciona individuals and is expressed in follicle cells as well as in hemocytes. Based on our data we propose that in the absence of MHC Ciona uses variable components of the complement system as individuality markers.

First identification of a chemotactic receptor in an invertebrate species: structural and functional characterization of Ciona intestinalis C3a receptor.

In mammals, the bioactive fragment C3a, released from C3 during complement activation, is a potent mediator of inflammatory reactions and exerts its functional activity through the specific binding to cell surface G protein-coupled seven-transmembrane receptors. Recently, we demonstrated a Ciona intestinalis C3a (CiC3a)-mediated chemotaxis of hemocytes in the deuterostome invertebrate Ciona intestinalis and suggested an important role for this molecule in inflammatory processes. In the present work, we have cloned and characterized the receptor molecule involved in the CiC3a-mediated chemotaxis and studied its expression profile. The sequence, encoding a 95,394 Da seven-transmembrane domain protein, shows the highest sequence homology with mammalian C3aRs. Northern blot analysis revealed that the CiC3aR is expressed abundantly in the heart and neural complex and to a lesser extent in the ovaries, hemocytes, and larvae. Three polyclonal Abs raised in rabbits against peptides corresponding to CiC3aR regions of the first and second extracellular loop and of the third intracellular loop react specifically in Western blotting with a single band of 98-102 kDa in hemocyte protein extracts. Immunostaining performed on circulating hemocytes with the three specific Abs revealed that CiC3aR is constitutively expressed only in hyaline and granular amoebocytes. In chemotaxis experiments, the Abs against the first and second extracellular loop inhibited directional migration of hemocytes toward the synthetic peptide reproducing the CiC3a C-terminal sequence, thus providing the compelling evidence that C. intestinalis expresses a functional C3aR homologous to the mammalian receptor. These findings further elucidate the evolutionary origin of the vertebrate complement-mediated proinflammatory process.

Protruding disordered loop of gC1qR is specifically exposed and related to antiapoptotic property in germ cell lineage.

We established a monoclonal antibody (MAb), 5G9, with the use of a fixed seminoma tissue from an archival paraffin-embedded specimen, as an immunogen. Without antigen retrieval, positive 5G9-immunohistochemical staining was confined mostly to primordial germ cells, spermatogonia and various germ cell tumors. 5G9 recognized a mitochondrial 32-kD protein with an isoelectric point of pH 4.2, identified as a multifunctional ubiquitous protein, receptor for globular head of C1q (gC1qR), whose epitope was mapped in a disordered loop connecting the beta3 and the beta4 strands. Reflecting the ubiquitous distribution of gC1qR, with antigen retrieval, 5G9 was found reactive to a wide range of normal and tumor tissues. Since several co-precipitated and phosphorylated bands were observed in various human cell lines but not in germ cell tumor cell lines by in vitro phosphorylation assay, we speculate that the epitope of gC1qR is specifically unmasked in the germ cell lineage. By reducing gC1qR by siRNA, a significant increase was observed in the number of apoptotic cells in ITO-II and TCam-2 cell lines, but to a lesser extent in the Colo201 colon cancer cell line, showing an antiapoptotic property of gC1qR in the germ cells. Since protein-protein interaction is partially preserved by fixation, archival paraffin-embedded specimens can be a valuable source of immunogens for generating monoclonal antibodies (MAbs) that recognize tissue-specific protein conformation.

Characterization of a C3a receptor in rainbow trout and Xenopus: the first identification of C3a receptors in nonmammalian species.

