Determination of the Disulfide Structure andN-Glycosylation Sites of the Extracellular Domain of the Human Signal Transducer gp130*

gp130 is the common signal transducing receptor subunit for the interleukin-6-type family of cytokines. Its extracellular region (sgp130) is predicted to consist of five fibronectin type III-like domains and an NH2-terminal Ig-like domain. Domains 2 and 3 constitute the cytokine-binding region defined by a set of four conserved cysteines and a WSXWS motif, respectively. Here we determine the disulfide structure of human sgp130 by peptide mapping, in the absence and presence of reducing agent, in combination with Edman degradation and mass spectrometry. Of the 13 cysteines present, 10 form disulfide bonds, two are present as free cysteines (Cys279 and Cys469), and one (Cys397) is modified byS-cysteinylation. Of the 11 potentialN-glycosylation sites, Asn21, Asn61, Asn109, Asn135, Asn205, Asn357, Asn361, Asn531, and Asn542 are glycosylated but not Asn224 and Asn368. The disulfide bonds, Cys112–Cys122 and Cys150–Cys160, are consistent with known cytokine-binding region motifs. Unlike granulocyte colony-stimulating factor receptor, the connectivities of the four cysteines in the NH2-terminal domain of gp130 (Cys6–Cys32 and Cys26–Cys81) are consistent with known superfamily of Ig-like domains. An eight-residue loop in domain 5 is tethered by Cys436–Cys444. We have created a model predicting that this loop maintains Cys469 in a reduced form, available for ligand-induced intramolecular disulfide bond formation. Furthermore, we postulate that domain 5 may play a role in the disulfide-linked homodimerization and activation process of gp130.

The family of cytokines that signal through the common receptor subunit gp130 (referred to as the gp130 cytokines) comprises interleukin (IL) 1 -6, leukemia inhibitory factor (LIF), ciliary neurotrophic factor, oncostatin M, cardiotrophin-1, herpesvirus IL-6, interleukin-11 (IL-11), and neurotrophin-1/B cell-stimulatory factor-3/cardiotrophin-like cytokine. These cytokines play a pivotal role in the immune, hematopoietic and nervous systems, as well as in bone metabolism, inflammation, and the acute-phase response (1)(2)(3)(4). The gp130 cytokine family are characterized by a four-␣-helix bundle structure, the helices being connected in an up-up-down-down arrangement by three polypeptide loops (3,5). These cytokines signal through their respective receptor systems that comprise a ligand-specific ␣-subunit and the common signal-transducing ␤-subunit, gp130. The shared use of gp130, in part, provides a molecular basis for the functional redundancy of the gp130 cytokines. The result of gp130-mediated signaling is the regulation of a variety of complex cellular processes such as proliferation, differentiation, and gene activation in a wide variety of adult tissue systems, owing to the ubiquitous expression of gp130 (6). Insights into the possible in vivo role of gp130 have come from transgenic and knock-out animal studies. Targeted inactivation of the gp130 gene results in a prenatal lethal phenotype that includes defects in the cardiac and hematopoietic systems (7)(8)(9). Chronic activation of gp130 signaling in a transgenic mouse model results in cardiac hypertrophy (10).
Intracellular signaling by the gp130 cytokines is initiated by the ligand first making low affinity (ϳ1 nM) contact with its cognate receptor ␣-subunit, which then recruits the signal transducing gp130 subunit with a resulting ligand-binding affinity of ϳ10 pM. Importantly, receptor activation depends on homodimerization or heterodimerization of gp130 (1,3). gp130 was initially cloned as a component of the IL-6 receptor complex (11). The binding of IL-6 to its cognate receptor subunit (IL-6R) induces the dimerization of gp130 (12) and the formation of a hexameric complex comprising two molecules each of IL-6, IL-6R, and gp130 (13,14). The IL-11 receptor complex is similar (15), whereas in the case of LIF (16), oncostatin M (17,18), cardiotrophin-1 (19), and ciliary neurotrophic factor (16), ligand binding induces heterodimerization of gp130 and the LIF receptor, another signal transducing subunit (20). Soluble forms of many of the gp130 cytokine receptors, including gp130, have been found in body fluids of different mammalian species (3). Soluble gp130 (sgp130) has been detected in human serum (21) and is most likely translated from alternatively spliced mRNA (22). sgp130 can neutralize IL-6⅐sIL-6R complexes, thereby acting as an antagonist (21).
