Structure-guided Optimization of the Interleukin-6 Trans-signaling Antagonist sgp130*

Binding of interleukin-6 (IL-6) to its specific receptor IL-6R is a prerequisite for the activation of the signal-transducing receptor glycoprotein 130 (gp130). A soluble form of the IL-6R (sIL-6R) in complex with IL-6 can activate cells lacking membrane-bound IL-6R (trans-signaling). IL-6-trans-signaling is counterbalanced by a naturally occurring, soluble form of gp130 (sgp130), whereby signaling via the membrane-bound IL-6R is not affected. Many inflammatory and neoplastic disorders are driven by IL-6 trans-signaling. By analysis of the three-dimensional structure of gp130 in complex with IL-6 and sIL-6R, we identified amino acid side chains in gp130 as candidates for the generation of sgp130 muteins with increased binding affinity to IL-6/sIL-6R. In addition, with information from modeling and NMR analysis of the membrane proximal domain of gp130, we generated a more stable variant of sgp130Fc. Proteins were tested for binding to the IL-6/sIL-6R-complex, for inhibition of IL-6/sIL-6R-induced cell proliferation and of acute phase gene expression. Several mutations showed an additive effect in improving the binding affinity of human sgp130 toward human IL-6/sIL-6R. Finally, we demonstrate the species specificity of these mutations in the optimal triple mutein (T102Y/Q113F/N114L) both in vitro and in a mouse model of acute inflammation.

IL-6 signals via a complex of IL-6, IL-6-␣-receptor (IL-6R), and two gp130 molecules (2) that leads to the transcription of target genes, such as acute phase response genes in hepatocytes (6,7). IL-6-type cytokines are involved in proliferation and differentiation processes, predominantly in the hematopoietic system, in neural cells and in the immune response (1,2). IL-6 has major functions in inflammatory reactions of the body (1,2). Mice with a targeted inactivation of the IL-6 gene are protected of rheumatoid arthritis (8,9) and multiple sclerosis (10). Furthermore, regenerative reactions like wound healing and liver regeneration are severely compromised in IL-6 Ϫ/Ϫ mice (11).
The expression of the IL-6R is limited mainly to hepatocytes and some leukocytes (12). Cells lacking IL-6R expression are not responsive to the cytokine IL-6. A soluble form of the IL-6R, however, can bind IL-6 with the same affinity as the membranebound form, and the complex of IL-6 and the soluble IL-6R (sIL-6R) can induce signaling in a process called trans-signaling (13,14). Because the IL-6R is only sparely expressed, IL-6 transsignaling dramatically increases the number of potential IL-6 target cells (12). We have shown that the naturally occurring soluble form of gp130 selectively inhibits IL-6 responses mediated by the soluble IL-6R without affecting responses via the membrane-bound IL-6R (15,16).
Recently it has been found that inhibition of the inflammatory cytokines TNF␣ (17) and  in rheumatoid arthritis and other inflammatory diseases can have dramatic therapeutic effects. More than 1 million patients have been treated so far with TNF␣-neutralizing agents (19). Global blockade of cytokines, however, is hampered by side effects including recurrent infections, highlighting the fact that cytokines exhibit pro-and anti-inflammatory properties (20). A fusion protein consisting of the extracellular region of gp130 and the Fc part of a human IgG1 antibody (sgp130Fc) selectively inhibits IL-6 trans-signaling without affecting responses via the membrane-bound IL-6R. Therefore, this protein preferentially inhibits pro-inflammatory activities of IL-6, as it has been postulated that IL-6 signaling via the membrane-bound receptor mainly contributes to anti-inflammatory reactions, whereas IL-6-trans-signaling mediates pro-inflammatory and chronic inflammatory states (21).
The structure of the extracellular portion of gp130 (domains 1-3) in complex with IL-6 and the extracellular part of the IL-6R (domains 2 and 3) has been solved by x-ray crystallography (22,23). We have inspected the contact areas between IL-6 and gp130 and have identified several amino acid side chains as candidates for site-directed mutagenesis to improve the ligand binding properties of gp130. It turned out that only mutations within the contact area to site III of IL-6 improved binding of gp130 to the IL-6/sIL-6R complex, whereas changes within the contact region to site II of IL-6 resulted in decreased binding. In a previous study, we have shown that sgp130Fc tends to form large amounts of aggregates (24). Based on the three-dimensional model structure of the extracellular domain 6 of gp130 (25), we generated an optimized version with a higher stability and a much lower tendency to form aggregates.
