Structures and mechanisms of Nudix hydrolases.

Nudix hydrolases catalyze the hydrolysis of nucleoside diphosphates linked to other moieties, X, and contain the sequence motif or Nudix box, GX(5)EX(7)REUXEEXGU. The mechanisms of Nudix hydrolases are highly diverse in the position on the substrate at which nucleophilic substitution occurs, and in the number of required divalent cations. While most proceed by associative nucleophilic substitutions by water at specific internal phosphorus atoms of a diphosphate or polyphosphate chain, members of the GDP-mannose hydrolase sub-family catalyze dissociative nucleophilic substitutions, by water, at carbon. The site of substitution is likely determined by the positions of the general base and the entering water. The rate accelerations or catalytic powers of Nudix hydrolases range from 10(9)- to 10(12)-fold. The reactions are accelerated 10(3)-10(5)-fold by general base catalysis by a glutamate residue within, or beyond the Nudix box, or by a histidine beyond the Nudix box. Lewis acid catalysis, which contributes 10(3)-10(5)-fold to the rate acceleration, is provided by one, two, or three divalent cations. One divalent cation is coordinated by two or three conserved residues of the Nudix box, the initial glycine and one or two glutamate residues, together with a remote glutamate or glutamine ligand from beyond the Nudix box. Some Nudix enzymes require one (MutT) or two additional divalent cations (Ap(4)AP), to neutralize the charge of the polyphosphate chain, to help orient the attacking hydroxide or oxide nucleophile, and/or to facilitate the departure of the anionic leaving group. Additional catalysis (10-10(3)-fold) is provided by the cationic side chains of lysine and arginine residues and by H-bond donation by tyrosine residues, to orient the general base, or to promote the departure of the leaving group. The overall rate accelerations can be explained by both independent and cooperative effects of these catalytic components.

Mechanistic implications of methylglyoxal synthase complexed with phosphoglycolohydroxamic acid as observed by X-ray crystallography and NMR spectroscopy.

Methylglyoxal synthase (MGS) and triosephosphate isomerase (TIM) share neither sequence nor structural similarities, yet the reactions catalyzed by both enzymes are similar, in that both initially convert dihydroxyacetone phosphate to a cis-enediolic intermediate. This enediolic intermediate is formed from the abstraction of the pro-S C3 proton of DHAP by Asp-71 of MGS or the pro-R C3 proton of DHAP by Glu-165 of TIM. MGS then catalyzes the elimination of phosphate from this enediolic intermediate to form the enol of methylglyoxal, while TIM catalyzes proton donation to C2 to form D-glyceraldehyde phosphate. A competitive inhibitor of TIM, phosphoglycolohydroxamic acid (PGH) is found to be a tight binding competitive inhibitor of MGS with a K(i) of 39 nM. PGH's high affinity for MGS may be due in part to a short, strong hydrogen bond (SSHB) from the NOH of PGH to the carboxylate of Asp-71. Evidence for this SSHB is found in X-ray, 1H NMR, and fractionation factor data. The X-ray structure of the MGS homohexamer complexed with PGH at 2.0 A resolution shows this distance to be 2.30-2.37 +/- 0.24 A. 1H NMR shows a PGH-dependent 18.1 ppm signal that is consistent with a hydrogen bond length of 2.49 +/- 0.02 A. The D/H fractionation factor (phi = 0.43 +/- 0.02) is consistent with a hydrogen bond length of 2.53 +/- 0.01 A. Further, 15N NMR suggests a significant partial positive charge on the nitrogen atom of bound PGH, which could strengthen hydrogen bond donation to Asp-71. Both His-98 and His-19 are uncharged in the MGS-PGH complex on the basis of the chemical shifts of their Cdelta and C(epsilon) protons. The crystal structure reveals that Asp-71, on the re face of PGH, and His-19, on the si face of PGH, both approach the NO group of the analogue, while His-98, in the plane of PGH, approaches the carbonyl oxygen of the analogue. The phosphate group of PGH accepts nine hydrogen bonds from seven residues and is tilted out of the imidate plane of PGH toward the re face. Asp-71 and phosphate are thus positioned to function as the base and leaving group, respectively, in a concerted suprafacial 1,4-elimination of phosphate from the enediolic intermediate in the second step of the MGS reaction. Combined, these data suggest that Asp-71 is the one base that initially abstracts the C3 pro-S proton from DHAP and subsequently the 3-OH proton from the enediolic intermediate. This mechanism is compared to an alternative TIM-like mechanism for MGS, and the relative merits of both mechanisms are discussed.

