Studies on the interleukin-6-type cytokine signal transducer gp130 reveal a novel mechanism of receptor activation by monoclonal antibodies.

The transmembrane glycoprotein gp130 belongs to the family of hematopoietic cytokine receptors. It represents the common signal transducing receptor component of the so called interleukin-6-type cytokines. For several cytokine receptors including gp130 it has been shown that receptor activation cannot only be achieved by the natural ligand but also by single monoclonal antibodies raised against the receptor ectodomain. These findings have been interpreted in a way that dimerization of cytokine receptors is sufficient for receptor activation. Here we show that the recently described gp130-activating antibody B-S12 actually consists of two different monoclonal antibodies. By subcloning of B-S12 the monoclonal antibodies B-S12-A5 and B-S12-G7 were obtained. The individual antibodies are biologically inactive, in combination they exert B-S12-like activity on hepatoma cells. On Ba/F3 cells stably transfected with gp130 a combination of B-S12-G7 with another monoclonal gp130 antibody, B-P8, is required to stimulate proliferation. Using gp130 deletion mutants we show that all three antibodies map to domains 2 and 3 of gp130 which constitute the cytokine binding module. The individual antibodies inhibit activation of the signal transducer by interleukin-6 and interfere with binding of interleukin-6 to gp130. Interestingly, the combination of B-S12-G7 and a Fab fragment of B-P8 retains biological activity. We conclude from our data that (i) the monoclonal antibodies activate gp130 by mimicking the natural ligand and (ii) enforcement of gp130 dimerization is not sufficient for receptor activation but additional conformational requirements have to be fulfilled.

The transmembrane glycoprotein gp130 belongs to the family of hematopoietic cytokine receptors. It represents the common signal transducing receptor component of the so called interleukin-6-type cytokines. For several cytokine receptors including gp130 it has been shown that receptor activation cannot only be achieved by the natural ligand but also by single monoclonal antibodies raised against the receptor ectodomain. These findings have been interpreted in a way that dimerization of cytokine receptors is sufficient for receptor activation. Here we show that the recently described gp130activating antibody B-S12 actually consists of two different monoclonal antibodies. By subcloning of B-S12 the monoclonal antibodies B-S12-A5 and B-S12-G7 were obtained. The individual antibodies are biologically inactive, in combination they exert B-S12-like activity on hepatoma cells. On Ba/F3 cells stably transfected with gp130 a combination of B-S12-G7 with another monoclonal gp130 antibody, B-P8, is required to stimulate proliferation. Using gp130 deletion mutants we show that all three antibodies map to domains 2 and 3 of gp130 which constitute the cytokine binding module. The individual antibodies inhibit activation of the signal transducer by interleukin-6 and interfere with binding of interleukin-6 to gp130. Interestingly, the combination of B-S12-G7 and a Fab fragment of B-P8 retains biological activity. We conclude from our data that (i) the monoclonal antibodies activate gp130 by mimicking the natural ligand and (ii) enforcement of gp130 dimerization is not sufficient for receptor activation but additional conformational requirements have to be fulfilled.
Fast and reliable intercellular communication is a requirement for the development, maintenance, and function of multicellular organisms. A central mode of communication is the release of mediators by one cell that are recognized specifically by the cognate receptors on the plasma membrane of target cells. Receptors act like a kind of molecular switches that respond to ligand binding with the triggering of cytoplasmic signal transduction pathways. These signaling cascades often target to nuclear DNA leading to specific cellular responses like growth, differentiation, or even apoptosis. The molecular mech-anisms of receptor activation by the ligand depends on the type of receptor involved. The commonly accepted activation mechanism for most receptors spanning the membrane only once is ligand-induced receptor dimerization (1,2). Important receptors sharing this feature belong to the families of receptor tyrosine kinases and cytokine receptors.
The ectodomain of gp130 is proposed to consist of an Nterminal Ig-like domain followed by a cytokine-binding module (CBM) and three fibronectin type III-like domains (14). The CBM is the hallmark of class I cytokine receptors and consists of two fibronectin type III-like domains that show as conserved features a distinct pattern of cysteine residues in the N-terminal domain and a WSXWS motif in the C-terminal domain (16). The structures of the CBM constituting domains 2 and 3 of gp130 have recently been solved (17,18). The Ig-like domain and the CBM of gp130 are required and sufficient to bind IL-6⅐IL-6R-as well as IL-11⅐IL-11R complexes (19,20). In comparison to the ectodomains of other cytokine receptors such as EpoR, IL-2R, IL-4R, or GHR that only consist of a single CBM (for review, see Ref. 21), the architecture of gp130 is relatively complex. The functional relevance of its structural features are not fully understood.
