Structure and mechanism of GDP-mannose glycosyl hydrolase, a Nudix enzyme that cleaves at carbon instead of phosphorus.

GDP-mannose glycosyl hydrolase (GDPMH) catalyzes the hydrolysis of GDP-mannose and GDP-glucose to GDP and sugar by substitution with inversion at C1 of the sugar. The enzyme has a modified Nudix motif and requires one divalent cation for activity. The 1.3 A X-ray structure of the GDPMH-Mg(2+)-GDP complex, together with kinetic, mutational, and NMR data, suggests a mechanism for the GDPMH reaction. Several residues and the divalent cation strongly promote the departure of the GDP leaving group, supporting a dissociative mechanism. Comparison of the GDPMH structure with that of a typical Nudix hydrolase suggests how sequence changes result in the switch of catalytic activity from P-O bond cleavage to C-O bond cleavage. Changes in the Nudix motif result in loss of binding of at least one Mg(2+) ion, and shortening of a loop by 6 residues shifts the catalytic base by approximately 10 A.

Kinetic and mutational studies of the number of interacting divalent cations required by bacterial and human methionine aminopeptidases.

Methionine aminopeptidases (MetAP) are responsible for the proteolytic removal of the initiator methionine from nascent proteins. This processing permits multiple posttranslational modifications and protein turnover. We have cloned, expressed in Escherichia coli, and purified the recombinant human mitochondrial MetAP isoform (MetAP1D). The full-length enzyme and a truncated form lacking the mitochondrial targeting sequence (residues 1-55) have been characterized as metal-requiring proteases, with Co2+ being the best activator. At the optimal pH (8.0), the kcat of MetAP1D of 0.39 min-1 is 280-fold lower, and the Km of the substrate Met-Pro-p-nitroanilide (576 microM) is 3-fold greater, than the respective kinetic parameters obtained with MetAP from E. coli, although MetAP1D is 61% homologous to E. coli MetAP and their circular dichroic spectra are nearly identical. MetAP1D thus appears to be a less efficient enzyme than other known MetAPs in vitro. At saturating substrate concentrations, a plot of Vmax versus free Co2+ shows sigmoidal metal activation of MetAP1D, both with and without an N-terminal His-tag, with a Hill coefficient (n) of 1.9 and a K0.5 of 0.40 microM. Similarly, E. coli MetAP shows n = 2.1 and K0.5 = 0.2 microM. Hence, at least two Co2+ ions, which may act cooperatively, are needed to promote catalysis, providing kinetic evidence for the functioning of both Co2+ ions of the binuclear complex found in the X-ray structure of E. coli MetAP [Roderick, S. L. and Matthews, B. W. (1993) Biochemistry 32, 3907-3912] and resolving a disagreement in the literature. The X-ray structure of the human cytosolic MetAP1 showed three Co2+ ions at the active site, with the third Co2+ coordinated by the conserved residue His 212 [Addlagatta, A., Hu, X., Liu, J. O., and Matthews, B. W. (2005) Biochemistry 44, 14741-14749]. Consistent with the structure, kinetic studies of the human cytosolic MetAP1 yielded a Hill coefficient (n) of 2.9 and a K0.5 of 0.26 microM for activation by Co2+, as well as a kcat of 25.5 min-1 and a Km of 740 microM for the substrate Met-Pro-p-nitroanilide. The H212A mutation decreased n to 2.2, decreased kcat 60-fold to 0.42 min-1, and increased K0.5 6.5-fold to 1.8 microM. The H212K mutation further decreased n to 1.4, decreased kcat 1800-fold to 0.014 min-1, and increased K0.5 158-fold to 41 microM. Hence, at least three Co2+ ions are needed to promote optimal catalysis by human MetAP1. Both mutations of His212 abolished the binding and/or the cooperativity of the third Co2+ ion, as indicated by the decreases in n and the increases in K0.5 of the remaining two Co2+ ions, but did not affect the Km of the substrate. The more damaging effects of the H212K mutation on both the Hill coefficient for Co2+ binding and the catalysis suggest that Lys 212 might directly compete with Co2+ for the third metal-binding site. Together, these results suggest that human MetAP1 is distinct from other members of the MetAP superfamily in the number of metal ions employed and likely mechanism of catalysis.

X-ray, NMR, and mutational studies of the catalytic cycle of the GDP-mannose mannosyl hydrolase reaction.

