Development by Genetic Immunization of Monovalent Antibodies (Nanobodies) Behaving as Antagonists of the Human ChemR23 Receptor

The G protein–coupled receptor (GPCR) ChemR23 is structurally related to receptors for chemoattractant molecules, such as formyl peptides (FPR1, FPR2, and FPR3), complement fragments (C5a and C3a), and PG D2 receptors, as well as to GPR1 and the orphan receptors GPR32 and GPR33 (1). ChemR23 is expressed by various leukocyte populations, including monocytes, macrophages, myeloid and plasmacytoid dendritic cells, NK cells, and microglial cells (2–4). ChemR23 expression was also described in nonleukocyte cell populations, including preadipocytes and adipocytes (5, 6), osteoclasts (7), chondrocytes (8), skeletal muscle cells (9), and endothelial cells (10).

Chemerin, a 18-kDa protein, was first identified as the natural ligand of ChemR23 (11). The structure of chemerin is unrelated to that of chemokines or other chemoattractant factors for leukocytes (12). It is predicted to share a “cystatin fold” with a set of extracellular proteins, which includes type 2 cystatins, cathelicidin precursors, and kininogen. Chemerin is secreted by most tissues (with the notable exception of leukocytes) as an inactive precursor, prochemerin, which is 143 aa long in humans. This precursor is processed by proteolytic enzymes within its C terminus, generating bioactive chemerin in inflammatory conditions (13, 14). As expected from the ChemR23 expression profile, chemerin promotes chemotaxis of monocytes, macrophages, dendritic cells, and NK cells (3, 4). As a chemotactic factor generated in inflammatory conditions and acting on cells involved in innate immune responses, chemerin was initially expected to behave as a proinflammatory agent. However, recent data point toward more complex activities that are pro- or anti-inflammatory, according to the disease model investigated (8, 15, 16). In addition to its role as a chemoattractant agent, chemerin was reported to be an adipokine regulating adipogenesis and adipocyte metabolism (17). Besides chemerin, resolvin E1 (RvE1) was proposed as a ligand for ChemR23 (18). RvE1 is an oxygenated product of eicosapentanoic acid and is believed to mediate the beneficial effects of this essential fatty acid in a number of inflammation-associated human disorders. However, the RvE1/ChemR23 connection is highly controversial, and our laboratory has been unable to confirm it (15, 19).

Although ChemR23 is the only receptor unambiguously mediating the chemotactic activities of chemerin, two other GPCRs were described to bind chemerin with high affinity: GPR1 and CCRL2 (20, 21). GPR1 is not expressed in leukocyte populations but is found in the CNS, skeletal muscle, skin, and adipose tissue (20). In response to chemerin exposure, GPR1 is a poor activator of classical G protein–mediated pathways but is internalized efficiently. Therefore, GPR1 might behave as a decoy receptor for chemerin. CCRL2 belongs to the chemokine receptor family and is closely related to CCR1, CCR2, CCR3, and CCR5. In humans, CCRL2 is expressed in monocytes, macrophages, dendritic cells, neutrophils, T cells, NK cells, mast cells, and CD34⁺ bone marrow precursors (22). Chemerin binding to this receptor does not seem to promote any signaling or internalization response. Therefore, it was suggested that CCRL2 might increase local concentrations of chemerin and present the protein to other surrounding cells expressing ChemR23 (21, 23).

In addition to conventional Abs, Camelidae produce a special type of Abs devoid of L chains that are called “H chain–only Abs” (24). Ags are bound through the 15-kDa N-terminal variable domain (V_HH) of the H chain, which is frequently referred to as a nanobody. For several years, nanobodies have generated a growing interest as diagnostic and therapeutic tools because of their potential advantages over conventional Abs (25, 26). Nanobodies were reported to display low immunogenicity, high solubility, superior stability, and greater tissue penetration. Their smaller paratope gives them access to cavities on the surface of proteins, epitopes not usually recognized by conventional Abs or engineered Fab and scFv fragments. Finally, they are easier and cheaper to produce in microbial-expressing hosts.

In the current study, we used genetic vaccination and phage display to generate nanobodies recognizing human ChemR23. These nanobodies were characterized as antagonists of this receptor and did not recognize murine ChemR23 or related human GPCRs. They bind ChemR23 with an affinity in the range of 100 nM, and their binding site overlaps with that of chemerin-9, a peptide corresponding to the C terminus of bioactive chemerin. As a bivalent construct, one of the nanobodies inhibited the activity of such peptide, with an IC₅₀ as low as 1.7 nM. Following fluorescent labeling, they detected human ChemR23 in FACS and immunofluorescence staining. Thus, these nanobodies are innovative tools to study the distribution and role of ChemR23.

