X-ray structure of glucose/galactose receptor from Salmonella typhimurium in complex with the physiological ligand, (2R)-glyceryl-b- D-galactopyranoside Sanjeewani Sooriyaarachchi 1 , Wimal Ubhayasekera 1 , Winfried Boos 2 and Sherry L. Mowbray 1 1 Department of Molecular Biology, Swedish University of Agricultural Sciences, Uppsala, Sweden 2 Department of Biology, University of Konstanz, Germany Glucose ⁄ galactose-binding protein (GBP) was the first sugar-binding protein for which roles in active trans- port [1] and chemotaxis [2] were demonstrated. The transport occurs via a typical ABC system [3] consist- ing of three components: the periplasmic binding protein (GBP, or alternatively, MglB) that acts as the primary recognition site; a membrane-bound permease (MglC); and a cytoplasmic module (MglA) that cou- ples the binding ⁄ hydrolysis of ATP to transmembrane transport of the cognate substrates. In Escherichia coli and Salmonella enterica serovar Typhimurium (S. ty- phimurium), both galactose and glucose are physiologi- cally important ligands [4,5]. As well as having affinity for the nonphysiological b-methyl-galactoside, from which the name Mgl is derived, it was recognized early that the GBP from E. coli also binds glyceryl-b-d-ga- lactopyranoside [6]. Further work showed that only the (2R) diastereomer was bound [7], consistent with the fact that only this stereoisomer (hereafter referred to as GGal) is found naturally as the polar head group of plant glycolipids. An estimated 16.6% of the total lipids in runner bean leaves represents GGal [8], and a similar abundance has been found in other plants, such as red clover [9]. Conjugated forms are common in both plants and animals. Interestingly, GGal is also a good substrate for all three components of the lac operon, i.e. b-galacto- sidase, the lactose transporter and thiogalactoside Keywords galactose uptake; glucose ⁄ galactose-binding protein; glyceryl galactoside; lactose uptake; Salmonella enterica serovar Typhimurium Correspondence S. L. Mowbray, Department of Molecular Biology, Swedish University of Agricultural Sciences, Box 590, Biomedical Center, SE-751 24, Uppsala, Sweden Fax: +46 18 53 6971 Tel: +46 18 471 4990 E-mail: mowbray@xray.bmc.uu.se Website: http://xray.bmc.uu.se/ (Received 13 December 2008, revised 31 January 2009, accepted 2 February 2009) doi:10.1111/j.1742-4658.2009.06945.x Periplasmic binding proteins are abundant in bacteria by virtue of their essential roles as high-affinity receptors in ABC transport systems and chemotaxis. One of the best studied of these receptors is the so-called glucose ⁄ galactose-binding protein. Here, we report the X-ray structure of the Salmonella typhimurium protein bound to the physiologically relevant ligand, (2R)-glyceryl-b-d-galactopyranoside, solved by molecular replace- ment, and refined to 1.87 A ˚ resolution with R and R-free values of 17% and 22%. The structure identifies three amino acid residues that are diag- nostic of ( 2R)-glyceryl-b-d-galactopyranoside binding (Thr110, Asp154 and Gln261), as opposed to binding to the monosaccharides glucose and galac- tose. These three residues are conserved in essentially all available glucose ⁄ galactose-binding protein sequences, indicating that the binding of (2R)- glyceryl-b-d-galactopyranoside is the rule rather than the exception for receptors of this type. The role of (2R)-glyceryl-b-d-galactopyranoside in bacterial biology is discussed. Further, comparison of the available struc- tures provides the most complete description of the conformational changes of glucose ⁄ galactose-binding protein to date. The structures follow a smooth and continuous path from the most closed structure [that bound to (2R)-glyceryl-b-d-galactopyranoside] to the most open (an apo structure). Abbreviations GBP, glucose ⁄ galactose-binding protein; GGal, [2R]-glyceryl-b- D-galactopyranoside; PDB, Protein Data Bank (http://www.rcsb.org). 