Abstract
The CXC chemokine receptor 3 (CXCR3) is predominantly expressed on T helper type 1 (Th1) cells that are involved in inflammatory diseases. The three CXCR3 ligands CXCL9, CXCL10, and CXCL11 are produced at sites of inflammation and elicit migration of pathological Th1 cells. Here, we are the first to characterize the pharmacological potencies and specificity of a CXCR3 antagonist, N-1R-[3-(4-ethoxy-phenyl)-4-oxo-3,4-dihydro-pyrido[2,3-d]pyrimidin-2-yl]-ethyl-N-pyridin-3-ylmethyl-2-(4-fluoro-3-trifluoromethyl-phenyl)-acetamide (NBI-74330), from the T487 small molecule series. NBI-74330 demonstrated potent inhibition of [125I]CXCL10 and [125I]CXCL11 specific binding (Ki of 1.5 and 3.2 nM, respectively) and of functional responses mediated by CXCR3, such as ligand-induced guanosine 5′-O-(3-[35S]thio)triphosphate ([35S]GTPγS) binding, calcium mobilization, and cellular chemotaxis (IC50 of 7 to 18 nM). NBI-74330 was selective for CXCR3 because it showed no significant inhibition of chemotactic responses to other chemokines and did not inhibit radioligand binding to a panel of nonchemokine G-protein coupled receptors. There was a striking difference in potencies among the three CXCR3 ligands, with CXCL11 >> CXCL10 > CXCL9. A comparison of the rank order of Ki values with the rank order of monocyte production levels of these three ligands revealed a precise inverse correlation, suggesting that the weaker receptor affinities of CXCL9 and CXCL10 were physiologically compensated for by an elevated expression, perhaps to maintain effectiveness of each ligand under physiological conditions.
Chemokines are a group of chemotactic cytokines classified by the position of two cysteine residues in the N-terminal end that are either adjacent (i.e., CC) or separated by another amino acid (i.e., CXC) (Zlotnik and Yoshie, 2000). Chemokines and their G-protein coupled receptors are important in basal leukocyte trafficking between lymphoid and nonlymphoid organs and during inflammation (Baggiolini et al., 1997). The CXC chemokine receptor 3 (CXCR3) is of particular interest because of its involvement in T cell-mediated inflammatory diseases. CXCR3 receptors are predominantly expressed on the majority of activated inflammatory T helper 1 (Th1) cells, but not on resting Th1 cells or on resting or activated Th2 cells, and bind three different ligands, CXCL9 (monokine induced by IFN-γ), CXCL10 (IFN-γ-induced protein-10), and CXCL11 (IFN-inducible T cell α chemoattractant) (Farber, 1997; Cole et al., 1998). These ligands are mainly produced by activated monocytes and macrophages that have been stimulated by the Th1 cell cytokine IFN-γ (Loetscher et al., 1998). Although IFN-γ is the primary mediator of chemokine release from these cells, other inflammatory cytokines can also induce expression of these ligands in other cell types, including polymorphonuclear neutrophils (Gasperini et al., 1999), keratinocytes, fibroblasts, endothelial cells (Gattass et al., 1994), and astrocytes (Salmaggi et al., 2002). Because IFN-γ-activated monocytes and macrophages are present in most Th1 immune-mediated inflammatory sites, a high prevalence of CXCR3 ligand expression is usually observed, especially in inflamed joints of rheumatoid arthritis (RA) patients (Wedderburn et al., 2000; Patel et al., 2001; Ruth et al., 2001; Mohan et al., 2002), in multiple sclerosis lesions (Balashov et al., 1999), during pancreatitis in type 1 diabetes (Frigerio et al., 2002), and during allograft rejection in animal models and transplantation patients (Agostini et al., 2001; Melter et al., 2001; Hancock et al., 2003).
CXCR3 expression on pathogenic Th1 cells plays an exceptionally strong role in animal models of arthritis. Anti-CXCL10 antibody treatment protected Lewis rats from adjuvant-induced arthritis and reduced the associated delayed type hypersensitivity (DTH) (Salomon et al., 2002). In fact, using either neutralizing antibodies or CXCL10-deficient mice, it was demonstrated that CXCR3 and its ligands, especially CXCL10, were required for DTH and contact-sensitivity responses to antigen (Gautam et al., 1994; Dufour et al., 2002; Salomon et al., 2002; Xie et al., 2003). CXCR3 also plays a role in the lymphocytic choriomeningitis virus model of type 1 diabetes, in which islet β-cells of the pancreas are induced by the disease process to produce CXCR3 ligands, and disease onset was substantially delayed in CXCR3-deficient mice (Frigerio et al., 2002). These studies validate CXCR3 as a target for pharmaceutical intervention, since preventing ligand-receptor interaction may alleviate these inflammatory conditions.