Activation of gp130 by IL-6-type cytokines triggers several cytoplasmic signal transduction cascades from which the Jak/ STAT pathway is understood in greatest detail. After cytokine binding Jak kinases that are constitutively associated with the cytoplasmic tail of gp130 are activated and subsequently phosphorylate tyrosine residues of gp130. These phosphotyrosine side chains serve as docking sites for latent transcription factors of the STAT family, mainly STAT3. Upon phosphorylation the STATs dimerize, translocate into the nucleus, and induce IL-6-type cytokine target genes (22,23).
Mechanistically, it is believed that IL-6 and IL-11, bound to their respective ␣-receptors, trigger dimerization of gp130 and that receptor dimerization is sufficient to launch signal transduction (4,24). The fact that several cytokine receptors or receptor tyrosine kinases can be activated by monoclonal antibodies has been interpreted to support the hypothesis that receptor dimerization is sufficient for receptor activation (25). In this study we show that efficient activation of gp130 requires the simultaneous action of two monoclonal antibodies and provide evidence that the mechanism of gp130 activation goes beyond simple receptor dimerization.

EXPERIMENTAL PROCEDURES
Chemicals, Proteins, and Cell Culture Media-Tran 35 S-label metabolic labeling reagent was purchased from ICN (Meckenheim, Germany). Recombinant human IL-6 was expressed in Escherichia coli, refolded, and purified as described by Arcone et al. (26). The specific activity was 10 8 units per mg of protein in the B9 cell proliferation assay (27). Soluble IL-6R was expressed in insect cells as described previously (28). The phycoerythrin-conjugated secondary antibody used for FACS analysis was purchased from Dianova (Hamburg, Germany). The frequently used PBS buffer contained 200 mM NaCl, 2.5 mM KCl, 8 mM Na 2 HPO 4 , and 1.5 mM KH 2 PO 4 . Dulbecco's modified Eagle's medium/ F-12, and RPMI-1640. Antibiotics were obtained from Life Technologies, Inc. (Eggenstein, Germany), fetal calf serum from Cytogen (Lohmar, Germany).
Generation of Antibodies against the Gp130 Ectodomain-Generation and characterization of the gp130 antibodies B-S12, B-P8, B-R3, B-P4, and B-K11 have been described previously (29). Briefly, BALB/c mice were immunized three times at 2-week intervals with 5 g of soluble human gp130. After 4 weeks mice were boosted with 2 g of sgp130 and killed after 4 days. Preparation and screening of hybridomas and production of ascites fluid was performed as described (30). From supernatants mAbs were purified on protein A columns (Bioprobe International, Tustin, CA). The antibodies B-S12-G7 and B-S12-A5 were obtained by subcloning of the B-S12-producing hybridomas.
Generation of Fab Fragments-Antibodies (1 mg/ml) were treated with the protease ficin (0.1 units/ml) (Sigma, Steinheim, Germany) in 50 mM Tris/HCl (pH 7.0), 2 mM EDTA, 10 mM cysteine for 30 min to 6 h. The digestion was stopped by the addition of N-ethylmaleimide (10 mM). 5 mg of protein A-Sepharose (Amersham Pharmacia Biotech) were added and after incubation for 12 h at 4°C the supernatant was used for further studies. Fab fragments prepared were analyzed by gel filtration. About 100 g of B-S12-G7 or B-S12-G7-Fab were loaded onto a Superdex 75 26/60 column (Amersham Pharmacia Biotech) at a flow rate of 2.5 ml/min. 20 g of B-P8 or B-P8-Fab were loaded onto an analytical Superdex 200 column (Amersham Pharmacia Biotech) at a flow rate of 0.8 ml/min. PBS was used as running buffer in all gel filtration experiments.
SDS-Polyacrylamide Gel Electrophoresis and Silver Staining-SDSpoylacrylamide gel electrophoresis was performed under standard conditions (31) using 15% polyacrylamide gels. Nonreducing gel loading buffer was prepared without mercaptoethanol. Probes containing mercaptoethanol were incubated at 95°C for 5 min previous to loading onto the gel. Silver staining of proteins was performed according to Blum et al. (32).
Induction of Acute Phase Protein Synthesis in HepG2 Cells-HepG2 cells were incubated in Dulbecco's modified Eagle's medium/F-12 with IL-6, antibodies, or Fab fragments (concentrations as given in Figs. 2A and 6A) for 16 h and metabolically pulse-labeled with [ 35 S]methionine/ cysteine (Tran 35 S-label) for 4 h. Induction of the acute phase protein ␣ 1 -antichymotrypsin was measured in cell culture supernatants by immunoprecipitation using a rabbit anti-human ␣ 1 -antichymotrypsin antiserum (DAKO, Hamburg, Germany). A 96-well microtiter plate (F96 MaxiSorp immunoplate, Nunc) was incubated with the antiserum (2.5 g/well in PBS). After 12 h, the wells were blocked with PBS containing 1% bovine serum albumin and subsequently cell supernatants were added. After 3 h at 37°C bound ␣ 1 -antichymotrypsin was released using Laemmli gel-loading buffer. Proteins were separated on 15% SDS-polyacrylamide gels and visualized by autoradiography.