The structural basis for the perturbed pKa of the catalytic base in 4-oxalocrotonate tautomerase: kinetic and structural effects of mutations of Phe-50.

The amino-terminal proline of 4-oxalocrotonate tautomerase (4-OT) functions as the general base catalyst in the enzyme-catalyzed isomerization of beta,gamma-unsaturated enones to their alpha,beta-isomers because of its unusually low pK(a) of 6.4 +/- 0.2, which is 3 units lower than that of the model compound, proline amide. Recent studies show that this abnormally low pK(a) is not due to the electrostatic effects of nearby cationic residues (Arg-11, Arg-39, and Arg-61) [Czerwinski, R. M., Harris, T. K., Johnson, Jr., W. H., Legler, P. M., Stivers, J. T., Mildvan, A. S., and Whitman, C. P. (1999) Biochemistry 38, 12358-12366]. Hence, it may result solely from a low local dielectric constant of 14.7 +/- 0.8 at the otherwise hydrophobic active site. Support for this mechanism comes from the study of mutants of the active site Phe-50, which is 5.8 A from Pro-1 and is one of 12 apolar residues within 9 A of Pro-1. Replacing Phe-50 with Tyr does not significantly alter k(cat) or K(m) and results in a pK(a) of 6.0 +/- 0.1 for Pro-1 as determined by (15)N NMR spectroscopy, comparable to that observed for wild type. (1)H-(15)N HSQC and 3D (1)H-(15)N NOESY HSQC spectra of the F50Y mutant demonstrate its conformation to be very similar to that of the wild-type enzyme. In the F50Y mutant, the pK(a) of Tyr-50 is increased by two units from that of a model compound N-acetyl-tyrosine amide to 12.2 +/- 0.3, as determined by UV and (1)H NMR titrations, yielding a local dielectric constant of 13.4 +/- 1.7, in agreement with the value of 13.7 +/- 0.3 determined from the decreased pK(a) of Pro-1 in this mutant. In the F50A mutant, the pK(a) of Pro-1 is 7.3 +/- 0.1 by (15)N NMR titration, comparable to the pK(a) of 7.6 +/- 0.2 found in the pH vs k(cat)/K(m) rate profile, and is one unit greater than that of the wild-type enzyme, indicating an increase in the local dielectric constant to a value of 21.2 +/- 2.6. A loss of structure of the beta-hairpin from residues 50 to 57, which covers the active site, and is the site of the mutation, is indicated by the disappearance in the F50A mutant of four interstrand NOEs and one turn NOE found in wild-type 4-OT. (1)H-(15)N HSQC spectra of the F50A mutant reveal widespread and large changes in the backbone (15)N and NH chemical shifts including those of Gly residues 48, 51, 53, and 54 causing their loss of dispersion at 23 degrees C and their disappearance at 43 degrees C due to rapid exchange with solvent. These observations confirm that the active site of the F50A mutant is more accessible to the external aqueous environment, causing an increase in the local dielectric constant and in the pK(a) of Pro-1. In addition, the F50A mutation decreased k(cat) 167-fold and increased K(m) 11-fold from those of the wild-type enzyme, suggesting an important role for the hydrophobic environment in catalysis, beyond that of decreasing the pK(a) of Pro-1. The F50I and F50V mutations destabilize the protein and decrease k(cat) by factors of 58 and 1.6, and increase K(m) by 3.3- and 3.8-fold, respectively.

Short, strong hydrogen bonds at the active site of human acetylcholinesterase: proton NMR studies.

