The Immunoglobulin-like Module of gp130 Is Required for Signaling by Interleukin-6, but Not by Leukemia Inhibitory Factor*

The transmembrane protein gp130 is a shared component of the receptor complexes for the interleukin-6 (IL-6)-type cytokines, which include IL-6, leukemia inhibitory factor (LIF) and oncostatin M (OSM). In addition to its role in the generation of high affinity receptors, gp130 is required for signal transduction by these cytokines. In the present study we have examined the role of the N-terminal located, extracellular immunoglobulin (Ig)-like module of gp130 in signal transduction by IL-6 and LIF. We have expressed wild-type human gp130 or three mutants in murine myeloid M1-UR21 cells that lack functional endogenous gp130 but express the IL-6 receptor (IL-6R) and the LIF receptor (LIFR). By measuring cellular responses, such as morphological changes upon differentiation, soft agar colony formation, and induction of tyrosine phosphorylation of the signal transducer and activator of transcription, STAT3, we show that signaling by IL-6, but not LIF, is significantly reduced by mutations in the Ig-like module of gp130. However, the binding of125I-labeled IL-6 or LIF is not affected by these mutations. We also present evidence that the Ig-like module forms part of the epitope of an anti-gp130 monoclonal antibody that neutralizes the bioactivity of IL-6, but not of LIF or OSM. The data suggest that gp130-activation by IL-6 and LIF requires different regions of gp130, that the Ig-like module of gp130 may be required for IL-6-induced gp130 dimerization, and that the stoichiometry of the high affinity IL-6 receptor-complex differs from those of the receptor-complexes for LIF and OSM.

The signal transducer gp130 is the shared receptor subunit, signal transducer, and affinity converter for the interleukin-6 (IL-6) 1 -type cytokines. The members of this family, which are IL-6, leukemia inhibitory factor (LIF), IL-11, oncostatin M (OSM), ciliary neurotrophic factor (CNTF) and cardiotrophin-1 (CT-1), are structurally related and exert many overlapping biological activities, such as the induction of acute phase proteins, regulation of hematopoiesis, and the control of proliferation and differentiation of neural cells (for a review, see Ref. 1). The shared usage of gp130 in part explains their overlapping functions.
In addition to gp130, the high affinity, signaling receptorcomplexes for the IL-6-type cytokines contain at least one other receptor subunit. For example, the biological activities of LIF are mediated by gp130 and the LIF receptor (LIFR) (2), OSM signals via gp130 in combination with either the LIFR (3) or the specific OSMR␤ (4), and IL-6 utilizes gp130 and the specific IL-6R (5). Recent results (6,7) suggest that the high affinity IL-6 receptor-complex consists of two molecules each of gp130, IL-6, and the IL-6R. The receptor-complexes for CNTF (8), and very possibly IL-11 (9), are also hexameric, whereas the high affinity receptor-complex for LIF is trimeric, containing one molecule each of gp130, LIF, and the signaling LIFR (10). The IL-6-type cytokines that induce trimeric receptor-complexes as well as those that induce hexameric complexes activate intracellular signaling through the dimerization of gp130 to a second signaling receptor, either gp130, the LIFR or the OSMR␤ (4,11,12). The formation of a hexameric receptor-complex occurs when the ligand (e.g. IL-6) requires a nonsignaling receptor (the IL-6R) for its association with gp130.
Gp130 belongs to the class I hemopoietin receptors, which are characterized by an extracellular cytokine binding domain (CBD) consisting of two fibronectin type III (FNIII)-like modules, each of ϳ100 residues. The N-and the C-terminal modules of the CBD normally contain four conserved cysteines and a WSXWS motif (where X is any amino acid), respectively (13). The extracellular region of gp130 contains 597 residues and consists of an N-terminal immunoglobulin (Ig)-like module followed by a CBD and three membrane-proximal FNIII modules (14). This region is most closely related to the corresponding region in the receptor for granulocyte-colony stimulating factor (G-CSFR) (15), with an amino acid identity of 26% (46% homology). A truncated form of gp130 lacking the cytoplasmic and transmembrane domains as well as the three membrane-proximal FNIII modules has been shown to bind to IL-6 in a complex with the extracellular domain of the IL-6R (16). Recent studies on the neutralizing effects of anti-gp130 monoclonal antibodies (mAbs) suggested that gp130 has different binding sites for the IL-6-type cytokines and their receptors (17,18); however, the specific binding interfaces on gp130 remain essentially uncharacterized.
