doi:10.1006/jmbi.2000.4387 available online at http://www.idealibrary.com on
`
`J. Mol. Biol. (2001) 306, 263±274
`
`Identification of the Domain in the Human
`Interleukin-11 Receptor that Mediates Ligand Binding
`
`Karin Schleinkofer1, Andrew Dingley2, Ingrid Tacken1
`Matthias Federwisch1, Gerhard MuÈ ller-Newen1, Peter C. Heinrich1
`Patricia Vusio3, Yannick Jacques3 and Joachim GroÈ tzinger1*
`
`1Institut fuÈ r Biochemie
`RWTH-Aachen
`UniversitaÈtsklinikum
`Pauwelsstr.30, 52057 Aachen
`Germany
`
`2Institut fuÈ r Physikalische
`Biologie, Heinrich Heine
`UniversitaÈt DuÈ sseldorf
`D-40225, DuÈ sseldorf and
`Forschungszentrum JuÈ lich
`D-52425 JuÈ lich, Germany
`
`3Cytokines et ReÂcepteurs
`Unite 463, 9 Quai Moncousu
`44035 Nantes Cedex 1
`France
`
`The interleukin-11 receptor (IL-11R) belongs to the hematopoietic receptor
`superfamily. The functional receptor complex comprises IL-11, IL-11R
`and the signal-transducing subunit gp130. The extracellular part of the
`IL-11R consists of three domains: an N-terminal
`immunoglobulin-like
`domain, D1, and two ®bronectin-type III-like (FNIII) domains and D2
`and D3. The two FNIII domains comprise the cytokine receptor-hom-
`ology region de®ned by a set of four conserved cysteine residues in the
`N-terminal domain (D2) and a WSXWS sequence motif in the C-terminal
`domain (D3). We investigated the structural and functional role of the
`third extracellular receptor domain of IL-11R. A molecular model of the
`human IL-11/IL-11R complex allowed the identi®cation of amino acid
`residues in IL-11R to be involved in ligand binding. Most of them were
`located in the third extracellular domain, which therefore should be able
`to bind with high af®nity to IL-11. To prove this prediction, domain D3
`of the IL-11R was expressed in Escherichia coli, refolded and puri®ed. For
`structural characterization, circular dichroism, ¯uorescence and NMR
`spectroscopy were used. By plasmon resonance experiments, we show
`that the ligand-binding capacity of this domain is as high as that one for
`the whole receptor. These results provide a basis for further structural
`investigations that could be used for the rational design of potential
`agonists and antagonists essential in human therapy.
`# 2001 Academic Press
`
`*Corresponding author
`
`Keywords: Interleukin-11 receptor; cytokine receptors; ligand binding;
`cytokine receptor-homology region
`
`Introduction
`
`Interleukin-11 is a pleiotropic cytokine playing
`an important role in hematopoiesis (Neben et al.,
`1994; Yonemura et al., 1992) and anti-in¯ammatory
`activities (Redlich et al., 1996; Trepicchio et al.,
`1996). Due to its thrombopoietic potential, it has
`been demonstrated in a phase II clinical trial that
`
`Abbreviations used: HSQC, hetero single quantum
`correlation; INEPT, insensitive nuclear enhanced
`polarization transfer; rmsd, root-mean-square
`differences; ru, resonance units; IL-11R, interleukin-11
`receptor; CNTF, ciliary neurotrophic factor;
`LIF, leukemia inhibitory factor; OSM, oncostatin M;
`CT-1, cardiotrophin-1; CRH, cytokine receptor
`homology region; hGH, human growth hormone;
`hGHR, hGH receptor.
`E-mail address of the corresponding author:
`achim@bionm1.biochem.rwth-aachen.de
`
`0022-2836/01/020263±12 $35.00/0
`
`recombinant IL-11 reduces chemotherapy-induced
`thrombocytopenia in breast cancer patients (Tepler
`et al., 1996). Furthermore, IL-11 was shown to play
`a critical role in female reproduction: female mice
`lacking the receptor for IL-11 are infertile due to a
`failure of trophoblast implantation (Robb et al.,
`1998).
`IL-11 belongs to the so-called IL-6-type family of
`cytokines, all sharing a four-helix-bundle fold
`(Bazan, 1990a; GroÈ tzinger et al., 1999). All members
`of this family, namely interleukin-6 (IL-6), interleu-
`kin-11 (IL-11), ciliary neurotrophic factor (CNTF),
`leukemia inhibitory factor
`(LIF), oncostatin M
`(OSM) and cardiotrophin-1 (CT-1) signal via a
`gp130 homodimer (Hibi et al., 1990) or a heterodi-
`mer of gp130 and the LIF receptor (Gearing et al.,
`1991) and/or OSM receptor (Mosley et al., 1996).
`Whereas IL-11 and IL-6 use a gp130 homodimer
`Singapore Exhibit 2010
`(Murakami et al., 1993; Yin et al., 1993), CNTF
`(Davis et al., 1993), LIF (Gearing et al., 1992) and
`Lassen v. Singapore et al.
`PGR2019-00053
`
`# 2001 Academic Press
`
`

