`
`Article
`
`The Dynamics of Signal Triggering
`in a gp130-Receptor Complex
`
`Rishi Matadeen,1,* Wai-Ching Hon,1 John K. Heath,2 E. Yvonne Jones,1 and Stephen Fuller1
`1 Division of Structural Biology, The Wellcome Trust Centre for Human Genetics, University of Oxford, Roosevelt Drive,
`Headington, Oxford OX3 7BN, United Kingdom
`2 Cancer Research UK, School of Biosciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, United Kingdom
`*Correspondence: rishi@strubi.ox.ac.uk
`DOI 10.1016/j.str.2007.02.006
`
`Open access under CC BY license.
`
`SUMMARY
`
`gp130 is a shared signal-transducing mem-
`brane-associated receptor for several hemato-
`poietic cytokines. The 30 A˚
`resolution cryo-
`electron microscopy (cryo-EM) structure of the
`Interleukin 11(IL-11)-IL-11 Receptor-gp130 ex-
`tracellular complex reveals the architecture
`and dynamics of this gp130-containing signal-
`ing complex. Normal-mode analysis reveals
`a repertoire of conformational changes that
`could function in signal triggering. This sug-
`gests a concerted mechanism of signaling in-
`volving all the components of the complex.
`This could provide a general mechanism of sig-
`nal transfer for cytokines utilizing the JAK-STAT
`signaling cascade.
`
`INTRODUCTION
`
`The cell-surface receptor gp130 transduces signals in-
`volved in the regulation of a wide variety of adult tissue
`systems, including the hematopoietic system, nervous
`system, bone, heart, adipose tissue, testes, liver, and
`muscle (Bravo and Heath, 2000). Signaling results from
`the extracellular, ligand-mediated formation of oligomeric
`receptor complexes. These complexes can differ in com-
`position and stoichiometry depending upon the ligand, but
`they require the incorporation of one or more molecules of
`gp130 (Hibi et al., 1990). The ligands are termed the
`gp130-binding cytokines and include Interleukin 6 (IL-6),
`Interleukin 11 (IL-11), Ciliary Neurotrophic Factor (CNTF),
`Oncostatin M (OSM), and Leukemia-Inhibitory Factor
`(LIF) (Kishimoto et al., 1995; Taga and Kishimoto, 1997).
`IL-6 and IL-11 signaling occurs via homodimerization of
`gp130 after the formation of a hexameric complex con-
`taining two molecules of gp130, two molecules of IL-6 or
`IL-11 ligand, and two molecules of a specific, nonsignaling
`IL-6 or IL-11 receptor (termed IL-6R and IL-11R, respec-
`tively) (Ward et al., 1994; Paonessa et al., 1995; Barton
`et al., 2000).
`Crystal structures of the gp130-binding cytokines IL-6,
`LIF, CNTF, and OSM have revealed a highly conserved
`four-a helix bundle topology (Bravo and Heath, 2000).
`
`The sequences of gp130, and other cognate receptors
`for the four-a helix bundle cytokines, are distinguished
`by a tandem pair of distinctive fibronectin type III (FNIII)-
`like domains in their N-terminal extracellular regions (Cos-
`man, 1993). This is termed the cytokine-binding homology
`region (CHR). Crystal structures have been determined for
`the CHR of gp130 (Bravo et al., 1998), the entire extracel-
`lular region of IL-6R (Ig-like plus cytokine-binding do-
`mains) (Varghese et al., 2002), and for complexes involv-
`ing soluble fragments of gp130 (Chow et al., 2001;
`Boulanger et al., 2003a, 2003b). These structural analy-
`ses, combined with functional studies, have provided sig-
`nificant insights into the key features and characteristics
`of the interaction surfaces required for affinity and speci-
`ficity in gp130-binding cytokine-receptor recognition.
`Many of the remaining questions concerning the trigger-
`ing of signaling by gp130-binding cytokines now relate to
`the larger-scale architecture and dynamics of the cyto-
`kine-receptor assemblies. The contribution of the N-termi-
`nal Ig-like domain of gp130 to the formation of the IL-6 sig-
`naling complex has been addressed by crystal structures
`from Garcia and coworkers (Chow et al., 2001; Boulanger
`et al., 2003b). However, little structural data are available
`to illuminate the role of the three FNIII-like domains that
`are predicted by sequence analysis to form the C-terminal
`half of the gp130 extracellular region. Biophysical data
`have indicated that interactions involving these domains
`do contribute to complex formation (Boulanger et al.,
`2003b). Furthermore, functional data point to a clear role
`for the FNIII-like domains of gp130 in signaling (Timmer-
`mann et al., 2002).