Here, we demonstrate that one of the mutated sgp130Fc proteins with only three amino acid exchanges is far more effective in blocking different biologic responses mediated by the IL-6/ sIL-6R complex in vitro. Moreover, using an animal model of acute inflammation, which is driven by the IL-6/sIL-6R complex (26, 27), we show that the sgp130Fc mutein is effective in blocking the progression of the inflammation from an acute to a chronic state. Our study therefore provides a general strategy to improve the efficacy of proteins, which are candidates for therapeutic applications.

EXPERIMENTAL PROCEDURES
Constructs-The sgp130Fc protein was produced as described (15). The optimized sgp130Fc containing a C-terminal extended part of the extracellular region of gp130 (amino acids 1-595) was fused to the Fc part (15). Muteins with single amino acid substitutions or combinations thereof were generated by overlapping primer PCR within three cassettes flanked by endonuclease restriction sites (XhoI/SspI, SspI/BclI, BclI/KpnI, respectively). These constructs were subcloned into the expression vector pcDNA3.1(ϩ) (Invitrogen, Karlsruhe, Germany).
Protein Purification-Cell supernatants were subjected to affinity chromatography (protein A-Sepharose (Roche Applied Sciences, Mannheim, Germany)) and eluted by 5 column volumes of 50 mM citric acid. Relevant fractions were concentrated, and the buffer exchanged for phosphate-buffered saline (PBS) using PD10 columns (GE Healthcare, Freiburg, Germany). The proteins were further purified by gel filtration on a Sephacryl S-200 HR (16/60) column (GE Healthcare) at a flow rate of 0.5 ml/min PBS. Fractions of 2.5 ml were collected, pooled, and concentrated. Protein concentrations were determined by recording absorption spectra in the range from 240 to 320 nm (31).
sgp130Fc ELISA and Hyper-IL-6 Titration Assay-The concentration of sgp130Fc and muteins derived from cell supernatants was measured by ELISA. Plates (Maxisorp, Nunc, Wiesbaden, Germany) were coated with 100 l of 10 g/ml protein A (Sigma) for 5 h at room temperature and blocked overnight at 4°C with 3% BSA and 5% sucrose in PBS. Supernatants were diluted 1:20 in 1% BSA plus 0.5% FBS in PBS, and 200 l of the dilution were incubated for 2 h at room temperature. Wells were blocked with 5% human serum (PAA) for 30 min to saturate residual protein A binding sites. Plates were then washed with PBS containing 0.05% Tween 20 (PBST) and incubated with 100 l of anti-human CD130 antibody (B-P4, Diaclone Research, Stamford, CT) diluted 1:5,000 in 1% BSA plus 1% FBS in PBS for 2 h. Detection was performed with anti-mouse IgG POD (ImmunoPure, Pierce) at 1:5,000 in 1% BSA plus 1% FBS in PBS for 1 h, the assay was developed with BM blue POD substrate (Roche Applied Science) and finally terminated by adding 1 M H 2 SO 4 followed by measurement at 450 nm. In a second round, equal amounts of sgp130Fc proteins in the supernatant were loaded and incubated with 1 ng/well Hyper-IL-6 for 1 h. Hyper-IL-6 was detected with the biotinylated anti-human IL-6R antibody BAF277 (R&D Systems, Minneapolis, MN) at a 1:2,500 dilution and streptavidin-POD conjugate (Roche Applied Science) at 1:5,000 in 1% BSA in PBS.
Proliferation Assay-Proliferation of BAF3/gp130 cells in response to Hyper-IL-6 or the human or murine IL-6/IL-6R complex (all components purchased from R&D Systems) and its inhibition by sgp130Fc, and the muteins was measured by [ 3 H]thymidine incorporation (32), Cell Titer Blue fluorescence assay or Cell Titer MTS colorimetric assay (both from Promega, Mannheim, Germany). For this purpose, 5 ϫ 10 3 BAF3/ gp130 cells were seeded per well in 96-well plates and cultured in a final volume of 100 l. Proliferation was induced by adding 1 ng/ml Hyper-IL-6 or 100 ng/ml human IL-6 plus 50 ng/ml human sIL-6R or 300 ng/ml murine IL-6 plus 150 ng/ml murine sIL-6R. The assay was performed with each value being determined in quadruplicate.