`

`doi:10.1006/jmbi.2000.4387 available online at http://www.idealibrary.com on
`
`J. Mol. Biol. (2001) 306, 263±274
`
`Identification of the Domain in the Human
`Interleukin-11 Receptor that Mediates Ligand Binding
`
`Karin Schleinkofer1, Andrew Dingley2, Ingrid Tacken1
`Matthias Federwisch1, Gerhard MuÈ ller-Newen1, Peter C. Heinrich1
`Patricia Vusio3, Yannick Jacques3 and Joachim GroÈ tzinger1*
`
`1Institut fuÈ r Biochemie
`RWTH-Aachen
`UniversitaÈtsklinikum
`Pauwelsstr.30, 52057 Aachen
`Germany
`
`2Institut fuÈ r Physikalische
`Biologie, Heinrich Heine
`UniversitaÈt DuÈ sseldorf
`D-40225, DuÈ sseldorf and
`Forschungszentrum JuÈ lich
`D-52425 JuÈ lich, Germany
`
`3Cytokines et ReÂcepteurs
`Unite 463, 9 Quai Moncousu
`44035 Nantes Cedex 1
`France
`
`The interleukin-11 receptor (IL-11R) belongs to the hematopoietic receptor
`superfamily. The functional receptor complex comprises IL-11, IL-11R
`and the signal-transducing subunit gp130. The extracellular part of the
`IL-11R consists of three domains: an N-terminal
`immunoglobulin-like
`domain, D1, and two ®bronectin-type III-like (FNIII) domains and D2
`and D3. The two FNIII domains comprise the cytokine receptor-hom-
`ology region de®ned by a set of four conserved cysteine residues in the
`N-terminal domain (D2) and a WSXWS sequence motif in the C-terminal
`domain (D3). We investigated the structural and functional role of the
`third extracellular receptor domain of IL-11R. A molecular model of the
`human IL-11/IL-11R complex allowed the identi®cation of amino acid
`residues in IL-11R to be involved in ligand binding. Most of them were
`located in the third extracellular domain, which therefore should be able
`to bind with high af®nity to IL-11. To prove this prediction, domain D3
`of the IL-11R was expressed in Escherichia coli, refolded and puri®ed. For
`structural characterization, circular dichroism, ¯uorescence and NMR
`spectroscopy were used. By plasmon resonance experiments, we show
`that the ligand-binding capacity of this domain is as high as that one for
`the whole receptor. These results provide a basis for further structural
`investigations that could be used for the rational design of potential
`agonists and antagonists essential in human therapy.
`# 2001 Academic Press
`
`*Corresponding author
`
`Keywords: Interleukin-11 receptor; cytokine receptors; ligand binding;
`cytokine receptor-homology region
`
`Introduction
`
`Interleukin-11 is a pleiotropic cytokine playing
`an important role in hematopoiesis (Neben et al.,
`1994; Yonemura et al., 1992) and anti-in¯ammatory
`activities (Redlich et al., 1996; Trepicchio et al.,
`1996). Due to its thrombopoietic potential, it has
`been demonstrated in a phase II clinical trial that
`
`Abbreviations used: HSQC, hetero single quantum
`correlation; INEPT, insensitive nuclear enhanced
`polarization transfer; rmsd, root-mean-square
`differences; ru, resonance units; IL-11R, interleukin-11
`receptor; CNTF, ciliary neurotrophic factor;
`LIF, leukemia inhibitory factor; OSM, oncostatin M;
`CT-1, cardiotrophin-1; CRH, cytokine receptor
`homology region; hGH, human growth hormone;
`hGHR, hGH receptor.
`E-mail address of the corresponding author:
`achim@bionm1.biochem.rwth-aachen.de
`
`recombinant IL-11 reduces chemotherapy-induced
`thrombocytopenia in breast cancer patients (Tepler
`et al., 1996). Furthermore, IL-11 was shown to play
`a critical role in female reproduction: female mice
`lacking the receptor for IL-11 are infertile due to a
`failure of trophoblast implantation (Robb et al.,
`1998).
`IL-11 belongs to the so-called IL-6-type family of
`cytokines, all sharing a four-helix-bundle fold
`(Bazan, 1990a; GroÈ tzinger et al., 1999). All members
`of this family, namely interleukin-6 (IL-6), interleu-
`kin-11 (IL-11), ciliary neurotrophic factor (CNTF),
`leukemia inhibitory factor
`(LIF), oncostatin M
`(OSM) and cardiotrophin-1 (CT-1) signal via a
`gp130 homodimer (Hibi et al., 1990) or a heterodi-
`mer of gp130 and the LIF receptor (Gearing et al.,
`1991) and/or OSM receptor (Mosley et al., 1996).
`Whereas IL-11 and IL-6 use a gp130 homodimer
`(Murakami et al., 1993; Yin et al., 1993), CNTF
`(Davis et al., 1993), LIF (Gearing et al., 1992) and
`
`0022-2836/01/020263±12 $35.00/0
`
`# 2001 Academic Press
`
`