`The complete extracellular portion of a gp130-binding
`cytokine-receptor signaling complex has evaded crystalli-
`zation. Lower-resolution EM studies provide an alternative
`route for structural analysis of the gp130-binding cytokine
`system (Skiniotis et al., 2005). Cryo-electron microscopy
`(cryo-EM) in particular allows for the investigation of the
`structural dynamics. We have used this approach for the
`entire extracellular region of the hexameric signaling com-
`plex formed by IL-11. IL-11 signals through a complex
`comprising a 2:2:2 stoichiometry of IL-11, IL-11R, and
`gp130 (Barton et al., 2000). In the following sections, we
`refer to the soluble form of this complex as IL-11H. Our
`single-particle cryo-EM reconstruction of IL-11H provides
`the first, to our knowledge, complete native structure for
`the extracellular portion of a gp130-binding cytokine
`
`Structure 15, 441–448, April 2007 ª2007 Elsevier Ltd All rights reserved 441
`
`Lassen - Exhibit 1022, p. 1
`
`
`
`Signal Triggering in a gp130-Receptor Complex
`
`Structure
`
`Figure 1. Cryo-EM of the IL-11-IL-11R-
`gp130 Complex
`(A) Part of a typical electron micrograph of the
`purified IL-11-IL-11R-gp130 complex. Arrows
`identify examples of the complexes within vit-
`reous ice. The scale bar represents 50 nm on
`the specimen scale.
`(B) A summary of the three-dimensional image
`analysis. Characteristic class averages ob-
`tained by multireference alignment and classi-
`fication are shown in row 1. Surface views from
`the three-dimensional structure of the IL-11-IL-
`11R-gp130 complex, in the Euler directions as-
`signed to the averages, are shown in row 2.
`Row 3 contains reprojected images from the
`three-dimensional structure in the Euler direc-
`tions found for the corresponding averages.
`
`signaling complex in vitrified water. The reconstruction re-
`veals a distinct head and two leg regions. The cryo-EM
`map is used to construct a discrete map via the tech-
`niques of vector quantization. This model structure, de-
`rived directly from the cryo-EM reconstruction, has en-
`abled the calculation of normal modes of the complex,
`which allows us to catalog the conformational changes
`that may participate in cellular communication. In addi-
`tion, the architecture of the cryo-EM map convincingly
`accommodates a modeled crystal structure of the hex-
`americ IL-6 signaling complex (which includes the mem-
`brane-distal half of gp130 [Boulanger et al., 2003b]), plus
`two ‘‘beads on a string’’ arrangements of FNIII domains
`(consistent with the membrane-proximal halves of two
`copies of gp130). The combination of the cryo-EM and
`normal-mode analyses allows the correlation of the den-
`sity distribution in the cryo-EM map to the calculated
`dynamical properties. This provides a novel insight into
`the mechanism of signal triggering, relevant to several
`gp130-containing receptor complexes.
`
`RESULTS
`
`Cryo-EM Structure of IL-11H
`The soluble IL-11H complexes were prepared as de-
`scribed in Experimental Procedures. A summary of the
`
`image analysis is shown in Figure 1. A total of 830 images
`were extracted from 15 micrographs and were subjected
`to single-particle analysis as described in Experimental
`Procedures.
`The extracellular portion of IL-11H is a flat, ring-shaped
`complex (Figures 2A–2C). The dimensions of the hexamer
`are 150 A˚ 3 150 A˚ 3 80 A˚ , and the central hole is 50 A˚
`in
`diameter (Figures 2A and 2B). The top of the complex
`(head, Figure 2) is larger than the bottom (legs, Figure 2B).
`The two-fold symmetric (C2) structure comprises identical
`subunits that twist toward a point of contact at the bottom
`of the molecule (Figures 2B and 2C). The two-fold axis
`of symmetry runs from the head across the central hole
`to the bottom of the legs. The resolution of the reconstruc-
`tion is 30 A˚ , as estimated from the spatial frequency in-
`tersection of
`the Fourier shell correlation (FSC) plot
`and the 3s threshold function corrected for the C2 point
`group symmetry (Orlova et al., 1997) (Figure 3). A resolu-
`tion of 30 A˚
`is also predicted by using a 0.2 crosscor-
`relation threshold (Rosenthal and Henderson, 2003). The
`drop in the FSC plot is steep, indicating that the resolu-
`tion estimate is robust and that the number of images
`used in the analysis is sufficient. The correlation co-
`efficient of the projections of the map and its enantiomer
`is 0.97, indicating a weakly handed structure (van Heel
`et al., 1996).