CHO-WTA11 cells, coexpressing apoaequorin and Gα₁₆, were cultured in Ham’s F12 medium supplemented with 10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 250 μg/ml Zeocin (Invitrogen, Groningen, The Netherlands) at 37°C with a constant supply of 5% CO₂ and split every 3 d. The cell lines expressing wild-type (WT) human or mouse ChemR23 were described previously (11, 27). To obtain the cell line expressing human ChemR23 fused at its C terminus with enhanced GFP (ChemR23-EGFP), the EGFP coding sequence was inserted in frame at the 3′ end of ChemR23 cDNA in a homemade bicistronic vector (pCDneo). The resulting plasmid (2 μg) was transfected into CHO-WTA11 cells using FuGENE transfection reagent (Roche), according to the manufacturer’s instructions. Selection of stably transfected cells was carried out in culture medium supplemented with 250 μg/ml G418 (Invitrogen). After 10–15 d of selection, individual colonies were transferred to 24-well plates, grown to confluence, trypsinized, and expanded further in 6-well plates. The identification of cell clones expressing the ChemR23-EGFP fusion was performed by measuring EGFP emission using FACS. A similar protocol was used to obtain the cell lines expressing human GPR1 or human CCRL2, cloned in pEFin3 (Euroscreen). The identification of cell clones expressing GPR1 or CCRL2 was performed by FACS using mouse mAb anti-GPR1 (kindly provided by Euroscreen; 1/20) or anti-CCRL2 (R&D Systems; 1/50).

Dubca cells (28) were cultured in RPMI 1640 medium supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin at 37°C with a constant supply of 5% CO₂ and split every 3 d. Dubca cells expressing human ChemR23 were obtained following electroporation of 20 μg the pCDneo-ChemR23 plasmid. After 10–15 d of selection in culture medium supplemented with 500 μg/ml G418 (Invitrogen), individual colonies were expanded as described above, and clones expressing human ChemR23 were selected by FACS using an anti-human ChemR23 mouse mAb (eBioscience; 1/50).

Human CD14⁺ monocytes were isolated from 1-d-old buffy coats derived from healthy donors (Blood Transfusion Center, Mechelen, Belgium), as described by Gouwy et al. (29). Human immature monocyte-derived dendritic cells (MDDCs) were generated by incubating purified peripheral blood CD14⁺ monocytes (1 × 10⁶ cells) in plastic culture dishes (55 cm²; International Medical Products) in 10 ml RPMI 1640 medium containing 10% FCS (Life Technologies Europe, Gent, Belgium), 50 ng/ml GM-CSF (Miltenyi Biotec, Bergisch Gladbach, Germany), and 20 ng/ml IL-4 (PeproTech, London, U.K.) at 37°C in humidified air with CO₂ (5%) for 6 d. Macrophages were differentiated from monocytes in the presence of recombinant human M-CSF (50 ng/ml; PeproTech) for 6–11 d. The purity of the cell preparations was evaluated by flow cytometry, and >85% of CD11b⁺/CD206⁺ macrophages were consistently obtained.

Camelid single-domain Ab fragments (V_HHs, nanobodies) specific for human ChemR23 were identified from Ig repertoires of genetically immunized llamas following a prime-boost regimen (30). The human ChemR23 coding sequence (GenBank accession number: Y14838.1 http://www.ncbi.nlm.nih.gov/genbank) was cloned into the mammalian expression vector pVAX1 (Invitrogen), and endotoxin-free plasmid DNA was prepared using the EndoFree Giga kit (QIAGEN), according to the manufacturer’s instructions. Four llamas (Llama glama) were immunized by intradermal administration of plasmid DNA (1–2 mg) at days 0, 17, 28, 42, and 74 using a DERMOJET device (AKRA DERMOJET, Pau, France). Twenty-three days after the last DNA immunization, 10⁷ Dubca cells overexpressing ChemR23 were injected s.c. ChemR23-specific serum conversion was monitored via flow cytometry during immunization (31). One week after the fourth DNA vaccination and 4 and 8 d after the cell boost, 100-ml blood samples were collected from each llama, and PBLs were prepared by Ficoll-Paque Plus density gradient centrifugation, according to the manufacturer’s instructions (LeucoSep tubes; Greiner Bio-One). Total RNA was extracted from the PBLs, as described by Chomczynski and Sacchi (32), and stored as ethanol precipitate at −80°C. First-strand cDNA synthesis was performed using hexanucleotide random priming and the SuperScript III First-Strand Synthesis System (Invitrogen), according to the manufacturer’s instructions. The nanobody repertoire was amplified and cloned in pXAP100, with minor modifications of the method described (33). pXAP100 is a pMESy4 derivative allowing expression of SfiI/BstEII cloned C-terminally His₆-cMyc–tagged nanobody. For the nested PCR, the 5′ primer was adapted to contain the SfiI restriction site.