The cDNA of human gp130 encodes a protein of 918 amino acids (11), including a signal peptide of 22 amino acids, an extracellular domain of 597 amino acids (sgp130), a transmembrane domain of 22 amino acids, and an intracellular domain of 277 amino acids. The amino acid sequence of murine (23) and rat gp130 (24) share an overall sequence identity of 85 and 88%, respectively, with that of human gp130. There are 11 potential N-linked glycosylation sequons in the extracellular domain of the predicted 101-kDa human gp130 (11), suggesting that the mature ϳ130-kDa gp130 (11) is highly glycosylated. The extracellular region of gp130 has a modular structure consisting of six domains of ϳ100 amino acids each. The NH 2terminal domain (domain 1) is predicted to be a member of the immunoglobulin superfamily (Ig-like) (25,26), which has "Greek key" ␤-sheet topology (27). In the Ig-like fold, neighboring ␤-strands form hydrogen bonds in an antiparallel fashion to form a ␤-pleated sheet, and two ␤-sheets pack against each other to produce a hydrophobic core. Domains 2-6 of the gp130 are classified as fibronectin type III (FN III)-like modules, a subclass of the ␤-sandwich fold (26). The topology of these domains is similar to those of the Ig-like modules, the notable exception being the "sheet switching" of ␤-strand D from the first ␤-sheet of an Ig-like domain to form ␤-strand CЈ on the second ␤-sheet of FN III domains. The two membrane distal FN III domains (domains 2 and 3) form the cytokine-binding region (CBR) that is characteristic of class I cytokine receptors (28,29). CBRs are characterized by two conserved disulfide bonds in their NH 2 -terminal domain and a WSXWS motif in their COOH-terminal domain.
The functional anatomy of the extracellular region of gp130 is still poorly understood. A truncated form of gp130 lacking the membrane-proximal FN III modules and the cytoplasmic and transmembrane domains has been shown to bind a complex of either IL-6⅐sIL-6R or LIF⅐sLIF receptor (30,31). Val 252 in the BC loop of the COOH-terminal domain of the CBR (domain 3 of gp130) has been implicated in the interaction of gp130 with IL-6⅐IL-6R (32) and IL-11⅐IL-11 receptor (33). The location of this residue corresponds to a tryptophan residue in the growth hormone receptor that has been shown to be critical for ligand binding (34), suggesting a conserved mode of ligand binding among the cytokine receptor superfamily. The Ig-like module of gp130 (domain 1) has been shown to be involved in the interaction of gp130 with ligands that induce homodimerization of gp130 (IL-6 (35, 36) and IL-11 (37)) and essential for the formation of high affinity hexameric complexes (15,35). This domain is thought to bind site III in IL-6 and IL-11 (35), and Ig-like modules in the GCSFR and LIF receptor are also thought to make contact with their cognate ligands (38,39). The three membrane-proximal FN III modules of gp130, although not directly involved in ligand binding, may be required for gp130 dimerization by ligands such as IL-6 (40) and IL-11 (41) or transmembrane signaling events such as stabilization and/or orientation of transmembrane receptor dimers (5).
