Mapping of Neutralizing Epitopes and the Receptor Binding Site of ...

Vol. 268, No. 13, Issue of May 5. pp. 9771-9779,1993 Printed in U.S.A.

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1993 by The American Society for Biochemistry and Molecular Biology, Inc

Mapping of Neutralizing Epitopes andthe Receptor BindingSite of Human Interleukin 18” (Received for publication, July 20, 1992, and in revised form, January 7, 1993)

Philip L. Simon$$,Vasant Kumarn, Jay S. Lillquist((**,Pradip Bhatnagar$$, RichardEinstein(IQ4, John Lee$, Terence PorterlIl,David GreenlIn, Ganesh Sathell, and Peter R. Young11 I(11 From the Departments of $Immunology, (Physical and Structural Chemistry, (IMolecularGenetics, $$Peptide Chemistry, and TlIProtein Biochemistry, SmithKline Beecham Pharmaceuticals, King of Prussia, Pennsylvania 19406-0939

immunity and inflammation (for reviews, see Oppenheim et al., 1986; Dinarello, 1991) has led to an increased effort to understand the biology and chemistry of this family of related proteins at the molecular level. Because much evidence has suggested that the continued presence of IL-1 may be a key factor in the persistence of chronic inflammatory diseases such as arthritis, there has been intense interest in blocking its activity through the discovery of receptor antagonists. Consequently, much attention hasbeen applied to thediscovery of the structuralfeatures of the proteins which allowthem to bind to their receptors on target cells and produce a biological effect. These include direct structural analysis by circular dichroism (Wingfield et al., 1987; Meyers et al., 1987), x-ray diffraction (Priestle et al., 1988,1989; Finzel et al., 1989; Graves et al., 1990; van Oostrum et al., 1991), three- and fourdimensional NMR (Clore et al., 1991; Stockman et al., 1992), mutational analysis (for review, see Cloreet al. (1991)),chemical (Meyers et al., 1987; Wingfield et al., 1989; Yem et al., 1989; Chollet et al., 1990), and biological (Livi et al., 1990, 1991) modification of the proteins. However, despite all of these studies we are far from a full understanding of the regions of the moleculewhich are required for biological activity, and many discrepancies remain. There are three related IL-1s that have been identified so far, two with biological activity (IL-la and IL-16) (Lomedico et al., 1984; Auron et al., 1984; March et al., 1985) and one a natural antagonist (IL-lra or IRAP) (Hannum et al., 1990; Eisenberg et al., 1990; Carter et al., 1990). Allof these bind with more or less similar affinity to thetwo types of receptor so far identified. These are the type I receptor found on T cells and fibroblasts (Sims et al., 1988), andthe type I1 receptor found on B cells and macrophages (McMahan et al., 1991). The ligands all share sequence (Eisenberg et al., 1991) and structural (Graves et al., 1990, Stockman et al., 1992) The identification of IL-1’ as one of the key mediators of similarities. Although significant, the sequence similarities among the different ligands are weak (typically only 20-30% * The costs of publication of this article were defrayed in part by amino acid identity, most of which is internal to the structhe payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 tures), so that the receptor binding region is not readily identified by these criteria (Priestleet al., 1989).The receptors solely to indicate this fact. 5 Present address: CIBA-GEIGY Pharmaceuticals RESlO6, 556 are also structurally similar, both consisting of an extracelMorris Ave., Summit, NJ 07901. lular ligand binding region with three immunoglobulin-like ** Present address: Central Research Division, Pfizer Inc., Groton, domains (Sims et al., 1988; McMahan et al., 1991). Both CT 06430. and receptors appear to be part of a family with similar $5 Present address: Dept. of Molecular Biology, Rhone-Poulenc ligands structural motifs which presently include both the IL-1s and PA Rorer Central Research NW14, 500 ArcolaRd.,Collegeville, the fibroblast growth factors (for a review, see Young(1992)). 19426. 11 11 To whom correspondence should be addressed Dept. of MolecUsing a combination of peptide antisera and monoclonal ular Genetics, SmithKline Beecham Pharmaceuticals, P. 0. Box 1539, antibody epitope mapping data, we have been able to define King of Prussia, PA 19406-0939. a face of IL-lP which binds neutralizing antibodies. Mutagen’ The abbreviations used are: IL-1, interleukin-1; IL-lra, interleu- esis of some of the residues in this face indicated that it kin 1 receptor antagonist; ELISA, enzyme-linked immunoabsorbent assay; mAb, monoclonal antibody; PBS, phosphate-buffered saline; overlapped with regions required for receptor binding and Fmoc, N-(9-fluorenyl)methyloxycarbonyl;TOCSY, total correlation biological activity. One mutation led to a protein with reduced spectroscopy; NOESY, nuclear Overhauser effect spectroscopy. bioactivity but wild type receptor binding, suggesting that a

