Portal de Periódicos da CAPES

Structural insights into the regulation of cohesion establishment by Wpl1

2013; Springer Nature; Volume: 32; Issue: 5 Linguagem: Inglês

10.1038/emboj.2013.16

1460-2075

Avradip Chatterjee, Silva Zakian, Xiaowen Hu, Martin R. Singleton,

Photosynthetic Processes and Mechanisms

Article8 February 2013Open Access Structural insights into the regulation of cohesion establishment by Wpl1 Avradip Chatterjee Avradip Chatterjee Macromolecular Structure and Function Laboratory, Cancer Research UK, London Research Institute, London, UK Search for more papers by this author Silva Zakian Silva Zakian Macromolecular Structure and Function Laboratory, Cancer Research UK, London Research Institute, London, UK Search for more papers by this author Xiao-Wen Hu Xiao-Wen Hu Macromolecular Structure and Function Laboratory, Cancer Research UK, London Research Institute, London, UK Search for more papers by this author Martin R Singleton Corresponding Author Martin R Singleton Macromolecular Structure and Function Laboratory, Cancer Research UK, London Research Institute, London, UK Search for more papers by this author Avradip Chatterjee Avradip Chatterjee Macromolecular Structure and Function Laboratory, Cancer Research UK, London Research Institute, London, UK Search for more papers by this author Silva Zakian Silva Zakian Macromolecular Structure and Function Laboratory, Cancer Research UK, London Research Institute, London, UK Search for more papers by this author Xiao-Wen Hu Xiao-Wen Hu Macromolecular Structure and Function Laboratory, Cancer Research UK, London Research Institute, London, UK Search for more papers by this author Martin R Singleton Corresponding Author Martin R Singleton Macromolecular Structure and Function Laboratory, Cancer Research UK, London Research Institute, London, UK Search for more papers by this author Author Information Avradip Chatterjee1, Silva Zakian1, Xiao-Wen Hu1 and Martin R Singleton 1 1Macromolecular Structure and Function Laboratory, Cancer Research UK, London Research Institute, London, UK *Corresponding author. Macromolecular Structure and Function Laboratory, Cancer Research UK, London Research Institute, 44 Lincoln's Inn Fields, London WC2A 3LY, UK. Tel.:+44 (0) 207 269 3148; Fax:+44 (0) 207 269 3258; E-mail: [email protected] The EMBO Journal (2013)32:677-687https://doi.org/10.1038/emboj.2013.16 There is a Have you seen? (March 2013) associated with this Article. PDFDownload PDF of article text and main figures. Peer ReviewDownload a summary of the editorial decision process including editorial decision letters, reviewer comments and author responses to feedback. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Correct segregation of duplicated chromosomes to daughter cells during mitosis requires the action of the cohesin complex. This tripartite ring-shaped molecule is involved in holding replicated sister chromatids together from S phase until anaphase onset. Establishment of stable cohesion involves acetylation of the Smc3 component of cohesin during replication by the Eco1 acetyltransferase. This has been proposed to antagonise the activity of another member of the cohesin complex, Wpl1. Here, we describe the X-ray structure of the conserved Wapl domain, and demonstrate that it binds the ATPase head of the Smc3 protein. We present data that suggest that Wpl1 may be involved in regulating the ATPase activity of cohesin, and that this may be subject to the acetylation state of Smc3. In addition, we present a structure of the Wapl domain bound to a functionally relevant segment of the Smc3 ATPase. Introduction Correct chromosome segregation in eukaryotes is ensured by physically linking duplicated sister chromatids during the S phase of the cell cycle and maintaining this connection until anaphase onset (Nasmyth, 2001). The linkage between sister chromatids is maintained by the ring-shaped cohesin complex (Haering et al, 2008), which is loaded onto chromosomes in late G1 phase and activated in a multi-step, replication-dependent manner (Skibbens et al, 1999; Tóth et al, 1999; Skibbens, 2009). The core of the cohesin complex consists of three proteins, Smc1, Smc3, and Scc1. Smc1 and Smc3 are members of the Structural Maintenance