The structure of the COOH-terminal domain of the gp130 CBR (domain 3) has been determined by NMR (42), and the complete CBR (i.e. domains 2 and 3) by x-ray crystallography (43); however, there is no complete structure for gp130 or any member of the gp130 family of receptors. To elucidate the tertiary structure of the gp130 extracellular region, we have purified human sgp130 using a Chinese hamster ovary (CHO) cell expression system (13,44). This form of human sgp130 contains 11 potential N-linked glycosylation sites and 13 cysteine residues. Four cysteines are located in the Ig-like domain (domain 1), four are in the NH 2 (45) containing three ␤-N-acetylglucosidase F (endo F) activities (F1, F2, and F3) as well as peptide N-glycosidase (46,47) was a kind gift from Dr. G. E. Norris (Massey University, New Zealand). HPLC grade solvents were from Mallinckrodt, and all other buffers and reagents (Analar grade) were from BDH. Anti-human gp130 monoclonal antibodies AM64, GPZ35, GPX7, and GPX22 (11,48) were from Dr. K. Yasukawa (TOSOH, Tokyo, Japan). All buffers and solutions were prepared with deionized water purified and polished by a tandem Milli-RO and Milli-Q system (Millipore).
Purification of the Extracellular or Soluble Domain of the Recombinant IL-6 Signaling Receptor gp130 (sgp130)-sgp130 was purified from the conditioned medium of CHO (G16) cells stably transfected with a plasmid (pECEdhfrgp620) that encodes the extracellular domain of gp130 truncated at amino acid 621 (44). Briefly, transfected CHO cells were grown in a large scale fermentation apparatus with a working volume of 1.25 liters (New Brunswick Celligen plus fermenter). Cell conditioned medium was concentrated (20-fold) by ultrafiltration using a Sartocon Mini apparatus (Sartorius, Germany) fitted with a 30,000 molecular weight cut-off filter. sgp130 was purified from concentrated CHO cell medium by binding to a 10-ml column of AM64-Sepharose and eluting sgp130 with 4 M MgCl 2 . sgp130 was further purified by preparative SEC (Ͼ95% pure, as judged by SDS-PAGE and Western blot analysis), and sgp130-containing fractions were pooled, and buffer was exchanged using a Fast-desalting TM column prior to storage at Ϫ20°C.
Labeling of Free Cysteine Residues in sgp130 -Prior to disulfide determination, free cysteine residues in sgp130 (400 g) were treated with a 5-fold molar excess of 4-vinylpyridine in 0.1 M Tris-HCl, pH 8.4, for 1 h at 25°C in the dark. The modified protein was buffer exchanged using a Fast-desalting TM column operated at 1 ml/min. 4-Vinylpyridine was chosen as the alkylating agent because S-␤-(4-pyridylethyl)-cysteine (Pec) containing peptides can be identified by their characteristic absorption spectra at 254 nm (51) and are also readily identified during ESI-MS by the presence of an ion of m/z 106 following collision-induced dissociation (CID), characteristic of the protonated S-pyridylethyl moiety (50).
Peptide Mapping of sgp130 -sgp130 (400 g) in 2 ml of 1% (w/v) ammonium bicarbonate, containing 2 mM calcium chloride, was digested at 37°C for 16 h with either trypsin or chymotrypsin at an enzyme to substrate ratio of 1:20. For pepsin digestion, sgp130 (400 g, ϳ4 nmol) in 5% formic acid was digested for 1 h at 37°C using an enzyme to substrate ratio of 1:20. Resultant peptides were fractionated by RP-HPLC on either a Vydac C18 column or a Brownlee RP-300 column at 0.5 ml/min at 45°C. The column eluent was split (ϳ1:160) post-detector, using a stainless steel Tee-union (52), whereas the remainder (99.4%) was collected for further analysis.
For disulfide-containing peptide identification, a portion of the digest (25%) was reduced with 10 mM DTT at 45°C for 1 h and rechromatographed under identical conditions. Peaks whose retention times shifted upon reduction were subjected to NH 2 -terminal sequence analysis.