Antibodies to synthetic peptides of human interleukin 18 (IL-1B) and to recombinant human IL-10 were used to identify epitopes of IL-1@associated with the neutralization of its biological activity. Analysis of antisera raised to 17 synthetic peptides derived from the mature IL-18 sequence showed that five regions and 120-133) (residues 6-15,49-80,58-80,92-101, were both immunoprecipitating and neutralizing. Using a hexamer epitope mapping method, comparison of the regions recognized by four neutralizing rabbit antisera with those recognized by a rabbit antiserum raised to denatured IL-1s suggested two further neutralizing epitopes, residues 39-48 and83-95. Finally, a neutralizingmonoclonal antibody wasshown to bind to the peptides 6-1 1and 87-95 by peptide binding and predomimutagenesis. All of theseregionsappear nantly on one face of IL- 18. The effect of mutations in residues 4-11 and 88-97, which lie within this face, on receptor binding andbiological activity was determined.Most of the mutationstested affected both receptor binding and activity, whereas mutations in another face of IL-lB (residues 74-80) had no effect. Purification of two of the mutants with reduced bioactivity and receptor binding and analysis by two-dimensional NMR indicated no gross changes in tertiary structure. A third mutant had reduced bioactivity in two differentbioassays but no change in receptor binding. Although two-dimensional NMR revealed no gross changes in conformation,small changes did occur at a site distal from that mutated. The data are consistent with other epitope mapping and receptor bindingmutagenesis data and suggest that the neutralizing antibodies and receptor recognize different but overlapping regions of IL-la.

9771

9772

Structure-Activity of IL-lp

second region is required for signal transduction. This latter mutation does not necessarily define a specific region on IL10 responsible for signal transduction, since two-dimensional NMR indicated small but significant structural changes at a regions distant from the mutation. EXPERIMENTALPROCEDURES