of Chromosomes family and contain a globular ABC ATPase domain connected to a dimer-forming hinge domain by long intramolecular coiled coils, which are proposed to entrap sister DNA strands (Hirano, 2002). Scc1 is a member of the kleisin family of proteins that bridges the ATPase heads of the Smc1–Smc3 ring (Schleiffer et al, 2003). Proteolytic cleavage of this subunit by separase at anaphase onset allows release and poleward movement of the chromosomes (Uhlmann et al, 2000). In budding yeast, all chromosome-bound cohesin is removed at this point. In vertebrates, by contrast, the bulk of cohesin is removed from chromosomes during prophase in a proteolysis-independent manner (Losada et al, 1998; Waizenegger et al, 2000; Sumara et al, 2002), while centromeric cohesin is protected until anaphase by the protein Sgo1, at which point it is proteolytically released (Kitajima et al, 2004; Salic et al, 2004; Tang et al, 2004; McGuinness et al, 2005). Prior to DNA replication, cohesin binds chromatin in a reversible manner that is not capable of maintaining cohesion and is thought to be held in this state by a complex containing the proteins Wpl1 and Pds5 (Tanaka et al, 2001; Gandhi et al, 2006; Kueng et al, 2006; Ben-Shahar et al, 2008; Rowland et al, 2009; Sutani et al, 2009), whose activity is antagonised by Eco1-mediated acetylation of the Smc3 protein during S phase (Ivanov et al, 2002; Bellows et al, 2003; Lengronne et al, 2006; Ben-Shahar et al, 2008; Unal et al, 2008; Zhang et al, 2008). In vertebrates, the stably chromatin-bound form of cohesin may be maintained by the binding of the protein sororin to Pds5 (Schmitz et al, 2007; Nishiyama et al, 2010) though it is unclear if this system is conserved in yeast. The Wapl gene was initially identified in Drosophila as a controller of heterochromatin structure and cohesion (Vernì et al, 2000). Subsequent studies in vertebrates have shown that the protein is involved in the non-proteolytic removal of cohesin during the prophase pathway as well as interphase (Gandhi et al, 2006; Kueng et al, 2006). Wapl shows homology to the budding yeast Rad61 protein (Game et al, 2003; Warren et al, 2004) which, when mutated, allows bypass of lethality associated with Eco1 inactivation (Ben-Shahar et al, 2008; Unal et al, 2008). Collectively, these results suggest that Wapl is involved in the destabilisation of chromatin-bound cohesin in an acetylation-dependent manner, and that budding yeast Rad61 (now also known as Wpl1) mediates a similar process. Currently, the molecular mechanisms of this complex pathway are not fully understood. In this paper, we present results of structural and biochemical studies of the Wpl1 protein. We describe the structure of the conserved domain of the Wpl1 protein, and present a biochemical analysis of its interactions with other members of the cohesin complex, which provides new insights into the process of cohesin establishment. Results Structure of the Wapl domain All Wapl orthologues contain a C-terminal conserved domain (Figure 1A; Supplementary Figure 1a), which defines the family. This domain is ∼45 kDa in size, and was predicted to be predominantly α-helical (Kueng et al, 2006). The non-conserved N-terminal of the protein is highly divergent in both length and composition, and is responsible for mediating interactions with Pds5 (Shintomi and Hirano, 2009). Secondary structure predictions of the protein suggest that this divergent N-terminal is likely to be intrinsically disordered, so we decided to further investigate the structure of the conserved C-terminal Wapl domain. We identified a truncation of the Wpl1 protein from Ashbya gossypii (transcript AAR187C) comprising the Wapl domain that yielded high-quality crystals. The A. gossypii Wpl1 (AgWpl1) protein is fractionally shorter than budding yeast equivalent, but shows a high degree of similarity and 24% identity over the entire sequence (Supplementary Figure 1b). The final construct crystallised contained residues 184–561 of AgWpl1, corresponding to residues 271–635 of budding yeast Wpl1 (ScWpl1). High quality diffraction data were obtained to a resolution of 2.1 Å, and the structure was determined using multiple wavelength anomalous diffraction of a selenomethionine-substituted