`

`264
`
`Ligand-binding Domain of the IL-11 Receptor
`
`CT-1 (Pennica et al., 1995) use a heterodimer of
`gp130 and LIF receptor for signal transduction.
`OSM is able to signal via a heterodimeric gp130/
`OSM receptor (Ichihara et al., 1997) or gp130/
`LIF receptor, respectively. The binding of IL-11,
`IL-6 and CNTF to their speci®c non-signaling
`a-receptors is a prerequisite for their interaction
`with the signal transducers, whereas OSM and LIF
`bind directly to their signal-transducing subunits.
`In the case of IL-11 and IL-6 gp130, homodimeriza-
`tion leads to the activation of the constitutively
`associated Janus tyrosine kinases (Jaks), which
`then phosphorylate tyrosine residues of the cyto-
`plasmic part of gp130. These phosphotyrosine resi-
`dues are speci®c docking sites for the transcription
`factors of the STAT family (Gerhartz et al., 1996)
`which then themselves become phosphorylated.
`The phosphorylated STATs dimerize and translo-
`cate into the nucleus to activate target gene
`expression (Heinrich et al., 1998; LuÈ tticken et al.,
`1994).
`Two isoforms of the human IL-11 receptor have
`been identi®ed that differ in their cytoplasmic
`domains (Cherel et al., 1995). One isoform of the
`human IL-11 receptor (IL-11Ra1) has a short cyto-
`plasmic domain and is similar to the human IL-6
`receptor and the murine IL-11 receptor. The other
`isoform, similar to the human CNTF receptor,
`lacks this domain (IL-11Ra2). Lebeau et al. (1997)
`described similar functions and properties of both
`isoforms when transfected with gp130 in Ba/F3
`cells. All
`receptors of
`the IL-6-type cytokines
`belong to the cytokine receptor class I
`family
`characterized by the presence of at least one cyto-
`kine receptor homology region (CRH) consisting of
`two ®bronectin type III-like domains (Yamasaki
`et al., 1988). The N-terminal domain of the CRH
`contains four conserved cysteine residues, whereas
`the C-terminal domain is characterized by a
`tryptophan-serine-X-tryptophan-serine
`(WSXWS)
`sequence motif (Bazan, 1990b). The extracellular
`part of the IL-11R is predicted to consist of an
`Ig-like domain (D1) and one CRH (D2 and D3)
`(Cherel et al., 1995). For several members of this
`receptor family, e.g. the IL-6R, the human growth
`hormone receptor, the IL-4 receptor, the prolactin
`receptor and the erythropoietin receptor,
`it has
`been shown that the two domains of the CRH are
`responsible for the interaction with their ligands
`(Wells & de Vos, 1996). In the case of IL-6R, the
`Ig-like domain stabilizes the receptor during intra-
`cellular traf®cking (Vollmer et al., 1999) but is not
`necessary for the assembly of a functional receptor
`(Vollmer et al., 1996; Yawata et al., 1993), whereas
`the third domain is suf®cient for ligand binding
`but not for gp130 association (OÈ zbek et al., 1998).
`So far, the three-dimensional structure of the
`complete CRH of human gp130 as well as its
`C-terminal domain has been solved by X-ray
`(Bravo et
`al., 1998) and NMR spectroscopy
`(Kernebeck et al., 1999), respectively. The latter con-
`sists of seven b-strands constituting a ®bronectin
`type III-like domain. It is structurally related to the
`
`homologous domains of the receptors for erythro-
`poietin, growth hormone and prolactin (Kernebeck
`et al., 1999). Since no structural
`information is
`available for the CRH of
`IL-11R, a molecular
`model of the IL-11/IL-11R complex has been built
`using the X-ray structures of the human growth
`hormone (hGH)/human growth hormone receptor
`complex (hGHR) (de Vos et al., 1992) and human
`CNTF, respectively, as a template (McDonald et al.,
`1995). This model allowed the prediction of amino
`acid residues within human IL-11 to be involved in
`the interaction with IL-11R and gp130, which was
`con®rmed by site-directed mutagenesis (Tacken
`et al., 1999). According to this model, amino acid
`residues within the IL-11R involved in ligand
`binding are located mainly in the third domain. To
`verify
`these predictions
`experimentally,
`this
`domain was expressed in Escherichia coli, refolded
`and puri®ed. Circular dichroism, ¯uorescence and
`NMR experiments were used to characterize the
`recombinant IL-11R-D3. We demonstrate that the
`binding capacity of this domain is as high as that
`for the full-length receptor, con®rming the model
`of the IL-11/IL-11R complex.
`
`Results
`
`Molecular modeling
`
`Figure 1(a) shows the molecular model of the
`human IL-11/IL-11R complex built as described
`in Material and Methods. To assess the qualitiy
`of the molecular model we calculated the amount
`of residues in the not-allowed regions of a Rama-
`chandran plot (IL-11 model, four residues out of
`166; IL-11R model, seven residues out of 202) and
`the rmsd values for equivalent Ca atoms between
`the ®nal models and the templates ( IL-11 model,
`0.8 AÊ ; IL-11R model, 1.1 AÊ ). The epitope of IL-11
`in contact with the IL-11R (site I) comprises the
`beginning of helix A, the end of helix D and the
`end of the AB-loop, and was identi®ed as the IL-
`11R binding epitope by site-directed mutagenesis
`(Tacken et al., 1999). The regions in IL-11R that are
`in contact with site I of IL-11 are located within
`domains D2 and D3 of IL-11R. Figure 1(b) shows
`the sequential alignments with homologous cyto-
`kines and their receptors of the regions involved in
`the interaction between the two molecules. Both
`domains consist of seven b-strands and interact
`with the ligand via their loop regions. These are, in
`particular, the EF-loop in D2 and the B0C-0 and
`F0G0-loops in D3, respectively.
`To identify the residues participating in the
`interaction area, the difference of the accessible
`surfaces (in AÊ 2) of IL-11R was determined in the
`free and ligand-bound state,
`respectively. The
`locations of
`the affected residues within the
`model of
`IL-11R are depicted in Figure 1(c).
`Residues
`contributing most
`to the interaction
`area are F187 and W188 in D2, and P250, H251,
`L253, F298, L299, D300 in D3, respectively. One
`residue (Q213)
`is located in the hinge region
`
`

