Development of a Human Interleukin-6 Receptor Antagonist”

Neutralizing monoclonal antibodies specific for hu- man interleukin-6 fIL-6) bind two distinct sites on the IL6 protein (sites I and X). Their interference with IL-6 receptor binding suggested that site I is a receptor-bind-ing site of KL-6, whereas site I1 is important for signal transduction. Mutagenesis of site I1 could therefore result in the isolation of IL-6 receptor antagonists. To test this hypothesis, a panel of IL-6 mutant proteins was constructed that did not bind to a site 11-specific mono- clonal antibody. One such site 11 mutant protein (with double su~titution of Gln-180 with Glu and Thr-183 with Pro) was found to be an antagonist of human IL-6. It was

Interleukin-6 (IL-6)1 is a multifixnctional cytokine playing a central role in host defense mechanisms (for reviews, see Refs. [1][2][3]. IL-6 exerts its multiple activities through interaction with specific receptors on the surface of target cells (4,5). The c D N h for two receptor chains have been cloned and code for transmembrane glycoproteins of 80 and 130 kDa (gp130) (6,7), which both belong to the large cytokine receptor superfamily (8)(9)(10). The 80-kDa IL-6 receptor (IL-6R) binds IL-6 with low affinity ( & --1 nM) without triggering a signal (11). The IL-6.80-kDa EL-6R complex s~bsequent~y associates with gp130, which then transduces the signal (7, 111. gp130 itself has no afEnity for IL-6 in solution, but stabilizes the IL-6.80-kDa IL-6R complex on the membrane, resulting in high affinity binding of IL-6 (& -10 PM) (7). Recently, it was found that gp130 is also a constituent of the receptors for leukemia inhibitory factor, oncostatin M, and ciliary neurotrophic factor (for erlands Foundation for Fundamental Research. The costs of publication * This work was supported by a grant (to J. P. J. B.) from the Nethof this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked ~a d v e~~e~n t~ in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
In a variety of human inflammatory, autoimmune, and neoplastic diseases, abnormal IL-6 production is obsewed and has been suggested to play a role in the pathogenesis (reviewed in Ref. 18). A causative role for IL-6 in the pathogenesis of multiple myeloma was indeed demonstrated by the observation that a d~i s t r a t i o n of an anti-IL-6 antibody to a patient with plasma cell leukemia could block myeloma cell prol~feration in the bone m~o w (19). Thus, i~i b i t o~s of IL-6 biological activity are useful to study i t s role in disease and could have broad therapeutic applications.
One strategy to neutralize IL-6 activity could be by inhibition of the ligandfreceptor interaction with specific receptor antagonista. The feasibility of such an approach was recently demonstrated with a natural occurring receptor antagonist for interleukin-1 (for review, see Ref. 21;Ref. 201, which is currently being tested for utitity in a variety of diseases 122). For IL-6, however, no natural receptor antagonist has been identified so far.
We speculated that one of these sites might be involved in binding to the 80-kDa IL-6R, whereas the other might be involved in an interaction with gp130. According to the IL-6R model, IL6 variants that bind normally to the 80-kDa IL-6R, but are somehow defective in inducing the interaction of the IL-6.80-kDa IL-6R complex with gp130, might be able to prevent heterodimerization of the receptor and function as receptor antagonists.
In this paper, we investigated the roles of sites I and 11 in IL-6Ireceptor interaction. We now show evidence for a role of site I in binding to the 80-kDa IL-6R and for site I1 in signal transduction because a biologically inactive hIL-6 mutant protein with mutations in site If could bind to the 80-kDa hIL-GR, but an~gonized the biolo~cal activity of wild-type hIL-6. Binding experiments suggest that signal ~ansduction cannot occur because the complex of the mutant protein and the 80-kRa IL-6R cannot associate with gp130. This is the first demonstration that it is possible to separate receptor binding from biological activity of hIL-6.

Antibodies and Cytokines
The production and purification of the ILS-specific mAbs have been described in detail (23). mAb B1 iLN1-73-10> was a kind gift of Dr. F. Di Padova (Sandoz Pharma, Preclinical Research, Basel, Switzer~and). The purified wild-type rhIL-6 preparation used throughout these experiments as a standard is derived from ~s c~r~c h~ coli carrying the HGF7 plasmid (24). tion on of rhIL-6.HGF7 has been described (23). The specific activity of purified rhIL-6.HGF7 as determined in the mouse B9 assay is -loe ~~~. Recombinant IFN-r was a kind gift from Genentech (San Francisco).

