The Ligands of CXC Chemokine Receptor 3, I-TAC, Mig, and IP10, Are Natural Antagonists for CCR3*

Th1 and Th2 lymphocytes express a different reper-toire of chemokine receptors (CCRs). CXCR3, the receptor for I-TAC (interferon-inducible T cell a -chemoattrac-tant), Mig (monokine induced by g -interferon), and IP10 (interferon-inducible protein 10), is expressed preferen-tially on Th1 cells, whereas CCR3, the receptor for eotaxin and several other CC chemokines, is characteristic of Th2 cells. While studying responses that are mediated by these two receptors, we found that the agonists for CXCR3 act as antagonists for CCR3. I-TAC, Mig, and IP10 compete for the binding of eotaxin to CCR3-bearing cells and inhibit migration and Ca 2 1 changes induced in such cells by stimulation with eotaxin, eotaxin-2, MCP-2 (monocyte chemottractant protein-2), MCP-3, MCP-4, and RANTES (regulated on activation normal T cell expressed and secreted). A hybrid chemokine generated by substituting the first eight NH 2 -ter- minal residues of eotaxin with those of I-TAC bound CCR3

pathophysiological roles, it is possible to distinguish between inflammatory and homing chemokines. Inflammatory chemokines are produced in most tissues under pathological conditions upon stimulation by cytokines and bacterial toxins, whereas homing chemokines are produced constitutively at homing sites.
In the past few years it was found that lymphocytes can express most chemokine receptors in relation to their state of maturation, activation, and differentiation. CCR3, CCR5, and CXCR3, for instance, are up-regulated in T cells by treatment with IL-2 and are expressed differentially in Th1 and Th2 cells (1)(2)(3). These observations explain how T cells with appropriate cytokine production and effector properties can be attracted specifically into diseased tissues. CCR3-expressing Th2 cells are recruited together with eosinophils, which express the same receptor, to sites of allergic inflammation as shown by immunochemical analysis of nasal polyps and atopic dermatitis lesions (6,7). In such infiltrates, Th2 cells are believed to promote inflammation by releasing IL-4 and IL-5 as priming and survival factors for eosinophil and basophil leukocytes (8). CCR5, on the other hand, is characteristic for Th1 cells, which also express high levels of CXCR3. Th1 cells accumulate in delayed-type hypersensitivity reactions and autoimmune inflammation (9,10).
CCR3 binds many different CC chemokines, namely eotaxin, eotaxin-2, eotaxin-3, RANTES, MCP-2, MCP-3, and MCP-4 (4). The eotaxins are highly selective for CCR3, whereas RANTES and the MCPs recognize additional CC chemokine receptors. Eotaxin is expressed in a wide variety of cells, including eosinophils, lymphocytes, macrophages, and endothelial and epithelial cells, and is critically involved in the regulation of the basal and inflammation-dependent traffic of eosinophils (11,12). In eotaxin-deficient mice and in animals treated with antibodies that neutralize eotaxin, eosinophil infiltration of the airways is markedly reduced (13)(14)(15). CXCR3 is highly expressed on T cells activated with IL-2 and binds selectively I-TAC, Mig, and IP10 (4). Of the three ligands, I-TAC has the highest receptor affinity and is the most potent agonist, as shown by chemotaxis and Ca 2ϩ mobilization assays (16). A notable feature of I-TAC, Mig, and IP10 is that their production is induced by interferon-␥, a cytokine that is typically associated with Th1 responses (16 -18). IP10, for instance, is expressed in skin lesions caused by delayed-type hypersensitivity, psoriasis, and tuberculoid leprosy, where interferon-␥ expression is enhanced.
While studying the activities of several chemokines on Th1 and Th2 cells, we observed that I-TAC, Mig, and IP10 act as antagonists for CCR3. These data suggest that CXCR3 agonists in addition to attracting CXCR3-bearing cells have the capacity to inhibit responses mediated via CCR3.

EXPERIMENTAL PROCEDURES
Chemokine Synthesis-All chemokines and chemokine analogs were synthesized chemically using tBoc (tertiary butyloxycarbonyl) solidphase chemistry (19). They were purified by high pressure liquid chromatography and analyzed by electron spray mass spectrometry. For each chemokine used, the mass determined by mass spectrometry corresponded to the expected value.
Cell Preparation and Culture-Eosinophils (Ͼ98% pure) were purified from venous blood of healthy volunteers (20). A human Th2 cell line generated from cord blood cells was kindly provided by Dr. C. Chizzolini (University Hospital, Geneva, Switzerland). These cells were expanded periodically by restimulation with phytohemagglutin in the presence of feeder cells (21).