Mutagenesis of HumanIL-lp-The cloning and expression of human IL-lp was described previously (Meyers et al., 1987). The specific biological activity of IL-lp is approximately 5 X lo8 units/ mg of protein based on the EL-4 bioassay (see below). To create sitespecific mutants of human IL-lp, the vector used for expression of mature 17-kDa IL-lP in Escherichia coli, pMGILl0 (Meyers et al., 1987),was altered in that the EagI site found 3' to the IL-1p coding region was changed to an EcoRI site using an oligonucleotide linker. The IL-lp coding regionwasmoved from this vector to a second related vector, pMGNco, which has a unique NcoI restriction site at the startof translation insteadof the NdeI site inpMGILlp (Lillquist et al., 1988). This NdeI site is duplicated in the IL-16 coding region and is therefore less convenient for further manipulations of the entire coding region. Specifically, a 691-base pair fragment containing the coding region of IL-lp starting atamino acid 1 of mature IL-lp (alanine) was removed from pMGILlp via a Ban1 digest and, following blunt ending with Klenow, was ligated into pMGNco cut with NcoI and EcoRV, phosphatased, and flush ended with Klenow. The final vector, pMGNcoILlp, expresses mature IL-1p at levels comparable topMGILlp.Eitherthe HindIII/EcoRI fragment from pMGILlB, which lacks the region of DNA encoding the first 13 amino acids of IL-lp, or the NsiI/BamHI fragment from pMGNcoILl0 was cloned into pUCl9 a t the corresponding HindIIJ/EcoRI or PstI/ BamHI sites, respectively. The IL-lpsequences were then mutagenized by the Kunkel procedure (Kunkel, 1985; Kunkel et al., 1987) using site-specific mixed oligonucleotides and theBio-Rad Mutagene Kit. The synthetic oligonucleotides employed for mutagenesis were of two kinds. The first kind, used for the individual mutations inresidues 4-10 and 88-97, was synthesized as a mixture of 19-mers in which the central codon was substituted in the first two positions with an equimolar mixture of the non-wild-type sequence (Goff et al., 1987). The second kind, used for mutating residues 76-80, was synthesized as if it were the wild-type sequence, but atseveral residues a low level of the three non-wild-type nucleotides was included in the reaction. In the present case a 37-mer was synthesized in which the first and second positions of seven codons included non-wild-type nucleotides at a concentration one-fourteenththat of the wild-type nucleotide at each position (Hutchinson et al., 1986). Deoxyoligonucleotideswere synthesized in an Applied Biosystems model 380A DNA synthesizer utilizing phosphoramidate chemistry (Gait, 1984). After transformation of the host JM101, the plaques were pooled by scraping and grown up in Luria Broth for the preparation of plasmid by standard methods (Maniatiset al., 1982). A HindIIIl EcoRI or NcoI/BanHI restriction fragment containing a mixture of mutant and wild-type IL-lp coding regions was then isolated from this plasmid preparation by gel electrophoresis and ligated directly into the equivalent sites of PMGlLlD or pMGNcoILl@prior to transformation of the expression strain AR58, which contains the temperature-sensitive cIa7 repressor (Meyers et al., 1987; Lillquist et al., 1988). Individual colonies were then assessed for expression of IL-1p after thermalinduction via Western blot (Lillquist et al., 1988). A rabbit polyclonal antiserum to mature IL-lp was used to detect the IL-lp in the soluble fraction of E. coli lysates, and levelswere quantitated relative to known concentrations of wild-type IL-lp using densitometry and/or scintillation counting. DNA from uninduced colonies was sequenced with the Sequenase system (U. S. Biochemical Corp.) using synthetic oligonucleotide primers within IL-lp or the adjacent noncoding sequences. IL-lp Peptides-All peptides were synthesized by solid phase methods, using either a Beckman model 990, Applied Biosystems model 430A, or a Vega model 250 synthesizer. Synthesis was performed on phenylacetomidomethyl, hydroxymethyl, or benzhydrylamine polystyrene resins (Applied Biosystems), using the t-butoxycarbonyl group-based chemistry. Appropriately protected amino acids were purchased from Protein Research Institute, Osaka, Japan or from Bachem. The peptides were deprotected and cleaved from the resin using hydrogen fluoride and anisole. When the peptide sequence contained Met, Cys, or Trp then the"low-high" deprotection method was used (Tam andMerrifield, 1987).All peptides were purified using reverse phase high performance liquid chromatography on either C-