Electrospray Mass Spectrometry-On-line MS analysis of peptide fractions was performed using either a Finnigan LCQ quadrupole ion trap mass spectrometer or a TSQ-700 triple quadrupole mass spectrometer, both equipped with an ESI source as described (50,53). S-Pyridylethylated peptides were identified by parent-ion scanning using the TSQ mass spectrometer by monitoring all peptides for the labile protonated S-pyridylethyl group (m/z 106) following CID. Source CID/single ion monitoring using the LCQ ion trap mass spectrometer was employed to identify S-pyridylethyl cysteine-containing peptides as described (52). Peptides were identified from their CID product ion spectra using either the Finnigan Xcalibur-Biomass software or Protein Prospector MS-Product algorithm (ProteinProspector Pacific-Rim mirror).
Homology Modeling-Domain 5 of sgp130 was modeled using the two FN III domains from the structure of the cytoplasmic tail of human integrin ␣ 6 ␤ 4 (54), which showed 19 and 16% amino acid sequence identity. These particular FN III domains were chosen as templates to minimize the insertions and deletions in the loop regions. The coordinates for the template were taken from the Protein Data Bank entry 1qg3. Domain 5 of sgp130 was manually aligned with the template of the cytoplasmic tail of integrin ␣ 6 ␤ crystal structure (54), conserving the hydrophobic and sequence patterns of the FN III ␤-sheets.   2. Comparative peptide mapping of a tryptic digest of sgp130. A, sgp130 was digested with trypsin as described under "Experimental Procedures," and generated peptides were fractionated by RP-HPLC using a Vydac octadecyl 5-m, 300 Å silica column (inner diameter, 250 ϫ 4.6 mm) with a gradient from 0 -70% B, where solvent A was 0.1% (v/v) trifluoroacetic acid and solvent B was 0.09% trifluoroacetic acid (v/v)/ acetonitrile. Flow rate was 0.5 ml/min, and temperature was 45°C. Peptides were monitored with four different wavelengths (215, 254, 280, and 290 nm), and spectral data were obtained on-the-fly by diode array detection. Cysteine-containing peptide fractions identified by NH 2 -terminal sequence analysis and MS analysis are annotated by the letter T (see also Table I). All peptide fractions were analyzed, and the obtained sequence information is in perfect agreement with the published amino acid sequence (11) (data not shown). B, chromatography of a portion (25%) of the tryptic digest of sgp130 from A following reduction with DTT (see "Experimental Procedures"). Chromatographic conditions are identical to A. The arrow indicates the position of T4 when chromatographed under nonreducing conditions (see A). Left inset, diode array spectra of T1 and T2. S-Pyridylethylated peptides (T1 and T2) were identified by their characteristic spectral properties (maxima at A 254 nm ) indicated by an arrow. Right inset, SDS-PAGE analysis of tryptic peptide fraction T5 and T6. Fractions T5-T6 from Fig. 2 (A) were pooled (10% of fraction T5 (1 ml) and T6 (0.9 ml)) and concentrated to 10 l by centrifugal lyophilization, diluted 3:1 (v/v) with nonreducing SDS sample buffer, and loaded onto a 4 -20% polyacrylamide precast gel, as described in Fig. 1. Protein was visualized by silver staining (76). Lane 1, protein molecular weight standards (Mark12™, Novex); lane 2, undigested gp130; lane 3, fraction T5-T6. The sample load was ϳ100 ng.