Native Polyacrylamide Gel Analysis-Native PAGE was carried out in 7.5% polyacrylamide gels with a Tris/HCl buffer system (pH 6.8 in the glycerol sample buffer and the stacking gel, pH 8.8 in the separating gel). Gels were run in a buffer of 25 mM Tris and 192 mM glycine at a constant voltage of 60 V until the samples had completely entered the stacking gel, then at 80 V for 4 -5 h. Proteins were subsequently fixed and silverstained according to standard procedures.
Western Blot Analysis-After SDS-PAGE and transfer to a polyvinylidene difluoride membranes (GE Healthcare), Western blot analysis was performed using a secondary peroxidaseconjugated antibody (ImmunoPure anti-mouse IgG horseradish peroxidase from goat, Pierce). Membranes were developed using the Amersham Biosciences ECL plus kit (GE Healthcare). The gp130 part of sgp130Fc was detected by the monoclonal mouse antibody anti-human CD130 clone B-P4 (Diaclone), which recognizes domain 4 of gp130.
Haptoglobin ELISA-An ELISA to detect the human acute phase protein haptoglobin was performed as described (33). Briefly, 10 5 HepG2 cells per well were seeded in 96-well plates and left to adhere overnight. Cells were washed twice with PBS (37°C) and starved in DMEM without FBS for 2 h. In the meantime, sgp130Fc and muteins were diluted and mixed with 5 ng/ml Hyper-IL-6 in DMEM without FBS and placed in the incubator for at least 30 min. Cells were incubated with 200 l of cytokine mixture for 20 h, and the amount of haptoglobin in the cell supernatant was determined by ELISA. Each value was determined in quadruplicate.
Surface Plasmon Resonance Studies-Plasmon resonance experiments were carried out with a ProteOn XPR36 protein interaction array system (Bio-Rad). The running buffer was PBS) with 0.005% Tween 20, pH 7.4 (PBST), and experiments were carried out at 25°C. The surface of a ProteOn GLC sensor chip was activated with the ProteOn amine coupling kit (4 mM EDAC/1 mM sulfo-NHS, flow rate 30 l/min, volume 40 l), and proteins were covalently coupled at 10 g/ml in 10 mM acetate buffer pH 4.5 (flow rate 30 l/min, volume 150 l). The respective levels of immobilization were 2,250 RU (resonance units) for sgp130Fc, 1,330 RU for the optimized sgp130Fc and 1,120 RU for mutein XIII. The concentrations of Hyper-IL-6 ranged from 80 to 2.5 nM in 2-fold dilutions in PBST, and the flow rate was 100 l/min. Association was monitored for 60 s before replacing the sample by PBST and monitoring the dissociation phase for another 600 s. Each sensogram set was referenced using the reference channel and was baseline-aligned. Sensograms were analyzed using the ProteOn Manager 2.0 software.
Animal Treatment-All procedures involving animals and their care were conducted in accordance to national and international laws and policies as well as the guidelines for animal care of the University of Kiel (acceptance no.: V 742-72241.121-3 (20-2/04) and (76-7/00)). Mice were maintained in a 12-h light/dark cycle under standard conditions and were provided with food and water ad libitum. Blood was drawn by tail bleeding or by cardiac puncture under general anesthesia.
Air Pouch Model-The air pouch model of local inflammation was performed with C57Bl/6 mice (34). In brief, mice were anesthetized and subcutaneous dorsal pouches were created by injection of 6 ml of sterile air. After 3 days, the pouches were reinjected with 4 ml of air. On day 6, 1 ml of 1% carrageenan (Sigma-Aldrich) in sterile PBS was injected into the pouches. sgp130Fc wild-type protein or muteins (10 g/mouse) or PBS were administered intraperitoneally 6 h before the carrageenan injection. Seventy-two hours after treatment, animals were sacrificed, and the pouches were washed with 3 ml of PBS. The lavage fluid was immediately cooled on ice and centrifuged at 5,000 rpm for 10 min at 4°C. The cells contained in the lavage fluid were analyzed by FACS (see below), and the supernatant was analyzed by ELISA for MCP-1 (DuoSet mouse MCP-1 ELISA Kit, R&D Systems) and sgp130Fc (DuoSet human sgp130 ELISA Kit, R&D Systems).
