Crystal structure of a cytokine-binding region of gp130
1998; Springer Nature; Volume: 17; Issue: 6 Linguagem: Inglês
10.1093/emboj/17.6.1665
ISSN1460-2075
Autores Tópico(s)HER2/EGFR in Cancer Research
ResumoArticle16 March 1998free access Crystal structure of a cytokine-binding region of gp130 Jerónimo Bravo Jerónimo Bravo Laboratory of Molecular Biophysics, The Rex Richards Building, South Parks Road, Oxford, OX1 3QU GB Search for more papers by this author David Staunton David Staunton Oxford Centre for Molecular Sciences, New Chemistry Building, South Parks Road, Oxford, OX1 3QT GB Search for more papers by this author John K. Heath John K. Heath Cancer Research Campaign Growth Factor Group, School of Biochemistry, The University of Birmingham, Edgbaston, Birmingham, B15 2TT GB Search for more papers by this author E.Yvonne Jones Corresponding Author E.Yvonne Jones Laboratory of Molecular Biophysics, The Rex Richards Building, South Parks Road, Oxford, OX1 3QU GB Oxford Centre for Molecular Sciences, New Chemistry Building, South Parks Road, Oxford, OX1 3QT GB Search for more papers by this author Jerónimo Bravo Jerónimo Bravo Laboratory of Molecular Biophysics, The Rex Richards Building, South Parks Road, Oxford, OX1 3QU GB Search for more papers by this author David Staunton David Staunton Oxford Centre for Molecular Sciences, New Chemistry Building, South Parks Road, Oxford, OX1 3QT GB Search for more papers by this author John K. Heath John K. Heath Cancer Research Campaign Growth Factor Group, School of Biochemistry, The University of Birmingham, Edgbaston, Birmingham, B15 2TT GB Search for more papers by this author E.Yvonne Jones Corresponding Author E.Yvonne Jones Laboratory of Molecular Biophysics, The Rex Richards Building, South Parks Road, Oxford, OX1 3QU GB Oxford Centre for Molecular Sciences, New Chemistry Building, South Parks Road, Oxford, OX1 3QT GB Search for more papers by this author Author Information Jerónimo Bravo1, David Staunton2, John K. Heath3 and E.Yvonne Jones 1,2 1Laboratory of Molecular Biophysics, The Rex Richards Building, South Parks Road, Oxford, OX1 3QU GB 2Oxford Centre for Molecular Sciences, New Chemistry Building, South Parks Road, Oxford, OX1 3QT GB 3Cancer Research Campaign Growth Factor Group, School of Biochemistry, The University of Birmingham, Edgbaston, Birmingham, B15 2TT GB *Corresponding author. E-mail: [email protected] The EMBO Journal (1998)17:1665-1674https://doi.org/10.1093/emboj/17.6.1665 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info The structure of the cytokine-binding homology region of the cell surface receptor gp130 has been determined by X-ray crystallography at 2.0 Å resolution. The β sandwich structure of the two domains conforms to the topology of the cytokine receptor superfamily. This first structure of an uncomplexed receptor exhibits a similar L-shaped quaternary structure to that of ligand-bound family members and suggests a limited flexibility in relative domain orientation of some 3°. The putative ligand-binding loops are relatively rigid, with a phenylalanine side chain similarly positioned to exposed aromatic residues implicated in ligand binding for other such receptors. The positioning and structure of the N-terminal portion of the polypeptide chain have implications for the structure and function of cytokine receptors, such as gp130, which contain an additional N-terminal immunoglobulin-like domain. Introduction Cytokines, generally in the form of secreted molecules, mediate intercellular signalling by high affinity interaction (KD ∼10−10 M) with the extracellular regions of specific cell surface receptors. This promotes receptor oligomerization which, in turn, triggers intracellular signalling cascades within the target cell. The ability of a given cytokine to elicit biological responses in a target cell is therefore dictated by the specificity of interaction between ligand and receptor. Gp130 is a transmembrane receptor which is required for signal transduction by a set of cytokines, the gp130 family, which have many significant biological functions of potential therapeutic interest (reviewed in Kishimoto et al., 1995). Gp130-mediated signalling has been implicated in