The structure of the T127L/S128A mutant of cAMP receptor protein facilitates promoter site binding.

The x-ray crystal structure of the cAMP-ligated T127L/S128A double mutant of cAMP receptor protein (CRP) was determined to a resolution of 2.2 A. Although this structure is close to that of the x-ray crystal structure of cAMP-ligated CRP with one subunit in the open form and one subunit in the closed form, a bound syn-cAMP is clearly observed in the closed subunit in a third binding site in the C-terminal domain. In addition, water-mediated interactions replace the hydrogen bonding interactions between the N(6) of anti-cAMP bound in the N-terminal domains of each subunit and the OH groups of the Thr(127) and Ser(128) residues in the C alpha-helix of wild type CRP. This replacement induces flexibility in the C alpha-helix at Ala(128), which swings the C-terminal domain of the open subunit more toward the N-terminal domain in the T127L/S128A double mutant of CRP (CRP*) than is observed in the open subunit of cAMP-ligated CRP. Isothermal titration calorimetry measurements on the binding of cAMP to CRP* show that the binding mechanism changes from an exothermic independent two-site binding mechanism at pH 7.0 to an endothermic interacting two-site mechanism at pH 5.2, similar to that observed for CRP at both pH levels. Differential scanning calorimetry measurements exhibit a broadening of the thermal denaturation transition of CRP* relative to that of CRP at pH 7.0 but similar to the multipeak transitions observed for cAMP-ligated CRP. These properties and the bound syn-cAMP ligand, which has only been previously observed in the DNA bound x-ray crystal structure of cAMP-ligated CRP by Passner and Steitz (Passner, J. M., and Steitz, T. A. (1997) Proc. Natl. Acad. Sci. U. S. A. 94, 2843-2847), imply that the cAMP-ligated CRP* structure is closer to the conformation of the allosterically activated structure than cAMP-ligated CRP. This may be induced by the unique flexibility at Ala(128) and/or by the bound syn-cAMP in the hinge region of CRP*.

Stressing-out DNA? The contribution of serine-phosphodiester interactions in catalysis by uracil DNA glycosylase.

The DNA repair enzyme uracil DNA glycosylase (UDG) pinches the phosphodiester backbone of damaged DNA using the hydroxyl side chains of a conserved trio of serine residues, resulting in flipping of the deoxyuridine from the DNA helix into the enzyme active site. We have investigated the energetic role of these serine-phosphodiester interactions using the complementary approaches of crystallography, directed mutagenesis, and stereospecific phosphorothioate substitutions. A new crystal structure of UDG bound to 5'-HO-dUAAp-3' (which lacks the 5' phosphodiester group that interacts with the Ser88 pinching finger) shows that the glycosidic bond of dU has been cleaved, and that the enzyme has undergone the same specific clamping motion that brings key active site groups into position as previously observed in the structures of human UDG bound to large duplex DNA substrates. From this structure, it may be concluded that glycosidic bond cleavage and the induced fit conformational change in UDG can occur without the 5' pinching interaction. The S88A, S189A, and S192G "pinching" mutations exhibit 360-, 80-, and 21-fold damaging effects on k(cat)/K(m), respectively, while the S88A/S189A double mutant exhibits an 8200-fold damaging effect. A free energy analysis of the combined effects of nonbridging phosphorothioate substitution and mutation at these positions reveals the presence of a modest amount of strain energy between the compressed 5' and 3' phosphodiester groups flanking the bound uridine. Overall, these results indicate a role for these serine-phosphodiester interactions in uracil flipping and preorganization of the sugar ring into a reactive conformation. However, in contrast to a recent proposal [Parikh, S. S., et al. (2000) Proc Natl. Acad. Sci. 94, 5083], there is no evidence that conformational strain of the glycosidic bond induced by serine pinching plays a major role in the 10(12)-fold rate enhancement brought about by UDG.

Cryosalts: suppression of ice formation in macromolecular crystallography.