The MODELLER program (55) was used to generate a model of FN III domain 5. The quality of the model was assessed as described (56) in particular using the ProsaII program (57)

RESULTS
Purification and Characterization of sgp130 -Purified sgp130 (yield, ϳ1.0 mg/liter of CHO cell conditioned medium) was homogeneous as judged by SDS-PAGE (Fig. 1A, lane 2) and 20 cycles of Edman degradation, which yielded the unique NH 2 -terminal sequence ELLDPXGYISPESPVVQLHS (data not shown). At cycle 6, no phenylthiohydantoin (PTH) amino acid was observed, which is consistent with the predicted cysteine at this position that would not be revealed by Edman degradation. This preparation of sgp130 was monomeric (apparent molecular mass, ϳ120 kDa) as judged by analytical SEC (data not shown), a value that is in agreement with the mass of ϳ100 -120 kDa determined by SDS-PAGE. This value is significantly higher than 67,914.3 Da, which is calculated from the amino acid composition (11). Upon treatment with either neuraminidase or a combination of neuraminidase and an endoglycosidase mixture, the mass of sgp130 was reduced to ϳ95 kDa and ϳ70 -80 kDa, respectively (Fig. 1A, lanes 3 and 4). These data suggest that the increased mass of sgp130 observed by both SDS-PAGE and analytical SEC is due to N-glycosylation. Pronounced charge heterogeneity of the mature sgp130 was observed by IEF (Fig. 1B, lane 2). Treatment of sgp130 with neuraminidase (Fig. 1B, lane 3) reduced the levels of the acidic forms of the molecule, whereas treatment with a mixture of neuraminidase and endoglycosidase (Fig. 1B, lane 4) further reduced the complexity. Taken together, these data indicate that the heterogeneity of the sgp130 preparation is primarily due to differential N-linked glycosylation The immunological integrity of sgp130 was evaluated by Western blot analysis using a panel of anti-sgp130 monoclonal antibodies (AM64, GPX7, GPX22, and GPZ35; Ref. 48). Although monoclonal antibody AM64 recognized sgp130 under both nonreducing and reducing conditions, GPX7, GPX22, and GPZ35 (35) only recognized conformational epitopes of sgp130 under nonreducing conditions (data not shown). The functional integrity of the sgp130 used in this study was assessed by its ability to bind the IL-6⅐sIL-6R binary complex as well as its ability to form a stable IL-6⅐sIL-6R⅐sgp130 ternary complex (13). The mass of the ternary IL-6 receptor complex observed by analytical SEC (ϳ380 kDa) (data not shown) was in good agreement with a value calculated for a hexameric complex of two molecules each of IL-6, IL-6R, and gp130 (13).
Identification of Unpaired Cysteine(s) in sgp130 -Soluble gp130 was alkylated with 4-vinylpyridine at pH 8.5, digested with trypsin, and then chromatographed by RP-HPLC using a trifluoroacetic acid/acetonitrile solvent system (Fig. 2). Using ESI-MS/MS, the parent-ion scan of the total tryptic digest of the S-pyridylethylated sgp130 identified three Pec containing peptides with molecular masses of 914.6 Da (T1), 2314.6 Da (T2), and 2548.4 Da (T3) (data not shown). Two of these Peccontaining peptides (T1 and T2) were also confirmed by their UV absorbance at A 254 nm (see tryptic peptide fractions T1 and T2 in Fig. 2A, inset).
Edman degradation identified tryptic peptide T1 as residues Cys 279 -Lys 285 , T2 as residues Cys 469 -Lys 489 , and T3 as residues Ser 387 -Lys 409 . Cysteines were observed as PTH-Pec derivatives at positions 279, 469, and 397 in these peptides, respectively (Table I). The yield of PTH-Pec in peptide T3 (Cys 397 ) was significantly lower than that observed for the PTH-Pec derivatives in T1 and T2. Further confirmation of the identity of these peptides was obtained by rechromatography and subsequent ESI-MS/MS analysis (Table I). Automated CID MS/MS analysis of the doubly charged ion of peptide T2 (m/z 1158.9) localized the difference in mass observed from the calculated mass to Cys 469 (Fig. 3).
To evaluate the extent of S-pyridylethylation of peptide fraction T3, this fraction (from Fig. 2A) was rechromatographed using a capillary RP-HPLC system coupled on-line to ESI-MS/MS (Fig. 4A). Although Edman degradation of T3 revealed only one sequence (residues Ser 387 -Lys 409 ), examination of the average mass spectrum of the major total ion current peak (Fig. 4B) reveals two peptide ions with molecular masses of 2548.4 and 2562.7 Da that differed by 14 Da. The charge state distribution patterns and relative ion intensities of these ions are very similar. The parent-ion scan of the 2548.4-Da peptide ion (Fig. 4C) revealed that this peptide was S-pyridylethylated (Figs. 4C and 4D). Because the mass difference between Spyridylethyl and S-cysteinyl moieties is 14 Da, the possibility that Cys 397 in T3 may also be cysteinylated was further examined.