`
`442 Structure 15, 441–448, April 2007 ª2007 Elsevier Ltd All rights reserved
`
`Lassen - Exhibit 1022, p. 2
`
`
`
`Structure
`
`Signal Triggering in a gp130-Receptor Complex
`
`Figure 3. Fourier Shell Correlation Plot from the IL-11H Data
`Set
`Fourier shell correlation curve from the IL-11H data set is shown in red
`plot. The 3s threshold curve is multiplied by O2 = 1.41 to account for
`the two-fold redundancy of the C2 pointgroup symmetry of the data
`(blue curve). The resolution of the reconstruction is 30 A˚ (correspond-
`ing to a spatial frequency of 0.033 A˚ 1). The position of 30 A˚ resolution
`is marked.
`
`model for IL-11H was then subjected to normal-mode
`analysis (NMA). This technique has been shown to be
`very useful in the study of protein motions (Tama and Sane-
`jouand, 2001). NMA provides a repertoire of possible con-
`formational changes based on the discrete model calcu-
`lated from the cryo-EM map. We can use this information
`as a complement to cryo-EM to visualize the dynamics of
`the structure. The extent of functionally important large-
`scale rearrangements of molecular structures and assem-
`blies is well represented by the lowest-frequency normal
`modes (Tama et al., 2003). ElNe´mo, the web interface to
`the Elastic Network Model, provides a tool to compute, vi-
`sualize, and analyze low-frequency normal modes of large
`macromolecules (Suhre and Sanejouand, 2004). Several
`motions show potential functional significance and are
`supported by the flexible regions in the cryo-EM map (Fig-
`ure 6; Movies S1–S3, see the Supplemental Data available
`with this article online). The lowest mode (mode 1, after the
`six global translation and rotation modes) describes the
`motion of the gp130 legs and reveals a change from an ex-
`tended to a compressed conformation (Movie S1). These
`motions are also seen in modes 2 and 3 (Movies S2 and
`S3, respectively), although these modes are at different
`normalized frequencies. Modes 1 and 3 comprise a rota-
`tion around the C2 symmetry axis of the hexamer, indicat-
`ing a conformational twist that could be important in posi-
`tioning intracellular components. Modes 2 and 3 show
`a ‘‘flapping’’ motion of the head region. Interestingly, these
`dynamical properties calculated from NMA are also sup-
`ported by the motions suggested by the density distribu-
`tion profile of the cryo-EM map (Figure 4).
`
`Fitting of the IL-6 Hexameric Complex Crystal
`Structure into the Cryo-EM Map
`There is no atomic structure of complete IL-11H;
`however, atomic structures are available from X-ray
`
`Figure 2. Orthogonal Views of the IL-11H Complex
`(A–C) The (A) top, (B) front, and (C) side views of the IL-11H complex
`are contoured at 2.5s.
`
`Flexibility in the Cryo-EM Map
`The electron potential of the structure gives rise to the ob-
`served density in a cryo-EM reconstruction. The strength
`of the observed density is modulated by variation in flexi-
`bility between the particles incorporated into the recon-
`struction. In the IL-11H cryo-EM structure, the head of
`the molecule is of stronger density than the legs (Figure 4).
`However, a relatively strong density is seen where the two
`protruding legs meet. Weak density areas exist between
`the head and legs of the complex and indicate mobile re-
`gions within the hexamer. Another weak area of density is
`located at the center of the head region (Figure 4).
`
`Elastic Normal Modes of the Discretized
`Cryo-EM Map
`The cryo-EM map was vector quantized to create a dis-
`crete, reduced description of its continuous shape/mass
`distribution (Wriggers et al., 1999). The docking program
`Situs (Wriggers et al., 1999) was used for the vector quan-
`tization of the cryo-EM electron density. A total of 50 so-
`called codebook vectors were chosen to describe the
`cryo-EM map, a sufficient number for reconstructions at
`a relatively low resolution (30 A˚ ) (Tama et al., 2002) (see
`Experimental Procedures). This discrete elastomechanical
`
`Structure 15, 441–448, April 2007 ª2007 Elsevier Ltd All rights reserved 443
`
`Lassen - Exhibit 1022, p. 3
`
`
`
`Signal Triggering in a gp130-Receptor Complex
`
`Structure
`
`brane-proximal domains of gp130 (Sharma et al., 1999).