`

`Ligand-binding Domain of the IL-11 Receptor
`
`265
`
`analysis
`two domains. This
`the
`connecting
`revealed that it is mainly the third domain that
`contributes
`to the interaction area, suggesting
`that this domain should be signi®cantly involved
`in ligand binding.
`
`shape of the spectrum shows that several distinct
`bands attributable to tyrosine and tryptophan side-
`chains are overlaid. The presence of such bands are
`indicative for a protein in its folded state (Grishina
`& Woody, 1994).
`
`Expression, refolding and purification of the
`third domain of IL-11R
`
`Thermal unfolding of the third domain of
`IL-11R
`
`Preliminary studies with the recombinant wild-
`type IL-11R-D3 were hampered by problems due
`to disul®de-linked dimers that were prone to
`form aggregates. Dimers could be detected by
`Western blotting using monoclonal antibodies
`against
`IL-11R-D3 (data not shown). To avoid
`aggregation and dimer
`formation,
`the only
`cysteine
`residue
`(C248)
`in
`IL-11R-D3 was
`mutated to an alanine residue. This mutated
`form of
`the third domain of IL-11R was used
`throughout all the following experiments.
`After expression in E. coli and lysis of cells, the
`protein was found only in inclusion bodies; no
`recombinant protein was detectable in the soluble
`fraction of the lysate. By SDS-PAGE (Figure 2(a))
`the expressed protein was visualized as a 14 kDa
`band. After solubilization of the inclusion bodies in
`guanidine hydrochloride, the protein was refolded
`and puri®ed by size-exclusion chromatography
`(Figure 2(b)).
`
`Circular dichroism spectroscopic
`characterization of the refolded IL-11R-D3
`
`The circular dichroism spectra of the refolded IL-
`11R-D3 in both the far and the near-UV are indica-
`tive of a protein in a folded state (Figure 3(a) and
`(b)). The positive peak at 232 nm re¯ects a positive
`lobe of a couplet, which can be attributed to inter-
`acting tryptophan side-chains (Grishina & Woody,
`1994). This band is observed in the far-UV CD
`spectrum of the third domain of gp130 (MuÈ ller-
`Newen et al., 1997) and IL-6R (OÈ zbek et al., 1998)
`as well as in the corresponding domain of the
`granulocyte-colony-stimulating
`factor
`receptor
`(Anaguchi et al., 1995). This positive peak seems to
`be characteristic for the WSXWS motif present in
`the C-terminal domain of the CRH in all class I
`cytokine receptors. Secondary structure analysis
`(Sreerama & Woody, 1994) of the far-UV CD spec-
`trum re¯ects the b-sheet character of the protein
`(IL-11R-D3, a-helix 5 %, b-sheet 49 %, turn 16 %),
`which is typical
`for a ®bronectin type III-like
`domain (MuÈ ller-Newen et al., 1997; OÈ zbek et al.,
`1998). The folded state of the protein is con®rmed
`by the near-UV CD spectrum (Figure 3(b)). The
`
`The thermal stability of the recombinant protein
`at pH 8.0 was studied by CD spectroscopy in the
`far-UV range. A series of far-UV CD spectra were
`recorded at different temperatures. The prominent
`band at 232 nm, which decreases with increasing
`temperature, was used to monitor the denaturation
`of the protein. Figure 4 shows the ellipticity at
`232 nm as a percentage of the native state of D3, as
`a function of temperature. The transition midpoint
`was estimated to be about 25 C, which indicates a
