The Orphan Receptor C5L2 Has High Affinity Binding Sites for Complement Fragments C5a and C5a des-Arg 74 *

The substantial variations in the responses of cells to the anaphylatoxin C5a and its desarginated form, C5adR 74 , suggest that more than one type of cell surface receptor for these ligands might exist. However, only a single receptor for C5a and C5adR 74 , CD88, has been characterized to date. Here we report that the orphan receptor C5L2/gpr77, which shares 35% amino acid iden-tity with CD88, binds C5a with high affinity but has a 10-fold higher affinity for C5adR 74 than CD88. C5L2 also has a moderate affinity for anaphylatoxin C3a, but cross-competition studies suggest that C3a binds to a distinct site from C5a. C4a was able to displace C3a, suggesting that C5L2, like the C3a receptor, may have a low binding affinity for this anaphylatoxin. Unlike CD88 and C3a receptor, C5L2 transfected into RBL-2H3 cells does not support degranulation or increases in intracellular [Ca 2 (cid:1) ] and is not rapidly internalized in response to ligand binding. However, ligation of C5L2 by anaphylatoxin did potentiate the degranulation response to cross-linkage of the high affinity IgE receptor by a pertussis toxin-sensitive mechanism. These results suggest that C5L2 is an anaphylatoxin-binding protein with unique ligand binding and signaling properties. Complement

Complement fragment C5a is a potent chemoattractant and anaphylatoxin that acts on all classes of leukocytes and on many other cell types including endothelial, smooth muscle, kidney, liver, and neuronal cells (1,2). In addition to its proinflammatory effects, C5a has recently been shown to protect cells against toxic insult and to stimulate proliferation in neurons and hepatocytes (3)(4)(5)(6), suggesting a wider role for C5a in homeostasis. C5a is rapidly desarginated by serum carboxypeptidase N to the less potent derivative C5adR 74 , 1 the first stage in deactivation of anaphylatoxin activity (7). The dR 74 form has a different spectrum of bioactivity to intact C5a; for instance, in human basophils, stimulation by intact C5a causes the release of lipid mediators (e.g. leukotriene C4) and cytokines (e.g. interleukin-4 and interleukin-13), whereas C5adR 74 stimulates only cytokine release (8). Antagonists can also discriminate between different cell types: a cyclic peptide is 30-fold more potent on human neutrophils than a linear peptide antagonist, but both peptides are equally potent on human umbilical artery macrophages (9). Wide variations in antagonist affinity have also been observed in different species, but the sequences of C5a receptor homologs in these species do not suggest an obvious mechanism for these variations (10). The molecular basis for the ability of different cell types to discriminate between agonists, antagonists, and intact C5a/ C5adR 74 has yet to be elucidated as only a single receptor for C5a (CD88), a member of the G protein-coupled receptor superfamily, has so far been cloned (11,12). CD88 is in a G protein-coupled receptor subfamily that contains the receptors for human C3a (C3aR), formyl peptide, and an orphan receptor, C5L2 (also known as gpr77) (13,14). C5L2 transcripts are widespread with expression demonstrated in spleen, testis, brain, heart, lung, liver, kidney, ovary, and colon and in granulocytes and dendritic cells but not monocyte-derived macrophages (13,14). Here we report that C5L2 has high affinity binding sites for both C5a and C5adR 74 , apparently with a distinct binding site for the related anaphylatoxin C3a. Unlike CD88, C5L2 couples poorly to G i -like G protein-mediated signaling pathways and does not undergo rapid receptor internalization in response to ligand binding.

EXPERIMENTAL PROCEDURES
Cell Lines and Culture Conditions-RBL-2H3 cells were routinely cultured in Dulbecco's modified Eagle's medium ϩ 10% (v/v) fetal calf serum, which was supplemented with 400 mg/liter G-418 for transfected cells, at 37°C in 5% CO 2 .