`Each segment consists of three FNIII domains. These
`components can be fitted into the IL-11H cryo-EM map.
`We have initially docked the crystal structure of the IL-6
`cytokine-binding complex into the head region of the
`IL-11H cryo-EM reconstruction by using the graphics
`program ‘‘O’’ (Kleywegt and Jones, 1997) to occupy as
`much of the head volume as possible. FNIII repeats from
`two segments of human fibronectin (Sharma et al., 1999)
`(PDB code 1FNH) were then inserted into the leg region
`of the cryo-EM map (one segment for each leg region) to
`represent
`the membrane-proximal domain of gp130.
`Each FNIII domain was rotated to satisfy the kink of the
`leg region. A model of complete extracellular IL-11H was
`thereby created. The fit of this hand-docked model and
`the cryo-EM map correlate with a coefficient of 0.54.
`The fit was refined by using the program URO (Navaza
`et al., 2002) (see Experimental Procedures). The result of
`the refined fit is shown in Table 1. The correlation between
`the observed and calculated structure factors is 0.656 and
`indicates a good fit; however, the R factor is high at 0.658,
`consistent with significant structural flexibility in the cryo-
`EM map compared to the modeled hexamer. Table 2
`shows values calculated from contributions of the cryo-
`EM reconstruction exclusively around the model. Conse-
`quently, only the solid, nonflexible portions of the recon-
`struction are considered in the optimization procedures.
`The correlation between the observed and calculated
`structure factors is 0.918, and the R factor 0.297, indicat-
`ing a good fit between the model and the nonflexible por-
`tions of the cryo-EM map.
`
`Composite Homology Model of IL-11H
`The modeled IL-11H and the cryo-EM map (Figure 5) have
`a correlation coefficient of 0.65 (Table 1). Our fit (Figure 5)
`places the FNIII domains representing the gp130 mem-
`brane-proximal domains in the density of the complex’s
`legs. The legs of the gp130 homodimer meet at their re-
`spective D6 domains (Figure 4). No contact is seen be-
`tween the two D5 domains, in contradiction to other
`models (Moritz et al., 2001). The homology models of
`IL-11R, and membrane-distal portions of
`the IL-11,
`gp130 fill the head region of the complex. The absence
`of density corresponding to D1 of the IL-11Rs could be ex-
`plained by the flexibility in the linker regions between the
`D1 and D2 domains. The density between the two cyto-
`kine ligands is relatively weak (Figure 4). This may arise
`from movements in D2 of gp130.
`
`Figure 4. Density Distribution in the Cryo-EM Map
`Percent occupancy is defined as the density distribution relative to the
`highest density in the map. The density is strongest at the head and
`central region of the legs. The region connecting the head and legs
`shows lower density, suggesting greater flexibility.
`
`crystallographic analysis of several components of the
`hexamer and structurally related molecules. We have
`taken the IL-6 cytokine-binding complex (Boulanger
`et al., 2003b) (PDB code 1P9M) to be structurally homolo-
`gous to an IL-11 cytokine-binding complex. An IL-11
`model has shown a ‘‘four-helix bundle’’ protein fold similar
`to that of IL-6 (Barton et al., 1999). In addition, the extra-
`cellular domains of IL-11R and IL-6R have a 24% se-
`quence identity (Curtis et al. 1997). FNIII repeats from
`two segment of human fibronectin (PDB code 1FNH)
`were taken to be homologous to the extracellular mem-
`
`Table 1. Result from the URO Fitting Procedure
`
`Fitting: URO
`
`a
`
`75.2
`
`b
`
`91.6
`
`g
`
`90.2
`
`X
`
`Y
`
`Z
`
` 0.0101
`
` 0.0079
`
`0.0178
`
`Cf
`
`65.6
`
`Rf
`
`65.8
`
`Fm
`
`29.6
`
`a, b, and g represent the Euler angles of the model; X, Y, and Z are the fractionary translations of the model; Cf is the correlation
`between the observed and calculated structure factors (3 100); Rf is the crystallographic R factor (3 100); and Fm is the value of the
`optimized function (quadratic misfit) (Navaza et al., 2002).