2116 FEBS Journal 276 (2009) 2116–2124 ª 2009 The Authors Journal compilation ª 2009 FEBS transacetylase [10]. The (2R), and not the (2S), diaste- reomer is formed by E. coli b-galactosidase during transfer of the galactosyl residue from any galactosyl donor (including lactose) to glycerol [7,11,12]. Further, unlike lactose itself, GGal is an excellent inducer for LacI, the repressor of the operon [13,14]. Considering these properties, one may be inclined to regard the name ‘lactose operon’ as a misnomer, as it seems likely that GGal, and not lactose, is the natural substrate of the system. Thus, GGal taken up by the Mgl trans- porter will induce expression of the lac operon, and so promote further uptake and utilization of the com- pound. Enterobacteriaceae, found in the gut of ani- mals, encounter GGal in large quantities via the ingestion of plant leaves (indeed, much more fre- quently than an adult mammal is exposed to the lac- tose contained in milk). In contrast to E. coli and most other Enterobacteriaceae, Salmonella has a deletion of the entire lac operon. However, because GGal can still be transported quite effectively by the Mgl transport system, it is expected that some other b-galactosi- dase(s) in Salmonella can be used to metabolize it. The K m of the Mgl transporter for GGal is 2.8 lm, comparable with the reported K d of GBP for this com- pound (3.2 lm) [6]. Measured using the same methods, the K m and K d values for galactose are similar, 0.5 and 1 lm, respectively; values for glucose are almost identi- cal to those of galactose, its C4 epimer [15]. Earlier crystal structures of GBPs from E. coli [16–18] and Salmonella [19–21] showed the basis of recognition for the monosaccharides. Here, we report the crystal struc- ture of Salmonella GBP in complex with GGal. We find that the protein provides a specific binding pocket for the d-glyceryl moiety, and that the amino acids lin- ing this pocket are highly conserved, reflecting the widespread importance of GGal as a bacterial carbon source. Results and Discussion Overall structure The structure of GBP in complex with GGal was determined by molecular replacement using the Salmo- nella GBP–Gal structure (PDB entry 1GCA) [21] as the search model, and refined to 1.87 A ˚ resolution with final R and R-free values of 17% and 22% (Table 1). Electron density was observed for all except residues 1–2 and 308–309 of the complete sequence in both molecules of the asymmetric unit. The structure is composed of two similar domains, each representing a b sheet sandwiched between two layers of a helices (Fig. 1A). Domain 1 is composed of residues 1–110 and 257–293; domain 2 includes residues 111–256 and 295–307. In each molecule, structural sodium and calcium ions are observed, bound in the loops following the first helices of domains 1 and 2, respectively. The EF-hand-like calcium site of domain 2 was described earlier, and tight binding of the ion was shown to con- tribute to the integrity of the protein structure [17,22]. The sodium site involves close interactions ( 2.3 A ˚ ) with Gly28-O, Ala31-O and Val34-O, as well as with well-ordered water molecules (2.3–2.6 A ˚ ). Although the concentration of sodium in the crystallization experiment (150 mm) falls within the generally accepted physiological range, no structural sodium ion was noted at the same position in earlier GBP struc- tures, from either Salmonella or E. coli. However, our inspection of the previous structures suggests that, in some cases, electron density modeled as a water mole- cule could actually be a sodium ion. One thiocyanate ion is also located in the asymmetric unit, based on the characteristic linear shape of the electron density, and the presence of 0.2 m NaSCN in the crystallization Table 1. Data collection and refinement statistics. Data collection a Environment ESRF ID14:4 Wavelength (A ˚ ) 0.955 Cell dimensions (A ˚ ) a = 36.4, b = 109.3, c = 150.7 Space group P2 1 2 1 2 1 Resolution (A ˚ ) 30.0–1.87 (1.97–1.87) Unique reflections 49 021 Average multiplicity 5 (5) Completeness (%) 96.4 (98.4) R merge b 13.5 (43.5) <(I) ⁄ r (I)> 9.9 (3.4) Refinement No. reflections (completeness, %) 46 530 (96%) Resolution range (A ˚ ) 30.0–1.87 R-factor, R-free (%) 17.0, 22.2 No. protein atoms (average B, A ˚ 2 ) c A molecule 2327 (9.6) B molecule 2329 (9.9) No. water molecules (average B, A ˚ 2 ) c 710 (21.3) No. ligand atoms (average B, A ˚ 2 ) c 34 (5.4) No. ions (average B, A ˚ 2 ) c Ca 2 (11.6) SCN 3 (10.6) Na 2 (10.2) Rms bond length (A ˚ ) 0.008 Rms bond angle (°) 1.052 Ramachandran plot outliers (n,%) d 4 (0.7%) a Values in parentheses are for the highest resolution shell. b R merge = P h P l jI hl ) ÆI h æ| ⁄ P h P l < I h >. c Calculated using MOLE- MAN [48]. d A stringent-boundary Ramachandran plot was used [49]. S. Sooriyaarachchi et al. GBP bound to (2R)-glyceryl-b- D-galactopyranoside FEBS Journal 276 (2009) 2116–2124 ª 2009 The Authors Journal compilation ª 2009 FEBS 2117 solutions; this site appears to have no links with struc- ture or function. The rms difference when all Ca atoms of the two molecules in the asymmetric unit are compared is 0.3 A ˚ , slightly greater than the expected coordinate error in the structures ( 0.1 A ˚ ). When the two domains are compared individually with a tightened cut-off of 0.5 A ˚ , it is seen that there is a very small (1.5°) difference in their relative orientations. A nearly perfect twofold axis (179°) relates the two molecules, with 750 A ˚ 2 on domain 1 of each molecule buried at the interface. Dimers have been reported previously for the E. coli protein under some conditions [23], however, inspection of a number of other GBP structures does not reveal any similar example, resulting from either non-crystallographic or crystallographic symmetry. GGal binding Electron density for the GGal ligand is clearly observed in the cleft between the two domains (Fig. 1). As illustrated in Fig. 2, 15 hydrogen bonds directly link protein and ligand, six of which arise from domain 1, and nine from domain 2. Two water mole- cules also make hydrogen bonds with the ligand; several other residues contribute hydrophobic inter- actions (Fig. 2). Most of these interactions have been identified previ- ously in complexes with glucose or galactose [20,21,24,25]. Asn91 is now shown to have an addi- tional role, forming a hydrogen bond to O2¢ of the glyceryl moiety. Asn256 was known to interact with O1 of the preferred b-sugars [26], and this role is preserved for the glycoside oxygen of GGal. Three other residues are exclusively linked to binding of the glyceryl moiety (marked with red ovals in Fig. 2B): N A B C Na + Ca +2 GGal Fig. 1. Structure of the GBP–GGal complex. (A) Overall structure of GBP, color-coded using a scheme going from blue at the N-termi- nus, through the rainbow to red at the C-terminus. The GGal ligand is shown in royal blue. Structural sodium and calcium ions are shown in red and blue, respectively. (B). Electron density of GGal in the final SIGMAA-weighted 2m|F o |–d|F c | map [50] contoured at 1 r = 0.49 e ⁄ A ˚ 3 . AB Fig. 2. Interactions in the binding site. (A) Stereoview of bound GGal showing GBP residues making hydrogen-bonding and aromatic inter- actions. (B) Schematic diagram of the hydrogen bonds between GBP and GGal. Interactions specific to the glyceryl moiety are marked with red ovals. GBP bound to (2R)-glyceryl-b- D-galactopyranoside S. Sooriyaarachchi et al. 2118 FEBS Journal 276 (2009) 2116–2124 ª 2009 The Authors Journal compilation ª 2009 FEBS Thr110 and Asp154 interact with O3¢, and Gln261 interacts with O2¢. These interactions increase the number of hydrogen bonds between the protein and ligand by five compared with the monosaccharides. The glyceryl moiety of GBP lies near the hinge of the protein, in a pocket that is otherwise filled only with water molecules (Fig. 3). Indeed, this pocket, which is lined by polar side chains, extends to the sur- face of the protein, suggesting that even longer com- pounds could be accommodated by GBP. However, it is not known what such compounds might be, or whether they could be accepted by the transport sys- tem. It is probably significant that the sugar unit of GGal lies closest to the portions of GBP that will make first contact with the permease, as deduced from mutagenesis studies summarized previously [27]. By presenting the sugar first, recognition by the permease can be largely independent of the presence or absence of the glyceryl moiety. Comparison with available sequences The presence of the equivalents to residues Thr110, Asp154 and Gln261 in a given GBP sequence would thus be expected to indicate GGal binding, as opposed to simply glucose ⁄ galactose binding. These residues are, in fact, well conserved in the sequences of proteins annotated as GBPs, some examples of which are given in Fig. 4. Asp154 and Gln261 are most tightly con- served, whereas Thr110 may be conservatively replaced by a serine residue; in more distant relatives, an alanine is