Nanobody-displaying bacteriophage particles were produced according to standard protocols following helper phage superinfection (33). ChemR23-specific phages were enriched by a first round of biopanning using solid-phase immobilized virus-like particles (VLPs) harboring human ChemR23 (custom made by Icosagen, Ossu, Estonia). After thorough washing to remove nonspecific binders, phages were eluted via trypsin treatment and amplified by infection of logarithmically growing Escherichia coli TG1. In a second selection round, phage particles were incubated with CHO cells expressing ChemR23 or VLPs. Subsequently, individual clones were picked, and periplasmic extracts were prepared as described (33). The specificity of individual nanobodies was assessed by flow cytometry, staining ChemR23-overexpressing CHO cells and WT CHO cells. Unfixed, unpermeabilized cells were incubated with 5-fold diluted periplasmic extracts containing the nanobody in FACS buffer (PBS, 10% FBS, 0.1% sodium azide) for 60 min at 4°C. After washing in FACS buffer, the bound nanobodies were detected with a FITC-conjugated anti-His Tag Ab (AbD Serotec; 1/50), and the median cell fluorescence (MCF) intensity was determined. The data were processed using FACSDiva software (Becton Dickinson).

The two nanobodies, CA4910 and CA5183, were expressed in E. coli WK6Su⁻ and purified by metal-affinity chromatography. Briefly, bacteria were grown in Terrific Broth supplemented with 100 μg/ml ampicillin, 0.1% glucose, and 2 mM MgCl₂ to an OD₆₀₀ of 0.6–0.9. The expression of nanobodies was induced by adding 1 mM isopropyl β-D-1-thiogalactopyranoside, and bacteria were grown overnight at 30°C. The bacteria were centrifuged at 7000 × g for 10 min, and the pellets were resuspended in 15 ml TES buffer (0.2 M Tris-HCl [pH 8], 0.5 M sucrose, 0.5 M EDTA) and incubated for 1 h at 4°C under mild agitation. A total of 30 ml 4-fold diluted TES buffer was added, and the samples were incubated for 45 min at 4°C under mild agitation. The samples were centrifuged at 8000 × g for 30 min at 4°C, and the supernatants containing nanobodies were collected for purification on Ni-NTA resin (QIAGEN) (33). Binding of nanobodies on Ni-NTA resin was performed at room temperature for 1 h, the columns were washed with 50 mM phosphate buffer (pH 6), 1 M NaCl, and nanobodies were eluted with 1 M NaCl in 50 mM sodium acetate buffer (pH 4.5). The protein solution was neutralized by adding 1 M Tris-HCl buffer (pH 7.5). Nanobody purity was verified by SDS-PAGE.

CHO cells were detached from culture dishes using PBS containing 5 mM EDTA, harvested by centrifugation (560 × g, 4°C, 3 min), and resuspended at a density of 10⁶ cells/ml in cold FACS buffer (PBS, 0.1% BSA, 0.1% sodium azide). One hundred microliters of the cell suspension was incubated with anti-receptor Ab. The Abs used were mouse mAbs anti-human ChemR23 (eBioscience; 1/50), anti-CCRL2 (R&D Systems; 1/50), anti-GPR1 (Euroscreen; 1/20), or anti-mouse ChemR23 (eBioscience; 1/50) or nanobodies (300 nM). After 1 h of incubation at 4°C, cells were washed with 2 ml cold FACS buffer and incubated for 30 min at 4°C in the dark with an FITC-conjugated anti-mouse IgG (Sigma; 1/100) for mAbs or with an FITC-conjugated anti-His Tag Ab (AbD Serotec; 1/50) for nanobodies. The cells were washed and resuspended in FACS buffer, and the fluorescence level was analyzed using an LSRFortessa flow cytometer (Becton Dickinson). The data were processed using FACSDiva software (Becton Dickinson). Dead cells (forward scatter < 25) were excluded from the analysis, and nonspecific fluorescence was determined using untransfected CHO cells.