Quality data collection for macromolecular cryocrystallography requires suppressing the formation of crystalline or microcrystalline ice that may result from flash-freezing crystals. Described here is the use of lithium formate, lithium chloride and other highly soluble salts for forming ice-ring-free aqueous glasses upon cooling from ambient temperature to 100 K. These cryosalts are a new class of cryoprotectants that are shown to be effective with a variety of commonly used crystallization solutions and with proteins crystallized under different conditions. The influence of cryosalts on crystal mosaicity and diffraction resolution is comparable with or superior to traditional organic cryoprotectants.

The 1.30 A resolution structure of the Bacillus subtilis chorismate mutase catalytic homotrimer.

The crystal structure of the Bacillus subtilis chorismate mutase, an enzyme of the aromatic amino acids biosynthetic pathway, was determined to 1.30 A resolution. The structure of the homotrimer was determined by molecular replacement using orthorhombic crystals of space group P2(1)2(1)2(1) with unit-cell parameters a = 52.2, b = 83. 8, c = 86.0 A. The ABC trimer of the monoclinic crystal structure [Chook et al. (1994), J. Mol. Biol. 240, 476-500] was used as the starting model. The final coordinates are composed of three complete polypeptide chains of 127 amino-acid residues. In addition, there are nine sulfate ions, five glycerol molecules and 424 water molecules clearly visible in the structure. This structure was refined with aniosotropic temperature factors, has excellent geometry and a crystallographic R factor of 0.169 with an R(free) of 0.236. The three active sites of the macromolecule are at the subunit interfaces, with residues from two subunits contributing to each site. This orthorhombic crystal form was grown using ammonium sulfate as the precipitant; glycerol was used as a cryoprotectant during data collection. A glycerol molecule and sulfate ion in each of the active sites was found mimicking a transition-state analog. In this structure, the C-terminal tails of the subunits of the trimer are hydrogen bonded to residues of the active site of neighboring trimers in the crystal and thus cross-link the molecules in the crystal lattice.

Heteronuclear NMR and crystallographic studies of wild-type and H187Q Escherichia coli uracil DNA glycosylase: electrophilic catalysis of uracil expulsion by a neutral histidine 187.

The nature of the putative general acid His187 in the reaction catalyzed by Escherichia coli uracil DNA glycosylase (UDG) was investigated using X-ray crystallography and NMR spectroscopy. The crystal structures of H187Q UDG, and its complex with uracil, have been solved at 1.40 and 1.60 A resolution, respectively. The structures are essentially identical to those of the wild-type enzyme, except that the side chain of Gln187 is turned away from the uracil base and cannot interact with uracil O2. This result provides a structural basis for the similar kinetic properties of the H187Q and H187A enzymes. The ionization state of His187 was directly addressed with (1)H-(15)N NMR experiments optimized for histidine ring spin systems, which established that His187 is neutral in the catalytically active state of the enzyme (pK(a) <5.5). These NMR experiments also show that His187 is held in the N(epsilon)()2-H tautomeric form, consistent with the crystallographic observation of a 2.9 A hydrogen bond from the backbone nitrogen of Ser189 to the ring N(delta)()1 of His187. The energetic cost of breaking this hydrogen bond may contribute significantly to the low pK(a) of His187. Thus, the traditional view that a cationic His187 donates a proton to uracil O2 is incorrect. Rather, we propose a concerted mechanism involving general base catalysis by Asp64 and electrophilic stabilization of the developing enolate on uracil O2 by a neutral His187.

The three-dimensional structures of two isoforms of nucleoside diphosphate kinase from bovine retina.

The crystal structures of two isoforms of nucleoside diphosphate kinase from bovine retina overexpressed in Escherischia coli have been determined to 2.4 A resolution. Both the isoforms, NBR-A and NBR-B, are hexameric and the fold of the monomer is in agreement with NDP-kinase structures from other biological sources. Although the polypeptide chains of the two isoforms differ by only two residues, they crystallize in different space groups. NBR-A crystallizes in space group P212121 with an entire hexamer in the asymmetric unit, while NBR-B crystallizes in space group P43212 with a trimer in the asymmetric unit. The highly conserved nucleotide-binding site observed in other nucleoside diphosphate kinase structures is also observed here. Both NBR-A and NBR-B were crystallized in the presence of cGMP. The nucleotide is bound with the base in the anti conformation. The NBR-A active site contained both cGMP and GDP each bound at half occupancy. Presumably, NBR-A had retained GDP (or GTP) from the purification process. The NBR-B active site contained only cGMP.