sometimes observed in this position. We conclude that GBP’s role in the binding and transport of GGal is widespread in nature. By contrast, the residues lining the ‘extension’ of the glyceryl pocket that reaches the surface are not conserved (Fig. 4). It should also be noted that a large number of sequences are annotated incorrectly, as periplasmic binding proteins of unknown specificity, lacI-type repressors or even enzymes (Fig. 4). Although designa- tion of a particular binding protein’s specificity should ultimately rely on a complete biochemical characteriza- tion, the patterns of conservation indicate that it is rather simple to distinguish GBPs from even their nearest relatives, the ribose-binding proteins. Examples of such features include residues Tyr10, His152 and Asp154, which are clearly present in the YP_087835.1 sequence (annotated as a RbsB), but replaced by other residues in the authentic ribose-binding proteins. In addition, the repressor sequences include a DNA-bind- ing headpiece, and so are consistently longer than those of the binding proteins, even if one includes their signal sequences; for example, the sequence of E. coli LacI is 363 residues, whereas the longest binding pro- tein of this type is typically 350 residues or fewer, and lacks the characteristic DNA-binding domain. Thus, modest improvements to the existing methods of anal- ysis ⁄ annotation would provide significant benefits, given that such proteins account for a large proportion of the bacterial genome. An unrelated type of glucose-binding protein has been identified in some bacteria; its fold is not similar to GBP, but rather to that of the larger maltose-bind- ing protein. This kind of protein is exemplified by the Thermus thermophilus protein, PDB entry 2B3B [28]. The mode of binding the monosaccharide is completely different in terms of orientation of the sugar, and inter- actions between protein and sugar, from that observed for GBP. Further, there appears to be no room within the structure to accommodate the additional glyceryl moiety. Thus, GGal binding is not expected to be a characteristic of this family of proteins. Conformational changes As described above, the two molecules in the asymmet- ric unit of our structure differ only slightly ( 1.5°) in their degree of opening. The similarity between the Fig. 3. Extension of the GGal site. Stereo- view of the residues lining the water-filled tunnel that extends from the glyceryl moiety to the surface of GBP are shown. S. Sooriyaarachchi et al. GBP bound to (2R)-glyceryl-b- D-galactopyranoside FEBS Journal 276 (2009) 2116–2124 ª 2009 The Authors Journal compilation ª 2009 FEBS 2119 Fig. 4. Sequence alignments. Representative sequences were identified by a BLAST search, and aligned using INDONESIA [45] after removal of the signal sequences using the SIGNAL P program [51]. Residues interacting directly (via either van der Waals interactions or hydrogen bonds) with the monosaccharide unit in the current complex are marked with cyan, and those specifically related to the glyceryl moiety with red. Residues lining the tunnel extending from the glyceryl site are marked in gray. The sequences were annotated as follows (number of resi- dues given in each case in parentheses): YP_001783460, periplasmic binding protein ⁄ LacI transcriptional regulator Haemophilus somnus 2336 (328); YP_087835.1, RbsB protein from Mannheimia succiniciproducens MBEL55E (330); ZP_01786351, galactose-1-phosphate uridylyl- transferase from Haemophilus influenzae 22.4-21 (331); ZP_01169389.1, probable galactoside ABC transporter from Bacillus sp. NRRL B-14911 (353); ZP_00134897.2, periplasmic component of ABC-type sugar transport system, Actinobacillus pleuropneumoniae serovar 1 str. 4074 (323); YP_720691.1, putative galactoside ABC transporter from Trichodesmium erythraeum IMS101 (342); ZP_02849935.1, periplasmic binding protein ⁄ LacI transcriptional regulator from Paenibacillus sp. JDR-2 (338); YP_001311499.1, periplasmic binding protein ⁄ LacI transcrip- tional regulator Clostridium beijerinckii NCIMB 8052 (356); ZP_02035313.1, hypothetical protein BACCAP_00909 from Bacteroides capillosus ATCC 29799 (333). 