Expression Vectors and Bacterial Strains
Construction of the expression vector pUK-IL-6 has been described (25). For expression of rhIL-6 or rhIL-6 mutant proteins with this vector, E. coli DH5a (Life "mologies, Inc.) was used as the host. The bacteriophage T7 promoter vector pET8c and expression strain E. coli BL21(DE3) (26) were a kind gift of Dr. G. h y n (Department of Biochemistry, University of Nijmegen, Nijmegen, The Netherlands).
Expression Library Construction a n d Recombinant DNA and

Sequencing Protocols
The vector pUK-IL-6 was used for construction of the library of rhIL-6 mutant proteins with randomly distributed substitutions of Gln-153-Thr-163. Construction of the library has been described in detail (27). Following transformation to E. coli DH5n, -1000 colonies were obtained. DNA manipulation procedures were performed as described (23,25). Nucleotide sequences of seleded mutants (see below) were obtained with cDNA-derived oligonucleotide primers on doublestranded DNA by using the Sequenase kit (United States Biochemical Corp.).
Preparation of E. coli Extracts for Library Screening with mAbs Four-hundred ampicillin-resistant colonies from the expression library were transferred to wells of 96-well flat-bottom microtiter plates (Nunc) containing 100 pl of LC amp medium (10 g of Bacto-Tryptone, 5 g of yeast extract, 8 g of NaCl, 2 ml of Tris baselliter supplemented with 100 p g / d ampicillin (final concentration)). Following overnight culture at 37 "C, baderia were lysed by addition of lysozyme to 1 m g / d and further incubation for 30 min at 37 "C. One in 10 dilutions of the crude extracts in phosphate-buffered saline, 0.02% Tween 20, 0.2% gelatin was directly tested for reactivity in sandwich enzyme-linked immunosorbent assays with site I-specific mAb CLB.IL-6/8 (mAbS(1)) or site 11-specific mAb CLB.ILfY16 (mAblG(I1)) coated on the plastic and biotinylated polyclonal goat anti-rhIL-6 as detecting antibody as described that bound to mAb&I), but not to mAblMI1). (23,27). The nucleotide sequences were determined for mutant proteins Preparation and Quantification of E. coli Extracts for Biological

Activity Measurements
To measure the biological activity of the mutant proteins that bound to mAb&I), but not to mAb16(II), overnight cultures of E. coli DH5a carrying the mutant constructs were diluted 1:50 in 250 ml of LC amp medium and subsequently cultured to an absorbance at 550 nm of 1.5. Bacteria were harvested by centrifugation, resuspended in 5 ml of lysis buffer (phosphate-buffered saline, 1% Tween 20, 10 nm EDTA, 2 nm phenylmethylsulfonyl fluoride), and lysed by sonication. To solubilize rhIL-&containing inclusion bodies, SDS was subsequently added to 1%. Mer 1 h of incubation at room temperature, SDS-insoluble material was removed by centrifugation (13,000 x g for 15 min). The biological activity of this SDS-solubilized material was directly measured in the m o w B9 and CESS assays starting from a 1:lOOO dilution. At this dilution, the SDS did not affect the bioassays used. The IL-6 variant protein concentration of these preparations was determined by a competitive inhibition radioimmunoassay with IL-&specific mAb CLBJL-6/7 coupled to Sepharose 4B (Pharmacia LKB, Uppsala) and 1ZsII-rhIL-6 in the presence of 0.1% SDS. Unlabeled rhIL-6 served as a standard. mAb CLB.ILW7 binds heat-and SDS-denatured IL-6 and recognizes IG6 residues Thr-143-Ala-146 as determined by pepscan analysis (28, 29k2 Expression a n d Purification of Mutant Proteins T163P and Q16OE,T163P from E. coli The IG6 cDNA inserts from the vedors puK-ILS T163P and pUK-IL-6 Q160E,T163P were subcloned in the vector PET&, and the plasmids were transformed to E. coli BL21fDE3) for expression (26). The rhIL-6 variants were subs~uently purified essentially as described (30). Briefly, the proteins were prepared from inclwion bodies by extraction with 6 M guanidine HC1, renaturation by dialysis against 25 m~ Tris (pH 8.5), and purification by anion-exchange chromatography. The preparations were free of contaminating E. coli-derived proteins as judged by SDS-polyacrylamide gel electrophoresis followed by Coomassie Blue or silver staining. In some of the preparations, two bands were observed corresponding to the full-length mature protein and an J. P. J. Brakenhoff, M. Hart, F. D i Padova, and L. A. Aarden, manuscript in preparation.