CHO cells were cultured on 22-mm glass coverslips for 48 h and washed in PBS. Nonspecific binding was prevented by a 30-min incubation at 4°C in PBS containing 10% FBS (FBS/PBS). The cells were then incubated for 1 h at 4°C with 1 μM nanobodies diluted in FBS/PBS. After two washes with FBS/PBS, the cells were incubated for 30 min at 4°C with a DyLight 549–conjugated anti-His Tag Ab (AbD Serotec) diluted (1/500) in FBS/PBS. After three rinses in FBS/PBS, cells were fixed for 15 min with 4% paraformaldehyde solution at room temperature and washed three times with PBS and once with water. Glass slides were mounted with ProLong DAPI mounting medium (Invitrogen) before viewing under a fluorescent microscope (Axioplan 2; Zeiss).

Purified CA4910 and CA5183 nanobodies were fluorescently labeled with DyLight 650 according to manufacturer’s instructions (Microscale Ab Labeling Kit; Pierce). Binding experiments were performed using DyLight 650-labeled nanobody as tracer. For saturation experiments, 100,000 WT or ChemR23-overexpressing CHO cells were incubated for 1 h at 4°C with increasing concentrations of labeled nanobodies (0.1–3000 nM) in 100 μl cold FACS buffer. After addition of 2 ml cold FACS buffer, the cells were harvested by centrifugation (560 × g, 4°C, 4 min) and resuspended in cold FACS buffer, and fluorescence was evaluated by FACS. K_d and B_max values were calculated by nonlinear regression (Prism software; GraphPad) using as total and nonspecific binding the MCF obtained with ChemR23-expressing and WT CHO cells, respectively. For competition-binding experiments, 100,000 ChemR23-expressing CHO cells were incubated for 1 h at 4°C in 100 μl cold FACS buffer containing 10 nM DyLight 650–labeled CA4910 or CA5183 and various concentrations of competitors [unlabeled nanobodies, chemerin (R&D Systems) or chemerin(149–157) nonapeptide]. Following a washing step with 2 ml cold FACS buffer, the fluorescence of cells was evaluated by FACS. IC₅₀ values were calculated by nonlinear regression using as nonspecific binding the MCF obtained in the presence of 3 μM unlabeled nanobody.

ChemR23-expressing CHO cells were detached from the plates by 5 mM EDTA in PBS and resuspended in binding buffer (50 mM HEPES [pH 7.4], 0.5% BSA, 5 mM MgCl₂, 1 mM CaCl₂). Binding was performed using 400,000 cells in a final volume of 100 μl and 0.2 nM radioiodinated [Y¹⁴⁵,F¹⁴⁹]-chemerin(145–157) (Phoenix Pharmaceuticals) as tracer. Total binding of the tracer was measured in the absence of competitor, and nonspecific radioligand binding was measured in the presence of 300 nM unlabeled peptide. For competition-binding assays, samples were incubated for 50 min at 27°C with increasing amounts of nanobodies. Bound and free radioligand were separated by filtration through GF/B filters presoaked for 16 h at 4°C in 1% BSA. Filters were washed three times with ice-cold binding buffer supplemented with 500 mM NaCl. IC₅₀ values were calculated using nonlinear regression (Prism software; GraphPad).

Calcium mobilization was measured in CHO cells expressing ChemR23 by an assay based on the luminescence of mitochondrial aequorin, as previously described (27). Briefly, cells were collected from plates with 5 mM EDTA in PBS, pelleted, resuspended at a density of 5.10⁶ cells/ml in DMEM/Ham’s F12 (Invitrogen) supplemented with 0.1% BSA, and incubated with 5 μM coelenterazine H (Promega) for 4 h at room temperature under gentle agitation in the dark. Cells were then diluted to a density of 10⁶ cells/ml and incubated for an additional hour. For the agonist assay, 50 μl cell suspension was added to microplate wells containing agonists diluted in a volume of 50 μl DMEM-F12. Calcium increase was evaluated by measuring, for 30 s, the luminescent signal (integration of area under the curve) resulting from the activation of the aequorin-coelenterazine complex using a Centro LB 960 luminometer (Berthold Technologies). The data were normalized for basal (0%, background removal) and maximal (100%) luminescence, corresponding to the signal measured following exposure to 20 μM ATP. For the antagonist assay, 25 μl cell suspension was added to 25 μl increasing concentrations of nanobodies and incubated for 45 min at room temperature and then 50 μl chemerin (1 nM) or chemerin(149–157) (5 nM) solution was added, and the luminescent signal (integration of area under the curve) was recorded for 30 s in a Centro LB 960 luminometer (Berthold Technologies). The data were normalized for basal (0%, background removal) and maximal (100%) luminescence, corresponding to the signal measured following exposure to 1 nM chemerin or 5 nM chemerin(149–157). Dose-response curves and IC₅₀ values were calculated using nonlinear regression (GraphPad Prism software).