`
`444 Structure 15, 441–448, April 2007 ª2007 Elsevier Ltd All rights reserved
`
`Lassen - Exhibit 1022, p. 4
`
`
`
`Structure
`
`Signal Triggering in a gp130-Receptor Complex
`
`Table 2. Result from the URO Fitting Procedure Calculated from Contributions of the Cryo-EM Reconstruction
`Exclusively around the IL-11H Model
`
`Fitting: URO
`
`a
`
`74.7
`
`b
`
`91.7
`
`g
`
`90.3
`
`X
`
`Y
`
`Z
`
` 0.0079
`
` 0.0193
`
`0.0175
`
`Cf
`
`91.8
`
`Rf
`
`29.7
`
`Fm
`
`8.0
`
`a, b, and g represent the Euler angles of the model; X, Y, and Z are the fractionary translations of the model; Cf is the correlation
`between the observed and calculated structure factors (3 100); Rf is the crystallographic R factor (3 100); and Fm is the value of the
`optimized function (quadratic misfit) (Navaza et al., 2002).
`
`DISCUSSION
`
`The IL-6-type hematopoeitic cytokines signal via the
`gp130 receptor. IL-11 and IL-6 initially bind to their spe-
`cific receptors, and the complex associates with mem-
`brane-bound gp130,
`forming the functional signaling
`complex. The cytokine-binding event occurs at D2 and
`D3 of the gp130s and is transmitted into the cell through
`D4, D5, and D6. The IL-11H cryo-EM map enables several
`regions of the functional complex to be localized by fitting
`a homology model. The ligand-binding region of the com-
`plex is situated at the head of the cryo-EM reconstruction
`(Figure 4). This region is composed of
`the bound
`
`into the
`
`Figure 5. Fitting of the IL-6-IL-6R-gp130 Model
`Cryo-EM Density
`The cryo-EM map is shown as a gray, semitransparent surface. The
`fitted model is shown as a ribbon representation. The IL-6 cytokines
`(representing IL-11) are shown in gold,
`the IL-6R (representing
`IL-11R) D2 and D3 domains are shown in blue, the gp130 homodimer
`membrane-distal D1–D3 domains are shown in green (gp130 head),
`and the gp130 homodimer membrane-proximal D4–D6 domains are
`shown in red (gp130 legs). The D1 domains of IL-6R (IL-11R) are not
`included. The position of
`the cell membrane is shown by the
`schematic.
`
`cytokines, the D1–D3 domains of the specific receptors,
`and the D1–D3 domains of the gp130s. The signal-trigger-
`ing region is located in the legs of IL-11H. This contains the
`membrane-proximal domains of the gp130s (D4–D6). The
`gp130 legs are readily located in the cryo-EM reconstruc-
`tion. The legs protrude from the head to meet and cross at
`their respective D6 domains. This indicates that signaling
`is transferred to the opposite side of the molecule from
`where the binding event takes place. Each D6 domain
`leads into a transmembrane region. The close coupling
`of the two D6 domains in the IL-11H structure implies
`a close association of these gp130 transmembrane re-
`gions. The D5 domains may be important in orientating
`the D6 domains since deletion abrogates signal transduc-
`tion (Kurth et al., 2000).
`The cryo-EM of particles within vitreous ice allows for
`the direct imaging of the sample density. The density dis-
`tributions within the cryo-EM maps obtained subse-
`quently have information pertaining to the dynamics of
`the object. The cryo-EM reconstruction, together with
`NMA on the discrete map, has allowed us to directly cal-
`culate potential movements between these domains and
`correlate the pattern of motions with density distributions
`of the map. The discrete map is derived from vector quan-
`tization of the cryo-EM map and thus contains information
`pertaining directly to the IL-11R cryo-EM reconstruction.
`The gp130 legs have a ‘‘c-like’’ configuration and conse-
`quently appear compressed in the hexameric complex
`(Figure 6). This allows structural adaptability of the mem-
`brane-proximal domains in the height of
`the ligand-
`binding domains with respect to the cell surface, as dem-
`onstrated by NMA (Figure 6; Movies S1–S3). Indeed, this
`adaptability suggests that unliganded gp130 is elongated
`so that the positioning of the cytokine-binding domains
`is more distal to the membrane than it is in bound
`gp130. The flexibility in the gp130 legs may be an impor-
`tant factor in orientating the cytoplasmic intracellular do-
`mains of gp130 and their associated Janus Kinases
`(JAKs), promoting signal transduction (Socolovsky et al.,
`1999). The rotation around the C2 symmetry axis shown
`by NMA would facilitate this association.