Despite discrepancies in the literature concerning the rank order potency of the three CXCR3 ligands, all are considered to be specific agonists for this receptor (Cole et al., 1998; Wang et al., 2000; Cox et al., 2001; Xanthou et al., 2003). Presumably, they function redundantly in vivo to ensure proper immunomodulatory tone, as suggested by the viability of CXCL10-deficient mice (Dufour et al., 2002); however, Cox et al. (2001) have asserted that CXCL10 and CXCL11 bind to different states of the receptor in an allotopic manner. It is therefore important to investigate whether small molecule nonpeptide antagonists can interfere with receptor activation to either of these ligands. Nonpeptide small molecule antagonists have been shown to interact with chemokine receptors, including CXCR3. For example, the CCR5 small molecule antagonist N,N-dimethyl-N-[4-[[[2-(4-methylphenyl)-6,7-dihydro-5H-benzocyclohepten-8-yl]carbonyl]amino]benzyl]tetrahydro-2H-pyran-4-aminium chloride (TAK779), also binds to murine, but not human, CXCR3 with high affinity and is effective in reducing severity and incidence of collagen-induced arthritis in the DBA/1 mouse model (Yang et al., 2002; Gao et al., 2003). These proof-of-concept experiments have led to the development of specific small molecule antagonists for CXCR3, and one such compound (T487) is currently being developed for psoriasis and RA by ChemoCentryx, Inc. (Mountain View, CA) in collaboration with Tularik/Amgen Biologicals (Thousand Oaks, CA) (Medina and Johnson, 2002; Medina et al., 2004). Herein, we report the pharmacological properties of a novel CXCR3 antagonist of the T487 series (Medina and Johnson, 2002) and relate its function to the pharmacology and expression of the three CXCR3 ligands.
Materials and Methods
NBI-74330 Synthesis. The quinazolinone-derived CXCR3 antagonist NBI-74330 (Fig. 1) was synthesized as described elsewhere (Medina and Johnson, 2002).
Cell Lines. The human H9 T lymphoma line [HTB-176; American Type Culture Collection (ATCC), Manassas, VA] was maintained with RPMI-1640 medium that was supplemented with 2 mM l-glutamine, 10 mM HEPES, 50 U/ml penicillin and 100 μg/ml streptomycin, 1 mM sodium pyruvate, (medium and supplements purchased from Mediatech (Herndon, VA), and 20% fetal bovine serum (FBS; Hyclone, Logan, UT). For generation of primary human T cells (PHA/IL-2 T cells), written and informed consent was obtained from subjects from whom peripheral blood mononuclear cells were obtained from whole blood via Ficoll separation with CPT Vacutainer tubes (BD Biosciences, Franklin Lakes, NJ). Peripheral blood mononuclear cells were resuspended (106 cells/ml) in Dulbecco's modified Eagle's medium (DMEM) supplemented with 2 mM l-glutamine, 10 mM HEPES, 50 U/ml penicillin, 50 μg/ml streptomycin, 1 mM sodium pyruvate, 1% nonessential amino acids (8.9 mg/l l-arginine, 15 mg/l l-asparagine, 13.3 mg/l l-aspartic acid, 14.7 mg/l l-glutamic acid, 7.5 mg/l glycine, 15.5 mg/l l-proline, and 10.5 mg/l l-serine; Mediatech), 10% FBS, and 10 μM β-mercaptoethanol (Sigma-Aldrich, St. Louis, MO) and cultured with phytohemagglutinin (PHA-M; 10 μg/ml; Sigma-Aldrich) and human IL-2 (100 U/ml; Roche Diagnostics, Indianapolis, IN) for 48 h. Cells were washed twice with fresh medium and cultured with IL-2 (100 U/ml) for an additional 12 days before use.