Virtually nothing is known about the structure, function, and evolutionary origins of the C3aR in nonmammalian species. Because C3aR and C5aR are thought to have arisen from the same common ancestor, the recent characterization of a C5aR in teleost fish implied the presence of a C3aR in this animal group. In this study we report the cloning of a trout cDNA encoding a 364-aa molecule (TC3aR) that shows a high degree of sequence homology and a strong phylogenetic relationship with mammalian C3aRs. Northern blotting demonstrated that TC3aR was expressed primarily in blood leukocytes. Flow cytometric analysis and immunofluorescence microscopy showed that Abs raised against TC3aR stained to a high degree all blood B lymphocytes and, to a lesser extent, all granulocytes. More importantly, these Abs inhibited trout C3a-mediated intracellular calcium mobilization in trout leukocytes. A fascinating structural feature of TC3aR is the lack of a significant portion of the second extracellular loop (ECL2). In all C3aR molecules characterized to date, the ECL2 is exceptionally large when compared with the same region of C5aR. However, the exact function of the extra portion of ECL2 is unknown. The lack of this segment in TC3aR suggests that the extra piece of ECL2 was not necessary for the interaction of the ancestral C3aR with its ligand. Our findings represent the first C3aR characterized in nonmammalian species and support the hypothesis that if C3aR and C5aR diverged from a common ancestor, this event occurred before the emergence of teleost fish.

Structural requirements for the intracellular subunit polymerization of the complement inhibitor C4b-binding protein.

C4b-binding protein (C4BP), an important inhibitor of complement activation, has a unique spider-like shape. It is composed of six to seven identical alpha-chains with or without a single beta-chain, the chains being linked by disulfide bridges in their C-terminal parts. To elucidate the structural requirements for the assembly of the alpha-chains, recombinant C4BP was expressed in HEK 293 cells. The expressed C4BP was found to contain six disulfide-linked alpha-chains. Pulse-chase analysis demonstrated that the recombinant C4BP was rapidly synthesized in the cells and the polymerized C4BP appeared in the medium after 40 min. The alpha-chains were polymerized in the endoplasmic reticulum (ER) already after 5 min chase. The polymerization process was unaffected by blockage of the transport from the ER to the Golgi mediated by brefeldin A or low temperature (10 degrees C). The C-terminal part of the alpha-chain (57 amino acids), containing 2 cysteine residues and an amphiphatic alpha-helix region, was required for the polymerization. We constructed and expressed several mutants of C4BP that lacked the cysteine residues and/or were truncated at various positions in the C-terminal region. Gel filtration analysis of these variants demonstrated the whole alpha-helix region to be required for the formation of stable polymers of C4BP, which were further stabilized by the formation of disulfide bonds.

Contribution of anaphylatoxin C5a to late airway responses after repeated exposure of antigen to allergic rats.

We attempted to elucidate the contribution of complement to allergic asthma. Rat sensitized to OVA received repeated intratracheal exposures to OVA for up to 3 consecutive days, and pulmonary resistance was then estimated for up to 6 h after the last exposure. Whereas the immediate airway response (IAR) in terms of R(L) tended to decrease in proportion to the number of OVA exposures, late airway response (LAR) became prominent only after three. Although premedication with two kinds of complement inhibitors, soluble complement receptor type 1 (sCR1) or nafamostat mesylate, resulted in inhibition of the IAR after either a single or a double exposure, the LAR was inhibited after the triple. Premedication with a C5a receptor antagonist (C5aRA) before every exposure to OVA also inhibited the LAR after three. Repeated OVA exposure resulted in eosinophil and neutrophil infiltration into the bronchial submucosa which was suppressed by premedication with sCR1 or C5aRA. Up-regulation of C5aR mRNA was shown in lungs after triple OVA exposure, but almost no up-regulation of C3aR. Pretreatment with sCR1 or C5aRA suppressed the up-regulation of C5aR expression as well as cytokine messages in the lungs. The suppression of LAR by pretreatment with sCR1 was reversed by intratracheal instillation of rat C5a desArg the action of which was inhibited by C5aRA. In contrast, rat C3a desArg or cytokine-induced neutrophil chemoattractant-1 induced cellular infiltration into the bronchial submucosa by costimulation with OVA, but these had no influence on the LAR. These differences might be explained by the fact that costimulation with OVA and C5a synergistically potentiated IAR, whereas that with OVA and either C3a or cytokine-induced neutrophil chemoattractant-1 did not. C5a generated by Ag-Ab complexes helps in the production of cytokines and contributes to the LAR after repeated exposure to Ag.