The chemotactic potency of chemerin and CCL3 on immature MDDCs in the presence or absence of nanobodies was determined in the Boyden microchamber assay (Neuro Probe, Cabin John, MD). Suspended immature MDDCs were collected from the dishes by pipetting up and down. Then the dishes were incubated with PBS for 10 min at 4°C, and the adhered immature MDDCs were detached by scraping. The immature MDDCs were pelleted by gentle centrifugation and resuspended in chemotaxis buffer (i.e., HBSS [Invitrogen] supplemented with 1 mg/ml human serum albumin [Belgian Red Cross, Leuven, Belgium]). Chemerin or CCL3 (positive control) was added to the lower compartment of the microchamber and separated from the upper compartment by a 5-μm pore size polyvinylpyrrolidone-treated polycarbonate membrane (GE Water & Process Technologies, Manchester, U.K.). The wells in the upper compartment of the Boyden chamber were filled with 50 μl cell suspension (1 × 10⁶ cells/ml) in the presence or absence of nanobodies (500 nM). After incubation at 37°C for 1.5 h in humidified air with 5% CO₂, the filters were removed and stained, and the cells migrated across the filter were counted using a microscope (×500 magnification). The chemotactic response displayed in the figures was calculated as the number of migrated cells to chemerin (minus cells that migrated spontaneously to the chemotaxis buffer)/number of migrated cells to CCL3 (minus cells that migrated spontaneously to the chemotaxis buffer). Data are shown as mean ± SEM.

To generate nanobodies directed against the extracellular domain of human ChemR23, four llamas were immunized by genetic vaccination following a prime-boost strategy. Immune responses were primed with five shots of 1–2 mg endotoxin-free pVAX1 plasmid DNA encoding human ChemR23. For the single-cell boost, ChemR23-overexpressing camelid cells (Dubca) were chosen as cell background to bias the response toward ChemR23 and to avoid immunization against other cell surface markers. Mounting of the ChemR23-specific immune response was monitored by testing the sera of animals, using FACS, on CHO cells expressing the receptor. Only one of the llamas (named 2T) showed a robust ChemR23-specific seroconversion. After DNA vaccination and after the cell boost, PBLs were collected from this animal, RNA was prepared, and a phage library displaying the V_HH repertoire was constructed (33). A first round of panning was performed on ChemR23-expressing VLPs, and second rounds of panning were made on ChemR23-expressing CHO cells or VLPs, using WT CHO cells or irrelevant VLPs as negative controls. After two rounds of panning, clear enrichment of ChemR23-specific phages was detected. Twenty-four clones resulting from two rounds of panning on ChemR23-expressing CHO cells and 48 clones selected by one round of panning on ChemR23-expressing CHO cells followed by one round on VLPs were analyzed further. Accordingly, from each clone, periplasmic extract was prepared to partially purify the encoded nanobody. Binding of these nanobodies to ChemR23 was assessed by flow cytometry on ChemR23-overexpressing CHO cells, using WT CHO cells as the negative control. Nanobodies resulting in MCF ≥ 5 over WT CHO cells were sequenced. Two nanobodies were identified (CA4910 and CA5183) to bind ChemR23, sharing a highly similar CDR3 sequence with few amino acid differences. Following the expression and purification, by affinity chromatography, on Ni-NTA, the specificity of the two nanobodies was reconfirmed by FACS. As shown in Fig. 1, CA4910 and CA5183 nanobodies bind to CHO cells expressing human ChemR23, but they do not recognize mouse ChemR23 or human GPR1 or CCRL2.