Molecular Modeling-To construct a model of the mouse IL-6/mouse IL-6R/human gp130 complex, the recently published structure of the human IL-6/IL-6R/gp130 complex was used as a template (23). According to the published alignment, amino acid residues of IL-6 and IL-6R were exchanged in the template (35). Insertions and deletions in the molecules were modeled using a data base search approach included in the software package WHATIF (36). For graphical representations, the program RIBBONS (37) was used.

RESULTS
Selection of Amino Acid Exchanges-IL-6 interacts with three receptor subunits (38,39). Site I is occupied by the specific IL-6R and site II is in contact with the cytokine binding module (domains 2 and 3) of gp130, whereas site III interacts with the Ig-like domain 1 (D1) of gp130 (Fig. 1, A and B) (23). When analyzing the three-dimensional structure of gp130 (D1-D3) in complex with IL-6 and the extracellular portion of the IL-6R, we focused on hydrophilic amino acid residues in gp130 participating in the interaction with IL-6 or IL-6R. Such residues were exchanged to hydrophobic side chains to achieve larger hydrophobic interaction areas and thereby increasing the affinity between the two molecules. The amino acid residue S251E was chosen to establish an additional salt bridge between gp130 and IL-6. As indicated in Fig. 1A, the amino acid side chains Ser- 251, Ser-279, Thr-285, and Lys-303 of gp130 are in contact with site II of IL-6, whereas amino acid side chains Thr-102, Gly-109, Gln-113 and Asn-114 of gp130 are in the contact region to site III of IL-6. In addition, combinations of all these mutations were generated (Fig. 1A).
Binding to Hyper-IL-6-The sgp130Fc cDNA and the mutated cDNAs were cloned into eukaryotic expression vectors and were used to transiently transfect HepG2 cells. For testing protein ability to bind Hyper-IL-6, a fusion protein of IL-6 and IL-6R (28), crude supernatants from these cells were used ( Fig. 2A) and tested by ELISA. Equal amounts of sgp130Fc protein were captured from the supernatants of transfected HepG2 cells by immobilized protein A. After incubation with Hyper-IL-6 and removal of the unbound material, the amount of bound Hyper-IL-6 was determined. Interestingly, mutations in the binding region of site II (muteins I to VI) resulted in reduced binding ability toward Hyper-IL-6 compared with the wild-type sequence. In contrast, all variants involving site III mutations, except for mutation VIII, exhibited similar or even higher binding abilities (muteins VII and IX to XIII) ( Fig. 2A).
Combination of all seven (site II and site III) mutations from muteins V and XIII in mutein XIV led to reduced binding most likely due to the effect of mutation IV, which was removed from the combination in mutein XV, resulting in a partial rescue of the Hyper-IL-6 binding ability ( Fig. 2A).
Purification and BAF3/gp130 Cell Assay-To verify these results in a cell-based assay, we purified the proteins by protein A-Sepharose affinity chromatography. The biologic activity of the purified muteins was tested on the cellular trans-signaling model cell line BAF3/gp130, which, due to stable transfection with a human gp130 cDNA, proliferates upon stimulation with IL-6/ sIL-6R or Hyper-IL-6 (28). This proliferation can be inhibited by sgp130Fc (15). As shown in Fig. 2B, proliferation of Hyper-IL-6-stimulated BAF3/gp130 cells was inhibited in a dose-dependent manner by sgp130Fc. Based on their inhibitory potential, the muteins could be divided into three categories. Muteins I to VI, VIII, XIV, and XV exhibited a considerably weaker or no inhibitory activity compared with wild-type sgp130Fc. Mutein VII showed only a slightly enhanced capacity of inhibition, whereas muteins IX to XIII exhibited a significantly higher potency in inhibiting BAF3/gp130 cell proliferation (Fig. 2B). Interestingly, the muteins with lower inhibitory potential all contained mutations in the gp130 contact region for site II of IL-6, whereas the muteins with increased inhibitory potential carried mutations within the gp130 contact region for site III. Moreover, the mutations V and VI abolished the inhibitory effect of sgp130Fc completely, but the activity of the corresponding muteins could partially be rescued by introducing beneficial mutations in the site III binding region (mutein XIV and XV, Fig. 2, A and B). It should be noted that mutein VII and IX showed higher binding to the IL-6/sIL-6R complex than the combined mutein XIII, although mutein XIII was most effective in the trans-signaling inhibition assay (Fig. 2B), which might be explained by differential protein stability (see below).
