An Anti-inflammatory Function for the Complement Anaphylatoxin C5a-binding Protein, C5L2*

C5L2 is an enigmatic serpentine receptor that is co-expressed with the C5a receptor on many cells including polymorphonuclear neutrophils. The apparent absence of coupling of C5L2 with G proteins suggests that this receptor may modulate the biological activity of C5a, perhaps by acting as a decoy receptor. Alternatively, C5L2 may affect C5a function through formation of a heteromeric complex with the C5aR, or it may utilize a G protein-independent signaling pathway. Here we show that in mice bearing a targeted deletion of C5L2, the biological activity of C5a/C5adesArg is enhanced both in vivo and in vitro. The biological role of C5L2 thus appears to be limiting to the pro-inflammatory response to the anaphylatoxin. Accordingly, up-regulation of C5L2 may be of benefit in inflammatory states driven by C5a, including sepsis, asthma, cystic fibrosis, and chronic obstructive lung disease.

During the proteolytic cascade of complement activation, small (ϳ10 kDa) cationic fragments of C3, C4, and C5 are released, known as the anaphylatoxins C3a, C4a, and C5a. The anaphylatoxins participate in host defense through initiating chemotaxis and activation of myeloid cells, enhancement of vascular permeability, contraction of smooth muscle, and as yet poorly understood functions on endothelial and epithelial cells (1,2). Conversely, inappropriate activation of complement in vivo may be involved in a number of autoimmune diseases, asthma, rheumatoid arthritis, and cardiac disease (3).
Biological insights into the function of the anaphylatoxins and their receptors have been obtained through gene deletion in mice. Mice deficient in the C5a receptor show slower clearance and hence higher mortality in a model of Pseudomonas pneumonia (4). In contrast, these animals are completely protected from pulmonary injury in a model of immune complex injury in the lung (5). C3a receptor-deficient mice do not develop airway hyperresponsiveness compared with wild type animals upon allergen sensitization and challenge (6,7). These mice have also shown a role for C3a/C3a receptor interactions in the clearance of bacteria (8).
Given the potential for the anaphylatoxins to cause injury, mechanisms exist to limit the duration of their biologic activity. Serum carboxypeptidase N is also recognized as an anaphylatoxin inactivator (9). All three anaphylatoxins contain carboxyl-terminal arginine residues. In the case of C3a and C4a, the desarginine derivatives are virtually incapable of binding to the C3a receptor. In the case of C5a, the biologic activity decreases by more than an order of magnitude, and the binding affinity of the desarginine derivative changes from subnanomolar to micromolar. In whole serum or plasma, removal of the arginine by carboxypeptidase is close to instantaneous. Control of the residual activity of C5a desArg is presumed to be through clearance and degradation of the ligand following receptor internalization.
In 2000, Ohno et al. (10) described an orphan receptor with significant homology to the C5a receptor that is expressed on immature (but not mature) dendritic cells. Chromosome analysis reveals that C5L2 is adjacent to the C5a receptor gene on human chromosome 19. Cain and Monk (11) and Okinaga et al. (12) demonstrated that C5L2 is a high affinity binding protein for C5a and C5a desArg . Some evidence was provided that C5L2 may be weakly coupled to G proteins in stably transfected RBL cells and potentiates IgE-mediated degranulation (11). Kalant et al. (13) further demonstrated that C5L2 binds C3a, C3a desArg , and C4a and suggested that C5L2 mediates the action of acylationstimulating protein (14). We have consistently been unable to detect binding of ligands other than C5a and C5a desArg , using both transiently transfected 293T cells and stably transfected L1.2 mouse lymphoblasts (12). 2 We have also been unable to detect an interaction of native C5L2 with G proteins, although when leucine 132 is mutated to arginine, a weak calcium response is observed upon stimulation with C5a (12). In a recent study by Gavrilyuk et al. (15), noradrenaline (NA) 3 was shown to up-regulate C5L2 message and protein in rat astrocytes; this correlated with an anti-inflammatory response induced by NA. Additionally, antisense oligonucleotides against C5L2 reversed some of the anti-inflammatory properties of NA.