2GX6 and 2IOY are authentic ribose-binding protein sequences for which structures are known [52] (M. J. Cuneo and H. W. Hellinga, unpublished results). GBP bound to (2R)-glyceryl-b- D-galactopyranoside S. Sooriyaarachchi et al. 2120 FEBS Journal 276 (2009) 2116–2124 ª 2009 The Authors Journal compilation ª 2009 FEBS two molecules indicates that their conformation is affected very little by differences in crystal packing. Comparison with the structures of Salmonella GBP in complex with galactose (1GCA) [21] and glucose (3GBP) [20] indicates that both are more open by  5°, as illustrated in Fig. 5A. The structure of the same protein, closed but without bound sugar (1GCG) [25], is even more open ( 7° compared with the new structures). A number of structures are also available for E. coli GBP (1GLG, 2GBP, 2HPH, 2FVY, 2FW0, 2IPN, 2IPM, 2IPL, 2GX6) [18,24–26] (M. J. Cueno and H. W. Hellinga, unpublished results), which given the 94% amino acid sequence identity, can be compared with Salmonella GBP with confidence. Least-squares superimposition of domain 1 of all of the GBP struc- tures is shown in Fig. 5B, illustrating the ‘fan’ of related conformations observed. The GGal complex is the most closed structure found to date, perhaps because of the significantly larger number of hydrogen bonds compared with the structures with simple sug- ars. The other structures represent a series of confor- mations that ‘link’ the GGal complex to the most open (apo, 2FW0) structure (by  37°) through similar motions at the hinge. As shown in Table 2, the three hinge strands do not contribute equally. Changes in relatively few main-chain dihedral angles (primarily ones in the first hinge segment, that near residue 110) account for most of the motion observed. Interestingly, Gly109 is a Ramachandran outlier in the closed struc- tures, but not in the most open one. We conclude that, like the ribose- and allose-binding proteins of the same structural class [29,30], GBP has a preferred conforma- tional pathway in its motions. However, inspection of Table 2 quickly shows that the motions are not of the same character in the three proteins, and that the three hinge segments contribute to different degrees. The A B Fig. 5. Conformational changes. (A) Stereo representation showing the different domain relationships seen when binding galactose (PDB entry 1GCA, gold) compared with GGal (A molecule, blue). Domain 1 of the two structures is superimposed. (B) Superposition of domain 1 in the available GBP structures from Salmonella and E. coli. The structures are colored progressing from blue (most closed) to green (most open) in the series: GGal, GGal molecule B (1.5°), 2GBP (1.7°), 1GLG (1.8°), 2IPN (2.0°), 2HPH (2.0°), 2IPM (2.0°), 2IPL (3.4°), 1GCA (5.1°), 3GBP (5.4°), 1GCG (7.0°), 2FVY (9.8°). 2FW0 (opened by 36.8°) was not shown for reasons of clarity. Table 2. Comparison of conformational changes. Structures of GBP (GBP–GGal versus PDB entry 2FW0), ribose-binding protein (2DRI versus 1URP) and allose-binding protein (1RPJ versus 1GUD) were compared with the delta-dihedral command of the program LSQMAN [44,48], which calculates Ca-Ca-Ca-Ca torsion angles. Only differences > 10° are shown for residues in the three hinge seg- ments of each protein; equivalent residues of the various structures are aligned. Where more than one molecule was present in the respective asymmetric unit, the A molecule was used for the calculation. Both open ribose- and allose-binding protein structures differ by 43° from their closed forms. The two proteins have 34% amino acid sequence identity to each other, and 28% and 25%, respectively, to Salmonella GBP. Protein GBP Ribose-binding protein Allose-binding protein Segment 1 Val108 10.3 Ile101 )14.4 Gly109 39.7 Ala102 )24.2 Thr110 18.4 Thr112 12.7 Asp111 )25.7 Set112 )18.9 Glu114 )10.9 Segment 2 Val254 21.4 Ile233 13.4 Val245 35.9 Ala234 )30.5 Ala246 )54.0 Gln235 12.9 Gln247 )10.1 Asn248 )12.9 Segment 3 Val291 )12.8 Pro262 )21.0 Val293 16.5 Val281 )11.5 Pro294 )15.0 Asp264 16.5 Asp282 )25.3 Tyr295 )13.9 Leu265 )44.0 Ser283 )26.8 Val296 10.5 Ile284 12.9 S. Sooriyaarachchi et al. GBP bound to (2R)-glyceryl-b- D-galactopyranoside FEBS Journal 276 (2009) 2116–2124 ª 2009 The Authors Journal compilation ª 2009 FEBS 2121 changes observed must be relevant both to the closing that traps bound sugars, and the opening required for a ligand’s release into the membrane-bound compo- nents of the ABC transport