Cloning of C5L2 and C4a and Transfection of RBL Cells-The C5L2 cDNA was cloned from a human brain (whole) Marathon-Ready cDNA (CLONTECH) by PCR using the sense primer 5Ј-GCGCGCAAGCTTG-CCACCATGTACCCATACGACGTCCCAGACTACGCTGGGAACG-ATTCTGTCAGCTAC-3Ј and the antisense primer 5Ј-GGGCCCGAAT-TCCTACACCTCCATCTCCGAGAC-3Ј. The added HindIII and EcoRI restriction sites are shown respectively in italics, the Kozak sequence used is shown underlined, and the added human influenza hemagglutinin (HA) tag on the sense primer is shown in bold. After authentication by sequencing, the full-length PCR product was digested with EcoRI and HindIII (Roche Molecular Biochemicals) and ligated into the expression vector pEE6hCMV.neo (Celltech). The C3aR cDNA, a generous gift of P. Gasque (Cardiff, United Kingdom), was inserted into PEE6hCMV.neo vector at the same site. Stable transfection of RBL-2H3 cells was achieved by electroporation as previously described (15). An anti-HA tag monoclonal antibody (clone 12CA5, Roche Molecular Biochemicals) or C3aR (clone P4B4, a generous gift from P. Gasque) was used to sort the highest 5% of transfected cells on a Becton-Dickinson Vantage flow cytometer in three rounds of cell sorting. C4a was cloned from the same human brain library as C5L2 using the sense primer 5Ј-CCGCCGGGATCCAACGTGAACTTCCAAAAGGCGA-3Ј and the antisense primer 5Ј-GCACCTGGTACCCTATTATCGTTGGAGGCCCG-CCT-3Ј; the added BamHI and KpnI restriction sites are shown respectively in italics.
Production of Anaphylatoxins-Expression and purification of the recombinant His 6 -tagged C5a, C5adR 74 , C3a, and C4a was performed under denaturing conditions as described previously (9). C4a was also expressed and purified under nondenaturing conditions by sonication in the presence of BugBuster Protein Extraction Reagent (Novagen) using * This research was funded by Arthritis Research Campaign Fellowship Grant M0543. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Cellular Activation Assays-Cellular activation was measured as the release of ␤-hexosaminidase from intracellular granules (16). The percentage of ␤-hexosaminidase release was calculated as a percentage of the maximal release (1 M C5a or C3a for CD88 and C3aR, respectively) or total cellular ␤-hexosaminidase (C5L2). EC 50 and standard error values were obtained by iterative curve fitting using GraphPad Prism 2.0. Alternatively activation was assayed as the increase in intracellular [Ca 2ϩ ] measured by flow cytometry of RBL cells labeled with the fluorescent indicator Fluo3AM (15).
Receptor Binding Assays-Competition binding assays were performed using 50 pM 125 I-C5a or 125 I-C3a (PerkinElmer Life Sciences) on adherent C5aR-transfected RBL cells in 96-well microtiter plates at 4°C as described previously (17). Binding curves were generated by incubating adherent cells in microtiter plates (at 55,000/well) with increasing concentrations of radiolabeled C5a and C3a in the presence or absence of 1 M unlabeled anaphylatoxin. The IC 50 , standard error values, and linear regression analyses were obtained by using Graph-Pad Prism 2.0.
Receptor Internalization Assays-These were performed as described previously (18). Transfected RBL cells were incubated with 100 nM ligand to stimulate receptor internalization, which was determined as the loss of surface receptor using specific monoclonal antibodies for CD88 (clone S5/1, Serotec), C3aR (clone P4B4), and HA tag at 10 g/ml.

RESULTS AND DISCUSSION
C5L2 and CD88 Have Very Similar Binding and Activationrelated Sequences-A sequence alignment between CD88, C3aR, and C5L2 is shown in Fig. 1. The N termini of CD88 and C5L2 contain several acidic residues that are characteristic of complement fragment receptors and that, for CD88 at least, form a part of the ligand binding site (19). Both CD88 and C3aR have been shown to have distinct ligand binding and activation sites (20). Receptor activation by ligand involves the engagement of charged and uncharged residues on the extracellular faces of the transmembrane helices. For CD88, these include Ile 116 , Val 286 , Arg 175 , Glu 199 , Arg 206 , and Asp 282 (21-23); C3aR has conserved residues at most analogous positions (Fig. 1). Interestingly C5L2 shares all of these residues except for Asp 282 , where there is a Glu residue (boxed residues in Fig. 1). Thus C5L2 has a similar acidic ligand-binding N-terminal domain to CD88 and a similar ligand activation domain.