In the current report, we describe a murine line deficient in C5L2, which retains the classic C5a receptor. Using a well characterized model of C5a-dependent immune complex pulmonary injury, we show that C5L2-deficient mice exhibit an exaggerated inflammatory response. The data support a function for C5L2 in limiting the effects of C5a/C5a desArg .

MATERIALS AND METHODS
Targeting the C5L2 Gene through Homologous Recombination-A genomic clone of ϳ11 kb containing the C5L2 gene was cloned from a mouse SV129 DNA library. A 6-kb genomic fragment including the C5L2 coding sequence was used to construct the targeting vector. Approximately 1 kb of coding sequence was excised and replaced with genes encoding green fluorescent protein and neomycin resistance under control of the phosphoglucokinase promoter. The mutant construct was subcloned into pPNT, linearized, and electroporated into CJ7 embryonic stem cells derived from mouse strain 129. Transfectants were selected with G418 and gancyclovir, and the correctly targeted event was screened by Southern blot. Six targeted lines out of ϳ300 doubly resistant colonies were isolated. Three of these were injected into blastocysts derived from C57/BL6 females. Chimeric males were bred to C57/BL6 females to yield germ line transmission of the targeted allele. Animals were backcrossed to C57/BL6 through at least five generations prior to initiating the studies presented.
Preparation of Recombinant Murine C5a-A cDNA encoding murine C5a was prepared by reverse transcription-PCR from mouse liver based on previously published sequence data (GenBank TM accession number M35525; Ref. 16) and cloned into p3XFLAG-CMV-9 (Sigma). The plasmid was transfected into HEK293T cells using calcium phosphate (17). After 6 h, cells were transferred to serum-free media containing 5 mM sodium butyrate and protease inhibitor mixture (Sigma) and cultured for an additional 48 h. Culture supernatants were harvested and the FLAG-murine C5a was purified by affinity chromatography on M2 anti-FLAG resin as described by the supplier (Sigma).
In Vitro Chemotaxis-Bone marrow cells were isolated from the femurs and tibias of C5L2Ϫ/Ϫ mice and wild type litteramtes by perfusion with sterile PBS. Cells were fluorescently labeled with calcein-AM (Molecular Probes), suspended in 20 mM HEPES, pH 7.5, 125 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 0.5 mM glucose, 0.2% bovine serum albumin, at 1 ϫ 10 7 /ml, and tested for chemotactic activity in modified Boyden chambers. Cells (0.1 ml) were placed in the upper wells of 3-micron, 6.4-mm FluoroBlok filter inserts (Falcon) with 0.6 ml of buffer containing 0 to 10 mM C5a or 8 M fMLP in the lower wells. Chambers were incubated at 37°C for 45 min and chemotaxis determined by measuring the fluorescence intensity (excitation: 485 nm, emis-* This work was supported in part by National Institutes of Health Grants HL36162 (to N. P. G.) and HL69511 (to C. G.) and by the Perlmutter Foundation at Children's Hospital. 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. 1  sion: 535 nm) passing to the underside of the filter. Measurements were determined in duplicate for three independent experiments. In Vivo Chemotaxis-Recombinant mouse C5a, 5 g in 500 l of sterile PBS, was injected intraperitoneally into C5L2Ϫ/Ϫ, C5aRϪ/Ϫ, or wild type mice. Four hours later, animals were sacrificed and the peritoneal cavity lavaged with 10 ml of PBS and the cellular infitrate quantified by hemocytometer.
Immune Complex-mediated Lung Injury-All mouse protocols were approved by the Animal Care and Use Committee of Children's Hospital. Mice matched for sex and age (female, 10 -12 weeks) were used for this model as described previously (5). Mice were anesthetized with ketamine and xylazine. A single incision was made at the neck, and salivary glands were parted by blunt dissection to expose the trachea. A single 30-l injection containing rabbit antibody to chicken egg albumin (300 g) diluted in normal saline was injected using a syringe with a 25-gauge needle, followed by injection of chicken egg albumin (20 mg/kg) into the tail vein. Four hours later, the mice were sacrificed, and bronchoalveolar lavage (BAL) was performed. The BAL fluid was analyzed for protein concentration, cytokine measurement, total cell count, and differential. The lung was perfused via the right ventricle with 5 ml of PBS, and individual lobes were removed for histology or homogenized and assessed for myeloperoxidase and cytokines (enzyme-linked immunosorbent assay).  Statistical Analysis-Arithmetric means and standard deviations were calculated using Prism 4.0a (Graphpad) software. Data were analyzed using Student's t test for unpaired samples, and statistical significance was defined as p Ͻ 0.05.

