The Chemoattractant Receptor-like Protein C5L2 Binds the C3a des-Arg77/Acylation-stimulating Protein*

The orphan receptor C5L2 has recently been described as a high affinity binding protein for complement fragments C5a and C3a that, unlike the previously described C5a receptor (CD88), couples only weakly to Gi-like G proteins (Cain, S. A., and Monk, P. N. (2002) J. Biol. Chem. 277, 7165–7169). Here we demonstrate that C5L2 binds the metabolites of C4a and C3a, C4a des-Arg77, and C3a des-Arg77 (also known as the acylation-stimulating protein or ASP) at a site distinct from the C5a binding site. The binding of these metabolites to C5L2 does not stimulate the degranulation of transfected rat basophilic leukemia cells either through endogenous rat G proteins or when co-transfected with human Gα16. C3a des-Arg77/ASP and C3a can potently stimulate triglyceride synthesis in human skin fibroblasts and 3T3-L1 preadipocytes. Here we show that both cell types and human adipose tissue express C5L2 mRNA and that the human fibroblasts express C5L2 protein at the cell surface. This is the first demonstration of the expression of C5L2 in cells that bind and respond to C3a des-Arg77/ASP and C3a. Thus C5L2, a promiscuous complement fragment-binding protein with a high affinity site that binds C3a des-Arg77/ASP, may mediate the acylation-stimulating properties of this peptide.

C5a and C3a have wide ranging effects in humans. Although initially described as leukocyte chemoattractants and anaphylatoxins, it is now clear that C5a and C3a are involved in microbial host defense, immune regulation (1), and protection against toxic insult (2)(3)(4)(5). C5a and C3a are also reported to have psychopharmacological effects on feeding and drinking behavior (6,7). Both complement fragments are rapidly desarginated by serum carboxypeptidase, which modulates their function. Although C5a des-Arg 74 retains most of the activity of intact C5a, albeit with a generally lower affinity for the C5a receptor (CD88), 1 C3a des-Arg 77 activity is profoundly reduced relative to C3a with respect to immunologic function. No binding of the C3a des-Arg 77 form to the previously cloned and characterized C3a receptor (C3aR) is observed in transfected RBL cells or mouse macrophage/monocytes (8) and, unlike C3a, C3a des-Arg 77 does not stimulate eosinophil chemotaxis (9), prostanoid production by guinea pig peritoneal macrophages and rat Kupffer cells (10), or human monocyte-like U937 cell degranulation (11). However, the following responses to C3a des-Arg 77 have been reported. (i) The cytotoxicity of NK cells is inhibited by both C3a and C3a des-Arg 77 (12). (ii) Cytokine production by human monocyte/macrophages and PBMC is enhanced by these ligands but inhibited in human tonsil-derived B cells (13,14). (iii) Histamine release from rat peritoneal mast cells is stimulated (15). In addition, C3a des-Arg 77 has well documented acylation-stimulating properties and increases triacylglycerol synthesis in human adipocytes, preadipocytes, and human skin fibroblasts (HSF), where this function as an acylation-stimulating protein (ASP) was initially characterized (16). This triglyceride-stimulating activity is also shared by C3a (17). One hypothesis explaining this pattern of responses is that cells may express two kinds of receptor; one, probably C3aR, binds only C3a, and another, as yet unidentified receptor, binds both C3a and C3a des-Arg 77 .
We have recently characterized a novel chemoattractantbinding protein, C5L2, that has high affinity for C5a, C5a des-Arg 74 , and C3a (18). Here we report that C5L2 also binds C3a des-Arg 77 /ASP and is expressed in three C3a des-Arg 77 / ASP-responsive cell types.
Stable Transfection of RBL and HEK 293 Cells-C5L2, C3aR and CD88-transfected RBL-2H3 and HEK 293 cells were produced as described (18). G ␣16 was cloned from human monocyte mRNA and authenticated by sequencing. Human G ␣16 and either C5L2 or CD88 were ligated into the bicistronic expression vector pIRES (Clontech). Stable transfection of RBL-2H3 cells with pIRES constructs was achieved by electroporation (20). Cells underwent three rounds of fluorescence-activated cell sorting (FACS) using anti-CD88 antibody (clone S5/1; Serotec) or anti-hemagglutinin peptide antibody (Roche Molecular Biochemicals, clone 12CA5) for C5L2-expressing cells, selecting the top 5% of receptor-positive cells in each round. HEK 293 cells were transfected (see below) and then sorted with two rounds of FACS using FLUOS-C3a des-Arg 77 /ASP binding, selecting the top 50% of the population of positive cells each time.
