Sugar and signal-transducer binding sites of the Escherichia coli galactose chemoreceptor protein.

D-galactose-binding (or chemoreceptor) protein of Escherichia coli serves as an initial component for both chemotaxis towards galactose and glucose and high-affinity active transport of the two sugars. Well-refined x-ray structures of the liganded forms of the wild-type and a mutant protein isolated from a strain defective in chemotaxis but fully competent in transport have provided a molecular view of the sugar-binding site and of a site for interacting with the Trg transmembrane signal transducer. The geometry of the sugar-binding site, located in the cleft between the two lobes of the bilobate protein, is novel in that it is designed for tight binding and sequestering of either the alpha or beta anomer of the D-stereoisomer of the 4-epimers galactose and glucose. Binding specificity and affinity are conferred primarily by polar planar side-chain residues that form intricate networks of cooperative and bidentate hydrogen bonds with the sugar substrates, and secondarily by aromatic residues that sandwich the pyranose ring. Each of the pairs of anomeric hydroxyls and epimeric hydroxyls is recognized by a distinct Asp residue. The site for interaction with the transducer is about 18 A from the sugar-binding site. Mutation of Gly74 to Asp at this site, concomitant with considerable changes in the local ordered water structures, contributes to the lack of productive interaction with the transmembrane signal transducer.

Preliminary crystallographic analysis of the Fab fragment of an antibody against HIV gp120.

Single crystals of the Fab fragment of a murine monoclonal antibody BAT123 (IgG1, kappa) raised against a dominant neutralizing determinant of gp120 of HIV that are suitable for X-ray structural analysis have been obtained. The thick prismatic plate crystals belong to space group P2(1)2(1)2 with unit cell dimensions of a = 177.42 A, b = 37.36 A and c = 73.30 A.

Preliminary crystallographic analysis of a Fab specific for the O-antigen of Shigella flexneri cell surface lipopolysaccharide with and without bound saccharides.

The Fab of a monoclonal anti-carbohydrate antibody, SYA/J6 (IgG3, kappa, murine), raised against the O-polysaccharide antigen of the cell surface lipopolysaccharide of variant Y Shigella flexneri, a Gram negative bacterium, has been crystallized in the unliganded form and in complex with tri- and pentasaccharide antigens. The three crystal forms belong to the tetragonal space group P4(3)2(1)2, or P4(1)2(1)2, with very similar unit cell dimensions and an asymmetric unit that contains one molecule of about 50,000 Daltons, and a fourth crystal form belongs to monoclinic space group P2(1) that contains four molecules of Fab in an asymmetric unit. Whereas diffractions of these crystals on an area detector-rotating anode system extend to only about 3.5 A resolution, those measured using an imaging plate and synchrotron radiation at the Photon Factory facility extend to 2.5 A.

Crystallization and preliminary X-ray crystallographic analysis of the 38-kDa immunodominant antigen of Mycobacterium tuberculosis.

The 38-kDa lipoprotein is one of the most potent cell surface immunogens of Mycobacterium tuberculosis in antibody-and T cell-mediated reactions. Using a pure recombinant form of the protein, we have recently shown that it binds phosphate much like that of the phosphate-binding protein (M(r) = 34.4 kDa) that is localized in the periplasm of Escherichia coli and is involved as an initial receptor for active transport of phosphate. The purified 38-kDa protein has been crystallized in 2 forms that are suitable for high-resolution structural analyses. One form belongs to the monoclinic space group P2(1) with unit cell dimensions of a = 67.42 A, b = 113.38 A, c = 42.68 A, and beta = 108.53 degrees. The other is of orthorhombic space group P2(1)2(1)2 with a = 125.46 A, b = 72.27 A, and c = 73.43 A. Both crystal forms diffract to about 2 A resolution on a fine focus rotating anode.

Novel stereospecificity of the L-arabinose-binding protein.

Tertiary structure refinement at 1.7 A resolution of the liganded form of L-arabinose-binding protein from Escherichia coli has revealed a novel binding site geometry which accommodates both alpha- and beta-anomers of L-arabinose. This detailed structure analysis provides new understanding of protein-sugar interaction, the process by which the binding protein minimizes the difference in the stability of the two bound sugar anomers, and the roles of periplasmic binding proteins in active transport.