systems. Differences in the direction of the motion could provide an additional level of specificity in the action of such systems. Experimental procedures Protein purification E. coli strain LA5709 [31], transformed with plasmid pBD10 [32], was used to overexpress GBP in Luria–Bertani medium containing 50 lgÆmL )1 ampicillin, as described previously [20,33]. Following expression, the osmotic (chloroform) shock fluid was removed and precipitated overnight using 60% (w ⁄ v) ammonium sulfate. The pellet was resuspended in 10 mm Tris ⁄ HCl buffer (pH 8.0), then dialyzed against the same buffer. The resulting sample was centrifuged at 5000 g at 4 °C for 15 min, passed through a membrane filter (0.22 lm) and concentrated (Vivaspin concentrator, 10 kDa cut-off, from Vivascience, Littleton, MA, USA). The con- centrated samples were purified using cation-exchange chro- matography, followed by anion exchange and gel filtration on a Superdex 75 16 ⁄ 60 column. The eluted fractions were analyzed by SDS ⁄ PAGE. To remove endogenously bound sugar, the purified pro- tein sample was treated with 8 m urea and incubated at room temperature for 30 min, then dialyzed in steps against 6, 4, 2, 1 and 0 m urea in 10 mm Tris ⁄ HCl buffer (pH 7.4) containing 1 mm CaCl 2 at 4 °C. The final concentrated protein sample was analyzed by SDS and native PAGE to confirm its homogeneity. Protein was stored in 10 mm Hepes (pH 7.0), 150 mm NaCl at )20 °C. Crystallization GBP was crystallized using the hanging-drop vapor diffu- sion method at room temperature. Drops were composed of 1.0 lL mother liquor [20% w ⁄ v poly(ethylene gly- col) 3350, 0.2 m NaSCN] and 1.0 lL of a solution com- posed of 0.29 mm (10 mgÆmL )1 ) protein and 0.60 mm GGal (synthesized as described earlier [11]). Crystal formation was facilitated by streak-seeding immediately after set-up. Prior to data collection, the thin plate-like crystals were stabilized by a cryoprotectant solution [35% w ⁄ v poly(eth- ylene glycol) 3350 in the same buffer] and then flash-cooled directly in liquid nitrogen. Data collection, structure solution, refinement and model building X-Ray data were collected at 100 K at beamline ID14:4 of the European Synchrotron Radiation Facility (ESRF, Grenoble, France). Data were processed with mosflm [34] and scaled with scala [35]. Analysis of the unit-cell content of GBP suggested that there would be two molecules in the asymmetric unit, consistent with a solvent content of 46% and a V m of 2.3 [36]. A relatively high R merge arose from some anisotropy in the data attributable to the thin, plate- like shape of the crystals. Molecular replacement with molrep [37], as implemented in the ccp4 interface [38,39], utilized the protein only of the unliganded form of GBP (PDB entry 1GCA [21] as the search model. The clear solution was improved with rigid-body and restrained refinement in refmac5 [40]. The protein was rebuilt as needed in o [41] and refined in a cyclical fashion. Waters were placed using the ARP ⁄ warp-solvent command in ccp4 [38]. Statistics for the data processing and final refined model are presented in Table 1. Structure factors and coordinates have been deposited at the PDB with the accession code 3GA5. Structural analysis, comparisons and figure preparation Similar proteins were located using blast [42]. Structures were obtained from the PDB [43] and compared using o and lsqman [44]. Similar sequences were aligned using indonesia [45]. Figures were prepared with the programs o, molscript [46], molray [47] and isis ⁄ draw (http://www.mdli.com). Acknowledgements This work was supported by grants from the Swedish Research Council (VR). We thank ESRF staff mem- bers for their support during the data collection. References 1 Boos W (1972) Structurally defective galactose-binding protein isolated from a mutant negative in the b-meth- ylgalactoside transport system of Escherichia coli. J Biol Chem 247, 5414–5424. 2 Hazelbauer GL & Adler J (1971) Role of the galactose binding protein in chemotaxis of Escherichia coli toward galactose. Nat New Biol 230, 101–104. 3 Davidson AL, Dassa E, Orelle C & Chen J (2008) Structure, function, and evolution of bacterial ATP-binding cassette systems. Microbiol Mol Biol Rev 72, 317–364. 