In the G-CSFR, both the Ig-like module and the CBD are required for ternary receptor complex formation with G-CSF (19). The importance of the Ig-like module in G-CSF signaling has also been implied from epitope-mapping studies of neutralizing anti-G-CSFR mAbs (20) and from results showing that a G-CSFR mutant lacking most of the Ig-like module was significantly impaired in its ability to transduce a mitogenic response upon stimulation with G-CSF (21). These data, and the fact that gp130 has the same modular structure as the G-CSFR (14,15), led us to propose a model of the hexameric IL-6 receptor complex, in which the Ig-like module of gp130 in one IL-6/IL-6R/gp130 trimer interacts with IL-6 in the opposite trimer (22) (see Fig. 1A). The "bridging" function of the Ig-like module of gp130 in the hexameric IL-6 receptor-complex would require two ligands in the complex, and we therefore predicted that the Ig-like module of gp130 is not involved in the formation of trimeric receptor complexes such as that for LIF (and possibly OSM) (22) (see Fig. 1B).
In the model of the hexameric IL-6 receptor complex, the interactions between IL-6 and the CBDs of the IL-6R and gp130 are based on the trimeric complex of growth hormone (GH) with its two identical receptors (GHR) (23). Using the GH terminology, site I on IL-6 binds the IL-6R, and site II binds gp130. The IL-6-type cytokines, however, have a third receptorbinding site, site III, that has no equivalent in GH (22). Site III in IL-6 (24) binds to gp130, whereas in LIF (25)(26)(27) site III binds to the LIFR. Sites II and III on IL-6 have been shown to interact with two different gp130 molecules in the hexameric receptor complex (7). In our model, the Ig-like module in one IL-6/IL-6R/gp130 trimer exerts a bridging function by interacting with site III on IL-6 in the opposite trimer (Fig. 1A). In this report, using three Ig-mutants of gp130 ((⌬Ig)gp130, which lacks the Ig-like module, (GR-Ig)gp130, a chimera with the Ig-like module of the human G-CSFR, and (ϩFlag)gp130, which has a Flag sequence at its N terminus), we present experimental evidence that the Ig-like module of gp130 is required for signaling by IL-6, but not LIF.

EXPERIMENTAL PROCEDURES
Cells-To generate IL-6-unresponsive cells, M1 cells (previously transfected with the pPGKNeo vector containing the neomycin resistance gene transcribed from the PGK promoter) were subjected to two rounds of mutagenesis with the DNA intercalator ICR-191. Typically, 8 ϫ 10 6 cells were incubated for 1.75 h in a humidified incubator at 37°C and 10% CO 2 in DMEM containing 10% (v/v) fetal bovine serum (FBS) (Trace Biosciences, Castle Hill, NSW, Australia), and 5 g/ml ICR-191 (Sigma-Aldrich, Castle Hill, NSW, Australia). To identify IL-6-unresponsive cells, ICR-191-treated cells (50,000 cells/35-mm dish) (Nunc Inc., Roskilde, Denmark) were grown in soft agar with 100 ng/ml IL-6. Large, tight colonies (consisting of cells that neither differentiated nor were clonally suppressed) were propagated in DMEM containing 10% (v/v) FBS, and 0.4 mg/ml G418 (Life Technologies, Inc.). One such unresponsive clone, the M1-UR21 cell line, lacks functional endogenous gp130 based on the following criteria: the cells did not bind 125 I-labeled LIF with high affinity and failed to react with anti-mouse gp130 antibodies (from T. Hirano and from Santa Cruz Biotechnology), as measured by immunoblotting and using a FACScan flow cytometer (Becton Dickinson, San Jose, CA). Furthermore, responsiveness to IL-6 and LIF could be rescued by transfection of these cells with gp130 cDNA.
Generation and Transfection of gp130 Mutants-An AccII/blunted SacI fragment of pBS130BES containing the cDNA of human gp130 was cloned into the EcoRV site of pcDNA1-Amp (Invitrogen) for mutagenesis by PCR or based on the method of Kunkel et al. (28). PCR reactions were performed on a PCR System 9600 thermal cycler (Perkin Elmer). The conditions used for mutagenesis were: 34 cycles of 20 s at 95°C (320 s in the first cycle), 20 s at 52°C, and 60 s at 72°C (360 s in the last cycle).
