Conformational changes in the AAA ATPase p97–p47 adaptor complex
2006; Springer Nature; Volume: 25; Issue: 9 Linguagem: Inglês
10.1038/sj.emboj.7601055
ISSN1460-2075
AutoresFabienne Beuron, Ingrid Dreveny, Xuemei Yuan, Valerie E. Pye, Ciaran Mckeown, Louise C. Briggs, Matthew J. Cliff, Yayoi Kaneko, Russell Wallis, Rivka L. Isaacson, John E. Ladbury, Stephen Matthews, Hisao Kondo, Xiaodong Zhang, Paul S. Freemont,
Tópico(s)RNA regulation and disease
ResumoArticle6 April 2006free access Conformational changes in the AAA ATPase p97–p47 adaptor complex Fabienne Beuron Fabienne Beuron Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Ingrid Dreveny Ingrid Dreveny Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Xuemei Yuan Xuemei Yuan Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Valerie E Pye Valerie E Pye Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Ciaran Mckeown Ciaran Mckeown Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Louise C Briggs Louise C Briggs Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Matthew J Cliff Matthew J Cliff Department of Biochemistry and Molecular Biology, University College London, London, UK Search for more papers by this author Yayoi Kaneko Yayoi Kaneko Cambridge Institute for Medical Research, University of Cambridge, Cambridge, UK PRESTO and SORST, Japan Science and Technology Corporation, Japan Search for more papers by this author Russell Wallis Russell Wallis Department of Biochemistry, University of Oxford, Oxford, UK Department of Infection, Immunity, and Inflammation, Medical Research Council Immunochemistry Unit, University of Leicester, Leicester, UK Search for more papers by this author Rivka L Isaacson Rivka L Isaacson Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author John E Ladbury John E Ladbury Department of Biochemistry and Molecular Biology, University College London, London, UK Search for more papers by this author Steve J Matthews Steve J Matthews Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Hisao Kondo Hisao Kondo Cambridge Institute for Medical Research, University of Cambridge, Cambridge, UK PRESTO and SORST, Japan Science and Technology Corporation, Japan Search for more papers by this author Xiaodong Zhang Xiaodong Zhang Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Paul S Freemont Corresponding Author Paul S Freemont Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Fabienne Beuron Fabienne Beuron Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Ingrid Dreveny Ingrid Dreveny Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Xuemei Yuan Xuemei Yuan Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Valerie E Pye Valerie E Pye Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Ciaran Mckeown Ciaran Mckeown Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Louise C Briggs Louise C Briggs Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Matthew J Cliff Matthew J Cliff Department of Biochemistry and Molecular Biology, University College London, London, UK Search for more papers by this author Yayoi Kaneko Yayoi Kaneko Cambridge Institute for Medical Research, University of Cambridge, Cambridge, UK PRESTO and SORST, Japan Science and Technology Corporation, Japan Search for more papers by this author Russell Wallis Russell Wallis Department of Biochemistry, University of Oxford, Oxford, UK Department of Infection, Immunity, and Inflammation, Medical Research Council Immunochemistry Unit, University of Leicester, Leicester, UK Search for more papers by this author Rivka L Isaacson Rivka L Isaacson Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author John E Ladbury John E Ladbury Department of Biochemistry and Molecular Biology, University College London, London, UK Search for more papers by this author Steve J Matthews Steve J Matthews Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Hisao Kondo Hisao Kondo Cambridge Institute for Medical Research, University of Cambridge, Cambridge, UK PRESTO and SORST, Japan Science and Technology Corporation, Japan Search for more papers by this author Xiaodong Zhang Xiaodong Zhang Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Paul S Freemont Corresponding Author Paul S Freemont Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK Search for more papers by this author Author Information Fabienne Beuron1, Ingrid Dreveny1, Xuemei Yuan1, Valerie E Pye1, Ciaran Mckeown1, Louise C Briggs1, Matthew J Cliff2, Yayoi Kaneko3,4, Russell Wallis5,6, Rivka L Isaacson1, John E Ladbury2, Steve J Matthews1, Hisao Kondo3,4, Xiaodong Zhang1 and Paul S Freemont 1 1Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington, London, UK 2Department