Crystal structures of open and closed forms of cyclo ⁄ maltodextrin-binding protein Naoki Matsumoto 1 , Mitsugu Yamada 2, *, Yuma Kurakata 1 , Hiromi Yoshida 2,3 , Shigehiro Kamitori 2,3 , Atsushi Nishikawa 1 and Takashi Tonozuka 1 1 Department of Applied Biological Science, Tokyo University of Agriculture and Technology, Japan 2 Graduate School of Medicine, Kagawa University, Japan 3 Life Science Research Center, Kagawa University, Japan Cyclodextrins (CDs) are torus-shaped molecules made up of six [a-cyclodextrin (a-CD)], seven [b-cyclodextrin (b-CD)] or eight [c-cyclodextrin (c-CD)] glucose resi- dues. The structure of CDs resembles a hollow, trun- cated cone with hydrophilic rims and a central hydrophobic cavity capable of hosting a large number Keywords crystal structure; cyclodextrin; maltodextrin- binding protein; sugar transporter; Thermoactinomyces vulgaris Correspondence T. Tonozuka, Department of Applied Biological Science, Tokyo University of Agriculture and Technology, 3-5-8 Saiwai-cho, Fuchu, Tokyo 183-8509, Japan Fax: +81 42 3675705 Tel: +81 42 3675702 E-mail: tonozuka@cc.tuat.ac.jp *Present address Research Unit for Quantum Beam Life Science Initiative, Japan Atomic Energy Agency, 1233 Watanuki, Takasaki, Gunma 370-1292, Japan Database The coordinates and structure factors of TvuCMBP–a-CD, TvuCMBP–b-CD and TvuCMBP–G4 have been deposited in the Protein Data Bank under the accession codes 2ZYM, 2ZYN, and 2ZYO, respectively. The revised coordinate of TvuCMBP–c-CD has been deposited in the Protein Data Bank under the accession code 2ZYK (Received 1 February 2009, revised 17 March 2009, accepted 23 March 2009) doi:10.1111/j.1742-4658.2009.07020.x The crystal structures of Thermoactinomyces vulgaris cyclo ⁄ maltodextrin- binding protein (TvuCMBP) complexed with a-cyclodextrin (a-CD), b-cyclodextrin (b-CD) and maltotetraose (G4) have been determined. A common functional conformational change among all solute-binding pro- teins involves switching from an open form to a closed form, which facili- tates transporter binding. Escherichia coli maltodextrin-binding protein (EcoMBP), which is structurally homologous to TvuCMBP, has been deter- mined to adopt the open form when complexed with b-CD and the closed form when bound to G4. Here, we show that, unlike EcoMBP, TvuCMBP– a-CD and TvuCMBP–b-CD adopt the closed form when complexed, whereas TvuCMBP–G4 adopts the open form. Only two glucose residues are evident in the TvuCMBP–G4 structure, and these bind to the C-domain of TvuCMBP in a manner similar to the way in which maltose binds to the C-domain of EcoMBP. The superposition of TvuCMBP–a-CD, TvuCMBP–b-CD and TvuCMBP–c-CD shows that the positions and the orientations of three glucose residues in the cyclodextrin molecules overlay remarkably well. In addition, most of the amino acid residues interacting with these three glucose residues also participate in interactions with the two glucose residues in TvuCMBP–G4, regardless of whether the protein is in the closed or open form. Our results suggest that the mechanisms by which TvuCMBP changes from the open to the closed conformation and maintains the closed form appear to be different from those of EcoMBP, despite the fact that the amino acid residues responsible for the initial bind- ing of the ligands are well conserved between TvuCMBP and EcoMBP. Abbreviations CD, cyclodextrin; EcoALBP, Escherichia coli D-allose-binding protein; EcoMBP, Escherichia coli maltodextrin-binding protein; G4, maltotetraose; TvuCMBP, Thermoactinomyces vulgaris cyclo ⁄ maltodextrin-binding protein; a-CD, a-cyclodextrin; b -CD, b-cyclodextrin; c-CD, c-cyclodextrin. 3008 FEBS Journal 276 (2009) 3008–3019 ª 2009 The Authors Journal compilation ª 2009 FEBS of insoluble chemicals through hydrophobic interac- tions [1,2]. We have studied the proteins involved in the CD metabolism of a thermophilic actinomycete, Thermoactinomyces vulgaris R-47 [3–9], and deter- mined the crystal structures of two enzymes, TVA I [4,5] and TVA II [6,7], both of which hydrolyze CDs and a polysaccharide pullulan. We have also deter- mined the crystal structure of a solute-binding protein from T. vulgaris, and evaluated its binding affinities for various sugars, using fluorescence measurements [8,9]. The protein showed high affinities for both CDs and maltodextrins, and thus was designated T. vulgaris cyclo ⁄ maltodextrin-binding protein ( TvuCMBP). Many bacterial solute-binding proteins bound to a variety of sugars have been reported, and their crystal structures have been compared [10–21]. Although the proteins are categorized into three classes, they consist of two similar globular domains linked by a two- stranded or three-stranded hinge region. A common drastic conformational change is found in this protein family, which involves switching from an open to a closed form through a large-scale hinge-bending motion. TvuCMBP structurally resembles maltodex- trin-binding proteins from Escherichia coli [18–20], Thermococcus litoralis [22], Pyrococcus furiosus [23], and Alicyclobacillus acidocaldarius [24], and E. coli maltodextrin-binding protein (Eco MBP) is one of the best characterized solute-binding proteins. EcoMBP binds to a membrane-bound MalFGK 2 translocation