-10-amino acid shorter IL-&derived degradation product. This degradation product does not significantly change the biological activity or the receptor binding of the full-length product because it generally constituted only 10-20% of the preparations. Moreover, carboxyl-terminal cleavage of only 5 amino acids of IL-6 reduces both the bioactivity and receptor binding affinity 1 0~f o l d (23): and amino-terminal cleavage of -10 amino acids results in a product that behaves very similar to the wild-type molecule (25). Protein concentration was determined both by measuring the absorbance of the preparations and by the Bradford method (64) using bovine serum albumin as a standard. Bradford and A, , ,data correlated best when assuming that the absorbance at 280 nm of a 10 mg/ml solution of IL-6 is 10.

fL-6 B i~s a y s
Mouse BSAssay-The hybridoma growth factor activity of rhIL-6 and variants was measured in the mouse B9 assay as described (31).
CESS Assay-B cell stimulatory fador-2 activity of rhIL-6 variants was measured as described (32). IL-&induced IgGl production by the cells was subsequently measured in a sandwich enzyme-linked immunosorbent assay using a mouse mAb specific for human IgGl (MH161l M , from the Department of Immune Reagents, Central Laboratory of the Netherlands Red Cross Blood Transfusion Service fC.L.B.1, Amsterdam) in combination with a horse radish peroxidase-conjugated mouse mAb specific for human IgG (MH16-1ME; C.L.B.) with a human serum as a standard (H00-1234; C.L.B.). Enzyme-linked immunosorbent assay procedures were as described above.
HepG2 Assay-The hepatocyte stimulating activity of rhIL-6 variants was assessed by measuring the induction of C1 esterase inhibitor production by HepG2 cells as described (33). Following culturing to confluency (5 x los cells in 0.5-ml wells (Costar) in Iscove's modified Dulbecco's medium supplemented with 5% fetal calf serum, 5 x M p-mercaptoethanol, 100 IU of penicillin, 100 d m l streptomycin, and 20 pg/ml human transferrin (Behringwerke, Marburg, Germany), HepG2 cells were washed twice and stimulated with serial dilutions of rhIL-6 or rhIL-6 mutant proteins for 48 h in the same medium in duplicate. In some experiments, cells were washed again after 24 h, and the cultures were continued for another 24 h in the presence of the same stimulus. This procedure results in a higher stimulation index. ARer the incubation period, C 1 esterase inhibitor synthesis was subsequently measured by sandwich radioimmunoassay with anti-C1 esterase inhibitor mAb RII coupled to Sepharose 4B and lZSI-labeled sheep polyclonal anti41 esterase inhibitor IgG with normal human plasma as a standard as described (34).

Binding Experiments
The inhibitory effect of the mAbs on IL-6 binding t o CESS cells was measured by using metabolically 36S-labeled rhIL6 as described (5,351. The inhibitory effect of rhIL-6 and of the rhIL-6 Q160E,T163P mutant protein on 1251-rhIL-6 binding to NIH-3T3 fibroblasts transfected with an 80-kDa IL6R expression vector was measured as described (36). Inhibition of binding of 1251-IL6 to the soluble IG6R was measured as follows. The extracellular ligand-binding domain of the 80-kDa IL-GR was expressed in NIH-3T3 fibroblasts as described (37). Culture supernatants of these cells were diluted 1:2 in 20 nm Tris-HCI (pH 7.51, 140 nm NaCl, 5 nm EDTA, 1% Triton X-100,2 nm methionine, 0.01% NaN3 and incubated with 5 x lo4 dpm 1261-IL-6 in the absence or presence of increasing concentrations of the rhIL-6 or Q160E,T163P protein for 2 h a t 4 "C. 1251-IL6-soluble hIL-6R complexes were immunoprecipitated using an 80-kDa receptor-specific antiserum and protein A-Sepharose. Sepharose-bound radioactivity was subsequently measured with a ycounter (37).