4 or (2-18 columns purchased from Vyadec. Molecular weights were determined using fast atom bombardment mass spectrometry. For the production of antipeptide antibodies, peptides were conjugated to keyhole limpet hemocyanin. 15 mgof carrier protein was dissolved in 5 ml of phosphate-buffered saline, pH 4, and incubated with 20 mg of water-soluble carbodiimide (l-ethyl-3(3-dimethlyaminopropy1)carbodiimide)at 0 "C for 15 min. 5 mg of peptide was added to thesolution and the pH adjusted to 9 using sodium carbonate. The mixture was allowed to stir overnight and was then dialyzed against distilled water. The retentate was lyophilized and reconstituted at the time of immunization. For epitope scanning (Geysen et al., 1984, 1986), the plates, amino acid derivatives, and the software for the design of peptide synthesis were from CRB England. Hexameric peptides were synthesized on polystyrene pins coated with polyacrylate using Fmoc-protected pentafluorophenol ester derivatives of amino acids. The coupling was monitored using Kaiser's test. The Fmoc group was removed using 40% piperidine in dimethylformamide, and side chain protecting groups were removed by trifluoroacetic acid containing 1%phenol and 10% mercaptoethanol. Several pins were analyzed for amino acid composition and were found to contain the expected quantities and molar ratios of amino acids. Antibodies to IL-1 and IL-1 Peptides-Polyclonal antibodies to recombinant human IL-lp and peptide/keyhole limpet hemocyanin conjugates were prepared in rabbits. Preimmunization sera were collected, and then each rabbit received subcutaneous injections into four sites on the back with a total of 1 ml of a 1:l mixture of 0.1 mg/ ml antigen and complete Freund's adjuvant. The rabbits were boosted with half the quantity of antigen mixed with incomplete Freund's adjuvant at 2-3-week intervals. Test bleeds were performed 7-10 days following booster injections and were monitored either by neutralization of biological activity (EL-4 assay, see below)or by ELISA (see below). For some experiments, the IgG fractions of the antisera were obtained using either the Beckman protein A antibody purification kit (Beckman Instruments) or by fast protein liquid chromatography using the protein A-Superose HR 10/2 column (Pharmacia LKB Biotechnology Inc.).Proteinconcentrations for purified antibody preparations were determined with the Bio-Rad protein assay kit, using human Ig as a standard. A neutralizing monoclonal antibody (which we designated as CISmAb) was purchased from Cistron Biotechnology, Pine Brook, NJ and has been characterized previously (Gaffney et al., 1987). Immunoassays-ELISAs were performed using Immunolon I1 plates (Dynatech, Alexandria, VA). Usually 100 ml of 50-100 ng/ml antigen, dissolved in 0.1 M bicarbonate buffer, pH 8.9, was used to coat the assay wells overnight at 4 "C. The plates were washed three times with 10% gelatin in PBS (0.137 M NaCl, 2.7 mMKC1, 1.3 mM KHzP04,9.5 mM Na2HPO4, pH 7.0) (gel/PBS) and then incubated for 1 h with the primary antibody diluted in gel/PBS. Following three washes with 0.05% Tween 20 in PBS (Tween/PBS), the plates were incubated for 1 h a t room temperature with the appropriate biotinconjugated secondary antibody (Vector Laboratories, Burlingame, CA) diluted in gel/PBS. Following three washes with Tween/PBS, the plates were incubated for 1 h a t room temperature with a 1:2,000 dilution of streptavidin-conjugated alkaline phosphatase (Bethesda Research Laboratories) in gel/PBS. Following three washes with Tween/PBS, alkaline phosphate substrate solution (p-nitrophenyl phosphate; Sigma) diluted in 10% diethanolamine, 0.5 mM MgClz, pH 9.8, was added to each well, and the plates were incubated in the dark for 15-45 min. Absorbance at 405 nM was measured using an automated plate reader (Bio-Tek Instruments, Winooski, VT). For the hexamer mapping method of Geysen et al. (1984, 1986),the pins were first blocked for 1 h with a supermixture of 1% ovalbumin, 1% bovine serum albumin, and 0.1% Tween 20 in PBS. Incubationswith primary antibody were performed overnight, with the antibody diluted in supermixture. The plates were then processed in the identical fashion as the ELISA starting with the secondary antibody incubation step. During incubations, the plates were agitated using a rotary shaker. Following each assay, the plates were stripped of adherent antibodies by sonication of the pins in 1%SDS, 0.1% 2-mercaptoethanol, 0.1 M sodium dihydrogen orthophosphate, pH 7, at 60 "C for 60 min. The pins were then washed with two changes of 60 "C distilled water, followed by a bath inboiling methanol for 2 min, and then air dried. Immunoprecipitations were performed at room temperature in 96well flat bottom tissue culture plates. Dilutions of antisera in PBS were combined with radiolabeled IL-lP in PBS containing 10% fetal bovine serum to a final volume of 100 ml. Followingincubation for 2 h, 100 ml of a 40% solution of polyethylene glycol 8000 (Sigma) was