the regulation of a wide variety of adult tissue systems, including haemopoesis, nervous system, bone, heart, adipose tissue, testes, liver and muscle (reviewed in Kishimoto et al., 1995). Targeted inactivation of the gp130 gene results in a complex pre-natal lethal phenotype including defects in cardiac and haematological function (Yoshida et al., 1996). In addition, chronic activation of gp130 signalling in a transgenic mouse model results in cardiac hypertrophy (Hirota et al., 1995). The gp130 family of ligands currently comprises interleukin-6 (IL-6), IL-11, herpes virus IL-6 (HSVIL-6), leukaemia inhibitory factor (LIF), oncostatin (OSM), cardiotrophin (CT-1) and ciliary neurotrophic factor (CNTF). The three-dimensional structures of three members of the gp130 family, LIF (Robinson et al., 1994), CNTF (McDonald et al., 1995) and IL-6 (Somers et al., 1997; Xu et al., 1997) have been defined by crystallographic or solution NMR techniques. This reveals that these cytokines share a common topology, being composed of four regions of α helix (helices A–D) linked by polypeptide loops in the 'up-up-down-down' conformation typical of the 'long chain' family of cytokines (Boulton et al., 1994). The signalling functions of gp130 are initiated by the ligand-mediated formation of oligomeric complexes with other specific partner receptors. Gp130 initially was cloned as an essential transmembrane component for signalling mediated by IL-6 (Hibi et al., 1990). This occurs via dimerization of gp130 (Murakami et al., 1993) following the formation of a hexameric complex containing two molecules of gp130, two molecules of IL-6 and two molecules of a soluble specific IL-6 receptor (Ward et al., 1994; Paonessa et al., 1995). A similar mechanism pertains to the case of IL-11, where homodimerization of gp130, and subsequent execution of signalling functions, is brought about by association with a complex of IL-11 and specific IL-11 receptors (Hilton et al., 1994; Karow et al., 1996). Gp130 was also cloned as a receptor required for signalling mediated by cytokines which associate with a second transmembrane receptor of the cytokine type–LIF-R (Gearing et al., 1991). These include OSM (Gearing et al., 1992; Liu et al., 1992), LIF and CNTF (Ip et al., 1992), and CT-1 (Pennica et al., 1995). In this case, signal transduction is initiated by ligand-mediated heterodimerization of gp130 and LIF-R (facilitated, in the case of CNTF, by association with a third non-signalling receptor component CNTF-R). Recently it has been discovered that OSM can also mediate signalling by heterodimerization of gp130 with a novel transmembrane signalling receptor of the cytokine type, OSM-R (Mosley et al., 1996). The intracellular signalling pathways activated by ligand-mediated homo- or heterodimerization of gp130 include activation of the receptor-associated JAK/Tyk tyrosine kinases (Boulton et al., 1994; Stahl et al., 1994), the STAT family of transcription factors (Stahl et al., 1995) and src-family tyrosine kinase pathways (Ernst et al., 1994). The sequence of the extracellular ligand-binding region of gp130 reveals that it is a member of the 'cytokine' superfamily of receptors characterized by a canonical cytokine-binding homology region (CHR) containing the 'WSXWS' motif, a proline-rich 'hinge' region and a characteristic spacing of cysteine residues (reviewed in Cosman, 1993). In addition, the extracellular region of gp130 contains an N-terminal module predicted to adopt a seven-stranded immunoglobulin-like conformation and, C-terminal to the CHR, three fibronectin type III (FN III) domains. Deletion studies have revealed that the gp130-CHR is sufficient for interaction with ligand (Horsten et al., 1995). Mutation studies of both IL-6 (Paonessa et al., 1995) and LIF (Hudson et al., 1996) have shown that this interaction involves topologically analogous receptor recognition epitopes (site II) in both ligands. A second, physically discrete, gp130 ligand recognition epitope (site III) has also been described for the interaction with IL-6 (Paonessa et al., 1995). This interaction requires regions of gp130 outside the CHR (Simpson et al., 1997; D.Staunton, K.R.Hudson and J.K.Heath, unpublished