`remarkably low thermal stability of this domain.
`Above 30 C, the protein precipitated.
`
`Fluorescence anisotropy decay measurements
`
`the time-resolved ¯uorescence
`The results of
`measurements are presented in Table 1. The aniso-
`tropy decay had to be ®tted with two rotational
`correlation times in order to achieve a good ®t. The
`shorter correlation time is in the typical range of
`side-chain motions (Kouyama et al., 1989) while the
`slower one is probably attributed to the overall
`motion of the protein. Assuming a spherical par-
`ticle,
`the theoretical
`rotational correlation time
`ftheor. can be calculated to 6.7 ns, which is shorter
`than the measured value of 12.2 ns. The higher
`value compared to the theoretically calculated one
`might be explained by the presence of the large
`unstructured, loose C-terminal part of the molecule
`and the non-spherical shape of
`the molecule,
`which both slow the rotational motion. On the
`other hand, the presence of high molecular mass
`oligomers can be neglected, as indicated by the
`low limiting anisotropy r1 (Table 2).
`
`NMR spectroscopy
`
`In order to facilitate the elucidation of the sol-
`ution structure of the third domain of IL-11R by
`NMR spectroscopy, we produced 15N-labeled pro-
`tein. The protein sample was prepared in the pre-
`sence of 7 mM arginine and concentrated to a ®nal
`concentration of about 0.2 mM. Figure 5 shows the
`1H-15N-HSQC-NMR spectrum of IL-11R-D3. The
`wide spread of the resonances indicates a folded
`state of the protein. The spectrum shows 92 reson-
`
`Table 1. Fluorescence intensity decay data (Bi is the fractional amplitude associated with the decay time ti; hti is the
`mean lifetime)
`
`B1 (%)
`
`4.8
`
`B2 (%)
`
`23.7
`
`B3 (%)
`
`71.5
`
`t1 (ns)
`
`0.56
`
`t2 (ns)
`
`3.87
`
`t3 (ns)
`
`7.40
`
`w2
`
`1.27
`
`hti (ns)
`
`6.23
`
`

`

`266
`
`Ligand-binding Domain of the IL-11 Receptor
`
`Figure 1 (legend opposite)
`
`ances. Since less than 20 amino acid residues are
`expected to be part of the unstructured C terminus
`and the 17 proline residues cannot be monitored in
`this experiment, the number of resonances ®ts well
`with the expected number of signals for a protein
`consisting of 126 amino acid residues.
`
`Surface plasmon resonance studies
`
`Surface plasmon resonance biosensor analysis
`was used to investigate whether, and to what
`extent, the IL-11R-D3 domain could contribute to
`
`the IL-11R/IL-11 interaction. Similar amounts of
`IL-11R-IL-2 fusion protein (Blanc et al., 2000) and
`IL-11R-D3 were immobilized on separate sensorch-
`ips. As shown in Figure 6(a), the anti-IL-11R mAb
`E24.2 bound to these two sensorchips with similar
`kinetics. The binding capacities show that similar
`numbers of IL-11R and IL-11R-D3 molecules were
`coupled to the matrices with respect to their mol-
`ecular masses. Trx-IL-11 fusion protein also bound
`with similar ef®ciencies to IL-11R-IL-2 and IL-11R-
`D3 (Figure 6(b)). Analysis of
`the sensorgrams
`(Table 3) indicates that the equilibrium dissociation
`
`Table 2. Fluorescence anisotropy decay data (ri are the anisotropies; fi the rotational correlation times)
`w2
`f1 (ns)
`f2 (ns)
`
`r2
`
`r1
`
`ro
`
`r1
`
`0.148
`0.162
`
`0.120
`0.028
`
`-
`0.119
`
`0.028
`0.015
`
`8.5
`0.8
`
`-
`12.2
`
`1.81
`1.44
`
`ftheor. (ns)
`
`6.7
`6.7
`
`