To generate (GR-Ig)gp130, the primary sequences of the Ig-like modules of gp130 and the human G-CSFR were aligned manually, and an EcoR V restriction enzyme site was introduced in corresponding positions in the cDNAs. In gp130, the EcoRV site was introduced at Gly 123 -Leu 124 (numbering as in Ref. 14) using the oligonucleotide (5Ј) GGAAT-CACAATAATTAGCGGATATCCTCCAGAAAAA (3Ј). The (5Ј) EcoRV/ (3Ј) XbaI fragment of gp130 encoding the remaining 3Ј residues to Gln 918 was ligated to a (5Ј) XbaI/(3Ј) EcoRV fragment of the G-CSFR, which encodes the predicted signal sequence and residues Glu 24 -Tyr 121 (20) (our numbering includes the 23-residue signal sequence and differs from Larsen, et al., Ref. 15), and cloned into XbaI-digested pEF-BOS (29) for expression. (GR-Ig)gp130, which was obtained following "domain swapping," thus contains residues Glu 24 -Tyr 121 of the G-CSFR, and Pro 125 -Gln 918 of gp130.
For expression in M1-UR21 cells, the cDNA of wt-gp130 was excised from pcDNA1-Amp using BamHI and cloned into the XbaI site of pEF-BOS using BamHI/XbaI adapters. (⌬Ig)gp130 and (ϩFlag)gp130 were inserted into BamHI-digested pEF-BOS. All constructs were verified by DNA sequencing using a model 370A DNA sequencer (Applied Biosystems, Foster City, CA). The restriction enzymes were from Boehringer Mannheim (Germany) and New England Biolabs (Beverly, MA). Typically, 20 g of pEF-BOS, containing a gp130 construct, and 2 g of pPGKPuropA vector, containing the puromycin resistance gene transcribed from the PGK promoter, were co-transfected into 4 ϫ 10 6 M1-UR21 cells, using gene pulsar cuvettes with 0.4 cm electrode gap (Bio-Rad) in a Bio-Rad gene pulser at 270 V and 960 microfarads. The cells were resuspended in 100 ml of DMEM containing 10% (v/v) FBS and plated into four 24-well dishes (Nunc). After 2 days in a humidified incubator at 37°C and 5% CO 2 , puromycin (Sigma-Aldrich, Castle Hill, NSW, Australia) was added to each well at a final concentration of 10  (22). B, the trimeric LIF receptor complex. For clarity of comparison with the IL-6 receptor complex, the receptor interaction involving site III on LIF has not been indicated. OSM may form a similar trimeric complex with gp130 and the LIFR or the specific OSMR␤. g/ml. Independently derived cell lines that were viable after 10 -12 days of selection were expanded.
For analysis of morphological changes upon differentiation, cells were plated in 6-well plates (Nunc) (10,000 cells in 2 ml/well) in DMEM containing 10% (v/v) FBS and various concentrations of mouse IL-6. After 5 days in a humidified incubator at 37°C and 5% CO 2 , the cells were analyzed by flow cytometry for IL-6-induced increases in forward and side scatter (FSC and SSC). Data from 10,000 events were collected.
Soft Agar Assay-The cells were plated into 35-mm culture dishes (Nunc) (ϳ 150 cells in 1 ml/dish) in DMEM containing 20% (v/v) FBS and 0.3% (w/v) dissolved bacto-agar (Difco). Prior to addition of the cell suspension, 0.1 ml of serial dilutions of mouse IL-6 or LIF, or medium as a negative control, were added to the dishes. After 7 days in a humidified incubator at 37°C and 5% CO 2 , the colonies were counted and scored for differentiation. Colonies with a halo of dispersed cells or composed entirely of dispersed cells were scored as differentiated.
STAT3 Phosphorylation Assay-The cells were serum-starved for 5 h in DMEM in a humidified incubator at 37°C and 5% CO 2 and stimulated with factor at 37°C. Immunoprecipitation and immunoblotting of the cell lysates was performed essentially as described (31).