For Scatchard analysis of wild-type rhIL-6 and Q160E,T163P binding to CESS cells, rhIL-6 purified from Chinese hamster ovary cells expressing hIL-6 (a kind giff from Dr. D. Fischer, Interpharm Laboratories, Nes-Ziona, Israel) and Q160E,T163P, purified as described above, were labeled with 12SI-Bolton-Hunter reagent (4000 Cilmmol, diiodinated; Amersham, Amersham, United Kingdom) essentially as described (4, 38). Briefly, 5 pg of rhIL-6 or mutant in 50 p l of borate buffer (50 m~ NaB03, 0.02% Tween 20 (pH 8.5)) was added to 500 pCi of dried Bolton-Hunter reagent. rhIL-6 was incubated for 30 min at room temperature, and Q160E,T163P for 15 min at 0 "C with occasional mixing. Both reactions were stopped by addition of 100 pl of 5 mg/ml glycine in phosphate-buffered saline for 10 min. 1261-Labeled rhIL-6 or   mutant was separated from lasI-labeled products of low molecular mass by gel filtration c~o m a t o~a p h y on a PD-10 column (Phannacia), sterile-filtered, and stored at 4 "C. The specific radioactivity of the preparations was determined by self-displacement analysis and was corrected for maximal binding capacity as described by Calvo et al. (39). The mauimal binding capacity was 38% for 1261-rhIL-6 and 16% for 1zGI-Q160E,T163P. The specific activity of both preparations was -160 Cilmmol.
Binding assays were performed as described by Shanafelt et at. (40). Briefly, CESS cells were hamested, washed on=, and resuspended in binding buf€er (RPMI 1640 medium, 10% fetal calf serum, 50 x m Hepes, 0.02% NaN3 (pH 7.4)). 2 x lo6 viable cellsipoint were incubated with decreasing concentrations of 1261-rkIG6 or 1261-Q160E,T~63P in triplicate at 4 "C with continuous agitation for 3 h. Nonspecific binding was determined by including unlabeled rhIL-6 or Q160E,T163P as appropriate at a 500-fold excess. Cell-bound radioactivity was separated from free radioactivity by oil centri~gation, and bound and total radioactivities were counted with a Cobra 5010 y-counter (40). The standard deviation of the triplicates was generally ~5 % .
The equilibrium binding data were analyzed using the LIGAND program (41). mAbs on biological activity with that on receptor binding of IL-6, human CESS cells (an Epstein-Barr ~~s -t r a n s f o~~ B cell line) were incubated with 35S-rhIL-6 in the presence of increasing concentrat~ons of the mAbs. Fig. 18 shows that m.AbBlfI) completely inhibited IL-6R binding at a concentration of 500 ng/ml, whereas mAb1aII) was unable to fully inhibit rhIL-6 binding even at a concentration of 250 pg/ml. At this concentration, the control mAb, mAb14(111), also inhibited receptor binding, S o , whereas mAb16(1I) inhibited biological activity of IL-6 more efficiently than mAbBl(I), also when tested with 35S-rhIL-6 used in Fig. LB (data not shown), it hardly interfered with the binding of IL-6 to CESS cells.

Site ZI m.Ab That Strongly Znhibits Biological Activity
C o~~r u c~i o n and Analysis of Expression Library of Site IZ ~~~n~ Proteins of IL-&There are multiple ways in which mAb16(11) could inhibit biological activity without preventing receptor bidding of IL-6. One of the possibilit~es was that site 11, the epitope recognized by m.Ab16, is directly involved in signal transduction of IL-6 through interaction with gp130. Mutagenesis in this region could therefore result in the isolation of IL-6 mutant proteins defective in signal ~ansduction, but not in receptor binding. In previous experiments, we had observed that binding of mAblG(I1) to rhIL-6 was abolished by substitution of Tq-158 (27). However, this substitution protein was fully active when tested in the mouse B9 assay. Random mutagenesis of the region around Trp-158 was therefore performed to identify other residues involved in mAblG(I1) binding, to perhaps uncover some that are also important to the biological activity of IL-6. A library of plasmids encoding rhIL-6 mutant proteins with randomly distributed substitutions of Gln-153-Thr-163 was constructed in the E. coli expression vector pUK-IL-6 (25). To select for variants with an intact receptorbinding site, the library was screened for mutant proteins that bound to a site I-specific d b , but not to site 11-specific d b 1 6 . The nucleotide sequences of plasmids encoding mutant proteins with this phenotype were subsequently determined. The phenotype and the biological activity of the mutants are depicted in Fig. 2. Single substitutions of (211-1-155, ksn-156, *-158, and Thr-163 disrupted the mAb16(11) epitope. From the double-and ~ple-substitution mutant proteins, it can be concluded that Leu-159, Gln-160, and Met-162 might also be im- portant for the mAbl6 epitope, but this has yet to be confirmed by analyzing mutant proteins with single substitutions of these residues.