observations). The function of the CHR of gp130 is therefore the association with partner ligands (alone or complexed with receptor) via their site II recognition epitopes. Crystal structures are available for four cytokine receptors containing a CHR; the growth hormone receptor (GHR; De Vos et al., 1992), the prolactin receptor (PRLR; Somers et al., 1994), the erythropoietin receptor (EPOR; Livnah et al., 1996) and the interferon-γ receptor (IFNγ-R; Walter et al., 1995). These four prototypes undergo an exclusively homodimerization mode of action with a restricted range of ligands; little is known currently of the detailed structural features of receptors which undergo heterodimerization in the presence of ligand or interact with multiple ligand and receptor partners. We report here the high resolution crystal structure of the cytokine-binding homology region of gp130. This structure provides the first detailed three-dimensional information for a receptor component crucial to the signalling complexes of a large family of growth factors (IL-6, IL-11, LIF, CNTF, CT-1 and OSM) allowing assessment of the molecular basis of specific recognition and ligand engagement. Results and discussion Expression and structure determination A soluble form of the gp130 cytokine-binding homology region (gp130-CHR) was expressed in Escherichia coli as a maltose-binding protein (MBP) fusion (D.Staunton, K.R.Hudson and J.K.Heath, in preparation). The final purified protein comprised four residues from the fusion linker, residues 100–303 of human gp130 (the CHR) and a further 14 residues corresponding to a three-alanine linker and c-myc tag. Surface plasmon resonance studies confirmed that this recombinant form of gp130-CHR binds OSM with an affinity (Kd ∼6.5×10−8 M) equivalent to the complete gp130 extracellular domain expressed in eukaryotic cells (D.Staunton, K.R.Hudson and J.K.Heath, in preparation). Gp130-CHR therefore retains structural features required for ligand engagement via site II. Crystallization trials yielded highly ordered crystals of space group C2221 (unit cell dimensions a = 84.5 Å, b = 132.3 Å, c = 121.9 Å) which contained two gp130-CHR molecules per crystallographic asymmetric unit. The non-crystallographic symmetry does not reveal any possible mode of receptor dimerization. The structure was determined by multiple anomalous dispersion (MAD) phasing techniques using X-ray diffraction data collected on BM14 at the European Synchrotron Radiation Facility (ESRF) from crystals of a selenomethionyl form of the protein. The structure has been refined to a crystallographic R-value of 21.5% for all data between 30 and 2.0 Å resolution. Crystallographic statistics are reported in Tables I and II. All 204 residues of the CHR are well ordered for one of the two copies in the crystallographic asymmetric unit, but residue 100 at the N-terminus and residues 212–213 of a loop region are disordered in the second copy; the final model also includes certain of the residues which derive from the expression construct (Figure 1A, see Materials and methods). Domain-wise superpositions of the two molecules in the crystallographic asymmetric unit show essentially identical structures (r.m.s. deviation for equivalent Cα atoms between 90 residues of the CHR N-terminal domain is 0.44 Å and between 99 residues of the CHR C-terminal domain is 0.32 Å). The following text focuses exclusively on the structure of the human gp130-CHR and refers to these residues by the intact gp130 numbering, 100–303. Figure 1.The structure of gp130-CHR. (A) Ribbon representation of the structure of gp130-CHR. Helical segments are shown in red and β strands in green. The full crystal structure is illustrated which includes, in addition to the structure of residues 100–303 of gp130, an extra three N-terminal residues and eight C-terminal residues which derive from the expression construct. The C-terminal α helix results from these latter residues. (B) Topology diagram of the two domains of gp130-CHR. Helices are represented by cylinders and β strands by arrows. The positions of the five cysteine residues are marked by black bars. (A) and all components of Figures 2, 3 and 5 were drawn using programs MOLSCRIPT (Kraulis, 