Metabolism of Activated Complement Component C3 Is Mediated by the Low Density Lipoprotein Receptor-related Protein/α2-Macroglobulin Receptor*

Complement component 3 (C3) and α2-macroglobulin evolved from a common, evolutionarily old, ancestor gene. Low density lipoprotein-receptor-related protein/α2-macroglobulin receptor (LRP/α2MR), a member of the low density lipoprotein receptor family, is responsible for the clearance of α2-macroglobulin-protease complexes. In this study, we examined whether C3 has conserved affinity for LRP/α2MR. Ligand blot experiments with human 125I-C3 on endosomal proteins show binding to a 600-kDa protein, indistinguishable from LRP/α2MR by the following criteria: it is competed by receptor-associated protein (the 39-kDa receptor-associated protein that impairs binding of all ligands to LRP/α2MR) and by lactoferrin and Pseudomonas exotoxin, other well known ligands of the multifunctional receptor. Binding of C3 is sensitive to reduction of the receptor and is Ca2+-dependent. All these features are typical for cysteine-rich binding repeats of the low density lipoprotein receptor family. In LRP/α2MR, they are found in four cassettes (2, 8, 10, and 11 repeats). Ligand blotting to chicken LR8 demonstrates that a single 8-fold repeat is sufficient for binding. Confocal microscopy visualizes initial surface labeling of human fibroblasts incubated with fluorescent labeled C3, which changes after 5 min to an intracellular vesicular staining pattern that is abolished in the presence of receptor-associated protein. Cell uptake is abolished in mouse fibroblasts deficient in LRP/α2MR. Native plasma C3 is not internalized. We demonstrate that the capacity to internalize C3 is saturable and exhibits a K D value of 17 nm. After intravenous injection, rat hepatocytes accumulate C3 in sedimentable vesicles with a density typical for endosomes. In conclusion, our ligand blot and uptake studies demonstrate the competence of the LRP/α2MR to bind and endocytose C3 and provide evidence for an LRP/α2MR-mediated system participating in C3 metabolism.

LRP/␣ 2 MR 1 is a member of the low density lipoprotein (LDL) receptor family, which includes the LDL receptor, LRP/␣ 2 MR, megalin (gp330), the very low density lipoprotein receptor, apolipoprotein E receptor 2, and LR8B (1). A gene closely related to that of LRP/␣ 2 MR was identified in Caenorhabditis elegans, showing that LRP/␣ 2 MR is an evolutionarily old molecule (2). LRP/␣ 2 MR was discovered by its homology to structures of complement components (3), and it was suggested that it might function together with the LDL receptor as receptor for apoE-containing lipoproteins executing the mass transport of chylomicron remnants (4). Later, LRP/␣ 2 MR was shown to be identical with the ␣ 2 MR (5,6). In recent years, a plethora of new ligands were reported, and LRP/␣ 2 MR was classified as a multifunctional receptor (1,7). Members of the LDL receptor family all consist of the same basic structural components: (i) a class of cysteine-rich repeats of approximately 40 amino acids, which are also present in the terminal complement components and are therefore referred to as complement type repeats (this sequence is believed to be responsible for binding of ligands); (ii) a second class of cysteine-rich repeats like those present in the epidermal growth factor precursor; (iii) an epidermal growth factor precursor homologous domain containing YWTD motifs, (iv) a single membrane-spanning segment; and (v) a cytoplasmic domain that contains one or more NPXY motifs, which are responsible for coated pit-mediated endocytosis (8,9). The ligand binding repeats are found in various members of the LDL receptor family assembled in clusters harboring 2-11 repeats, thereby varying affinity for different ligands. The study of the avian equivalents of members of the LDL receptor family profoundly refined the knowledge about the physiologic function of this receptor class as apoE is not expressed in avians. Similar to mass transport of chylomicron remnants into liver, LR8 on chicken oocyte membranes was found to mediate mass transport of egg yolk components into developing oocytes. (1,10).