Cholinesterases use a Glu-His-Ser catalytic triad to enhance the nucleophilicity of the catalytic serine. We have previously shown by proton NMR that horse serum butyryl cholinesterase, like serine proteases, forms a short, strong hydrogen bond (SSHB) between the Glu-His pair upon binding mechanism-based inhibitors, which form tetrahedral adducts, analogous to the tetrahedral intermediates in catalysis [Viragh, C., et al. (2000) Biochemistry 39, 16200-16205]. We now extend these studies to human acetylcholinesterase, a 136 kDa homodimer. The free enzyme at pH 7.5 shows a proton resonance at 14.4 ppm assigned to an imidazole NH of the active-site histidine, but no deshielded proton resonances between 15 and 21 ppm. Addition of a 3-fold excess of the mechanism-based inhibitor m-(N,N,N-trimethylammonio)trifluoroacetophenone (TMTFA) induced the complete loss of the 14.4 ppm signal and the appearance of a broad, deshielded resonance of equal intensity with a chemical shift delta of 17.8 ppm and a D/H fractionation factor phi of 0.76 +/- 0.10, consistent with a SSHB between Glu and His of the catalytic triad. From an empirical correlation of delta with hydrogen bond lengths in small crystalline compounds, the length of this SSHB is 2.62 +/- 0.02 A, in agreement with the length of 2.63 +/- 0.03 A, independently obtained from phi. Upon addition of a 3-fold excess of the mechanism-based inhibitor 4-nitrophenyl diethyl phosphate (paraoxon) to the free enzyme at pH 7.5, and subsequent deethylation, two deshielded resonances of unequal intensity appeared at 16.6 and 15.5 ppm, consistent with SSHBs with lengths of 2.63 +/- 0.02 and 2.65 +/- 0.02 A, respectively, suggesting conformational heterogeneity of the active-site histidine as a hydrogen bond donor to either Glu-327 of the catalytic triad or to Glu-199, also in the active site. Conformational heterogeneity was confirmed with the methylphosphonate ester anion adduct of the active-site serine, which showed two deshielded resonances of equal intensity at 16.5 and 15.8 ppm with phi values of 0.47 +/- 0.10 and 0.49 +/- 0.10 corresponding to average hydrogen bond lengths of 2.59 +/- 0.04 and 2.61 +/- 0.04 A, respectively. Similarly, lowering the pH of the free enzyme to 5.1 to protonate the active-site histidine (pK(a) = 6.0 +/- 0.4) resulted in the appearance of two deshielded resonances, at 17.7 and 16.4 ppm, consistent with SSHBs with lengths of 2.62 +/- 0.02 and 2.63 +/- 0.02 A, respectively. The NMR-derived distances agree with those found in the X-ray structures of the homologous acetylcholinesterase from Torpedo californica complexed with TMTFA (2.66 +/- 0.28 A) and sarin (2.53 +/- 0.26 A) and at low pH (2.52 +/- 0.25 A). However, the order of magnitude greater precision of the NMR-derived distances establishes the presence of SSHBs at the active site of acetylcholinesterase, and detect conformational heterogeneity of the active-site histidine. We suggest that the high catalytic power of cholinesterases results in part from the formation of a SSHB between Glu and His of the catalytic triad.

Potency of Michael reaction acceptors as inducers of enzymes that protect against carcinogenesis depends on their reactivity with sulfhydryl groups.

Induction of phase 2 enzymes and elevations of glutathione are major and sufficient strategies for protecting mammals and their cells against the toxic and carcinogenic effects of electrophiles and reactive forms of oxygen. Inducers belong to nine chemical classes and have few common properties except for their ability to modify sulfhydryl groups by oxidation, reduction, or alkylation. Much evidence suggests that the cellular "sensor" molecule that recognizes the inducers and signals the enhanced transcription of phase 2 genes does so by virtue of unique and highly reactive sulfhydryl functions that recognize and covalently react with the inducers. Benzylidene-alkanones and -cycloalkanones are Michael reaction acceptors whose inducer potency is profoundly increased by the presence of ortho- (but not other) hydroxyl substituent(s) on the aromatic ring(s). This enhancement correlates with more rapid reactivity of the ortho-hydroxylated derivatives with model sulfhydryl compounds. Proton NMR spectroscopy provides no evidence for increased electrophilicity of the beta-vinyl carbons (the presumed site of nucleophilic attack) on the hydroxylated inducers. Surprisingly, these ortho-hydroxyl groups display a propensity for extensive intermolecular hydrogen bond formation, which may raise the reactivity and facilitate addition of mercaptans, thereby raising inducer potencies.