After the initial purification by affinity chromatography, we found that about 50% of the material was present as aggregates. These aggregates could be detected by Western blotting even under denaturing and reducing conditions (Fig. 3A). To examine whether these aggregates were also present under native conditions, we performed size-exclusion chromatography (Fig. 3D). About 50% of the material eluted in the void volume, indicating the presence of aggregated sgp130Fc. An NMR-supported three-dimensional model of the sixth (membrane proximal) domain of gp130 (25) revealed that 11 amino acid residues missing in the original sgp130Fc protein constitute the last ␤-strand of domain 6 (Fig. 3B). Therefore, in a new construct, we added these 11 FIGURE 2. Binding ability and biological activity of sgp130Fc muteins. A, supernatants of HepG2 cells transfected with plasmids coding for sgp130Fc protein or muteins were analyzed by ELISA. Equal amounts of the sgp130Fc proteins were bound to the plates and incubated with 1 ng/ml Hyper-IL-6. After washing, the amount of bound Hyper-IL-6 was determined, the Hyper-IL-6/sgp130Fc mutein ratio was calculated and normalized to the Hyper-IL-6/sgp130Fc value. B, sgp130Fc and muteins were purified from the HepG2 supernatants by protein A-Sepharose affinity chromatography and analyzed with respect to their ability to inhibit cell proliferation of BAF3/gp130 cells. Cells were cultured with 1 ng/ml Hyper-IL-6 and the indicated amounts of purified wild-type sgp130Fc or muteins, and proliferation of the cells was quantified by [ 3 H]thymidine incorporation. OCTOBER 3, 2008 • VOLUME 283 • NUMBER 40 amino acid residues to prevent misfolding and subsequent aggregation of this domain.

Optimization of IL-6 Trans-signaling Antagonist sgp130
After establishing stably transfected CHO-K1 cell lines expressing the optimized sgp130Fc version or the muteins VII, IX, XI, XII, and XIII, we analyzed these molecules by native gel electrophoresis and gel filtration (Fig. 3, C and D). As anticipated, the amount of native sgp130Fc was dramatically increased, with almost no aggregation observed for the optimized version as well as all muteins expressed (VII, IX, and XI to XIII). Throughout the following experiments, the most promising mutein XIII was used.
Proliferation and Inhibition of Acute Phase Induction-To examine the influence of the increased stability on the biological activity, we investigated the original sgp130Fc, the optimized sgp130Fc with elongated domain 6 and mutein XIII in a proliferation assay. Fig. 4A demonstrates that the elongated sgp130Fc version is able to inhibit Hyper-IL-6-induced proliferation to a much larger extent than the original sgp130Fc. Furthermore, the introduction of the mutation XIII into the optimized construct caused another drastic improvement of its inhibitory capacity (Fig. 4A).
As BAF3/gp130 cells are murine cells and express transfected human gp130 cDNA, we verified the results with human hepatoma cells (HepG2). Upon IL-6 stimulation (40) or stimulation with the IL-6/sIL-6R complex (41), HepG2 cells show an induction of acute phase proteins and can be used as a model system for the response to liver inflammatory processes. The acute phase protein haptoglobin secreted by these cells in response to Hyper-IL-6 was quantified by ELISA (33). The ability to inhibit the Hyper-IL-6 response in a dose-dependent manner by the optimized sgp130Fc is shown in Fig. 4B. Mutein XIII exhibited a significantly higher reduction of haptoglobin secretion as compared with optimized sg130Fc (Fig. 4B).
Binding Kinetics-To examine whether the improved inhibitory potential of the sgp130Fc optimized protein and the mutein XIII is reflected in the binding kinetics, we used surface plasmon resonance to quantify the k on and k off rates. sgp130Fc proteins were immobilized on the affinity sensor chip and binding of Hyper-IL-6 was measured. The sensograms are depicted in Fig. 4, C and D for the optimized sgp130Fc and mutein XIII, respectively. For comparison, we also performed these measurements for the original sgp130Fc. From these sensograms, the k on and k off constants and the affinity constants K D were calculated. Compared with the original sgp130Fc, the optimized version with elongated domain 6 shows an increased k on , whereas k off is virtually identical ( Table 1). As a consequence, the affinity constant K D is decreased, demonstrating the higher stability of the optimized version. In case of mutein XIII, the higher inhibitory capacity is reflected in a moderately increased k on and a much lower k off and, consequently, in a 4-fold decreased affinity constant K D . The change in the k off results from an energetically more stable complex, whereas the complex formation is only slightly affected compared with the optimized sgp130Fc.