Analysis of STAT Activation in HepG2 Cells by Electrophoretic Mobility Shift Assay-HepG2 cells were stimulated at 37°C for 15 min with IL-6 or monoclonal antibodies (concentrations as given in legend to Fig. 2B). Preparations of nuclear extracts and electrophoretic mobility shift assays were performed as described (33). A mutated doublestranded oligonucleotide corresponding to the c-fos promoter (m67SIE: 5Ј-GATCCGGGAGGGATTTACGGGAAATGCTG-3Ј) which provides STAT3-and STAT1-binding sites was used as 32 P-labeled probe. Protein-DNA complexes were separated on a 4.5% polyacrylamide gel containing 7.5% glycerol in 23 mM Tris, 23 mM boric acid (pH 8.0), 0.5 mM EDTA at 20 V/cm for 4 h. Gels were fixed in 10% (v/v) methanol, 10% (v/v) acetic acid in aqeous solution for 30 min and subsequently dried and analyzed by autoradiography.
Flow Cytometry-Ba/F3-gp130 cells or COS7 cells transfected with expression vectors encoding gp130 deletion mutants (for generation of constructs, see Refs. 19 and 20) were collected, washed, and resuspended in cold PBS containing 5% fetal calf serum and 0.1% sodium azide. Subsequently, cells were incubated on ice with 4 g/ml gp130 antibodies or Fab fragments, respectively. Cells were washed with cold PBS/azide and incubated with R-phycoerythrin-conjugated anti-mouse IgG Fab fragment at a 1:50 dilution. Again, cells were washed with cold PBS/azide and then resuspended in 400 l of PBS/azide followed by flow cytometry analysis using a FACScalibur (Beckton Dickinson).
IL-6⅐sIL-6R⅐sGp130 Ternary Complex Formation Assay-To analyze ternary complex formation in the presence of different antibodies, IL-6 (400 ng/ml) was incubated for 2 h at 4°C with biotinylated sIL-6R (12.5 ng/ml), sgp130 (25 ng/ml), and various amounts of antibodies (see Fig.  4, B and C) in PBS containing 1% bovine serum albumin. 96-Well microtiter plates (F96 MaxiSorp immunoplate, Nunc) were coated overnight at room temperature with the monoclonal gp130 antibody B-P4 (0.5 g/well in PBS). The plates were incubated with saturation buffer (0.1 M Tris, 20% sucrose, 0.1% sodium azide, pH 7.7) for 2 h at room temperature. After three washes with PBS, 0.02% Tween 20 was added to the samples and incubated for 2 h at 37°C. Subsequently the plates were washed again and streptavidin poly-horseradish peroxidase (100 ng/ml in PBS, 1% bovine serum albumin) was added and incubated for 45 min at room temperature. After a final wash, substrate solution (100 g/ml tetramethylbenzidine in 0.1 M sodium acetate (pH 5.5) and 0.003% H 2 O 2 ) was added. After incubation for 30 min the color reaction was stopped with 2 M H 2 SO 4 and the absorbance at 450 nm determined using an enzyme-linked immunosorbent assay reader (SLT-Labinstruments, Austria).

B-S12 Consists of Two Different Monoclonal Antibodies-In
previous studies several monoclonal antibodies (mAb) raised against the ectodomain of gp130 have been characterized in respect to their antagonistic or agonistic activities. Among these B-S12 has been described to be a potent agonist exerting IL-6-type cytokine-like responses on cells expressing gp130 (24,29). The mAb B-P8 acted synergistically by supporting the agonistic activity of B-S12. In an epitope competition assay we found that B-S12 displaced B-P8, but B-P8 even at high concentrations did not displace B-S12 (29). Furthermore, using a similar assay Autissier et al. (35) reported that B-S12 recognizes two different epitopes and therefore may be a mixture of two antibodies. These observations prompted us to re-examine the nature of B-S12. After application of small amounts of B-P8 or B-S12 (0.5 g) to SDS-PAGE under reducing conditions the heavy chain and light chain of B-P8 appear as single bands (Fig. 1A, lane 1) whereas the heavy chain of B-S12 appears as a double band of two species of slightly different molecular masses (lane 2). After application of a larger amount of B-S12 also the light chain is clearly visible as a double band of two species showing a more pronounced difference in electrophoretic mobilities (lane 3). After deglycosylation of the antibodies, the heavy chains shift to lower molecular masses but the double band remains suggesting that the two species are not due to differential glycosylation of the B-S12 heavy chain (not shown). These data suggest that B-S12 may indeed represent a mixture of two mAb. Therefore, the B-S12 secreting hybridoma cells were subcloned and the antibodies produced by the individual clones were analyzed. The resulting antibodies can be divided into two groups due to their appearance in SDS-PAGE. In Fig. 1B two representatives of each group are shown (B-S12-A5/B-S12-A9 and B-S12-A12/B-S12-G7, respectively). The electrophoretic mobilities of the heavy and light chains of the individual antibodies when compared with B-S12 indicate that the agonistic antibody previously described is a mixture of two monoclonal antibodies. For further studies one antibody of each group was chosen, namely B-S12-A5 and B-S12-G7.