Transient Transfection of HEK 293 Cells-HEK 293 cells were seeded into 6-well plates at 1 ϫ 10 6 cells/well the day before transfection. C5L2 in vector pEE6hCMV.neo (Celltech) or C3aR in vector pcDNA1/AMP (Invitrogen) at 2 g of DNA/well was transfected with LipofectAMINE 2000 (5 l/well) (Invitrogen) according to the manufacturer's protocol. Cells were assayed for binding/uptake 3 days post-transfection.
Production of Anaphylatoxins-Expression and purification of the recombinant His 6 -tagged C5a, C5a des-Arg 74 , and C3a were performed under denaturing conditions as described (21). Recombinant C4a, C4a des-Arg 77 , and C3a des-Arg 77 were expressed and purified under nondenaturing conditions by sonication in the presence of BugBuster Protein Extraction Reagent (Novagen) using manufacturer's conditions. Plasma C3a des-Arg 77 /ASP and plasma C3a were purified as described previously (17).
Fluorescent Labeling of C3a des-Arg 77 /ASP and C3a-C3a des-Arg 77 / ASP and C3a were labeled with FLUOS (Roche Molecular Biochemicals) at a molar ratio of 1:10 (ligand to FLUOS) for 2 h according to the manufacturer's recommendations. Labeled ligand was separated from free FLUOS on a Sephadex G25 M column and stored in aliquots at Ϫ80°C.
Radiolabeled Ligand Competition Receptor Binding Assays-Competition binding assays were performed using 50 pM 125 I-C5a or 125 I-C3a (PerkinElmer Life Sciences) on adherent C3aR-, CD88-, or C5L2-transfected RBL cells in 96-well microtiter plates (55,000 cells/well) at 4°C as described previously (22). Competition assays for HSF, 3T3-L1, U937, and HEK 293 were performed using 1 nM 125 I-C3a or 125 I-C3a des-Arg 77 /ASP on adherent cells in 96-well microtiter plates. Competition curves were generated by preincubating adherent cells with increasing concentrations of unlabeled complement fragments. The IC 50 , standard error values and linear regression analyses were obtained by using GraphPad Prism 2.0 or Sigma Plot.
Production of Antiserum against C5L2-Antiserum was raised in rabbits using the extracellular N-terminal sequence of human C5L2 (MGNDSVSYEYGDYSDLSDRPVDC) coupled to keyhole limpet hemocyanin, as described previously (23). The serum recognized RBL cells transfected with human C5L2 (but not untransfected control cells) at dilutions as low as 1/10,000, and binding to C5L2 was totally inhibited by preincubation of serum with 100 g/ml immunizing peptide.
Fluorescence-activated Cell Scanning for Ligand Binding/Uptake Assays-Cells were incubated with the indicated concentrations of FLUOS-labeled C3a des-Arg 77 /ASP or C3a for 30 min at 37°C in binding buffer (24) and washed three times with cold binding buffer. Cells were then detached with 0.25% trypsin/0.02% EDTA in phosphatebuffered saline (PBS), fixed with 1% paraformaldehyde, washed with 0.3% PBS, and assayed by FACS. For anti-human C5L2 binding, cells were released from the culture dishes with non-enzymatic cell dissociation solution (Sigma), pelleted (600 ϫ g, 5 min), resuspended with anti-C5L2 antiserum (1:2000 in 3% bovine serum albumin in PBS), and incubated at 4°C for 60 min. Again, cells were pelleted, washed twice with PBS, and resuspended in fluorescein isothiocyanate-labeled antirabbit IgG, (Sigma) at (1:1000 dilution in 3% bovine serum albumin in PBS) and incubated at 4°C for 60 min. Finally, cells were pelleted, washed twice, and resuspended in 0.3% paraformaldehyde in PBS for FACS analysis.
Cellular Activation Assays-Cellular activation was measured as the release of ␤-hexosaminidase from RBL intracellular granules (25) or as the stimulation of triglyceride synthesis in HSF and 3T3-L1 cells (17). For ␤-hexosaminidase assays, EC 50 and standard error values were obtained by iterative curve fitting using GraphPad Prism 2.0. For triglyceride synthesis, cells were incubated with 100 M [ 3 H]oleate complexed to albumin (molar ratio 5:1) for 4 h. Triglyceride synthesis was calculated as [ 3 H]oleate incorporation into triglyceride.