complex comprising two permease domains, MalF and MalG, and two copies of the ATPase subunits, MalK [25–29]. The maltodextrin transport is driven by the energy provided from ATP hydrolysis by MalK, and induced by the complementary binding of the malto- dextrin-loaded EcoMBP to the external face of the transmembrane subunits of MalF and MalG. Large conformational changes of the EcoMBP—MalFGK 2 complex are induced, and maltodextrin permeates from the binding site of EcoMBP to the cytoplasm through the translocation pathway [25–29]. Unlike linear malto- dextrins, CDs are nonphysiological ligands for E. coli, and EcoMBP reportedly adopts an open form when complexed with b-CD [19]. We have previously reported the crystal structure of TvuCMBP complexed with c-CD at a moderate 2.5 A ˚ resolution [8]. Although the overall structure of TvuCMBP resembles that of EcoMBP (the amino acid sequence identity is 30.1%), TvuCMBP–c-CD has been determined as the closed form. Here, we present the crystal structures of TvuCMBP complexed with a-CD, b-CD and maltotetraose (G4) at higher resolutions. We also re-refined the coordinates of TvuCMBP– c-CD, and compared them, to make clear the mecha- nism of the ligand binding. It is interesting to note that, unlike EcoMBP, TvuCMBP–a-CD and TvuCMBP– b-CD were determined to be in the closed form, whereas TvuCMBP–G4 adopted the open form. Results and Discussion Overall structures of TvuCMBP–ligand complexes The crystal structures of TvuCMBP liganded with a-CD, b-CD and G4 have been determined at 1.7, 1.8 and 1.55 A ˚ resolutions, respectively (Table 1). The crys- tals of TvuCMBP in complex with a-CD and b-CD belong to the monoclinic system with space group C2, the unit cell parameters of which are similar to those of selenomethionine-substituted TvuCMBP–c-CD (the cell dimensions are a = 85.3 A ˚ , b = 49.3 A ˚ , c = 87.6 A ˚ , and b = 94.9°) [8], and contain one molecule in an asymmetric unit. On the other hand, the crystal of TvuCMBP–G4 belongs to space group P2 1 2 1 2 1 , with one molecule in an asymmetric unit. In Ramachandran plots, 91.1% (TvuCMBP–a-CD), 91.1% (TvuCMBP– b-CD) and 93.0% (TvuCMBP–G4) of residues were shown to be in the most favored regions, and no residue was in the generously allowed regions or disallowed regions, as calculated by the program procheck of ccp4 [30]. The electron density (2F o ) F c ) maps for the three complex structures contoured at 1r show continuous density for almost all main chain atoms, but the N-termi- nal segments (residues 1–16, a-CD complex; 1–14, b-CD complex; 1–12, G4 complex) are not visible, as previ- ously described in a report on TvuCMBP–c-CD [8]. The overall structures of the TvuCMBP–ligand com- plexes are shown in Fig. 1A,B. Like many other bacte- rial sugar-binding proteins, TvuCMBP consists of two globular domains, an N-domain and a C-domain, linked by three hinge regions. The structures of TvuCMBP– a-CD and TvuCMBP–b-CD are almost identical to that of TvuCMBP–c-CD. In contrast, TvuCMBP–G4 exhib- its a markedly different conformation from the TvuCMBP–CD complexes (Fig. 1C). Specifically, the N-domain and C-domain of TvuCMBP–G4 are far apart, and the cleft is opened up to the solvent. We have reported that TvuCMBP–c-CD structurally resem- bles the closed form of EcoMBP [8]. Superposition of TvuCMBP–G4 and the open form of EcoMBP shows that the conformations of these two structures are very similar (Fig. 1D). These results indicate that TvuCMBP–a-CD and TvuCMBP–b-CD adopt the closed form, whereas TvuCMBP–G4 adopts the open form. In the crystal structures of EcoMBP complexed with reduced maltooligosaccharides (maltotriitol and maltotetraitol), P2 1 crystals (determined as the open N. Matsumoto et al. Open ⁄ closed cyclo ⁄ maltodextrin-binding protein FEBS Journal 276 (2009) 3008–3019 ª 2009 The Authors Journal compilation ª 2009 FEBS 3009 form) and C2 crystals (determined as the closed form) were obtained at 4 and 23 °C, respectively [21], but, in this study, both of the P2 1 2 1 2 1 (TvuCMBP–G4) and C2 (TvuCMBP–a-CD and TvuCMBP–b-CD) crystals were grown at 20 °C under almost the same buffer conditions. Although the overall conformations of TvuCMBP–G4 and TvuCMBP–a-CD are distinct, the rmsd of their N-domains (except for an a-helix of residues 322–333, i.e. residues 17–127 and 282–321) is 0.531 A ˚ , and that of their C-domains (except for a loop of residues 191–198, i.e. residues 131–190, 199–278, and 338–397) is 0.516 A ˚ , indicating that the open and the closed forms of each domain are almost identical, despite differences in the overall structures. This result suggests that the confor- mational change is attributable to the flexible hinge region connecting the two rigid domains. Structure of TvuCMBP–G4 Although the whole map of TvuCMBP–G4 is well defined, the electron density map for the ligand, G4, was weak, and only two glucose residues were identi- fied in the omit map (Fig. 2A). The positions of elec- tron density, however, appear to be similar to those of two glucose residues, consisting of c-CD bound to the C-domain of TvuCMBP–c-CD (Fig. 2B). The glucose residues of c-CD have been previously numbered from Glc-1(c-CD) to Glc-8(c-CD) from the comparison with EcoMBP–G4 (Fig. 2C) [8]. Therefore, we