1991), with modifications by R.Esnouf (Esnouf, 1997), and RASTER3D (Merrit and Murphy, 1994). Download figure Download PowerPoint Table 1. gp130-CHR data collection statistics Wild-type Seλ1 Seλ2 Seλ3 Seλ4 Seλ5 Wavelength (Å) 1.030 9790 0.9791 0.9535 0.9793 0.9789 Unique 48 189 20 549 20 280 22 400 20 264 19 303 19.4 (5.0) 23.2 27.2 24.8 18.8 28.6 Rmerge (%) 5.8 (20.6) 5.1 4.3 4.2 4.2 5.0 Resolution (Å) 30–2.0 30–2.55 30–2.55 30–2.50 30–2.55 30–2.55 Completeness (%) 98.8 (96.6) 93.6 93.5 92 93.5 89.4 Rmerge = Σ|I − | / Σ . Values in parentheses correspond to the highest resolution shell (2.07–2.00 Å). Table 2. gp130-CHR structural refinement statistics Resolution range (Å) 30–2.0 Completeness (%) 95.0 No. of reflections (F>0) 44 627 Rcryst (%) 21.5 Rfree (%) 25.0 No. of non-hydrogen atoms Protein 3359 Water 287 Sulfate 30 R.m.s.d. from ideality Bond lengths (Å) 0.006 Bond angles (°) 1.40 Dihedrals (°) 25.36 Improper (°) 1.11 Average B-factor (Å2) Main chain 19.5 Side chain 20.3 Water 19.1 Rcryst = Σ||Fobs| − |Fcalc||/Σ|Fobs|. Rfree is as for Rcryst but calculated for a test set comprising reflections not used in the refinement (7.5%). Structure description and comparison with other CHR structures As anticipated from sequence analysis, the topology of gp130-CHR is similar to those of the three other class 1 receptors of the cytokine superfamily (Cosman, 1993) for which structures have been determined (hGHR, de Vos et al., 1992; hPRLR, Somers et al., 1994; EPOR, Livnah et al., 1996). The CHR comprises two specialized FN III domains. The basic structural scaffold for each domain consists of a β sandwich primarily formed from a three-strand (A, B, E) and four-strand (C, C′, F, G) β sheet. These domains are connected by a short 310 helix and are oriented such that the whole molecule has an approximate L shape (Figure 1). Gp130-CHR D1 The N-terminal domain (D1; residues 103–192) has the standard arrangement of A, B, C, C′, E, F and G β strands (Figures 1 and 2). One notable feature, unique to gp130-CHR D1, is the division of strand G into two approximately equal portions. This is the result of a β bulge and, as discussed below, shows similarity to the WSXWS motif of the second domain. In common with other examples of this fold, gp130-CHR D1 contains two disulfides which link the A and B strands (Cys112–Cys122) and the C′ and E strands (Cys150–Cys160). The lengths of the interstrand loops are similar to those in the EPOR D1 (with the exception of the shorter region between strand C′ and E, and a longer FG loop) and show little significant increase in flexibility relative to the core of the domain (as judged from crystallographic B-factors). Figure 2.Gp130-CHR domain 1. (A) Comparison of D1 of gp130 (green) and D1 of EPOR (brown). The C′ and G strands are broken into two portions in gp130, and there is no D strand. (B) The β bulge in strand G. The main-chain atoms of β strands F and G are shown in stick representation, Ser187 is shown in ball and stick representation, and hydrogen bonds are denoted by broken lines. (C) The distinctive C′,C,F,G sheet in gp130. The C′,C,F,G strands are shown for gp130 (green) and EPOR (brown) positioned on the basis of a whole domain superposition (performed using the program SHP; Stuart et al., 1979) to illustrate the novel nature of the top half of the gp130 β sheet. Download figure Download PowerPoint Given this general level of topological equivalence, it is somewhat surprising to find that structural superpositions with other CHR D1 structures (Figure 2A) show relatively poor agreement within the core framework (for example, 1.0 Å r.m.s. deviation for 53 structurally equivalent Cα atoms with EPOR D1). As illustrated in Figure 2B, this discrepancy arises from the distinctive angle of β strands C, F and G in the upper half of the GFCC′ sheet. A superposition on gp130-CHR D2 shows a more extensive match over these main secondary structure elements (1.2 Å r.m.s. deviation for 64 structurally equivalent Cα atoms). The key feature, common to both gp130-CHR D1 and D2, is the region of extended polypeptide chain (residues 102–107 of D1, residues 198–204 of D2) which packs tightly against the edge of the β sandwich between strands B and G before