NMR evidence for a short, strong hydrogen bond at the active site of a cholinesterase.

Cholinesterases (ChE), use a Glu-His-Ser catalytic triad to enhance the nucleophilicity of the catalytic serine. It has been shown that serine proteases, which employ an Asp-His-Ser catalytic triad for optimal catalytic efficiency, decrease the hydrogen bonding distance between the Asp-His pair to form a short, strong hydrogen bond (SSHB) upon binding mechanism-based inhibitors, which form tetrahedral Ser-adducts, analogous to the tetrahedral intermediates in catalysis, or at low pH when the histidine is protonated [Cassidy, C. S., Lin, J., Frey, P. A. (1997) Biochemistry 36, 4576-4584]. Two types of mechanism-based inhibitors were bound to pure equine butyrylcholinesterase (BChE), a 364 kDa homotetramer, and the complexes were studied by (1)H NMR at 600 MHz and 25-37 degrees C. The downfield region of the (1)H NMR spectrum of free BChE at pH 7.5 showed a broad, weak, deshielded resonance with a chemical shift, delta = 16.1 ppm, ascribed to a small amount of the histidine-protonated form. Upon addition of a 3-fold excess of diethyl 4-nitrophenyl phosphate (paraoxon) and subsequent dealkylation, the broad 16.1 ppm resonance increased in intensity 4.7-fold, and yielded a D/H fractionation factor phi = 0.72+/-0.10 consistent with a SSHB between Glu and His of the catalytic triad. From an empirical correlation of delta with hydrogen-bond length in small crystalline compounds, the length of this SSBH is 2.64+/-0.04 A, in agreement with the length of 2.62+/-0.02 A independently obtained from phi. The addition of a 3-fold excess of m-(N,N, N-trimethylammonio)trifluoroacetophenone to BChE yielded no signal at 16.1 ppm, and a 640 Hz broad, highly deshielded proton resonance with a chemical shift delta = 18.1 ppm and a D/H fractionation factor phi = 0.63+/-0.10, also consistent with a SSHB. The length of this SSHB is calculated to be 2.62+/-0.04 A from delta and 2.59+/-0.03 A from phi. These NMR-derived distances agree with those found in the X-ray structures of the homologous acetylcholinesterase complexed with the same mechanism-based inhibitors, 2.60+/-0.22 and 2.66+/-0.28 A. However, the order of magnitude greater precision of the NMR-derived distances establish the presence of SSHBs. We suggest that ChEs achieve their remarkable catalytic power in ester hydrolysis, in part, due to the formation of a SSHB between Glu and His of the catalytic triad.

GDP-mannose mannosyl hydrolase catalyzes nucleophilic substitution at carbon, unlike all other Nudix hydrolases.

GDP-mannose mannosyl hydrolase (GDPMH) from Escherichia coli is a 36. 8 kDa homodimer which, in the presence of Mg(2+), catalyzes the hydrolysis of GDP-alpha-D-mannose or GDP-alpha-D-glucose to yield sugar and GDP. On the basis of its amino acid sequence, GDPMH is a member of the Nudix family of enzymes which catalyze the hydrolysis of nucleoside diphosphate derivatives by nucleophilic substitution at phosphorus. However, GDPMH has a sequence rearrangement (RE to ER) in the conserved Nudix motif and is missing a Glu residue characteristic of the Nudix signature sequence. By (1)H NMR, the initial hydrolysis product of GDP-alpha-D-glucose is beta-D-glucose, indicating nucleophilic substitution with inversion at C1' of glucose. Substitution at carbon was confirmed by two-dimensional (1)H-(13)C HSQC spectra of the products of hydrolysis in 48.4% (18)O-labeled water which showed an additional C1' resonance of beta-D-glucose with a typical upfield (18)O isotope shift of 18 ppb and an intensity of 47.6% of the total signal. No (18)O isotope-shifted resonances (<4%) were found in the (31)P NMR spectrum of the GDP product. Thus, unlike all other Nudix enzymes studied so far, GDPMH catalyzes nucleophilic substitution at carbon rather than at phosphorus. A small solvent kinetic deuterium isotope effect on k(cat) of 1.76 +/- 0.25, independent of pH over the range of 6.0-9.3, suggests that the deprotonation of water may be part of the rate-limiting step.