A Combination of Two Monoclonal Antibodies Is Required to Induce Acute Phase Protein Synthesis and STAT Activation in Hepatoma Cells-To analyze the biological activities of the individual antibodies, HepG2 human hepatoma cells that respond to IL-6-type cytokine stimulation with the production of acute phase proteins were used. HepG2 cells were incubated at different concentrations with single gp130 mAb or a combination of two antibodies. Incubation of HepG2 cells with B-S12 leads to the induction of synthesis of the acute phase protein ␣ 1 -antichymotrypsin that is indistinguishable from stimulation by IL-6 ( Fig. 2A). Even at high concentrations (up to 50 g/ml) neither B-P8 alone nor the B-S12 constituting antibodies B-S12-A5 and B-S12-G7 alone induce prominent activation of ␣ 1 -antichymotrypsin synthesis. The combination of B-S12-A5 and B-S12-G7, however, exerts B-S12-like activity. These findings suggest that two antibodies are required for activation of gp130. Interestingly, the combination of B-P8 and B-S12-G7 was biologically active as well whereas the combination of B-P8 and B-S12-A5 was completely inactive indicating that gp130 is not activated by any combination of two antibodies.
Activation of gp130 by IL-6-type cytokines leads to tyrosine phosphorylation of the transcription factor STAT3 and to a lesser extent STAT1. Upon phosphorylation the STATs homoor heterodimerize, translocate into the nucleus, and bind to enhancers of IL-6-type cytokine target genes. The electrophoretic mobility shift assay presented in Fig. 2B shows that STAT activation in response to stimulation of gp130 with monoclonal antibodies also requires the specific action of two monoclonal antibodies. Neither B-P8 nor B-S12-A5 nor B-S12-G7 alone led to any significant activation of STAT3 or STAT1 in HepG2 cells. Again the combination of B-S12-A5 and B-S12-G7 (that constitute B-S12) and the combination of B-S12-G7 and B-P8 is biologically active whereas the combination of B-S12-A5 and B-P8 failed to induce STAT activation. Combination of B-S12-G7 with another gp130 mAb (B-K11) did not lead to STAT activation indicating that not any pair of gp130 mAb comprising B-S12-G7 is able to activate gp130.
A More Specific Combination of Two Monoclonal Antibodies Is Required to Stimulate Proliferation of Ba/F3-Gp130 Cells-To confirm the biological activities of the antibody combinations, another cellular response to activation of gp130 was FIG. 1. Analysis of gp130 antibodies by SDS-PAGE. A, B-P8 (0.5  g) and B-S12 (0.5 g or 2.5 g); or B, B-S12-A5, B-S12-A9, B-S12-A12, and B-S12-G7 (0.5 g each) were loaded on SDS-polyacrylamide gels as indicated. After electrophoresis proteins were visualized by silver staining. The relative molecular mass (M r ) of marker proteins and their electrophoretic mobilities as well as the location of heavy chains (hc) and light chains (lc) are indicated.

FIG. 2. Agonistic activities of gp130 antibodies on HepG2 cells.
A, HepG2 cells were stimulated for 16 h with IL-6 (20 ng/ml) or antibodies (total concentrations: 5 or 50 g/ml; i.e. in the case of two antibodies: 2.5 or 25 g/ml of each) or left unstimulated (Ϫ) as indicated. After metabolic labeling of proteins with [ 35 S]cysteine/methionine for 4 h, secreted ␣ 1 -antichymotrypsin (␣ 1 -ACT) was immunoprecipitated from cell supernatants, separated by SDS-PAGE, and visualized by autoradiography. B, after stimulation of HepG2 cells with IL-6 (20 ng/ml) or antibodies (total concentrations 20 g/ml) as indicated, nuclear extracts were prepared and analyzed for STAT activation by electrophoretic mobility shift assay using a radioactively labeled probe derived from the c-fos promotor. studied. Ba/F3 cells stably transfected with gp130 (Ba/F3-gp130 (34)) proliferate upon stimulation with IL-6 and soluble IL-6R (sIL-6R). Ba/F3-gp130 cells were incubated with IL-6 and sIL-6R or with different combinations of antibodies and subsequently cell proliferation was measured. Any antibody or antibody combination that was inactive on HepG2 cells also failed to stimulate growth of Ba/F3-gp130 cells (Fig. 3A). Surprisingly, even at high concentrations (up to 333 nM) neither B-S12 (not shown) nor a combination of B-S12-A5 and B-S12-G7 efficiently stimulated growth of Ba/F3-gp130 cells (Fig. 3A) although these antibodies acted agonistically on HepG2 cells. On the other hand, the combination of B-S12-G7 and B-P8 acted as an agonist stimulating Ba/F3-gp130 cells dose dependently (Fig. 3B). At high agonist concentrations, the complexes of IL-6 and sIL-6R or antibody stimuli led to an almost identical maximal response. Half-maximal proliferation was achieved with 0.08 nM IL-6 in the presence of 20 nM sIL-6R or a total antibody concentration of about 0.8 nM using the combination of equimolar amounts of B-S12-G7 and B-P8. Since B-S12-G7 and B-P8 exert the highest agonistic potential that most closely mimicks IL-6 activity, these antibodies were chosen to further investigate the underlying mechanism of gp130 activation.