used both of these proteins in our comparative analysis. Superposi- tion of the C-domains of TvuCMBP–G4 and EcoMBP–G4 indicated that the third and fourth glu- cose residues from the nonreducing end are located close to the electron density for the ligand in the TvuCMBP–G4 structure. On the basis of Glc-3(c-CD) and Glc-4(c-CD) of TvuCMBP–c-CD (Fig. 2B), the model for two glucose residues of G4 was readily placed. According to the numbering of EcoMBP–G4 and TvuCMBP–c-CD, the two glucose residues are labeled Glc-(a) and Glc-(b) (Fig. 2A–C). It is unclear whether Glc-(a) and Glc-(b) exactly match with the third and fourth glucose residues, respectively, of G4, Table 1. Data collection and refinement statistics. Complex G4 a-CD b-CD c-CD Protein Data Bank ID 2ZYO 2ZYM 2ZYN 2ZYK Data collection Beamline PF BL5A PF BL6A PF-AR NW12 PF-AR NW12 a Space group P2 1 2 1 2 1 C2 C2 C2 a Cell dimensions a(A ˚ ) 48.3 83.2 82.4 167.4 a b(A ˚ ) 79.8 46.3 46.5 95.3 a c(A ˚ ) 90.5 85.6 85.4 117.1 a b (°) 90 94.3 94.1 131.6 a Resolution range (A ˚ ) 50–1.55 (1.61–1.55) b 50–1.8 (1.86–1.80) b 50–1.7 (1.76–1.70) b 50–2.5 (2.66–2.50) a,b Measured reflections 298 866 88 067 100 796 181 528 a Unique reflections 49 866 28 257 33 896 47 691 a Completeness (%) 96.6 (78.8) b 91.9 (83.6) b 94.5 (97.2) b 100.0 (100.0) a,b I ⁄ r(I) 33.6 (3.1) b 34.6 (6.5) b 25.2 (3.4) b 24.0 (8.6) a,b R merge 0.062 (0.267) b 0.040 (0.245) b 0.081 (0.390) b 0.069 (0.217) a,b Refinement statistics R work 0.192 0.216 0.216 0.222 R free 0.221 0.257 0.254 0.285 rmsd Bond lengths (A ˚ ) 0.006 0.006 0.006 0.008 Bond angles (°) 0.9 1.0 1.0 1.2 Number of atoms Protein 2992 2964 2977 11 856 Ligand 23 66 77 352 Water 597 422 389 538 a Values are from [8]. b The values for the highest-resolution shells are given in parentheses. Open ⁄ closed cyclo ⁄ maltodextrin-binding protein N. Matsumoto et al. 3010 FEBS Journal 276 (2009) 3008–3019 ª 2009 The Authors Journal compilation ª 2009 FEBS and, unlike for the open form of TvuCMBP–G4, clear electron density for all the four glucose residues has been identified in the closed form of EcoMBP–G4 [20]. A possible explanation is that there could be multiple binding patterns in TvuCMBP–G4. The C-domain side of the sugar-binding cleft has been categorized into four regions, C-I (residues 170– 177), C-II (residues 227–232), C-III (residues 248–251), and C-IV (residues 359–365) [8]. Aromatic side chains of Trp360 (at region C-IV) and Tyr175 (at region C-I) stack with Glc-(a) and Glc-(b), respectively (Fig. 2A). A B C D Fig. 1. Overall structures of TvuCMBP in complexes with ligands. Complex structures with a-CD and G4 are shown in (A) and (B), respectively. The overall structure of TvuCMBP–b-CD is essentially identical to that of TvuCMBP–a-CD. N-domain, C-domain, hinge regions and ligands are shown in yellow, blue, red and green, respectively. Only two glucose residues are seen in TvuCMBP–G4. (C) The Ca backbones of TvuCMBP in complexes with a-CD (blue), b-CD (magenta), c-CD (green), and G4 (orange). C-domains of their structures are superposed. Ligands are represented as stick models. (D) Comparison of the Ca backbones of TvuCMBP–G4 (orange) and EcoMBP–b-CD (cyan). A B C D Fig. 2. Stereo views of the C-domain involved in G4 binding in TvuCMBP and related structures. Regions C-I, C-II, C-III and C-IV are in blue, yellow, magenta and red, respectively. The ligands are shown in gray. Three aromatic residues (two Trp and one Tyr) stacking with the glucose residues are indicated. (A) TvuCMBP–G4. An F o ) F c omit map at the 2.0r contoured level is in purple, and only two glucose residues, labeled (a) and (b), are seen in the map. (B) TvuCMBP–c-CD. The glucose residues of c-CD are labeled from 1 to 8. (C) EcoMBP–G4 (Protein Data Bank ID: 4MBP). The glucose residues of G4 are labeled from 1 to 4. (D) EcoMBP–maltose (Protein Data Bank ID: 1ANF). N. Matsumoto et al. Open ⁄ closed cyclo ⁄ maltodextrin-binding protein FEBS Journal 276 (2009) 3008–3019 ª 2009 The Authors Journal compilation ª 2009 FEBS 3011 Comparison of the C-domain of TvuCMBP–G4 with EcoMBP–maltose reveals that the two glucose residues, Glc-(a) and Glc-(b), of TvuCMBP–G4 and the maltose molecule bound to EcoMBP are located at structurally identical positions, and Trp340 and Tyr155 of EcoMBP are structurally identical to Trp360 and Tyr175 of TvuCMBP (Fig. 2A,D). These results sug- gest that Trp360 and Tyr175 play the key role in anchoring the G4 molecule. A schematic representation of the interaction between TvuCMBP and G4 is presented in Fig. 3A. There are numerous water molecules in the cavity formed by the N-domain and C-domain. Although the G4 molecule is located much closer to the C-domain than to the N-domain, Glc-(a) and Glc-(b) interact with Asp83 and Arg84 from the N-domain, and Glu129 (hinge-1), directly or through the water mole- cules (Figs 3A and 4A). The N-domain side of the sugar-binding cleft consists of three loops, regions N-I (residues 25–33), N-II (residues 56–61), and N-III (resi- dues 80–85), and Arg83 and Arg84 are located in region N-III. In EcoMBP, the ligand-induced move- ment of Glu111 has been proposed to be the triggering mechanism for the motion that enables the other domain to participate in the