derivative. Statistics relating to the X-ray analysis are presented in Supplementary Table 1. The majority of the peptide chain is clearly defined in the electron density, with only short breaks between residues 544–548 (chain A) and 189–194, 214–221, 311–312 (chain B). The overall model refinement statistics are appropriate to the resolution of the experimental data. Figure 1.Structure of the Wapl domain. (A) Domain structure of four Wapl orthologues showing the location of the conserved Wapl domain (blue). The length of the protein is indicated. Pale blue bars indicate the location of the Pds5-interacting FGF repeats in the human protein. (B) Ribbon diagram of AgWpl1184–561, coloured blue (N-terminal) to red (C-terminal). The protein is subdivided into three indicated domains with the HEAT repeats comprising domain 2. (C) Secondary structural elements of the A. gossypii Wapl domain. Helices are numbered, and position of the canonical HEAT repeats highlighted. The position of the helical insertion in repeat HEAT3 is indicated. (D) Structure-based superimposition of the central Wpl1 HEAT repeats onto the PP2A repeats (pdb i.d. 1b3u). Wpl1 is shown in blue and PP2A in yellow. (E) Electrostatic surface of the Wapl domain showing positive potential in blue and negative in red. (F) Surface conservation properties of the Wapl domain. Red indicates areas of high conservation, white of low. Download figure Download PowerPoint The Wapl domain forms an elongated, peanut-shaped structure of ∼90 × 35 Å (Figure 1B). The fold is defined by 20 consecutive α-helices and may be divided into 3 subdomains (Figure 1C). Domain 1 (residues 189–220) forms an anti-parallel helical 'cap' to the structure, and is followed by six HEAT repeats that form the central domain of the protein. The C-terminal lobe (domain 3, residues 420–561) is composed of a flattened 4-helix bundle capped by two anti-parallel helices at the extreme C-terminal. Three-dimensional homology searches performed using the DALI server (Holm and Sander, 1993) readily identify the central HEAT core of the protein and the Wapl repeats may be aligned to those of human PP2A (Groves et al, 1999) with a Cα r.m.s. deviation of 3.9 Å over 72 residues (Figure 1D). No obvious structural homologues to domains 1 and 3 were identified. The topology of the helices in these domains is the same as in the HEAT repeat, but the three-dimensional structure has substantially diverged from the basic repeat architecture. An unusual feature of the structure is an insertion of a long helix and connecting loop (residues 301–329) into repeat HEAT3 that projects away from the main body of the protein, somewhat like an opposable thumb against a palm formed by domain 1 (Figure 1C). Analysis of the electrostatic surface of the protein (Figure 1E) showed that the N-terminal domain is highly charged with prominent acidic and basic patches, while the C-terminal domain is generally acidic or uncharged. Mapping sequence conservation onto the molecular surface identifies regions of particular interest, two corresponding to the charged patches in the N-lobe of the protein, and a conserved hydrophobic patch on the inner face of the C-terminal lobe (Figure 1F). Based on our structural analyses and multiple sequence alignments of the Wapl domain, we have defined five sequence motifs that constitute these three surface features (Figure 2). Motifs 1 and 2 form the N-terminal acidic and basic regions, while motifs 3, 4, and 5 form the C-terminal hydrophobic patch. Figure 2.Conserved motifs in the Wapl domain. The positions of the five sequence- and structure-based motifs are mapped onto the surface representation of the protein, and multiple sequence alignments of the corresponding residues in six fungal Wpl1 orthologues indicated. The residues in the A. gossypii sequence are indicated against electrostatic representations of the protein surface. Motif 2 is relatively weakly conserved at a sequence level, but characterised by a basic-residue rich tract that falls in the α8 helical insertion and connecting loop. The sequences shown are from V. polyspora, Z. rouxii, S. cerevisiae, L. thermotolerans, A. gossypii, and K. lactis. Download figure Download PowerPoint Screening for interactions