GDP-mannose hydrolase catalyzes the hydrolysis with inversion of GDP-alpha-D-hexose to GDP and beta-D-hexose by nucleophilic substitution by water at C1 of the sugar. Two new crystal structures (free enzyme and enzyme-substrate complex), NMR, and site-directed mutagenesis data, combined with the structure of the enzyme-product complex reported earlier, suggest a four-stage catalytic cycle. An important loop (L6, residues 119-125) contains a ligand to the essential Mg2+ (Gln-123), the catalytic base (His-124), and three anionic residues. This loop is not ordered in the X-ray structure of the free enzyme due to dynamic disorder, as indicated by the two-dimensional 1H-15N HMQC spectrum, which shows selective exchange broadening of the imidazole nitrogen resonances of His-124 (k(ex) = 6.6 x 10(4) s(-1)). The structure of the enzyme-Mg2+-GDP-mannose substrate complex of the less active Y103F mutant shows loop L6 in an open conformation, while the structure of the enzyme-Mg2+-GDP product complex showed loop L6 in a closed, "active" conformation. 1H-15N HMQC spectra show the imidazole N epsilon of His-124 to be unprotonated, appropriate for general base catalysis. Substituting Mg2+ with the more electrophilic metal ions Mn2+ or Co2+ decreases the pKa in the pH versus kcat rate profiles, showing that deprotonation of a metal-bound water is partially rate-limiting. The H124Q mutation, which decreases kcat 10(3.4)-fold and largely abolishes its pH dependence, is rescued by the Y103F mutation, which increases kcat 23-fold and restores its pH dependence. The structural basis of the rescue is the fact that the Y103F mutation shifts the conformational equilibrium to the open form moving loop L6 out of the active site, thus permitting direct access of the specific base hydroxide from the solvent. In the proposed dissociative transition state, which occurs in the closed, active conformation of the enzyme, the partial negative charge of the GDP leaving group is compensated by the Mg2+, and by the closing of loop L2 that brings Arg-37 closer to the beta-phosphate. The development of a positive charge at mannosyl C1, as the oxocarbenium-like transition state is approached, is compensated by closing the anionic loop, L6, onto the active site, further stabilizing the transition state.

Transient state kinetic studies of the MutT-catalyzed nucleoside triphosphate pyrophosphohydrolase reaction.

The MutT pyrophosphohydrolase, in the presence of Mg2+, catalyzes the hydrolysis of nucleoside triphosphates by nucleophilic substitution at Pbeta, to yield the nucleotide and PP(i). The best substrate for MutT is the mutagenic 8-oxo-dGTP, on the basis of its Km being 540-fold lower than that of dGTP. Product inhibition studies have led to a proposed uni-bi-iso kinetic mechanism, in which PP(i) dissociates first from the enzyme-product complex (k3), followed by NMP (k4), leaving a product-binding form of the enzyme (F) which converts to the substrate-binding form (E) in a partially rate-limiting step (k5) [Saraswat, V., et al. (2002) Biochemistry 41, 15566-15577]. Single- and multiple-turnover kinetic studies of the hydrolysis of dGTP and 8-oxo-dGTP and global fitting of the data to this mechanism have yielded all of the nine rate constants. Consistent with an "iso" mechanism, single-turnover studies with dGTP and 8-oxo-dGTP hydrolysis showed slow apparent second-order rate constants for substrate binding similar to their kcat/Km values, but well below the diffusion limit (approximately 10(9) M(-1) s(-1)): k(on)app = 7.2 x 10(4) M(-1) s(-1) for dGTP and k(on)app = 2.8 x 10(7) M(-1) s(-1) for 8-oxo-dGTP. These low k(on)app values are fitted by assuming a slow iso step (k5 = 12.1 s(-1)) followed by fast rate constants for substrate binding: k1 = 1.9 x 10(6) M(-1) s(-1) for dGTP and k1 = 0.75 x 10(9) M(-1) s(-1) for 8-oxo-dGTP (the latter near the diffusion limit). With dGTP as the substrate, replacing Mg2+ with Mn2+ does not change k1, consistent with the formation of a second-sphere MutT-M2+-(H2O)-dGTP complex, but slows the iso step (k5) 5.8-fold, and its reverse (k(-5)) 25-fold, suggesting that the iso step involves a change in metal coordination, likely the dissociation of Glu-53 from the enzyme-bound metal so that it can function as the general base. Multiple-turnover studies with dGTP and 8-oxo-dGTP show bursts of product formation, indicating partially rate-limiting steps following the chemical step (k2). With dGTP, the slow steps are the chemical step (k2 = 10.7 s(-1)) and the iso step (k5 = 12.1 s(-1)). With 8-oxo-dGTP, the slow steps are the release of the 8-oxo-dGMP product (k4 = 3.9 s(-1)) and the iso step (k5 = 12.1 s(-1)), while the chemical step is fast (k2 = 32.3 s(-1)). The transient kinetic studies are generally consistent with the steady state kcat and Km values. Comparison of rate constants and free energy diagrams indicate that 8-oxo-dGTP, at low concentrations, is a better substrate than dGTP because it binds to MutT 395-fold faster, dissociates 46-fold slower, and has a 3.0-fold faster chemical step. The true dissociation constants (KD) of the substrates from the E-form of MutT, which can now be obtained from k(-1)/k1, are 3.5 nM for 8-oxo-dGTP and 62 microM for dGTP, indicating that 8-oxo-dGTP binds 1.8 x 10(4)-fold tighter than dGTP, corresponding to a 5.8 kcal/mol lower free energy of binding.