ligand binding [18]. In TvuCMBP, Glu129 in hinge-1 is identified as the corre- sponding residue, and may be responsible for trigger- ing the hinge-bending motion, together with Arg83 and Arg84 located in region N-III of the N-domain. We have reported that Leu59 (at region N-II) inter- acts with the central cavity of c-CD and appears to play the key role in binding to the sugar among the residues of the N-domain [8]. In TvuCMBP–G4, how- ever, Leu59 seems not to interact with the G4 mole- cule, as the closest distance between G4 [O2 of Glc-(a)] and Leu59 (atom CD2) is 7.6 A ˚ (Fig. 4A). In the closed form of EcoMBP, Trp62 (at region N-III) is found at the center of the inner curvature of the oligo- saccharides, suggesting that Trp62 of EcoMBP func- tions similarly to Leu59 of TvuCMBP. However, Trp62 of EcoMBP can interact with the oligosaccharides A B C D Fig. 3. Schematic drawing of the amino acid residues interacting with the glucose residues [labeled (a) and (b)] of G4 (A), the struc- turally shared three glucose residues (labeled 2, 3, and 4) of CDs (B), and the structurally nonconserved glucose residues of a-CD (labeled 5, 6, and 1) (C) and b-CD (labeled 5, 6, 7, and 1) (D). The figures are based on a cartoon generated by the program LIGPLOT [33]. The C-domain and N-domain sides are categorized into four (C-I–C-IV) and three (N-I–N-III) regions, respectively. Symbols: open circle, oxygen atom; closed circle, carbon atom; gray circle, nitrogen atom; dashed line, hydrogen bond. Residues involved in hydrophobic interactions are illustrated. Open ⁄ closed cyclo ⁄ maltodextrin-binding protein N. Matsumoto et al. 3012 FEBS Journal 276 (2009) 3008–3019 ª 2009 The Authors Journal compilation ª 2009 FEBS through the water molecules even in the open form EcoMBP–b-CD (Fig. 4B) [19], and this seems to be the most important difference between the open forms of TvuCMBP and EcoMBP. Structures of TvuCMBP–a-CD and TvuCMBP–b-CD In both TvuCMBP–a-CD and TvuCMBP–b-CD, omit maps show clear density for the CD molecules (Fig. 5A,B). The bound a-CD and b-CD are ellipsoidal torus-shaped molecules. Crystallographic studies of CDs alone have shown that the CD rings of a-CD [31] and b-CD [32] are distorted because of conformational strain, and a-CD and b-CD bound to TvuCMBP appear to have conformations similar to the CDs described in these reports. The structures of TvuCMBP–a-CD, TvuCMBP–b-CD and TvuCMBP– c-CD were superposed, and three glucose residues of all the CD molecules overlaid remarkably well (Fig. 5C). The glucose residues of TvuCMBP–c-CD have been numbered according to the numbering for EcoMBP–G4 [20], and here the glucose residues of a-CD and b-CD were labeled in the same manner. The three structurally homologous glucose residues of the CDs were numbered Glc-2–Glc-4, and the glucose residues of a-CD, b-CD and c-CD were designated Glc-1(a-CD)–Glc-6(a-CD), Glc-1(b-CD)–Glc-7(b-CD), and Glc-1(c-CD)–Glc-8(c-CD), respectively, as shown in Fig. 5A,B. The angles C4–O4¢–C1¢ produced by two glucose residues were calculated with the program geomcalc of ccp 4 (Table 2). The angles between the structurally shared glucose residues, Glc-2–Glc-3 and Glc-3–Glc-4, are close to the mean angles of each CD molecule; for example, those of Glc-2(a-CD)–Glc- 3(a-CD) and Glc-3(a-CD)–Glc-4(a-CD) are 153.0° and 152.6°, respectively, similar to the mean angle of a-CD, 151.7°. In contrast, the angles of Glc-1(a-CD)– Glc-2(a-CD) (146.6°), Glc-4(a-CD)–Glc-5(a-CD) (144.2°) and Glc-5(b-CD)–Glc-6(b-CD) (144.1°) are much smaller than the others (150–170°), resulting in distortion of the CD molecules. The interaction between the structurally shared three glucose residues, Glc-2(CD), Glc-3(CD), and Glc- 4(CD), was analyzed with the program ligplot (Fig. 3B) [33]. The amino acid residues that interact with Glc-2(CD), Glc-3(CD) and Glc-4(CD) are almost identical to those that interact with Glc-(a) and Glc- (b) (Fig. 3A), regardless of whether the conformation is the closed or open form. The nature of the interac- tions between the two is, however, different. In the open form, Arg84, Asp83 and Glu129 form hydrogen bonds with O3 of Glc-(a), O3 of Glc-(b), and O2 of Glc-(b), respectively, directly or through the water molecules, but in the closed form, Arg84, Asp83 and Glu129 form hydrogen bonds with different atoms, O2 of Glc-2(CD), O3 of Glc-3(CD), and O3 of A B Fig. 4. Stereo views of N-domains involved in ligand binding in the open forms of TvuCMBP [TvuCMBP–G4 (A)] and EcoMBP [EcoMBP–b-CD, Protein Data Bank ID: 1DMB (B)]. Regions N-I, N-II, N-III and hinge-1 are in green, orange, cyan and pink, respectively. The ligands are shown in gray. Hydrogen bonds and water molecules hydrogen-bonded to the ligands or amino acid residues are indicated in blue. N. Matsumoto et al. Open ⁄ closed cyclo ⁄ maltodextrin-binding protein FEBS Journal 276 (2009) 3008–3019 ª 2009 The Authors Journal compilation ª 2009 FEBS 3013 Glc-4(CD), respectively. In addition, Leu59 interacts with the ligand molecule when only TvuCMBP adopts the closed form, as described in the previous section. The structurally nonconserved glucose residues, Glc- 5(a-CD)–Glc-1(a-CD) and Glc-5(b-CD)–Glc-1(b-CD), were also analyzed with the program ligplot (Fig. 3C,D). Many