Characterization of a powerful high affinity antagonist that inhibits biological activities of human interleukin-13.

Interleukin-13 (IL-13), a predominantly Th2-derived cytokine, appears to play a central pathological role in asthma, atopic dermatitis, allergic rhinitis, some parasitic infections, and cancer. We hypothesized that an IL-13 antagonist may have profound therapeutic utility in these conditions. We, therefore, mutagenized human IL-13 in which Glu at position 13 was substituted by a Lys residue. This highly purified recombinant IL-13 variant, IL-13E13K, bound with 4-fold higher affinity to the IL-13 receptor than wild-type IL-13 but retained no detectable proliferative activity on the TF-1 hematopoietic cell line. IL-13E13K competitively inhibited IL-13- and IL-4-dependent TF-1 proliferation. It also inhibited IL-13-induced STAT-6 (signal transduction and activator of transducer-6) activation in immune cells and cancer cells and reversed IL-13-induced inhibition of CD14 expression on human primary monocytes. These results demonstrate that high affinity binding and signal generation can be uncoupled efficiently in a ligand receptor interaction. These results also suggest that IL-13E13K may be a useful antagonist for the treatment of allergic, inflammatory, and parasitic diseases or even malignancies in which IL-13 plays a central role.

Interleukin-13 (IL-13), a predominantly Th2-derived cytokine, appears to play a central pathological role in asthma, atopic dermatitis, allergic rhinitis, some parasitic infections, and cancer. We hypothesized that an IL-13 antagonist may have profound therapeutic utility in these conditions. We, therefore, mutagenized human IL-13 in which Glu at position 13 was substituted by a Lys residue. This highly purified recombinant IL-13 variant, IL-13E13K, bound with 4-fold higher affinity to the IL-13 receptor than wild-type IL-13 but retained no detectable proliferative activity on the TF-1 hematopoietic cell line. IL-13E13K competitively inhibited IL-13-and IL-4-dependent TF-1 proliferation. It also inhibited IL-13-induced STAT-6 (signal transduction and activator of transducer-6) activation in immune cells and cancer cells and reversed IL-13-induced inhibition of CD14 expression on human primary monocytes. These results demonstrate that high affinity binding and signal generation can be uncoupled efficiently in a ligand receptor interaction. These results also suggest that IL-13E13K may be a useful antagonist for the treatment of allergic, inflammatory, and parasitic diseases or even malignancies in which IL-13 plays a central role.
Cytokine receptors for hematopoietic growth factors with a four ␣-helix structure exist largely as homodimers or heterodimers (34). For example, receptors for erythropoietin, thrombopoietin, granulocyte colony-stimulating factor, growth hormone, prolactin, and leptin can exist as homodimers (34). In this class of receptors, a single cytokine binds to two identical subunits and causes receptor internalization and signal transduction. Heterodimeric cytokine receptors generally consist of a major cytokine binding subunit and a signaling subunit (or shared subunit) (34). In this class of receptors, signaling subunits are often shared with more than one cytokine. For example, gp130 receptor subunit is shared with receptors for IL-6, IL-11, leukemia inhibitory factor, ciliary neurotrophic factor, cardiotrophin-1, and oncostatin M. IL-2R␥ subunit is shared by IL-2, IL-4, IL-9, and IL-15 receptors. The common ␤-subunit is shared by receptors for IL-3, IL-5, and GM-CSF (34). Similarly, IL-4R␣ subunit is shared by the IL-4R and IL-13R system (14,(35)(36)(37). In heterodimeric receptor systems, usage of intracellular signaling mechanism(s) is generally also shared (34).
One of the difficulties in understanding the interaction between ligand and the shared receptor subunits is that the subunit itself usually binds its ligand with low affinity. To overcome this problem, numerous cytokine antagonists have been generated by site-directed mutagenesis, which has been shown to utilize heterodimeric receptor systems. These mutants have clarified the crucial role of a particular residue in the ligand-receptor interaction. Among them, various antagonistic muteins including mGM-CSF (38,39), hIL-5 (40), human IL-6 (41, 42), human leukemia inhibitory factor (43,44), human IL-15 (45), human and murine IL-2, and human IL-4 (46,47) have been produced. However, no antagonist of murine or human IL-13 has been produced.