`A three-dimensional EM map of the related IL-6-IL-6Ra-
`gp130 hexameric complex in negative stain has been
`calculated (Skiniotis et al., 2005). This shows a related
`structure consisting of an elongated bipartite head do-
`main and two leg domains, one of which is represented
`by a straight density and the other of which is represented
`by a kinked density—an arrangement that distorts the
`
`Structure 15, 441–448, April 2007 ª2007 Elsevier Ltd All rights reserved 445
`
`Lassen - Exhibit 1022, p. 5
`
`
`
`Signal Triggering in a gp130-Receptor Complex
`
`Structure
`
`Figure 6. Schematic of the Movements
`of IL-11H Calculated from NMA
`Three orthogonal views (i, ii, iii) of the discrete
`map of IL-11H, composed of 50 codebook
`vectors (purple spheres), are shown for the
`three lowest-frequency normal modes (A–C).
`The black arrows indicate the directions of
`motion.
`
`structural C2 symmetry. The distortion of the map could
`be due to preparation artifacts of the negative staining
`technique, which additionally eliminates the dynamics of
`the structure.
`Structural and functional data for the cytokine EPO and
`its receptor (EPOR) have indicated that there are stringent
`requirements for the relative positioning of the two signal-
`ing receptors when crosslinked by the cytokine ligand if
`signal is to be transduced (Livnah et al., 1996, 1998). In ad-
`dition the orientations of the extracellular EPOR D1 and D2
`domains in the signal-competent complex provide a sensi-
`tive adjustment for the positioning of their cytoplasmic
`components involved in signal transduction, resulting in
`the modulation of the signal intensity (Wilson and Jolliffe,
`1999). The signaling assemblies involving gp130 are
`based on a more complex architecture. Consequently,
`the orientations of the IL-11H extracellular domain could
`provide more extended, exquisite adjustments of cyto-
`plasmic components that modulate signal intensity. These
`orientations are suggested by NMA through the motions
`and multiple configurations of the IL-11H discrete map,
`in particular with the movements of the gp130 mem-
`brane-proximal domains. Indeed, all of the component
`molecules contribute to the dynamics of the hexamer;
`thus, a concerted mechanism may be involved in reorien-
`tating the intracellular components into a configuration
`that allows for signal transduction. Initiation of the JAK/
`STAT pathway in response to IL-11 involves JAK-1 and
`predominantly STAT3 (Dahmen et al., 1998).
`Cryo-EM, together with NMA, has now illuminated the
`relative positioning and dynamics of the three mem-
`brane-proximal FNIII-like domains in IL-11H. Furthermore,
`functional data point to a clear role for the FNIII-like do-
`mains of gp130 in signaling (Timmermann et al., 2002). Cy-
`
`tokines such as LIF, OSM, and CNTF utilize a heterodimer
`of gp130 and the LIF receptor (LIFR) to transduce a signal
`(OSM can use the OSM receptor [OSMR] instead). LIFR is
`similar to gp130, although it possesses an extra N-terminal
`(membrane-distal) copy of a cytokine-binding domain (He
`et al., 2005). The signaling complexes formed with these
`factors may have a similar conformation and dynamics in
`the membrane-proximal regions as the gp130s in IL-11H.
`Consequently, signal triggering and signal modulation
`could occur in a similar fashion. However, further structural
`and dynamic studies on these liganded receptor com-
`plexes and their unliganded receptor complexes are
`needed to confirm the mechanism of signal activation.
`
`EXPERIMENTAL PROCEDURES
`
`Expression and Purification of the IL-11-IL-11R-gp130 Complex
`Murine IL-11 (residues 22–199) was subcloned into the pET15b vector
`(Novagen), and the His6-tagged fusion protein, expressed in E. coli,
`was purified by metal-chelating affinity chromatography (TALON, Clo-
`netech). The extracellular regions of murine IL-11R (residues 1–361)
`and gp130 (residues 1–617) were expressed as secreted, soluble,
`C-terminally tagged Fc fusion constructs in stably transfected Dro-
`sophila S2 cells and transiently transfected human epithelial kidney
`293T cells, respectively. Both soluble receptors were purified from
`the media by adsorption to protein A Sepharose (Amersham Biosci-
`ences) and were eluted from the matrix after on-column cleavage
`with the rhinovirus 3C protease. The three affinity-purified subunits
`were mixed with IL-11 in molar excess and concentrated, and the
`complex was purified by gel filtration with the Superdex 200 (10/30)
`column (Amersham Biosciences). The purified complex used for
`cryo-EM imaging was in a buffer of 5 mM HEPES (pH 7.5), 100 mM
`NaCl, 1 mM DTT, and 0.25% b-octylglucopyranoside.