Human CXCR3 was polymerase chain reaction-amplified from a thymus cDNA library using primers incorporating native stop and optimized (Kozak, 1981) start codons. The full-length gene was subcloned into the pcDNA5/FRT/V5-His-TOPO expression vector (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. This construct was transfected into Flp-In Chinese Hamster Ovary (CHO) cells using Lipofectamine 2000 according to the manufacturer's instructions (Invitrogen), and isogenic stable transfectant cell pools were selected in hygromycin B (200 μg/ml; Invitrogen). The rat basophilic leukemia cell line RBL-2H3 (ATCC CRL-2256; American Type Culture Collection) was transfected with a plasmid generated from pBSII into which was added the spleen focus-forming virus long terminal repeat, an SV-40 splice site, an SV-40 Poly A site, and the neomycin gene. The CXCR3 gene was placed under control of the spleen focus-forming virus long terminal repeat. The recombinant vector was transfected as described above, and stable transfectant clonal populations were selected in geneticin (G418) (250 μg/ml; Calbiochem, San Diego, CA). These two cell lines were maintained as monolayers in DMEM supplemented with 10% FBS, 2 mM glutamine, 12.5 mM HEPES, 50 U/ml penicillin, and 50 μg/ml streptomycin. All cells were cultured at 37°C in humidified incubators with a 5% CO2 atmosphere.
Radioligand Binding. Cell membrane fractions were prepared as previously described (Hoare et al., 2003) and resuspended in 50 mM HEPES, 10 mM MgCl2, 100 mM NaCl, and 1 mM CaCl2, pH 7.2 for use in competitive radioligand binding reactions. Reactions were performed in duplicate and consisted of 25-μl unlabeled chemokine (R&D Systems Inc., Minneapolis, MN) at indicated concentrations, 25-μl radiolabeled chemokine ligand (∼70 nM; [125I]CXCL11 and [125I]CXCL10 with specific activities of 1500 and 2200 Ci/mmol, respectively; PerkinElmer Life and Analytical Sciences, Boston, MA), and 50-μl membrane protein (5 μg) added sequentially in assay buffer (50 mM HEPES, 10 mM MgCl2, 100 mM NaCl, 1 mM CaCl2, and 0.1% BSA, pH 7.2) to low-binding 96-well plates (Corning, Acton, MA). The reaction was allowed to reach equilibrium by incubation at room temperature for 45 min while shaking. The amount of bound radioligand was determined by harvesting membranes via filtration through a UniFilter GF/C filter plate (PerkinElmer Life and Analytical Sciences) using a UniFilter-96 vacuum manifold (PerkinElmer Life and Analytical Sciences) (filters were pretreated with 1% polyethylenimine), washing twice with 400-μl wash buffer (10 mM HEPES, 5 mM MgCl2, 1 mM CaCl2, and 500 mM NaCl, pH 7.3), and measuring radioactivity by liquid scintillation using a TopCount NXT (PerkinElmer Life and Analytical Sciences). The dissociation half-life of [125I]CXCL11 was measured using CXCR3-CHO membranes that were allowed to equilibrate with radiolabel (∼70 nM) for 30 min (as described above) prior to the addition of excess cold CXCL11 (31 nM final) in the presence or absence of different concentrations of NBI-74330. Membrane-bound [125I]CXCL11 was assessed in duplicate along with nonspecific binding ([125I]CXCL11 plus excess cold CXCL11 added at the same) and total binding ([125I]CXCL11 without inhibitors) at each time point on the same plate.
Calcium Mobilization. RBL cells stably expressing CXCR3 were seeded (5 × 104 cells/well) in black 96-well clear bottom plates (Corning/Costar; Corning) and allowed to adhere for 24 h at 37°C in a humidified chamber with 5% CO2. Cells were loaded with the calcium-sensitive dye Calcium-3 before agonist was added as described by the manufacturer's instructions (Molecular Devices, Sunnyvale, CA). Agonists were added at different concentrations to cells in the absence or presence of antagonist (preincubated for 10 min), and calcium mobilization was assessed every 2 s during a 100-s duration using either an Image Trak (PerkinElmer Life and Analytical Sciences) or a Fluorescent Imaging Plate Reader (Molecular Devices), and relative fluorescent intensity units (RFU) are reported.
[35S]GTPγS Exchange. [35S]GTPγS exchange assays were conducted in triplicate in 96-well plates using 1.7-μg CXCR3-CHO membranes in binding buffer (50 mM HEPES, 10 mM MgCl2, 100 mM NaCl, and 1 mM EDTA, pH 7.2), saponin (5 μg), [35S]GTPγS (0.5 nM; 1250 Ci/mmol; PerkinElmer Life and Analytical Sciences), GDP (10 μM), and various concentrations of chemokines in a total volume of 100 μl. The reaction was allowed to reach equilibrium by incubation at room temperature for 1 h. The amount of bound [35S]GTPγS was determined by harvesting membranes via filtration through a UniFilter GF/C filter plate (PerkinElmer Life and Analytical Sciences) using a UniFilter-96 vacuum manifold (PerkinElmer Life and Analytical Sciences), washing twice with 400-μl wash buffer (50 mM Tris and 5 mM MgCl2, pH 7.4), and measuring radioactivity by liquid scintillation counting using a TopCount NXT (PerkinElmer Life and Analytical Sciences).