Human ␣ 2 M is a macromolecular protease scavenger consisting of four subunits with an M r of 180,000 each. It is an acute phase protein inhibiting a wide range of proteases by forming a complex with them. Proteases cleave the molecule at the "bait region," leading to the hydrolysis of the internal thioester followed by an extensive conformational transition of the ␣ 2 M molecule. The conformational change results in the entrapment of the protease. Simultaneously, recognition sites are exposed enabling the ␣ 2 M molecule to interact with its specific cell surface receptor, LRP/␣2MR (11). The resulting complex is then internalized by LRP/␣ 2 MR-mediated endocytosis. The transformation ("activation") of the ␣ 2 M molecule can also be induced by a direct nucleophilic attack in a process in which the thioester bond is hydrolyzed without initial alteration of the bait region. This form of ␣ 2 M is also recognized by LRP/␣ 2 MR (5, 6).
Complement component 3 (C3) is a key molecule of the complement system. The classical and the alternative pathway lead to the formation of a convertase that cleaves C3 to C3a (molecular mass, 9 kDa) and C3b (178 kDa), resulting in the hydrolysis of the thioester bond. The subsequent dramatic conformational change in the molecule is a pivotal step in the complement activation (12). The classical pathway is activated by antibody-antigen reactions. In the alternative pathway, the initial reaction and its conformational consequences are not as clearly defined as in the classical pathway. Initial complement deposition on surfaces is thought to arise from spontaneously activated C3 (C3*). The inherent instability of the internal thioester in the native C3 molecule spontaneously and continuously generates small amounts of fluid-phase and bound C3* ("tick over" activation). A number of proteins control activated C3* from triggering unnecessarily the amplification loop of the complement system. The spontaneous hydrolysis rate of the thioester bond in vitro, under physiologic conditions, is estimated to amount to approximately 1%/h. In adult humans, approximately 240 mg of C3* have to be removed continuously per day by a system with high capacity.
Similar to in vivo tick over C3, activated C3* can be produced in vitro by treatment of native C3 with methylamine or with chaotropic agents or by slowly freezing and thawing of the protein and is referred to as C3(N)* (N stands for nucleophils, such as methylamine, ammonia, or water (13)).
C3 and ␣ 2 M belong to a group of evolutionarily related thioester-containing proteins (14,15). In the horseshoe crab, a single protein called Limulus ␣-macroglobulin complementlike protein possesses functional properties of both sides of the superfamily, namely, protease inhibition and participation in a hemolytic system (16). Although with the evolutionary development, thioester proteins exhibited marked functional divergence, structural homologies persist. This led us to examine whether LRP/␣ 2 MR is capable of binding and mediating endocytosis of certain forms of C3*.
We found that LRP/␣ 2 MR bound C3* in ligand blots. Very likely, the cysteine-rich complement type repeats were involved. Other ligands of LRP/␣ 2 MR, such as lactoferrin, receptor-associated protein (RAP), and Pseudomonas exotoxin, competed with C3*. Confocal microscopy demonstrated that fluorescent-labeled C3* but not native C3 was taken up into cells with features typical for receptor-mediated endocytosis. In the presence of RAP, which abolishes all binding to LRP/␣ 2 MR, uptake into cells was abolished. We found that in the intact rat, some endogenous C3* was present in liver endosomes. Uptake into liver and recovery in endosomes occurred very effectively when C3* was injected as a bolus, intravenously. From these data, we propose that LRP/␣ 2 MR plays an active role in the clearance of activated C3*.

EXPERIMENTAL PROCEDURES
Materials-We obtained chemicals from Sigma and Na 125 I for protein iodination from NEN Life Science Products. Bovine lactoferrin was a generous gift from Morinaga Co.. Mouse fibroblasts deficient in LRP/ ␣ 2 MR expression were obtained from J. Herz (University of Texas Health Science Center, Dallas, TX) (17).
Preparation and Definition of C3* Species-C3 was obtained from Chemicon (Temecula, CA) or prepared following an adapted protocol of Janatova (18). Briefly, 250 ml of freshly donated blood was used in phosphate buffers with all conductivity tests done with a Lovibond-Checkit-Micro (0 -1990 S). Batch adsorption of C4 and C3 was performed on Q-Sepharose (Fast Flow, Sigma Q 1126). The fractions containing C3, eluted by a salt gradient (100 -200 mM NaCl), were pooled and prepared for a hydrophobic interaction chromatography using phe-nyl-agarose (Sigma P-8901) to remove C5 and factor H. C3 purity and content in the fractions was determined by SDS-PAGE and by slotblotting (Hoefer, Vienna, Austria) with immunodetection with a rabbit anti-human C3c-antibody (DAKO) and visualized with peroxidase-coupled goat anti-rabbit IgG antibody and ECL-detection (Pierce). Purified material migrated as a double band of approximately 180 kDa. The rabbit anti-human C3-antibody (DAKO) recognized both bands (estimated purity, Ͼ95%). No signal was detected with an anti-C4 antibody (DAKO).
