Identification of Amino Acid Residues Critical for Biological Activity in Human Interleukin-18*

Interleukin-18 (IL-18) is a pro-inflammatory cytokine, and IL-18-binding protein (IL-18BP) is a naturally occurring protein that binds IL-18 and neutralizes its biological activities. Computer modeling of human IL-18 identified two charged residues, Glu-42 and Lys-89, which interact with oppositely charged amino acid residues buried in a large hydrophobic pocket of IL-18BP. The cell surface IL-18 receptor α chain competes with IL-18BP for IL-18 binding, although the IL-18 receptor α chain does not share significant homology to IL-18BP. In the present study, Glu-42 was mutated to Lys and Lys-89 to Glu; Glu-42 and Lys-89 were also deleted separately. The deletion mutants (E42X and K89X) were devoid of biological activity, and the K89E mutant lost 95% of its activity. In contrast, compared with wild-type (WT) IL-18, the E42K mutant exhibited a 2-fold increase in biological activity and required a 4-fold greater concentration of IL-18BP for neutralization. The binding of WT IL-18 and its various mutants to human natural killer cells was evaluated by competition assays. The mutant E42K was more effective than WT IL-18 in inhibiting the binding of 125I-IL-18 to natural killer cells, whereas the three inactive mutants E42X, K89E, and K89X were unable to compete with 125I-IL-18 for binding. Similarly, WT IL-18 and the E42K mutant induced degradation of Iκ-Bα, whereas the three biologically inactive mutants did not induce degradation. The present study reveals that Glu-42 and Lys-89 are critical amino acid residues for the integrity of IL-18 structure and are important for binding to cell surface receptors, for signal transduction, and for neutralization by IL-18BP.

The cytokine interleukin-18 (IL-18), 1 a member of the IL-1 superfamily, elicits several biological activities that initiate and promote host defense and inflammation following infection or injury. Unlike IL-1, human and murine macrophages constitutively expressed IL-18 in the absence of disease (1). The cDNA for human and murine IL-18 revealed the absence of a signal peptide (2,3), and it was predicted that the IL-18 precursor would be cleaved by the IL-1␤ converting enzyme (ICE, caspase-1) to generate a biologically active molecule (4). Although the hallmark for IL-18 activity is the induction of Th1 cytokines, e.g. interferon-␥ (IFN␥), IFN␥ production by IL-18 requires co-stimuli, such as IL-2, IL-15, or more commonly IL-12. IL-12 enhances the induction of IFN␥ by increasing the expression of both IL-18R␣ and IL-18R␤ chains (5,6).
IL-18R␣ is the ligand-binding receptor chain (7), whereas IL-18R␤ appears to be the signal-transducing chain (5,8). The IL-18 and IL-1 receptor chains share structural homology in their extracellular domains, which consist of three immunoglobulin-like repeats (9,10). The IL-18R␣ chain was originally cloned using degenerate primers of IL-1R␣ type I (11) and therefore was included in the IL-1 receptor family. The IL-18R␤ chain was similarly identified by homology to the IL-1 receptor accessory protein (8,12). Upon binding to its receptors, IL-18 induces activation of IL-1 receptor-associated kinase and TNF receptor-associated factor-6 (TRAF-6) and nuclear translocation of nuclear factor -B (NF-B) (13,14). Four new homologues of the IL-1 family (IL-1H) have been identified by homology searches of human and mouse expressed sequence tag libraries (15,16). Of these, IL-1H4 binds to IL-18R␣ chain (17).
IL-18-binding protein (IL-18BP) is a secreted protein that binds IL-18 and neutralizes its biological activities. IL-18BP has a single Ig domain but bears little homology to either chain of the IL-18R complex. It binds IL-18 with a high affinity (400 pM) and neutralizes the biological activity of IL-18 in vitro and in vivo (18,19). IL-18BP is constitutively expressed in healthy subjects and circulates at plasma concentrations of 2-5 ng/ml (20). IL-18BP is induced by IFN␥ in various cells, suggesting that it serves as a negative feedback inhibitor of the Th1 immune response (21). Viral homologues of IL-18BP are encoded by Molluscum contagiosum as well as other members of the Poxvirus family (18). Viral IL-18BP was expressed and was shown to bind and neutralize human IL-18 (22). Interestingly, skin infections of M. contagiosum exhibit a paucity of macrophages and T-cells despite a large number of infected epithelial cells, suggesting a natural viral defense against immune-mediated elimination by the host.