Chemotaxis. The Millipore MultiScreen-MIC assay (Millipore Corporation, Billerica, MA) was performed as previously described (Ott et al., 2004). Briefly, serial dilutions of 150-μl chemokines (R&D Systems, Inc.) were prepared in phenol red-free DMEM supplemented with 2 mM l-glutamine, 10 mM HEPES, and 1 mM sodium pyruvate (Mediatech) and added to the 96-well MultiScreen-MIC receiver plate. Cells (100 μl) in serum-free DMEM were added to the upper wells of the 5-μm MultiScreen-MIC filter plate. The MultiScreen-MIC receiver and filter plates were assembled as described in the manufacturer's instructions. After incubation at 37°C, the assay was terminated by rinsing the top of the MultiScreen filter with 200 μl of phosphate-buffered saline. Cells that migrated to the bottom chamber (i.e., receiver plate) were collected by the centrifugation of the receiver plate with the filter attached (400g for 15 min), and then the filter was removed and discarded. The number of migrated cells in the bottom chamber was detected using the DNA-intercalating Cyquant dye (Molecular Probes, Eugene, OR), and fluorescence intensity (RFU) was detected using 480-nm/530-nm emission/excitation wavelengths with an LJL Analyst fluorimeter (Molecular Devices).
Monocyte Culturing and Chemokine Assessment. The study protocol was approved by institutional review boards at The Barbara Davis Center for Childhood Diabetes (Denver, CO) and was conducted according to The Declaration of Helsinki Principles. Written and informed consent was obtained from 22 subjects (seven males and 15 females). Monocytes were obtained from each subject by first isolating peripheral blood mononuclear cells from whole blood via Ficoll separation with CPT Vacutainer tubes (BD Biosciences) and then separating the monocyte fraction by negative selection with a magnetic cell sorting bead separation kit (Miltenyi Biotec Inc., Auburn, CA). These monocytes (99% CD14+/HLA class II+ by flow cytometry) were seeded at 105 cells in 100 μl per well of 96-well flat-bottom tissue culture-treated plates (Costar) and incubated for 2 h at 37°C, 5% CO2, in a humidified chamber to allow monocytes to adhere and spread. Monocytes were stimulated with lipopolysaccharide (LPS; Escherichia coli, 0111:B4; Sigma-Aldrich) and recombinant human IFN-γ (BD PharMingen, San Diego, CA) in RPMI-1640 medium supplemented with 2 mM l-glutamine, 10 mM HEPES (Mediatech), 50 U/ml penicillin and 100 μg/ml streptomycin (Invitrogen), and 10% FBS (Cambrex Bio Science Walkersville, Inc., Walkersville, MD), and culture-conditioned medium was collected at 24 h. Culture-conditioned medium was stored at -20°C until the assessment of chemokine levels (i.e., human CXCL9, CXCL10, and CXCL11) using sandwich ELISA (R&D Systems, Inc.).
Analysis of Dose-Response Curves. Values for EC50, IC50, and Ki were calculated using the one-affinity state competition curvefitting model with an F test best fit using Prism 4.0 software (Graph-Pad Software, San Diego, CA). Radioligand dissociation data were analyzed using a monoexponential decay function (Prism 4.0).