Crystal structure of Escherichia coli uracil DNA glycosylase and its complexes with uracil and glycerol: structure and glycosylase mechanism revisited.

The DNA repair enzyme uracil DNA glycosylase (UDG) catalyzes the hydrolysis of premutagenic uracil residues from single-stranded or duplex DNA, producing free uracil and abasic DNA. Here we report the high-resolution crystal structures of free UDG from Escherichia coli strain B (1.60 A), its complex with uracil (1.50 A), and a second active-site complex with glycerol (1.43 A). These represent the first high-resolution structures of a prokaryotic UDG to be reported. The overall structure of the E. coli enzyme is more similar to the human UDG than the herpes virus enzyme. Significant differences between the bacterial and viral structures are seen in the side-chain positions of the putative general-acid (His187) and base (Asp64), similar to differences previously observed between the viral and human enzymes. In general, the active-site loop that contains His187 appears preorganized in comparison with the viral and human enzymes, requiring smaller substrate-induced conformational changes to bring active-site groups into catalytic position. These structural differences may be related to the large differences in the mechanism of uracil recognition used by the E. coli and viral enzymes. The pH dependence of k(cat) for wild-type UDG and the D64N and H187Q mutant enzymes is consistent with general-base catalysis by Asp64, but provides no evidence for a general-acid catalyst. The catalytic mechanism of UDG is critically discussed with respect to these results.

Nucleoside diphosphate kinase from bovine retina: purification, subcellular localization, molecular cloning, and three-dimensional structure.

The biochemical and structural properties of bovine retinal nucleoside diphosphate kinase were investigated. The enzyme showed two polypeptides of approximately 17.5 and 18.5 kDa on SDS-PAGE, while isoelectric focusing revealed seven to eight proteins with a pI range of 7.4-8.2. Sedimentation equilibrium yielded a molecular mass of 96 +/- 2 kDa for the enzyme. Carbohydrate analysis revealed that both polypeptides contained Gal, Man, GlcNAc, Fuc, and GalNac saccharides. Like other nucleoside diphosphate kinases, the retinal enzyme showed substantial differences in the Km values for various di- and triphosphate nucleotides. Immunogold labeling of bovine retina revealed that the enzyme is localized on both the membranes and in the cytoplasm. Screening of a retinal cDNA library yielded full-length clones encoding two distinct isoforms (NBR-A and NBR-B). Both isoforms were overexpressed in Escherichia coli and their biochemical properties compared with retinal NDP-kinase. The structures of NBR-A and NBR-B were determined by X-ray crystallography in the presence of guanine nucleotide(s). Both isoforms are hexameric, and the fold of the monomer is similar to other nucleoside diphosphate kinase structures. The NBR-A active site contained both a cGMP and a GDP molecule each bound at half occupancy while the NBR-B active site contained only cGMP.

Crystal structure of apo-cellular retinoic acid-binding protein type II (R111M) suggests a mechanism of ligand entry.

The crystal structure of unliganded mutant R111M of human cellular retinoic acid-binding protein type II (apo-CRABPII (R111M)) has been determined at 2.3 A and refined to a crystallographic R-factor of 0. 18. Although the mutant protein has lower affinity for all-trans-retinoic acid (RA) than the wild-type, it is properly folded, and its conformation is very similar to the wild-type. apo-CRABPII (R111M) crystallizes in space group P1 with two molecules in the unit cell. The two molecules have high structural similarity except that their alpha2 helices differ strikingly. Analyses of the molecular conformation and crystal packing environment suggest that one of the two molecules assumes a conformation compatible with RA entry. Three structural elements encompassing the opening of the binding pocket exhibit large conformational changes, when compared with holo-CRABPII, which include the alpha2 helix and the betaC-betaD and betaE-betaF hairpin loops. The alpha2 helix is unwound at its N terminus, which appears to be essential for the opening of the RA-binding pocket. Three arginine side-chains (29, 59, and 132) are found with their guanidino groups pointing into the RA-binding pocket. A three-step mechanism of RA entry has been proposed, addressing the opening of the RA entrance, the electrostatic potential that directs entry of RA into the binding pocket, and the intramolecular interactions that stabilize the RA.CRABPII complex via locking the three flexible structural elements when RA is bound.