The Individual Agonistic Monoclonal Antibodies Map to the CBM of Gp130, Partially Neutralize IL-6 Activity on Ba/F3-gp130 Cells, and Interfere with IL-6⅐IL-6R⅐Gp130 Ternary Complex Formation-By immunoblot analysis of recombinant gp130 domains in our previous study (29), B-S12 and B-P8 were mapped to the CBM (D2-D3) of gp130. In order to define the epitopes of the individual mAb B-S12-G7 and B-S12-A5, gp130 deletion constructs were overexpressed in COS-7 cells and binding of the mAb was analyzed by FACS. As a control the gp130 mAb B-P4 that maps to the membrane-proximal part of gp130 (D4-D6) was also included. The data collected by FACS are summarized in Table I. A gp130 construct (D2-D6) lacking the Ig-like domain (D1) is recognized by all mAb indicating that D1 of gp130 is not part of the epitopes. A construct (D4-D6) lacking the membrane-distal part of gp130 is recognized by B-P4 but not by any of the agonistic mAb. A construct comprising the membrane-distal part of gp130 (D1-D3) followed by the transmembrane and cytoplasmic regions, although weakly expressed, is recognized by all agonistic mAb but not by B-P4. From this recognition pattern it can be concluded that all the individual agonistic mAb map to the CBM (D2-D3) of gp130. A competition binding assay based on the replacement of a biotinylated antibody by an excess of unbiotinylated antibody for binding to gp130 revealed that B-S12-A5 and B-P8 recognize overlapping epitopes (Table II).
Since the gp130 CBM is crucial for binding of IL-6-type cytokines we further investigated to what extent the agonistic mAb interfere with IL-6 signaling. Ba/F3-gp130 cells were stimulated with IL-6⅐sIL-6R in the presence of varying amounts of the individual agonistic mAb. The gp130 mAb B-R3 and B-K11 that were previously described to act as an antagonist or to be without biological activity, respectively, were included in the assay. As shown in Fig. 4A, all agonistic mAb significantly impair activation of gp130 by IL-6⅐sIL-6R, whereas B-K11 shows no antagonistic activity. They are less potent antagonists than B-R3 which at a concentration of 20 nM (3 g/ml) completely inhibits proliferation. Even at high concentrations of the agonistic mAb (up to 50 g/ml) no total inhibition of the IL-6 response is achieved. B-R3 shows the lowest IC 50 (2 nM), followed by B-P8 (5 nM) and B-S12-A5 (45 nM). Only at a rather high concentration of B-S12-G7 (333 nM) is a 50% inhibition of the IL-6 response achieved.
Next we investigated whether the IL-6 antagonizing activity of the individual agonistic mAb is due to interference with IL-6⅐IL-6R⅐gp130 ternary complex formation. Soluble gp130 was incubated with B-S12-G7, B-S12-A5, B-P8, B-K11, or without antibody. To allow ternary complex formation, IL-6 and biotinylated sIL-6R were added. The mixture was transferred to an enzyme-linked immunosorbent assay plate precoated with the gp130 mAb B-P4 that maps to the membrane-proximal part of gp130 and therefore does not interfere with ligand binding. After several washing steps detection of biotinylated sIL-6R by the use of avidin conjugated to horseradish peroxidase is indicative of ternary complex formation, since neither in the absence of IL-6 nor sgp130 were significant amounts of sIL-6R detectable (Fig. 4B). In contrast to B-K11 that does not interfere with ternary complex formation, all agonistic mAb compete with IL-6⅐sIL-6R complexes for binding to gp130 (Fig.  4C). The strength of interference correlates with the antagonistic activities of the individual mAb (compare Fig. 4, A and  C). Thus, the agonistic mAb at least partially occupy IL-6⅐IL-6R binding epitopes of gp130. Even at high antibody concentrations residual binding of IL-6⅐sIL-6R complexes is observed; most pronounced in the case of B-S12-G7. This residual activity may be due to the formation of trimeric complexes of lower affinity engaging D1 of gp130, since this domain is not occupied by any of the agonistic antibodies. We conclude from these data that the agonistic mAb activate gp130 by mimicking the action of natural ligands such as the IL-6⅐IL-6R complex.