RESULTS AND DISCUSSION
C5L2 Is a Promiscuous Complement Fragment-binding Protein-We have shown previously that C5L2 has binding sites for C5a, C5a des-Arg 74 , C4a, and C3a (18). Here we show that the des-Arg 77 forms of C4a and C3a are also ligands for this receptor when expressed in the RBL-2H3 cell line (Fig. 1, A and B, and Table I) and can compete strongly with 125 I-C3a for C5L2 binding (Fig. 1A). In contrast, C4a des-Arg 77 and C3a des-Arg 77 /ASP cannot compete effectively with 125 I-C5a for C5L2 or CD88 binding (Fig. 1B, and Table I). Although C3aR and C5L2 bind C3a with similar affinities, C3aR has no detectable affinity for C3a des-Arg 77 /ASP (Table I). Similarly, although C4a can compete with 125 I-C3a for binding to both C3aR and C5L2, suggesting a similar affinity for both receptors, C4a des-Arg 77 is Ͼ50-fold more effective at competing with 125 I-C3a binding at C5L2 than at C3aR (Table I). The data suggest either that C5L2 has two conformations with different ligand binding profiles or that the receptor has two binding sites. As we have shown previously that the B max values for 125 I-C3a and 125 I-C5a binding to C5L2-transfected RBL cells are identical (18), the most likely explanation is that a single form of C5L2 has separate binding sites. We propose that one site binds 125 I-C3a and C3a des-Arg 77 /ASP, at which all of the complement fragments except C5a des-Arg 74 can compete with similar affinities, and that the second high affinity site, which preferentially binds 125 I-C5a, can only be competed by C5a des-Arg 74 and, to a lesser extent, C4a.
C3a des-Arg 77 /ASP Binds Directly to C5L2 but Not to C3aR or CD88 -Because recombinant C3a des-Arg 77 /ASP can clearly compete with 125 I-C3a (but not C5a) for binding to C5L2, we then directly measured the affinity of C3a des-Arg 77 /ASP for C5L2 using protein purified from human plasma as C3a des-Arg 77 /ASP and tested for acylation-stimulating bioactivity. Plasma-purified human C3a des-Arg 77 /ASP and C3a were both labeled with FLUOS. Increasing concentrations of C3a des-Arg 77 /ASP were incubated with HEK 293 cells transiently transfected with C5L2, and binding and uptake were assessed by flow cytometry (Fig. 2A). FLUOS-C3a des-Arg 77 /ASP clearly binds to C5L2 with half-maximal fluorescence intensity at ϳ3 nM, whereas mock-transfected cells ( Fig. 2A, inset) show no binding of C3a des-Arg 77 /ASP, even at a high concentration of 10 nM. For comparison purposes, the binding of FLUOS-C3a to HEK 293 cells transiently transfected with C3aR is shown (Fig.  2B) with half-maximal binding of FLUOS-C3a at 2.5 nM. In   2C). C3a des-Arg 77 /ASP binding was further examined in cells that are responsive to the acylation-stimulating properties of C3a des-Arg 77 /ASP and compared with that in HEK cells transfected with C3aR and CD88. 125 I-C3a des-Arg 77 /ASP does not bind to C3aR-transfected HEK cells and does not compete with 125 I-C3a (Table II), as found previously (26). Similarly, Bt 2 -cAMP-differentiated U937 macrophages (which are reported to express the C3a receptor and respond to C3a) demonstrated no specific C3a des-Arg 77 /ASP binding (data not shown). The result was also negative for undifferentiated U937 cells (data not shown). Also, C3a des-Arg 77 /ASP does not bind to HEK 293 cells transfected with CD88 (binding of 125 I-C3a des-Arg 77 / ASP, mock transfection 100% Ϯ 4%, n ϭ 6; irrelevant receptor transfection, 102% Ϯ 11%, n ϭ 6; CD88 transfection, 110% Ϯ 22%, n ϭ 6). Similar results were obtained for 125 I-C3a binding to CD88 (irrelevant receptor transfection, 100% Ϯ 6%, n ϭ 6; CD88 transfection, 99% Ϯ 17, n ϭ 6). By contrast, human skin fibroblasts, which respond to C3a des-Arg 77 /ASP by increasing triglyceride synthesis (27), bind both 125 I-C3a des-Arg 77 /ASP and 125 I-C3a with high affinity (Table II). As observed in C5L2transfected RBL cells, unlabeled C3a des-Arg 77 /ASP is slightly less effective at competing for 125 I-C3a binding than unlabeled C3a in both HSF-and C5L2-transfected RBL cells (Tables II  and I, respectively), whereas C3a was an effective competitor for 125 I-C3a des-Arg 77 /ASP binding (Table II). Thus, C5L2 has binding characteristics that overlap with both CD88 and C3aR but also has the unique ability to bind C3a des-Arg 77 /ASP, which parallels the binding characteristics of HSF cells.
