Neutrophil-Mediated Maturation of Chemerin: A Link between Innate and Adaptive Immunity¹ Free

Host defense against infection requires an integrated response of both the innate and adaptive arms of the immune system. At the gateway between innate and adaptive immunity are dendritic cells (DC),³ which are professional APCs that play a key role in the initiation and regulation of immune responses against invading pathogens (1, 2). Given the increasing prevalence of chronic inflammatory diseases and cancer, and the ambivalent role of APC in the physiopathology of these diseases, a growing interest has emerged for a better understanding of the mechanisms of APC recruitment. The involvement of chemokines in the recruitment of APC to inflammatory sites has designated a number of chemokine receptors as attractive targets for therapeutic applications in this area. However, their usefulness is often impaired because the inflammatory chemokines/chemokine receptor system is redundant and pleiotropic and none of the chemokine receptors is specific for APC (3, 4). We have recently identified a novel protein, chemerin, as the natural ligand of the previously orphan receptor ChemR23 (renamed chemerinR) (5). ChemerinR, also known as Dez in mice (6), exhibits a unique expression pattern among leukocyte populations that is expressed specifically in macrophages and immature DCs, the two major classes of APC (7). Chemerin was isolated from a human ascitic fluid secondary to ovarian carcinoma, and was characterized as a potent chemoattractant factor for macrophages and DCs. Recently, chemerin was shown to recruit human plasmacytoid DCs ex vivo, a property shared only by stromal cell-derived factor-1 so far (8, 9). In this context, the novel chemerin/chemerinR system represents an attractive candidate for future drug development. Interestingly, chemerin is synthesized as a secreted precursor, prochemerin, which is poorly active, but converted into a full agonist of chemerinR by proteolytic removal of the last six amino acids (5, 10). Such C-terminal cleavage processing is characteristic of cathelicidins that share a cystatin-like fold with prochemerin. This family includes precursors of bactericidal peptides (cathelicidins) (11), precursors of mediators active on leukocytes through G protein-coupled receptors (prokininogen, cathelicidin precursors) (12, 13) as well as cysteine protease inhibitors (cystatins). However, proteolytic enzymes that mediate prochemerin processing remain thus far uncharacterized and their identification might help to better understand in vivo chemerin generation, regulation, and function in specific human diseases. Maturation of prochemerin constitutes a critical limiting step in the chemerin-dependent recruitment of APC. We expected that bioactive chemerin would be produced during the initial phases of inflammation, preceding APC recruitment. Given the early infiltration by neutrophils during these early phases, we hypothesized that these cells could contribute to active chemerin generation. In support of this hypothesis, data reported in this study show for the first time that neutrophils, through the release of serine proteases cathepsin G (CG) and human leukocyte elastase (HLE), mediate the conversion of prochemerin into active chemerin. The generation of active chemerin by neutrophils, and its chemotactic properties on APCs, strongly support the early hypothesis that this novel mediator constitutes an important link between innate and adaptive immunity and play a central role in the initiation of immune responses.

Recombinant chemerin and prochemerin were obtained as described (5). Purified HLE, mast cell chymase and CG were obtained from Calbiochem. Proteinase 3 was purchased from Elastin Products. The purity of HLE and CG was ≥95% as determined by SDS-PAGE, according to the manufacturer. The other proteases, dextran T-500, cytochalasin B, fMLP, LPS, and protease inhibitors were purchased from Sigma-Aldrich. Secretory leukocyte protease inhibitor (SLPI) was purchased from R&D Systems.

Intracellular Ca²⁺ release in a CHO-K1 cell line expressing the human ChemR23 receptor was measured as described (5, 10) by a functional assay based on the luminescence of mitochondrial aequorin (14). Results were expressed as relative light units or as a percentage of the endogenous response to 20 μM ATP.