4 Rotman B, Ganesan AK & Guzman R (1968) Trans- port systems for galactose and galactosides in Escheri- chia coli. II. Substrate and inducer specificities. J Mol Biol 36, 247–260. 5Mu ¨ ller N, Heine HG & Boos W (1985) Characteriza- tion of the Salmonella typhimurium mgl operon and its gene products. J Bacteriol 163, 37–45. GBP bound to (2R)-glyceryl-b-D-galactopyranoside S. Sooriyaarachchi et al. 2122 FEBS Journal 276 (2009) 2116–2124 ª 2009 The Authors Journal compilation ª 2009 FEBS 6 Boos W (1969) The galactose binding protein and its relationship to the b -methylgalactoside permease from Escherichia coli. Eur J Biochem 10, 66–73. 7 Silhavy TJ & Boos W (1973) A convenient synthesis of (2R)-glyceryl-b-d-galactopyranoside – substrate for b-galactosidase, lactose repressor, galactose-binding protein, and b-methylgalactoside transport system. J Biol Chem 248, 6571–6574. 8 Sastry PS & Kates M (1964) Lipid components of leaves. V. Galactolipids, cerebrosides, and lecithin of runner-bean leaves. Biochemistry 3, 1271–1280. 9 Weenink RO (1961) Acetone-soluble lipids of grasses and other forage plants. 1. Galactolipids of red clover (Trifolium pratense) leaves. J Sci Food Agric 12, 34–38. 10 Musso RE & Zabin I (1973) Substrate specificity and kinetic studies on thiogalactoside transacetylase. Bio- chemistry 12, 553–557. 11 Boos W (1982) Synthesis of (2R)-glycerol-ortho-b-d-ga- lactopyranoside by b-galactosidase. Meth Enzymol 89, 59–64. 12 Boos W, Lehmann J & Wallenfels K (1966) Asymmetri- scher Galaktosyltransfer auf Glycerin mit b-Galaktosi- dase aus E. coli. Carbohydr Res 1, 419–420. 13 Burstein C, Cohn M, Kepes A & Monod J (1965) Role du lactose et de ses produits metaboliques dans l’induc- tion de l’operon lactose chez Escherichia coli. Biochim Biophys Acta 95, 634–639. 14 Boos W & Wallenfels K (1968) Untersuchungen zur Induktion der Lac enzyme. 2. Die Permeation von Galactosylglycerin in Escherichia coli. Eur J Biochem 3, 360–363. 15 Anraku Y (1968) Transport of sugars and amino acids in bacteria. I. Purification and specificity of the galac- tose- and leucine-binding proteins. J Biol Chem 243, 3116–3122. 16 Vyas NK, Vyas MN & Quiocho FA (1983) The 3 A ˚ resolution structure of a deuterium–galactose-binding protein for transport and chemotaxis in Escherichia coli. Proc Natl Acad Sci USA 80, 1792–1796. 17 Vyas NK, Vyas MN & Quiocho FA (1987) A novel cal- cium-binding site in the galactose-binding protein of bac- terial transport and chemotaxis. Nature 327, 635–638. 18 Vyas MN, Vyas NK & Quiocho FA (1994) Crystallo- graphic analysis of the epimeric and anomeric specificity of the periplasmic transport ⁄ chemosensory protein receptor for d-glucose and d-galactose. Biochemistry 33, 4762–4768. 19 Mowbray SL & Petsko GA (1983) The X-ray structure of the periplasmic galactose-binding protein from Sal- monella typhimurium at 3.0 A ˚ resolution. J Biol Chem 258, 7991–7997. 20 Mowbray SL, Smith RD & Cole LB (1990) Structure of the periplasmic glucose ⁄ galactose receptor of Salmonella typhimurium . Receptor 1 , 41–53. 21 Zou JY, Flocco MM & Mowbray SL (1993) The 1.7 A ˚ refined X-ray structure of the periplasmic glucose ⁄ galac- tose receptor from Salmonella typhimurium. J Mol Biol 233, 739–752. 22 Luck LA & Falke JJ (1991) 19 F NMR studies of the d-galactose chemosensory receptor. 2. Ca(II) binding yields a local structural change. Biochemistry 30, 4257–4261. 23 Rasched I, Shuman H & Boos W (1976) The dimer of the Escherichia coli galactose-binding protein. Eur J Biochem 69, 545–550. 24 Vyas NK, Vyas MN & Quiocho FA (1988) Sugar and signal-transducer binding sites of the Escherichia coli galactose chemoreceptor protein. Science 242, 1290– 1295. 25 Flocco MM & Mowbray SL (1994) The 1.9 A ˚ X-ray structure of a closed unliganded form of the periplasmic glucose ⁄ galactose receptor from Salmonella typhimuri- um. J Biol Chem 269, 8931–8936. 26 Borrok MJ, Kiessling LL & Forest KT (2007) Confor- mational changes of glucose ⁄ galactose-binding protein illuminated by open, unliganded, and ultra-high-resolu- tion ligand-bound structures. Protein Sci 16, 1032–1041. 27 Bjo ¨ rkman AJ, Binnie RA, Cole LB, Zhang H, Hermod- son MA & Mowbray SL (1994) Identical mutations at corresponding positions in two homologous proteins with nonidentical effects. J Biol Chem 269, 11196– 11200. 