Some Site 1 1 Mutant Proteins Are Inactive on Human CESS
Cell-The biological activity of crude extracts of the various mutant proteins was subsequently measured both in the mouse B9 assay and in IgGl production by human CESS cells. All mutant proteins were biologically active in the mouse B9 assay. However, although very active in the mouse B9 assay, no activity could be detected in the rhIL-6.Tl63P and rhIL-6.Q160E,T163P mutant protein preparations on CESS cells (Fig. 2).
Selective Agonism of Q160E,T163P on Human Cells--To confirm the above observation, both IL-6 mutant proteins were purified and tested for biological activity on CESS cells and on a second IL-6-responsive cell line of human origin. Fig. 3 (A and  B ) shows representative dose-response curves of the mutant proteins in the two assays. In Table I, the specific activities of the mutant proteins in these assays are depicted, together with the specific activities in the mouse B9 assay. The T163P mutant protein was active in all assays, but with a lower specific activity than that of wild-type rhIG6. As with the crude protein, the purified Q160E,T163P mutant protein again did not induce IgGl synthesis by the CESS cells, not even when tested at a 106-fold higher concentration than the wild-type protein (Fig.   3A). At these high concentrations, this variant protein caused a weak increase in the production of the acute-phase protein C1 esterase inhibitor by HepG2 cells. This weak response was  characterized by a strongly reduced maximal response as compared to wild-type rhIL-6 ( Fig. 3B). On mouse B9 cells, the specific activity of Q160E,T163P was G7-fold lower than that of rhIL-6 ( Table I).
Antagonism of IL-6 Action on CESS and HepG2 Cells by Q160E,T163P-We subsequently tested whether these mutant proteins were able to antagonize the biological activity of wildtype rhIL-6. Fig. 4 (A and B ) shows that the Q160E,T163P mutant protein completely inhibited the wild-type IL-6 activity on CESS and HepG2 cells. In these experiments, 50% inhibition of IL-6 activity in CESS and HepG2 assays was observed with -50 ng/ml and 1 pg/ml Q160E,T163P respectively, corresponding to 20-and 200-fold the concentration of wild-type rhIL-6 used to stimulate the cells. No antagonistic activity of the T163P mutant protein could be detected (data not shown). The inhibitory effect of Q160E,T163P on IL-6 activity in both the CESS and HepG2 assays could be reversed by high concentrations of wild-type rhIL-6, suggesting that the inhibitory mechanism is competitive inhibition of IL-6 receptor binding by Q160E,T163P (data not shown).
Specificity of Antagonism by Ql 60E, Tl63P-The production of the acute-phase protein C1 esterase inhibitor by HepG2 cells can be increased in response to both IL-6 and IFN-y via separate mechanisms (33). 'h demonstrate the specificity of inhibition by the double-mutant protein, we tested whether Q160E,T163P could inhibit IFN-y-induced C1 esterase inhibitor synthesis by the HepG2 cells. Because IFN-y is more potent than IL-6 in this assay, we tested the effect of Q160E,T163P over a concentration range of IFN-y. No inhibitory effects of Q160E,T163P were observed at any IFN-y concentration tested. An example is shown in Fig. 5, demonstrating that the C1 esterase inhibitor synthesis induced by 5 ng/ml wild-type rhIL-6 was inhibited to background levels, whereas that induced by 1 ng/ml IFN-y was unimpaired. Binding to 80-kDa IL-6R by Q160E,T163P-The fact that the mutant protein Q160E,T163P could still be recognized by site I-specific mAb8 and that it could antagonize wild-type IL-6 activity on CESS and HepG2 cells suggested that the 80-kDa IL-6R-binding site of the mutant protein was still intact. -To test this hypothesis, binding of this variant to the 80-kDa IL-6R (to both the membrane-bound and -soluble forms) was measured. In Fig. 6 A , the capacity of Q160E,T163P to inhibit binding of '2sI-IL-6 to NIH-3T3 fibroblasts transfected with an expression vector encoding the 80-kDa IL-6R (36) was compared to that of wild-type rhIL-6. Fig. 6B shows the effect of the mutant protein and wild-type rhIL-6 on binding of lz51-IL-6 to the soluble IL-6R. The results indicate that the mutant protein Q160E,T163P was 3-4-fold less efficient than wild-type rhIL-6 in inhibiting binding of lz5I-rhIL-6 to the 80-kDa IL-6R.