To this end, peptide fraction T3 was reduced with DTT and immediately reanalyzed by capillary RP-HPLC/ESI-MS. It can be seen from the total ion current profile (Fig. 5A) that the minor peak (fraction I) corresponds to the 2548.4-Da Pec-modified form of tryptic peptide Ser 387 -Lys 409 (compare Figs. 5B and 4D). The major peak (Fig. 5A, fraction II), which was resolved from the Pec peptide-containing fraction, has a mass of 2443.0 Da (Fig. 5C) (Fig. 6). Taken together, these data indicate that Cys 397 is cysteinylated (119 Da) and that this cysteine modification is readily removed upon reduction with DTT. A comparison of the two forms of the tryptic peptide fraction T3 (2548.4-and 2562.7-Da peptide ions) in sgp130 indicated that the major cysteine modification was S-cysteinylation (97%) with ϳ3% as the free thiol (as revealed by S-pyridylethylation following treatment with 4-vinylpyridine).
Determination of Disulfide Bonds of sgp130 by Proteolytic Digestion and Peptide Mapping-A portion (25%) of the tryptic digest of S-pyridylethylated sgp130 was reduced with 10 mM DTT at pH 8.0 and rechromatographed by RP-HPLC using chromatographic conditions identical to those employed for the separation of the nonreduced S-pyridylethylated tryptic digest of sgp130 (Fig. 2). A comparison of panels A and B in Fig. 2 revealed that, upon reduction, only peptide fraction T4, had an altered retention time, suggesting that it is a disulfide-containing peptide. Edman degradation of fraction T4 (Table I) identified two peptides, T4a APXITDWQQEDGTVHR (residues 442-457) and T4b YILEWXVLSDK, (residues 431-441).
Further analysis of peptide fraction T4 by MALDI-TOF MS yielded an average molecular mass of 3220.7 Da (Fig. 7), which is in good agreement with the mass observed by ESI ion trap MS (3222.4 Da). The ion (m/z 1611.9) observed by MALDI-TOF MS is the doubly charged form of the parent peptide ion (m/z 3221.7). In addition, two further peptide ions of (m/z 1856.0) and (m/z 1368.6) were observed (Fig. 7). Presumably, these peptide ions are the pseudomolecular ions corresponding to reduced forms of peptide T4a and T4b (Table I), respectively, that have arisen by prompt fragmentation (58) of the disulfide linkage.
No peptides from the NH 2 -terminal region of sgp130, including four potential disulfide-containing peptides, were obtained using this approach. Close inspection of the peptide map ( Fig. 2A) revealed a shallow peak (peptide fraction T5-T6) eluting late in the gradient. Edman degradation of these pooled fractions yielded two peptide sequences in low yield, one corresponding to the NH 2 terminus of sgp130 and the other commencing at residue 63 (data not shown). Nonreducing SDS-PAGE analysis of peptide fraction T5-T6 revealed a single protein band of ϳ50 kDa ( Fig. 2A, inset), suggesting that this fraction contained at least domains 1 and 2 of sgp130, held together by disulfide bonds, consistent with a partially digested NH 2 -terminal "core" polypeptide.
To establish the cysteine connectivities of the Ig-like domain of the molecule, S-pyridylethylated sgp130 was subjected to extensive digestion (48 h) using a high concentration of trypsin (enzyme to substrate ratio 1:5). Peptide mapping of this tryptic digest under reducing and nonreducing conditions (data not shown) revealed the identity of two disulfide-containing peptide fractions. One of these (T7) was identified as Cys 150 -Cys 160 (Table I), whereas the other fraction (T8) contained cysteine residues Cys 6 , Cys 26 , Cys 32 , and Cys 81 linked by disulfide bonds. To determine the connectivities of Cys 6 , Cys 26 , Cys 32 , and Cys 81 , tryptic peptide fraction T8 was subdigested with chymotrypsin. Using a combination of Edman degradation and MS analysis of the resultant peptides, disulfide bond linkages were established between Cys 6 -Cys 32 and Cys 26 -Cys 81 (Table I).