A novel interaction between type IV pili of Neisseria gonorrhoeae and the human complement regulator C4B-binding protein.

C4b-binding protein (C4BP) is an important plasma inhibitor of the classical pathway of complement activation. Several bacterial pathogens bind C4BP, which may contribute to their virulence. In the present report we demonstrate that isolated type IV pili from Neisseria gonorrhoeae bind human C4BP in a dose-dependent and saturable manner. C4BP consists of seven identical alpha-chains and one beta-chain linked together with disulfide bridges. We found that pili bind to the alpha-chain of C4BP, which is composed of eight homologous complement control protein (CCP) domains. From the results of an inhibition assay with C4b and a competition assay in which we tested mutants of C4BP lacking individual CCPs, we concluded that the binding area for pili is localized to CCP1 and CCP2 of the alpha-chain. The binding between pili and C4BP was abolished at 0.25 M NaCl, implying that it is based mostly on ionic interactions, similarly to what have been observed for C4b-C4BP binding. Furthermore, the N-terminal part of PilC, a structural component of pili, appeared to be responsible for binding of C4BP. Membrane cofactor protein, previously shown to be a receptor for pathogenic N. gonorrhoeae on the surface of epithelial cells, competed with C4BP for binding to pili only at high concentrations, suggesting that different parts of pili are involved in these two interactions. Accordingly, high concentrations of C4BP were required to inhibit binding of N. gonorrhoeae to Chang conjunctiva cells, and no inhibition of binding was observed with cervical epithelial cells.

Interaction of C1q and the collectins with the potential receptors calreticulin (cC1qR/collectin receptor) and megalin.

Several proteins have been identified as candidate cell-surface receptors for the complement protein C1q. Some of these also interact with the structurally-related collectin proteins. Previous descriptions of C1q-binding properties of cells, and information on the cellular distribution of candidate receptors suggest that there is more than one physiologically relevant receptor for C1q. Two such candidate receptors, cell-surface calreticulin (also referred to as cC1qR or collectin receptor) and megalin are discussed in this review.

The C1q-binding cell membrane proteins cC1q-R and gC1q-R are released from activated cells: subcellular distribution and immunochemical characterization.

Two types of widely coexpressed cell surface C1q-binding proteins (C1q-R): a 60-kDa calreticulin-homolog which binds to the collagen-like "stalk" of C1q and a 33-kDa protein with affinity for the globular "heads" of the molecule, have been described. In this report, we show that the two molecules are also secreted by Raji cells and peripheral blood lymphocytes and can be isolated in soluble form from serum-free culture supernatant by HPLC purification using a Mono-Q column. The two purified soluble proteins had immunochemical and physical characteristics similar to their membrane counterparts in that both bound to intact C1q and to their respective C1q ligands, cC1q and gC1q. In addition, N-terminal amino acid sequence analyses of the soluble cC1q-R and gC1q-R were found to be identical to the reported sequences of the respective membrane-isolated proteins. Ligand blot analyses using biotinylated membrane or soluble cC1q-R and gC1q-R showed that both bind to the denatured and nondenatured A-chain and moderately to the C-chain of C1q. Moreover, like their membrane counterparts, the soluble proteins were found to inhibit serum C1q hemolytic activity. Although cC1q-R was released when both peripheral blood lymphocytes and Raji cells were incubated in phosphate-buffered saline for 1 hr under tissue culture conditions, gC1q-R was releasable only from Raji cells, suggesting that perhaps activation or transformation leading to immortalization is required for gC1q-R release. Subcellular fractionation of Raji cells and analyses by enzyme-linked immunosorbent assay and Western blotting showed that the two molecules are present in the cytosolic fractions as well as on the membrane. The data suggest that soluble forms of both C1q-binding molecules are released from cells and that these molecules may play important roles in vivo as regulators of complement activation.

Distribution of receptors of collagen and globular domains of C1q in human lung fibroblasts.