of the amino acid residues (Asn26, Leu59, Glu170, Ala230, Asn234, Asn247, and Trp250) participate in the interaction with both a-CD and b-CD. The structures of TvuCMBP–a-CD and TvuCMBP–b-CD are, however, essentially identical, and the side chains of these amino acid residues are oriented in the same directions (Fig. 5C). All of the hydrogen bonds between TvuCMBP and the noncon- served glucose residues are mediated through the water molecules. These results suggest that the arrangement A B C Fig. 5. CD-binding sites in TvuCMBP. (A) Stereo view of the N-domain involved in a-CD binding. Regions N-I, N-II and N-III are shown in green, orange and cyan, respectively. The ligands (gray) and an F o ) F c omit map at the 2.0r contoured level (purple) are shown. The amino acid residues chosen on the basis of analysis with the program LIGPLOT [33] are indicated. The glucose residues Glc-1(a-CD)–Glc-6 (a-CD) are labeled 1–6. (B) Stereo view of the N-domain involved in b-CD binding. Color representations are as in (A). The glucose residues Glc-1(b-CD)–Glc-7(b-CD) are labeled 1–7. (C) Superposition of TvuCMBP in complexes with a-CD (blue), b-CD (magenta), and c-CD (green). Amino acid residues interacting with the CDs are illustrated as stick models. Three aromatic residues, Tyr175, Trp250, and Trp360, and some amino acid residues shown in Fig. 3C,D are illustrated as stick models. The glucose residues Glc-2(CD), Glc-3(CD) and Glc-4(CD) are labeled 2, 3 and 4, respectively. Table 3. Differences in Ca-torsion angles between TvuCMBP–G4 and TvuCMBP–a-CD in the vicinity of the hinge regions. Differ- ence ={ [/(TvuCMBP–G4) ) /(TvuCMBP–a-CD)] 2 +[w(Tv uCMBP–G4) ) (TvuCMBP–a-CD)] 2 } 1 ⁄ 2 . Hinge-1 Hinge-2 Hinge-3 Residue number Difference (°) Residue number Difference (°) Residue number Difference (°) 127 9.4 278 5.4 331 7.6 128 15.1 279 17.1 332 20.1 129 10.4 280 11.0 333 11.0 130 14.3 281 7.2 334 32.0 131 5.5 282 5.2 335 21.7 – – – – 336 75.6 – – – – 337 61.7 – – – – 338 3.4 Table 2. Angles C4–O4¢–C1¢ produced by two glucose residues in the CD molecules. –, not applicable. Numbers of the two glucose residues 1–2 2–3 3–4 4–5 5–6 6–1 or 6–7 7–1 or 7–8 8–1 Mean a-CD (°) 146.6 153.0 152.6 144.2 157.7 156.2 – – 151.7 b-CD (°) 153.8 158.7 163.2 162.3 144.1 163.3 165.8 – 158.7 c-CD (°) 168.3 164.2 161.1 165.2 167.1 159.7 172.3 159.8 164.7 Open ⁄ closed cyclo ⁄ maltodextrin-binding protein N. Matsumoto et al. 3014 FEBS Journal 276 (2009) 3008–3019 ª 2009 The Authors Journal compilation ª 2009 FEBS of the water molecules is important for binding to the ligands. Hinge bending We determined the range of the hinge region bending by comparing the open and closed conformations of TvuCMBP and superposing the N-domains or C-domains (Fig. 6). The critical conformational change seems to be centered around Ile128–Ser130 (hinge-1) and Lys279–Val281 (hinge-2). Hinge-1 and hinge-2 form a part of the b-strands (Fig. 6A,B), whereas hinge-3 (Asn334–Val337) seems to be a turn linking two a-helices (Fig. 6C,D). The interpretation of the bending motion of hinge-3 is, however, very compli- cated, and a superposition of individual domains indi- cated that the conformational changes are found in an a-helix (residues 322–333) and a loop (residues 191– 198), both of which are located near Asn334–Val337. Differences in the Ca-torsion angles of hinge-1 and hinge-2 between TvuCMBP–G4 and TvuCMBP–a-CD were calculated using the program cns (Table 3) [34]. Between the two complexes, the differences in hinge-1 and hinge-2 are only 7.2–17.1°, but these values are enough to convert the entire conformation, because hinge-1 and hinge-2 are located in the center of TvuCMBP. Many similarities can be seen between unliganded EcoMBP [18] and TvuCMBP–G4. As with EcoMBP, upon hinge bending, not only are the two domains of the TvuCMBP apart from each other, but the N-domain is twisted anticlockwise around the hinge region relative to the C-domain. In measuring the hinge bend, the dihedral angle was defined by four atoms, Tyr175–Ca (C-domain), Phe278–Ca (hinge-2), Ser130–Ca (hinge-1), and Leu59–Ca (N-domain), and the difference in the dihedral angles between TvuCMBP–G4 and TvuCMBP–a-CD was calculated. The twist angle was also defined by the difference in the dihedral angles of four atoms, Leu58–Ca (N-domain), Gly280–Ca (hinge-2), Lys282–Ca (C-domain), and Arg84–Ca (N-domain). The bend and twist angles of TvuCMBP are 24° and 7°, respectively. Similar measurements have been reported for EcoMBP, and the bending and twist angles (35° and A B C D E Fig. 6. Structures of the hinge regions of TvuCMBP–a-CD and TvuCMBP–G4. (A, B) Comparison of the residues in and around hinge-1 and hinge-2. Colors: blue, TvuCMBP–a-CD (closed form); orange, TvuCMBP–G4 (open form). N-domains (A) or C-domains (B) of TvuCMBP–a-CD and TvuCMBP–G4 were superposed. Water molecules hydrogen-bonded to hinge-1 or hinge-2 are shown in ball representation. (C, D) Comparison of the residues in and around hinge-3. N-domains (C) or C-domains (D) of TvuCMBP–a-CD and TvuCMBP–G4 were superposed, and hinge-3 (residues 334–337) and Ca backbones of two a-helices (residues 322–333 and 338– 348) are illustrated. Color representations are as described in (A) and (B). (E) Alignment of