Species-specific Enhanced Activity of sgp130Fc Mutein XIII-In all experiments described above we used the human IL-6 and the human IL-6 receptor or Hyper-IL-6, which is based on the human IL-6 and sIL-6R sequences. As a prerequisite for in vivo studies in mice, we investigated whether the mutein XIII had the same enhanced binding properties toward mouse IL-6 and mouse IL-6 receptor. For this purpose, we used the three-dimensional structure of the human IL-6/IL-6R/gp130 complex to generate a model complex of mouse IL-6/mouse IL-6R and human gp130 and compared these structures with the human complex. The phenylalanine in mutation IX (Q113F) was hypothesized to enlarge a cluster of aromatic side chains in the interface of gp130 and the human IL-6R and thereby strengthen the binding (Fig. 5A), which we indeed observed. Inspection of the mouse/human complex revealed that the mouse IL-6R lacks one of the aromatic side chains (F155) (Fig. 5A), and therefore the enhanced binding affinity of mutein XIII observed in the human situation should not be effective in the mouse/hu-man system. To verify this prediction, we used the cell line BAF3/gp130 (expressing human gp130, see above) to test the inhibition of the optimized sgp130Fc and mutein XIII using the murine IL-6/sIL-6R complex for stimulation. As predicted, the mutein XIII is not more effective than optimized sgp130Fc with murine IL-6/sIL-6R (Fig. 5B). The same holds true for mutein XI (N114L): the increased hydrophobicity of the N114L mutations in the human system is compensated by the amino acid exchange (R117/M116) in mouse as compared with human IL-6 (data not shown). Therefore, the enhanced activity of sgp130Fc mutein XIII is restricted to human IL6/sIL6R complexes.
In Vivo Studies-sgp130Fc has recently been shown to exhibit promising therapeutic potential for the treatment of chronic inflammatory diseases including Crohn disease and rheumatoid arthritis (16,42). This suggestion was based on the fact that sgp130Fc could be used to block disease progression in animal models of these diseases (21). It was therefore interesting to ask whether the optimized sgp130Fc mutein was able to FIGURE 4. Biological activity and binding affinities of sgp130Fc and its muteins. A, proliferation of BAF3/gp130 cells and its inhibition by the original (dotted) or optimized (hatched) sgp130Fc or mutein XIII (filled). Subsequently, the number of viable cells was determined by Cell Titer-Blue fluorescence assay (***, p Ͻ 0.001). B, acute phase response measured by expression of haptoglobin after stimulation of human HepG2 cells with 5 ng/ml Hyper-IL-6 and its inhibition by increasing amounts of the optimized sgp130Fc versus mutein XIII (***, p Ͻ 0.001; **, p Ͻ 0.01; *, p Ͻ 0.05). C and D, sensograms of the binding kinetics of Hyper-IL-6 to the optimized sgp130Fc (C) and mutein XIII (D) measured by surface plasmon resonance. sgp130Fc or mutein XIII were immobilized on the surface, and the indicated concentrations of Hyper-IL-6 were added. Optimization of IL-6 Trans-signaling Antagonist sgp130 OCTOBER 3, 2008 • VOLUME 283 • NUMBER 40

JOURNAL OF BIOLOGICAL CHEMISTRY 27205
inhibit biologic activities of the IL-6/sIL-6R complex in vivo.
We have recently shown that in the mouse air pouch model, the IL-6/sIL-6R complex is important to drive the inflammatory process from the acute neutrophilic state to the more chronic state governed by mononuclear cells (26,27). In this model, the infiltration of mononuclear cells is mediated by the CC-chemokine MCP-1, which is induced in the lining cells by the IL-6/ sIL-6R complex, but not by IL-6 alone. We therefore used this model to study whether the infiltration of cells and the secretion of MCP-1 could be modulated by the optimized sgp130Fc and mutein XIII. Injection of optimized sgp130Fc and mutein XIII clearly decreased the number of infiltrating cells to the same extent (Fig. 6A). In addition, the ratio of infiltrating neutrophils and monocytes is similar after injection of both molecules (Fig. 6B). Although the concentration of mutein XIII in the inflamed area is slightly higher compared with the optimized sgp130Fc (Fig. 6C), mutein XIII exhibits the same inhibitory effect on the concentration of MCP-1 (Fig. 6D). Therefore, the conclusion derived from our model shown in Fig. 5A is in good agreement with the situation in vivo, demonstrating that the enhanced activity of sgp130Fc mutein XIII is restricted to human IL-6/sIL-6R complexes.