starting strand A at Asn109 in D1 and Asn205 in D2. The tight packing of this part of the polypeptide chain against the core of the β sandwich is mediated by the insertion of proline residues Pro103 and Pro107 in D1, and Pro200 and Pro203 in D2. The incorporation of this feature necessitates the shift in orientation of the upper part of the GFC sheet. In gp130-CHR D1, this is achieved through the distinctive β bulge in strand G at residues 187–189 stabilized by the hydrogen bonding of the Ser187 hydroxyl to the main-chain nitrogen of Val176 in strand F (Figure 2C). An identical function is performed by the two serine residues in the canonical WSXWS motif of CHR D2 structures, with the superposition of the gp130-CHR D1 and D2 domains indicating that Ser187 and Ser292 are structurally equivalent. None of the other class 1 CHR D1 structures contain the equivalent length of polypeptide tightly clamped between the B and G β strands or the β bulge in strand G. Gp130-CHR D2 The C-terminal domain (D2; residues 200–300) conforms to the standard A,B,E and GFCC′ β sheet arrangement. Of the two domains of the CHR, the second appears to be generally the more structurally conserved within the cytokine receptor family. Structural superpositions indicate closest similarity to the EPOR domain (1.05 Å r.m.s. deviation for 81 structurally equivalent Cα atoms). This arises primarily from the shorter length of the β strands in these two molecules compared with those in other members of the superfamily. The only secondary structure element to not correspond closely in position between the gp130-CHR and EPOR D2 structures is the C′ strand. In common with the other class-1 CHR D2 structures, the interstrand loops in gp130-CHR D2 show relatively limited mobility (as judged from crystallographic B-factors), with the exception of the AB loop. This loop is stabilized by a crystal contact in one copy of the gp130-CHR but in the other copy the AB loop is exposed to solvent and is disordered in the electron density map (residues 212 and 213). The domain contains one free cysteine residue (Cys279) which is buried within the core of the β sandwich. This precludes it from involvement in disulfide-linked homodimerization of gp130 during IL-6-related cytokine signalling (Murakami et al., 1993). The WSXWSX sequence (gp130 residues 288–293), a defining feature of this receptor superfamily, is situated in the N-terminal portion of strand G (Figure 3A and B) and has an essentially identical double β bulge structure to that of the homologous region in EPOR. The two successive β bulges are stabilized by hydrogen bonds from the hydroxyls of Ser289 and Ser292 to the main-chain nitrogens of Cys279 and Ile277 respectively. As in the other class-1 CHR D2 structures, the side chains of Trp288 and Trp291 participate in an extended π-cation system, stacking between the side chains of strand F residues Arg276, Arg278 and Met280. This feature is considerably more extended in gp130-CHR D2 than in EPOR D2, however, since it also involves side chains from residues Arg240 and Trp247 in strand C and at the start of strand C′ respectively. The absence of these latter interactions in EPOR may underlie the difference in the position of its C′ strand relative to that in gp130-CHR D2. The extended π-cation system in gp130 is most closely matched by that in PRLR. Figure 3.Gp130-CHR domain 2 and interdomain linker. (A) Schematic diagram of the WSXWS box. The side chains of residues contributing to this structural feature are colour coded according to the secondary structure element from which they originate. (B) Atomic coordinates and electron density map for the WSXWS box. The refined coordinates are displayed with the original 2.9 Å resolution electron density map calculated using MAD phases followed by density modification (program DM, see Materials and methods). The map is contoured at 1σ in program O. (C) Schematic diagram of the D1–D2 linker region. Strand L in the linker region hydrogen-bonds to strand A in DL and the WSXWS box region of the polypeptide chain prior to the start of strand G in D2. Download figure Download PowerPoint The interdomain region and relative domain orientation As observed for the other