C5L2 Binds Multiple Complement Fragments-C5L2 was cloned from a human brain cDNA library and expressed as a stable transfectant in the rat basophilic leukemia line RBL-2H3. Antibodies against an N-terminal HA peptide, inserted after the initiating methionine of C5L2, was used to select the top 5% of expressing cells by fluorescence-activated cell sorting in three rounds of cell sorting. Anti-HA antibody, but not anti-CD88 or anti-C3aR monoclonal antibody, recognized these cells (data not shown). Binding assays were performed using 125 I-C5a and 125 I-C3a to determine the ligand specificity. The specific binding curve indicates that specific, saturable binding of C5a occurs (Fig. 2a) with a receptor number calculated from the B max value of 39,736 Ϯ 5,993/cell, mean Ϯ S.E., n ϭ 3. C3a also binds specifically (Fig. 2b) with a similar number of binding sites (25,652 Ϯ 10,237/cell, mean Ϯ S.E., n ϭ 3; not significantly different from the C5a binding site number), but the calculated affinity for C5a was higher than that for C3a. Ligand specificity was investigated further using competition binding analysis, preincubating cells with a number of poten- tial ligands prior to the addition of 125 I-C5a or 125 I-C3a (Table  I). Using 125 I-C5a, the IC 50 for C5a was similar to that observed with CD88 in RBL cells, but C5L2 had a 10-fold lower IC 50 for C5adR 74 than CD88 (Table I). In contrast, C3a and C4a were very poor competitors for 125 I-C5a binding to both C5L2 and CD88 (Table I). A very different pattern was observed using 125 I-C3a: C5a, C4a, and C3a all had similar IC 50 values for 125 I-C3a binding to C5L2 (Table I), whereas C5adR 74 , a very effective competitor for 125 I-C5a binding, had no detectable ability to compete for 125 I-C3a binding (Table I). This pattern is clearly different to that observed for C3aR (Table I) where both C3a and C4a competed much more effectively with 125 I-C3a than C5a or C5adR 74 . These data demonstrate that C5L2 has a high affinity binding site for C5a and C5adR 74 and that C3a (and possibly also C4a) is a low affinity ligand for C5L2. It is likely that the binding sites for C5a and C3a on C5L2 are distinct because of the complex pattern of competition between ligands, in particular the complete failure of C5adR 74 to compete with 125 I-C3a. The location of the C3a binding site is, however, unclear as C5L2 does not have the very large second extracellular loop that appears to form the binding site for C3a on C3aR (24). The binding of C4a to C3aR has been previously observed (25), albeit with a larger difference between C3a and C4a affinities than observed here. The relatively high binding activity of the recombinant C4a used here may be due to the production process, which did not include the denaturation/ refolding step used for C5a and C3a. Denaturation of C4a in urea during purification appeared to destroy both the C5L2 and C3aR binding activity.
C5L2 Couples Weakly to Intracellular Signaling Pathways-The ligand binding data suggested the possibility that the high affinity of C5L2 for C5adR 74 might explain the sensitivity of some cell types to this ligand. We examined this by assessing the ability of C5L2 to activate transfected RBL cells using the degranulation response to ligand binding. The C5L2 ligands (C5a, C4a, C3a, and C5adR 74 ) were not able to support degranulation at concentrations of up to 3 M (Fig. 3a). In contrast, C5a and C5adR 74 (EC 50 ϭ 8 and 21 nM, respectively) could activate CD88-transfected RBL cells (Fig. 3b), and C3a (EC 50 ϭ 52 nM) but not C4a could activate C3aR-transfected RBL cells (Fig. 3c). C5L2 also did not increase degranulation even when RBL cells were primed with phorbol 12-myristate 13-acetate (100 nM), a treatment that enhances the response to suboptimal stimuli in RBL cells (26), for 10 min prior to the addition of ligand (data not shown) (17). The failure of C4a to activate C3aR has been previously reported (25). The signaling activity of C5L2 was also assessed as the increase in intracellular Ca 2ϩ using Fluo3AM-labeled RBL cells, an assay previously shown to be 10-fold more sensitive to ligand concentration than degranulation (15). Cells expressing C5L2 did not respond to 100 nM C5a, C5adR 74 , C3a, or C4a (Fig. 3d), whereas RBL cells expressing CD88 responded to C5a and C5adR 74 with robust increases in cellular fluorescence (Fig. 3e), and RBL cells expressing C3aR responded to C3a but not to C4a (Fig. 3f). Identical patterns of activity were observed using 1 M ligand (data not shown). The absence of intracellular Ca 2ϩ signaling is not due to low receptor number because the receptor expression level for C5L2 (ϳ40,000/cell) is actually higher than CD88 expression (ϳ36,000/cell (17)). We then examined whether ligand binding to C5L2 could prime RBL cells for a subsequent stimulus through the tyrosine kinase-coupled high affinity IgE receptor, Fc⑀RI. C5L2-transfected RBL cells were incubated with IgE DNP and activated by addition of 100 ng/ml HSA-DNP. Pretreatment for 10 min with 100 nM C5a, C3a, C4a, and C5adR 74 caused small but significant increases in the secretory response to HSA-DNP (Fig. 4). Untransfected cells did not show any increased response to HSA-DNP (Fig. 4), and the pretreatment of C5L2-transfected cells with pertussis toxin at a dose that could completely inhibit the