`

`Ligand-binding Domain of the IL-11 Receptor
`
`267
`
`Figure 1. Molecular model of the human IL-11/IL-11R complex. (a) Ribbon representation of the model of the
`IL-11/IL-11R complex. The helices of IL-11 (grey) are designated A, B, C, D. The two domains of the CRH of IL-11R
`(yellow) are designated D2 and D3, respectively. The b-strands of these domains are designated A, B, C, D, E, F, G
`(D2) and A0, B0, C0, D0, E0, F0, and G0 (D3). (b) Sequential alignments of IL-11, IL-6, CNTF and of IL-11R, IL-6R, GHR.
`Shown are regions within the AB-loop and helix D of IL-11 and of the EF-loop in IL-11R-D2 and of the B0C0 and
`F0G0-loop in D3, respectively. Bars above the sequences indicate a-helical or b-strand regions in IL-11 and IL-11R.
`(c) Location of the residues whose surface accessibilties are affected by ligand binding in the IL-11R model. Residues
`with changes of their surface accessibilties greater than 10 AÊ are coloured in red, smaller than 10 AÊ are coloured
`blue. Residues within D2 are labeled green and residues within D3 are labeled orange. The broken lines indicate
`approximately the hinge region between the two domains.
`
`constant depicting Trx-IL-11 binding to IL-11R-D3
`(KD 49 nM) was close to that measured for Trx-IL-
`11 binding to IL-11R-IL-2 (KD 22 nM). Trx-IL-11
`had a slightly higher af®nity for IL-11R-IL-2 than
`for IL-11R-D3, mainly because of a higher associ-
`ation constant. Previous results have shown that
`the af®nity of IL-11 for Ba/F3 cells expressing the
`transmembrane IL-11R is in the range of 10-50 nM
`(Curtis et al., 1997).
`If one calculates the equilibrium dissociation
`constants in terms of free energies of interaction
`( G ˆ RTlog(1/KD)), binding of IL-11 to IL-11R-
`IL-2 or to IL-11R-D3 is accompanied by free energy
`changes of 4.6 or 4.4 kcal/mol, respectively, indi-
`cating that the IL-11R-D3 accounts for about 95 %
`of the free energy of binding to IL-11.
`
`Discussion
`
`The second and the third extracellular domain of
`IL-11R form a cytokine-binding module (CRH)
`characteristic for all members of
`the cytokine
`receptor class I family. This CRH consists of two
`®bronectin type III-like domains, of which the
`N-terminal domain contains a set of four conserved
`cysteine residues and the C-terminal domain
`a WSXWS motif (Bazan, 1990b; Sprang & Bazan,
`1993; Wells & de Vos, 1996). The
`speci®c
`a-receptors of the cytokines IL-6, IL-11 and CNTF
`belong to this family and consist of one CRH and
`an N-terminal Ig-like domain. In the case of IL-6R,
`it has been shown that the Ig-like domain is not
`required for ligand binding (Vollmer et al., 1996).
`
`Table 3. Kinetic (kon, association, koff dissociation) and equilibrium (KD dissociation) constants for the binding of
`Trx-IL-11
`
`IL-11R-IL-2
`IL-11R-D3
`
`kon (M1 s1)
`9.0(0.8)  104
`3.3(0.5)  104
`
`koff (s1)
`2.0(0.08)  103
`1.6(0.04)  103
`
`KD (M)
`22(3.0)  109
`48(8.0)  109
`
`