To produce an IL-13 antagonist, we created a mutation in the IL-13 molecule. The selection of the residue to be mutated was based on the knowledge of mutants produced for the IL-4 family of lymphokines (e.g. GM-CSF, IL-5, IL-4, and IL-13) (48,49). When the amino acid sequence of ␣-helix A of these molecules was aligned, a conserved structure was identified. Mutation in one of the Glu residue in the conserved helical structure produced molecules with altered interaction with their recep-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

EXPERIMENTAL PROCEDURES
Materials-Sequence-specific oligonucleotide primers were synthesized at Bioserve Biotechnologies (Laurel, MD). The pET based expres-

FIG. 1. Prediction of the hot residue in the ␣-helix A of IL-13 molecule.
A, alignment of ␣-helices A between mGM-CSF, hIL-5, hIL-4, and hIL-13. Helical residues are in bold. The structurally conserved residues are boxed, whereas consensus buried hydrophobic side chains are shown in reverse (white characters in black box). *, marked Glu of mGM-CSF, hIL-5, and hIL-4 were reported to interact with shared receptor subunits. B, aligned ␣-helices of hIL-13, hIL-4, and mGM-CSF are also shown in helical wheel projections (1 turn every 3.6 residues). Hydrophobic and buried residues may constitute the inner core and are indicated in a box. The aligned and proposed hot Glu are marked with asterisk (*). sion vector (Novagen, Madison, WI) was used for construction of mutein clone. Plasmids were amplified in E. coli, DH5␣ (Life Technologies, Inc.), and DNA was extracted using plasmid purification kits (Qiagen, Chatsworth, CA). Restriction endonucleases and DNA ligase were obtained from New England Biolabs (Beverly, MA), Life Technologies, Inc., Panvera (Madison, WI), and Roche Molecular Biochemicals. TF-1 human erythroleukemia cell line was purchased from ATCC (Manassas, VA). PM-RCC renal cell carcinoma cell line was established in our laboratory (50). THP-1, human monocytic cell line, TORY, virus-immortalized B cell, and KSY-1 AIDS-related Kaposi's sarcoma cell line were obtained and maintained as previously described (51).
Construction of Plasmids Encoding IL-13R112D and IL-13E13K-The mutagenesis of IL-13 gene was performed using cDNA of wtIL-13 (20) as a template. Sense primer 5Ј-agg aga tat aca tat gtc ccc agg ccc tgt gcc tcc ctc tac agc cct cag gaa gct cat tga gga-3Ј and antisense primer 5Ј-taa ttt gcc cga att cag ttg aac cgt ccc tcg cg-3Ј were used to mutate Glu-13 to Lys and incorporate NdeI and EcoRI restriction enzyme sites at the 5Ј and 3Ј termini, respectively. Construction of the expression vector for IL-13R112D was described before (51). After subcloning the PCR products, the fragment was restricted by NdeI and EcoRI and inserted into an expression vector. We confirmed the existence of mutation and restriction sites by sequencing of the plasmid.
Expression and Purification of Recombinant Proteins-Expression and purification of wtIL-13, IL-13 mutants, and IL-4 was carried out by similar techniques as previously reported (51,52). wtIL-13, IL-13 mutants, and IL-4 were produced in inclusion bodies.
Cell Proliferation Assays-Proliferation assays were performed as described previously (51,53). Briefly, 1 ϫ 10 4 TF-1 cells/well were cultured in 96-well plates in RPMI with 5% fetal bovine serum. Varying concentrations of wtIL-13 or IL-4 and/or IL-13 mutein were added to the wells, and the cells were cultured for ϳ2 days. Tritiated thymidine (0.5 Ci) was added to each well 6 -12 h before the plates were harvested in a Skatron cell harvester (Skatron, Inc., Sterling, VA). Glass fiber filter mats were counted in a ␤ Plate counter (Wallac, Gaithersburg, MD).
