INTRODUCTION

Interleukin-8 (IL-8)¹ mediates the migration of neutrophils to sites of inflammation and tissue injury. This pro-inflammatory protein belongs to a large family of structurally related chemotactic cytokines, termed chemokines, that are subdivided into two distinct groups, CXC chemokines, for which the first two cysteines are separated by one amino acid, and CC chemokines, for which these cysteines are adjacent (2). Structure-function studies have demonstrated that the NH₂-terminal sequence of IL-8 is essential for receptor binding and neutrophil activation. In particular, deletion or mutagenesis of the three residues, Glu⁴-Leu⁵-Arg⁶ (ELR motif), which immediately precede the first cysteine, leads to complete loss of biological activity (3, 4). The ELR motif is a characteristic feature of all CXC chemokines that act via IL-8 receptors. Besides IL-8 the members of this group include GRO, GRO, GRO, NAP-2, ENA-78, and GCP-2 (2).

Two high affinity IL-8 receptors, which according to revised nomenclature are now termed CXCR1 and CXCR2 instead of either IL-8R1 and 2 or IL-8RA and B, have previously been identified on human neutrophils (5, 6). These receptors belong to the G-protein-coupled seven transmembrane segment class (2, 7). Both receptors bind IL-8 with high affinity (K_d 0.5-3 nM); however, only CXCR2 binds the other neutrophil-activating and ELR motif-bearing chemokines with high affinity (8-10). CXCR1 and CXCR2 share a high degree of sequence similarity within the membrane spanning domains but differ significantly within the extracellular and intracellular loops and the NH₂- and COOH-terminal domains. These observations suggest that the two receptors not only possess distinct ligand-binding properties but also could transduce different post-receptor signals. The development of molecules that selectively block these receptors could lead to useful tools for the characterization of IL-8 receptor signaling.

Previously we demonstrated that NH₂-terminal truncation of IL-8 and amino acid modification of the ELR motif can lead to IL-8 receptor antagonists (1). In this study we have examined NH₂-terminally modified analogs of IL-8, GRO, and PF4 for inhibition of neutrophil function and selective binding to CXCR1 or CXCR2 bearing cells. Using this approach chemokine antagonists were identified that discriminate between the two IL-8 receptors and selectively block their function.

EXPERIMENTAL PROCEDURES

All reagents were purchased from Merck, Fluka, or Sigma. RPMI 1640 medium and additional cell culture supplements, including G418 (geneticin), were obtained from Life Technologies, Inc. The expression vectors (pcDNA-1/pSVneo and pcDNA-3) for the generation of stable transfectants were from the Invitrogen Corp., and Na¹²⁵I was from Amersham Corp.

The chemokines and chemokine derivatives listed in Table I were synthesized with tertiary-butyloxycarbonyl chemistry and automated solid-phase methods as described previously (4, 11). (R)IL-8,NMeLeu² was prepared with N-methylleucine at position 25. This has been shown to prevent dimer formation in IL8(4-72) (12).

Table I. NH2-terminal sequences and binding characteristics of CXC chemokines and corresponding analogs with antagonistic properties Amino acid replacements are underlined. Kd values were determined in 125I-IL-8(1-72) competition binding experiments performed with Jurkat cells expressing either CXCR1 or CXCR2. Results from two independent experiments are shown. In all experiments IL-8(1-72) was included as a control, and the mean Kd values ± S.D. obtained from eight independent experiments were 8 ± 2 nM for CXCR1 and 6 ± 3 nM for CXCR2. NH2-terminal residues Kd (nM) CXCR1 CXCR2 Exp. 1 Exp. 2 Exp. 1 Exp. 2  IL-8(1-72) SAKELRCQCIK.. 8 7 6 8 (R)IL-8 RCQCIK.. 450 500 300 700 (R)IL-8,NMeLeua RCQCIK.. 250 300 150 350 (AAR)IL-8 AARCQCIK.. 250 400 85 60 GRO(1-73) ASVATELRCQCLQ.. 300 400 8 10 (R)GRO RCQCLQ.. 15,000 10,000 250 300 (ELR)PF4b ELRCLCVK.. 500 500 12 13 (R)PF4 RCLCVK.. 20,000 20,000 300 750 a (R)IL-8 with an N-methylated leucine at position 25 that has been shown to prevent dimerization in IL-8-(4-72) (12). b PF4 in which the NH2-terminal sequence EAEEDGDLQ was replaced by ELR (19).