The internal thioester in the ␣-chain was analyzed by using the DIC assay (18). C3 that reacted with nucleophils and incorporated these and is therefore in the process of the conformational rearrangement is designated C3(N)* according to Pangburn (19). When the conformational change is completed, the asterisk is dropped. This conformation, now designated C3(N), has also been referred to as C3* or C3b-like C3, because it is similar to the proteolytically generated C3b. For analysis, the samples were boiled in the presence of 3% SDS without reducing reagents at 95°C for 15 min and then reduced in electrophoresis sample buffer with 0.1 M dithiothreitol at 95°C for 5 min. After separation on SDS-polyacrylamide, gels were stained with Coomassie Brilliant Blue. Alternatively, they were blotted on nitrocellulose sheets and detected with the antibody against C3. Reduced native C3 without DIC treatment gave two bands with electrophoretic mobility values of 115 (uncleaved ␣-chain) and 75 kDa (␤-chain), whereas after DIC treatment, it showed an additional band at 40 kDa resulting from the cleavage of the reactive ␣-chain. We used C3 preparations in which the 40-kDa band was absent or minor; upon three cycles of freezing in liquid nitrogen and thawing, these preparations developed as described (19). To study the difference in cell uptake between native and activated C3, we used fresh human serum and serum that was freeze/thaw cycled three times to activate C3.
Antibodies and Subcellular Fractions-For the LRP/␣ 2 MR antibody, polyclonal rabbit anti-LRP/␣ 2 MR antiserum was obtained from rabbits immunized with LRP/␣ 2 MR from rat liver endosomal fractions purified with a glutathione S-transferase (GST)-RAP affinity column. Recombinant GST and rat 39-kDa fusion protein (designated GST-RAP) were produced following the procedure of Herz et al. (20). Subcellular fractions (oocyte membranes from chicken, plasma membranes, and endosomes from rat liver) were prepared essentially as described (10,21,22). For the labeling of C3, C3* was radiolabeled with Na 125 I, using Iodo-Beads (Pierce), to a specific activity of 1000 cpm/ng and was dialyzed extensively against TBS-EDTA (50 mM Tris, pH 7.4, 0.15 M NaCl, 0.1 mM EDTA) before use. For cell uptake studies, C3* was fluorescentlabeled with BODIPY-succinimidylester (Molecular Probes, Leiden, The Netherlands). A solution of 335 g of BODIPY in 10 l of Me 2 SO was coupled in 20-fold molar excess with C3* (1 mg/ml in 0.1 M sodium bicarbonate) according to the manufacturer's protocol. Uncoupled dye was removed by passing the solution over a PD-10 column (Amersham Pharmacia Biotech).
Electrophoresis and Blotting-Controls with preparation of C3 were performed on 4 -15% Phastgels with SDS-buffer strips according to the manufacturer's protocols (Amersham Pharmacia Biotech). Separation of endosomal fractions and LR8 preparations was done using 4 -12% SDS gradient gels in a Bio-Rad apparatus followed by transfer to nitrocellulose according to the manufacturer's protocols. Immunodetection was done essentially as described (23) using HRP-conjugated second antibodies (Bio-Rad) and the ECL reagent (Amersham Pharmacia Biotech) according to the specifications of the manufacturer. ECL signal was quantitated by a Bio-Rad Fluor-S MultiImager. For quantification of bound radioactive ligands, counting was performed in a Packard Instant Imager (Packard Canberra, Vienna, Austria).