Results
Characterization of Binding Affinities for NBI-74330 and the CXCR3 Ligands. To characterize the pharmacological properties of NBI-74330 (Fig. 1), we first determined its binding affinity in a competition binding assay with the CXCR3 ligand, CXCL11 (Fig. 2). NBI-74330 demonstrated potent inhibition of [125I]CXCL11 specific binding to membranes prepared from transfected CHO cells expressing CXCR3 (CXCR3-CHO) (Ki = 3.6 nM, Fig. 2). NBI-74330 was 12- and 3.5-fold more potent than CXCL9 (Ki = 45.2 nM) and CXCL10 (Ki = 12.5 nM), respectively, at inhibiting [125I]CXCL11 binding to CXCR3-CHO cell membranes, whereas CXCL11 (Ki = 0.069 nM) was 51-fold more potent than NBI-74330 (Fig. 2). Similar results were obtained with another transfected cell line (CXCR3-RBL) and with endogenously expressing CXCR3 cells (i.e., H9 and PHA/IL-2 T cells) (Table 1). The mean Ki values for each ligand varied by only 2- to 5-fold among the four cell types, demonstrating consistent pharmacological activity of the compound and CXCR3 ligands among diverse cell types expressing human CXCR3 (Table 1). In addition, the affinity range for the three natural CXCR3 ligands was striking, in that the Ki of CXCL11 was 1700-fold lower than that of CXCL9 and 350-fold lower than that of CXCL10 (Table 1). Despite this diversity, NBI-74330 demonstrated similar affinities in competition with either [125I]CXCL10 (mean Ki = 1.5 nM, n = 4; data not shown) or [125I]CXCL11 binding to CXCR3-CHO membranes (Table 1). Note that [125I]CXCL9 is not commercially available, and the low binding affinity would most likely preclude an accurate assessment of the inhibitory potency of NBI-74330 at that site (data not shown). NBI-74330 did not inhibit specific radioligand binding to a panel of other GPCRs, including the melanin concentrating hormone receptor 1, melanocortin receptor 4, histamine receptor 1, and corticotropin releasing factor receptor 1 (data not shown), thereby demonstrating specificity of NBI-74330 for the CXCR3 receptor.
Inhibition of CXCR3 Function by NBI-74330. Because NBI-74330 binds to CXCR3 with high affinity, we next determined whether the compound could effectively inhibit ligand-induced functional responses of CXCR3 such as calcium mobilization and [35S]GTPγS binding. RBL cells expressing CXCR3 receptors were used for calcium mobilization because this cell type endogenously expresses the promiscuous-binding Gα15 G-protein subunit. This subunit has been demonstrated to mobilize intracellular calcium stores during activation of GPCRs that couple to the Gαi subunit, such as the chemokine receptors (Gu et al., 2003). Expression of CXCR3 in other cell lineages did not result in a measurable calcium response (data not shown). CXCL9, CXCL10, and CXCL11 induced calcium mobilization in CXCR3-RBL cells with the same rank order of potencies as they did in inhibiting radioligand binding (EC50 values for CXCL9, CXCL10, and CXCL11 were 22, 7, and 0.1 nM, respectively; Fig. 3A). NBI-74330 inhibited calcium mobilization in response to CXCL11 and CXCL10 with an IC50 value of 7 nM for both ligands used at their EC80 concentrations (i.e., 1 nM for CXCL11 and 30 nM for CXCL10) (Fig. 3B). The relatively narrow effective dose range of NBI-74330 appears to be characteristic of the calcium mobilization response only. The time course of this assay prevents a steady-state measurement, and therefore the kinetic profile of the compound could disrupt a typical dose-dependent inhibition spanning two log units. Again, because of the weak potency of CXCL9, an accurate EC80 was not attainable, which precluded competition studies with NBI-74330. NBI-74330 specifically inhibited CXCR3-mediated calcium mobilization because there was no evidence of negative interactions downstream of CXCR3 activation. That is, the small molecule failed to inhibit lysophosphatidic acid-induced calcium mobilization in CXCR3-RBL cells (GPCR endogenously expressed on RBL cells; data not shown).
Ligand-induced G protein activation, as measured by [35S]GTPγS binding, is a membrane proximal event that can be used as a functional corollary for in vivo systems. CXCL9, CXCL10, and CXCL11 stimulated [35S]GTPγS exchange in CXCR3-CHO membranes with EC50 values of 260, 47, and 0.1 nM, respectively (Fig. 4A). The rank order of potency for this assay was similar to that observed with radioligand binding and calcium mobilization. NBI-74330 inhibited [35S]GTPγS binding induced by EC80 concentrations of 0.3 nM CXCL11 and 300 nM CXCL10 with IC50 values of 10.8 and 9.3 nM, respectively (Fig. 4B). NBI-74330 also dose-dependently inhibited CXCL11-induced [35S]GTPγS binding in membranes of cells endogenously expressing CXCR3 (i.e., H9 cells, IC50 value 5.5 nM; data not shown).