superfamily members, residues of the interdomain linker region (gp130 residues 193–197) form a 310 helix. Additionally, residues 198 and 199 in gp130-CHR form a short β strand. This novel structural element hydrogen-bonds to both strand A of D1 and the WSXWS region at the N-terminus of strand G in D2, providing an extra constraint on the relative orientation and positioning of the two domains (Figure 3C). The resultant juxtaposition of D1 and D2 produces a tight interdomain interface which, excluding the contribution of the linker, buries ∼350 Å2 of solvent-accessible surface. This is contributed mainly by residues in strand A and the EF loop of D1 (I113, E116 and Y168) and residues from the BC loop and strand G of D2 (I227, V230, I231 and Y287). Superposition of the two copies of gp130-CHR in the crystallographic asymmetric unit reveals a 3° difference in the relative orientation of their domains. This appears to originate from very slight changes in the main-chain torsion angles for linker residues V198–K199. The structural constraints imposed by the linker 310 helix and β strand plus the hydrophobic, close-packed nature of the interface, argues against any more substantial degree of interdomain orientational freedom than the observed 3° range. The overall shape of the molecule can be quantified in terms of a tilt angle (defined as the angle between the long axes, running approximately parallel to the β strands, in the two domains; Bork et al., 1996). With a tilt angle of 78°, the relative domain orientation in gp130-CHR corresponds most closely to the general 'L-shaped' (∼90° tilt angle) arrangement characteristic of the other class 1 members of the superfamily (hGHR, hPRLR and EPOR) rather than the more upright (∼50° tilt angle) arrangement of the more distantly related class 2 members IFNγ-R and TF or the more extreme 120° or so of tilt very recently observed in the natural killer inhibitory receptor (Fan et al., 1997). A detailed comparison (combining the effect of tilt angle and twist in the orientation of D1 relative to the D2 axis) reveals that the precise interdomain orientation in gp130-CHR differs by some 23° from that of the most closely related quaternary structure, that of EPOR. Sequence comparisons with other species The human gp130-CHR sequence employed in this study was aligned with the homologous regions from mouse, rat and Xenopus gp130 (Figure 4). This reveals that 74/204 (36%) of residues in this region are conserved amongst all versions of the gp130-CHR. It is notable that the majority (64) of these shared residues are in a relatively buried location in the human gp130-CHR structure. This indicates that these conserved residues most probably play a role in determining the structural framework of gp130 rather than ligand recognition. It follows that the putative ligand recognition epitopes of gp130 may exhibit variation between species. Figure 4.Sequence alignment and solvent accessibility of gp130 sequences. The CHR region of human (Swissprot accession No. P40189, Hibi et al., 1990), murine (Q00560, Saito et al., 1992), rat (P40190, Wang et al., 1992) and Xenopus (K.Chien, A.Grace and J.Chen, personal communication) gp130 sequences were aligned using the progressive pairwise algorithm of Feng and Doolittle (1987) implemented in the Pileup programme of the GCG package followed by minor manual editing. Solvent exposure scores were calculated using DSSP implemented in the program Turbo-6 (Roussel and Cambillau, 1989). Scores 0–50 were assigned a value of 1, 51–100:2, 101–150:3 and 151–200:4. Conserved residues are coloured in orange and residues which are identical to human gp130 are coloured in red. The figure was generated using the programme Alscript (Barton, 1993). Download figure Download PowerPoint Implications for ligand recognition The topological similarity of gp130 to hGHR and hPRLR, systems for which ligand binding has been structurally and functionally well characterized (Cunningham and Wells, 1989; reviewed in Sprang and Bazan, 1993; Wells et al., 1993), permits the identification of candidate structural features of gp130 that mediate ligand recognition via site II. The cognate site II gp130 recognition epitopes have been defined for