DISCUSSION
There are four major findings in this study. First, upon changing amino acid residues within the gp130 contact sites to the binding sites II and III of IL-6, we identified mutations which increased the binding affinity of sgp130Fc to the IL-6/ sIL-6R complex. Surprisingly, only gp130 mutations contacting site III of IL-6 resulted in increased binding, whereas all changes at the site II interface decreased the binding affinity. Secondly, as expected from theoretical considerations (43), we could demonstrate that the increased affinity of the mutated sgp130Fc protein was entirely due to a lowered off-rate of the IL-6/sIL-6R complex bound to sgp130Fc. Thirdly, close inspection of the gp130 D6 domain and modeling of its three-dimensional structure resulted in the prediction that the originally used sgp130Fc protein did not contain the entire membrane proximal ␤-sheet (25). We hypothesized that this was responsible for the observed tendency of the sgp130Fc to aggregate. Indeed, the optimized version of sgp130Fc with an extended C terminus of domain six did not aggregate and, consequently, was more stable in solution. When the two features, affinityincreasing mutations and the enhanced stability, were combined in one molecule, the resulting sgp130Fc protein was about 100-fold more potent in blocking the biologic activity of the IL-6/sIL-6R complex in vitro. Finally, by comparing the structures of the human IL-6/IL-6R/gp130 complex with the mouse IL-6/ mouse IL-6R/human gp130 complex, which is the relevant complex in the mouse air pouch model, we realized that the sgp130Fc mutations shown effective in the human system might not enhance the binding affinity of this molecule in the mouse system. This prediction of species specificity could be verified in vitro and in vivo. An important consequence of this finding is that the optimized sgp130Fc protein will have to be tested in primate models such as the cynomalgus monkey model.
The mechanism by which IL-6/IL-6R binds to the two binding sites of gp130 is still under debate. Although our data do not FIGURE 5. The enhanced activity of sgp130Fc mutein XIII is restricted to human IL-6/sIL-6R complexes. A, detail of the interaction sites of gp130 (green), IL-6 (pink), and IL-6R (gray). In the left panels (human), the interaction between human wild type (wt) gp130 and mutant Q113F with human IL-6 and human sIL-6R is shown. The right panels (mouse) show the respective interactions with murine IL-6 and sIL-6R. In the all-human complex, an aromatic cluster is formed which is hypothesized to account for the increased affinity of muteins containing Q113F. B, proliferation of BAF3/gp130 cells and its inhibition by optimized sgp130Fc or mutein XIII was measured in response to the human or murine IL-6/sIL-6R complex (MTS assay; OD at 490 nm). provide an insight into the mechanism of complex formation, it is interesting to note that the deleterious mutations contacting site II could be rescued by amino acid exchanges at the site III interface, suggesting independent binding events rather than a sequential order of binding of the IL-6/sIL-6R complex to the gp130 homodimer.
The sgp130Fc protein has been shown to selectively inhibit IL-6 trans-signaling mediated by the sIL-6R without affecting classic signaling via the membrane bound IL-6R (15,16). Global blockade of the cytokines IL-6 and TNF␣ in patients has revealed that some patients suffer from recurrent and partly life-threatening infections due to the inhibition of anti-inflammatory properties of these cytokines (17,18,20). We have suggested that one advantage of the sgp130Fc protein for therapeutic interventions may be the fact that many physiological IL-6 responses, which rely on classic IL-6 signaling, are not inhibited by this protein (21). This makes this protein a promising candidate for therapeutic intervention in the treatment of chronic inflammatory diseases, such as Crohn disease and rheumatoid arthritis, without the blockade of the hepatic acute phase response, which is an integral part of the innate immune response and which is triggered by IL-6 via the membranebound IL-6R on hepatocytes.