`
`Cryo-Microscopy and Three-Dimensional Image Processing
`Purified IL-11H, at a concentration of 1 mg/ml, was applied to a holey
`carbon film and freeze plunged into liquid ethane after blotting off
`
`446 Structure 15, 441–448, April 2007 ª2007 Elsevier Ltd All rights reserved
`
`Lassen - Exhibit 1022, p. 6
`
`
`
`Structure
`
`Signal Triggering in a gp130-Receptor Complex
`
`excess fluid. The vitrified samples were imaged in a Philips CM200
`FEG at liquid nitrogen temperatures by using a Gatan 626 cryoholder
`and cryotransfer system. Images were taken at defocus values ranging
`from 3 to 5 mm and at an electron optical magnification of 50,000.
`Micrographs were digitized by using a UMAX Powerlook 3000 at
`a step size of 8.3 mm, corresponding to 1.67 A˚ on the specimen scale.
`This resulted in 160 3 160 pixel IL-11H images. These images were
`binned by a factor of 2 for final analysis, resulting in a pixel sampling
`distance of 3.3 A˚ .
`Image processing was performed with the
`IMAGIC-5 software package (van Heel et al., 1996). The full data set
`consisted of 830 images from 15 micrographs. The IL-11H particles
`were interactively selected by using the DISPLAY module. CTF com-
`pensation was performed on images as described (Matadeen et al.,
`1999). The reconstruction was calculated de novo; the images were
`first aligned by using the reference-free alignment-by-classification
`procedure,
`followed by multistatistical analysis (MSA)
`(van Heel,
`1989). Averages of the data set were used to center the images within
`the data set; three rounds of iteration were performed. MSA was then
`used to generate class averages. Initially, 40 class averages were gen-
`erated and used for further center refinement by using multireference
`alignment (MRA). The images were subjected to five cycles of MRA
`and MSA. Relative orientations of ten averaged images (those with
`the lowest variance) were determined by using angular reconstitution
`(van Heel, 1987). The orientations obtained were used to calculate
`a three-dimensional map by using the exact filter back-projection algo-
`rithm (Harauz and van Heel, 1986). Projections from the map were
`compared to their corresponding averages as a measure of the consis-
`tency of the analysis. A total of 70 projections were calculated from the
`map and were used in a new round of MRA followed by MSA. From
`100 averages, 50 (consisting of a total of 622 particles) were selected
`for angular reconstitution, and a new three-dimensional map was cal-
`culated by using the resultant orientations. The process of analysis
`was iteratively refined until stability in alignment, resulting orientations,
`and FSC were observed.
`
`Fitting of the IL-11H Model
`The X-ray crystallographic model of the IL-6 cytokine-binding complex
`(Boulanger et al., 2003b) together with six FNIII repeats (Sharma et al.,
`1999) were manually docked into the IL-11H cryo-EM map by using the
`program ‘‘O’’ (Kleywegt and Jones, 1997). The fitting refinement pro-
`cedure was performed in URO (Navaza et al., 2002). This program
`uses an adapted rigid-body refinement to perform the fitting of molec-
`ular models to EM reconstructions in reciprocal space. Six positional
`variables are used in the refinement (Euler angles a, b, and g and trans-
`lational parameters x, y, and z). Data from 260 A˚ to 30 A˚ were used in
`the refinement procedure. Four optimization cycles were performed
`until no shifts in the coordinates were observed.
`
`Representation of the IL-11H Cryo-EM Map and Model
`Volume rendering and density representation were performed in
`Opendx (http://www.opendx.org) and pymol (DeLano, 2002). For final
`representation of the results, the high-frequency components of the
`three-dimensional map are suppressed by a Gaussian low-pass filter
`with a 1/e width corresponding to 30 A˚ . Figure 4 was prepared
`with RASMOL (Sayle and Milner-Whilte, 1995), BOBSCRIPT (Esnouf,
`1997), and RASTER3D (Merritt and Bacon, 1997). The codebook vec-
`tors were calculated by using the qvol program in Situs (Wriggers et al.,
`1999). A total of 50 codebook vectors were calculated, and the resul-
`tant discretized map was subjected to NMA with the ElNemo web
`server (http://igs-server.cnrs-mrs.fr/elnemo/) by using a cut-off value
`of 30 A˚ .