The ability of purified nanobodies to label ChemR23 by immunofluorescence was evaluated on nonpermeabilized CHO cells expressing an engineered ChemR23 receptor fused to EGFP at its C terminus. Cells expressing the ChemR23-EGFP fusion and emitting green fluorescence were also labeled with the two nanobodies detected by DyLight 549–conjugated anti-His mAb (Fig. 2). Although EGFP fluorescence allowed detection of the receptors on the plasma membrane and in intracellular compartments, nanobodies only labeled cell surface receptors. Control experiments performed with cells expressing unrelated receptors (Fig. 2) or WT CHO cells, as well as incubation with irrelevant nanobodies or with a secondary Ab only, did not show any signal above background fluorescence (data not shown).

To estimate the affinity of CA4910 and CA5183 for ChemR23, the two nanobodies were labeled with DyLight 650 fluorochrome. Saturation-binding experiments were performed via flow cytometry using DyLight 650–labeled CA4910 (DL650-CA4910) or CA5183 (DL650-CA5183) as tracers. K_d values were calculated from saturation curves from MCF of CHO cells, expressing (total binding) or not expressing (nonspecific binding) ChemR23, following incubation with increasing concentrations of the tracer. DL650-CA4910 and DL650-CA5183 bound ChemR23 with K_d values of 130 ± 30 and 160 ± 10 nM, respectively (Fig. 3). We next evaluated whether the two nanobodies recognize similar epitopes by performing competition-binding experiments using DL650-CA4910 and DL650-CA5183 as tracers. We found that CA4910 and CA5183 displaced DL650-CA4910 binding with IC₅₀ values of 127 ± 4 and 187 ± 6 nM, respectively. Similarly, CA4910 and CA5183 displaced DL650-CA5183 binding with IC₅₀ values of 150 ± 10 and 190 ± 9 nM, respectively (Fig. 4). These data show that CA4910 and CA5183 share a common epitope, in agreement with the sequence similarity of these two nanobodies. The IC₅₀ values obtained from competition experiments were similar to the K_d values derived from saturation-binding curves, suggesting that labeling of nanobodies with DL650 did not affect their interaction with ChemR23.

We further tested whether the two nanobodies were able to affect the binding of chemerin, the natural ligand of ChemR23, on CHO cells expressing the human receptor. In a first set of experiments, we used the same competition-binding assays as above, with DL650-CA4910 and DL650-CA5183 as tracers. We found that chemerin displaced DL650-CA4910 binding with an IC₅₀ value of 100 ± 20 nM, and it displaced DL650-CA5183 binding with an IC₅₀ value of 66 ± 6 nM (Fig. 4). These results suggest that CA4910 and CA5183 share a common or overlapping binding site with chemerin. Therefore, we performed additional competition-binding experiments using the iodinated peptide [¹²⁵I]-[Y¹⁴⁵,F¹⁴⁹]-chemerin(145–157) as tracer, which is the sole high-affinity radioiodinated tracer commercially available for ChemR23. It is derived from a bioactive C-terminal peptide of chemerin (13). In agreement with previous experiments, we found that CA4910, CA5183, and chemerin(149–157) completely displaced [[¹²⁵I]]-[Y¹⁴⁵,F¹⁴⁹]-chemerin(145–157) binding with IC₅₀ values of 40 ± 6, 21 ± 4, and 23 ± 6 nM, respectively (Fig. 5). An irrelevant nanobody was also tested, but it was unable to compete for tracer binding (data not shown).

Cell-based signaling assays were performed to evaluate whether the two nanobodies were able to modulate the activity of ChemR23. Using a calcium-mobilization assay based on aequorin luminescence performed on CHO cells expressing ChemR23, the nanobodies were unable to activate the receptor (data not shown). However, both nanobodies inhibited chemerin- or chemerin(149–157)–induced intracellular calcium release (Fig. 6). However, inhibition of signaling was partial when full-size chemerin was used as the agonist, with IC₅₀ values of 60 ± 10 nM for CA4910 and 35 ± 3 nM for CA5183. The nanobodies behaved as full antagonists when the nonapeptide chemerin(149–157) was used as agonist, with IC₅₀ values of 17 ± 2 nM for CA4910 and 11 ± 4 nM for CA5183. An irrelevant nanobody did not modulate the activity of ChemR23 (data not shown).