Chemokine Receptor Transfectants-Murine pre-B 300-19 cells that stably express chemokine receptors were generated by transfection. cDNAs for CXCR1 (22) Receptor Binding and Functional Assays-Competition binding assays were performed with CCR3-B300-19 cells using 125 I-eotaxin labeled by the Bolton-Hunter procedure (37). Briefly, the maximal binding of labeled eotaxin was determined by measuring binding at saturating concentrations; the bindability was about 20%. 5 ϫ 10 6 cells were incubated with 4 nM 125 I-eotaxin and increasing concentrations of unlabeled competitor (10 Ϫ9 -3 ϫ 10 Ϫ5 M) in 200 l of RPMI 1640 containing Hepes (25 mM, pH 7.4), bovine serum albumin (10 mg ml Ϫ1 ), and sodium azide (0.1%). The incubations were carried out for 30 min at 4°C, and cell-associated radioactivity was separated immediately by spinning the cells through a 2:3 mixture of diacetylphthalate and dibutylphthalate. The cpm that specifically bound to cells was calculated by subtracting the nonspecific cpm (the cpm bound in the presence of 100-fold molar excess of unlabeled eotaxin) from the total cpm that was bound to the cells. Dissociation constants (K values) were determined by Scatchard analysis using the computer program LIGAND (38). Ca 2ϩ mobilization was assayed in Fura-2-loaded cells after single or sequential stimulation with chemokines or chemokine analogs by recording [Ca 2ϩ ] i -related fluorescence changes (39). The rate of the change was expressed as the percentage of Fura-2 saturation/s. Chemotaxis was assessed in 48-well Boyden microchambers (Neuro Probe Inc., Cabin John, MD) using polyvinylpyrrolidone-free polycarbonate membranes (Poretics Corp., Livermore, CA) with 5-m pores (30). Cell suspensions and chemokine dilutions were made in RPMI 1640 containing 1% pasteurized plasma protein (Swiss Red Cross Laboratory, Bern, Switzerland) and 20 mM Hepes, pH 7.4. Migration was allowed to proceed for 2 h, and migrated cells were counted at a ϫ1000 magnification in 5 fields/well. All determinations were performed in triplicate.
Receptor Internalization-Chemokine-induced internalization was assayed as described (40). Briefly, CCR3-B300-19 or CXCR3-B300-19 cells were incubated for 40 min at 37°C with the chemokine to be tested at a 1 M concentration. After washing twice with phosphate-buffered saline, surface-bound ligands were removed by exposure to 50 mM glycine buffer, pH 3.0, containing 100 mM NaCl for 1 min followed by washing with phosphate-buffered saline. Receptor expression was then determined by flow cytometry (see "Receptor Binding and Functional Assays"), and the relative fluorescence intensity was calculated (41).

Inhibition of Chemotaxis-
The migration of CCR3-B300-19 cells and of human eosinophils in response to eotaxin was inhibited by I-TAC, Mig, and IP10, as shown for I-TAC in Fig.  1, A and B. The inhibition of chemotaxis induced by optimum concentrations of eotaxin was concentration-dependent and was complete at 100 -1000 nM (Fig. 1, C and D). I-TAC and Mig were equally effective and somewhat more potent than IP10. None of the CXCR3 ligands induced chemotaxis of CCR3-B300-19 cells or eosinophils (not shown).
Inhibition of Ca 2ϩ Mobilization-As shown in Fig. 2, the [Ca 2ϩ ] i rise induced by eotaxin in CCR3-bearing cells was decreased in a concentration-dependent manner by pretreatment with I-TAC, which was completely inhibitory between 100 and 1000 nM. I-TAC, on the other hand, did not induce [Ca 2ϩ ] i changes in CCR3-B300-19 cells or eosinophils, even at high concentrations, confirming that it is devoid of agonistic activity on CCR3 and indicating that it is a pure antagonist. Experiments with Th2 cells, which express CCR3 and CXCR3, show that blockade of CCR3 can occur concomitantly with the activation of CXCR3 (Fig. 2B). Marked [Ca 2ϩ ] i changes were observed after stimulation with increasing concentrations of I-TAC, which in turn prevented the response to eotaxin. Ca 2ϩ mobilization induced by eotaxin in CCR3-B300-19 cells and Th2 cells was inhibited in a concentration-dependent manner by all three CXCR3-selective chemokines (Fig. 2, C and D). I-TAC was the most potent antagonist, followed by Mig and IP10. In agreement with the chemotaxis assays, these results show that I-TAC, Mig, and IP10 significantly inhibit CCR3-dependent responses of Th2 cells at concentrations as low as 1 nM ( Fig. 2B). CCR3-B300-19 cells were less susceptible, presumably because they express higher numbers of receptors (Fig.  2C).