degranulatory response to ligation of CD88 (10 ng/ml for 4 h (30)) also inhibited the effects of C5L2 ligands on the HSA-DNP response (Fig. 4). It appears that a low level of pertussis toxin-sensitive G proteindependent signal transduction can occur following ligand binding to C5L2. The relatively weak coupling of C5L2 to G protein (probably G i ) is not surprising because C5L2 does not have a sequence corresponding to the DRY motif found in most chemoattractant and chemokine receptors (Fig. 1); CD88 has 132 DRF, C3aR has DRC, but C5L2 has DLC (Fig. 1). The arginine residue of this motif in particular has been shown to be important in coupling to G proteins; mutation of the analogous residue in formyl peptide receptor inhibits signaling because of uncoupling from G protein (27). In addition, the third intracellular loop of C5L2 is much shorter than that of CD88 and C3aR (Fig. 1) and lacks Ser/Thr residues that may be protein kinase C phosphorylation sites as well as a conserved basic region ( 239 KTLK in CD88). Mutation of these Ser/Thr residues to Ala in CD88 inhibits signaling but not ligand binding (18), suggesting that this loop plays an essential role in G protein coupling. RBL cells are regarded as an excellent model system for the expression of granulocyte chemoattractant receptors (28) with similar G proteins and other receptor-associ- ated molecules. Platelet-activating factor receptor transfected into RBL cells (29) couples primarily to pertussis toxin-insensitive G proteins, demonstrating that different types of G proteins are available for stimulus-secretion coupling in the RBL cell line. We conclude therefore that C5L2, despite being expressed at high levels on granulocytes (13), couples only poorly to the normal range of G proteins for a granulocyte chemoattractant receptor. The possibility that C5L2 has additional signaling functions that do not require G i protein activation cannot be excluded. This might occur, for example, by the receptor phosphorylation-dependent mechanism utilized by C3aR for chemokine production (30).
C5L2 Does Not Undergo Ligand-dependent Internalization-CD88 and C3aR both undergo ligand-dependent internalization (25) when expressed in RBL cells, and so the ability of ligands (100 nM C5a, C5adR 74 , C4a, and C3a) to stimulate C5L2 internalization was investigated using anti-HA antibody to measure surface expression. None of the C5L2 ligands induced internalization after a 10-min incubation at 37°C, although 55% of surface CD88 and 75% of surface C3aR internalized after 10 min of treatment by 100 nM C5a/C5adR 74 or C3a, respectively (Fig. 5). The broad expression pattern and ligand preference of C5L2 initially suggested the possibility that C5L2 may act as a "sink" for excess anaphylatoxin following activation of the complement cascade, analogous to the Duffy and D6 promiscuous chemokine-binding proteins that may function in the buffering and presentation of chemokines (31). As it is not rapidly internalized, C5L2 is unlikely to be involved in anaphylatoxin clearance but might act as a reservoir of cell surface-associated anaphylatoxin to aid chemotaxis or to buffer anaphylatoxin concentrations during an inflammatory response. Alternatively, although internalization and receptor desensitization are not directly correlated, the retention of ligated C5L2 at the cell surface may be involved in prolonging signaling beyond the rapid responses normally stimulated by C5a.
In conclusion, we have shown that C5L2 has high affinity binding sites for C5a and C5adR 74 and also binds C3a and C4a with a similar affinity to C3aR. However, C5L2 couples poorly to the intracellular signaling and internalization machinery used by other chemoattractant receptors. The functions of this novel anaphylatoxin-binding protein remain to be defined.  4. Potentiation by C5L2 ligands of the degranulation response to cross-linkage of the high affinity IgE receptor. RBL cells transfected with C5L2 or untransfected control cells were incubated overnight with 1 g/ml IgE DNP and then treated with buffer or 100 nM C5a, C5adR 74 , C4a, or C3a for 15 min prior to the addition of the cross-linking agent HSA-DNP at 100 ng/ml. Degranulation was assessed as the secretion of ␤-hexosaminidase. In some cases cells were pretreated with 10 ng/ml pertussis toxin (PT) for 4 h prior to the addition of C5L2 ligands. Results are shown as a percentage of the release stimulated by 100 ng/ml HSA-DNP in the absence of anaphylatoxin (control ϭ 100) and are means of three to six separate experiments performed in triplicate Ϯ S.E. Significantly different from untransfected RBL cell response: ns, p Ͼ 5%; *, p Ͻ 5%; **, p Ͻ 0.5% (t test).

FIG. 5. Ligand-dependent internalization of chemoattractant receptors.
Transfected RBL cells were incubated with the stated ligands at 37°C for 10 min. After quenching in ice-cold buffer, surface expression of receptors was measured by adding antibodies specific for the N termini of CD88 and C3aR and the N-terminal HA tag of C5L2 and quantifying bound antibody levels by flow cytometry. The results are shown as a percentage of the untreated control cell expression and are the means Ϯ S.E. of three separate experiments performed in duplicate. Significantly different from control (ϭ100): *, p Ͻ 5%; ns, p Ͼ 5% (one sample t test). NA, no addition.