the primary structures of hinge-1 and hinge-2 of TvuCMBP and related solute-binding proteins. PfuMBP, P. furiosus maltodextrin-binding protein (amino acid sequence iden- tity = 33.3%) [23]; TliTMBP, Th. litoralis trehalose ⁄ maltose-binding protein (26.3%) [22]; AacMBP, A. acidocaldarius maltose ⁄ maltodex- trin-binding protein (30.2%) [24]; TthGBP, Thermus thermophilus glucose-binding protein (23.9%) [35]. The numbering of the amino acid sequences is given. Residues that are identical to those in TvuCMBP are shown in red. N. Matsumoto et al. Open ⁄ closed cyclo ⁄ maltodextrin-binding protein FEBS Journal 276 (2009) 3008–3019 ª 2009 The Authors Journal compilation ª 2009 FEBS 3015 8°, respectively) are roughly comparable to those of TvuCMBP. The structures of TvuCMBP-G4 and unli- ganded EcoMBP are almost identical, with a 1.84 A ˚ rmsd (Fig. 1D). The open and closed forms of several solute-binding proteins have been previously determined, and their hinge-bending motions have been analyzed [14,15,18– 21]. In E. coli d-allose-binding protein (EcoALBP), two water molecules embedded in the hinge appear to act as ballbearings, and residues with hydrophobic side chains located at the terminus of the hinge regions are proposed to function as grease [15]. The hinge regions of EcoALBP and TvuCMBP share similar structural features, although EcoALBP is categorized as a class I solute-binding protein [11,12], whereas TvuCMBP is a class II solute-binding protein, and the primary struc- tures of their hinge regions have virtually no similarity. In TvuCMBP, many water molecules are identified in the vicinity of Glu129 at hinge-1, and amino acid resi- dues with hydrophobic side chains, Ala127, Ile128, Phe278, Ile279, and Val281, are found at the termini of hinge-1 and hinge-2 (Fig. 6A,B). The primary struc- tures of hinge-1 and hinge-2 of TvuCMBP and related proteins [18–23,35] were aligned on the basis of struc- tural comparison, and EcoMBP appears to be the most similar to TvuCMBP (Fig. 6E). Insights into the conformational change of TvuCMBP Comparing and contrasting the EcoMBP and TvuCMBP structures has provided some interesting mechanistic insights. The crystal structures of EcoMBP indicated that EcoMBP–b-CD adopts the open form [19], whereas the complexes with linear maltodextrins, such as maltose, maltotriose, and G4, adopt the closed form [18,20]. The reason why EcoMBP–b-CD adopts the open form might be simple. In the closed form, the cleft of the sugar-binding site of EcoMBP is very nar- row, and CDs may not fit as readily as linear maltodext- rins, which are half-buried when EcoMBP adopts the closed form [8,20]. These observations suggest that the closed form of EcoMBP may cause steric hindrance with b-CD. In contrast, TvuCMBP–a-CD, TvuCMBP–b-CD and Tvu CMBP–c-CD have a closed conformation, whereas TvuCMBP–G4 remains in the open conforma- tion. It is, however, impossible to use steric hindrance as the explanation for the open conformation of TvuCMBP-G4, because the CD molecules are larger than the G4 molecule. The study of EcoMBP using paramagnetic NMR indicated that the predominantly open form coexists in rapid equilibrium with a minor closed species in the absence of ligand [36], indicating that the conformation of the solute-binding protein could be determined by the ratio of the open fi closed rate to the closed fi open rate. Comparison of the open forms of TvuCMBP and EcoMBP has revealed a possible explanation for the shifting of the equilibrium. Many of the residues inter- acting with Glc-(a) and Glc-(b) in the open form of TvuCMBP–G4 are conserved in EcoMBP. A report on EcoMBP has shown that Tyr155, Trp230 and Trp340 form the key nonpolar stacking interaction between aromatic residues [19], and these residues are strictly conserved in TvuCMBP as Tyr175, Trp250, and Trp360, respectively. We previously evaluated the affinities between TvuCMBP and various oligosaccha- rides by measuring the fluorescence intensities, indi- cating that the K d values for CDs and linear maltodextrins are almost identical [8]. Moreover, the amino acid sequences of hinge-1 and hinge-2 of TvuCMBP are very similar to those of EcoMBP (Fig. 6E), suggesting that these proteins’ mechanisms for initial ligand binding and hinge bending could be similar. In EcoMBP, Glu111 is has been reported to be significant in the ligand-induced hinge–twist interdo- main motion [20], and Trp62 appears to be important for the open fi closed conformational change. The functionally equivalent residue to Glu111 has been identified as Glu129, whereas no equivalent residue to Trp62 is found in TvuCMBP (Fig. 4). These observa- tions suggest that the mechanism of the open fi - closed conformational change in TvuCMBP is largely similar to that of EcoMBP, but short linear maltodext- rins might not affect the open fi closed conforma- tional change as readily in TvuCMBP as they do in EcoMBP. By comparison, the mechanisms of the closed fi open conformational changes seem to differ between these two proteins. In the closed form of EcoMBP (EcoMBP–G4; Protein Data Bank ID: 4MBP), five hydrogen bonds, Glu45 OE2–Tyr341 OH, Ala96 O–Asn332 ND2, Arg98 O–Asn332 ND2, Asp65 OD2– Trp340 NE1, and Ser233 OG–Asp296 O, form directly between