The inhibitory action of NBI-74330 on CXCR3 function was confirmed at the cellular level in chemotaxis assays with the endogenous CXCR3-expressing H9 cells. The difference in functional activities among the three ligands was also apparent in the H9 cell chemotactic response, in which CXCL11 showed not only a dramatically greater potency than the other two ligands but also induced a much greater degree of chemotaxis (CXCL9 and CXCL11 EC50 values were 24.5 and 0.5 nM, respectively, and CXCL10 was not potent enough to elicit a substantial response; Fig. 5A). Consistent with its potencies in other functional assays, NBI-74330 inhibited CXCL11-induced chemotaxis in these cells with an IC50 of 3.9 nM (Fig. 5B). A similar potency of NBI-74330 inhibition of CXCL11-induced chemotaxis of PHA/IL-2 T cells was also observed (IC50 of 6.6 nM; data not shown).
Functional Selectivity of NBI-74330 for CXCR3. Next we investigated the specificity of NBI-74330 for CXCR3 in chemotaxis experiments using H9 cells, which express the chemokine receptors CXCR4 and CCR7 in addition to CXCR3 (Ott et al., 2004) (Fig. 6). H9 cell chemotaxis was measured in response to the chemokine ligands CXCL11, CXCL12, and CCL19 that bind CXCR3, CXCR4, and CCR7, respectively (with predetermined EC80 values of 3, 10, and 100 nM, respectively), in the presence or absence of an IC90 concentration of NBI-74330 (i.e., 100 nM; see Fig. 5B). The mean background number of cells that migrated to medium alone was subtracted from all other mean values, and the percentage of maximum migration in the presence of NBI-74330 was reported. NBI-74330 inhibited CXCL11-induced chemotaxis by >99% but had no significant effect on chemotaxis induced by the other chemokines tested.
NBI-74330 Inhibitory Mechanism of Action. To further elucidate the inhibitory mechanism of NBI-74330 at CXCR3, a Schild regression analysis was performed using the CXCL11-induced [35S]GTPγS exchange (CXCR3-CHO) and calcium mobilization assays (CXCR3-RBL). Increasing concentrations of NBI-74330 were used to inhibit different concentrations of CXCL11 from inducing [35S]GTPγS binding. NBI-74330 had a dramatic effect on the maximum stimulatory activity of CXCL11 (i.e., Emax) but had no significant effect on the respective EC50 values (Fig. 7A). These data are consistent with characteristics of an insurmountable noncompetitive antagonist (Christopoulos and Kenakin, 2002). Using the calcium mobilization assay in CXCR3-expressing RBL cells as a different functional measure, we confirmed the noncompetitive nature of the antagonism of NBI-74330 (Fig. 7B). A similar profile of NBI-74330 antagonism was also observed using endogenously expressing PHA/IL-2 T cells in the [35S]GTPγS exchange assay (data not shown). To further investigate the insurmountable nature of NBI-74330, we determined whether this compound showed an allosteric potential in radioligand dissociation experiments, which would be plausible for an insurmountable antagonist. NBI-74330 did not affect the dissociation half-life of [125I]CXCL11, even when relatively high concentrations of the antagonist were used (i.e., t1/2 of [125I]CXCL11 dissociation at 0, 5, 50, 500, and 5000 nM NBI-74330 were 12.4, 10.3, 13.4, 12.0, and 9.5 min, respectively; Fig. 7C), suggesting that the noncompetitive mechanism of NBI-74330 does not entail an allosteric action. Alternatively, an alteration of the dissociation rate of [125I]CXCL11 would have indicated that NBI-74330 binds the receptor-[125I]CXCL11 complex, in which negative cooperativity would have reduced, whereas positive cooperativity would have increased, the affinity for the radiolabel. Nevertheless, there was no change in affinity of CXCR3 for [125I]CXCL11 in the presence of NBI-74330.
Inverse Correlation of Binding Affinity and Expression Levels of CXCR3 Ligands. We used CXCL10 and CXCL11 to characterize the inhibitory activity of NBI-74330 because the potency of CXCL9 in radioligand binding and functional assays was too weak to accurately evaluate antagonist potency. In fact, the wide range in potency among the three CXCR3 ligands was striking, spanning roughly 1000- to 2000-fold in Ki values (see Table 1). Because the potency of CXCL11 is similar to those of other chemokine GPCR ligands (i.e., Ki values are usually in the high pM range), the Ki values of CXCL9 were exceptionally weak (i.e., 29 to 61 nM) relative to CXCL11. This observation brings into question whether CXCL9 is a true ligand for CXCR3. To address the physiological significance of this variation in potencies, we determined whether differences in CXCR3 ligand expression under normal conditions correlated with the observed differences in receptor affinities.