The lack of sequence similarity between IL-18BP and the ligand-binding IL-18R␣ chain is a challenge to identifying amino acid residues that are critical for binding of IL-18 and inhibiting its biological activity. Many cytokines have circulating soluble receptors constituting the ligand-binding extracel-lular domain of the cell surface receptor. These soluble receptors are produced either from receptor mRNA splice variants or by proteolytic cleavage of the cell surface receptor. Circulating soluble receptors exhibit lower affinity for their ligands compared with the cell surface receptors. In contrast, IL-18BP is not the counterpart of the cell surface IL-18R␣ chain, and its affinity is comparable with the cell surface IL-18 receptor. 2 The molecular modeling of IL-18 and IL-18BP predicts regions of interaction having matching hydrophobic and electrostatic interactions on the two counterparts (19). In the present study, we performed a mutational analysis to study the role of human IL-18 residues Glu-42 and Lys-89 in binding and biological activity. These amino acid residues were selected based on the molecular modeling. Glu-42 was mutated to Lys and Lys-89 to Glu. In addition, deletion mutants of these amino acids were studied. The functionality of the IL-18 mutants was assessed by induction of cytokines, induction of IB degradation, competition for receptor binding using 125 I-IL-18, and neutralization by IL-18BP. The results reveal the importance of Glu-42 and Lys-89 in IL-18 structure and function.

MATERIALS AND METHODS
Reagents-RPMI 1640 culture medium was purchased from Invitrogen and supplemented with 10 mM L-glutamine, 24 mM NaHCO 3 , 10 mM HEPES, 50 units/ml penicillin, and 50 g/ml streptomycin (Cellgro, Waukesha, WI). FBS was obtained from Invitrogen. IL-2 was purchased from R&D Systems (Minneapolis, MN). Recombinant human IL-12, IL-15, and TNF-␣ were gifts provided by PeproTech (Rocky Hill, NJ). Recombinant IL-18BP was a gift provided by Serono pharmaceutical research institute (SPRI, Geneva, CH). All restriction endonucleases and Taq DNA polymerase were purchased from Invitrogen. Talon affinity resin was obtained from CLONTECH (Palo Alto, CA) and used as recommended. For Western blot analysis, polyclonal rabbit anti-I-B␣ was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The secondary antibody, donkey anti-rabbit peroxidase conjugated antibody was purchased from Jackson ImmunoResearch (West Grove, PA) and developed using enhanced chemiluminescence from PerkinElmer Life Sciences.
Radioiodination of Human IL-18 -Mature human IL-18 (10 g) in borate buffer was radiolabeled with 125 I Bolton-Hunter reagent (PerkinElmer Life Sciences) according to the manufacturer's instructions. The reaction was stopped with excess glycine. Unbound 125 I was removed by chromatography on Sephadex G-25 equilibrated in PBS containing 0.25% gelatin and 0.02% sodium azide. Each fraction was counted, and the protein peak fractions were pooled. The pooled radiolabeled 125 I-IL-18 contained ϳ4.2 ϫ 10 7 cpm/g and was used for competition binding assays and for cell surface receptor cross-linking.
Primer 1 corresponds to the 5Ј open reading frame of proIL-18 cDNA, extended at its 5Јwith an EcoRI site. Primer 2 contained a Glu-42 to Lys mutation. PCR with these primers provided the 5Јhalf of the proIL-18 cDNA. The 3Ј half of the proIL-18 cDNA was prepared by PCR with primers 3 and 4. Primer 3 contained the Glu-42 to Lys mutation, whereas primer 4 corresponded to the 3Ј end of the proIL-18 cDNA, extended by a BamHI site. The two PCR products were purified by electrophoresis on 1% agarose gel, and the bands were eluted using a gel extraction system (Invitrogen). These two cDNA segments were mixed at a 1:1 ratio and used as the template for the second-step PCR to generate a complete human IL-18 cDNA in which Glu-42 is mutated to Lys. The complete E42K proIL-18 cDNA was ligated into Bluescript vector (Stratagene, La Jolla, CA) using the EcoRI and BamHI (Invitro-gen) restriction sites. The following mutants, Glu-42 to deletion (E42X), Lys-89 to Glu (K89E), and Lys-89 to deletion (K89X), were generated in the same manner using primers 5, 6, and 7, respectively, instead of primer 3. For Escherichia coli expression, each of the five IL-18 cDNA inserts was ligated into pPROEX TM HTa vector, which adds an His 6 tag at the N terminus of proIL-18 (Invitrogen) using the EcoRI and XbaI sites.