`

`268
`
`Ligand-binding Domain of the IL-11 Receptor
`
`the third
`Figure 2. Expression and puri®cation of
`domain of IL-11R. (a) SDS-PAGE of the third domain of
`IL-11R after expression in E. coli: M, marker proteins
`used: 67 kDa, 45 kDa, 29 kDa, 24 kDa, 14 kDa; CL, total
`cell
`lysate; SN, supernatant of cell
`lysate; IB, washed
`inclusion bodies; P, after gel-®ltration. (b) Elution pro®le
`of the refolded third domain of IL-11R after size-exclu-
`sion chromatography.
`(A),
`(B) and (C) denote void
`volume,
`third domain of
`IL-11R and waste volume,
`respectively.
`
`The same might be valid for the IL-11R, since a
`fusion protein of IL-11 and IL-11R-D2,3 is suf®cient
`to induce biological activity (P¯anz et al., 1999).
`The epitope on human IL-11 (site I) responsible for
`the interaction with its speci®c a-receptor IL-11R
`has been de®ned by a combined approach of mol-
`ecular modeling and site-directed mutagenesis
`(Tacken et al., 1999). The ®rst detailed study on
`ligand receptor interaction was done on the growth
`hormone/growth hormone receptor complex (de
`Vos et al., 1992), which has served as a paradigm
`for the interaction of cytokines with their receptors.
`Both domains of the CRH of the GHR contribute
`
`Figure 3. CD spectra of the third domain of IL-11R in
`the (a) far, and (b) near UV.
`
`equally to site I af®nity (Clackson & Wells, 1995).
`In contrast, the C-terminal domain of the IL-6R
`CRH (IL-6R-D3) has only a tenfold lower binding
`compared with the whole receptor. A KD of
`385 nM was determined for IL-6R-D3 (OÈ zbek et al.,
`1998), whereas the KD for the whole receptor is
`34 nM (WeiergraÈ ber et al., 1995).
`Based on the X-ray structure of the GH/GHR,
`complex we built a molecular model of the IL-11/
`IL-11R complex. This model reveals that residues
`within IL-11R-D2, F(187) and W(188), although less
`accessible in the complex compared with the
`uncomplexed molecule, show only a few contacts
`to the AB-loop of IL-11. This loop is much shorter
`than in the case of IL-6 and should therefore con-
`tribute less to complex formation. This has led us
`to the conclusion that
`the binding capacity of
`IL-11R-D3 might be even more pronounced than in
`the case of the IL-6/IL-6R complex, suggesting an
`
`

`

`Ligand-binding Domain of the IL-11 Receptor
`
`269
`
`suf®cient for ligand binding and that IL-11R-D2
`plays only a minor role in ligand binding. In the
`case of IL-6R, it has been demonstrated that the
`corresponding domain might be important
`to
`recruit the signal-transducing subunit gp130. It is
`also conceivable that the D3 domain of IL-11R is
`needed for the speci®city of ligand receptor recog-
`nition. The low thermal stability of IL-11R-D3 may
`re¯ect that IL-11R-D2 is necessary for interdomain
`stabilization in the full-length receptor. IL-11R-D3
`N-glycosylation as a stabilizing factor can be
`ruled out, since the protein does not contain an
`N-glycosylation consensus sequence.
`To summarize, the third domains of IL-11R and
`IL-6R are suf®cient for ligand binding, whereas the
`second domains seem to be involved in (a) inter-
`domain stabilization or (b) supplying speci®city
`or (c) formation of an interface to the signal-trans-
`ducing subunit or (d) they are needed to induce a
`conformational change in the ligand that might be
`a prerequisite for the assembly of the ®nal signal-
`ing complex. The stoichiometry of this complex is
`still under debate. A tetrameric receptor complex
`for
`the
`IL-6-type
`cytokines was
`suggested
`by GroÈ tzinger et al. (1997). In this model, site I of
`IL-6 or IL-11 binds to the CRH of their speci®c
`a-receptors. At low concentrations of IL-6/IL-6R
`complexes, a tetramer assembles via sites II and III
`of
`the cytokine with two molecules of gp130,
`which are believed to form a preformed dimer
`at
`the cell surface. At high IL-6/IL-6R concen-
`trations, the tetramer is able to bind an additional
`
`Figure 4. Thermal stability of the third domain of
`IL-11R. Thermal unfolding of
`the third domain of
`IL-11R was monitored by recording a series of far-UV
`CD spectra with increasing temperatures. Normalized
`232 nm ellipticities are plotted as a function of tempera-
`ture. The transition midpoint was estimated to be about
`25 C.
`
`important role of IL-11R-D3 in ligand binding. In
`the present work, this was veri®ed experimentally
`by binding studies using the recombinant protein
`IL-11R-D3. The KD values of the whole IL-11R and
`the IL-11R-D3 are of the same order of magnitude.
`Our
`results
`clearly show that
`IL-11R-D3 is
`
`Figure 5. 1H-15N-HSQC spectrum of the third domain of the IL-11 receptor at 8 C.
`
`