Stable Jurkat transfectants bearing functional receptors for either CXCR1 or CXCR2 were generated as described previously (13, 14). These transfected cell lines were maintained at 37 °C, 5% CO₂ in RPMI 1640 containing 10% fetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin, 100 mg/ml streptomycin, and 0.8 mg/ml G418.

IL-8 iodination and competition binding studies with Jurkat transfectants were performed as described previously for neutrophils (8). Briefly, transfected cells bearing either CXCR1 or CXCR2 (1.2-2.0 × 10⁶ cells/assay point) were incubated at 4 °C in 3 nM ¹²⁵I-IL-8 for 90 min in the absence or presence of increasing concentrations of an unlabeled competing ligand (0.1-10,000 nM). The binding parameters were determined according to Schumacher and von Tscharner (15).

Cells expressing either CXCR1 or CXCR2 were loaded with fura-2-acetoxymethyl ester (0.1 nmol/10⁶ cells), as described previously (16). The cells were then stimulated with a chemokine or analog, and the fluorescence-related [Ca²⁺]_i changes were monitored. Antagonists were added 60 s before the stimulus.

Human neutrophils were isolated from buffy coats of donor blood, supplied by the Swiss Central Laboratory Blood Transfusion Service, SRK, as described previously (17). Chemokine-mediated elastase release was assessed using cytochalasin-B (5 µg/ml)-pretreated neutrophils stimulated with either IL-8 or GRO in the presence or absence of an IL-8 receptor antagonist (10-10,000 nM) (17). Superoxide production was determined using a superoxide dismutase-inhibitable cytochrome c reduction assay (18). Briefly, 10⁶ cells/ml of Krebs-Ringer buffer were stimulated at 37 °C, and the reduction of cytochrome c (80 µM horse heart, type III) was monitored at 550-540 nm. Cells were treated with varying concentrations of antagonist and 60 s later stimulated with 25 nM IL-8. The production of superoxide was quantified using an extinction coefficient of 19.1 mM¹ × cm¹ (18).

RESULTS AND DISCUSSION

Jurkat cells transfected with cDNA corresponding to either CXCR1 or CXCR2 have previously been shown to transmit signals for the mobilization of cytosolic free Ca²⁺, chemotaxis, and the activation of mitogen-activated protein kinase in response to IL-8 (13, 14). As shown in Table I both transfected receptors have similar affinity for IL-8 with K_d values ranging between 5 and 10 nM. The number of binding sites expressed on the two cell types is also similar, 13,300 ± 3,500 and 16,000 ± 4,500 per cell for CXCR1 and CXCR2 transfectants, respectively. These results indicate that Jurkat cells expressing either CXCR1 or CXCR2 are suitable for the characterization of specific IL-8 receptor antagonists. NH₂-terminally modified analogs of IL-8 had previously been shown to antagonize the effects of IL-8 on human neutrophils (1). Here the effect of several truncated CXC chemokines on CXCR1 and CXCR2 was tested to establish potential receptor preferences. The derivatives used are listed in Table I. (R)IL-8,NMeLeu designates an IL-8 analog with an N-methylated leucine at position 25, a modification that prevented dimer formation in IL-8(4-72) (12). The monomeric analog was similar in efficacy as (R)IL-8. (ELR)PF4 is a previously described variant of PF4 with the NH₂-terminal sequence preceding the first cysteine replaced by ELR. In contrast to native PF4, (ELR)PF4 has potent neutrophil activating properties (19).

Competition binding studies performed with the transfectants showed that (R)IL-8, (R)IL-8,NMeLeu, and (AAR)IL-8 compete equally well for ¹²⁵I-IL-8 binding to CXCR1 and CXCR2, although (AAR)IL-8 appears to possess a slightly higher affinity for CXCR2 than CXCR1. This is consistent with the behavior of IL-8, which binds with high affinity to both receptors. In contrast, (R)GRO and (R)PF4 competed for ¹²⁵I-IL-8 binding to CXCR2, with K_d values comparable to those observed for (R)IL-8 but were unable to displace ¹²⁵I-IL-8 from CXCR1 (Table I and Fig. 1). The agonist (ELR)PF4 bound with high affinity to CXCR2 only. This indicates binding to CXCR2 but not to CXCR1. In this respect the PF4 analogs are similar to the GRO proteins.