Structure-Activity of IL-1P added to each well, and the plates were incubated on ice for 30 min. The precipitated antibody-ligand complexes were separated from free ligand by filtration onto glass fiber filters presoaked in 0.1% polyethyleneamine using a Brandel cell harvestor (Brandel Instruments, Rockville, MD). The radioactivity on the cut filters was determined using a y-scintillation spectrometer (Beckman Instruments). Receptor Binding Assay for IL-lfl-The radioreceptor binding assay for IL-lp, using EL-4 cell-derived membrane and '251-labeledIL-10, was performed as described previously (Lillquist et al., 1988). Radiolabeled IL-10 prepared by the IODO-GEN method had specific activities of500-1,000Ci/mM. For binding experiments using either EL-4 membrane fractions or solubilized EL-4 membrane, the radiolabeled IL-10 was further purified by fast protein liquid chromatography hydrophobic interaction chromatography using an HR 5/5 phenyl-Superose column (Pharmacia). Sample volumes of 100-200 pl of radiolabeled IL-10 in 1.7 M ammonium sulfate, pH 7, were applied to the column equilibrated in the same buffer. Proteins were eluted with a decreasing gradient of ammonium sulfate using 0.1 M sodium phosphate as the elution buffer. This procedure routinely separated unlabeled from monoiodinated IL-10 and thus increased the specific activity to approximately 2,200Ci/mM. Values of K . for receptorligand interactions were calculated using the program LIGAND (Munson and Rodbard, 1980). Biological Assays for ZL-I-The EL-4 cell bioassay for IL-1 was performed as described previously (Simon et al., 1985). The induction of prostaglandin E, in HeLa cells was measured as follows. HeLa cells were plated on 24-well dishes at 8.5 X IO4 cells/well. The cells were grown overnight prior to the assay. The growth medium was removed, the cells rinsed with PBS, and200 ml ofDulbecco's modified Eagle's medium (GIBCO) containing 0.1% fetal bovine serum and the respective concentration of agonist was applied to thecells. After a 6-h incubation, 100 mlof medium was removed and assayed for prostaglandin E, by radioimmunoassay (Amersham Corp.) according to the manufacturer's directions. Control samples containing no IL14 were processed in parallel to determine background levels of prostaglandin E,. Purification of IL-16 Mutants-Purification of Arz4 -+ Val, Leu' + Ala; and Thr' + 'Gly IL-10 followed the previously published procedure for wild-type IL-10 (Meyers et al., 1987) with some minor modifications. Yields and elution conditions for the mutants were similar to wild-type IL-10, and the proteins behaved identically on size exclusion chromatography. Amino-terminal sequencing of the Thrg + Gly mutant indicated a similar distribution of amino termini to wild-type IL-lfl (Meyers et al., 1987). Stability of wild-type and Thrg + Gly IL-10 was assessed by incubating 500 ng of each with target EL-4 cells for 0 and 16 h, harvesting supernatants, anddetermining IL-10 levels via SDS-polyacrylamide gel electrophoresis and Western blotting, followed by incubation with a rabbit anti-human IL-10 antiserum as described above. No degradation of either protein occurred during the 16-h incubation time. Two-dimensional NMR--IL-1/3 samples were concentrated and exchanged into 100 mM deuterated sodium acetate buffer, pH 5.4, by ultrafiltration. The final concentration of the protein was about 2 mM. All NMR experiments were carried out on a Bruker AM600 or AMX500 in the 25-35 "Ctemperature range. Double quantum filtered COSY (Rance etal., 1983),TOCSY (Braunschweiler and Ernst,1983; Davis and Bax, 1985; Griesenger et al., 1988) and NOESY experiments were performed using standard pulse sequences. Mixing times in NOESY and TOCSY experiments were 100-150 ms and 50 ms, respectively. The SCUBA (stimulated cross-peaks under bleached alphas) technique (Brown et al., 1988), with a total delay of 60 ms, was used to recover a-protons under the solvent resonance. FTNMR and FELIX programs (Hare Research Inc.) were used for off-line data processing.