of Biochemistry and Molecular Biology, University College London, London, UK 3Cambridge Institute for Medical Research, University of Cambridge, Cambridge, UK 4PRESTO and SORST, Japan Science and Technology Corporation, Japan 5Department of Biochemistry, University of Oxford, Oxford, UK 6Department of Infection, Immunity, and Inflammation, Medical Research Council Immunochemistry Unit, University of Leicester, Leicester, UK *Corresponding author. Centre for Structural Biology, Division of Molecular Biosciences, Imperial College London, South Kensington Campus, Biochemistry Building, South Kensington, London SW7 2AZ, UK. Tel.: +44 20 7594 5327; Fax: +44 20 7594 3057; E-mail: [email protected] The EMBO Journal (2006)25:1967-1976https://doi.org/10.1038/sj.emboj.7601055 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info The AAA+ATPase p97/VCP, helped by adaptor proteins, exerts its essential role in cellular events such as endoplasmic reticulum-associated protein degradation or the reassembly of Golgi, ER and the nuclear envelope after mitosis. Here, we report the three-dimensional cryo-electron microscopy structures at ∼20 Å resolution in two nucleotide states of the endogenous hexameric p97 in complex with a recombinant p47 trimer, one of the major p97 adaptor proteins involved in membrane fusion. Depending on the nucleotide state, we observe the p47 trimer to be in two distinct arrangements on top of the p97 hexamer. By combining the EM data with NMR and other biophysical measurements, we propose a model of ATP-dependent p97(N) domain motions that lead to a rearrangement of p47 domains, which could result in the disassembly of target protein complexes. Introduction p97 belongs to the diverse AAA+ (ATPase Associated with various cellular Activities) family of proteins involved in a variety of apparently unrelated cellular processes. AAA+ proteins are characterised by the presence of one or two conserved AAA domains (∼230–250 residues, D1 and D2 domains), which contain the Walker A and B motifs. AAA+ proteins are generally oligomeric (typically hexamers) and are believed to have common functions in the unfolding or disassembly of macromolecules or multiprotein complexes (Ogura and Wilkinson, 2001). Several p97 cofactors have been identified that link p97 with a variety of cellular activities (reviewed in Dreveny et al, 2004b). For example, p97 binds to the Ufd1–Npl4 adaptor complex mediating events during endoplasmic reticulum (ER)-associated degradation (ERAD, Bays et al, 2001; Ye et al, 2001; Braun et al, 2002; Jarosch et al, 2002), nuclear envelope formation (Hetzer et al, 2001) and spindle disassembly after mitosis (Cao et al, 2003). p97 also functions in membrane fusion utilising the p47 adaptor protein to mediate the reformation of Golgi membranes and reassembly of the nuclear envelope at the end of mitosis (reviewed in Meyer, 2005; Uchiyama and Kondo, 2005). The p97–p47 complex is proposed to disassemble t-t SNARE (target-target soluble N-ethylmaleimide-sensitive factor attachment protein receptor) complexes (Acharya et al, 1995), in a role reminiscient of that of the homologous AAA+ ATPase NSF (N-ethylmaleimide-sensitive factor) and its adaptor α-SNAP (soluble NSF attachment protein) in disassembling v-t SNAREs (vesicle-target-SNARE). During Golgi regrowth, it is proposed that p47 mediates the binding of p97 to the t-SNARE syntaxin 5 post fusion (Kondo et al, 1997; Rabouille et al, 1998). An additional protein, VCIP135 (valosin-containing protein, VCP/p97–p47 complex-interacting protein, p135), binds to the p97–p47-syntaxin 5 complex and dissociates it using ATP hydrolysis events in p97, preparing syntaxin 5 for further rounds of membrane fusion (Uchiyama et al, 2002). Recently, NSF–α-SNAP complexes and the p97–p47–VCIP135 complexes were shown to act sequentially in cell cycle-dependent reformation of the ER network (Kano et al, 2005). These events also involve the t-SNARE syntaxin 18 (Kano et al, 2005). Interestingly, p47 was shown to bind to ubiquitin via its N-terminal UBA domain (Meyer et al, 2002; Yuan et al, 2004b) and this binding is essential for efficient p97–p47 mediated reassembly of Golgi cisternae (Meyer et al, 2002). In yeast, p47 orthologues interact with ubiquitylated proteins in vivo and are described to be involved in ubiquitin mediated proteolytic events, probably delivering substrates to the 26S proteasome (Hartmann-Petersen et al, 2004; Schuberth et al, 2004). p97 comprises three domains, the N-terminal cofactor binding domain and two AAA ATPase domains (D1 and D2). Three-dimensional (3D) reconstructions of p97 in different nucleotide states