Because previous studies showed that generation of bi- or multivalent nanobodies resulted in increased affinity, we engineered a bivalent CA4910 construct. For that purpose, a second CA4910 coding sequence, preceded by a GGGGSGGGG linker, was inserted in-frame at the 3′ end of CA4910 cDNA in a derivative pXAP100 vector in which the sequence encoding gene III was deleted. In saturation-binding experiments, this bivalent nanobody displayed a moderate increase in affinity (K_d = 95 ± 4 nM) compared with monovalent CA4910 (K_d = 150 ± 20 nM). Similarly, we found that the bivalent nanobody completely displaced [[¹²⁵I]]-[Y¹⁴⁵,F¹⁴⁹]-chemerin(145–157) binding in a competition-binding assay, with increased potency (IC₅₀ = 18 ± 4 nM) compared with the monovalent nanobody. Interestingly, we found that bivalent CA4910 almost completely inhibited chemerin-induced intracellular calcium release with an IC₅₀ of 17 ± 2 nM, whereas monovalent CA4910 behaved as a partial antagonist (Fig. 6). Moreover, the bivalent nanobody was 30-fold more potent at inhibiting intracellular calcium release promoted by the chemerin(149–157) nonapeptide, displaying an IC₅₀ value in the low nanomolar range (1.7 ± 0.3 nM, Fig. 6).

The ability of nanobodies to detect ChemR23 on human primary cells was tested. Monocytes were collected from buffy coats of healthy donors and derived into immature dendritic cells or macrophages using standard culture conditions. Expression of ChemR23 was monitored by FACS, using a commercial mouse mAb as reference, as well as DyLight 650–conjugated CA4910 (monovalent or bivalent) and CA5183 nanobodies. As shown in Fig. 7A, labeling of ChemR23 on the surface of human monocyte-derived macrophages or dendritic cells was as efficient with the fluorescently labeled nanobodies as with the reference mouse mAb.

The chemotactic potency of chemerin on human immature MDDCs was evaluated in the presence or absence of nanobodies. In the absence of nanobodies, chemerin tested at 30, 100, and 300 ng/ml concentrations promoted a chemotactic response. The maximal chemotactic response was obtained at different chemerin concentrations, depending on the donor. Therefore, the responses to the three chemerin concentrations were averaged and normalized according to the migration obtained for the reference chemokine, CCL3, used at 3 ng/ml (100%), after deduction of background migration toward control medium. The overall relative migration response of chemerin was 12 ± 2% (Fig. 7B). In the presence of 500 nM of monovalent or bivalent CA4910 nanobodies, the relative migration response toward chemerin was reduced to 6 ± 1% (p = 0.004) and 2 ± 1% (p < 0.001), respectively. Addition of 500 nM of an irrelevant nanobody had no significant effect. Similarly, none of the nanobodies tested significantly modified the CCL3-induced chemotaxis of immature MDDCs (mean, 87 ± 9%).

Over the past years, nanobodies have generated a growing interest as new diagnostic and therapeutic tools because of their potential advantages over conventional Abs (25, 26). Multimembrane-spanning proteins, such as GPCRs and ion channels, are extremely difficult to purify as native proteins. Consequently, the generation of good Abs that recognize the native conformation can be challenging. Following a procedure combining genetic vaccination, phage display, and biopanning on cells and VLPs, we identified nanobodies targeting ChemR23 and modulating its function. Indeed, we developed two nanobodies that specifically recognize human ChemR23, but not human GPR1 and CCRL2, the two other chemerin receptors. They bind ChemR23 with moderate affinity and share a common binding site, because they were both able to interfere with the binding of the other. This is consistent with the fact that the sequences of the two nanobodies are closely related, because they vary at only a few amino acid positions. We also found that chemerin, the natural ligand of ChemR23, completely inhibited the binding of the nanobodies. However, the measured IC₅₀ was relatively high compared with the established K_d of chemerin for ChemR23, suggesting that the binding sites of chemerin and the nanobodies overlap but are not identical. Similarly, the nanobodies completely inhibited the binding of a peptide tracer derived from the chemerin C terminus. Consistent with these binding data results, we found that the nanobodies were able to antagonize chemerin- and chemerin(149–157)-induced intracellular calcium release. However, the inhibition was partial when chemerin was used as agonist and complete when the chemerin(149–157) nonapeptide was used as agonist. Altogether, these results show that the two nanobodies share a common binding site that overlaps with that of chemerin and strongly suggest that the common epitope recognized by the nanobodies is close to the binding pocket recognized by the C terminus of chemerin, which likely involves the extracellular part of the transmembrane helix bundle of ChemR23. Indeed, although chemerin is structurally different from chemokines, it was suggested that these proteins could bind to their respective receptors according to a two-step binding model (34). Accordingly, the cystatin-like domain of chemerin would interact with the N terminus and extracellular loops of ChemR23, whereas the C-terminal peptide of chemerin, known to trigger activation of the receptor, would engage the helix bundle of the receptor (13, 19). Although this hypothesis regarding ChemR23 has not been confirmed experimentally, our present results are consistent with such a hypothesis. This was further supported by the results obtained with the bivalent CA4910 nanobody that we generated. Indeed, this bivalent nanobody displayed only a slight increase in affinity in a saturation-binding assay, but its efficacy as an antagonist of chemerin-induced intracellular calcium mobilization was enhanced significantly, probably resulting from increased nanobody size and steric hindrance. The bivalent nanobody also displayed a much higher potency (30-fold) as antagonist when the chemerin(149–157) nonapeptide was used as agonist, reaching low nanomolar IC₅₀ values. Similarly, the bivalent CA4910 nanobody inhibited chemerin-induced chemotaxis of human immature MDDCs more efficiently.