Protein Expression and Purification-Each of the five pPROEX TM -HTa/IL-18 plasmids was transformed into the competent E. coli strain DH5␣ (Invitrogen) and expressed as described previously (23). An overnight culture (25 ml) was added to 450 ml of LB medium containing 100 g/ml ampicillin and was grown to a density of 0.6 -1 A 600 . Protein expression was induced by adding isopropylthiogalactoside (0.3 mM), and incubation continued at 37°C with shaking for 3 h. Bacteria were harvested by centrifugation (5,000 ϫ g for 15 min at 4°C), and the pellet was suspended in 30 ml of Talon buffer (50 mM NaH 2 PO 4 , 20 mM Tris-HCl, 100 mM NaCl, pH 8). Cells were lysed by sonication (2 ϫ 30-s bursts) on ice. The soluble protein was clarified by centrifugation (4,000 ϫ g, 30 min, 4°C) and applied to a 2-ml mini-Talon column. The column was washed with 30 bed volumes of Talon buffer and eluted with 6 ml of 100 mM imidazole in Talon buffer. The elutant was dialyzed against Factor Xa buffer (20 mM Tris-HCl, 150 mM NaCl, 2 mM CaCl 2 ) at 4°C for 20 h. Aliquot (0.2 ml) of the dialyzed protein was incubated with 4 g of Factor Xa (New England Biolabs, Beverly, MA) for 4 h at room temperature in the presence of 2 mM phenylmethanesulfonyl fluoride (Invitrogen). The mixture was dialyzed in Talon buffer and passed over a Talon column to remove the pro-piece and uncleaved proIL-18, yielding mature IL-18 mutants. Purity and quantitation of the products was determined by 10% SDS-PAGE under reducing conditions and by staining with Coomassie Blue.
Cell Lines-The original NK92 cell line was derived from a patient and obtained as a kind gift from Dr. Hans Klingerman (Rush Medical Center, Chicago, IL). The human NK cell line used in the present studies was a subclone of the original cell line (24). NK cells were maintained in supplemented RPMI 1640 medium containing 10% FBS, 50 pg/ml IL-2 and 200 pg/ml IL-15 (Peprotech). The KG-1 cell line was obtained from ATCC and maintained in RPMI 1640 medium containing 10% FBS. For bioassays, NK and KG-1 cells were suspended at 0.5 ϫ 10 6 cells/ml in the RPMI 1640 medium (0.2 ml) in 96 flat well plates in the presence of 0.5 ng/ml IL-12 or 10 ng/ml TNF-␣, respectively. Different concentrations of IL-18 mutants were then added, and after 16 -20 h at 37°C in humidified air with 5% CO 2 the culture supernatants were collected for cytokine measurements.
Measurement of Cytokines-The liquid-phase electrochemiluminescence method was used to measure IFN␥ (25) and IL-8 (26) in cell culture supernatants. The electrochemiluminescence assay for IL-6 was developed using polyclonal and monoclonal antibodies purchased from R&D Systems. The monoclonal anti-human IL-6 was biotinylated, and the affinity-purified goat anti-human IL-6 was labeled with ruthenium as previously described (26). The amount of electrochemiluminescence was determined using an Origen Analyzer (Igen, Gaithersburg, MD). The limit of detection for IFN␥ was 62 pg/ml, and the limit of detection for IL-6 and IL-8 was 40 pg/ml.
Immunoblotting-KG-1 or NK cells (3 ϫ 10 6 cells) were individually stimulated with WT IL-18 or the different IL-18 mutants for 30 min at 37°C. The cells were washed three times with PBS. The cell pellets were suspended in 3ϫ the packed cell volume in buffer consisting of 20 mM Tris (pH 7.6), 0.4 M NaCl, 0.2 mM EDTA, 20% glycerol, 1.5 mM MgCl 2 , 2 mM dithiothreitol, 0.4 mM phenylmethanesulfonyl fluoride, 1 mM Na 3 VO 4 , and 2 g/ml each of leupeptin, pepstatin, and aprotinin. The suspended cells were then subjected to three freeze and thaw cycles. Soluble protein was obtained by centrifugation (14,000 ϫ g for 15 min at 4°C). Each 20-l sample protein was electrophoresed in 10% SDS-PAGE, transferred to nitrocellulose paper, and incubated with rabbit antibody to I-B (0.2 g/ml) overnight at 4°C. Peroxidase-labeled donkey anti-rabbit IgG was added and developed using enhanced chemiluminescence from PerkinElmer Life Sciences.