Human neutrophils were isolated from healthy donors. Briefly, after sedimentation with 3% dextran T-500 (Sigma-Aldrich) in isotonic NaCl, the leukocyte-rich supernatant was pelleted, resuspended in saline, and centrifuged at 400 × g for 40 min on Lymphoprep (Axis-Shield), for the removal of lymphocytes and monocytes. Remaining erythrocytes were lysed by addition of 1 vol of 0.2% NaCl, and tonicity was restored after 30 s by the addition of 1 vol of 1.6% NaCl. Cells were washed once and resuspended in the adequate buffer. With the exception of dextran sedimentation, all steps were conducted at 4°C. Water was checked as endotoxin- and pyrogen-free.

PMN (5 × 10⁵ cells/ml in HBSS with or without 0.01% BSA) were incubated for 30 min at 37°C with 5 μg/ml cytochalasin B and 0.5 μM fMLP. Alternatively, the cells were incubated for 5 min at 37°C with 1 ng/ml LPS, then for 30 min with 10 nM fMLP. Following stimulation, cells were centrifuged, and supernatants were used in the prochemerin conversion assay.

Recombinant human prochemerin (220 nM) was incubated in neutrophil-conditioned medium containing 0.01% BSA for 30 min at 37°C before being assayed for activation of ChemR23-expressing CHO-K1 cells. For the testing of enzyme inhibitors, the media were preincubated with inhibitors for 30 min at 37°C, after which 220 nM prochemerin was added and incubated for another 30 min at 37°C before being assayed. Potential interference of inhibitors with the aequorin assay was evaluated by testing the inhibitors alone and testing the ability of 5 nM chemerin to stimulate the chemerinR-expressing cell line in the presence of inhibitors.

Recombinant human prochemerin (44 nM) was incubated for 15 min at 37°C in 25 mM HEPES buffer, pH 7.4, containing 0.01% BSA and various amounts of purified human proteases (300 ng/ml to 3 pg/ml), before being assayed for activation of ChemR23-expressing CHO-K1 cells.

BALB/c mice were injected with human recombinant chemerin or the prochemerin COOH-terminal octapeptide FSKALPRS. Sera were tested by ELISA, and immune mice were used to generate mAbs by standard hybridoma technology, using the NSO myeloma cell line (15). The 5H6 mAb was shown to recognize specifically prochemerin but no mature chemerin, whereas the 7F6 mAb recognizes both active chemerin and prochemerin.

Recombinant human prochemerin (10 ng) was incubated for 30 min at 37°C in 25 mM HEPES buffer, pH 7.4, containing 0.01% BSA in the presence of 0.1 ng of purified human CG, elastase, or both, and the reaction was stopped by mixing with SDS-PAGE sample buffer. Proteins were separated by SDS-PAGE on a 15% polyacrylamide gel for ∼4 h and electrotransferred to nitrocellulose membranes. The blots were soaked into a solution of 2% skim milk, then incubated for 2 h with the 7F6 mAb, recognizing both active chemerin and prochemerin (dilution 1/3000), or with the anti-prochemerin 5H6 mAb (dilution 1/30,000). Bound Abs were detected using a peroxidase-labeled anti-mouse IgG and a chemiluminescence kit (Amersham). Quantitative gel analysis was performed using a GS-800 densitometer and Quantity One software (Bio-Rad).

Prochemerin (44 nM) was incubated for 30 min at 37°C in a BSA-free HEPES buffer in the presence of purified HLE or CG (15 ng/ml), and the reaction was stopped by heating at 95°C. Samples were then adjusted to pH 8.0 and incubated with 250 ng of trypsin (Promega) for 1 h at 37°C. The digested peptides were purified by solid-phase extraction (C18 StageTips; Proxeon Biosystems), eluted onto a metallic MALDI target, dried and mixed with 1.5 μl of matrix mix (2 mg/ml 2,5-dihydroxybenzoic acid and 10 mg/ml α-cyano-4-hydroxycinnamic acid, 2 mM fucose, 5 mM ammonium acetate). For the tests with PMN-conditioned media, prochemerin (220 nM) was incubated in a BSA-free medium for 1 h at 37°C. Following trypsin digestion, the peptides were separated onto a C18 reverse-phase 1 × 50 mm column (Vydac), which was submitted to a 5–95% CH₃CN gradient at a rate of 2.5%/min in 0.1% trifluoroacetic acid. Fraction collection was adjusted in a manner to eluate the three potential COOH-terminal peptides of chemerin or prochemerin in a single fraction for direct mass spectrometry analysis. The HPLC fractions were vacuum-dried and then resuspended in 1.5 μl of matrix mix onto the MALDI target. Mass spectrometry analysis was performed on a Q-TOF Ultima Global mass spectrometer equipped with a MALDI source (Micromass) and calibrated using the monoisotopic masses of tryptic and chymotryptic peptides from BSA. Ionization was achieved using a nitrogen laser (337-nm beam, 10 Hz) and acquisitions were performed in a V mode reflectron position. Microsequencing was obtained after argon-induced fragmentation after selection of the parent ion.