The known ability of nanobodies to bind to cavities, in contrast to classical Abs, also fits with the expected interaction of the C-terminal nonapeptide with the transmembrane helix bundle of ChemR23. Therefore, the nanobodies generated in this study might constitute interesting tools to further investigate the two-step binding model of chemerin to its receptor. Although the C-terminal peptides of human and mouse chemerin cross-react fully with human and mouse ChemR23, the nanobodies do not recognize the mouse receptor efficiently. Therefore, the binding site of nanobodies and of the chemerin C terminus is overlapping but different. Additional experiments involving site-directed mutagenesis will be necessary to investigate more precisely the respective residues involved in chemerin and nanobody binding.

The precise role of chemerin in physiological and pathological processes is incompletely understood. Chemerin was described initially as a chemoattractant factor for leukocyte populations expressing ChemR23. However, prochemerin is expressed constitutively by different tissues in noninflammatory conditions, and its bioactivity depends primarily on the complex proteolytic processing of its C terminus. Numerous proteases belonging to different classes and involved in various pathways were described to regulate chemerin bioactivity (14, 35–39), rendering the regulation of this system particularly complex. In addition, ChemR23 expression was described in many cell populations other than leukocytes, including preadipocytes and adipocytes (5, 6), skeletal muscle cells (9), and endothelial cells (10), and additional roles for the chemerin/ChemR23 system were proposed in the control of lipid and glucose metabolism (9, 17, 40), blood pressure (41), and angiogenesis (10). Finally, chemerin was demonstrated to bind with high affinity to two additional receptors: GPR1 and CCRL2. However, ChemR23 is the only receptor through which biological activities of chemerin have unambiguously been ascribed, including in vivo. Although chemerin is a chemotactic factor generated in inflammatory conditions, it behaves as a pro- or anti-inflammatory agent, according to the disease model studied (15, 16, 42–45). GPR1 and CCRL2 are receptors with limited and no signaling properties, respectively, and their roles are poorly understood. Only one antagonist of ChemR23 has been developed: the small molecule CCX832. It was described as an inhibitor of chemerin signaling through ChemR23 across multiple species (human, mouse, and rat) using radioligand-binding and calcium-mobilization assays (41). No antagonist of GPR1 or CCRL2 has been described. The nanobodies described in this article might constitute useful tools to delineate the complex activities of chemerin as antagonists able to discriminate the effects of chemerin mediated by ChemR23, GPR1, and CCRL2. We showed that our nanobodies detect ChemR23 on the surface of human leukocyte populations as efficiently as a mouse mAb widely used as reference and that a bivalent form was able to inhibit the migration of MDDCs toward chemerin. Because nanobodies can easily be engineered and modified, they also will constitute the basis of additional tools useful for detecting ChemR23 on cells and in tissue sections.

In conclusion, we generated and characterized two specific nanobodies recognizing human ChemR23 and demonstrated that they behave as antagonists. We also developed a bivalent nanobody that displays enhanced antagonist activity, including on human leukocyte populations ex vivo. Thus, these nanobodies constitute new original tools to study the distribution, pharmacology, and role of the chemerin/ChemR23 system in physiological and pathological conditions.

We thank INSTRUCT, part of the European Strategy Forum on Research Infrastructures, and the Hercules Foundation Flanders for Nanobody discovery support.

The authors have no financial conflicts of interest.