obtained by cryo-electron microscopy (cryo-EM) and small angle X-ray scattering have shown that the N-domains are mostly flexible and together with the D1 and D2 domains can undergo significant conformational change during ATP binding and hydrolysis (Zhang et al, 2000; Rouiller et al, 2002; Beuron et al, 2003; Davies et al, 2005). Crystal structures of p97 at low resolution (3.5–4.7 Å) in different nucleotide states reveal a consistent picture of a hexameric arrangement of subunits with N-domains interacting with D1 in the D1 hexamer plane and with only minor conformational changes (DeLaBarre and Brunger, 2003, 2005; Huyton et al, 2003). The adaptor protein p47 contains three domains connected by long flexible regions. The solution structures for all three p47 domains (UBA, SEP and UBX) and dynamics of p47 fragments have been reported (Yuan et al, 2001, 2004b; Soukenik et al, 2004). A 2.9 Å crystal structure of p97(N–D1) hexamer bound to p47(UBX) domain revealed that p47 interacts with the p97(N) domain via a conserved loop within the UBX domain (S3/S4) that inserts into a hydrophobic pocket between the two p97(N) subdomains (Dreveny et al, 2004a). Despite this body of data, no detailed information exists regarding the structural arrangements within the full-length p97–p47 complex, other than low-resolution 2D EM projections (Rouiller et al, 2000). Six peripheral densities were attributed to p47, which is not consistent with biochemical studies suggesting a p47 trimer binding to a p97 hexamer (Kondo et al, 1997). One of the major cellular roles for p97 is to provide mechanical force via ATPase activity to disassemble or unfold target proteins or protein complexes, a force which is mostly thought to be mediated through various adaptor proteins in vivo, although direct substrate binding has also been reported (Ye et al, 2003). We have combined cryo-EM, high-field NMR and other biophysical methods to capture and analyse the nucleotide-dependent conformational states of the physiological p97–p47 complex. A comparison between these states points to possible conformational changes that are necessary for p97 action on target substrates specific to the p97–p47 pathway and thus provide new insights into p97 function. Interestingly, we found that the p97–p47 complex resembles the NSF 20S complex structurally despite the differences between these ATPases and corresponding adaptor proteins (Brunger and DeLaBarre, 2003; Furst et al, 2003). Results and discussion p47 trimerisation and stoichiometry of the p97–p47 complex In this study, we have carried out experiments, which allowed a qualitative appreciation of the oligomeric status of freshly purified full-length p47 using short timescale NMR experiments. Our earlier NMR studies on the p47(SEP–UBX) fragment revealed a weak propensity for oligomerisation via the SEP domain (Yuan et al, 2004b). Detailed analysis of 15N relaxation data indicated that the likely oligomeric state is either a dimer or trimer (Yuan et al, 2004b). SEP domain NMR signals are not observable in 1H-15N HSQC spectra of full-length p47, which can be attributed to their fast relaxation properties, revealed by 1D relaxation measurements and the fact that they become visible upon perdeuteration (Supplementary Figure 1A). The extent of broadening is similar to our earlier studies (Yuan et al, 2004b) and therefore consistent with the involvement of the SEP region in formation of either a dimeric or trimeric p47 species. Support for a trimeric p47 has been provided by gel-filtration and light scattering data (Kondo et al, 1997) as well as crosslinking experiments (Yuan et al, 2004b). Furthermore, analytical ultracentrifugation data presented in this study provides additional evidence for full-length p47 existing as a trimer (Supplementary Figure 1B). As the individual UBA, SEP and UBX domains possess relaxation properties consistent with monomeric species (overall correlation times ∼4–6 ns) (Yuan et al, 2001, 2004b; Soukenik et al, 2004), we were able to investigate elements of the trimerisation interface by comparing chemical shifts between spectra of trimeric p47 and the separate, monomeric domains (Figure 1A) (Soukenik et al, 2004; Yuan et al, 2004b). Several well-resolved shifts occur in the SEP domain and delineate a contiguous patch at one edge of the β-sheet that could form part of the trimerisation boundary (Figure 1B). No perturbations are observed in the UBA or UBX domains and spectral overlap precludes the detailed comparison of the linker regions. These data combined with rigid-body docking using the program SymmDock (Schneidman-Duhovny et al, 2005) provide the means to construct a working model for the trimeric arrangement of SEP (Figure 1B). Interestingly, the N- and C-termini extend in opposite directions but are proximal to the trimer interface, this supports the premise that neighbouring linker regions are also important determinants for trimer formation. Figure 1.