`

`270
`
`Ligand-binding Domain of the IL-11 Receptor
`
`1992) served as a template for the three-dimensional
`model of IL-11R. The initial sequential alignment of IL-
`11R and GHR were generated with the Husar program
`package (EMBL Heidelberg, Germany). Based on this
`alignment, amino acid residues were exchanged in the
`template. Insertions and deletions in IL-11R were mod-
`eled by using a database approach included in the soft-
`ware package WHATIF (Vriend, 1990). The database
`was searched for a peptide sequence of the appropriate
`length, which was ®tted to the template. All loops were
`selected from the database so as to give a minimum rms
`distance between the ends of the loops. In a last step,
`the three-dimensional structural models were energy-
`minimized,
`using
`the
`steepest-descent
`algorithm
`implemented in the GROMOS force-®eld (van Gunsteren
`et al., 1996). To create the IL-11/IL-11R complex, the cor-
`responding IL-11 and IL-11R molecules were oriented to
`each other using the GH/GHR complex as a template
`(De Vos et al., 1992). In a ®rst step, the IL-11 molecule
`was ®tted onto the GH structure (using only the Ca
`positions of the helices A, B, C and D) and the IL-11R
`molecule was ®tted onto the GHR structure, which is
`bound to site I of GH. The next steps were performed in
`an iterative manner. First, the interaction areas were
`inspected for overlapping side-chains. These unfavorable
`contacts were then eliminated by rotating them properly.
`Second, the accessible surface was calculated for this
`complex to ®nd cavities in the interaction area.
`If
`possible, these cavities were ®lled by adjustment of side-
`chains from their neighborhood. These complexes were
`then energy-minimized and again analyzed for unfavor-
`able contacts and cavities in the interaction area. This
`procedure was repeated until a low-energy conformation
`of the complex was reached. Accessible surfaces were
`calculated using the algorithm implemented in the soft-
`ware package WHATIF (Vriend, 1990). For graphical
`representation, the Ribbons program (Carson, 1991) and
`GRASP was used (Nicholls et al., 1991). All programs
`were run on a Silicon Graphics Indigo work station. The
`residues of IL-11R-D3 were numbered according to the
`dataset with the accession code I37891.
`
`Construction of an IL-11R-D3 expression vector
`
`All enzymes used for DNA ampli®cation and modi®-
`cation were obtained from Boehringer Mannheim (Man-
`nheim, Germany) and AGS (Heidelberg, Germany). The
`IL-11R-D3-encoding region of the IL-11R was ampli®ed
`by PCR using IL-11R-cDNA as a template (Cherel et al.,
`1995). NcoI and BamHI sites were introduced into the
`50 and 30-primers to enable cloning of the ampli®ed DNA
`in the NcoI and BamHI sites of the prokaryotic expression
`vector pET3d (Studier et al., 1990) (sense primer, 50 GGT
`GGT GCC ATG GAG AGC ATC TTG CGC CCT GAC
`30; antisense primer, 50 CCG GAA GCT TAC TCC ACC
`TCT GGC TGC GT 30). The C248A IL-11R-D3 mutant
`was created by PCR overlap (Ho et al., 1989) using the
`following oligonucleotide and respective antisense oligo-
`nucleotide (50 TCC TGG CCG GCC CAG CCC CAC TTC
`CTG 30; 50 GTG GGG CTG GGC CGG CCA GGA GGC
`AGG 30). PCR was performed in a total volume of 100 ml
`of PCR buffer (10 mM KCl, 20 mM Tris HCl (pH 8.8),
`10 mM (NH4)2SO4, 2 mM MgSO4, 0.1 % (v/v) Triton
`X-100, 0.2 mM each dNTP) containing 0.25 mg of tem-
`plate DNA, 0.75 unit of Vent DNA polymerase and
`1 mM IL-11R-D3 sense and antisense primer. PCR was
`performed with 30 cycles of denaturation (one minute,
`93 C), primer annealing (two minutes, 50 C) and
`elongation (two minutes, 72 C). PCR products were
`
`Figure 6. Surface plasmon resonance studies. Biosen-
`sor analysis of IL-11 and anti-IL-11R interaction with
`IL-11R and IL-11R-D3. Similar amounts of IL-11R-IL-2
`fusion protein or IL-11R-D3 domain were immobilized
`on parallel CM5 sensorchips.
`(a) The sensorgrams
`depicts the binding of anti-IL-11R mAb E24.2 or (b) of
`Trx-IL-11 to IL-11R-IL-2 (continuous lines) or to IL-11R-
`D3 (broken lines). E24.2 and Trx-IL-11 were used at con-
`centrations of 5 mg/ml and 10 mg/ml, respectively. The
`association phase starts at zero time and the dissociation
`phase is initiated at ten minutes.
`
`forming a hexameric non-
`IL-6/IL-6R complex,
`(GroÈ tzinger
`al.,
`1999).
`signaling complex
`et
`Recent data suggest a mechanism of a sequential
`recruitment of the two gp130 molecules by the
`IL-6/IL-6R complex in which site III occupancy is
`a prerequisite for the site II interaction (unpub-
`lished results). In analogy, the interaction of the
`IL-11R-D2 or IL-6R-D2 with IL-11 or IL-6, respect-
`ively, might induce a conformational change in site
`III of the ligand required for the interaction with
`th