28 Cuneo MJ, Changela A, Warren JJ, Beese LS & Hel- linga HW (2006) The crystal structure of a thermophilic glucose binding protein reveals adaptations that inter- convert mono and di-saccharide binding sites. J Mol Biol 362, 259–270. 29 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. 30 Magnusson U, Chaudhuri BN, Ko J, Park C, Jones TA & Mowbray SL (2002) Hinge-bending motion of d-allose binding protein from Escherichia coli: three open conformations. J Biol Chem 277, 14077–14084. 31 Mu ¨ ller N, Heine HG & Boos W (1982) Cloning of mglB, the structural gene for the galactose-binding pro- tein of Salmonella typhimurium and Escherichia coli. Mol Gen Genet 185, 473–480. 32 Benner-Luger D & Boos W (1988) The mglB sequence of Salmonella typhimurium LT2; promoter analysis by gene fusions and evidence for a divergently oriented gene coding for the mgl repressor. Mol Gen Genet 214, 579–587. 33 Willis RC & Furlong CE (1974) Purification and prop- erties of a ribose-binding protein from Escherichia coli. J Biol Chem 249 , 6926–6929. 34 Leslie AG (1999) Integration of macromolecular diffrac- tion data. Acta Crystallogr D 55, 1696–1702. S. Sooriyaarachchi et al. GBP bound to (2R)-glyceryl-b-D-galactopyranoside FEBS Journal 276 (2009) 2116–2124 ª 2009 The Authors Journal compilation ª 2009 FEBS 2123 35 Evans PR (1993) Data reduction. In Proceedings of CCP4 Study Weekend on Data Collection and Processing (Sawyer L, Isaac N & Bailey S, eds), pp. 114–122. Daresbury Laboratory, Warrington. 36 Matthews BW (1968) Solvent content of protein crys- tals. J Mol Biol 33, 491–497. 37 Vagin A & Teplyakov A (1997) MOLREP: an auto- mated program for molecular replacement. J Appl Crys- tallogr 30, 1022–1025. 38 Collaborative Computing Project Number 4 (1994) The CCP4 Suite – programs for protein crystallography. Acta Crystallogr D 50, 760–763. 39 Potterton E, Briggs P, Turkenburg M & Dodson E (2003) A graphical user interface to the CCP4 program suite. Acta Crystallogr D 59, 1131–1137. 40 Murshudov GN, Vagin AA & Dodson EJ (1997) Refinement of macromolecular structures by the maxi- mum-likelihood method. Acta Crystallogr D 53, 240– 255. 41 Jones TA, Zou JY, Cowan SW & Kjeldgaard M (1991) Improved methods for building protein models in elec- tron density maps and the location of errors in these models. Acta Crystallogr 47, 110–119. 42 Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, Miller W & Lipman DJ (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 25, 3389–3402. 43 Berman HM, Westbrook J, Feng Z, Gilliland G, Bhat TN, Weissig H, Shindyalov IN & Bourne PE (2000) The Protein Data Bank. Nucleic Acids Res 28, 235– 242. 44 Kleywegt GJ & Jones TA (1997) Detecting folding motifs and similarities in protein structures. Macromol Crystallogr B 277, 525–545. 45 Madsen D, Johansson P & Kleywegt GJ (2002) Indone- sia: an integrated sequence analysis system. http:// xray.bmc.uu.se/dennis/. 46 Kraulis PJ (1991) Molscript – a program to produce both detailed and schematic plots of protein structures. J Appl Crystallogr 24, 946–950. 47 Harris M & Jones TA (2001) Molray – a web interface between O and the POV-Ray ray tracer. Acta Crystal- logr D 57, 1201–1203. 48 Kleywegt GJ (1997) Validation of protein models from C-alpha coordinates alone. J Mol Biol 273, 371–376. 49 Kleywegt GJ & Jones TA (1996) Phi ⁄ psi-cology: Rama- chandran revisited. Structure 4, 1395–1400. 50 Read RJ (1986) Improved Fourier coefficients for maps using phases from partial structures with errors. Acta Crystallogr A 42, 140–149. 51 Bendtsen JD, Nielsen H, von Heijne G & Brunak S (2004) Improved prediction of signal peptides: Sig- nalP 3.0. J Mol Biol 340, 783–795. 52 Cuneo MJ, Tian Y, Allert M & Hellinga HW (2008) The backbone structure of the thermophilic Thermo- anaerobacter tengcongensis ribose binding protein is essentially identical to its mesophilic E. coli homolog. BMC Struct Biol 8, 20. GBP bound to (2R)-glyceryl-b-D-galactopyranoside S. Sooriyaarachchi et al. 2124 FEBS Journal 276 (2009) 2116–2124 ª 2009 The Authors Journal compilation ª 2009 FEBS . of the larger maltose-bind- ing protein. This kind of protein is exemplified by the Thermus thermophilus protein, PDB entry 2B3B [28]. The mode of binding. studied of these receptors is the so-called glucose ⁄ galactose-binding protein. Here, we report the X-ray structure of the Salmonella typhimurium protein bound