As several chemokines bind and activate CCR3, it was important to test the effect of the antagonists on responses induced by different ligands. As shown in Fig. 3, I-TAC prevented the [Ca 2ϩ ] i changes elicited in CCR3-B300-19 cells by eotaxin, eotaxin-2, MCP-2, MCP-3, MCP-4, and RANTES, demonstrating that the antagonistic effect is not dependent on the CCR3 agonist used. It was also important to test the specificity of the antagonists to block chemokine receptor-mediated responses. In a panel of 14 receptor-transfected B300-19 cell lines, I-TAC (the most potent of the three CCR3 antagonists) abrogated the [Ca 2ϩ ] i changes in response to stimulation with the appropriate chemokine only in CCR3-expressing cells (Fig. 4). I-TAC was slightly inhibitory on CCR5 (30 -40% decrease of the [Ca 2ϩ ] i rise in three experiments) but had no effect on all other receptors, indicating that its antagonistic activity is restricted largely to CCR3.
Effect of Eotaxin and MCP-4 on CXCR3-mediated Responses-It has been reported that eotaxin and MCP-4, which are agonists for CCR3, bind to CXCR3 and that eotaxin prevents [Ca 2ϩ ] i changes induced by IP10 (42). In view of the observed antagonism for CCR3, we tested the effect of eotaxin and MCP-4 on CXCR3-B300-19 cells that were stimulated with I-TAC, Mig, and IP10. Neither eotaxin nor MCP-4 inhibited Ca 2ϩ mobilization or chemotaxis induced by I-TAC, Mig, and IP10, indicating that their effect on CXCR3 was negligible (data for I-TAC are shown in Fig. 5).
CCR3 Antagonist Obtained by Chemokine Modification-It has been suggested that chemokines dock onto receptors by interacting with the loop region that follows the second cysteine and that the docking facilitates a triggering of the receptor by the NH 2 -terminal domain (43)(44)(45)(46). In the attempt to enhance the antagonistic effect, we synthesized a hybrid che-

FIG. 2. Inhibition of CCR3-mediated [Ca 2؉ ] i changes by I-TAC, Mig, and IP10. CCR3-B300-19 cells (A and C) and Th2 cells (B and D)
loaded with Fura-2 were exposed to increasing concentrations of I-TAC, Mig, or IP10 and stimulated after 60 s with 3 nM eotaxin. mokine by substituting the NH 2 -terminal region of eotaxin with that of I-TAC. As shown in Fig. 6, I-TAC/EoH1, corresponding to eotaxin with the first eight amino acids of I-TAC, was about 5-fold more potent than I-TAC itself as an inhibitor of CCR3-dependent [Ca 2ϩ ] i changes and chemotaxis. I-TAC/ EoH1 was also tested on CXCR3-bearing cells and was found to have moderate agonistic and no antagonistic activity (data not shown).
CCR3 Binding Studies-The relative affinities of I-TAC, I-TAC/EoH1, Mig, and IP10 to CCR3 were determined by binding competition assays with 125 I-labeled eotaxin (Fig. 7). All antagonists fully displaced labeled eotaxin. I-TAC/EoH1 was the most potent competitor (K d 4.5 Ϯ 1.0 nM, n ϭ 3) followed by eotaxin (K d 13.5 Ϯ 1.9 nM, n ϭ 3) and I-TAC (K d 65.0 Ϯ 7.7 nM, n ϭ 3), whereas the affinity of Mig (K d 4065 Ϯ 1, 231 nM, n ϭ 3) and IP10 (K d 1582 Ϯ 154 nM, n ϭ 3) was comparatively low. Overall, the binding data are in agreement with the observed antagonistic activities. The affinity of Mig, however, was lower than expected.
CCR3 Internalization-Binding of chemokines leads to a rapid receptor internalization, which is not observed on binding of antagonists (40,47). Internalization was determined in cells expressing either CCR3 or CXCR3 by flow cytometry measurements of surface receptors before and after ligand exposure. As shown in Fig. 8, in CCR3-bearing cells, receptor uptake was induced by eotaxin only. In CXCR3-bearing cells, on the other hand, receptor uptake was observed with I-TAC, Mig, IP10, and I-TAC/EoH1 but not with eotaxin. Together with the functional data, these results show that the CXCR3 ligands lack agonistic activity and act on CCR3 as pure antagonists. DISCUSSION This paper shows that I-TAC, Mig, and IP10 are potent antagonists for CCR3 and prevent the responses of Th2 cells and eosinophils to CCR3-binding chemokines.