Fibroblasts are the predominant cell type responsible for the synthesis of collagen and other matrix elements in normal and fibrotic lungs. We have previously reported that human lung fibroblasts are heterogeneous in C1q binding and that subpopulations differing in C1q binding can be isolated and subcultured. We have investigated the distribution of receptors for C1q-collagen domain (cC1q-R) and globular domain (gC1q-R) in adult human lung fibroblasts. Fibroblasts were isolated from cultures of adult human lung explants in medium containing fresh- or heated plasma-derived human sera and separated by FACS-cell sorting into populations binding to C1q with high- (HF) and low- (LF) fluorescence. The cC1q-R was obtained from fibroblast membrane preparations by affinity chromatography through an anti-cC1q-R antibody column and its distribution was determined by Western analysis. The presence of gC1q-R was determined by immunoblots using an anti-gC1q-R antibody raised against a synthetic peptide. The results showed that a 54 kD protein crossreacting with anti-cC1q-R antibody was produced by LF cells, but it was barely detectable in HF cultures. Immunostaining with anti-cC1q-R antibody revealed that most of the cells in LF cultures were positive while the HF cells were negative. A 38 kD protein recognized by anti-gC1q-R antibody was produced by lung fibroblasts; however, no differences were detected in its distribution between LF and HF cultures. SDS-polyacrylamide gel electrophoresis of membrane proteins binding to an affinity column of C1q-globular fragment showed that the HF cultures contain a approximately 51 kD protein, which was a minor component in LF membranes. These data show that cC1q-R is expressed predominantly by a population of human lung fibroblasts, while the 38 kD gC1q-R is produced by all cells. Another 51 kD protein appears to be produced by a separate population of fibroblasts which does not express cC1q-R. Our results indicate that two lung fibroblast subtypes may be distinguished based on production of the 54 kD putative cC1q-R and another 51 kD protein which binds to C1q-globular domain.

A selection system to study C5a-C5a-receptor interactions: phage display of a novel C5a anaphylatoxin, Fos-C5aAla27.

Binding and effector domains of the human anaphylatoxin C5a have been determined by either site directed mutagenesis or synthetic peptide studies. However, the lack of specific selection methods, which allow direct investigation of C5a-C5a-receptor interaction made these studies laborious. To overcome these limitations we have constructed a novel Fos-C5a expressed on the tip of a filamentous phage. To guarantee for a free C-terminus which is required for C5a activity C5a cDNA was cloned into the phagemid vector pJuFo. Helper phage infection of pJuFc-C5a transformed cells resulted in a mutant phage displaying Fos-C5a on its surface. However studies with Bt2cAMP differentiated U937 cells revealed that phage displayed Fos-C5a is functional inactive. Subsequently we replaced a nonconserved cysteine residue at position 27 by alanine and obtained Fos-C5aAla27. Both the purified and the phage displayed Fos-C5aAla27 proteins were functional active and induced enzyme release from differentiated U937 cells. In addition, purified Fos-C5aAla27 exhibited the same binding profile as compared to rhC5a. Fos-C5aAla27 displaying phages were mixed with phage harboring only the pJuFo plasmid at a ratio of 10(6). After four successive rounds of panning on differentiated U937 cells Fos-C5aAla27 phages were enriched to 100% as shown by C5a-specific ELISA. We expect this approach to prove helpful for studying C5a-C5a-receptor interactions. i.e. to screen C5a libraries for high affinity binders with agonistic or antagonistic properties directly on cells.

The binding protein for globular heads of complement C1q, gC1qR. Functional expression and characterization as a novel vitronectin binding factor.