Primary and tertiary structures of the Fab fragment of a monoclonal anti-E-selectin 7A9 antibody that inhibits neutrophil attachment to endothelial cells.

The murine monoclonal IgG1 antibody 7A9 binds specifically to the endothelial leukocyte adhesion molecule-1 (E-selectin), inhibiting the attachment of neutrophils to endothelial cells. The primary and three-dimensional structures of the Fab fragment of 7A9 are reported. The amino acid sequence was determined by automated Edman degradation analysis of proteolytic fragments of both the heavy and light chains of the Fab. The sequences of the two chains are consistent with that of the IgG1 class with an associated kappa light chain with two intrachain disulfide bridges in each of the heavy and light chains. The tertiary structure of the antibody fragment was determined by x-ray crystallographic methods at 2.8 A resolution. The F(ab')2 molecule, treated with dithiothreitol, crystallizes in the space group P2(1) 2(1) 2(1) with unit cell parameters a = 44.5 A, b = 83.8 A, and c = 132.5 A with one Fab molecule in the asymmetric unit. The structure was solved by the molecular replacement method and subsequently refined using simulated annealing followed by conventional least squares optimization of the coordinates. The resulting model has reasonable stereochemistry with an R factor of 0.195. The 7A9 Fab structure has an elbow bend of 162 degrees and is remarkably similar to that of the monoclonal anti-intercellular adhesion molecule-1 (ICAM-1) antibody Fab fragment. The 7A9 antigen combining site presents a groove resembling the structure of the anti-ICAM-1 antibody, and other antibodies raised against surface receptors and peptides. Residues from the six complementary determining regions (CDRs) and framework residues form the floor and walls of the groove that is approximately 22 A wide and 8 A deep and that is lined with many aromatic residues. The groove is large enough to accommodate the loop between beta-strands beta4 and beta5 of the lectin domain of E-selectin that has been implicated in neutrophil adhesion (1).

Crystal structure of the disulfide-stabilized Fv fragment of anticancer antibody B1: conformational influence of an engineered disulfide bond.

A recombinant Fv construct of the B1 monoclonal antibody that recognizes the LewisY-related carbohydrate epitope on human carcinoma cells has been prepared. The Fv is composed of the polypeptide chains of the VH and VL domains expressed independently and isolated as inclusion bodies. The Fv is prepared by combining and refolding equimolar amounts of guanidine chloride solubilized inclusion bodies. The Fv is stabilized by an engineered interchain disulfide bridge between residues VL100 and VH44. This construct has a similar binding affinity as that of the single-chain construct (Benhar and Pastan, Clin. Cancer Res. 1:1023-1029, 1995). The B1 disulfide-stabilized Fv (BldsFv) crystallizes in space group P6(1)22 with the unit cell parameters a = b = 80.1 A, and c = 138.1 A. The crystal structure of the BldsFv has been determined at 2.1-A resolution using the molecular replacement technique. The final structure has a crystallographic R-value of 0.187 with a root mean square deviation in bond distance of 0.014 A and in bond angle of 2.74 degrees. Comparisons of the BldsFv structure with known structures of Fv regions of other immunoglobulin fragments shows closely related secondary and tertiary structures. The antigen combining site of BldsFv is a deep depression 10-A wide and 17-A long with the walls of the depression composed of residues, many of which are tyrosines, from complementarity determining regions L1, L3, H1, H2, and H3. Model building studies indicate that the LewisY tetrasaccharide, Fuc-Gal-Nag-Fuc, can be accommodated in the antigen combining site in a manner consistent with the epitope predicted in earlier biochemical studies (Pastan, Lovelace, Gallo, Rutherford, Magnani, and Willingham, Cancer Res. 51:3781-3787, 1991). Thus, the engineered disulfide bridge appears to cause little, if any, distortion in the Fv structure, making it an effective substitute for the B1 Fab.