Human monocytes from peripheral blood produce substantial levels of the three CXCR3 ligands when activated with LPS in combination with IFN-γ (Loetscher et al., 1998), but a rigorous assessment of the production of all three ligands simultaneously has not been reported. Therefore, we cultured the same number of activated monocytes from 20 subjects for 24 h and assessed levels of CXCR3 ligands in the conditioned medium via ELISA (Fig. 8A). A comparison of the mean chemokine levels from these 20 subjects revealed that the rank order of expression of the three ligands was reversed from that of their binding affinities, such that CXCL9 levels were the highest (1087 pg/ml), followed by CXCL10 (571 pg/ml) and CXCL11 (3.6 pg/ml) (Table 1). In fact, the mean Ki values for each ligand showed a precise inverse correlation with the respective mean expression levels, with an r2 correlation coefficient of 0.99 (Fig. 8B). Thus, this analysis suggests that the weaker receptor affinities of CXCL9 and CXCL10 relative to that of CXCL11 are physiologically compensated for by a precise elevation in expression levels, perhaps to maintain effectiveness and activity of each ligand. The physiological consequences of this inverse relationship remain to be determined.
Discussion
The chemokine receptor CXCR3 and its ligands have been implicated in several immune-mediated inflammatory diseases, and a few reports have described the vast differences in potencies of the three CXCR3 ligands. These two observations, in addition to an absence of published reports characterizing nonpeptide CXCR3 antagonists, prompted us to investigate the pharmacology of each ligand in comparison to a small molecule antagonist at the CXCR3 receptor. The results of this study provide a foundation of pharmacological information that should lead to an improved understanding of therapeutics directed at this chemokine receptor.
The CXCR3 antagonist T487 is currently being evaluated for efficacy in phase II clinical trials for psoriasis and RA, and the preclinical status of the compound has recently been reported at The 29th National Medicinal Chemistry Symposium (Medina et al., 2004). Because the molecular structure of T487 is not yet available, we characterized the pharmacological properties of a related compound, NBI-74330, from the same series (Medina and Johnson, 2002; Fig. 1). NBI-74330 was a potent and efficacious inhibitor of [125I]CXCL11 and [125I]CXCL10 binding to CXCR3 expressed on a variety of cell types with Ki values from 1.5 to 16 nM. Similar potencies of this antagonist were observed in the inhibition of CXCL11- and CXCL10-induced calcium mobilization, [35S]GTPγS binding, and CXCL11-induced chemotaxis (IC50 values were 3 to 10 nM). The Ki values of NBI-74330 were similar to those disclosed for T487 (i.e., 7 to 8 nM; Medina et al., 2004). NBI-74330 appears to be selective for CXCR3 because it did not affect chemotactic responses by the human H9 T lymphoma cell line in response to CXCL12 and CCL19, and it did not interfere with calcium mobilization induced by lysophosphatidic acid or radioligand specific binding to several nonchemokine GPCRs.
Our analysis of the inhibitory mechanism of NBI-74330 suggests a noncompetitive action, as defined by Schild analyses of both the [35S]GTPγS and calcium mobilization assays, in which the maximum signal induced by CXCL11 was dose-dependently reduced with NBI-74330. Dissociation half-life experiments demonstrated a nonallosteric action of NBI-74330, a characteristic that can be associated with noncompetitive antagonism. Note that NBI-74330 competed equally for either the [125I]CXCL11 and [125I]CXCL10 binding sites and also functionally antagonized receptor stimulation via either of these ligands with similar potencies, also supporting a nonallosteric mechanism of antagonism. Indeed, binding kinetics of antagonist compounds may play an important role in their mechanism of action (Swinney, 2004). That is, if the dissociation rate of NBI-74330 was significantly longer than that of CXCL11 in assays that are not measured at equilibrium (i.e., [35S]GTPγS exchange and calcium mobilization assays), then antagonist-occupied receptors would be functionally removed from the receptor population long enough to allow for the appearance of insurmountable antagonist activity. Such interesting kinetics can be observed with a variety of commercially available drugs, including aspirin (irreversible antagonist) and Celecoxib (COX-2 inhibitor and pseudo-irreversible antagonist).