between the Wapl domain and the cohesin complex It has previously been shown that Wapl directly interacts with Pds5 in both yeast and vertebrates via the N-terminal domain (Gandhi et al, 2006; Kueng et al, 2006; Shintomi and Hirano, 2009; Sutani et al, 2009), but the function of the conserved C-terminal domain was unknown. We therefore decided to analyse the interactions between this domain and other proteins in the cohesin complex in more detail. To identify putative interacting partners, we performed a screen against a membrane peptide array comprising the AgSmc1 and AgSmc3 ATPase heads and AgScc1. The arrays consisted of 21-mer peptides covering the proteins at a 3-residue resolution. These were probed with the isolated Wapl domain. We obtained strong hits against three sequences in AgSmc3, labelled WIS1, WIS2, and WIS3 (Wapl Interacting Sequence; Figures 3A and B and Supplementary Figure 2a), as well as some potential hits, primarily on the N-terminal of AgScc1 (Supplementary Figure 2b). Interestingly, it is the N-terminal of Scc1 that interacts with Smc3, raising the possibility that the Wapl domain may interact with both proteins simultaneously, independently of Pds5 (Shintomi and Hirano, 2009). However, these potential Scc1 interactions have not been further tested in the current study. The AgSmc3 target peptides identified all mapped to regions predicted to be solvent exposed based on analysis of the homologous ScSmc1 crystal structure (Haering et al, 2004; Figure 3C) and are all highly conserved between Smc3 orthologues. WIS1 lies across two strands and the interconnecting loop at the top of the external β-sheet of the ATPase domain. Intriguingly, the connecting loop in the centre of the sequence also contains lysines K107 and K108 that correspond to the Eco1-targetted lysines K112 and K113 in ScSmc3. WIS2 constitutes two short α-helices and connecting loops that lie over the ATP-binding sites. These connect the so-called Q-loop that interacts with the bound nucleotide to the base of the coiled-coil domain. The WIS3 site is located at the base of the outer helix that forms the coiled-coil domain and extends through a short loop (disordered in the crystal structure) to the adjacent β-strand. There is some uncertainty in mapping the exact location of this site to the Smc1 structure, as the sequence of this region differs significantly between Smc3 and Smc1. Figure 3.Screening of potential Wapl domain interactors. (A) Peptide array screen of potential Wapl domain binding partners in the AgSmc3/AgSmc1 ATPase heads and AgScc1. The likely positive hits are boxed and numbered. (B) Schematic of Smc3 primary structure showing the position of the target peptides. The coiled-coil domain is omitted for clarity. (C) Structure of the ScSmc1 protein (pdb i.d. 1w1w) with the sequences homologous to WIS1–WIS3 indicated in the same colour scheme as in (A) and (B) mapped onto the ribbon diagram of one monomer. Download figure Download PowerPoint Binding site validation and analysis To validate the hits obtained in the peptide array, and obtain quantitative data on the binding, we analysed the interactions between the peptides and the Wapl domain by fluorescence polarisation (FP) assays. We used peptides corresponding to the A. gossypii sequences carrying an amino-terminal fluorescein group and purified recombinant Wapl domain. The binding curves and derived constants are shown in Figure 4. All three peptides bound with high micromolar affinities. We next analysed a variant of the WIS1 in which lysines K107 and K108 were replaced by ε-acetyl lysine, representing the acetylated state. Interestingly, this modification substantially reduced, but did not eliminate binding of the peptide to the Wapl domain. We next decided to verify these results and the affect of lysine acetylation in the context of the intact proteins by analysing the binding of the Wapl domain to an intact Smc3 globular head. We made a construct containing the full Saccharomyces cerevisiae Smc3 ATPase domain with the coiled coils and hinge replaced by a short flexible linker (Supplementary Figures 2c and d) to create a monopartite ATPase head (ScSmc3AH). One complication that arose was the finding that the Smc3 heads cannot be acetylated