Purification and 125 I-labeling of Cytokines-Mouse IL-6 was produced as described (32), mouse LIF was from AMRAD Pharmacia Biotech (Boronia, Victoria, Australia), and human OSM was from Pepro Tech Inc. (Rocky Hill, NJ). The purity of IL-6 was confirmed by Nterminal amino acid sequencing and mass spectrometry, and the protein concentration was determined by amino acid analysis. Mouse LIF was radiolabeled using a modified iodine monochloride method as described (33), and mouse IL-6 was radiolabeled using di-125 I-Bolton-Hunter reagent (NEN Life Science Products) (34). The specific activities were determined by self-displacement analysis (35) and varied from 0.5-1.2 ϫ 10 6 cpm/pmol.
Equilibrium Binding Assay-The assays were performed essentially as described (33). The cells (2-12 ϫ 10 6 cells/data point in duplicates) were incubated for at least 3 h on ice with 125 I-labeled factor in the presence of 0 -25 nM of unlabeled competitor, or with various concentrations of 125 I-labeled factor in the absence of competitor, in a total volume of 100 l. Nonspecific binding was defined as binding in the presence of 500 nM of unlabeled competitor. The data were examined using the curve-fitting program LIGAND (36).

Generation and Expression of gp130 Constructs in M1-UR21 Cells
To examine the involvement of the Ig-like module of gp130 in signaling by IL-6-type cytokines, the following human gp130 constructs were generated: (⌬Ig)gp130, which lacks the Ig-like module, (GR-Ig)gp130, a chimera with the Ig-like module of the human G-CSFR, and (ϩFlag)gp130, which has a Flag sequence at its N terminus (Fig. 2) (see also "Experimental Procedures").
Wild-type gp130 and the mutants were subcloned into the expression vector pEF-BOS and transfected into murine myeloid M1-UR21 cells, as described under "Experimental Procedures." In contrast to M1 cells, which differentiate into macrophages and undergo clonal suppression in response to IL-6 and LIF (37), M1-UR21 cells are unresponsive to these cytokines because of lack of functional endogenous gp130 (see "Experimental Procedures" and Fig. 4C). The M1-UR21 cells do, however, express the IL-6R and the LIFR as well as the intracellular machinery necessary for the physiological response to IL-6 and LIF (data herein). Transfection of wt-gp130 into M1-UR21 cells thus resulted in restoration of IL-6-and LIF-responsiveness, and mutations in gp130 that influence the bioactivity of these cytokines can readily be identified in this cell system.
Independently derived viable cell lines were expanded and analyzed for expression of gp130 by flow cytometry using anti-gp130 mAbs and isotype-matched mAbs as negative controls. Cells transfected with wt-gp130 or the mutants, but not nontransfected M1-UR21 cells (cf. Fig. 7), all yielded cell lines expressing anti-gp130 mAb-reactive protein on the surface, as exemplified using mAb B-T6 (Fig. 3). Additionally, (ϩFlag)gp130and (GR-Ig)gp130-expressing cell lines were also reactive with an anti-Flag mAb and a conformation-sensitive mAb recognizing the G-CSFR-Ig like module, respectively (Fig. 3).

The Ig-like Module of gp130 Is Involved in Signaling by IL-6, but Not by LIF
Vacuolization Assay-Three or four independently derived cell lines from each transfection with wt-gp130 or the mutants were stimulated for 5 days with various concentrations of IL-6, prior to measurement by flow cytometry of cell size and vacuolization (Fig. 4). Increases in these two parameters correlate with differentiation of the cells into macrophages and are reflected in increases in FSC and SSC, respectively (Fig. 4A). Maximal stimulation of M1-UR21 cells transfected with wt-gp130 was seen with 30 ng IL-6/ml and resulted in a shift of Ͼ65% of cells from the lower left (LL) quadrant to the upper left (UL), upper right (UR), and lower right (LR) quadrants (Fig. 4B). Increased cell size and vacuolization of the wt-gp130 transfectants was also seen at lower concentrations of IL-6 (EC 50 ϭ 1.4 ng/ml), with shifts of approximately 60, 50, 25, and 3% at 10, 3, 1, and 0.3 ng/ml, respectively. Stimulation of (ϩFlag)gp130 and (⌬Ig)gp130 transfectants with 30 ng IL-6/ml resulted in shifts of approximately 20% (S.E. ϭ 3.5%) and 25% (S.E. ϭ 14.5%), respectively, and smaller shifts were obtained with lower concentrations of IL-6 (Fig. 4B). Cells transfected with (GR-Ig)gp130, however, showed no significant changes in morphology at 0 -30 ng of IL-6/ml (Fig. 4B). In a separate experiment, duplicate samples of one cell line of (GR-Ig)gp130 were stimulated with a higher dose of IL-6. Whereas the positive control (wt-gp130 transfectants) differentiated at near maximal level in response to 20 ng of IL-6/ml, (GR-Ig)gp130, and nontransfected M1-UR21 cells as a negative control, remained undifferentiated even at 800 ng of IL-6/ml (Fig. 4C).