B-S12-G7 and Fab Fragments of B-P8 Are Sufficient for gp130 Activation-Monoclonal antibodies are believed to activate type I transmembrane receptors by virtue of their bivalency. In order to learn more about the underlying mechanism of gp130 activation by the two mAb B-S12-G7 and B-P8, Fab fragments of both antibodies were prepared. Antibodies were incubated with the protease ficin. After completion of the limited proteolysis, as judged by the disappearance of the heavy chain, the protease was inactivated and the preparation was incubated with protein A-Sepharose to remove Fc fragments and residual undigested antibodies. The quality of the Fab fragments was analyzed by SDS-PAGE under nonreducing and reducing conditions (Fig. 5A), indicating that neither residual intact antibody nor F(ab) 2 fragments were present. Furthermore, the purity and molecular mass of the Fab fragments were analyzed under nondenaturing conditions by size exclusion chromatography. The Fab fragments appeared as monomeric proteins with the expected molecular masses of about 45 kDa (Fig. 5B). By FACS analysis it is shown that the Fab fragments retained their ability to bind to gp130 (Fig. 5C).
HepG2 cells were stimulated with different combinations of antibodies and Fab fragments (Fig. 6A). B-S12-G7 and B-P8-Fab induced ␣ 1 -antichymotrypsin synthesis comparable to the combination of the intact antibodies, whereas B-S12-G7-Fab and B-P8 as well as both Fab fragments did not lead to significant induction of the acute phase protein. A combination of B-S12-G7 and B-S12-A5-Fab is also biologically active on HepG2 cells whereas B-S12-G7-Fab combined with intact B-S12-A5, as well as both Fab fragments, are inactive (not shown). This suggests a similar role of B-P8 and B-S12-A5 in gp130 activation. The fact that B-P8 and B-S12-A5 recognize at least overlapping epitopes (Table II) support this assumption. For the antibody pair B-S12-G7/B-P8 and the derived Fab fragments similar biological activities were observed on Ba/F3-gp130 cells (Fig. 6B): replacement of B-P8 by its Fab fragment retained almost full biological activity of the protein. The other combination of antibody and Fab fragment as well as the two Fab fragments were inactive. We conclude from these findings that the two mAb B-S12-G7 and B-P8 fulfill different functions that are required for receptor activation. DISCUSSION In this study we investigated the mechanism of gp130 activation by monoclonal antibodies. We show that the previously described agonistic antibody B-S12 (24,29) consists of two different species. By subcloning of B-S12, two different monoclonal antibodies, B-S12-G7 and B-S12-A5, were obtained. On the cells used in this study the individual antibodies are devoid of any biological activity, in combination they exert B-S12-like activity. The combination of B-S12-G7 with another gp130 mAb, B-P8, previously described to synergize with B-S12 (29) was even more potent in activating gp130 on different cell types. Autissier et al. (35) generated 37 mAb against the ectodomain of gp130. None of these were able to activate gp130 on myeloma cells whereas some combinations of antibodies were biologically active (35). Therefore, it seems to be a general phenomenon that gp130 is not efficiently activated by a single monoclonal antibody. Up to now, although many monoclonal antibodies to gp130 have been generated and tested, not a single one is known to exert IL-6-like activity on different cells expressing gp130. Thus, in contrast to what has been stated previously based on results obtained with B-S12 (24,29), homodimerization of gp130 by a single monoclonal antibody is not sufficient to efficiently trigger signal transduction.
What is the relationship between activation of gp130 by monoclonal antibodies and by natural ligands such as IL-6? Using deletion constructs of gp130 we have shown that the mabs as analyzed by FACS Gp130 deletion mutants were transiently overexpressed in COS-7 cells and recognition of the proteins by antibodies was analyzed by FACS. Percentage of cells shifting to higher fluorescence intensities compared to untransfected cells (% gated) are given. The schematic drawing on the right represents the domain organization of gp130. Conserved cysteines in domain 2 (D2) as well as the WSXWS motif in domain 3 (D3) are indicated by bars.

TABLE II Competition binding assay
The competition binding assay was performed as described earlier (29). Labeled antibody was either displaced (ϩ) or not displaced (Ϫ) from gp130 by an excess of unlabeled antibody.