in vitro with Eco1 (Borges et al, 2010). We therefore made a mutant version of the protein, where lysines 112 and 113 were replaced with asparagines as this has previously been shown to recapitulate the effect of acetylation (Ben-Shahar et al, 2008; Unal et al, 2008). The affinities were quantified using a surface-based bio-layer interferometry system (ForteBio Octet) with the ScWpl1259–647 being coupled to the surface and ScSmc3AH in solution. The data obtained allowed excellent fits for both association and dissociation curves with no non-specific binding observed to the sensor or GST tag (Supplementary Figure 2e). The binding parameters are shown in Table I. The full-length intact protein domains demonstrated considerably higher affinity compared to the experiments with the isolated peptides, presumably because of cooperative binding at multiple sites. As can be seen, Smc3AH K112N, K113N reproducibly shows a significantly lower affinity for the Wapl domain than the wild-type protein. Although the lysine to asparagine mutation bypasses the requirement for Eco1 in vivo, it probably does not represent a true acetyl lysine mimicking mutation in vitro. We would expect the difference in affinity to be greater against the natural acetylated substrate as the terminal methyl group of ε-acetyl lysine is chemically more distinct from the lysine ε-amine than the amide moiety in asparagine. Analysis of other lysine mutants (Rowland et al, 2009) suggests that neutralisation of the basic lysine side chain is the key function of acetylation, rather than formation of a specific side-chain interaction. Figure 4.Quantification of peptide binding. Fluorescence polarisation binding curves for the peptides WIS1–WIS3, and the K107Ac K108Ac derivative of WIS1, together with derived binding constants. Download figure Download PowerPoint Table 1. ScWpl1259–647-ScSmc3AH binding affinities Proteins KDa (nM) Kon (1/Ms × 103) Koff (1/s × 10−4) R2 Wild-type Smc3/Wild-type Wpl1 47 5.22±0.07 2.48±0.08 0.996 Smc3 K112N K113N 88 4.88±0.06 4.31±0.11 0.993 Smc3 L163A L166A D168A 104 4.77±0.06 4.97±0.12 0.993 Smc3 L163D L166D 102 5.88±0.08 5.97±0.12 0.989 Wpl1 N493A N622A L532E 82 4.20±0.03 3.44±0.06 0.998 Wpl1 N493A N622A L577R 77 5.54±0.06 4.28±0.10 0.993 Wpl1 N493E N622E 126 3.92±0.05 4.92±0.09 0.995 Wpl1 D271A D272A 167 2.37±0.03 3.95±0.06 0.998 Wpl1 K376Q K377Q K382Q K383Q 200 2.80±0.04 5.60±0.07 0.997 Structure of the Wapl domain bound to Smc3154–163 To gain structural insights into interaction between Wpl1 and cohesin, we attempted to determine the crystal structure of the Wapl domain bound to the Smc3 target peptides. We were successful in obtaining a structure of the protein bound to the WIS2 peptide at 2.01 Å resolution (Supplementary Table 1). Clear electron density was seen for residues 154–163 of Smc3 and allowed unambiguous assignment of the sequence (Figures 5A and B; and Supplementary Figure 3a). The peptide lies in the hydrophobic pocket on the surface of Wpl1 formed by motifs 3, 4, and 5. Residues 159–163 form a single α-helical turn, with the remainder of the peptide in an extended conformation. The primary interaction between the peptide and Wpl1 is formed between the conserved leucine residues L158 and L161 of Smc3 projecting into the hydrophobic cavity formed by helices α16, α17, α19, and α20 (Figure 5C; Supplementary Figure 3b). Acidic residues E162 and E163 form polar interactions primarily with main and side chains of N419, H548, and N549, as well as main chain carbonyl hydrogen bonds to the N547 side-chain amide. The N-terminal end of the peptide is more loosely bound to Wpl1 and curves away from the side of the binding cavity. Comparison to the equivalent helix in the Smc1 crystal structure (Figure 5D) shows that the C-terminal end of the peptide remains in a helical conformation. The terminal hydrophobic and acid pairs of residues that interact with Wpl1 are in an equivalent position to the chemically similar residues in Smc1. However, the N-terminal of the peptide is somewhat unwound and deviates from the Smc1 structure. It is not clear if this reflects the conformation that exists in unbound Smc3, though the N-terminal end of the peptide diverges considerably in amino-acid composition between Smc1 