`
`Supplemental Data
`Supplemental Data include movies showing the dynamics of the
`IL-11H schematized in Figure 6 and are available at http://www.
`structure.org/cgi/content/full/15/4/441/DC1/.
`
`ACKNOWLEDGMENTS
`
`We acknowledge L. Lyne for assistance with tissue culture and B.
`Gowen for support in data collection. R.M. is supported by the Well-
`come Trust, and S.F. is a Wellcome Trust Principal Research Fellow.
`E.Y.J. is a Cancer Research UK Principal Research Fellow. The work
`was funded by Wellcome Trust grant H5RCZRO (S.F.) and by Cancer
`Research UK program grant C375/A3964 (E.Y.J.).
`
`Received: May 24, 2006
`Revised: January 10, 2007
`Accepted: February 15, 2007
`Published: April 17, 2007
`
`REFERENCES
`
`Barton, V.A., Hudson, K.R., and Heath, J.K. (1999). Identification of
`three distinct receptor binding sites of murine interleukin-11. J. Biol.
`Chem. 274, 5755–5761.
`
`Barton, V.A., Hall, M.A., Hudson, K.R., and Heath, J.K. (2000). Interleu-
`kin-11 signals through the formation of a hexameric receptor complex.
`J. Biol. Chem. 275, 36197–36203.
`
`Boulanger, M.J., Bankovich, A.J., Kortemme, T., Baker, D., and
`Garcia, K.C. (2003a). Convergent mechanisms for recognition of diver-
`gent cytokines by the shared signalling receptor gp130. Mol. Cell 12,
`577–589.
`
`Boulanger, M.J., Chow, D.C., Brevnova, E.E., and Garcia, K.C.
`(2003b). Hexameric structure and assembly of the interleukin-6/IL-6
`a-receptor/gp130 complex. Science 300, 2101–2104.
`
`Bravo, J., and Heath, J.K. (2000). Receptor recognition by gp130 cyto-
`kines. EMBO J. 19, 2399–2411.
`
`Bravo, J., Staunton, D., Heath, J.K., and Jones, E.Y. (1998). Crystal
`structure of a cytokine-binding region of gp130. EMBO J. 17, 1665–
`1674.
`
`Chow, D., He, X., Snow, A.L., Rose-John, S., and Garcia, K.C. (2001).
`Structure of an extracellular gp130 cytokine receptor signalling com-
`plex. Science 291, 2150–2155.
`
`Cosman, D. (1993). The hematopoietin receptor superfamily. Cytokine
`5, 95–106.
`
`Curtis, D.J., Hilton, D.J., Roberts, B., Murray, L., Nicola, N., and
`Begley, C.G. (1997). Recombinant soluble interleukin-11 (IL-11) recep-
`tor a-chain can act as an IL-11 antagonist. Blood 90, 4403–4412.
`
`Dahmen, H., Horsten, U., Kuster, A., Jacques, Y., Minvielle, S., Kerr,
`I.M., Ciliberto, G., Paonessa, G., Heinrich, P.C., and Muller-Newen,
`G. (1998). Activation of the signal transducer gp130 by interleukin-11
`and interleukin-6 is mediated by similar molecular interactions. Bio-
`chem. J. 331, 695–702.
`
`DeLano, W.L. (2002). The PyMOL Molecular Graphics System (http://
`www.pymol.org).
`
`Esnouf, R.M. (1997). An extensively modified version of MolScript that
`includes greatly enhanced coloring capabilities. J. Mol. Graph. Model.
`15, 112–133.
`
`Harauz, G., and van Heel, M. (1986). Exact filters for general geometry
`three dimensional reconstruction. Optik 73, 146–156.
`
`He, W., Gong, K., Smith, D.K., and Ip, N.Y. (2005). The N-terminal
`cytokine binding domain of LIFR is required for CNTF binding and
`signalling. FEBS Lett. 579, 4317–4323.
`
`Hibi, M., Murakami, M., Saito, M., Hirano, T., Taga, T., and Kishimoto,
`T. (1990). Molecular cloning a