Half-of-the-sites binding of reactive intermediates and their analogues to 4-oxalocrotonate tautomerase and induced structural asymmetry of the enzyme.

4-Oxalocrotonate tautomerase (4-OT), a homohexameric enzyme, converts the unconjugated enone, 2-oxo-4-hexenedioate (1), to the conjugated enone, 2-oxo-3-hexenedioate (3), via a dienolic intermediate, 2-hydroxymuconate (2). Pro-1 serves as the general base, and both Arg-11 and Arg-39 function in substrate binding and catalysis in an otherwise hydrophobic active site. Although 4-OT exhibits hyperbolic kinetics and no structural asymmetry either by X-ray or by NMR, inactivation by two affinity labels showed half-site stoichiometry [Stivers, J. T., et al. (1996) Biochemistry 35, 803-813; Johnson, W. H., Jr., et al. (1997) Biochemistry 36, 15724-15732], and titration of the R39Q mutant with cis,cis-muconate showed negative cooperativity [Harris, T. K., et al. (1999) Biochemistry 38, 12343-12357]. To test for anticooperativity during catalysis, 4-OT was titrated with equilibrium mixtures (> or = 81% product) of the reactive dicarboxylate or monocarboxylate intermediates, 2 or 2-hydroxy-2,4-pentadienoate (4), respectively, in three types of NMR experiments: two-dimensional 1H-15N HSQC titrations of backbone NH and of Arg N epsilonH resonances and one-dimensional 15N NMR titrations of Arg N epsilon resonances. All titrations showed substoichiometric binding of the equilibrium mixtures to 3 +/- 1 sites per hexamer with apparent dissociation constants comparable to the Km values of the intermediates. Compound 4 also bound 1 order of magnitude less tightly at another site, suggesting negative cooperativity. Consistent with negative cooperativity, asymmetry of the resulting complexes at saturating levels of 2 and 4 is indicated by splitting of the backbone NH resonances of 11 residues and 10 residues of 4-OT, respectively. The dicarboxylate competitive inhibitor, (2E)-fluoromuconate (5), with a KI of 45 +/- 7 microM, also exhibited substoichiometric binding to 3 +/- 1 sites per hexamer, with a KD of 25 +/- 18 microM, and splitting of the backbone NH resonance of L8. The monocarboxylate inhibitors (2E)- (6) and (2Z)-2-fluoro-2,4-pentadienoate (7) showed much weaker binding (KD = 3.1 +/- 1.3 mM), as well as splitting of two and five backbone NH resonances, respectively, indicating asymmetry of the complexes. The N epsilon resonances of both Arg-11 and Arg-39 were shifted downfield, and that of Pro-1N was broadened by all ligands, consistent with the major catalytic roles of these residues. Structural pathways for the site-site interactions which result in negative cooperativity are proposed on the basis of the X-ray structures of free and affinity-labeled 4-OT. Selective resonance broadenings induced by the binding of inactive analogues and active intermediates indicate residues which may be mobilized during reversible ligand binding and during catalysis, respectively.

Mutational, structural, and kinetic evidence for a dissociative mechanism in the GDP-mannose mannosyl hydrolase reaction.