Monocyte-derived DCs, generated as previously described (5), were loaded with 5 μM fura-2 (Molecular Probes) for 30 min at 37°C in the dark (10⁷ cells/ml in HBSS without phenol red but containing 0.1% BSA). The loaded cells were washed twice, resuspended at 10⁶ cells/ml, kept for 15 min at 4°C in the dark, and transferred into the quartz cuvette of a luminescence spectrometer LS50B (PerkinElmer). Ca²⁺ mobilization in response to human chemerin was measured by recording the ratio of fluorescence emitted at 510 nm after sequential excitation at 340 and 380 nm.

Previous studies have demonstrated that the proteolytic processing of prochemerin into chemerin is essential for the APC chemotactic activity of the protein (5, 10). This processing affects the C-terminal end of the protein, located after the last cysteine involved in the disulfide bonds that presumably stabilize the cystatin fold domain of this secreted protein. Given the established role of PMN as the predominant migrating cell during the early stages of inflammation, we investigated whether conditioned media from purified neutrophils have the ability to convert prochemerin into biologically active chemerin. The neutrophil-mediated inflammatory response is dependent on the release of the content of their cytoplasmic granules upon activation, a mechanism termed degranulation (16). Human PMN were isolated from peripheral blood, and stimulated with potent inducers of degranulation, cytochalasin B and fMLP or LPS and fMLP (17). Human recombinant prochemerin was incubated with conditioned media, and its conversion into chemerin was tested by measuring the activation of the chemerinR using an aequorin-based calcium mobilization assay. Fig. 1,a shows that incubation of prochemerin with conditioned media from unstimulated PMN for 30 min resulted in a weak specific activity on chemerinR, reflecting basal proteolytic conversion of the precursor. This specific activity was however significantly increased after induction of PMN degranulation by cytochalasin B and fMLP, and a similar activation of the chemerinR was observed when PMN were primed with LPS and challenged with fMLP (Fig. 1 b).

Neutrophil granules have been shown to constitute an important reservoir of proteolytic enzymes, including the serine proteases CG, HLE, and proteinase 3, as well as various metalloproteases. To determine the nature of the activity involved in prochemerin processing, we next tested a range of protease inhibitors on the neutrophil-mediated chemerin generation, using the aequorin assay. We selected class-specific inhibitors active on serine, cysteine, aspartate, or metalloprotease. As shown in Fig. 1 c, the serine protease inhibitors were the most active, as PMN-induced chemerin production was significantly decreased by pretreatment with AEBSF (4-(2-aminoethyl)benzenesulfonyl fluoride) (69%), aprotinin (51%) and 3,4-dichloroisocoumarin (45%). Inhibitors of aspartic proteases (pepstatin), cysteine proteases (E-64), or metalloproteases EDTA had no effect in this assay. Together, these results indicate that activated PMN have the ability to generate biologically active chemerin from its precursor, and that this proteolytic processing is mediated by neutrophil-derived serine proteases released following PMN degranulation.