and Smc3, which might account for the differing secondary structure. This helix is disordered in certain structures of related ABC ATPases (Hopfner et al, 2000), and other studies have suggested it can adopt multiple conformations (see Discussion) indicating that the form bound to Wpl1 might relate to a specific intermediate in the catalytic cycle. Figure 5.Details of Wpl1–Smc3 interactions. (A) Ribbon diagram of the AgWpl1184–561–WIS2 complex. The Wapl domain is shown as a ribbon, and electron density for the peptide in an unbiased |Fo|−|Fc| omit map contoured at 2σ is depicted in red. (B) Stereo close-up of the electron density for the WIS2 peptide calculated as in (A). The peptide is indicated in yellow. (C) Close-up of the peptide-binding site. The hydrophobic surface on the Wapl domain comprising motifs 3, 4, and 5 (green, blue, and red, respectively). The WIS2 peptide is shown in pink. Residues from Wpl1 (blue) or Smc3 (pink) that participate in interactions are labelled. (D) Comparison of the WIS2 peptide as bound to the Wapl domain (pink) and the equivalent sequence in the Smc1 crystal structure (cyan). The matching residues at the C-terminal of the helix that participate in interactions with Wpl1 are labelled. Download figure Download PowerPoint Biochemical and in vivo characterisation of Wpl1–Smc3 interactions Our experiments have shown that mutations in the WIS1 peptide that mimic Smc3 acetylation reduce the affinity of Wpl1 for Smc3, and these mutations have previously been shown to be active in vivo, by allowing eco1-1 bypass (Ben-Shahar et al, 2008; Unal et al, 2008; Rowland et al, 2009). To further study the function of other interactions involving WIS2, WIS3, and the conserved Wapl motifs (Figure 6A), we carried out binding studies using structure-guided mutants that we predict would affect the Wpl1–WIS2 interface, as well as on conserved residues in Wpl1 motifs 1 and 2. Correct folding of all the mutant proteins was verified by thermal denaturation assays and gel filtration (Supplementary Figure 4). Affinities were measured using purified recombinant ScWpl1259–647 and ScSmc3AH as described previously. Mutations in the Wpl1–WIS2 interface were made in both proteins, and resulted in an approximate halving of affinity (Figure 6B; Table I), with similar effects being caused by mutations in either protein. The strongest effect was observed with the ScWpl1 N493E N622E (N419 and N547 in A. gossypii) mutations that our structure predicts would adversely interact with the conserved acidic residues at the C-terminal end of the WIS2 peptide. We next tested mutations in the conserved Wapl motifs 1 and 2, which lie at the N-terminal tip of the domain. Motif 1 forms an acidic patch, and includes the D272G mutation that has previously been identified as an eco1-1 suppressor (Rowland et al, 2009), while motif 2 is the basic helix that projects from the side of the protein (Figure 2). We made conservative mutations that would be predicted to neutralise these charged patches, but not affect the folding of the protein. Both sets of mutations substantially reduced the binding of Wpl1 to Smc3, with a four-fold reduction in affinity seen with motif 2 mutations (Figure 6B; Table I). Inactivation or deletion of Wpl1 allows bypass of lethality associated with the eco1-1 alleles. We tested the function of our binding mutants by introducing the proteins into an eco1-1 strain from which WPL1 had been deleted. The mutants were under the control of their endogenous promoter and were C-terminally HA tagged to allow detection. Western blot analysis verified that the mutant and wild-type proteins were expressed at the same level (Figures 6C and D). While re-introduction of wild-type Wpl1 totally blocks growth at the restrictive temperature, charge-neutralising mutations in motifs 1 and 2 rescue the cells, correlating with the reduction in Smc3 binding affinity observed in vitro, and demonstrating that the Wpl1–Smc3 interaction is physiologically relevant. Surprisingly, mutations in the WIS2 helix did not show strong rescue phenotypes (data not shown). This may reflect the fact that a simple weakening (as opposed to total disruption) of the interface is insufficient to affect the physiological function of the