No High Affinity Binding of Q160E,T163P to CESS Cells-The above data suggest that Q160E,T163P is inactive on human cells because, although it can efficiently bind to the 80-kDa IL-6R, the complex of mutant and receptor is deficient in triggering signal transduction through gp130. To test (2700 sitedcell), respectively. However, both with E. coli-derived 12sI-rhIL-6 (data not shown) and with Chinese hamster ovary cell-derived glycosylated rhIL-6, we always observed two binding sites. In Fig. 7, the results from representative experiments are displayed in Scatchard plots. ' h o binding sites were statistically significant for wild-type rhIL-6 ( Fig. 7 A F = 26.14 ( p = 0.001) for a one-uersus two-site fit), with Kd values of 13 PM (coefficient of variation (CV) = 40%, 430 sitedcell (CV = 30%)) and 360 PM (CV = 31%, 3200 sitedcell (CV = 9%)). Q160E,T163P, however, exhibited only a single class of binding sites, with a& of 500 PM (Fig. 7B; CV = 59%, 7000 sitedcell (CV = 49%)). A two-site fit of the 1261-Q160E,T163P binding data was not statistically significant (F = 0.03 ( p = 0.9710) for a oneuersus two-site fit). From this experiment, we concluded that Q160E,T163P is inactive on CESS cells because the binding of the mutant.80-kDa IG6R complex to gp130 cannot occur. DISCUSSION In this paper, we show for the first time that receptor binding and signal transduction of hIL-6 can be uncoupled. This is evidenced by the following observations. 1) The Q160E,T163P Selective Agonism of Q160E,T163P--On HepG2 cells, at high concentrations of the Q160E,T163P mutant protein, we reproducibly observed a small partial agonist activity, characterized by a maximal response of -1&20% of that of wild-type rhIL-6 (Fig, 38). This might indicate that the antagonist.80-kDa IL6R complex still exhibits some affhity for gp130, which we I&-6 Receptor Antagonist 91 did not detect in the binding experiments with CESS cells. For the recently described hIL-4 (44) and mouse IL-2 (45) antagonist variants, the response also varied between cell lines studied and could be explained by differences in sensitivity of the cell lines to the respective wild-type cytokines: in the more sensitive cell tines, fulI receptor occupancy was not required to elicit a maximal response. Partial agonist activity of a mutant that was inactive on insensitive cells was explained by occupancy of spare functional receptors on the very responsive cell types, compensating for the receptor activation defects of the mutants (44, 45). This expIanation cannot be applied to our results, however, because CESS cells, which were nonresponsive to the Q160E,T163P mutant, were more sensitive to wildtype rhIL6 than HepG2 cells: for HepG2 cells, half-maximal stimulat~on was achieved at 70 PM rhIL-6, whereas 10 PM induced half-maximal stimulation of CESS cells (Table I). A!though we have as yet no explanation for this observation, our results might be due to differences in IL-6 receptors on CESS and HepG2 cells. F'ietzko et al. (43) recently showed that the gp130 molecule on HepG2 cells seems to differ in molecular mass from that on leukocytes and postulated the existence of'a third receptor chain necessary for high affinity binding, in analogy with the IL-2 system (46). Furthermore, there might be differences in the relative numbers of high and low Ptffinity receptors on CESS and HepG2 cells (Fig. 7A) (43).