Cell Culture and Microscopy-Fibroblasts were maintained in monolayer and used for experiments as described previously (22,24). Cells grown on coverslips (Nunc) were incubated with BODIPY-labeled C3* (2 g/ml) for the indicated times and observed in a Zeiss Axiovert 135 microscope with fluorescence equipment. Experiments were duplicated with unlabeled C3*, and localization was detected with anti-C3 antibody (DAKO) and visualized with a fluorescein isothiocyanate-labeled second antibody. Where indicated, cell nuclei were stained with bisbenzimid dye (Hoechst 33258). Intracellular localization was determined using a Carl Zeiss laser scanning microscope with HeNe laser excitation, and sections of 1 m were made sequentially in the z plane. For cell binding and uptake studies, C3 was biotinylated with EZ-Link TM Biotin Hydrazide (Pierce) according to the manufacturer's protocol. Unreacted material was removed by dialysis. Mouse fibroblasts devoid of LRP expression, and wild type mouse fibroblasts expressing LRP (17) were seeded into 24-well plates and grown in RPMI 1640 medium (Life Technologies, Inc.) for 24 h. To start binding and uptake, the culture medium was replaced by medium containing biotinylated C3*. After 3 h, equilibrium of uptake and elimination was reached, determined by receptor activity. Cells were washed three times and lysed, the content of C3 detected on slot blots using HRP-conjugated anti-biotin antibody (Vector) with ECL reagent (Amersham Pharmacia Biotech) and quantitated by a Bio-Rad Fluor-S MultiImager. Unspecific binding measured in LRP defective cells was subtracted to calculate binding parameters.
All results are one representative of experiments performed at least twice.

RESULTS
Binding Proteins for C3*-We reasoned from the inherent genetic relationship of C3 and ␣ 2 M that LRP/␣ 2 MR may bind C3. Thus, we performed ligand blotting experiments using liver subcellular fractions to screen for proteins binding C3* with high affinity. For ligand overlays, we used C3 treated to favor production of thioester cleaved forms (C3*) as tested by DIC-C3 electrophoresis (see under "Experimental Procedures"). With gradient PAGE, we separated plasma membrane and endosomal proteins from rat liver and ligand blotted with 125 I-C3*. In Fig. 1, we show that in plasma membrane, the dominant binding protein was an approximately 80-kDa protein. However, C3* binding was insensitive to addition of RAP, indicating that this protein was not a member of the LDL receptor family. In addition, a faint band at 600 kDa was visible. In endosomes, the 600-kDa protein was the prominent binding protein visualized with C3*, and binding to it was abolished when RAP was present in the incubation mixture. The electrophoretic migration and immunological response of this band was indistinguishable from that of LRP/␣ 2 MR (not shown). The RAP-insensitive binding protein at 80 kDa was barely visible. Having established that at least one of the binding proteins for C3* in liver might be LRP/␣ 2 MR, we tested whether this function is specific for LRP/␣ 2 MR or extends to other members of the LDL receptor family. We used chicken oocyte membrane extracts, which are a rich source of LR8, the chicken homologue of mammalian very low density lipoprotein receptor. We have previously shown that this receptor was able to bind some typical ligands of LRP/␣ 2 MR (25). Using chicken oocyte membrane extracts, LR8 (95 kDa) was detected as the sole binding protein for C3*. When RAP was added, binding disappeared in the same way as when EDTA was added.
Examination of Specificity of Binding-Specific features of the binding were tested by varying cofactors, inhibitors, and competitors specific for LRP/␣ 2 MR (Fig. 2). Equal amounts of endosomal protein were examined by ligand blotting, and the amount of C3 that bound to LRP/␣ 2 MR was quantitated by densitometric scanning of the ECL signal. When Ca 2ϩ was removed by addition of EDTA (10 mM) to the incubation buffer, binding was abolished. Again, complete inhibition of binding was achieved when RAP was added to the incubation mixture. Treatment of the endosomal proteins bound to the nitrocellulose with dithiothreitol before incubation with C3 reduced the binding signal to 10% of normal. This feature was extensively described to be due to the reduction of cysteine disulfide bridges that are crucial for maintaining the tertiary structure of the ligand binding domains of members of the LDL receptor family. In addition, Pseudomonas exotoxin, which is one of the ligands described for LRP (26), reduces the binding to 20% of normal. Lactoferrin, which is another example of an immediate immune response molecule that was also shown to be a ligand of LRP/␣ 2 MR (22,23,27,28), competed as efficient as Pseudomonas exotoxin.