the N-domain and C-domain, and appear to lock and stabilize the closed form of EcoMBP. On the other hand, the ligand-binding cleft of the closed form of TvuCMBP is much wider than that of EcoMBP (a surface model of TvuCMBP–c-CD has been illustrated previously [8]), and no direct hydrogen bond is observed in the interdomain interactions. To maintain the closed form of TvuCMBP, glucose residues of CDs, especially Glc-5(CD), Glc-6(CD), Glc-7(CD), Glc-8(CD), and Glc-1(CD), contribute to the forma- tion of many hydrogen bonds and may function as glue between the N-domain and C-domain. Open ⁄ closed cyclo ⁄ maltodextrin-binding protein N. Matsumoto et al. 3016 FEBS Journal 276 (2009) 3008–3019 ª 2009 The Authors Journal compilation ª 2009 FEBS Some bacteria and archaea have been proposed to have a specific uptake mechanism for CDs and to uti- lize CDs as a carbon source [37–39]. A bacterium, Klebsiella oxytoca, has been reported to have a CD metabolic pathway [39,40], and the identity of the pri- mary structures between the solute-binding protein, CymE, from K. oxytoca and TvuCMBP (34.6%) is higher than that between EcoMBP and TvuCMBP (30.1%), although the residue corresponding to Leu59 of TvuCMBP is not conserved in K. oxytoca CymE. Pajatsch et al. [41] investigated the properties of a CD-specific porin, CymA, from K. oxytoca. A mutant of K. oxytoca with a lesion of the LamB maltoporin gene was still able to grow on linear maltodextrins, and they concluded that CymA has a role in the uptake of both CDs and linear maltodextrins. The channel properties of CymA were measured from titra- tion experiments of the membrane conductance with sugars, indicating that CymA binds both CDs and linear maltodextrins, but the stability constant range for binding of linear maltodextrins (K = 310– 32 000 m –1 ) is much smaller than the range for binding CDs (K = 20–91 m )1 ) [42]. It is likely that TvuCMBP possesses porins similar to CymA, and a high concen- tration of linear maltodextrins may allow TvuCMBP to adopt the closed form. Experimental procedures Expression and purification of TvuCMBP The expression, purification and preparation of TvuCMBP in complex with a-CD, b-CD or G4 was performed using the same procedure as for TvuCMBP–c-CD [8,9]. Briefly, in the step of purification with amylose resin (New England Biolabs, Ipswich, MA, USA), TvuCMBP was eluted with 5 mm a-CD, b-CD or G4 instead of c-CD. The eluted TvuCMBP was bound to the ligands, and allowed us to obtain the a-CD, b-CD and G4 complexes. The samples were further purified using cation exchange chromatography (Hiload SP-Sepharose HR 16 ⁄ 10 column, 1.6 · 10 cm; GE Healthcare, Chalfont St Giles, UK) with elution by an NaCl gradient. Protein concen- trations were determined by the measurement of absorbance at 280 nm, as previously described [8]. Crystallization and data collection TvuCMBP–a-CD, TvuCMBP–b-CD and TvuCMBP–G4 were crystallized at 20 °C, using the hanging drop vapor diffusion method with the same conditions, where 1 lLof TvuCMBP–ligand solution (8 mgÆmL )1 )in50mm Mes ⁄ NaOH (pH 6.0) was mixed with the same volume of well solution (20% polyethylene glycol 6000, 50 mm Mes, pH 6.2). The obtained crystals were transferred to a solution consisting of 34% polyethylene glycol 6000 and 50 mm Mes (pH 6.0), and frozen in a 100 K nitrogen stream. Their dif- fraction data were collected at beamlines PF BL5A, PF BL6A and PF-AR NW12 (Photon Factory, Tsukuba, Japan). The data were processed and scaled with the program hkl2000 (Table 1) [43]. Model building and structure refinement The structures of TvuCMBP–a-CD, TvuCMBP–b-CD, and TvuCMBP–G4 were solved by molecular replacement with the program molrep in the ccp4 suite [30], and a model of TvuCMBP–c-CD (Protein Data Bank ID: 2DFZ) was employed as a probe model. To solve the open TvuCMBP– G4 structure, the probe model was divided into the N-domain and C-domain, and an adequate result was computed. The refinement was carried out using the program cns [34], and manual adjustment and rebuilding of the model were carried out with the programs xfit [44] and coot [45]. Superpositioning of TvuCMBP and other protein structures and the calculation of the rmsd were carried out with the pro- gram superpose in the ccp4 suite. Figures were generated using pymol (http://www.pymol.org/). Sequence identities were calculated using the program sim on the ExPASy server (http://www.expasy.org/tools/sim-prot.html). Based on the coordinates of TvuCMBP–a-CD and TvuCMBP–b-CD, incorrect rotamer assignments (especially Leu59) were discovered in the previous coordinate of TvuCMBP–c-CD (Protein Data Bank ID: 2DFZ, 2.5 A ˚ resolution) and were rebuilt with the program coot. The R work and R free values were converged at 0.222 and 0.285, respectively, and in a Ramachandran plot calculated with the program procheck of ccp4, no residues were present in the disallowed regions or the generously allowed regions, and 90.5% of residues were in the most favored region (Table 1). Acknowledgements We thank Hayashibara Biochemical Laboratories Inc. and Ensuiko Sugar Refining Co., Ltd for providing various sugars. This work was supported by a grant- in-aid for Scientific Research (20570103) from the Ministry of Education, Culture, Sports, Science and Technology of Japan. This research was performed with the approval of the Photon Factory Advisory Committee (2007G010 and 2008G013), the National Laboratory for High Energy Physics, Tsukuba, Japan. References 1 Davis ME & Brewster ME (2004) Cyclodextrin-based pharmaceutics: past, present and future. Nat Rev Drug Discov 3, 1023–1035. N. Matsumoto et al. Open ⁄ closed cyclo ⁄ maltodextrin-binding protein FEBS Journal 276 (2009) 3008–3019 ª 2009 The Authors Journal compilation ª 2009 FEBS 3017 [...]