C3a des-Arg 77 Binding to C5L2 Does Not Stimulate Degranulation in C5L2-Transfected RBL Cells-We have shown previously that C5a, C5a des-Arg 74 , C4a, and C3a binding to C5L2 does not stimulate either an increase in intracellular Ca 2ϩ or the degranulation of transfected RBL cells due to weak coupling to endogenous G i -like G proteins (18). We also examined the effects of C3a des-Arg 77 /ASP and C4a des-Arg 77 and found that these ligands did not stimulate degranulation in transfected RBL cells at concentrations of up to 10 M (data not shown). In addition, there was no effect of these two ligands on either CD88 or C3aR activation of degranulation (Table III) although the expected responses to C5a, C5a des-Arg 74 , and C3a, respectively, are robust. Neither recombinant nor plasmapurified C3a des-Arg 77 /ASP (nor any other ligand) is able to activate endogenous G proteins in C5L2-transfected RBL cells.  Co-expression of C5L2 with G␣ 16 Does Not Enable a Degranulatory Response-The C5a receptor CD88 can couple effectively to the pertussis toxin (PT)-sensitive G proteins G i2 and G i3 (28) and also to the toxin-insensitive G q -family member, G 16 (29,30). We reasoned that the moderate response of ligand coupling to C5L2 could be due to the absence of human G 16 from RBL cells, which we tested by co-transfecting cells with human G␣ 16 and either CD88 or C5L2. The bicistronic vector pIRES was used to increase the likelihood that equal amounts of receptor and G protein would be expressed in transfected cells. With transfection of CD88 alone (Fig. 3A), increasing concentrations of PT inhibit the degranulation response. In co-transfected cells (Fig. 3B), CD88 clearly couples strongly to G␣ 16 , and the degranulation response to C5a is resistant to doses of PT that could substantially inhibit degranulation in cells transfected with CD88 alone. At a higher dose of PT (10 ng/ml), a small inhibition of degranulation is observed, presumably due to stabilization of interactions between free ␤␥ subunits and ADP-ribosylated G ␣i . In C5L2ϩG ␣16 co-transfected cells, treatment with high concentrations (1 M) of intact or des-Arg complement fragments still does not stimulate degranulation (Fig. 3C). It appears unlikely that C5L2 couples to G proteins usually associated with leukocyte chemoattractant receptors, although this does not eliminate the possibility of coupling to other signaling pathways.
C3a des-Arg 77 /ASP Stimulates Triglyceride Synthesis in Human Skin Fibroblasts but Not in Cells Expressing C3a Receptor-In HSF, both C3a des-Arg 77 /ASP and C3a can stimulate triglyceride synthesis (TGS) at levels comparable with insulin, a hormone well known to influence cellular triglyceride levels (Table IV). C3a des-Arg 77 /ASP appears to act via stimulation of the protein kinase C pathway (31), and stimulation of this pathway by the phorbol ester PMA also results in increased TGS (Table IV). Bioactivity of C3a is not dependent on conversion of C3a to the des-arginated form, C3a des-Arg 77 /ASP, because the presence of the carboxypeptidase inhibitor (Plummer's inhibitor) has no effect on C3a bioactivity (Table IV). Increased TGS is not simply a response to C3a binding, however, as C3aR-transfected HEK cells and Bt 2 -cAMP-differentiated U937 monocytic cells (which express the C3aR and bind C3a) do not respond with an increase in TGS to either C3a or C3a des-Arg 77 /ASP (Table IV). However, these cell types may lack all or part of the machinery to mount an increase in TGS, as there is no significant response to treatment with PMA or insulin (Table IV).