Mutational, kinetic, and NMR studies of the roles of conserved glutamate residues and of lysine-39 in the mechanism of the MutT pyrophosphohydrolase.

The MutT enzyme catalyzes the hydrolysis of nucleoside triphosphates (NTP) to NMP and PP(i) by nucleophilic substitution at the rarely attacked beta-phosphorus. The solution structure of the quaternary E-M(2+)-AMPCPP-M(2+) complex indicated that conserved residues Glu-53, -56, -57, and -98 are at the active site near the bound divalent cation possibly serving as metal ligands, Lys-39 is positioned to promote departure of the NMP leaving group, and Glu-44 precedes helix I (residues 47-59) possibly stabilizing this helix which contributes four catalytic residues to the active site [Lin, J. , Abeygunawardana, C., Frick, D. N., Bessman, M. J., and Mildvan, A. S. (1997) Biochemistry 36, 1199-1211]. To test these proposed roles, the effects of mutations of each of these residues on the kinetic parameters and on the Mn(2+), Mg(2+), and substrate binding properties were examined. The largest decreases in k(cat) for the Mg(2+)-activated enzyme of 10(4.7)- and 10(2.6)-fold were observed for the E53Q and E53D mutants, respectively, while 97-, 48-, 25-, and 14-fold decreases were observed for the E44D, E56D, E56Q, and E44Q mutations, respectively. Smaller effects on k(cat) were observed for mutations of Glu-98 and Lys-39. For wild type MutT and its E53D and E44D mutants, plots of log(k(cat)) versus pH exhibited a limiting slope of 1 on the ascending limb and then a hump, i.e., a sharply defined maximum near pH 8 followed by a plateau, yielding apparent pK(a) values of 7.6 +/- 0.3 and 8.4 +/- 0.4 for an essential base and a nonessential acid catalyst, respectively, in the active quaternary MutT-Mg(2+)-dGTP-Mg(2+) complex. The pK(a) of 7.6 is assigned to Glu-53, functioning as a base catalyst in the active quaternary complex, on the basis of the disappearance of the ascending limb of the pH-rate profile of the E53Q mutant, and its restoration in the E53D mutant with a 10(1.9)-fold increase in (k(cat))(max). The pK(a) of 8.4 is assigned to Lys-39 on the basis of the disappearance of the descending limb of the pH-rate profile of the K39Q mutant, and the observation that removal of the positive charge of Lys-39, by either deprotonation or mutation, results in the same 8.7-fold decrease in k(cat). Values of k(cat) of both wild type MutT and the E53Q mutant were independent of solvent viscosity, indicating that a chemical step is likely to be rate-limiting with both. A liganding role for Glu-53 and Glu-56, but not Glu-98, in the binary E-M(2+) complex is indicated by the observation that the E53Q, E53D, E56Q, and E56D mutants bound Mn(2+) at the active site 36-, 27-, 4.7-, and 1.9-fold weaker, and exhibited 2.10-, 1.50-, 1.12-, and 1.24-fold lower enhanced paramagnetic effects of Mn(2+), respectively, than the wild type enzyme as detected by 1/T(1) values of water protons, consistent with the loss of a metal ligand. However, the K(m) values of Mg(2+) and Mn(2+) indicate that Glu-56, and to a lesser degree Glu-98, contribute to metal binding in the active quaternary complex. Mutations of the more distant but conserved residue Glu-44 had little effect on metal binding or enhancement factors in the binary E-M(2+) complexes. Two-dimensional (1)H-(15)N HSQC and three-dimensional (1)H-(15)N NOESY-HSQC spectra of the kinetically damaged E53Q and E56Q mutants showed largely intact proteins with structural changes near the mutated residues. Structural changes in the kinetically more damaged E44D mutant detected in (1)H-(15)N HSQC spectra were largely limited to the loop I-helix I motif, suggesting that Glu-44 stabilizes the active site region. (1)H-(15)N HSQC titrations of the E53Q, E56Q, and E44D mutants with dGTP showed changes in chemical shifts of residues lining the active site cleft, and revealed tighter nucleotide binding by these mutants, indicating an intact substrate binding site. (ABSTRACT TRUNCATED)