Crystal structure of calcium-independent subtilisin BPN' with restored thermal stability folded without the prodomain.

The three-dimensional structure of a subtilisin BPN' construct that was produced and folded without its prodomain shows the tertiary structure is nearly identical to the wild-type enzyme and not a folding intermediate. The subtilisin BPN' variant, Sbt70, was cloned and expressed in Escherichia coli without the prodomain, the 77-residue N-terminal domain that catalyzes the folding of the enzyme into its native tertiary structure. Sbt70 has the high-affinity calcium-binding loop, residues 75 to 83, deleted. Such calcium-independent forms of subtilisin BPN' refold independently while retaining high levels of activity [Bryan et al., Biochemistry, 31:4937-4945, 1992]. Sbt70 has, in addition, seven stabilizing mutations, K43N, M50F, A73L, Q206V, Y217K, N218S, Q271E, and the active site serine has been replaced with alanine to prevent autolysis. The purified Sbt70 folded spontaneously without the prodomain and crystallized at room temperature. Crystals of Sbt70 belong to space group P2(1)2(1)2(1) with unit cell parameters a = 53.5 A, b = 60.3 A, and c = 83.4 A. Comparison of the refined structure with other high-resolution structures of subtilisin BPN' establishes that the conformation of Sbt70 is essentially the same as that previously determined for other calcium-independent forms and that of other wild-type subtilisin BPN' structures, all folded in the presence of the prodomain. These findings confirm the results of previous solution studies that showed subtilisin BPN' can be refolded into a native conformation without the presence of the prodomain [Bryan et al., Biochemistry 31:4937-4945, 1992]. The structure analysis also provides the first descriptions of four stabilizing mutations, K43N, A73L, Q206V, and Q271E, and provides details of the interaction between the enzyme and the Ala-Leu-Ala-Leu tetrapeptide found in the active-site cleft.

Structure and function of the xenobiotic substrate-binding site and location of a potential non-substrate-binding site in a class pi glutathione S-transferase.

Complex structures of a naturally occurring variant of human class pi glutathione S-transferase 1-1 (hGSTP1-1) with either S-hexylglutathione or (9R,10R)-9-(S-glutathionyl)-10-hydroxy-9, 10-dihydrophenanthrene [(9R,10R)-GSPhen] have been determined at resolutions of 1.8 and 1.9 A, respectively. The crystal structures reveal that the xenobiotic substrate-binding site (H-site) is located at a position similar to that observed in class mu GST 1-1 from rat liver (rGSTM1-1). In rGSTM1-1, the H-site is a hydrophobic cavity defined by the side chains of Y6, W7, V9, L12, I111, Y115, F208, and S209. In hGSTP1-1, the cavity is approximately half hydrophobic and half hydrophilic and is defined by the side chains of Y7, F8, V10, R13, V104, Y108, N204, and G205 and five water molecules. A hydrogen bond network connects the five water molecules and the side chains of R13 and N204. V104 is positioned such that the introduction of a methyl group (the result of the V104I mutation) disturbs the H-site water structure and alters the substrate-binding properties of the isozyme. The hydroxyl group of Y7 forms a hydrogen bond (3.2 A) with the sulfur atom of the product. There is a short hydrogen bond (2.5 A) between Y108 (OH) and (9R, 10R)-GSPhen (O5), indicating the hydroxyl group of Y108 as an electrophilic participant in the addition of glutathione to epoxides. An N-(2-hydroxethyl)piperazine-N'-2-ethanesulfonic acid (HEPES) molecule is found in the cavity between beta2 and alphaI. The location and properties of this HEPES-binding site fit a possible non-substrate-binding site that is involved in noncompetitive inhibition of the enzyme.