IL-13 Receptor Binding Studies-wtIL-13 was labeled as previously described (2,51). The specific activity of radiolabeled IL-13 was 26 Ci/g. The equilibrium binding studies were performed as described elsewhere (2,51). Briefly, 5 ϫ 10 5 cells in 100 l of binding buffer were incubated at 4°C for 2 h with 125 I-IL-13 (200 or 500 pM) in the absence or presence of various concentrations of unlabeled wtIL-13 or IL-13 mutant. Receptor-bound 125 I-IL-13 was separated from unbound 125 I-IL-13. The cell pellets were counted in a ␥ counter (Wallac).
CD14 Regulation by IL-13-Primary monocytes were cultured at 1 ϫ 10 7 cells/ml for 48 h with 1 ng/ml wtIL-13 with or without 1 g/ml IL-13E13K. Staining of the cells was performed as described elsewhere (51). The fluorescence data were collected on a FACScan/C32 (Becton Dickinson, San Jose, CA). The results were analyzed with the CELLQuest (Becton Dickinson) program.
Protein Synthesis Inhibition Assay-Protein synthesis inhibition assay was performed as previously described (3,51). In brief, 1 ϫ nally, cells were washed and harvested on a fiberglass filtermat, and cell associated radioactivity was measured in a ␤ Plate counter (Wallac). The concentration of IL-13PE38QQR at which 50% inhibition of protein synthesis (IC 50 ) occurred was calculated.
Sequence Alignment and Molecular Modeling of IL-13 Receptor CRH Domain and IL-13-The sequence of CRH domains of various cytokines were aligned by the Bestfit program of GCG software (Genetics Computer Group, Inc., Madison, WI). Helical wheel analyses were also performed using GCG software. Percent similarity and identity of extracellular domains between IL-13R␣1 and IL-2R␥ subunits were 40.5 and 31.9%, respectively. These numbers indicate reasonable sequence similarity justifying the use of IL-2R␥ subunit as a template for mod-eling IL-13R␣1 subunit. Conserved sequence patterns such as the WSXWS motif and four conserved Cys residues between ␤-strands of IL-13R␣1 and IL-2R␥ were perfectly aligned. The alignment of sequences between IL-4 and IL-13 was also performed as previously reported (20,54). The similarity and identity of ␣-helix A and D of IL-13 to the known structure of IL-4 was in a similar range, as observed for IL-2R␥ and IL-13R␣1 subunits. However, the similarity of ␣-helix B and C could not be reasonably determined (54).
The coordinate of the CRH domain of the IL-4R␣ subunit was also used in our model that was obtained from protein data bank entry 1ILL. The model building and refinement procedures were based on the procedure previously described in detail (55). An initial model was built using the homology module of InsightII (Molecular Simulations Inc., San Diego, CA). Small loops and splices were created and handled such that the energy was kept at minimum for best model. The structures were finally refined using the Discover program (Molecular Simulations Inc., San Diego, CA).

Sequence Alignment of ␣-Helix A between IL-13, IL-4, mGM-
CSF. and hIL-5-When the ␣-helix A of hIL-13, hIL-4, mGM-CSF, and hIL-5 were aligned, glutamic acid residues were found to be aligned perfectly in all of these molecules. In IL-13, this Glu was located at position 13 (Fig. 1A). Helical wheel analysis of IL-13, IL-4, and mGM-CSF suggested that hydrophobic residues clustered in one side of the ␣-helix (Fig. 1B). These hydrophobic residues may be buried in the core of the molecules, as shown for IL-4 and hGM-CSF (56,57). ␣-Helix A of hIL-5 did not show a cluster of hydrophobic residues (data not shown). Further pattern analysis demonstrated that the aligned critical residues are located in the other side of the hydrophobic buried cluster. Based on these analyses and the fact that Glu in IL-4 and mGM-CSF interact with shared receptor subunits, it is predicted that Glu at position 13 in IL-13 molecule is the "hot residue," and it may also interact with shared receptor subunit.