Binding and Cross-linking of 125 I-IL-18 to NK Cells-125 I-IL-18 (2 ϫ 10 6 cpm/50 ng) was incubated with 10 7 NK cells in the absence or presence of 100-fold molar excess of unlabeled IL-18 for 1 h at 4°C. For chemical cross-linking, disuccinimidyl suberate (Pierce) was added as described (18) for 20 min. After chemical cross-linking, the cells were washed twice in PBS and then processed for extraction of membrane proteins. Briefly, 0.5 ml of PBS was added to the cell pellet then homogenized for ten bursts on ice. The homogenate was centrifuged for 10 min at 500 ϫ g. The supernatant was collected and centrifuged for 15 min at 14,000 ϫ g at 4°C. The membrane protein pellet was dissolved in sample buffer and analyzed by 10% SDS-PAGE under reducing conditions.
Competition Binding Assays-NK cells were suspended in a binding solution of RPMI 1640 medium containing 2% FBS and 0.02% sodium azide at 1 ϫ 10 6 cells/0.2 ml. The cells were incubated with 100 ng of unlabeled WT IL-18 or IL-18 mutants for 10 min at 4°C. Thereafter, 125 I-IL-18 (2 ϫ 10 5 cpm/4.5 ng) was added for 20 min on ice. The cells were then washed twice with the binding solution, and bound radioactivity was determined in the cell pellet.
Statistical Analysis-Data are expressed as the mean Ϯ S.E. Group means were compared by analysis of variance (ANOVA) using Fisher's least significant difference. Statistical significance was accepted within 95% confidence limits. ANOVA and correlation analyses were performed with the statistical packages Statview TM 512ϩ (Brain Power, Inc., Calabasas, CA).

Sequence Similarity among Various IL-18R Ligands-
The amino acid sequences of human and mouse IL-18 as well as the human IL-1H4 short variant (IL-1H4s) are shown in Fig. 1. The amino acid sequences of human and mouse IL-18 and human IL-1H4 short variant reveal a 16.8% identity. The short IL-1H4 mRNA spliced variant (17) has a higher degree of alignment with the human IL-18 amino acid sequence than the long form splice variant of IL-1H4 (not shown). The ICE (caspase-1) cleavage site of IL-18 and a potential ICE cleavage site of IL-1H4 are underlined, respectively. Glu-42 and Lys-89 (shown in bold) are two amino acid residues predicted to be involved in the binding of human IL-18 to IL-18BP (19). These two amino acids are conserved between human IL-18 and human IL-1H4s. The Lys-89 residue is conserved in the mouse IL-18 as well (Fig. 1). These conserved residues were chosen for site-directed mutagenesis within human IL-18.
Generation and Purification of IL-18 Mutants-We first generated the IL-18 precursor (proIL-18) and exchanged the ICE cleavage site for a Factor Xa site for the purpose of producing mature IL-18 mutants by Factor Xa cleavage (23). In addition to the WT proIL-18, a two-step PCR procedure was used to generate four proIL-18 mutants: Glu-42 to Lys (E42K), Glu-42 to deletion (E42X), Lys-89 to Glu (K89E), and Lys-89 to deletion (K89X). The proIL-18 mutant contained a His 6 tag in its N terminus to facilitate affinity purification. Following expression in E. coli, each of the five proIL-18 recombinant proteins was soluble and could be purified using Talon affinity chromatography. The affinity-purified proIL-18 was first cleaved with Factor Xa, and then the N terminus His 6 tag pro-piece of IL-18 was removed by a second Talon affinity purification. The ma-ture IL-18, analyzed by SDS-PAGE and Coomassie Blue staining, revealed the 18-kDa mature IL-18 protein and its various point mutants (Fig. 2).