Soft Agar Assay-Two representative independently derived cell lines from each transfection were chosen for further analysis of differentiation capacity in a soft agar colony assay (Fig.  5). M1-UR21 cells transfected with wt-gp130 fully differentiated in response to IL-6 (EC 50 ϭ 0.4 ng/ml) (Fig. 5A). Cells expressing (ϩFlag)gp130 were over 10-fold less responsive to IL-6 (EC 50 ϭ 4.8 ng/ml). The (⌬Ig)gp130 transfectants, at 820 ng of IL-6/ml (a concentration corresponding to Ͼ2000-fold the EC 50 of wt-gp130 transfectants), showed approximately 50% of the maximal differentiation of wt-gp130 transfectants, but maximal stimulation (and therefore their EC 50 ) was not obtained within the concentration range of IL-6 in this assay. (GR-Ig)gp130 transfectants were essentially unresponsive to IL-6, their differentiation being indistinguishable from background levels at all concentrations of IL-6 assayed (Fig. 5A).
STAT 3 Phosphorylation Assay-One cell line from each transfection was also analyzed for intracellular signaling, as measured by the inducible tyrosine phosphorylation of the signal transducer and activator of transcription, STAT3 (Fig. 6). The activation of STAT3 is a critical event in gp130-mediated terminal differentiation of M1 cells (38,39). When stimulated with IL-6, cells transfected with wt-gp130, but not (GR-Ig)gp130, yielded induction above background of tyrosine phosphorylated STAT3 (Fig. 6). (ϩFlag)gp130 transfectants also responded to IL-6, albeit considerably weaker than wt-gp130, whereas IL-6 induced little or no detectable tyrosine phosphorylation of STAT3 in the (⌬Ig)gp130 transfectants. Upon stimulation with LIF or OSM, wt-gp130 and the gp130 mutants showed similar levels of induction above background of tyrosine phosphorylated STAT3 (Fig. 6).
Equilibrium Binding-One cell line from each transfection was assessed for binding of 125 I-labeled IL-6 or 125 I-LIF in the presence of unlabeled competitor. Scatchard transformation of the data yielded a single class of IL-6-binding sites on wild-type M1 cells (K D ϭ 5.6 ϫ 10 Ϫ10 M) and on M1-UR21 cells transfected with wt-gp130, (ϩFlag)gp130, (⌬Ig)gp130 or (GR-Ig)gp130 (K D ϭ 2.8, 2.2, 3.5, and 7.9 ϫ 10 Ϫ10 M, respectively) ( Table I). The preparation of 125 I-IL-6 that was used for these experiments bound to mouse hybridoma 7TD1 cells with K D ϭ 5.1 ϫ 10 Ϫ10 M, however, no low affinity binding sites for IL-6 could be detected above background on nontransfected M1-UR21 cells (Table I).

The Ig-like Module of gp130 Forms Part of the Epitope of mAb B-T2
The anti-gp130 mAb B-T2 has been recently shown to inhibit the bioactivity of IL-6 on the human erythroleukemia cell line TF1 (18), but its epitope remains to be determined. Further evidence for the importance of the Ig-like module of gp130 in IL-6 signaling comes from flow cytometry studies measuring the binding of mAb B-T2 to M1-UR21 cells transfected with wt-gp130 or the mutants (Fig. 7). As a control in these experiments, nontransfected M1-UR21 cells were incubated with mAb B-S12, which has the same isotype as B-T2 and recognizes the C-terminal part of the CBD of gp130 (17). As expected, mAb B-S12 failed to react with the nontransfected cells, but it rec- ognized the M1-UR21 cells transfected with either wt-gp130 or the mutants (Fig. 7, panels A-D). In contrast, mAb B-T2 recog- but express the IL-6R, the LIFR, and the intracellular signaling molecules necessary for a biological response to these cytokines. Upon stimulation with IL-6, the wt-gp130-transfected cells differentiated into macrophages, as shown by vacuolization and soft agar colony assays, and responded by activation of intracellular signaling, as measured by tyrosine phosphorylation of STAT3. Cells transfected with (ϩFlag)gp130 or (⌬Ig)gp130, however, were significantly impaired in IL-6-responsiveness, and (GR-Ig)gp130 was essentially unresponsive to IL-6. The Ig-like module of the latter was recognized by a conformation-dependent mAb (20), suggesting that the lack of IL-6-responsiveness was not mainly because of gross misfolding. Moreover, because the LIF-responsiveness of the Ig-mutants was similar to wt-gp130, it appears unlikely that their decreased IL-6-responsiveness was caused by gross structural perturbations elsewhere in the molecules; LIF could thus be seen as a positive control in these assays. The data show that the Ig-like module of gp130 is required for signaling by IL-6, but not LIF.