In contrast to its activity on human cells, the Q160E,T163P mutant protein was nearly fully active on mouse cells, with a &7-foId reduced specific activity as compared to wild-type rhIL-6 in the mouse B9 assay. At first glance, this seemed to be due to the extremely high sensitivity of B9 cells to IL6: halfmaximal proliferation is induced by 0.08 PM, corresponding to a receptor occupancy of only OB%, assuming a high affbity X d for the mouse IL-6R of -10 PM. However, when we tested the Q160E,T163P mutant protein on the insensitive mouse plasmacytoma cell line T1165, which requires 10 PM IL-6 for halfmaximal activation, as with B9 cells, the specific activity was -15% of that of wild-type rhIL-6. This suggests that in the mouse system, the interaction of the Ql6OE,Tl63P.~0-k~a IL-6R complex with gp130 is not as strongly affected as in the human system and that the double mutant should still be able to bind with high affnity to mouse cells. Further e x~r i m e n t s are in progress to resolve the above issues.
Localization of Gln-160 and Thr-163 in Putative Tertiary Structure of hZL-6"-Unfortunately, the tertiary structure of IL-6 is unknown. Based on homology c o m~r i~n s , however, IL-6 belongs to the large p u p of cytokines that have an antiparallel four-a-helical bundle core structure similar to that of growth hormone, which include granulocytdcolony-stimulating factor, myelomonocytic growth factor, erythropoietin, and prolactin and also oncostatin M, leukemia inhibitory factor, and ciliary neurotrophic factor (8,47,48). For IL-6, the GH structure indeed seems to be a useful working model: Fig. 8 shows the localization of sites I and 11 in the hypothetical three-dimensional model for hIL-6 based on the GI3 structure. mAb~l(I} was shown to recognize residues at both the amino and carboxyl termini of IL6, which fits with close proximity of helices A and I) in the model 123). Site I and 11-specific mAbs are capable of forming a sandwich with monomer rML-6 in enzyme-linked immunosorbent assays, which also agrees with the model (23). Site I on IL-6 and residues close to it are likely to be the 80-M)a receptor-binding site. 1) Site I-specific mAbs strongly inhibit receptor binding, and 2) site I colocalizes with Ser-178-Arg-185, which were recently shown to be essential for biological activity and receptor binding of IL-6 (23,(49)(50)(51)(52)(53). Site I also maps closely to a region essential to bioactivity in the NHz-terminal at-helix (Ile-30-Asp-35) (25, 29). This region on hIL-6 partially colocalizes with binding site 1 on human GHC, . activities of IG6. Detailed information concerning the in vivo sites of production and the activities of IG6 and cytokines with similar activities is required before considering these options. Development of IG6 inhibitor strategies seems very appropriate regarding the speed with which diseases are discovered in which IG6 seems to play a role (60,61). which was identified as a patch consisting of three discontinuous segments: the loop between GH residues 54 and 74 (the A-B loop), the COOH-terminal half of helix D, and, to a lesser extent, the M12-terminal region of helix A (54, 55). Whether the A-B loop in hIG6 is also part of the 80-kDa binding site is as yet unknown.
Gln-160 and Thr-163 are located in the C-D loop and a t the beginning of helix D, respectively, and are part of site 11. This region does not colocalize with site 2 on human GH, which consists of residues near the N H 2 terminus, and on the hydrophilic faces of helices A and C (56). It does colocalize, however, with one of the regions of greatest similarity between neurupoietic and hematopoietic cytokines (including IG6, leukemia inhibitory factor, oncostatin M, and ciliary neurotrophic factor), which was called the "Dl motif" (47). Because gp130 is part of the receptor complex of IG6 (12,13) and is essential for signal transduction by IG6, ciliary neurotrophic factor, leukemia inhibitory factor, and oncostatin M (57,581, it would be interesting to test the effect of mutations in the corresponding Dl regions of these cytokines on biological activity and gp130 interaction. Conclusion-In this report, we described for the first time the isolation of an IG6 analog that could antagonize the activity of wild-type hIG6 in some in vitro bioassays. The dose range in which this molecule inhibited IG6 activities in these bioassays is similar to that reported for the natural occumng IGlra (59). The pleiotropy of IG6 and the broad distribution of its receptors make selective therapeutical application of receptor antagonists difficult to envision. However, the observation that it is possible to isolate an IG6 variant that is active in some (but not all) bioassays may point to the possibility of isolating selective agonists of IG6 that retain useful activities, but antagonize IG6 activity where desired. Otherwise, combination of an IG6 antagonist with other cytokines that share some (but not all) of the functions of IG6 might in part restore the useful