GDP-mannose hydrolase (GDPMH) catalyzes the hydrolysis of GDP-alpha-d-sugars by nucleophilic substitution with inversion at the anomeric C1 atom of the sugar, with general base catalysis by H124. Three lines of evidence indicate a mechanism with dissociative character. First, in the 1.3 A X-ray structure of the GDPMH-Mg(2+)-GDP.Tris(+) complex [Gabelli, S. B., et al. (2004) Structure 12, 927-935], the GDP leaving group interacts with five catalytic components: R37, Y103, R52, R65, and the essential Mg(2+). As determined by the effects of site-specific mutants on k(cat), these components contribute factors of 24-, 100-, 309-, 24-, and >/=10(5)-fold, respectively, to catalysis. Both R37 and Y103 bind the beta-phosphate of GDP and are only 5.0 A apart. Accordingly, the R37Q/Y103F double mutant exhibits partially additive effects of the two single mutants on k(cat), indicating cooperativity of R37 and Y103 in promoting catalysis, and antagonistic effects on K(m). Second, the conserved residue, D22, is positioned to accept a hydrogen bond from the C2-OH group of the sugar undergoing substitution at C1, as was shown by modeling an alpha-d-mannosyl group into the sugar binding site. The D22A and D22N mutations decreased k(cat) by factors of 10(2.1) and 10(2.6), respectively, for the hydrolysis of GDP-alpha-d-mannose, and showed smaller effects on K(m), suggesting that the D22 anion stabilizes a cationic oxocarbenium transition state. Third, the fluorinated substrate, GDP-2F-alpha-d-mannose, for which a cationic oxocarbenium transition state would be destabilized by electron withdrawal, exhibited a 16-fold decrease in k(cat) and a smaller, 2.5-fold increase in K(m). The D22A and D22N mutations further decreased the k(cat) with GDP-2F-alpha-d-mannose to values similar to those found with GDP-alpha-d-mannose, and decreased the K(m) of the fluorinated substrate. The choice of histidine as the general base over glutamate, the preferred base in other Nudix enzymes, is not due to the greater basicity of histidine, since the pK(a) of E124 in the active complex (7.7) exceeded that of H124 (6.7), and the H124E mutation showed a 10(2.2)-fold decrease in k(cat) and a 4.0-fold increase in K(m) at pH 9.3. Similarly, the catalytic triad detected in the X-ray structure (H124- - -Y127- - -P120) is unnecessary for orienting H124, since the Y127F mutation had only 2-fold effects on k(cat) and K(m) with either H124 or E124 as the general base. Hence, a neutral histidine rather than an anionic glutamate may be necessary to preserve electroneutrality in the active complex.

Inverse thinking about double mutants of enzymes.

The quantitative effect of a second damaging mutation on a mutated enzyme may be additive, partially additive, synergistic, antagonistic, or absent, in the double mutant. Each of these five possible types of interactions has its own mechanistic explanation [Mildvan, A. S., Weber, D. J., and Kuliopulos, A. (1992) Arch. Biochem. Biophys. 294, 327-340]. Additive effects indicate independent functioning of the two residues in the process being studied, such as catalysis (k(cat)) or substrate binding (K(S)). Departures from additivity reflect interaction of the two residues. Thus, partial additivity indicates cooperativity, synergy indicates anticooperativity, and antagonism indicates opposing structural effects of the two mutations. No additional effects represent limiting cases of either partial additivity or antagonism. A significant conceptual simplification is achieved by applying inverse thinking, namely, by using the parameters of the double mutant rather than those of the wild-type enzyme as the reference point. To explain partially additive effects on k(cat), inverse thinking starts with the k(cat) of the double mutant. Restoring only one residue increases k(cat) by the factor A. Restoring only the other residue increases k(cat) by the factor B. Restoring both residues is shown to increase k(cat) by a factor greater than A x B, with the excess directly measuring the cooperativity. Similarly, inverse thinking provides simpler and more intuitive explanations of synergistic and antagonistic effects, as illustrated by specific examples.

Mutational, NMR, and NH exchange studies of the tight and selective binding of 8-oxo-dGMP by the MutT pyrophosphohydrolase.

The solution structure of the ternary MutT enzyme-Mg(2+)-8-oxo-dGMP complex showed the proximity of Asn119 and Arg78 and the modified purine ring of 8-oxo-dGMP, suggesting specific roles for these residues in the tight and selective binding of this nucleotide product [Massiah, M. A., Saraswat, V., Azurmendi, H. F., and Mildvan, A. S. (2003) Biochemistry 42, 10140-10154]. These roles are here tested by mutagenesis. The N119A, N119D, R78K, and R78A single mutations and the R78K/N119A double mutant showed very small effects on k(cat) (

Inhibition of reverse transcription in vivo by elevated manganese ion concentration.