To further examine whether neutrophil-derived serine proteases are involved in chemerin generation, we next tested the ability of purified CG, HLE, and proteinase 3 to convert prochemerin into its active form. Prochemerin was incubated for 15 min with the proteases (3 pg/ml to 300 ng/ml), and the reaction mixtures tested for their activity on chemerinR. In parallel, we also tested a range of other human serine proteases, including plasmin, urokinase, thrombin, kallikrein, and mast cell chymase. Fig. 2,a shows that plasmin, urokinase, thrombin, kallikrein, chymase, and proteinase 3 did not promote the generation of bioactive chemerin, whereas CG and HLE efficiently converted prochemerin, down to low enzyme concentrations (0.3 ng/ml). In control experiments performed in the absence of prochemerin, it was verified that none of the enzymes displayed activity on chemerinR by themselves (data not shown). Moreover, Western blot analysis was used to determine whether the conversion of prochemerin into active chemerin involved the C-terminal part of the precursor, as expected from our previous observations (5, 10). mAbs were generated to discriminate the unprocessed prochemerin from chemerin activated by COOH-terminal truncation. As shown in Fig. 2 b, both proteins were immunodetected as a doublet presenting an apparent m.w. of 17,000. The 5H6 mAb, raised against the C-terminal octapeptide of prochemerin detects prochemerin but not chemerin. Maturation of prochemerin with CG, HLE or both, led to a decrease of the immunoreactivity, as a result of the loss of the C-terminal epitope recognized by the 5H6 mAb. Quantitative analysis revealed a decrease of the signal by 20, 45, and 80% for CG, HLE, and both enzymes, respectively. The apparent size of the protein, as well as its total amount following detection by the 7F6 mAb that recognizes both active chemerin and prochemerin, were not affected by the proteolytic treatment. These observations indicate that the C-terminal end of prochemerin is the only target site of CG and HLE and that these two proteases do not cleave prochemerin in its other domain. Altogether, these results demonstrate that, among the three neutrophil-derived serine proteases, CG and HLE have both the ability to generate active intact chemerin from its precursor and are most probably both involved in neutrophil-induced chemerin generation.

CG and HLE are enzymes that display different cleavage site specificities: CG was described to hydrolyze peptide bonds involving an aromatic residue, whereas HLE preferentially cleaves bonds downstream to small, nonpolar residues, particularly valine or serine (18). To determine the precise nature of the active chemerin forms generated by these enzymes, prochemerin was incubated with 0.5 ng of CG or HLE, and the products were analyzed by mass spectrometry, after trypsin treatment, with an emphasis on the C-terminal end of the protein. Proteinase 3, which does not induce bioactive chemerin generation, was also used in this assay. As expected, in the absence of proteases, a tryptic peptide with a molecular mass of 2031.92 Da was observed, corresponding to amino acids 141 to 158 of unprocessed prochemerin (Fig. 3,a). Following elastase treatment, an additional nontryptic peak appeared, with a molecular mass of 1903.82 Da. Fragmentation-induced microsequencing demonstrated that this peak corresponded to prochemerin cleavage after the serine in position 157 (Fig. 3,b). This C-terminal end corresponds to the major natural bioactive form of chemerin previously purified from human ascitic fluid (5). Interestingly, after incubation with CG, the 1903.82-Da peptide was not recovered, but another nontryptic peak of 1816.79 Da was found, resulting from prochemerin cleavage after the phenylalanine in position 156 (Fig. 3,c). As expected, these two peptides were not observed following prochemerin incubation with proteinase 3 (Fig. 3, b and c). Our data demonstrate therefore that in vitro processing of prochemerin by neutrophil-derived serine proteases results in the production of two distinct C-terminal variants of bioactive chemerin. Although HLE cleaves the Ser¹⁵⁷-Lys¹⁵⁸ bond, CG cleaves the Phe¹⁵⁶-Ser¹⁵⁷ bond, in agreement with the preferred cleavage sites reported previously for these two enzymes. It is worth noting that none of the other serine proteases tested in the functional assay was able to generate any of these two peptides (data not shown). Also, thorough analysis of the entire mass spectrum resulting from prochemerin processing by CG and HLE did not identify additional cleavage sites for these two proteases, in agreement with the Western blotting experiments as described (data not shown).