The Ig-mutants had a similar affinity for binding IL-6 as did the wt-gp130 transfectants, implying that the lack of IL-6responsiveness of (GR-Ig)gp130 cells was not because of the absence of IL-6R. Scatchard analyses of IL-6-binding to the transfectants yielded K D ϭ 2.2-7.9 ϫ 10 Ϫ10 M, in good agree-ment with K D ϭ 5.6 ϫ 10 Ϫ10 M on M1 cells. According to the literature, the binding of mouse IL-6 to mouse target cells can be described by a single site model with a (cell type-dependent) K D ϭ 10 Ϫ10 -10 Ϫ11 M (40). The preparation of 125 I-IL-6 that was used for the binding experiments on the transfectants bound to mouse 7TD1 cells with a similar affinity (K D ϭ 5.1 ϫ 10 Ϫ10 M) to that obtained by Coulie et al. (40) (K D ϭ 9 ϫ 10 Ϫ10 M). Based on the following, it is likely that the K D ϭ 2.2-7.9 ϫ 10 Ϫ10 M of IL-6 binding to the transfectants (or the M1 cells) corresponds to high affinity binding. 1) The affinity of IL-6 binding to nontransfected M1-UR21 cells, which express only low affinity receptors (IL-6R), was too low for an accurate estimation of the K D . However, upon transfection of wt-gp130 or the Ig-mutants, measurable binding of IL-6 was obtained (Table I). 2) The transfected cells express fewer binding sites for IL-6 than LIF ( Table I), suggesting that the IL-6R is the limiting component.
3) The major part of 125 I-IL-6 that was specifically bound to M1 cells dissociated slowly (data not shown), consistent with the characteristics of high affinity binding (33). Scatchard analyses of the binding of LIF to M1 cells and the transfectants yielded K D ϭ 0.5-1.2 ϫ 10 Ϫ10 M, similar to K D ϭ 0.3 ϫ 10 Ϫ10 M previously determined for this ligand on M1 cells (33). Taken together, these studies suggest that mutations in the Ig-like module of gp130 do not significantly affect IL-6 nor LIF binding and that human gp130 can substitute for mouse gp130 without significant loss of affinity.
The biological activities of IL-6 on the Ig-mutants are in agreement with the model of the hexameric IL-6 receptor complex shown in Fig. 1A. The absence of an Ig-like module may reduce the interactions within the complex, and the highly negatively charged Flag-peptide may induce charge repulsion, in both cases without fully preventing the formation of the signaling complex. The substitution of the Ig-like module of gp130 for that of the G-CSFR may result in charge repulsion and/or steric hindrance within the complex and fully impair the formation of the signaling complex. The binding of IL-6 to the mutants, however, appears not to support a model wherein the Ig-like module binds to (site III on) IL-6. Instead, the data suggest that this module may be required for IL-6-induced gp130 dimerization, perhaps by stabilizing the signaling IL-6 receptor-complex and/or bringing together two IL-6/IL-6R/ gp130 trimers. It could thus be envisaged that the Ig-like module forms part of the gp130 homodimerization interface (for example by dimerizing with the Ig-like module of gp130 in the opposite IL-6/IL-6R/gp130 trimer), or that it binds to the IL-6R in the opposite trimer (cf. Fig. 1A). Such interactions would also be consistent with the results showing that it is not required for LIF-signaling (cf. Fig. 1B). Studies are ongoing to determine the interactions of the Ig-like module of gp130 within the signaling IL-6 receptor complex and to identify the site on gp130 that interacts with site III on IL-6 (7,24).