Kinetic, stereochemical, and structural effects of mutations of the active site arginine residues in 4-oxalocrotonate tautomerase.

Three arginine residues (Arg-11, Arg-39, Arg-61) are found at the active site of 4-oxalocrotonate tautomerase in the X-ray structure of the affinity-labeled enzyme [Taylor, A. B., Czerwinski, R. M., Johnson, R. M., Jr., Whitman, C. P., and Hackert, M. L. (1998) Biochemistry 37, 14692-14700]. The catalytic roles of these arginines were examined by mutagenesis, kinetic, and heteronuclear NMR studies. With a 1,6-dicarboxylate substrate (2-hydroxymuconate), the R61A mutation showed no kinetic effects, while the R11A mutation decreased k(cat) 88-fold and increased K(m) 8.6-fold, suggesting both binding and catalytic roles for Arg-11. With a 1-monocarboxylate substrate (2-hydroxy-2,4-pentadienoate), no kinetic effects of the R11A mutation were found, indicating that Arg-11 interacts with the 6-carboxylate of the substrate. The stereoselectivity of the R11A-catalyzed protonation at C-5 of the dicarboxylate substrate decreased, while the stereoselectivity of protonation at C-3 of the monocarboxylate substrate increased in comparison with wild-type 4-OT, indicating the importance of Arg-11 in properly orienting the dicarboxylate substrate by interacting with the charged 6-carboxylate group. With 2-hydroxymuconate, the R39A and R39Q mutations decreased k(cat) by 125- and 389-fold and increased K(m) by 1.5- and 2.6-fold, respectively, suggesting a largely catalytic role for Arg-39. The activity of the R11A/R39A double mutant was at least 10(4)-fold lower than that of the wild-type enzyme, indicating approximate additivity of the effects of the two arginine mutants on k(cat). For both R11A and R39Q, 2D (1)H-(15)N HSQC and 3D (1)H-(15)N NOESY-HSQC spectra showed chemical shift changes mainly near the mutated residues, indicating otherwise intact protein structures. The changes in the R39Q mutant were mainly in the beta-hairpin from residues 50 to 57 which covers the active site. HSQC titration of R11A with the substrate analogue cis, cis-muconate yielded a K(d) of 22 mM, 37-fold greater than the K(d) found with wild-type 4-OT (0.6 mM). With the R39Q mutant, cis, cis-muconate showed negative cooperativity in active site binding with two K(d) values, 3.5 and 29 mM. This observation together with the low K(m) of 2-hydroxymuconate (0.47 mM) suggests that only the tight binding sites function catalytically in the R39Q mutant. The (15)Nepsilon resonances of all six Arg residues of 4-OT were assigned, and the assignments of Arg-11, -39, and -61 were confirmed by mutagenesis. The binding of cis,cis-muconate to wild-type 4-OT upshifts Arg-11 Nepsilon (by 0.05 ppm) and downshifts Arg-39 Nepsilon (by 1.19 ppm), indicating differing electronic delocalizations in the guanidinium groups. A mechanism is proposed in which Arg-11 interacts with the 6-carboxylate of the substrate to facilitate both substrate binding and catalysis and Arg-39 interacts with the 1-carboxylate and the 2-keto group of the substrate to promote carbonyl polarization and catalysis, while Pro-1 transfers protons from C-3 to C-5. This mechanism, together with the effects of mutations of catalytic residues on k(cat), provides a quantitative explanation of the 10(7)-fold catalytic power of 4-OT. Despite its presence in the active site in the crystal structure of the affinity-labeled enzyme, Arg-61 does not play a significant role in either substrate binding or catalysis.