We next examined whether CG and HLE are both responsible for the neutrophil-mediated chemerin generation ex vivo. Prochemerin was incubated with conditioned medium from neutrophils activated by cytochalasin B and fMLP, assayed for biological activity on the chemerinR, and analyzed by mass spectrometry. To avoid contamination of the spectrum with a large number of signals resulting from the proteolysis of the N-terminal domain of prochemerin and other proteins present in the medium, peptides generated by tryptic digestion were first separated by reverse-phase chromatography. The three potential COOH-terminal peptides of prochemerin or chemerin were shown to elute at the same elution time and the corresponding fraction was analyzed by mass spectrometry. As shown in Fig. 4,a, both the 1903.82- and 1816.79-Da peptides, corresponding respectively to HLE and CG cleavage, could clearly be detected. Natural inhibitors of neutrophil serine proteases are found in plasma, including the SLPI, which act as a specific inhibitor of both CG and HLE (18). The effect of SLPI was evaluated on PMN-induced chemerin generation. Mass spectrometry analysis showed that the signals corresponding to the two active variants of chemerin were significantly decreased following pretreatment with SLPI (Fig. 4,a). Furthermore, in the functional assay, chemerinR activation was abolished when SLPI (1 μM) or plasma (20%) were added to the conditioned medium of cytochalasin B/fMLP-activated PMN (Fig. 4,b). Together, these data clearly demonstrate that neutrophil-mediated chemerin generation is performed by CG and HLE, both released in the extracellular milieu following neutrophil degranulation. The two chemerin variants generated by neutrophils are identical with the forms identified initially in human ascitic fluids (5, 10). These variants, characterized by mass spectrometry following HPLC purification, as previously described (5), were shown to activate human monocyte-derived immature DCs, for which the expression of ChemR23 has been established (Fig. 4 c).

The control of inflammatory mediator generation is a fundamental mechanism for the recruitment of leukocytes to inflamed tissues. In most instances, the production of a bioactive cytokine is regulated at the level of gene transcription, as most cytokines are secreted as active proteins. However, some cytokines and other regulatory molecules are secreted as inactive precursors that require proteolytic processing to be converted into biologically mature forms (19, 20, 21). Chemerin, a specific chemoattractant factor for APC, belongs to this category. Nevertheless, the nature of the protease that mediates the precise maturation of chemerin remains so far undetermined. As infiltration of inflammatory sites by APC is generally preceded by infiltration of PMN, we have explored in this study the hypothesis that PMN might be involved in the conversion of chemerin in vivo. Our results showed that PMNs isolated from human blood are able, following activation and degranulation, to efficiently convert inactive prochemerin into active chemerin. We further demonstrated that this maturation process can be attributed to the release of the serine proteases CG and HLE, both of which cleave specifically the C-terminal extremity of prochemerin. Mass spectrometry analysis of processed prochemerin revealed that the activity of these two enzymes leads to the production of the two previously shown distinct forms of bioactive chemerin, lacking respectively the last seven (chemerin 1–156) or six (chemerin 1–157) amino acids of the inactive precursor (10).

The involvement of neutrophils in the generation of chemerin is an interesting observation for the understanding of chemerin functions in vivo. As phagocytes recruited in the early stages of an inflammatory reaction, neutrophils represent the first line of defense against pathogenic invaders, and are therefore essential effector cells of innate immunity (22). The activation of neutrophils leads to their degranulation and to the release of microbicidal products and other agents involved in host defense mechanisms, including reactive oxygen species, cationic peptides, eicosanoids, and proteolytic enzymes. Neutrophils can also contribute to the development of adaptive immunity, as a number of products prestored in neutrophil granules were shown to attract cells involved in the initiation of specific immune responses, including DCs. As examples, the microbicidal peptides defensins are chemotactic for immature DCs and T lymphocytes (23), the human cathelicidin LL-37 recruits PMNs, monocytes, and T cells (12), whereas CG and azurocidin were shown to be chemotactic for PMN and mononuclear cells (24).