(A) Overlay of 1H-15N HSQC NMR spectra for 2H, 15N-labelled p47 (black) and a mixture of the individual 15N-labelled UBA, SEP and UBX domains from p47 (red). Key shifted resonances are labelled. (B) Ribbon representation of the p47(SEP) domain with chemical shift perturbations illustrating the likely trimer interface mapped in red (left) and a model for the trimer obtained by rigid-body docking (right) (top and side views). (C) Isothermal titration calorimetry of full-length p97–p47 and (D) p97–p47(SEP–UBX) domains showing a 1:0.5 and 1:1 stoichiometry respectively. The 1:0.5 and 1:1 ratios are indicated by a red and black dotted line in both plots, respectively. Download figure Download PowerPoint Much controversy exists regarding the binding stoichiometry of p47 to p97 with data supporting both 3 and 6 p47 molecules per p97 hexamer (Kondo et al, 1997; Rouiller et al, 2000). In resolving this ongoing debate, we took extra care in producing, assessing and testing the activity of our p97–p47 complexes (Supplementary Figure 2A and B). To establish the stoichiometry of our p97–p47 complexes, we used isothermal titration calorimetry (Figures 1C and D). A ratio of 0.5 p47 monomer to 1 p97 protomer was observed with a Kd of 0.5 μM confirming that three p47 bind to a hexameric p97 (Figure 1C). In contrast, the stoichiometry of a shorter p47 fragment, comprising the SEP–UBX domains, was found to be of 1 p47 fragment bound to 1 p97 protomer (Figure 1D). Furthermore, en face 2D EM projection averages of the respective complexes show that p97–p47(SEP–UBX) has significant additional six-fold density at the periphery of the p97 hexamer whereas the full-length complex contains central density and no additional peripheral density (data not shown). Scanning transmission electron microscopy mass measurements also confirm that the particles are no larger than a p97 hexamer in complex with 3 p47 (data not shown) consistent with the calorimetric data and analysis of the complex by native gel electrophoresis (Figure 1C and Supplementary Figure 2A). Taken together our data firmly establish that three p47 molecules bind to a single p97 hexamer and indicate that p47 trimerisation mediated by the SEP domain and parts of the N-terminal linker are important determinants for this stoichiometry. Cryo-EM reconstructions of the p97–p47 complex We have obtained cryomicrographs of p97–p47 complex in the presence of ADP or AMPPNP (Figure 2 and Supplementary Table I). Manual picking of single particles was straightforward because of the good contrast of the particles (Figure 2A), and the resulting class averages (Figure 2C) presented a range of different views that were used for the angular reconstitution. 3D reconstructions were computed independently from separate initial models. The distribution of Euler angles for the AMPPNP and ADP data sets are shown in Figure 2B. Class averages obtained in the final iteration of the multireference alignments (MRA)/multivariate statistical analysis (MSA) and angle assignment correspond well to matching projections of the 3D model (Figure 2C). The parallel-striated appearance of p97 side views (Beuron et al, 2003; Figure 5A) is replaced in all data sets by a flat appearance of the bottom striation and additional density found at the top of the upper striation with small satellite densities connecting to this additional density (Figure 2C, arrows). Figure 2.Cryo-EM and class averages of the p97–p47 complex. (A) Field of a cryomicrograph of p97–p47 incubated with AMPPNP showing a range of different views with protein density shown in black. (B) Distribution of viewing angles of the class sums used for calculating the 3D models of the p97–p47 complex in the AMPPNP and ADP conformations. The γ-angles are clustered between (−60°, +60°) because of the use of the three-fold symmetry. (C) For each nucleotide condition, the top row contains representative p97–p47 class averages (protein is shown in white) and the bottom row the corresponding reprojection of the 3D map in the direction of the orientation assigned to the average. For the complex in the presence of AMPPNP, three classes are circled that represent typical top, side and tilted views (left to right). The arrows point to small satellite densities that are connected to the rest of the particle. Download figure Download PowerPoint The