Recombinant Protein Isolation and Purification-Recombinant wtIL-13, IL-13E13K, and IL-13R112D in which the 112th Arg (R) residue of IL-13 molecule was substituted for Asp (D) were expressed in E. coli and purified from inclusion bodies as previously described (51). After purification, each recombinant protein was analyzed using SDS-polyacrylamide gel electrophoresis and stained with Coomassie Blue. Each protein showed a prominent single band at ϳ13 kDa with purity of at least 95% (Fig. 2).
IL-13E13K Blocks Proliferative Activity of IL-13 and IL-4 -TF-1 erythroleukemia cells proliferate in response to IL-13 (51). We, therefore, measured proliferative activity of wtIL-13 and IL-13E13K (Fig. 4A) either alone or in combination of both (Fig.  4, B and C). As expected, wtIL-13 stimulated the growth of TF-1 cells in a concentration-dependent manner (51). In contrast, IL-13E13K did not show any proliferative activity (Fig. 4B). This result indicated that inserting a mutation at position 13 completely suppressed its agonistic activity and that the amino acid residue at position 13 seemed essential for the IL-13-induced proliferation of TF-1 cells. To determine the effect of IL-13E13K on wtIL-13 induced proliferation of TF-1 cells, we cultured cells in the presence of 1 g/ml IL-13E13K and various concentrations of wtIL-13. Interestingly, IL-13E13K blocked the mitogenic activity of wtIL-13 (Fig. 4B). This block of IL-13 mitogenic activity appeared to be concentration-dependent (Fig. 4C). A 100 -333fold excess of IL-13E13K completely neutralized wtIL-13-induced mitogenic activity. Because IL-4 has similar biological activities to IL-13 and receptors for both cytokines share two subunits with other (14,18), we investigated whether IL-13E13K can also suppress the mitogenic response induced by IL-4. As shown in Fig.  4D, IL-13E13K completely neutralized IL-4-induced mitogenic activity.
IL-13E13K Can Neutralize the Down-regulation of CD14 Expression by wtIL-13 on Human Primary Monocytes-IL-13 has been shown to down-regulate CD14 expression on monocytes (23,51). Therefore, we investigated whether IL-13E13K can nullify the down-regulating activity induced by wtIL-13. As shown in Fig. 5, wtIL-13 suppressed CD14 expression on monocytes, and IL-13E13K completely neutralized the effect of wtIL-13. For example, IL-13 decreased the mean channel number (mean fluorescence intensity) in the gated region from 591 to 492 (p Ͻ 0.01). IL-13E13K reversed this effect, and the mean channel number recovered to 600.

DISCUSSION
In this report we describe the production and characterization of a powerful antagonist of IL-13 that was produced by site-directed mutagenesis of an amino acid in the predicted ␣-helix A of the IL-13 molecule. Our results demonstrate conclusively that Glu at position 13 in the IL-13 molecule is of crucial importance for the potency and signal generation. The antagonistic activity of IL-13 mutein IL-13E13K was determined based on (a) 4 -8-fold better displacement of 125 I-IL-13 binding on cancer cells, (b) the inhibition of wtIL-13-induced proliferation of TF-1 cells, (c) neutralization of wtIL-13-induced down-modulation of CD14 expression in primary monocytes, (d) inhibition of wtIL-13-induced activation of STAT-6 in monocytic cell line, and (e) inhibition of cytotoxicity mediated by IL-13PE38QQR. Thus, the antagonistic activities of IL-13E13K were evident in cells that expressed type I, type II, and type III IL-13 receptors.