An additional cysteine connectivity in sgp130 was identified following S-pyridylethylation, deglycosylation (see "Experimental Procedures"), and then digestion with pepsin. Peptide mapping of the resultant peptic digest of S-pyridylethylated sgp130 (400 g) under reducing and nonreducing conditions (data not shown) revealed the presence of a disulfide-containing peptide P1 (Table I). Edman degradation and MS analysis established the identity of peptide P1 as a linear peptide (residues Ile 98 -Leu 132 ) containing the disulfide linkage Cys 112 -Cys 122 (Table I) (Table II). Further confirmation of N-glycosylated asparagine residues 61, 205, 357, 361, 531, and 542 was provided by Asn/Asp conversion in both tryptic and peptic sgp130 peptides following endoglycosidase F treatment (59) (see Asn/Asp conversion at position 109 in peptide P1 (Table I)). Asparagine residues 224 (NPS) and 368 (NLT) were identified as PTH-asparagine during Edman degradation indicating that these residues are not N-glycosylated. DISCUSSION To date there is no complete three-dimensional structure for the extracellular domain of gp130 or any member of the gp130 family of receptors. The results presented here for the cysteine connectivity pattern and carbohydrate attachment sites of sgp130 provide a foundation for the determination of the tertiary structure of this molecule and, ultimately, of the IL-6⅐IL-6R⅐gp130 complex. Mass spectrometric and NH 2 -terminal sequence analysis of proteolytic digests of sgp130 identified five disulfide linkages (Cys 6 -Cys 32 , Cys 26  The two disulfide linkages of the NH 2 -terminal Ig-like domain of gp130 (Cys 6 -Cys 32 and Cys 26 -Cys 81 ) (Fig. 9A) are found between the A strand and B-C loop and between the B and F strands, respectively, with the second disulfide linkage characteristic of a typical I-type Ig-like fold (25). The amino acid sequences of the Ig-like domains of gp130 and GCSFR are homologous, suggestive of an identical fold, yet their disulfide patterns differ (Fig. 9A). The percentage amino acid sequence identity is 19% over the whole Ig-like domain, with 30% identity over the region encompassing the first three cysteines (Fig. 9A).
The two consecutive disulfide linkages in the NH 2 -terminal FN III domain of the CBR of gp130 (Cys 112 -Cys 122 and Cys 150 -Cys 160 ) correspond to those experimentally determined by a peptide mapping approach for the CBR of other class I cytokine receptor families, namely GCSFR (61, 62), growth hormone receptor (63), and leptin receptor (64). These cysteine connectivities have also been confirmed by the x-ray crystal structure analysis of the CBR of gp130 (43), the growth hormone-growth hormone receptor complex (65), the GCSF⅐GCSFR complex (66), the growth hormone receptor-prolactin receptor complex (67), the erythropoietin receptor-peptide agonist complex (68), the IL-4⅐IL-4R complex (69), and IL-12R␣ (70). An inspection of the gp130 CBR three-dimensional structure reveals that the N-linked carbohydrate moieties found at Asn 109 (A strand, domain 2), Asn 135 (C strand, domain 2), and Asn 205 (A strand domain 3) are consistent with these asparagines being solvent exposed, whereas the potential N-glycosylation site at Asn 224 would not be expected to be glycosylated because this asparagine residue is partially buried in the structure (43).