low-pass filtered, three-fold symmetrised 3D reconstructions are shown as surface representations in Figure 3 and Supplementary movies. The complex adopts a typical barrel-shaped arrangement formed by the stacking of D1 and D2 hexameric rings. In both maps, the top and bottom rings have an overall diameter of ∼110 Å and a combined height of 95 Å, which corresponds well to the diameter of p97 and the height of stacked D1 and D2 rings in crystal structures (DeLaBarre and Brunger, 2003, 2005; Huyton et al, 2003). Significant common density is found above the D1 plane at the top of the particle and consists of a large flat 'cap' or 'plug' sitting on the three-fold symmetry axis, from which other densities project. The height of the maps increases to ∼135 Å when including the 'plug' and extends to ∼150 Å when enclosing the satellite densities ('antennae') at the top of the AMPPNP map. The presence and connectivity of these antennae to the 'plug' in the AMPPNP map is clearly visible on the class averages as well as in the corresponding reprojection (Figure 2C) and density sections through the map (Supplementary Figure 3). The same is observed for the 3D reconstruction of p97–p47 in the absence of nucleotide (data not shown). Figure 3.3D reconstructions of p97–p47 single particles imaged by cryo-EM. Surface representations of the three reconstructions low-pass filtered using a 20 Å Gaussian mask are shown (A) in the presence of AMPPNP and (B) in the presence of ADP. Three views are shown: bottom (looking from D2 to D1), side (perpendicular to the symmetry axis) and top (looking from D1 to D2). The position of the N, D1 and D2 domains of p97 are indicated. The 'propeller' and 'antennae' emerging from the central plug are indicated and represent densities attributed to p47. The mesh superimposed in Figure 3A (side view) shows the connection between the 'antennae' and central 'plug' in the non band-pass filtered AMPPNP surface map. The scale bar represents 50 Å. Download figure Download PowerPoint The p97–p47 reconstructions also display a number of unique features. In the presence of AMPPNP, the D2 ring has no central pore (Figure 3A, bottom view) and small densities protrude outwards from the outer tip of the D1 ring (Figure 3A, side view and vertical density sections in Supplementary Figure 3). Connected 'antennae' densities emerge from three points on the central 'plug' and are contiguous giving rise to a ring-like appearance. In the presence of ADP, the D2 ring has a pore of ∼20 Å in diameter located at the end of a funnel shaped conduit (Figure 3B, bottom view). The particle is wider at the D1–D2 interface and three 'arms' emerge from the central 'plug' forming a three-fold symmetric 'propeller-like' structure that links back onto the top of three D1 domains. Similarities between the cryo-EM reconstructions of the p97–p47 and NSF–-SNAP–SNARE complexes Recently, the 3D structure of NSF (D1 ATP hydrolysis mutant) in complex with α-SNAP–SNARE (20S complex) was determined by cryo-EM and single-particle methods in the presence of a mixture of ADP and ATP (EM3D accession code 1059; Furst et al, 2003). NSF is a close homologue of p97 and dissociates v-t SNARE complexes, thereby reactivating them to mediate further rounds of membrane fusion. Interestingly, the 3D reconstruction of the NSF 20S complex (filtered to 20 Å resolution) is similar to p97–p47 both in size and overall shape (Figure 4). The central 'plugs' are comparable as are the positions of protrusions emanating form D1 attributed to N-domains (see below). The volume of the NSF 20S reconstruction accounts mainly for NSF due to the six-fold averaging used in the reconstruction of the complex (Furst et al, 2003). Owing to the conformational heterogeneity, the NSF 20S reconstruction does not contain all the density for the α-SNAP–SNARE proteins with the 'plug' density representing only a fraction of the whole α-SNAP–SNARE complex (Furst et al, 2003). These similarities are interesting given the absence of any sequence and structural homology between p47 and α-SNAP, the corresponding adaptor for NSF. Furthermore, the residues implicated in α-SNAP binding by NSF(N) are located on the opposite side of the p47 binding site on p97(N) (Dreveny et al, 2004a). Despite the differences including ATPase domain arrangements between NSF and p97 (Brunger and DeLaBarre, 2003; Furst et al, 2003), both in complex with their respective adaptor protein (adaptor protein and parts of SNARE complex for NSF) show an overall similarity at this resolution. Especially, the similar location of the adaptor proteins on top of the hexamers is intriguing (Figure 4B