Stabilization of charges on isolated ionic groups sequestered in proteins by polarized peptide units.

Electrostatic interactions are of considerable importance in protein structure and function, and in a variety of cellular and biochemical processes. Here we report three similar findings from highly refined atomic structures of periplasmic binding proteins. Hydrogen bonds, acting primarily through backbone peptide units, are mainly responsible for the involvement of the positively charged arginine 151 residue in the ligand site of the arabinose-binding protein, for the association between teh sulphate-binding protein and the completely buried sulphate dianion, and for the formation of the complex of the leucine/isoleucine/valine-binding protein with the leucine zwitterion. We propose a general mechanism in which the isolated charges on the various buried, desolvated ionic groups are stabilized by the polarized peptide units. This mechanism also has broad application to processes requiring binding of uncompensated ions and charged ligands and stabilization of enzyme reaction charged intermediates, as well as activation of catalytic residues.

A novel calcium binding site in the galactose-binding protein of bacterial transport and chemotaxis.

The refined 1.9-A resolution structure of the periplasmic D-galactose-binding protein (GBP) reveals a calcium ion surrounded by seven ligands, all protein oxygen atoms. A nine-residue loop (amino-acid positions 134-142), which is preceded by a beta-turn and followed by a beta-strand, provides five ligands from every second residue. The last two ligands are supplied by the carboxylate group of Glu 205. The entire GBP Ca2+-binding site adopts a conformation very similar to the site in the 'helix-loop-helix' or 'EF-hand' unit commonly found in intracellular calcium-binding proteins, but without the two helices. Structural analyses have also uncovered the sugar-binding site some 30 A from the calcium and a site for interacting with the membrane-bound trg chemotactic signal transducer approximately 45 A from the calcium. Our results show that a common tight calcium binding site of ancient origin can be tethered to different secondary structures. They also provide the first demonstration of a metal-binding site in a protein which is involved in bacterial active transport and chemotaxis.

Substrate specificity and affinity of a protein modulated by bound water molecules.

Water molecules influence molecular interactions in all biological systems, yet it is extremely difficult to understand their effects in precise atomic detail. Here we present evidence, based on highly refined atomic structures of the complexes of the L-arabinose-binding protein with L-arabinose, D-fucose and D-galactose, that bound water molecules, coupled with localized conformational changes, can govern substrate specificity and affinity. The atoms common to the three sugars are identically positioned in the binding site and the same nine strong hydrogen bonds are formed in all three complexes. Two hydrogen-bonded water molecules in the site contribute further to tight binding of L-arabinose but create an unfavourable interaction with the methyl group of D-fucose. Equally tight binding of D-galactose is attained by the replacement of one of the hydrogen-bonded water molecules by its--CH2OH group, coordinated with localized structural changes which include a shift and redirection of the hydrogen-bonding interactions of the other water molecule. These observations illustrate how ordered water molecules can contribute directly to the properties of proteins by influencing their interaction with ligands.

Comparison of the periplasmic receptors for L-arabinose, D-glucose/D-galactose, and D-ribose. Structural and Functional Similarity.