9773 TABLEI

Summary of binding activities of antipeptide 1gG Peptide antiserum

Immunizing peptide"

IL-lB

1-10 6-15 11-27 27-37 33-57 47-55 49-80 58-80 74-88 80-101 92-101 96-105 120-133 140-153

-

+

+ + + + + + + + + +

+ + +

+/-

-

+-

+-

+ + + + +

Immunoprecipitation'

Inhibitiond

cpm

%

100' 73 f 18 16 f 1 7 f 5 32 f 8 21 f 5 62 f 14 39 f 22 3-t3 17 f 13 77 f 1 2 7 f 7 54 19 54'

28 f 27 82 f 12 7 f 7 11 t 11 0 0 69 & 3 64 f 11 0 13 -t 13 86 f 6 23 f 23 69 t 9 0'

*

ELISA reactivity of the antibodies prepared against synthetic IL-

l p peptides to the same peptides used to prepare the antisera. Two

peptides did not yield antibodies with reactivity to theoriginal peptide or to IL-10 (peptides 23-33 and 125-142). ELISA reactivity of the antipeptide sera to rIL-lP. Immunoprecipitation of lZ5I-IL-lpby the antipeptide sera. 20 pl ofIgG (3 mg/ml) and 20 p1 of phenyl-Sepharose-purified '251-IL-10 (11,000 cpm = 50 PM final concentration) were incubated for 2 h at room temperature. The immunoprecipitates were collected and counted as described under "Experimental Procedures." Because the absolute amount of IL-10 precipitated varied from experiment to experiment, the results have been expressed as a percentage of the cpm immunoprecipitated by the 1-10 antiserum. The average of two experiments is shown. Neutralizing activity of the protein A-purified antipeptide antiM antibody and 1 X 10"' M rIL-10. The bodies tested at 1 X results of two experiments have been averaged. e Only one set of data generated.

tively. Immunoblot analysis using the same antisera generally produced the same results as the ELISAs (data not shown). Thirteen of 14 antisera showed ELISA reactivity to the immunizing peptide, and nine showed reactivity to IL-1p. In one instance (peptide 1-10) reactivity to IL-lp was seen in the absence of reactivity to the peptide. Two peptides produced no peptide or IL-1P reactive antisera (23-33 and 112-127). The capacity of the proteinA-purified antipeptide antibodies to immunoprecipitate radiolabeled IL-16 is shown in the fourth column of Table I. Antibodies to peptides 1-10, 6-15, 33-57, 49-80, 58-80, 92-101, 120-133, and 140-153 precipitated IL-10 to differing extents, indicative of recognition of solution phase IL-1p. Antibodies to peptides 27-37, 80-101, and 96-105 bound IL-lP in the ELISA but demonstrated only a weak ability to immunoprecipitate themolecule. Antibodies to peptides 33-57 and 58-80 bound IL-lP in solution but not in the solid phase ELISA, suggesting that the appropriate surface residues were buried or altered in conformationwhen IL-lp was bound to an ELISAplate. WhentestedintheEL-4IL-1 bioassay, five synthetic peptidescorrespondingto four nonoverlapping regions inRESULTS duced antibodies that neutralized IL-lp biological activity by 58-80, 92-101, and more than 40%: peptides 6-15, 49-80, Neutralizing Epitopes Defined by Antisera to Synthetic IL1p Peptides-We used a panel of rabbit antisera to 16 syn- 120-133 (Table I, fifth column). Two other peptide antibodies thetic peptides encompassing the entire aminoacid sequence that overlap these five gave variable results (1-10 and 96of the mature IL-1pmolecule to identify the surface epitopes 105). None of the antibodies inhibited IL-la activity or IL-2 of IL-1P by (a) solid phase binding of the antibodies to IL-lP activity, indicating that the inhibitionwas specific for IL-lp. 140-153, those or IL-10 peptides (ELISA),(b) immunoprecipitation of radi- With theexception of the antiserum to peptide olabeled IL-10, and (c) neutralization of the biological activity sera that were strongly positive in the immunoprecipitation assay were inhibitoryinthe biological activity assay. The of IL-10 in the EL-4 bioassay. The results of several ELISAs are summarized in second the antiserum to the 120-133 peptide cross-reacted with the 49and third columnsof Table I, which present the reactivityof 80 peptide andvice versa. This may be because of recognition the antisera to the immunizing peptide and to IL-lP, respec- of sequences in common between the two peptides (i.e. resi-