and C), although further higher resolution structures would be necessary to substantiate any possible mechanistic analogies. Figure 4.Superposition of the EM reconstructions of p97–p47 and the NSF 20S complex. The NSF 20S complex (Furst et al, 2003) was band pass filtered to a resolution of 20 Å. (A) Top view of the superposition of the NSF 20S complex (green) and the p97–p47 AMPPNP map (blue mesh; lacking the antennae). (B) Side view of the superimposed maps. (C) Vertical central section of superimposed maps. The N, D1 and D2 domains of p97 are indicated. Both maps are characterised by a central plug on top of the D1 hexameric ring, which is attributed to α-SNAP–SNARE densities for the NSF 20S complex and corresponds to the SEP domains of the p47 adaptor in the p97–p47 complex. Download figure Download PowerPoint Distinct nucleotide-dependent conformations of the p97–p47 complex When comparing our previous p97 maps (Beuron et al, 2003) with the new p97–p47 maps obtained under identical nucleotide conditions, we find good overall agreement for p97 (Figure 5A). This comparison allows us to identify density for p47 as being on top of the D1 ring ('plug' and 'antennae'). Apart from these clear extra densities, there are a number of other distinct differences. In the AMPPNP conformations (Figure 5A, top), the D1 and D2 rings have similar diameters with D1 stacking on top of D2. In p97–p47 the D2 ring is closed, whereas it is open in p97 alone. In the ADP conformations (Figure 5A, bottom), the D1 and D2 rings also show good agreement although in contrast to AMPPNP, the D2 ring is open in p97–p47 but closed in p97 alone. These differences in p97 clearly reflect p47 binding and illustrate the limitations of interpreting p97 nucleotide-dependent conformational changes in the absence of bound adaptor proteins. Superimposed surface rendered representations for the p97–p47 AMPPNP (blue) and ADP (orange; Figure 5B) as well as vertical sections of the same comparison (Figure 5C) show that the maps differ substantially in both shape and protruding densities above the D1 plane. Intriguingly, thin densities emerge from the same point in the D1 ring in both maps, though projecting in different directions (arrows in Figure 5D). In the ADP state, p97 adopts a barrel shape with the widest point between the D1 and D2 domains, although the D1 ring in the ADP state has a slightly wider diameter (∼120 Å). By comparison, the D1 and D2 domains in the AMPPNP state have similar diameters with the D1 domain stacking directly on top of D2 (Figure 5C). This results in an apparent counter-clockwise rotation in D2 of ∼10° (viewing down the trimer axis from top) in the ADP map compared to the AMPPNP map (Figure 5E). Proposed p47 domain arrangements in the p97–p47 complex In order to interpret further the domain arrangements in the p97–p47 reconstructions, we performed a number of labelling studies. Electron micrographs of negatively stained nanogold labelled p47 N-termini (preceding the UBA) clearly show the gold marker at the centre of the hexamer in both the AMPPNP and nucleotide-free samples (Supplementary Figure 4A). In contrast, we observe a different labelling pattern in the presence of ADP (Supplementary Figure 4A, middle) reflecting a dispersed distribution of the observed gold particles, consistent with flexible or untethered p47 N-termini (Supplementary Figure 1C). In agreement with the location of p47 domains above the p97 hexamer, we observe both syntaxin 5 and ubiquitin (known p97 target proteins) as binding to the top of the p97–p47 complex (Supplementary Figure 4B and C). Using the mapping information and comparisons with previous p97 reconstructions, we have provisionally assigned p47 domains within our reconstructions. These provisional assignments were further guided by the following key pieces of information: (1) existing high-resolution structures of p97, p47 and the p97(N–D1)–p47(UBX) complex (DeLaBarre and Brunger, 2003, 2005; Huyton et al, 2003; Dreveny et al, 2004a; Yuan et al, 2004b); (2) three p47 molecules bind to one p97 hexamer; (3) p47 binds to p97(N) domains via two binding sites, the p47(UBX) domain (Dreveny et al, 2004a) and the linker region between SEP and UBX (residues ∼246–273) (Uchiyama et al, 2002; Bruderer et al, 2004), which is confirmed by chemical shift perturbation data measured from CRIPT-TROSY NMR spectra of the p97–p47 complex (Supplementary Figure 1D); (4) p47 has a propensity to form trimers in solution mediated by the p47(SEP) domain and flanking regions, for which we propose a possible model (Figure 1B). A number of chemical shifts for interfacial resid
Referência(s)