In addition to the Ig-like module of gp130, the formation and the stability of the IL-6 receptor-complex is dependent on interactions involving, for example, the CBDs of gp130 and the IL-6R (42)(43)(44), and possibly the membrane-proximal FNIII modules of gp130. 2 Homodimerization of gp130 is a prerequisite for IL-6-induced signaling across the cell membrane (11,12), suggesting that the cytoplasmic and/or transmembrane domains of gp130 participate in stabilizing interactions within the signaling IL-6 receptor complex. We have recently shown that an N-terminal Flag-tagged "soluble" form of gp130 (sgp130-FLAG) is deficient in the formation of a hexameric IL-6 receptor complex. 2,3 These results are consistent with the re- a The data represent the mean value of at least two assays on one independently derived cell line. The assays were performed as described under "Experimental Procedures." b The data represent the mean value of at least two assays on one independently derived cell line (except for (ϩFlag)gp130, where only one assay was performed). The assays were performed as described under "Experimental Procedures." c ND, not detectable. were incubated with the isotype-matched anti-gp130 mAbs B-T2 (shaded graphs) or B-S12 (thin lines), and binding was measured by flow cytometry as described under "Experimental Procedures." Nontransfected M1-UR21 cells incubated with mAb B-S12 (solid lines) are shown as a control. duced IL-6 responsiveness of the M1-UR21 cells transfected with the Ig mutants of gp130, indicating that the interactions within the signaling IL-6 receptor complex are similar to those in the hexameric solution complex. The deficiency in hexameric receptor complex formation by sgp130-FLAG may therefore, in part, be because of the absence of the transmembrane and cytoplasmic domains of gp130. Conversely, the IL-6-responsiveness of the (ϩFlag)gp130 transfectants, which is significantly reduced compared with wt-gp130 transfectants, may involve the formation of transiently stabilized signaling receptor complexes, which can tentatively be explained by the homodimerization of the transmembrane and cytoplasmic domains of gp130. A similar observation has been made previously with (Q159E, T162P)IL-6, a site III mutant of IL-6, which is deficient in the formation of a stable hexameric receptor-complex (45), but has residual biological activity on human hepatoma HepG2 cells (24).
The IL-6-type cytokines may be categorized as those that are thought to form hexameric receptor complexes and those that may form trimeric complexes (22). The former are IL-6 (6, 7), CNTF (8), and presumably IL-11 (9). This group may also include CT-1 (22). The latter are LIF (10) and OSM. In this report we showed that the Ig-like module forms part of the epitope of the anti-gp130 mAb B-T2. This mAb neutralizes the mitogenic activities on human TF1 cells of the cytokines that form hexameric receptor complexes but not of those that form trimeric complexes (18). Whereas the hexameric IL-6 receptor complex is supported by stoichiometry analyses (6, 7), a tetrameric IL-6 receptor complex without such experimental foundation has also been proposed (46). Similar to the Paonessa model of the hexameric complex (7), the tetrameric model (consisting of one molecule each of IL-6 and IL-6R, and two molecules of gp130) does not invoke a role for the Ig-like module of gp130. In this model, sites I and II on IL-6 bind to the CBD of one IL-6R and one gp130, respectively, and the second molecule of gp130 interacts with site III on IL-6 through an undefined interface (46). The data on mAb B-T2 are, therefore, in agreement with a model of the hexameric IL-6 receptor complex wherein the Ig-like module is required for complex formation.
OSM signals via heterodimerization of gp130 with either the LIFR (3) or the specific OSMR␤ (4). A specific nonsignaling receptor, the equivalent of the IL-6R, appears not to be necessary for OSM receptor complex formation, suggesting that it forms trimeric rather than hexameric receptor complexes. We have shown that the Ig-like module of gp130 is not required for the biological effects of OSM, supporting the notion that this cytokine induces a trimeric OSM/LIFR/gp130-or OSM/ OSMR␤/gp130-signaling receptor complex similar to the trimeric LIF receptor complex (cf. Fig. 1B).
The activation of gp130 by IL-6 and LIF requires different regions of gp130. It remains to be determined whether, in addition to the Ig-like module, there are other modules in the extracellular domain of gp130 that are selectively utilized by the members of the IL-6-type cytokines and their receptors. IL-6 is involved in a variety of neoplastic and chronic diseases such as plasmacytoma/myeloma, Castleman's disease, and rheumatoid arthritis (41). The Ig-like module of gp130 is a possible target for specific antagonists of IL-6 activity.