Cell Uptake of C3*-We next determined C3* internalization mediated by LRP/␣ 2 MR on the cellular level. Human fibroblasts, which express internalization-competent LRP/␣ 2 MR, were grown in chamber slides. We incubated the cells with fluorescent-labeled C3* either 60 min at 4°C to demonstrate plasma membrane binding, or 30 min at 37°C for uptake into cells. In the plasma membrane labeling experiment, an even surface distribution of the labeled ligand was seen (Fig. 3B). Exposing cells for 30 min caused a distinct vesicular pattern to be exhibited (Fig. 3C). In confocal planar section series (Fig.  3A), the late stages of endocytosis were clearly located in the midsections of the cell, with the perinuclear compartments positive for C3* fluorescence. In the presence of RAP (Fig. 3D), intracellular perinuclear distribution of fluorescent labeled C3* was completely abolished. To further demonstrate the competence of LRP/␣ 2 MR we studied the uptake of C3* in the mouse fibroblasts deficient in LRP/␣ 2 MR. In contrast to wild type cells (Fig. 3, LRPϩ), the mutant fibroblasts do not internalize C3. This experiment proved on a genetic level that LRP/ ␣ 2 MR mediates uptake of C3*.
Analysis of Cell Uptake-Using the mouse fibroblasts, we determined the kinetic parameters of uptake of C3* by LRP/ ␣ 2 MR. Cells were incubated with biotin-labeled C3* at 37°C

FIG. 1. Ligand blots with 125 I-C3* of rat liver plasma membrane (PM), endosomal (E), and chicken oocyte membrane (OM) proteins.
Extracts were separated by SDS-PAGE on 4 -8% gels under nonreducing conditions and blotted onto nitrocellulose. Strips with rat liver plasma membranes (PM) and rat liver endosomes (E) were incubated with 125 I-labeled C3* (10 6 cpm/g, 2 g/ml) in the absence (RAP Ϫ) and in the presence of 32 g/ml GST-RAP (RAP ϩ). Radioactivity was visualized by Packard Instant Imager. C3* bound by proteins of chicken oocyte membrane extract (OM) was immunodetected by overlay with anti-C3 followed by anti-rabbit HRP-coupled antibody and visualized by the ECL procedure. The approximate positions of 100 kDa (RAP-insensitive binding and chicken LR8) and 600 kDa (RAP-sensitive binding to LRP/␣ 2 MR) are shown in the left lane.

FIG. 2. Binding of 125 I-C3* to endosomal proteins in the presence of inhibitors and competitors.
Rat liver endosomal fractions (100 g of protein per lane) were separated by 4 -12% gradient SDS-PAGE (nonreducing) and blotted onto nitrocellulose membranes. Individual nitrocellulose strips were subjected to ligand blot analysis with radiolabeled C3* alone (2 ϫ 10 6 cpm/ml) and with the indicated additions: EDTA, 10 mM EDTA during incubation and washing; DTT, preincubation with 100 mM dithiothreitol for 1 h at 37°C; PE, 1 mg/ml Pseudomonas exotoxin during incubation with C3*; RAP, 100 nM GST-RAP; LF, 3 M bovine lactoferrin. Autoradiographs were quantitated by a Packard Instant Imager. for 3 h, and the cell-associated C3 transferred to nitrocellulose. It was quantitated with an anti-biotin antibody that was labeled with HRP. A linear signal in the concentration range between 0.5 and 5 ng was accomplished with ECL signal detection in the Bio-Rad Fluor-S MultiImager. Unspecific binding was determined accordingly but using LRP-deficient fibroblasts and subtracted from the total binding measured in LRPϩ fibroblasts. In Fig. 4, it can be seen that specific uptake of C3* into LRPϩ fibroblasts is saturable. At saturation, approximately 4 ng of C3 per 10 5 cells were taken up within a 3-h incubation period (40,000 molecules/cell/h). In Fig. 4, inset, we show Scatchard analysis of these data, giving a K D of 17 nM (mean, 15 Ϯ 3).
Our initial hypothesis was that a conserved structural similarity between ␣ 2 M and C3 determines the ability of C3* to be recognized by the LRP/␣ 2 MR. To validate this suggestion, we added a 10-fold molar excess of methylamine-activated ␣ 2 M to some of the incubation mixtures and measured C3* uptake. This addition resulted in an inhibition of uptake of C3* of 75%. The same reduction was found with the addition of a 50-fold molar excess of GST-RAP (not shown).