Interestingly, IL-13E13K not only suppressed IL-13-induced proliferation of TF-1 cells, it also inhibited IL-4-induced mitogenic response in a dose-dependent manner. These results suggest that IL-13E13K may interact with the IL-4R␣ subunit that is shared between IL-13R and IL-4R complexes (35)(36)(37). In that regard, IL-13E13K is similar to muteins of mGM-CSF and hIL-5. The mGM-CSF antagonist E21A and the hIL-5 antagonist E13Q had mutations at the interface with a shared signaling subunit (38 -40). On the other hand, IL-4E9K, in which Glu-9 was substituted by Lys, did not result in an antagonist but was incapable of binding to purified IL-4R␣-shared subunit (47). IL-13E13K is different from hIL-6 and hIL-15 antagonists that have mutations at ␣-helix D or near the C terminus of the molecule. These antagonists were found to be very specific and could not suppress bioactivity of the related cytokines that share a signaling subunit, e.g. hIL-6 antagonists termed DFRD, DFFD, and DFLD did not antagonize human oncostatin M or human leukemia inhibitory factor, which share the gp130 subunit (41). Similarly, the antagonist of hIL-15 did not antagonize hIL-2 activity, which shares the IL-2R␥ subunit with IL-15R (45).
Various studies including the crystal structure of IL-4 and its receptor components demonstrate that IL-4 interacts with IL-4R␣ and IL-2R␥ subunits (47,56). ␣-Helix A of IL-4 molecule interacts with IL-4R␣ subunit, and ␣-helix D may interact with IL-2R␥ subunit. Because the conserved residues in the IL-2R␥ subunit are well aligned with the IL-13R␣1 subunit, we created a model of interaction between the CRH domains of IL-13R␣1, IL-4R␣, and IL-13 (Fig. 8). This model was created based on our hypothesis that IL-13 interacts with IL-13R␣1 and IL-4R␣ sub-FIG. 7. Wild-type and mutant IL-13 block the cytotoxic activity of IL-13PE38QQR on U251 and PM-RCC cells. One thousand cells/well were cultured in Leu-free media containing various concentrations of IL-13-cytotoxin (IL-13PE38QQR) and 1 g/ml wtIL-13 or its muteins (A) or containing 1 ng/ml IL-13PE38QQR with or without various concentration of wtIL-13 or its muteins (B) overnight before the addition of 1 Ci of tritiated Leu. Cells were then incubated for 4 h and harvested, and radioactivity was counted with a ␤ counter. The data are the average of quadruplicate determinations, with the error bars representing the S.D. within a data set. Experiments were repeated several times. units simultaneously. The model suggests that the receptor binding interface of the IL-13 molecule is located (at least) in ␣-helix A and D and that ␣-helix D interacts with IL-13R␣1 subunit, and ␣-helix A may interact with IL-4R␣ subunit. Hot Glu in hIL-5, mGM-CSF, and hIL-4 were proposed to interact with shared receptor subunits (38 -40, 46, 47, 61). Our model suggests that the hot Glu in the IL-13 molecule may also interact with the IL-4R␣ subunit, which is shared between the IL-13/IL-4 receptor system (Fig. 8). Future studies involving a three-dimensional structure of IL-13 and a receptor complex determined by x-ray crystallography will confirm these hypotheses.
The mechanism of antagonistic activity of IL-13 caused by single amino acid substitution is not known. This mutation may cause inappropriate aggregation of receptor subunits or intracellular signaling molecules. Alternatively, a mutation in IL-13 molecule may have created a novel surface in the ligand itself that binds the receptor and alters conformation in a manner incapable of signaling. Future studies will investigate these possibilities.
In conclusion, IL-13E13K is a powerful antagonist that may neutralize the effect of IL-13 in various disease processes such as bronchial asthma, allergic rhinitis, and atopic dermatitis. In addition, since IL-13 is an autocrine growth factor for Hodgkin/ Reed-Sternberg tumor cells, it is possible that IL-13E13K may also play a significant role in the therapy of Hodgkin's disease.