All the antagonists that we have examined here possess an arginine residue immediately preceding the first cysteine (Table I). This arginine is essential for the agonist activity of IL-8 and cannot be substituted with other amino acids (19). IL-8 analogs in which the integrity of the arginine in the ELR has been lost no longer bind to receptors. This has also been demonstrated with the PF4 agonist or antagonist that as shown here bind only CXCR2. We therefore conclude that the RCXC motif is required for the binding of the antagonists to both receptors and that a second binding domain determines the selectivity for CXCR1 or CXCR2. In agreement with this suggestion, evolutionary distance analysis of CXC chemokines, based upon sequence comparisons, shows that GRO and PF4, the parent molecules of (R)GRO and (R)PF4 which are selective for CXCR2, are more closely related to each other than to IL-8 (19). Evidence for the involvement of the secondary binding domain in receptor selectivity was also obtained in recent structure-activity studies using mutations of human and rabbit IL-8 and mutants of IL-8 and GRO (20, 21).

To demonstrate the receptor preference of the modified derivatives at the level of signaling and to establish their antagonistic properties, the effects on IL-8-mediated Ca²⁺ mobilization were determined. Fura-2-loaded Jurkat transfectants were treated with increasing concentrations (50-1000 nM) of the analog under test and were then stimulated with 50 nM IL-8. Addition of the analogs elicited no [Ca²⁺]_i response, as shown for (R)GRO and (AAR)IL-8 in Fig. 2A, indicating that they lack agonistic effects. The subsequent response to IL-8, however, was inhibited in a concentration-dependent manner by all analogs when CXCR2-transfected cells were used, which is consistent with their action as antagonists. In CXCR1 transfectants no inhibition was observed with (R)GRO and (R)PF4 confirming their selectivity for CXCR2 (Fig. 2).

A characteristic response of neutrophils stimulated by IL-8 and related chemokines is the release of granule enzymes (17, 22, 23). This process can be elicited by stimulation via CXCR1 or CXCR2 (24). We have determined the effect of the analogs on the release of elastase by neutrophils stimulated with IL-8 (10 nM) or GRO (30 nM). At 30 nM, GRO acts exclusively via CXCR2 (10). Treatment of the neutrophils with (AAR)IL-8, (R)IL-8,NMeLeu, (R)GRO, or (R)PF4 alone did not stimulate elastase release (data not shown); however, all four antagonists markedly inhibited the release induced by GRO (Fig. 3). In contrast, only (AAR)IL-8 and (R)IL-8,NMeLeu inhibited the response mediated by IL-8. These results demonstrate that (R)PF4 and (R)GRO selectively block CXCR2 and that stimulation of CXCR1 alone is sufficient for mediating exocytosis induced by IL-8. This observation is in agreement with our recent finding that CXCR2-specific monoclonal antibodies block elastase release in response to GRO but not in response to IL-8 (24).

Effect of IL-8 antagonists on elastase release induced in human neutrophils by IL-8 and GRO.

IL-8 induces superoxide production in human neutrophils through the activation of the NADPH oxidase (25, 26). By measuring the superoxide dismutase-inhibitable reduction of cytochrome c, we found that the rate of superoxide production in response to 25 nM IL-8 could be inhibited by prior addition of (AAR)IL-8 or (R)IL-8,NMeLeu. In contrast no effect was observed with (R)PF4 and (R)GRO which act on CXCR2 (Fig. 4). We have previously shown that the generation of superoxide in response to IL-8 is mediated only by CXCR1 (24). The present data extend this observation and emphasize the primary role of CXCR1 in the activation of the respiratory burst. Moreover, the results suggest that antagonists can be used to selectively block neutrophil functions; for example (R)GRO does not affect the respiratory burst but inhibits other activities, as described above.

The results of this study show that receptor antagonists can be generated that are selective both at the receptor and functional level. (R)GRO and (R)PF4 are selective antagonists for CXCR2, the receptor that has high affinity for all neutrophil-activating chemokines. Since both antagonists discriminate between the two IL-8 receptors, as shown for superoxide production and elastase release, they can be used to further investigate the differences in the signal transduction pathways of CXCR1 and CXCR2 in neutrophils. As a next step it will be important to find antagonists that are specific for CXCR1. Our results indicate that such antagonists would allow the selective inhibition of the respiratory burst without affecting other neutrophil responses like chemotaxis and granule enzyme release. Furthermore, by combining specific antagonists with signal transduction inhibitors, it may be possible to selectively control the functions of neutrophils.