Cell Uptake of Native and Activated C3-In plasma, C3 is an abundant protein occurring with a concentration of roughly 1-2 mg/ml. In the experiment described above, we determined that saturation of LRP/␣ 2 MR-mediated uptake occurs above 50 nM. We thus incubated LRPϩ fibroblasts with fresh serum diluted to this concentration of C3 for 45 min at 37°C and determined intracellular C3 by immunofluorescence techniques as described. In Fig. 5, it can be seen that incubation with native C3 does not result in detectable uptake. We then activated the serum by three cycles of freezing in liquid nitrogen and thawing at 40°C. In DIC assays, this treatment notably inhibited the appearance of the 40-kDa fragment. Uptake experiments with this activated form of C3* resulted in considerable intracellular localization. The addition of RAP again prohibited this uptake, strongly indicating that C3 has to be activated to allow uptake into cells mediated by LRP/␣ 2 MR.
Liver Uptake of Intravenously Injected C3*-These experi-ments were done to determine whether uptake of C3* in vivo followed the kinetics of liver uptake seen with other ligands of LRP/␣ 2 MR. Furthermore, we wanted to find out what molecular weight form of C3 accumulates in endosomes. In in vivo turnover experiments with 125 I-C3*, we found a considerable fraction of C3* to be removed from the plasma and taken up within 20 min into the liver (not shown). In Fig. 6, we show the results of liver uptake experiments with a bolus injection of 250 g of unlabeled C3* and a saline-injected rat (untreated). The intracellular distribution of C3* recovered in liver subcellular fractions (40,000 ϫ g, vesicles) was analyzed by zonal rotor density gradient centrifugation. In Fig. 6, top panel, the density gradient profile is given in g/ml. Immunodetection of C3 in the density gradient fractions after injection of saline (untreated) and a mass bolus of unlabeled C3* are shown in the bottom two panels. Clearly, two different localization patterns of C3* were seen in immunoblot analysis of the density gradient fractions with anti-C3 polyclonal antibody detecting rat and human C3. In animals injected with saline, we detected immunoreactive material at the top of the gradient (density, 1.05 g/ml), corresponding to endogenous rat C3. This material was not particlebound, indicating that it was likely native C3 of extracellular origin. Some C3-positive material was located in endosomes, which is in accordance with the hypothesis that endogenous tick over reaction in the rat produces a form of C3* that is taken up by endocytosis. In rats injected with C3*, immunopositive material accumulated markedly in the 1.12 g/ml density range. These subcellular fractions have been previously characterized to contain predominantly endosomes (21). We also demonstrated that these fractions exhibit the highest concentration of LRP/␣ 2 MR, whereas RAP is located preferentially in fractions with higher density (23). C3 in endosomal fractions was sedimentable by centrifugation, verifying that it was enclosed in vesicles (not shown). The molecular weight of the immunopositive material was indistinguishable from the injected material, indicating that a processing prior to uptake did not significantly alter migration in gel electrophoresis.

DISCUSSION
C3 and ␣ 2 M evolved from a common ancestor in which two specific functions, protease trapping and cell lysis, were combined, as described in the arthropod Limulus polyphemus (16). The thioester bonds in complement component C3 and the protease inhibitor ␣ 2 M have traditionally been thought of as fulfilling the dual roles of mediating covalent attachment and maintaining a thermodynamically unfavorable conformational state of the native (nonactivated) protein. The thus stabilized conformation keeps binding sites buried, to be uncovered for further reaction only after activation. The removal of the activated ancestral molecule that was not consumed by the complement cascade reactions was presumably performed by an evolutionarily old receptor. We speculated that LRP/␣ 2 MR, which is expressed in mammals, as well as C. elegans, might have preserved affinity not only for ␣ 2 M but for C3* as well. Whether this feature is archaic or was preserved because of physiologic benefits will be an interesting task to determine. Our experiments clearly demonstrate that C3* specifically binds to LRP/␣ 2 MR immobilized on nitrocellulose and is internalized by LRP/␣ 2 MR in cell culture. In serum, C3 is present in concentrations over 1 g/liter, whereas in ligand binding experiments, concentrations as low as 2 g/ml gave a positive signal. However, we never observed binding of C3* using total fresh serum in the ligand blot experiments. Thus, we believe that although transition of C3 from the native to the activated form continuously takes place, these forms are rapidly removed and consequently in a very low steady state concentration in serum.