A binding protein for the globular head domains of complement component C1q, designated gC1qR, recently described to be present on vascular and blood cells (Ghebrehiwet, B., Lim, B.-L., Peerschke, E. I. B., Willis, A. C., and Reid, K. B. M. (1994) J. Exp. Med. 179, 1809-1821 was expressed in recombinant form in bacteria to investigate its functional and structural properties. The recombinant gC1qR was found to be functional because tetramerization of the 24.3-kDa polypeptide occurred as described for the native protein, and the binding of the ligand C1q by recombinant gC1qR was indistinguishable from binding shown by gC1qR isolated from Raji cells. Recombinant gC1qR immobilized to microspheres was used to search for additional binding proteins unrelated to C1q. Surprisingly, it was found that vitronectin or complexes containing vitronectin were retained from plasma or serum, and subsequent analysis revealed the specific binding of the ternary vitronectin-thrombin-antithrombin complex to gC1qR. Because the thrombin-antithrombin complex was unable to interact with gC1qR, direct binding with vitronectin was investigated in a purified system. The heparin binding multimeric form of vitronectin but not the plasma form of vitronectin was found to bind specifically to gC1qR isolated from Raji cell membrane as well as to recombinant gC1qR. This interaction was saturable (KD approximately 20 nM) and inhibitable by glycosaminoglycans such as heparin but not by chondroitin sulfate. C1q and vitronectin did not compete with each other for binding to gC1qR, and both ligands seem to interact with different parts of the gC1qR because a truncated version of recombinant gC1qR lacking the N-terminal 22-amino acid portion hardly interacted with vitronectin but bound C1q as well as the intact gC1qR. These findings establish gC1qR as a novel vitronectin-binding protein that may participate in the clearance of vitronectin-containing complexes or opsonized particles or cooperate with vitronectin in the inhibition of complement-mediated cytolysis.

Identification of functional domains on gC1Q-R, a cell surface protein that binds to the globular "heads" of C1Q, using monoclonal antibodies and synthetic peptides.

A membrane protein (33 kDa) that binds to the globular "heads" of C1q (gC1q-R) has been recently described. The full length cDNA encoding gC1q-R has been cloned, expressed in E. coli and using the purified recombinant protein (rgC1q-R) as an immunogen, a panel of IgG monoclonal antibodies (MAb) has been produced by fusion of spleen cells from hyperimmunized BALB/c mice with NSO mouse myeloma partners. From this fusion, 60 anti-gC1q-R hybridomas were selected and evaluated for their ability to (1) discriminate between the mature form (MF) of gC1q-R (residues 74-282) and a truncated form (TF) lacking residues 74-95, which contains a major C1q binding site, (2) recognize two functionally defined synthetic peptides derived from the NH2-(XN18) and COOH-(XC15) terminus of gC1q-R, and (3) bind to microtiter well fixed intact Raji cells. Several clones were identified: MAbs 46.23 and 60.11 (IgG1 kappa), reacted strongly with ELISA plate-fixed intact Raji and K562 cells, MF, and the XN18 peptide, but had poor or no reactivity with TF; MAbs 74.5.2 > 25.15 (IgG1 kappa) recognized both MF and TF and are directed against epitopes in the XC15 peptide that contains a binding site for high-molecular-weight kininogen and Factor XII.

Expression cloning of the human C3a anaphylatoxin receptor (C3aR) from differentiated U-937 cells.

A cDNA clone encoding the human C3a anaphylatoxin receptor (C3aR) was isolated from a pcDNAI/Amp expression library prepared from U-937 cells which had been differentiated with dibutyryl cAMP to a macrophage-like phenotype. The cDNA clone contained an insert of 4.3 kbp and was able to confer to transfected human HEK-293 cells the capacity to bind specifically iodinated human C3a. Chinese hamster ovary cells co-transfected with this cDNA clone and a G-protein alpha subunit (G alpha-16) became functionally responsive to C3a and a C3a analog synthetic peptide, as measured by increased phosphoinositide hydrolysis. As inferred from the cDNA sequence, the clone encodes a 482-residue polypeptide with seven hydrophobic membrane-spanning helices and a high homology to the human C5a and formyl-Met-Leu-Phe receptors. Uniquely among the family of G-protein coupled receptors, the C3aR contains an exceptionally large second extracellular loop of approximately 175 residues. Northern hybridizations revealed an approximately 2.3-kb transcript as the major and an additional approximately 3.9 kb-transcript as a minor transcription product of the C3aR. The C3aR appears to be widely expressed in different lymphoid tissues, as shown by Northern hybridizations, providing evidence for a central role of the C3a anaphylatoxin in inflammatory processes.