The primary sequence of the receptor for L-arabinose or Ara-binding protein (ABP) composed of 306 residues is very different from the D-glucose/D-galactose-binding protein (GGBP) which consists of 309 residues. Nevertheless, superimpositioning of the well-refined high resolution structures of ABP in complex with D-galactose and the GGBP in complex with D-glucose shows very similar structures; 220 of the residues (or about 70%) have a root mean square deviation of 2.0 A. From the superpositioning, nine pairs of continuous segments (consisting of 8-51 residues), mainly alpha-helices and beta-strands that form the core of the two lobes of the bilobate proteins were found to exhibit strong sequence homology. The equivalenced structures and aligned sequences show that many of the polar, as well as aromatic residues, in the sugar-binding sites located in the cleft between the two lobes are highly conserved. Surprisingly, however, the exact mode of binding of the D-galactose in ABP is totally different from that of the D-glucose in GGBP. Using the structurally aligned sequences of the ABP and GGBP as a template, we have matched the sequence of the ribose-binding protein (RBP) which consists of 271 residues with the ABP/GGBP pair. Although the nine aligned segments of all three proteins show little sequence identity, they have significant homology. Four additional segments of RBP were matched only with GGBP, leading to the alignment of about 90% of the RBP sequence with the GGBP sequence. Many of the conserved residues in the binding sites of ABP and GGBP matched with similar residues in RBP. Additional observations indicate that the GGBP/RBP pair is more closely related than the ABP/RBP or ABP/GGBP pair. All three binding proteins, which may have diverged from a common ancestor, serve as primary receptors for bacterial high affinity active transport systems. Moreover, GGBP and RBP, but not ABP, also act as receptors for chemotaxis. An exposed site located in one domain, which includes Gly74, for interacting with the trg transmembrane signal transducer that is involved in triggering chemotaxis has been located in the structure of GGBP (Vyas, N.K., Vyas, M.N., and Quiocho, F.A. (1988) Science 242, 1290-1295). Whereas the site is absent in the structure of ABP, it is strongly predicted to be present in RBP which shares the same trg transducer with GGBP. The knowledge-based alignment of RBP further revealed two possible additional peripheral chemotactic sites that show high structural and sequence similarity between GGBP and RBP only. At least one of these sites, together with the one proven to exist in the other domain, could be used by the signal transducer with which both binding proteins interact in a way which the substrate-loaded "closed cleft" structure could be discriminated from the unliganded "open cleft" form by the transducer.

Crystallographic analysis of the epimeric and anomeric specificity of the periplasmic transport/chemosensory protein receptor for D-glucose and D-galactose.

The D-glucose/D-galactose-binding protein (M(r) = 33,000) found in the periplasm of bacterial cells serves as the primary high-affinity receptor of active transport for and chemotaxis toward both sugar epimers. This protein from Escherichia coli binds D-glucose with a Kd of 2 x 10(-7) M, which is about 2 times tighter than D-galactose. The 2.0-A resolution crystal structure of the binding protein complexed with D-galactose has been refined to a crystallographic R-factor of 0.167. This structure, combined with that previously refined for the complex with D-glucose [Vyas, N.K., Vyas., M. N., & Quiocho, F. A. (1988) Science 242, 1290-1295], provides understanding, in atomic detail, of recognition of sugar epimers and anomers. In the two complex structures, the sugar ring is positioned identically in the binding site, and each hydroxyl group common to both is involved in very similar cooperative hydrogen-bonding interactions with protein residues and ordered water molecules. Only the beta-anomer of both monosaccharides is bound, with Asp154 OD1 primarily responsible for accepting a hydrogen bond from the anomeric hydroxyl. Recognition of both sugar epimers is accomplished principally by hydrogen bonding of Asp14 OD1 with the equatorial OH4 of D-glucose and OD2 with the axial OH4 of D-galactose. These results are reconciled with equilibrium and fast kinetics data, which indicate binding of both anomers of the two sugars, and further compared with sugar recognition by other periplasmic sugar-binding proteins with specificities for arabinose/galactose/fucose, maltooligosaccharides, and ribose.

The 3 A resolution structure of a D-galactose-binding protein for transport and chemotaxis in Escherichia coli.

X-ray diffraction studies of a D-galactose-binding protein essential for transport and chemotaxis in Escherichia coli have yielded a model of the polypeptide chain backbone. An initial polyalanine backbone trace was obtained at 3.2 A resolution by the molecular replacement technique, using a polyalanine search model derived from the refined structure of the L-arabinose-binding protein. Concurrently, a 3 A resolution electron-density map of the D-galactose receptor was determined from multiple isomorphous replacement (MIR) phases. The properly transformed initial polyalanine model superimposed on the MIR electron-density map proved to be an excellent guide in obtaining a final trace. The few changes made in the polyalanine model to improve the fit to the density were confined primarily to the COOH-terminal peptide and some loops connecting the elements of the secondary structure. Despite the lack of significant sequence homology, the overall course of the polypeptide backbone of the D-galactose-binding protein is remarkably similar to that of the L-arabinose-binding protein, the first structure in a series to be solved from this family of binding proteins. Both structures are elongated (axial ratios of 2:1) and composed of two globular domains. For both proteins, the arrangements of the elements of the secondary structure in both domains are identical; both lobes contain a core of beta-pleated sheet with a pair of helices on either side of the plane of the sheet. The four major hydrophobic clusters that stabilize the structure of the L-arabinose-binding protein are also present in the D-galactose-binding protein.