The search for chemokine antagonists began several years ago when chemokine receptor blockade was recognized as a possible therapeutic approach for inflammatory diseases. It is well established that antagonists can be generated by modifying the NH 2 -terminal region of natural chemokines (48). Studies were performed first with IL-8 and other ELR chemokines yielding antagonists for CXCR1 and CXCR2 (49,50). The same principle proved valid for CC chemokines, as shown by the effects obtained upon NH 2 -terminal truncation of MCP-1, MCP-3, and RANTES (37, 51, 52). These studies indicate that, as a rule, receptor recognition depends on structural motifs located in the loop region of chemokines that follows the second cysteine. Extensive structure-activity studies of SDF-1 (stromal cell-derived factor-1) led to the proposal of a two-step interaction of chemokines with their receptors, involving specific docking via the loop region and subsequent triggering by the NH 2 -terminal region preceding the first cysteine (43, 53). In some cases, however, NH 2 -terminal truncation did not yield derivatives with antagonistic properties. This result was observed with eotaxin and IP10, which lose the capacity to bind to CCR3 and CXCR3, respectively, when only a few NH 2 -terminal residues are deleted. 2 It was also shown that dipeptidyl-peptidase IV (CD26) reduces the activity of eotaxin by cleaving off the first two NH 2 -terminal residues (54).
The observation that the naturally occurring I-TAC, Mig, and IP10 block CCR3 was unexpected, because normally CXC and CC chemokines discriminate precisely between CXC and CC chemokine receptors. Our data suggest that CCR3 and CXCR3, despite their overall sequence identity of only 34%, share sufficient structural similarity within domains that determine the binding of I-TAC, Mig, and IP10, on the one hand, and the binding and triggering by eotaxins, MCPs, and RAN-TES on the other. The existence of binding-relevant homology between CXCR3 and CCR3 is suggested in particular by the observation that the potency ranking for CCR3 antagonism, I-TAC Ͼ Mig Ͼ IP10, as shown in the present study, and the potency ranking for agonistic activity via CXCR3 as determined by Cole et al. (16) are the same. It is also noteworthy that NH 2 -terminal truncation of IP10 leads to a loss of agonistic activity on CXCR3 as well as a loss of antagonistic effects on CCR3 (data not shown). We found that CXCR3 ligands block CCR3, but we were unable to demonstrate the converse; this contrasts with the report of Weng et al. (42) and suggests that CCR3 ligands are unlikely to exert biologically relevant effects via CXCR3. It has been reported that murine secondary lymphoid tissue chemokine has agonistic activity on murine CXCR3 (55). Human secondary lymphoid tissue chemokine, on the other hand, is inactive on human and murine CXCR3 (Ref. 56; data not shown).
Attempts to design high affinity antagonists by NH 2 -terminal truncation of eotaxin have been unsuccessful. This paper presents an alternative approach. After realizing that CXCR3 and CCR3 receptors may have homologous binding domains, we synthesized a hybrid by replacing the NH 2 -terminal region of eotaxin with that of I-TAC. Eotaxin was chosen because it has the highest affinity for CCR3 and I-TAC because it is the best CCR3 antagonist. Substitution of the NH 2 -terminal region led to the loss of receptor triggering activity, whereas retaining high affinity binding yielded a chemokine analog with higher affinity to CCR3 than eotaxin and I-TAC.
Inflammatory and immune reactions are characterized by the production of chemokines in the affected tissues, leading to the infiltration of leukocytes that bear the appropriate receptors. The local expression of chemokines is often induced by cytokines. In the context of this paper, it is important to realize that I-TAC, Mig, and IP10, which attract CXCR3-bearing cells, are induced by interferon-␥ (16 -18), whereas eotaxin, a specific agonist for CCR3, is induced by IL-4 (67). The infiltrate observed in the presence of interferon-␥ is rich in Th1 cells, whereas Th2 cells predominate under the influence of IL-4. The present observations suggest that CXCR3-selective chemokines enhance this polarization by acting as antagonists of CCR3 and thus inhibiting the infiltration of Th2 cells, in addition to their effect as attractants of Th1 cells via CXCR3. This paper describes a new mechanism for the regulation of leukocyte recruitment by chemokines based on the combination of agonistic and antagonistic effects.