The Biological Activity of the IL-18 Mutants-WT IL-18 and its four mutants were tested for biological activity in NK and KG-1 cells by the ability to induce IFN␥ and IL-6 in the presence of IL-12 or TNF-␣, respectively (as co-stimulants), as well as IL-8 in the absence of any co-stimulant. WT IL-18 and E42K were active in both cell lines in all of the assays, whereas E42X and K89E mutants were inactive even at high concentrations. Interestingly, the E42K mutant was more potent than WT IL-18 (Fig. 3, A-D).
The four mutants and WT IL-18 were then examined for the ability to be neutralized by increasing concentrations of IL-18BP. As shown in Fig. 4, E42K required a 4-fold greater concentration of IL-18BP compared with the amount of IL-18BP needed to neutralize the WT IL-18. This difference was observed at the sensitive segment of the IL-18BP dose-response (50% inhibition). We also observed this difference over the entire dose-response range of IL-18BP. For example, WT IL-18 at 40 ng/ml is completely neutralized by 28 ng/ml of IL-18BP, whereas the E42K mutant required 112 ng/ml (also a 4-fold difference).

Binding of IL-18 and Its Mutants to the IL-18
Receptor on NK Cells-125 I-IL-18 was chemically cross-linked to NK cells in the presence or absence of the unlabeled WT IL-18. SDS-PAGE of the solubilized cell membranes revealed a cross-linked complex having an apparent molecular mass of 93 kDa. This band was not seen when cross-linking was performed in the presence of a 100-fold excess unlabeled WT IL-18 (Fig. 5A). The apparent molecular mass of the band corresponded to a 1:1 complex of IL-18 and IL-18R␣ chain. A weaker band (Ͼ200 kDa) was seen as well, and it probably corresponds to the ternary complex of IL-18 and the two receptor chains.
The results of the cross-linking of 125 I-IL-18 provided a basis for competition assays of 125 I-IL-18 binding with each of the four mutants on NK cells. The two biologically active forms of IL-18, WT IL-18 and E42K, specifically competed with 125 I-IL-18 for binding. In contrast, the three biologically inactive mutants did not affect the binding of 125 I-IL-18 to the NK cells even when present in the same high molar excess as the unlabeled WT IL-18 and E42K (Fig. 5B). Conserved amino acids in all of the sequences are shown in the Consensus lines. Alignment was generated using the Expert Protein Analysis System (ExPASy) with an additional manual adjustment. The underlined area indicates the ICE cleavage sites, and the amino acids in bold print indicate conserved charged amino acid residues that represent predicted binding residues of IL-18 to IL-18BP (19).

Induction of Ik-B␣ Degradation by WT IL-18 and Its
Mutants-To further localize the effect of IL-18 mutations to the receptor-binding level, we tested the IL-18-induced degradation of I-B␣ in both KG-1 and NK cells. Following activation by either WT IL-18 or E42K phosphorylated I-B␣ was rapidly degraded in KG-1 and in NK cells as determined by immunoblotting with anti I-B␣ antibody (Fig. 6, A and B). In contrast, the IL-18 mutants E42X, K89E, and K89X, which are inactive in stimulating the production of IFN␥, IL-6, and IL-8, were unable to induce degradation of I-B␣. I-B␣ appears as a double band (indicated by an arrow) representing two I-B␣ isoforms likely derived from mRNA splicing. The upper band of the doublet is degraded upon treatment of either cell type with biologically active forms of IL-18. The higher molecular mass band at ϳ100 kDa is nonspecific and indicates that an equal amount of protein was present in each lane.

DISCUSSION
The biological responses to IL-18 were initially focused on its ability to induce IFN␥ and to act as a Th1 cytokine. However, recent studies have shown that IL-18 also contributes to the Th2 immune response (reviewed in Ref. 27). For example, IL-18 plays an important role in IgE synthesis (28) and induces IL-13 (24). Moreover, IL-18 exhibits inflammatory properties that are independent of induction of IFN␥ (29). IL-18 also participates in the pathological processes that follow ischemia re-oxygenation (30). Therefore, interest in IL-18 has broadened to include diverse biological functions.