RESULTS AND DISCUSSION
The murine C5L2 gene was targeted using the construct shown in Fig. 1A. As shown in Fig. 1B (upper panel), HindIII-digested DNA revealed the expected 5.7-kb restriction fragments for the targeted gene hybridizing with probe B, confirming homologous recombination. Three independent ES cell clones transmitted through the germ line and produced viable offspring in the expected Mendelian ratios. The mice are fecund, grow normally, and have a normal weight and life span.
Because the C5L2 locus is only ϳ15 kb away from the C5aR locus, it was important to demonstrate normal expression of the C5aR message and functional receptor protein. As shown in Fig. 1B (lower panel), BamHIdigested genomic DNA revealed identical restriction fragments hybridizing with the C5aR gene for both wild type and C5L2 deficient animals. Similarly, Northern blots demonstrated normal C5aR mRNA (Fig. 1C) in wild type and C5L2Ϫ/Ϫ mice but no C5L2 mRNA in C5L2-deficient animals. Furthermore, as indicated by the data shown in Fig. 2, the C5a receptor protein exhibits apparently normal function, as intraperitoneal instilla-tion of a single dose of recombinant murine C5a (5 g) into C5L2-deficient and wild type mice yielded indistinguishable neutrophilic infiltrates that were increased ϳ2-fold over C5aRϪ/Ϫ animals (p Ͻ 0.05, between wild type or C5L2Ϫ/Ϫ and C5aRϪ/Ϫ mice; n ϭ 3-5 animals per group). C5a receptor-deficient animals yielded cell numbers comparable with injection of PBS alone (Fig. 2).

FIGURE 3. Pulmonary immune complex (IC) injury in C5L2؊/؊ mice results in greater influx of inflammatory cells compared with wild type (WT) animals.
Four hours after tail vein injection of ovalbumin and intratracheal instillation of anti-ovalbumin IgG, mice were sacrificed, lungs were lavaged, and cells were quantified. A, C5L2Ϫ/Ϫ mice exhibited significantly greater total cells compared with wild type animals. **, p ϭ 0.008; n ϭ 5-7 mice per group. B, lavaged neutrophil levels were also significantly elevated in C5L2-deficient animals. ***, p ϭ 0.0005 for neutrophils; n ϭ 5-7 mice per group. No difference was observed in monocytes. Control animals received either antigen alone (Ag) or antibody alone (Ab). . C5L2-deficient mice release significantly more IL6 and TNF␣ than wild type animals following immune complex (IC) injury. Lung homogenates were assessed for IL6 and TNF␣ levels by enzyme-linked immunosorbent assay as described under "Materials and Methods." The cytokine levels assessed were each ϳ4-fold elevated in C5L2Ϫ/Ϫ mice compared with wild type animals. *, p ϭ 0.03; **, p ϭ 0.002; n ϭ 3 animals per group. Immune complex lung injury, an intrapulmonary version of a reverse passive Arthus reaction, has been demonstrated to be dependent on C5a acting on the C5a receptor (5,18). A recent report elegantly demonstrated the mechanism for this phenotype (19). Fc␥ receptor activation by immune complexes results in non-complement-mediated production of C5a by alveolar macrophages. C5a signaling through the classical C5aR and G␣ i results in a decrease in expression of the anti-inflammatory Fc␥RIIB without affecting the inflammatory Fc␥RIII. As a result, signaling through the latter receptor stimulates production of TNF␣ and CXCR2 ligands, which are central to the generation of hemorrhagic pneumonitis.