Cys 279 , Cys 397 , and Cys 469 located in domains 3, 4, and 5, respectively, do not form disulfide bonds and occur as either a free thiol or in the case of Cys 397 , a cysteinylated form (Fig. 8). Previously, we have reported the presence of both free and cysteinylated cysteine residues in the human IL-6R (52). The occurrence of free cysteines has also been reported for GCSFR (62) and leptin receptor (64). Cys 279 , which is present as a free thiol in sgp130, is not conserved among the species variants of gp130 whose sequences have been determined to date (i.e. mouse (23), rat (24), chicken (71), and frog 2 ). Analysis of the gp130 CBR crystal structure (43) indicates that Cys 279 is situated on the B strand of domain 3 and is partially buried. Cys 397 , which is located at the end of the G strand of domain 4 and likely forms part of the hinge region with domain 5, was found to be S-cysteinylated. This finding is consistent with Cys 397 being solvent exposed.
Cys 469 , located in domain 5, is the only free cysteine con-served in all species of gp130 studied to date. As such, Cys 469 may be involved in the homodimerization related intracellular signaling of this molecule that is reported to involve disulfidebond formation (12). The free thiol Cys 469 is located at the NH 2 terminus of the F strand flanked by conserved lysine and tyrosine residues and would be predicted to be solvent exposed and hence modified. Our homology modeling data predict that although the free thiol Cys 469 is orientated away from the hydrophobic core of domain 5, it is shielded from the solvent by a short eight-residue "protective loop" tethered by the disulfide bond Cys 436 -Cys 444 in strands C and CЈ of domain 5 (Fig. 10). Thus, a conformational change in gp130 would be required to reposition this protective loop and expose Cys 469 to possible disulfide bond formation with another gp130 molecule. Mü ller-Newen and colleagues (73) have recently demonstrated, using a panel of monoclonal antibodies, that the enforcement of gp130 dimerization alone is not sufficient for receptor activation and that an additional conformation requirement is needed. A similar phenomenon has also been reported for the erythropoietin receptor (74) and is also applicable to other types of receptors from both eukaryotic and prokaryotic origins (75), where dimerization alone is not sufficient for intracellular signaling but a further conformational change within the receptor dimer induced by ligand binding may be essential to propagate the signal across the cell membrane. It is tempting to speculate that the suggested conformational switch of the gp130 dimer required for activation of intracellular signaling (73) may involve the movement of the protective loop, thereby exposing the buried thiol (Cys 469 ) in such a way that it can partake in disulfide bond formation with another gp130 molecule. The role of FN III domain 5 and the conserved free Cys 469 residue in gp130 homodimerization and intracellular signaling process must await further studies.  (78) representation of domain 5 is shown in two orientations, one rotated by 90°around the vertical axis relative to the other. ␤-Strands are labeled according to the alignment in Fig. 9B, and the free thiol Cys 469 and disulfide linkage Cys 436 -Cys 444 are shown in ball and stick representation. The C-CЈ loop covers the free thiol Cys 469 , rationalizing why this otherwise solvent exposed residue is not modified. The model follows the human integrin ␣ 6 ␤ 4 domain 2 (77) most closely, particularly in the C-CЈ loop region.  (25). Cysteine connections determined experimentally for sgp130 and soluble GCSFR (62) are indicated as a thick line. Amino acid sequence data were taken from the NCBI data base (human gp130, accession number 4504674; human GCSFR, accession number 729564) by their respective accession numbers. B, alignment of the amino acid sequence of FN III domain 5 (Gln 400 -Pro 497 ) of human gp130 and FN III domain of human integrin ␣ 6 ␤ 4 . Amino acid sequences were manually aligned. The disulfide bond between Cys 436 and Cys 444 is shown as a thick line, and the free cysteine (Cys 469 ) is shown with an asterisk. Letters above sequence indicate positions of strands, and their length is indicated by the underlined sequence of integrin ␣ 6 ␤ 4 as determined from the x-ray crystal structure obtained from the Protein Data Bank Data Base code 1qg3 (77). Amino acid sequence data for gp130 was taken from the NCBI data base accession number 4504674. Gaps to improve the alignment of sequences are indicated by hyphens. Residue numbers for the mature form of gp130 are shown next to the sequence.