Labeled antibody
Unlabeled antibody B-S12-A5 B-S12-A9 B-S12-A12 B-S12-G7 B-P8 agonistic antibodies all map to the CBM of gp130. By mutagenesis studies this region of gp130 was identified to be important for the interaction of gp130 with IL-6⅐IL-6R as well as IL-11⅐IL-11R complexes (20,34). It can easily be envisaged that occupying these domains by the individual agonistic mAb interferes with binding of IL-6⅐IL-6R to gp130 and as a consequence inhibits signal transduction as shown by our experiments. We postulate from the interference of the agonistic mAb with IL-6 responses and ternary complex formation that the antibodies mimic the activity of the cytokine. This explains why only a narrow subset of antibody pairs exert IL-6-like responses. Thus, activation of gp130 by pairs of monoclonal antibodies is a highly specific process. Knowing how mAb activate gp130 may give hints to a better understanding of the mechanism by which gp130 is activated upon binding of the natural ligands.
What is the underlying molecular mechanism leading to activation of gp130 by two monoclonal antibodies? For activation of gp130 by IL-6, three independent receptor-binding sites on the surface of the cytokine are required. Site I binds the IL-6R ␣-subunit, sites II and III recruit the gp130 dimer (for review, see Ref. 36). In the hexameric model of the IL-6-receptor ternary complex consisting of two molecules of each IL-6, IL-6R, and gp130, both gp130 molecules contact site II as well as site III of IL-6 (37). Thus, the gp130 dimer is fixed twice by the natural ligand. This situation is mimicked to some extent by two monoclonal antibodies. In the tetrameric model of the receptor complex consisting of two gp130 and one molecule of each IL-6 and IL-6R, the gp130 dimer is fixed only once since each gp130 molecule contacts IL-6 only once (38). To mimic this scenario dimerization by a single antibody would be sufficient. Independent from these models of receptor activation by the natural ligand other possibilities of receptor activation by two mAb can be envisaged. Two monoclonal antibodies recognizing different epitopes are able to cluster gp130 on the cell surface. This cannot be achieved by a single mAb. Another explanation is that one antibody enforces gp130 dimerization by virtue of its bivalency whereas the second antibody helps to adjust the active conformation of the gp130 dimer. Indeed, by solving the structures of the EpoR with Epo (39) or agonistic (40) and antagonistic peptides (41) it has been shown that the orientation of the receptor chains in the dimer is crucial for signal transduction. Recently it has been suggested for EpoR that the receptor exists as a preformed dimer on the cell surface (42,43). In this case the ligand does not actively induce dimerization, but switches the conformation of the dimer from an inactive to an active state. If gp130 exists as a preformed dimer on the cell surface, it cannot be excluded that bivalency of the antibodies is not required at all. FIG. 4. Inhibitory activities of monoclonal antibodies. A, Ba/F3-gp130 cells were stimulated with IL-6 (5 ng/ml) and sIL-6R (1 g/ml) in the presence of varying amounts (50 -0.024 g/ml; 333-0.16 nM) of the antibodies B-P8, B-S12-G7, B-S12-A5, B-R3, or B-K11. After 72 h a tetrazolium compound was added as a substrate and incubated for 5 h at 37°C. Subsequently, the absorbance at 450 and 690 nm was measured. The difference of absorbances corresponds to the number of metabolically active cells (XTT proliferation assay). 100% correspond to proliferation in the absence of antibody. Ⅺ, B-P8; E, B-S12-G7; ⌬, B-S12-A5; *, B-R3; ϫ, B-K11. B, 12.5 ng/ml (250 pM) biotinylated sIL-6R (bsIL-6R) was incubated either with the combination of IL-6 (400 ng/ml; 20 nM) and sgp130 (25 ng/ml; 250 pM) or with IL-6 or sgp130 alone. Aliquots were transferred to a 96-well microtiter plate precoated with the gp130 mAb B-P4. After several washing steps avidin conjugated to peroxidase was added. Binding of avidin-peroxidase to bsIL-6R was detected by the conversion of a chromogenic substrate that was measured as an increase of absorbance at 450 nm. C, a ternary complex formation assay as described in B was performed in the presence of various concentrations (16.65-1332 pM) of gp130 mAb B-P8, B-S12-G7, B-S12-A5, or B-K11. 100% corresponds to the absorbance in the absence of antibodies (symbols as in A).
To distinguish between the above possibilities Fab fragments of the agonistic gp130 mAb B-S12-G7 and B-P8 were prepared. The Fab fragments retain the ability to bind to gp130 but are devoid of bivalency and therefore do not actively enforce dimerization. Surprisingly, we found that in the presence of intact B-S12-G7 B-P8 can be replaced by its Fab fragment without major loss of bioactivity. The combination of intact B-P8 with a Fab fragment of B-S12-G7 is inactive as well as the combination of both Fab fragments. These findings exclude most of the above possibilities. It seems that the antibody B-S12-G7 actively enforces or strengthens gp130 dimerization because this antibody is inactive when replaced by its Fab fragments. B-P8 does not act by virtue of its bivalency since a Fab fragment is biologically active. We suggest that B-P8 adjusts the active conformation of gp130 that cannot be achieved by B-S12-G7 alone. In analogy we hypothesize that the natural ligand also has to fulfill both objectives: enforcement of gp130 dimerization and adjustment of the active conformation.