Receptor affinities (i.e., Ki values) and stimulatory potencies in the [35S]GTPγS binding, calcium mobilization, and chemotaxis assays (i.e., EC50 values) among the three CXCR3 ligands were dramatically different in that the rank order of potencies was CXCL11 CXCL10 > CXCL9. The Ki and EC50 values for CXCL11 were usually 2000-fold and 200-fold, respectively, greater than those for CXCL9, and those for CXCL10 were usually 3- to 5-fold greater than those for CXCL9. The greater disparity of potencies observed in the membrane-based assays (i.e., radioligand binding and [35S]GTPγS binding, ∼2000-fold) among the three CXCR3 ligands relative to that of EC50 values of the whole cell functional assays (∼200-fold) may be explained by the intracellular amplification process of receptor-mediated signals that occur substantially downstream of the receptor. This concept is important in supporting that, although CXCL9 and CXCL10 have relatively weak affinities for CXCR3, they are in fact modestly effective at stimulating the necessary cellular activity for inflammatory cell migration. Nevertheless, the relatively low affinity of CXCL9 and CXCL10 for CXCR3 brings into question the biological significance of these ligands and whether they bind exclusively to CXCR3 or use an, as yet undefined, additional receptor with high affinity. Although past studies demonstrated similar binding affinities of CXCL10 and CXCL11 (Wang et al., 2000), our results here are in line with more recent studies that show CXCL11 binds CXCR3 with the highest affinity and that CXCL9 and CXCL10 have much weaker affinities (Cole et al., 1998; Cox et al., 2001; Xanthou et al., 2003). Investigations of possible mechanisms for such diverse potencies among these chemokines have led to hypotheses of differential binding states of CXCR3 governed by G-protein coupling (Cole et al., 1998; Cox et al., 2001) and differential utilization of receptor binding and activation sites that have been investigated via chimeric receptor studies (Xanthou et al., 2003).
We further addressed the issue of the biological significance of weak affinities of CXCL9 and CXCL10 by relating the relative in vitro expression of these ligands with their Ki values. A rigorous evaluation of the expression of all three ligands by IFN-γ-induced monocytes from 20 subjects showed a striking and precise inverse correlation between the differences in ligand potencies and expression. Our results are consistent with reports of CXCR3 ligand expression under pathological conditions such as during allograft rejection (Hancock et al., 2001) and in inflamed synovial tissue and fluid from RA patients (Wedderburn et al., 2000; Patel et al., 2001; Ruth et al., 2001; Mohan et al., 2002). However, few (Mach et al., 1999; Hancock et al., 2001) have investigated the expression of all three CXCR3 ligands simultaneously. We evaluated the precise differences in production among the ligands from the same cell type from healthy individuals. Our analysis strongly suggests a coevolutionary relationship between the ligand-receptor interaction and ligand expression such that the weak receptor affinities of CXCL9 and CXCL10 may be compensated for by elevated expression to ensure that each ligand achieves a significant effectiveness in vivo. In addition, all three CXCR3 ligand genes are adjacent to one another on a distinct human chromosomal locus, 4q21.21, supporting their close evolutionary relationship.
To date, the T487 series of compounds appears to be the only CXCR3 selective small molecule antagonist series reported, which shows efficacy in animal models of arthritis (Medina et al., 2004). Reports on its efficacy in phase II clinical trials for RA are pending. If such results are positive, new avenues could open for discovery of small molecule CXCR3 antagonist compounds that would be useful for treating not only RA but other Th1-mediated inflammatory conditions as well.
Acknowledgments
We thank Dr. Rich Maki and Dr. Dimitri Grigoriadis for critical review of the manuscript.
Footnotes
-
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
-
doi:10.1124/jpet.105.083683.
-
ABBREVIATIONS: CXCR, CXC chemokine receptor; Th, T helper type; CXCL, CXC chemokine ligand; IFN-γ, interferon-γ; RA, rheumatoid arthritis; DTH, delayed type hypersensitivity; CCR, chemokine receptor; NBI-74330, N-1R-[3-(4-ethoxy-phenyl)-4-oxo-3,4-dihydro-pyrido[2,3-d]pyrimidin-2-yl]-ethyl-N-pyridin-3-ylmethyl-2-(4-fluoro-3-trifluoromethyl-phenyl)-acetamide; [35S]GTPγS, guanosine 5′-O-(3-[35S]thio)triphosphate; ATCC, American Type Culture Collection; FBS, fetal bovine serum; DMEM, Dulbecco's modified Eagle's medium; PHA, phytohemagglutinin; IL-2, interleukin 2; CHO, Chinese hamster ovary; RBL, rat basophilic leukemia; RFU, relative fluorescent intensity unit; LPS, lipopolysaccharide; ELISA, enzyme-linked immunosorbent assay; GPCR, G-protein coupled receptor.
- Received January 13, 2005.
- Accepted March 3, 2005.
- The American Society for Pharmacology and Experimental Therapeutics