Both C3a and C3a des-Arg 77 /ASP bind to the C5L2 receptor expressed in RBL cells and HSF with comparable affinity, suggesting that C5L2 may be the C3a des-Arg 77 /ASP receptor on HSF. As C5L2 has already been shown to bind several complement fragments, we examined the acylation-stimulating properties of other C5L2 ligands in cells that respond to C3a des-Arg 77 /ASP. Even at higher concentrations than those usu-  also has no effect on triglyceride synthesis or binding of C3a des-Arg 77 /ASP. 2 These results suggest that the triglyceride synthesis stimulation is both peptide-and receptor-specific, with both C3a des-Arg 77 /ASP and C3a as the appropriate ligands interacting with the receptor C5L2. All ligands that stimulate C5L2, C3aR, or CD88 to increase TGS or degranulation also act as competitors for either 125 I-C3a or 125 I-C5a binding. The converse is not true; some C5L2 ligands (e.g. C5a) bind C5L2 but fail to activate the receptor (as assessed by TGS). C4a also binds to both C3aR and C5L2 receptors but activates neither, whereas C3a binds to and activates both receptors, but induces different responses (degranulation versus TGS). Activation requires binding to the appropriate receptor, but ligand binding per se does not necessarily cause activation. This may be explicable in terms of the physical separation of binding and activation sites on chemoattractant receptors such as CD88 (25,32). The two binding sites tentatively identified on C5L2 may also have different roles, one involved solely in ligand binding and one involved in both binding and activation of TGS. Thus, C5a, which binds to the first site on C5L2, may be able to sterically hinder the binding of ligands that interact primarily with the second site (C3a and C3a des-Arg 77 /ASP) without activation of receptor. The ability of C5a to influence binding to the second site is presumably dependent on the C-terminal Arg residue, as C5a des-Arg 74 cannot compete for 125 I-C3a binding to C5L2.
C5L2 mRNA and Cell Surface Protein Are Expressed in Adipose Tissue, Skin Fibroblasts, and 3T3-L1 Preadipocytes-Although C3a des-Arg 77 /ASP is regarded as biologically inactive in most myeloid systems, the acylation-stimulating properties of this complement fragment are well documented in adipocytes and related cells (33). We therefore investigated the expression of C5L2 in human adipose tissue, HSF, and 3T3-L1 preadipocytes, because fibroblasts, preadipocytes, and adipocytes are all known to respond directly to C3a and C3a des-Arg 77 /ASP by an increase in triglyceride synthesis (Table  IV) and glucose transport (17). We performed RT-PCR using species-specific sets of primers to detect expression in human adipocytes, HSF, and mouse 3T3-L1 preadipocyte mRNA. Both primer sets (human and murine) produced a band as seen on polyacrylamide electrophoresis gels at sizes similar to those expected for a C5L2 transcript (Fig. 4). As the DNA markers are standardized for agarose gels and not polyacrylamide gels, the human adipose tissue PCR product was extracted from an agarose gel and sequenced. We confirmed the authenticity of the transcript as that of C5L2. By contrast, RT-PCR of RNA from the human monocytic cell line U937 and non-transfected HEK 293 cells did not result in any PCR product using C5L2 primers despite equal levels of glyceraldehyde-3-phosphate dehydrogenase (Fig. 4).
These results were further confirmed using an antiserum specific to the N-terminal region of human C5L2. FACS analysis clearly demonstrates that HSFs (Fig. 5A) express endogenous C5L2 on their cell surface, although the fluorescent intensity was lower than that of HEK 293 cells overexpressing stably transfected C5L2 (Fig. 5B). In contrast, untransfected HEK 293 cells did not bind the anti-serum (Fig. 5C). As the antiserum does not appear to recognize murine C5L2, cells transfected with mouse C5L2 were negative (data not shown), and we were unable to test for the expression of C5L2 on the surface of the murine 3T3-L1 cells.
In summary, we have shown that adipocytes, HSF, and 3T3-L1 preadipocytes, cell types that have been shown to bind both C3a and C3a des-Arg 77 /ASP and to respond to these ligands with increased triglyceride synthesis, also express C5L2. C5L2 binds both ligands with high affinity, suggesting that it may be a functional C3a des-Arg 77 /ASP and C3a receptor when expressed in appropriate cell types. In contrast, C5a and C5a des-Arg 74 , which bind preferentially to a different site on C5L2, do not stimulate triglyceride synthesis. The role of C5L2 in cellular responses to complement fragments is clearly complex and remains to be elucidated. , and untransfected HEK 293 cells (C) were detached nonenzymatically and incubated at 4°C with either rabbit anti-C5L2 (continuous line) or rabbit non-immune serum (NI serum; broken line) as control. After washing, the cells were incubated with goat anti-rabbit IgG conjugated to fluorescein isothiocyanate. After washing and fixing with paraformaldehyde, cellular fluorescence was measured by FACS.