Another interesting observation is that the mode of receptor activation restricts the biological response. Both antibody combinations induce acute phase protein synthesis in hepatoma There is no evidence for any contamination of Fab fragments by undigested mAb or F(ab) 2 fragments (expected molecular mass of about 100 kDa). Under reducing conditions the mAb dissociate into heavy (hc) and light chains (lc), the Fabs dissociate into light chain and the digested heavy chain. The light chain of B-S12-G7 is slightly degraded by Ficin treatment without affecting its binding capability (see C). In the Fab preparations no residual intact heavy chain is visible. B, the sizes of the prepared Fab fragments compared with undigested antibodies under nondenaturing conditions were analyzed by size exclusion chromatography. B-S12-G7-Fab (closed line) or B-S12-G7 (dotted line) were separated on a Supderdex 17  . C, the receptor binding capabilities of the Fab fragments were analyzed by FACS using Ba/F3-gp130 cells. Cells were incubated either with phycoerythrin-conjugated secondary antibody alone (black) or with Fab fragments (Fab, dotted lines) or intact antibodies (mAb, closed lines) previous to addition of secondary antibody. The shift of cell counts to higher fluorescence intensities (FL2-H) indicate Fab or mAb binding, respectively. The lower fluorescence intensities after Fab binding is partially due to lower secondary antibody binding capability of Fab fragments compared with mAb.
FIG. 6. Agonistic activities of combinations of gp130 monoclonal antibodies and Fab fragments. A, HepG2 cells were left unstimulated (Ϫ) or were stimulated with IL-6 (25 ng/ml) or combinations of mAb and Fab fragments at different concentrations as indicated (e.g. 10 mg/ml B-S12-G7 ϩ B-P8-Fab means 5 mg/ml of each B-S12-G7 and B-P8-Fab). Secretion of ␣ 1 -antichymotrypsin (␣ 1 -ACT) was analyzed as described in the legend to Fig. 2A. B, Ba/F3-gp130 cells were stimulated with varying amounts of different combinations of mAb and Fab fragments. Total concentrations of 0.0024 -2.5 mg/ml were applied (e.g 2.5 mg/ml of B-S12-G7 ϩ B-P8-Fab means 1.25 mg/ml of each B-S12-G7 and B-P8-Fab). Proliferation of Ba/F3 cells was analyzed as described in the legend to Fig. 3. OE, B-S12-G7 ϩ B-P8; Ⅺ, B-S12-G7-Fab ϩ B-P8; छ, B-S12-G7 ϩ B-P8-Fab; ⌬, B-S12-G7-Fab ϩ B-P8-Fab. cells comparable to IL-6. However, on Ba/F3-gp130 cells only the combination of B-S12-G7 and B-P8 exerts IL-6-like responses whereas the combination of B-S12-G7 and B-S12-A5 even at high concentrations fails do stimulate robust proliferation. Thus, B-S12-G7 and B-S12-A5 selectively induce acute phase protein synthesis in hepatoma cells without stimulating a proliferative response in pro-B cells. One may speculate that the two antibody combinations induce active gp130 dimers of slightly different conformations that influence the accessibility of the cytoplasmic part for the diverse signaling molecules and therefore lead to the observed discrepancies. For instance, tyrosine 759 of the gp130 cytoplasmic part recruits the tyrosine phosphatase SHP2 to the activated receptor that is required for proliferation of Ba/F3-gp130 cells (44) but not for acute phase gene induction in hepatoma cells (45). It is of pharmacological interest that in principal a cytokine receptor can be turned on by artificial agonists in a way that distinct biological responses can be selectively elicited.
One intact antibody and a Fab fragment of a second antibody seem to be the minimal requirement for gp130 activation by antibodies. To our knowledge this has not been reported previously for any other receptor. For other cytokine receptors like the EpoR or GHR it has been shown that they can be activated by single monoclonal antibodies although only a small subset of antibodies are biologically active (25). It is tempting to speculate that the complex ectodomain architecture of gp130 is the reason for the requirement of that special mechanism of activation by antibodies. Autoantibodies that activate receptors may lead to autoimmune diseases. Possibly the ectodomain of the ubiquitously expressed signal transducer gp130 has evolved to prevent stimulation by single antibodies since agonistic autoantibodies to gp130 would lead to dysregulation of gene expression, differentiation, and proliferation in many organs and tissues.