interaction, or that the molecular consequences of the mutations are not manifest by simple suppression of Eco1 deficiency. Figure 6.Mutagenesis studies of the Wpl1–Smc3 interaction. (A) Diagram of the Wapl domain surface showing the location of the conserved motifs 1–5. (B) Binding constants of ScWpl1259–647-ScSmc3AH mutants. The wild-type proteins are shown in white. Other mutants are coloured as follows: Smc3 WIS1—orange, Smc3 WIS2—pink, Wpl1 motifs 3,4,5—green, Wpl1 motif 1—red, Wpl1 motif 2—blue. Error bars are calculated from global fits of data obtained at eight different concentrations. (C) In vivo yeast viability assay of mutants in motifs 1 and 2. All mutants grow at the permissive temperature (left). At the restrictive temperature, mutants in motif 1 (D271A D272A) and motif 2 (K376Q K377Q K382Q K383Q) suppress lethality associated with loss of Eco1 function, indicating Wpl1 inactivation. (D) Cellular expression of the mutant proteins was verified by western blot. Expression levels of the mutant proteins were approximately similar to expression of wild-type Wpl1, both from the HA-tagged endogenous locus (lane 4) and introduced on an integrating plasmid (lane 5). Download figure Download PowerPoint Discussion In this study, we present the first structure of a member of the cohesin complex other than the core Smc1/Smc3/Scc1 subunits and show that it can directly bind the Smc3 ATPase via a conserved domain. Two of the three putative binding sites in Smc3 (WIS1 and WIS2) have been previously identified as being functionally important in either Smc3 or related ABC ATPases, and in this work, we show that mutations that are predicted to weaken binding to the third site involving the WIS3 demonstrate strong in vivo phenotypes. We discuss these points in more detail below. Further, we have shown that the strength of the interaction between Wpl1 and Smc3 is partially dependent on the acetylation state of the ATPase domain. The Smc1–Smc3 heterodimer forms a functional ATPase, whose activity is necessary for correct loading of the cohesin complex onto chromosomes (Arumugam et al, 2003; Weitzer et al, 2003). The molecular details of ATP binding and hydrolysis by the ABC-ATPase superfamily have been extensively studied (Hopfner and Tainer, 2003), but how these processes relate to cohesion establishment is still unclear. Interestingly, the WIS2 site binding identified in this work has been previously proposed to be involved in ATP-driven conformational changes in other members of the ABC ATPase family, and there are precedents for protein–protein interactions involving regions structurally homologous to the WIS2. The sequence constitutes part of two short helices in Smc3 that link the nucleotide-interacting Q-loop (Hopfner et al, 2000) to the base of the coiled-coil domain (Figure 7A). These helices (termed as signature-coupling helices, SCH) have been previously described in the structurally homologous Rad50 ABC ATPase as being involved in the transduction of nucleotide-driven conformational changes (Williams et al, 2011). In Rad50, it has been proposed that nucleotide binding splays these helices apart, which in turn alters the orientation of the coiled coils relative to the ATPase domain. This might function to alter the association of other proteins with the ATPase domains. In other ABC ATPases involved in membrane transport processes, such as the BtuCD vitamin transporter (Locher et al, 2002) the transmembrane subunit binds the ATPase domain at a sequence directly adjacent to the Q-loop, so as to connect ATP-driven motion to conformational changes in a partner protein (Figure 7B). This raises the possibility that Wpl1 binding to Smc3 may modulate the ATPase activity of the cohesin complex, or transduce ATP-driven conformational transitions. Figure 7.Conservation and functional implications of the WIS2 helix. (A) Close up of the Smc1 ATPase active site, showing the location of the signature-coupling helices (purple and orange) between the Q-loop (pink) and the coiled coils (cyan). The predicted position of the WIS2 peptide bound to Wpl1 is depicted in purple. (B) Superimposition of ScSmc1 with the ATPase domain of Rad50 (pdb i.d. 3qkt) and BtuD (pdb i.d. 1l7v). The conserved glutamine that de