Structure-Activity of IL-lp

9774

dues 57-61 are Pro-Val-Ala-Leu and residues 130-133 are Pro-Val-Phe-Leu). Neutralizing Epitopes of IL-lp Defined by Monoclonal and Polyclonal Antibodies Prepared against Mature IL-Ip-We used the epitope mapping methodof Geysen et al. (1984,1986) to map theresidues recognized by monoclonal and polyclonal antibodies raised to IL-lP or IL-lp peptides. In this method, each of the possible 148 hexamers of the mature IL-lpmolecule was synthesized separately and tested for its binding of antibodies to IL-1p. We confirmed the veracity of this method with the neutralizing antipeptide antiserum to peptide 6-15, which bound as expected to the 10-15 hexamer of IL-lp (Fig.

,750

A

1

,650

,550 ,450

,350 ,250 ,150

.050 ,450

B

,350

1).

Five different rabbit polyclonal antisera to human recombinant IL-lp, four neutralizing (panels A - f ) ) and one nonneutralizing (panel E ) were similarly mapped (Fig. 2). Prebleed sera and antisera to IL-la did not demonstrate any activity to hexamers of IL-1p by this method. At least three out of four of the neutralizing antisera tested recognized six major epitopes (residues 1-6,39-48,83-95,115-121,122-135, and 133-142) and oneminorepitope (65-71) withslight differences in the particularresidues recognized, particularly in the carboxyl terminus. Several of these epitopes were also recognized by anonneutralizing antiserum raised against biologically inactive SDS-polyacrylamide gel electrophoresispurified recombinant IL-lp (Meyers et al., 1987), with the clear exception of the 39-48 and 85-93 epitopes (Fig. 2 E ) . The hexamer mappingof a neutralizingmurine monoclonal antibody (CIS-mAb) to IL-lp(Fig. 3) suggested two separate and discontinuousepitopes inthe IL-1/3 linear sequence, residues 6-11 and 87-95. Epitope Mapping of CIS-mAb by Mutagenesis-To confirm the epitopes determinedby hexamer peptide mapping,we also monitored the binding of CIS-mAb to various site-specific mutant forms of IL-1p created in these regions. We have shown previously that wild-type IL-1p can be isolated in a

.250 ,150 ,050

.400

D

,300 ,200

0.30,100

0

0.26-

0.30 0.26

0.22-

0.22 0.18 0.14

0.1 8-

0.10

A B

0.06

0.14-

0.02 10

0.10-

0.D6

1

I

I I

RESIDUE FIG. 1. Epitope mapping of a rabbit antiserum prepared with IL-lB synthetic peptide 6-15. The whole antiserum was diluted 1:500 in supermixture, and the assay was performed as described under “Experimental Procedures.” The results represent the mean absorbance of duplicate peptide pins.

30

50

70

90

110

130 150

RESIDUE FIG. 2. Epitope mapping of five different rabbit antisera to recombinant human IL-ID. The whole antisera were diluted 1:5,000 in supermixture and the assay was performed as in Fig. 1. Panels A-D are the assay of four different neutralizing rabbit antisera, each from a different rabbit panel A , R675; panel B , S32; panel C, R486; panel D, R261. Panels D and E are the results from a neutralizing and nonneutralizing serum, respectively, derived from the same rabbit (R261).

fully active form from the soluble fraction of E. coli lysates (Meyers etal., 1987; Lillquist etal., 1988). Typically we obtained as much as 50-100 pg/ml (or >lo7 units/ml) of cleared lysate, which allowed us to screen the activity of various mutants of IL-Ip directly withoutfurther purification. IL-16 levels were estimated by comparison withknown quan-

Structure-Actiuity of IL-IP

9775

RESIDUE

FIG. 3. Epitope mapping of a neutralizing mouse monoclonal antibody to human recombinant IL-18 (CIS-mAb),performed as in Fig. 1.