... Rodseth LE (1997) Extensive features of tight oligosaccharide binding revealed in high-resolution structures of the maltodextrin transport ⁄ chemosensory receptor Structure 5, 997–1015 Duan X, Hall JA, Nikaido H & Quiocho FA (2001) Crystal structures of the maltodextrin ⁄ maltose-binding protein complexed with reduced oligosaccharides: flexibility of tertiary structure and ligand binding J Mol Biol 306, 1115–1126... W, Hulsmann A, Schneider E & Mowbray SL (2004) X-ray structures of the maltose-maltodextrin-binding protein of the thermoacidophilic bacterium Alicyclobacillus acidocaldarius provide insight into acid stability of proteins J Mol Biol 335, 261–274 Hall JA, Ganesan AK, Chen J & Nikaido H (1997) Two modes of ligand binding in maltose-binding protein of Escherichia coli Functional significance in active.. .Open ⁄ closed cyclo ⁄ maltodextrin-binding protein N Matsumoto et al 2 Aachmann FL, Otzen DE, Laesen KL & Wimmer R (2003) Structural background of cyclodextrin protein interactions Protein Eng 16, 905–912 3 Ichikawa K, Tonozuka T, Uotsu-Tomita R, Mizuno M, Nishikawa A & Sakano Y (2005) Bacterial and archaeal enzymes homologous to glucoamylase: characterization and subsite affinities of a glucoamylase... (2003) CymA of ¨ Klebsiella oxytoca outer membrane: binding of cyclodextrins and study of the current noise of the open channel Biophys J 85, 876–885 43 Otwinowski Z & Minor W (1997) Processing of X-ray diffraction data collected in oscillation mode Methods Enzymol 276, 307–326 44 McRee D (1992) XtalView: a visual protein crystallographic software system for XII ⁄ XView J Mol Graph 10, 44–47 45 Emsley... Quiocho FA (1992) Crystallographic evidence of a large ligandinduced hinge–twist motion between the two domains of the maltodextrin binding protein involved in active transport and chemotaxis Biochemistry 31, 10657– 10663 Sharff AJ, Rodseth LE & Quiocho FA (1993) Refined ˚ 1.8-A structure reveals the mode of binding of b-cyclodextrin to the maltodextrin binding protein Biochemistry 32, 10553–10559 Quiocho... Tonozuka T & Sakano Y (2002) Crystal structures and structural comparison of Thermoactinomyces vulgaris R-47 a-amylase ˚ I (TVAI) at 1.6 A resolution and a-amylase II (TVAII) ˚ at 2.3 A resolution J Mol Biol 26, 443–453 5 Abe A, Tonozuka T, Sakano Y & Kamitori S (2004) Complex structures of Thermoactinomyces vulgaris R-47 a-amylase 1 with malto-oligosaccharides demonstrate the role of domain N acting as a... maltodextrin-binding protein FEBS J 274, 2109–2120 9 Yopi, Tonozuka T, Sakai H & Sakano Y (2002) Cloning of a gene cluster for dextrin utilization from Thermoactinomyces vulgaris R-47 and characterization of the cyclodextrin-binding protein J Appl Glycosci 49, 107–114 10 Vyas NK, Vyas MN & Quiocho FA (1991) Comparison of the periplasmic receptors for l-arabinose, d-glucose ⁄ d-galactose, and d-ribose Structural and. .. Kumagai H, Ashida H et al (2008) Structural and thermodynamic analyses of solute-binding protein from Bifidobacterium 3018 14 15 16 17 18 19 20 21 22 23 24 25 longum specific for core 1 disaccharide and lactoN-biose I J Biol Chem 283, 13165–13173 Bjorkman AJ & Mowbray SL (1998) Multiple open ¨ forms of ribose-binding protein trace the path of its conformational change J Mol Biol 279, 651–664 Magnusson U,... Hinge-bending motion of d-allose-binding protein from Escherichia coli: three open conformations J Biol Chem 277, 14077–14084 Nikaido H & Hall JA (1998) Overview of bacterial ABC transporters Methods Enzymol 292, 3–20 Lee SJ, Bohm A, Krug M & Boos W (2007) The ABC ¨ of binding -protein- dependent transport in Archaea Trends Microbiol 15, 389–397 Sharff AJ, Rodseth LE, Spurlino JC & Quiocho FA (1992) Crystallographic... FA, Davidson AL & Chen J (2007) Crystal structure of a catalytic intermediate of the maltose transporter Nature 450, 515–521 29 Oldham ML, Davidson AL & Chen J (2008) Structural insights into ABC transporter mechanism Curr Opin Struct Biol 18, 726–733 30 Collaborative Computational Project (1994) The CCP4 suite: programs for protein crystallography Acta Crystallogr D Biol Crystallogr 50, 760–763 31 Manor . the conformation of the solute-binding protein could be determined by the ratio of the open fi closed rate to the closed fi open rate. Comparison of the open forms of TvuCMBP and EcoMBP has revealed. Asp83 and Glu129 form hydrogen bonds with different atoms, O2 of Glc-2(CD), O3 of Glc-3(CD), and O3 of A B Fig. 4. Stereo views of N-domains involved in ligand binding in the open forms of TvuCMBP. for EcoMBP, and the bending and twist angles (35° and A B C D E Fig. 6. Structures of the hinge regions of TvuCMBP–a-CD and TvuCMBP–G4. (A, B) Comparison of the residues in and around hinge-1 and hinge-2.