HLE and CG constitute, with proteinase 3, the three major serine proteases of the azurophil granules of neutrophils. CG and HLE are also found in the granules of monocytes, but in quantitative terms, neutrophils constitute by far the main source (25, 26). The main physiological function of these proteases is commonly thought to be intralysosomal degradation of engulfed cell debris or microorganisms. However, following neutrophil stimulation, CG, HLE, and proteinase 3 are rapidly released from cytoplasmic granules into the extracellular space, and evidence has accumulated over the past few years suggesting that HLE and CG also play crucial roles in extracellular proteolytic processes at inflammatory sites (27). Activation of neutrophils has also been shown to result in an increase in the catalytically active membrane-bound form of the proteinases on the neutrophil cell surface (27, 28, 29). The functional relevance of the neutrophil granules content in host defense mechanisms is illustrated by Chediak-Higashi syndrome, a rare genetic disease characterized by a defect in azurophil granules release. These patients suffer from severe bacterial infections and often develop an atypical lymphoproliferative syndrome (30). If neutrophil-derived proteases constitute fundamental components of physiological immune responses, the excessive, prolonged, or inappropriate activity of these enzymes can also play a pivotal role in physiopathological events leading to serious tissue damage and dysfunction (31, 32).

The chemerin form generated by HLE corresponds to the major natural active form purified originally from human ascitic fluid (5). Interestingly, the variant generated by CG was also found, although less abundantly, in the human ascitic fluid, and was shown to be active from a previous structure-function analysis (10). We have indeed demonstrated that a nonapeptide derived from the chemerin C terminus, ¹⁴⁹YFPGQFAFS¹⁵⁷ displayed biological activity on the chemerin receptor similar to that of the full-size protein. Removal of Ser¹⁵⁷ was relatively well tolerated. Removal of two amino acids or addition of a single amino acid at the C terminus of this peptide abolished its biological activity, indicating that very precise prochemerin processing is required for the generation of active chemerin. Our results provide therefore the first evidence that two different bioactive C-terminal chemerin variants can specifically be produced by primary immune cells, i.e., human neutrophils.

A remaining question is how prochemerin synthesis is regulated in vivo. Prochemerin transcripts have been detected in most human tissues tested, whereas no transcripts were found in peripheral blood leukocyte populations (5). In addition, high amounts of active chemerin were found in a diverse set of human inflammatory fluids (5). Although further studies will be necessary to understand the events that regulate prochemerin secretion in vivo, the available data suggest that inactive prochemerin might be produced constitutively by tissues in basal conditions. It would be converted into bioactive chemerin only in inflammatory conditions, following degranulation of infiltrating neutrophils, contributing to the enhancement of an Ag-specific immune response (Fig. 5). High amounts of CG and HLE are released upon neutrophil degranulation. Up to 1 μg of CG was reported to be released per million of activated PMN (33), and stimulated PMNs have been shown to have ∼100 ng of cell surface HLE or CG per 10⁶ cells (27, 34). According to our results in vitro, this is by far sufficient for proteolytic conversion of prochemerin into active chemerin. The conversion of prochemerin by unstimulated PMN is attributed to a partial activation of the cells during the purification process. It is however not excluded that other proteases might contribute to the activation of chemerin in vivo. In this context, the involvement of proteases of the coagulation cascade has been suggested recently (8).

In summary, this study provides strong evidence that bioactive chemerin generation can be mediated by the serine proteases CG and HLE following degranulation of neutrophils. This report is the first description of a precise mechanism leading to the generation of chemerin from its inactive secreted precursor. Our results strongly support the hypothesis that chemerin and its receptor play a pivotal role as a link between innate and adaptive immunity. They provide a novel mechanism by which tissue-infiltration by neutrophils can promote recruitment of APC at inflammatory sites, and suggest the involvement of chemerin in pathological processes characterized by an early neutrophil infiltration. Prochemerin processing constitutes also a novel regulatory function for elastase and CG in local inflammatory processes.

We thank Dominique Revets for expert technical assistance.

The authors have no financial conflict of interest.