TABLE I1 Reactivity of CIS-mAb to mutations of ZL-1p Residue Mutation IC500 w/ml

90

Wild type SeP 510 >3,000 Leu6 Asn'

155 Arg GlY Glu Ala

670

CYS 380 GlY >3,000 Trp 1,450 Gln 1,250 Glu 860 Lys@ Val 120 150 GlY Leu 390 Asnm Phe 120 180 Arg 130 GlY 130 CYS Tyr* 1,660 Arg Val 1,150 Ser 520 Leu 1,320 a IC50is the concentration of CIS-mAb required to inhibit 50% of the activity of 1unit/rnl IL-lp in the EL-4 bioassay. Thrg

tities of IL-10 via Western blot. We measured the binding of the CIS-mAb to IL-10 via neutralization of biological activity (Table 11). Several mutations in the two epitope regions analyzed, particularly residues 6, 9, and 90, lowered the ICso for neutralization of CIS-mAb to IL-10, confirming the role of these regions in antibody recognition. IL-10 mutants containing changes in residues outside of the two areas (for example residues 74-80) did not affect neutralization (data not shown). Effect of Mutations on Biological Actiuityand Receptor Binding-In examining the peptide regions ofLL-10 which bind the various neutralizing antibodies described above, it was

'.

:

II

FIG. 4. Position of neutralizing epitopes on the three-dimensional structure of IL-18.Residues in magenta represent the regions 1-11, 58-71, and 87-95. For orientation, several residues are identified on this face. The line of white residues inserted into the region defined by the magenta residues corresponds to part of the 3948 epitope. The open end of the barrel is to theright of the neutralizing epitope.

observed that most of them could befound predominantly on one face of the molecule as illustrated in Fig. 4, whichincludes residues 1-11, 58-71, and 87-95. This face is adjacent to the open end of the 0 barrel and also includes the polyclonal epitope at residues 39-48. The ability of many antibodies

Structure-Activity of IL-lp

9776

binding to this region to block biological activity suggests that it represents a region that overlaps with the receptor binding site. We therefore examined the IL-1@ mutants for biological activity and IL-1 type I receptor binding on murine EL-4 cells (Table 111).Three categories of mutants could be found. First, there were thosein which activityandreceptorbinding changed in parallel. The most dramatic reductionswere seen at Arg4, Leu', Thrg, Leu", Lysg3,Lysg4,and Metg5. Less dramatic reductions include Ser5, Cys', Tyrg', and Glug6.On the other hand we saw 3- and 5-fold increases in both activity and binding, respectively, a t Lysg2+ Arg. Second, there were mutations that resulted in significant reductions in activity relative to receptor binding. This was most dramatically illustrated by Thrg + Gly, and two other changes at this residue showed the effect to a lesser extent. The Leu' + Ala change also fell into this class. The third category of mutations had relatively little effect on activity and receptorbinding. These were illustrated by multiple TABLEI11 Activities of IL-Ip mutants Site-specific mutagenesis Residue

Mutation

Arg4

Ser' Leu' Asn' cyss Thrg

Leu"

LysM AsnBg

TyrW

Lys92 Lys93 Lys" Metg5 G1uS Lys97

19 vi1 ND, not determined.

Specific activity

Receptor binding K .

% wild-type

% wild-type

12 57 47 4 7 120 29 31 5 16 61 27 0.4 47 20 0.5 0.5 0.01 0.2 0.7 0.1 0.2 120 84 94 112 90 131 120 41 47 43 41 76 306 9 123 47 39 19 7 70 22 84 102 35 37

9 36

ND" 23

ND 208 32 40 60 25 25

ND 100 208 145 13 0.3