Model of lactose repressor core based on alignment with sugar-binding proteins is concordant with genetic and chemical data.

Using primary sequence similarity to arabinose-binding protein, D-glucose/D-galactose-binding protein, and ribose-binding protein (Vyas, N. K., Vyas, M. N., and Quiocho, F. A. (1991) J. Biol. Chem. 266, 5226-5237; Mowbray, S. L., and Cole, L. B. (1992) J. Mol. Biol. 225, 155-175), the core domain (residues 62-323) of the bacterial regulatory protein lac repressor has been aligned to these sugar-binding proteins of known structure. Although the sequence identity is not striking, there is strong overall homology based on two separate matrix scoring systems (minimum base change per codon (MBC/C) and amino acid homology per residue (AAH/R)) (mean score: MBC/C < 1.25, AAH/R > 5.50; random sequences: MBC/C = 1.45, AAH/R = 4.46). Similarly, the predicted secondary structure of the repressor exhibits excellent agreement with the known secondary structures of the sugar-binding proteins. Using this primary sequence alignment, the tertiary structure of the core domain of the lac repressor has been modeled based on the known structures of the sugar-binding proteins as templates. While the structure deduced for the repressor is hypothetical, the model generated allows a comparison between the predicted tertiary arrangement and the wealth of genetic and chemical data elucidated for the repressor. Important residues involved in operator and sugar binding and in protein assembly have been identified using genetic methods, and placement of these residues in the model is consistent with their known function. This approach, therefore, provides a means to visualize the core domain of the lac repressor that allows interpretation of genetic and chemical data for specific residues and rational design of future experiments.

Reporter molecules as probes of DNA conformation: structure of a crystalline complex containing 2-methyl-4-nitroaniline ethylene dimethylammonium hydrobromide--5-iodocytidylyl (3'-5')guanosine.

2-Methyl-4-nitroaniline ethylene dimethylammonium hydrobromide forms a crystalline complex with the self-complementary dinucleoside monophosphate, 5- iodocytidylyl (3'-5')guanosine. The crystals are tetragonal, with a = b = 32.192 A and c = 23.964 A, space group P4(3)2(1)2. The structure has been solved to atomic resolution by Patterson and Fourier methods, and refined by full matrix least squares. 5- Iodocytidylyl (3'-5')guanosine molecules are held together in pairs through Watson-Crick base-pairing, forming an antiparallel duplex structure. Nitroaniline molecules stack above and below guanine-cytosine pairs in this duplex structure. In addition, a third nitroaniline molecule stacks on one of the other two nitroaniline molecules. The asymmetric unit contains two 5- iodocytidylyl (3'-5')guanosine molecules, three nitroaniline molecules, one bromide ion and thirty-one water molecules, a total of 160 atoms. Details of the structure are described.

Predicted structure of the sugar-binding site of the lac repressor.

The lactose repressor protein from Escherichia coli binds sugars, primarily galactosides, which modulate its interactions with operator DNA and thereby affect synthesis of the lac metabolic enzymes. The affinity of the repressor for operator DNA is decreased by binding inducer sugars and increased by binding anti-inducer sugars. Based on regions of the primary structure implicated by genetic methods to be involved in sugar binding, amino acid sequence homology between L-arabinose-binding protein (ABP) and lac repressor has recently been reported. The sugar-binding sites for these two proteins might be expected to have similar structural features, as both bind L-arabinose and D-galactose. The high resolution structure of ABP reported in the accompanying article provides complete definition of amino acids in the sugar-binding site. By identification of homologous residues in the lac repressor, we have now predicted the structure of the portion of the repressor sugar-binding site which accommodates the galactosyl moiety. This prediction provides the first potential view of the inducer/anti-inducer site in the repressor protein.