Hence, the present studies were intended to have a better understanding of ligand-receptor interactions for the response to IL-18. We introduced point mutations and deletions in IL-18 to identify amino acid residues involved in binding to the IL-18 receptor. Point mutations are useful in understanding and manipulating protein-protein interactions. For example, single point mutations in IL-1␤, IL-5, or IL-6 have converted these molecules from being potent agonists to functional receptor antagonists (31)(32)(33). In the case of IL-18, molecular modeling of the IL-18/IL-18BP complex predicted two amino acid residues in IL-18 (Glu-42 and Lys-89) that are critical for binding to IL-18BP (19). We show here that mutations of these amino acid residues affect the biological activity of IL-18, as measured in two cell types, KG-1 and NK cells (Fig. 3, A-D). The importance of these two charged amino acid residues is also suggested by their conservation in human and mouse IL-18, as well as in human IL-1H4 (Fig. 1). IL-1H4, a homologue of IL-1, binds to the IL-18R␣ chain (17). Previously, we demonstrated that exchanging Glu-42 or Lys-89 to Ala increased the biological activity of the resulting mutants severalfold (34). In the present study, we show that exchanging Glu-42 to Lys and Lys-89 to Glu or the deletion of either Glu-42 or Lys-89 resulted in distinctly different biological activities. The exchange of Lys-89 to Glu abolished IL-18 activities. The deletion of Glu-42 or Lys-89 resulted in nearly inactive molecules, suggesting the importance of these residues in binding to the IL-18 receptor. On the other hand, the exchange of Glu-42 to Lys enhanced biological activity compared with WT IL-18. The biological activities of the mutants were consistent using two cell lines for the induction of IL-8, IL-6, and IFN␥. Moreover, the induction of IFN␥ by the mutants was also consistent using primary cells in human peripheral blood mononuclear cells (data not shown).
The present data suggest that these point mutations resulted in significant structural changes affecting the affinity of IL-18 for the IL-18R␣ chain. The amino acids Glu-42 and Lys-89 of IL-18 were initially identified by molecular modeling as putative binding sites for IL-18BP (19). However, IL-18BP does not share significant homology to the IL-18R␣ chain (15.1%). In fact, the third domain of IL-1RI and IL-18R␣ share a greater identity (20.7%), and both consist of three Ig-like domains. In contrast, there is only one Ig-like domain in IL-18BP (19). Although there are differences in primary amino acid sequences of IL-18BP and IL-18R␣, the mutants suggest that IL-18R␣ chain and IL-18BP likely share a three-dimensional similarity.
Mutant E42K exhibits enhanced biological activity as well as increased resistance to neutralization by IL-18BP (Fig. 4). In fact, the Glu-42 to Lys mutant was 4-fold less sensitive to neutralization by IL-18BP as compared with WT IL-18. These data indicate that Glu-42 and Lys-89 in IL-18 are critical for binding to both IL-18R␣ and IL-18BP. The binding of IL-1H4 to IL-18R␣ may also include similar binding sites because the Glu-42 and Lys-89 are conserved amino acid residues present in IL-1H4 (Fig. 1). Although IL-1H4 binds to IL-18R␣ chain (17), to date there are no reports of an activity for IL-1H4.
We have also demonstrated that the inactive mutants failed to compete with 125 I-IL-18 for binding to receptors on NK cells, thereby identifying those amino acid residues needed for binding to the IL-18 cell surface receptors (Fig. 5B). The biological activities of cytokines usually correlate with the receptor binding affinities of the ligands; however, exceptions exist. For example, the human IL-18BP isoform c, which has an affinity for IL-18 of ϳ10-fold less than for human IL-18BP isoform a, exhibits a similar ability to neutralize IL-18 (19). In the case of IL-1 mutants, receptor affinities are either unchanged or modestly lower, but the biological activities are decreased 100-fold (35).
IL-18 induces the activation of several transcription factors: NF-B, STAT3, and AP-1 following the cascade of kinases MYD88, IRAK, and TNF receptor-associated factor-6 (TRAF-6) (36 -40). We observed a consistent difference between biologically active (WT IL-18 and E42K) and inactive (E42X, K89E, and K89X) IL-18 mutants and the degradation of I-B␣. Thus, we conclude that IL-18-dependent I-B␣ degradation in KG-1 and NK cells fails to release NF-B for nuclear translocation, which results in the induction of IFN␥, IL-6, and IL-8 in either KG-1 or NK. It is unlikely, however, that the IL-18 point mutations described in this paper are receptor antagonists since they were unable to prevent the binding of 125 I-IL-18 to NK cells even at 20-fold molar excess.