Cystic fibrosis transmembrane conductance regulator: solution structures of peptides based on the Phe508 region, the most common site of disease-causing DeltaF508 mutation.

Most cases of cystic fibrosis (CF), a common inherited disease of epithelial cell origin, are caused by the deletion of Phe508 located in the first nucleotide-binding domain (NBF1) of the protein called CFTR (cystic fibrosis transmembrane conductance regulator). To gain greater insight into the structure within the Phe508 region of the wild-type protein and the change in structure that occurs when this residue is deleted, we conducted nuclear magnetic resonance (NMR) studies on representative synthetic 26 and 25 amino acid peptide segments. 2D 1H NMR studies at 600 MHz of the 26-residue peptide consisting of Met498 to Ala523 in 10% DMSO, pH 4.0, at 25 degrees C show a continuous but labile helix from Gly500 to Lys522, based on both NH-NH(i,i+1) and alphaH-NH(i,i+1) NOEs. Phe508 within this helix shows only short-range (i,

Solution structure of Delta 5-3-ketosteroid isomerase complexed with the steroid 19-nortestosterone hemisuccinate.

The solution structure of the ketosteroid isomerase homodimer complexed with the product analogue 19-nortestosterone hemisuccinate (19-NTHS) was solved by heteronuclear multidimensional NMR methods using 1647 distance restraints, 77 dihedral angle (phi) restraints, and 67 hydrogen bond restraints per monomer. The refined secondary structure of each subunit consists of three alpha-helices, eight beta-strands, four turns, and two beta-bulges. The beta-strands form a mixed beta-sheet. One of the five proline residues, Pro-39, is cis and begins a nonclassical turn. A self-consistent ensemble of 15 tertiary/quaternary structures of the enzyme dimer-steroid complex, with no distance violations greater than 0.35 A, was generated by simulated annealing and energy minimization with the program X-PLOR. The mean pairwise RMSD of the secondary structural elements was 0.63 A for the average subunit and 1.25 A for the dimer. Within each subunit, the three alpha-helices are packed onto the concave surface of the beta-sheet with a groove between them into which the steroid binds at a site defined by 14 intermolecular distances. In the productive complex, Tyr-14, from alpha-helix 1, approaches both Asp-99 and the 3-keto group of 19-NTHS while, from beta-strand 1, the carboxylate of Asp-38 approaches the beta-face of the steroid near C4 and C6, between which it transfers a proton during catalysis. Thus the solution structure of the isomerase-steroid complex can accommodate the catalytic diad mechanism in which Asp-99 donates a hydrogen bond to Tyr-14 which in turn is hydrogen bonded to the 3-oxygen of the steroid. While direct hydrogen bonding of Asp-99 to the steroid oxygen is less likely, it cannot be excluded. All other interactions of the steroid with the enzyme are hydrophobic. The dimer interface, which is between the convex surfaces of the beta-sheets, is defined by 28 intersubunit NOEs between hydrophobic residues in the 13C-filtered NOESY-HSQC spectrum of a 13C/12C-heterolabeled dimer. Both hydrophobic and polar interactions occur at the dimer interface which contains no space that would permit additional steroid binding. Comparison of the complexed enzyme with the solution structure of the free enzyme [Wu et al. (1997) Science 276, 415-418] reveals that the three helices change position in the steroid complex, becoming more closely packed onto the concave surface of the beta-sheet, thus bringing Tyr-14 closer to Asp-99 and the substrate. Comparison of the enzyme-steroid complex in solution with the free enzyme in the crystalline state reveals similar differences between the positions of the helices.

Three-dimensional structure of the HTLV-II matrix protein and comparative analysis of matrix proteins from the different classes of pathogenic human retroviruses.