The data are therefore consistent with the hypothesis that a C3b-like form (C3*), which was activated in an activation process similar to the tick over process, is recognized by LRP/␣ 2 MR and is rapidly removed from the circulation, much like activated ␣ 2 M.
Our experiments also demonstrate that type A-like cysteinerich repeats, which constitute the ligand-binding domain of the LDL receptor family, are responsible for the high affinity of LRP/␣ 2 MR to C3*. These ligand binding repeats are found in various members of the LDL receptor family assembled in 2-11-fold clusters, thereby varying binding parameters for different ligands. Interestingly, we could demonstrate that the described function does not require more than one cluster because LR8, a chicken homologue of the very low density lipoprotein receptor bearing eight repeats, was also able to bind C3*. ApoE, a ligand of LRP/␣ 2 MR in mammals, is not expressed in birds. C3, in contrast, is present and functional in these animal species as well (29). Thus, our findings might help to refine our view about evolutionary aspects about structure/function relationship of members of this receptor family.
Although a vast number of data exist, not all details of conformational changes in the activation of native C3 are known. To this end, it is reasonable to assume an activation process, taking place in vitro and similar in vivo, that produces a form of C3* that can be recognized by LRP/␣ 2 MR. We interpret the data from our in vivo uptake studies to mean that in the intact rat, such minor but continuously generated C3* in plasma is internalized into endosomes continuously. We therefore obtained a weak immunopositive signal in the endosomal fraction in the control experiments. With injection of a bolus of 250 g of C3*, the endogenous activated C3* was outnumbered, and uptake into liver endosomes occurred rapidly. The amount of C3* that accumulated in endosomes of rats injected with a bolus of C3* was at least 1 order of magnitude higher, and the molecule migrated to the same molecular weight position as the injected material. This indicates that no proteolytic processing of the injected C3* occurred prior to the uptake into liver and makes participation within the complement cascade unlikely. Although we do not have direct evidence from whole animal studies that RAP abolishes uptake of C3 into liver, this uptake is very likely dependent on LRP/␣ 2 MR, as we observed the same effect of RAP in fibroblasts and hepatocytes.
In view of a recently proposed function of LRP/␣ 2 MR (1), it might serve as a carrier over barriers, transporting C3 to compartments that are otherwise not accessible for plasma proteins.
FIG. 5. Binding and internalization of native and activated C3 into fibroblasts expressing LRP/␣ 2 MR . To the culture medium of mouse fibroblasts expressing LRP, freshly prepared serum (native C3) and serum treated to activate C3 (C3*) as described was added. The concentration of C3 was set to saturating quantity (100 nM). C3 was localized by immunofluorescence, and nuclei were visualized by Hoechst dye. To determine the effect of LRP/␣ 2 MR, activated C3 was also coincubated with a 50-fold molar excess of GST-RAP (C3* ϩ RAP).
FIG. 6. Liver subcellular distribution of intravenously injected C3*. A postmitochondrial vesicle fraction from rat liver homogenate was analyzed by centrifugation on a sucrose density gradient (density, 1.05-1.15 g/ml). Eight consecutive density fractions from untreated rats (injected with saline) and rats that were injected 250 g of C3* 20 min prior to removal of the livers were separated by 4 -12% SDS-PAGE under nonreducing conditions and blotted onto nitrocellulose. C3 in these fractions was detected by anti-C3 antibody and visualized by a second antibody (anti-IgG) coupled to HRP and development by an ECL system until saturation of the strongest signal was achieved.
Finally, an interesting, newly discovered function of C3 could involve LRP/␣ 2 MR in C3 metabolism. It was recently reported that chylomicrons stimulate C3 synthesis in adipocytes by a factor of 100. In this situation, C3 is cleaved to C3a-and C3b-like portions. The first is used in the processing of the acylation stimulation protein governing fatty acid removal from the plasma; the latter has no known function and is likely to be removed (30,31). Indeed LRP/␣ 2 MR is present on adipocytes, and it is tempting to infer its participation in this metabolic reaction.