two ligands, LIF (Hudson et al., 1996) and IL-6 (Savino et al., 1994). In both cases, these consist of a small number (4–6) of solvent-exposed residues located in the adjacent helices A and C of the ligand. This suggests that the site II ligand recognition site of gp130 will also be formed from relatively few solvent-exposed residues forming a complementary binding site. The hGH-hGHR complex (De Vos et al., 1992) and the related hGH-hPRLR complex (Somers et al., 1994) reveal that recognition of the ligand via site II involves solvent-exposed residues located in three loops linking the main β strand elements. The first of these is a prominent aromatic residue located in the loop between strands E and F of D1 (Trp104 in both hGHR and hPRLR). In the human gp130-CHR structure, the analogous EF loop contains a similar prominent, solvent-exposed residue Phe169 (Figure 5). Sequence alignment of gp130 sequences (Figure 4) suggests that the analogous residue is present in rat gp130 but is replaced by the conservative substitution of a tyrosine residue in the mouse and Xenopus proteins. Sequence alignment of the equivalent region in other receptors for members of the long chain cytokine family reveals that a non-polar residue (Phe, Tyr or Trp) in the predicted EF loop of D1 is a common feature (data not shown). Figure 5.The putative ligand-binding region in gp130-CHR compared with hGHR. Two orthogonal views are shown for gp130-CHR (green) and hGHR (yellow). Loops implicated in ligand binding for the hGH–hGHR complex, and their equivalents in the gp130-CHR structure, are denoted in red. The side chains of the structurally equivalent residues Phe169 and Trp104 are shown in ball and stick representation. Download figure Download PowerPoint The other two potential binding sites are located in D2 in the form of the loops linking strands BC (gp130 residues 226–230) and FG (gp130 residues 281–285). A non-conservative substitution mutant V230D, which is located in the BC loop, results in loss of affinity for IL-6–IL-6R (Horsten et al., 1997). V230 is, however, relatively buried in the three-dimensional structure (DSSP score of relative exposure 23%), and the effect of this mutation on ligand recognition may, therefore, be indirect. The FG loop has also been implicated recently in the recognition of granulocyte colony-stimulating factor (GCSF) by the D2 domain of the GCSFR CHR (Yamasaki et al., 1997). In addition, non-conservative substitution mutants of two exposed residues (G286W and K285E) in this region of gp130 result in loss of affinity for IL-6–IL-6R (Horsten et al., 1997), suggesting that the FG loop does indeed play a significant role in ligand recognition. The equivalent regions of gp130 proteins from other species exhibit conservation, but not identity, of residues in these regions. A key feature of gp130 is its ability to interact with a range of ligands in the context of a number of other partner receptors. In cases such as LIF and OSM (Hudson et al., 1996), the ligand is able to interact with gp130 with high affinity on its own. In other cases such as IL-6 and IL-11, high affinity interaction between gp130 and the ligand requires that ligand is associated with a partner receptor (IL-6R and IL-11R respectively). This suggests that there may exist additional sites on gp130 which are involved in interaction with partner receptors. Mutagenesis of the IL-6R (Yawata et al., 1993) and the IL-11R (M.A.Hall, P.Bilinski, A.Gossler and J.K.Heath, unpublished observations) have revealed that specific non-conservative substitutions of residues in the membrane-proximal region of the predicted D2 in these receptors can block the ability of the ligand–receptor complex to interact with gp130. Inspection of the dimeric hGHR complex (de Vos et al., 1992) shows that D2 of each hGHR partner receptor is in apposition, forming a receptor dimer interface. It is likely, therefore, that non-conservative mutations in the dimer interface could disrupt the formation of a high affinity complex. It may therefore be anticipated that the analogous dimer interface region of gp130 (formed from the AB loop and strand E of D2) may also include sites of receptor–receptor recognition. It is o
Referência(s)