Since this phenotype is virtually "black and white" in comparing wild type and C5a receptor-deficient mice, we elected to ascertain the functional role of C5L2 in this model. Mice were given ovalbumin via tail vein injection, and anti-ovalbumin IgG was instilled into the trachea. After 4 h, animals were sacrificed and analyzed for inflammatory parameters. The bronchoalveolar lavage fluid was quantified for both total cells (Fig. 3A) and for neutrophils and monocytes (Fig. 3B). A dramatic increase in both total cells (8.3 Ϯ 1.1 ϫ 10 5 for C5L2Ϫ/Ϫ versus 3.2 Ϯ 1.0 ϫ 10 5 for wild type mice; p ϭ 0.008, n ϭ 5 per group) and neutrophils (2.70 Ϯ 0.56 ϫ 10 5 for C5L2Ϫ/Ϫ versus 1.07 Ϯ 0.56 ϫ 10 5 for wild type; p ϭ 0.0005, n ϭ 5 per group) was observed in C5L2-deficient mice compared with wild type animals. The cellular influx in the lungs of control mice of either strain treated with antibody alone or antigen alone was significantly reduced, as observed previously (5). No difference in influx of monocytes was apparrent. This phenotype for C5L2Ϫ/Ϫ mice is opposite the result obtained with mice deficient in the classical C5a receptor (5).
As indicated by the data of Fig. 4, the increased cellular influx correlated with an increase in the concentration of TNF␣ and IL6 in lung homogenates and in TNF␣ levels in BAL (data not shown). Both cytokines were increased ϳ4-fold in immune complex treated C5L2Ϫ/Ϫ mice compared with wild type animals (TNF␣: 166 Ϯ 15 pg/ml for C5L2Ϫ/Ϫ versus 38 Ϯ 7 pg/ml for wild type mice; p ϭ 0.002, n ϭ 3 per group; IL6: 1117 Ϯ 251 pg/ml for C5L2Ϫ/Ϫ versus 263 Ϯ 63 pg/ml for wild type mice; p ϭ 0.03, n ϭ 3 per group). These findings were confirmed by histologic examination of the lavage fluid and lung parenchyma as shown in Fig. 5.
C5L2-deficient mouse cells are also more responsive to C5a than wild type cells expressing both receptors in in vitro chemotaxis assays. Bone marrow cells isolated from C5L2Ϫ/Ϫ mice or wild type littermates were examined for their ability to migrate to 10 nM C5a relative to 8 M fMLP. As shown in Fig. 6 chemotaxis of C5L2Ϫ/Ϫ cells to C5a was ϳ40% greater relative to the response to 8 M fMLP than cells from wild type mice (C5L2Ϫ/Ϫ cells responded to 10 nM C5a with 7.9 Ϯ 0.4-fold increase in chemotactic cells over 8 M fMLP compared with 4.8 Ϯ 0.3-fold more cells for wild type; p ϭ 0.008, n ϭ 3 independent experiments performed in duplicate). This difference is substantially less than the 250 -300% increase in cellular influx in the lungs following immune complex injury (Fig. 3), likely reflecting absence of the amplification resulting from secondary release of additional chemotactic factors observed in this model. The observation of similar neutrophil influx into the peritoneal cavity of wild type versus C5L2Ϫ/Ϫ mice following injection of 5 g C5a (Fig. 2) is likely the result of inaccuracies inherent in this assay and reflective of the relatively small number of animals tested.
The overall expression level of mRNA for C5L2 is significantly lower than that for the C5a receptor and generally does not have a 1:1 correspondence among tissues (20). The median expression values for C5L2 are approximately one-third that of the C5aR. In some murine tissues, for example, pancreas and spleen, high levels of C5aR expression are associated with below median expression of C5L2, while in lactating mammary gland and skeletal muscle, the mRNA levels of for the two receptors are nearly equal. Clearly, the significance of these differences is subject to evaluation of the protein levels, but it suggests that C5L2 acts independently of the C5a receptor to counteract C5a/C5a receptor-mediated inflammation. Whether it does so by reducing the effective C5a levels available to act at the classical C5a receptor as suggested by Gao et al. (21), by triggering an anti-inflammatory signaling pathway, or both, awaits further study.