The matrix protein performs similar roles in all retroviruses, initially directing membrane localization of the assembling viral particle and subsequently forming a stable structural shell associated with the inner surface of the mature viral membrane. Although conserved structural elements are likely to perform these functions in all retroviral matrix proteins, invariant motifs are not evident at the primary sequence level and three-dimensional structures have been available for only the primate lentiviral matrix proteins. We have therefore used NMR spectroscopy to determine the structure of the matrix protein from human T-cell leukemia virus type II (HTLV-II), a member of the human oncovirus subclass of retroviruses. A total of 577 distance restraints were used to build 20 refined models that superimpose with an rmsd of 0.71 A for the backbone atoms of the structured regions. The globular HTLV-II matrix structure is composed of four alpha-helices and a 3(10) helix. Exposed basic residues near the C terminus of helix II form a putative membrane binding surface which could act in concert with the N-terminal myristoyl group to anchor the protein on the viral membrane surface. Clear structural similarities between the HTLV-II and HIV-1 matrix proteins suggest that the topology and exposed cationic membrane binding surface are likely to be conserved features of retroviral matrix proteins.

Comparison of the NMR and X-ray structures of the HIV-1 matrix protein: evidence for conformational changes during viral assembly.

The three-dimensional solution- and solid-state structures of the human immunodeficiency virus type-1 (HIV-1) matrix protein have been determined recently in our laboratories by NMR and X-ray crystallographic methods (Massiah et al. 1994. J Mol Biol 244:198-223; Hill et al. 1996. Proc Natl Acad Sci USA 93:3099-3104). The matrix protein exists as a monomer in solution at low millimolar protein concentrations, but forms trimers in three different crystal lattices. Although the NMR and X-ray structures are similar, detailed comparisons have revealed an approximately 6 A displacement of a short 3(10) helix (Pro 66-Gly 71) located at the trimer interface. High quality electron density and nuclear Overhauser effect (NOE) data support the integrity of the X-ray and NMR models, respectively. Because matrix apparently associates with the viral membrane as a trimer, displacement of the 3(10) helix may reflect a physiologically relevant conformational change that occurs during virion assembly and disassembly. These findings further suggest that Pro 66 and Gly 71, which bracket the 3(10) helix, serve as "hinges" that allow the 3(10) helix to undergo this structural reorientation.

Three-dimensional structure of the human immunodeficiency virus type 1 matrix protein.

The HIV-1 matrix protein forms an icosahedral shell associated with the inner membrane of the mature virus. Genetic analyses have indicated that the protein performs important functions throughout the viral life-cycle, including anchoring the transmembrane envelope protein on the surface of the virus, assisting in viral penetration, transporting the proviral integration complex across the nuclear envelope, and localizing the assembling virion to the cell membrane. We now report the three-dimensional structure of recombinant HIV-1 matrix protein, determined at high resolution by nuclear magnetic resonance (NMR) methods. The HIV-1 matrix protein is the first retroviral matrix protein to be characterized structurally and only the fourth HIV-1 protein of known structure. NMR signal assignments required recently developed triple-resonance (1H, 13C, 15N) NMR methodologies because signals for 91% of 132 assigned H alpha protons and 74% of the 129 assignable backbone amide protons resonate within chemical shift ranges of 0.8 p.p.m. and 1 p.p.m., respectively. A total of 636 nuclear Overhauser effect-derived distance restraints were employed for distance geometry-based structure calculations, affording an average of 13.0 NMR-derived distance restraints per residue for the experimentally constrained amino acids. An ensemble of 25 refined distance geometry structures with penalties (sum of the squares of the distance violations) of 0.32 A2 or less and individual distance violations under 0.06 A was generated; best-fit superposition of ordered backbone heavy atoms relative to mean atom positions afforded root-mean-square deviations of 0.50 (+/- 0.08) A. The folded HIV-1 matrix protein structure is composed of five alpha-helices, a short 3(10) helical stretch, and a three-strand mixed beta-sheet. Helices I to III and the 3(10) helix pack about a central helix (IV) to form a compact globular domain that is capped by the beta-sheet. The C-terminal helix (helix V) projects away from the beta-sheet to expose carboxyl-terminal residues essential for early steps in the HIV-1 infectious cycle. Basic residues implicated in membrane binding and nuclear localization functions cluster about an extruded cationic loop that connects beta-strands 1 and 2. The structure suggests that both membrane binding and nuclear localization may be mediated by complex tertiary structures rather than simple linear determinants.