Mutations in PMR1, a yeast gene encoding a calcium/manganese exporter, dramatically decrease Ty1 retrotransposition. Ty1 cDNA is reduced in pmr1 mutant cells, despite normal levels of Ty1 RNA and proteins. The transposition defect results from Mn(2+) accumulation that inhibits reverse transcription. Cytoplasmic accumulation of Mn(2+) in pmr1 cells may directly affect reverse transcriptase (RT) activity. Trace amounts of Mn(2+) potently inhibit Ty1 RT and HIV-1 RT in vitro when the preferred cation, Mg(2+), is present. Both Mn(2+) and Mg(2+) alone activate Ty1 RT cooperatively with Hill coefficients of 2, providing kinetic evidence for a dual divalent cation requirement at the RT active site. We propose that occupancy of the B site is the major determinant of catalytic activity and that Mn(2+) at this site greatly reduces catalytic activity.

Mutational, kinetic, and NMR studies of the mechanism of E. coli GDP-mannose mannosyl hydrolase, an unusual Nudix enzyme.

GDP-mannose mannosyl hydrolase (GDPMH) is an unusual Nudix family member, which catalyzes the hydrolysis of GDP-alpha-D-mannose to GDP and the beta-sugar by nucleophilic substitution at carbon rather than at phosphorus (Legler, P. M., Massiah, M. A., Bessman, M. J., and Mildvan, A. S. (2000) Biochemistry 39, 8603-8608). Using the structure and mechanism of MutT, the prototypical Nudix enzyme as a guide, we detected six catalytic residues of GDPMH, three of which were unique to GDPMH, by the kinetic and structural effects of site-specific mutations. Glu-70 (corresponding to Glu-57 in MutT) provides a ligand to the essential divalent cation on the basis of the effects of the E70Q mutation which decreased kcat 10(2.2)-fold, increased the dissociation constant of Mn2+ from the ternary E-Mn2+-GDP complex 3-fold, increased the K(m)Mg2+ 20-fold, and decreased the paramagnetic effect of Mn2+ on 1/T1 of water protons, indicating a change in the coordination sphere of Mn2+. In the E70Q mutant, Gln-70 was shown to be very near the active site metal ion by large paramagnetic effects of Mn2+ on its side chain -NH2 group. With wild-type GDPMH, the effect of pH on log(kcat/K(m)GDPmann) at 37 degrees C showed an ascending limb of unit slope, followed by a plateau yielding a pK(a) of 6.4, which increased to 6.7 +/- 0.1 in the pH dependence of log(kcat). The general base catalyst was identified as a neutral His residue by the DeltaH(ionization) = 7.0 +/- 0.7 kcal/mol, by the increase in pK(a) with ionic strength, and by mutation of each of the four histidine residues of GDPMH to Gln. Only the H124Q mutant showed the loss of the ascending limb in the pH versus log(kcat) rate profile, which was replaced by a weak dependence of rate on hydroxide concentration, as well as an overall 10(3.4)-fold decrease in kcat, indicating His-124 to be the general base, unlike MutT, which uses Glu-53 in this role. The H88Q mutant showed a 10(2.3)-fold decrease in kcat, a 4.4-fold increase in K(m)GDPmann, and no change in the pH versus log(kcat) rate profile, indicating an important but unidentified role of His-88 in catalysis. One and two-dimensional NMR studies permitted the sequence specific assignments of the imidazole HdeltaC, H(epsilon)C, N(delta), and N(epsilon) resonances of the four histidines and defined their protonation states. The pK(a) of His-124 (6.94 +/- 0.04) in the presence of saturating Mg2+ was comparable to the kinetically determined pK(a) at the same temperature (6.40 +/- 0.20). The other three histidines were neutral N(epsilon)H tautomers with pK(a) values below 5.5. Arg-52 and Arg-65 were identified as catalytic residues which interact electrostatically with the GDP leaving group by mutating these residues to Gln and Lys. The R52Q mutant decreased kcat 309-fold and increased K(m)GDPmann 40.6-fold, while the R52K mutant decreased kcat by only 12-fold and increased K(m)GDPmann 81-fold. The partial rescue of kcat, but not of K(m)GDPmann in the R52K mutant, suggests that Arg-52 is a bifunctional hydrogen bond donor to the GDP leaving group in the ground state and a monofunctional hydrogen bond donor in the transition state. Opposite behavior was found with the Arg-65 mutants, suggesting this residue to be a monofunctional hydrogen bond donor to the GDP leaving group in the ground state and a bifunctional hydrogen bond donor in the transition state. From these observations, a mechanism for GDPMH is proposed involving general base catalysis and electrostatic stabilization of the leaving group.

Interactions of the products, 8-oxo-dGMP, dGMP, and pyrophosphate with the MutT nucleoside triphosphate pyrophosphohydrolase.

The MutT enzyme from E. coli, in the presence of a divalent cation, catalyzes the hydrolysis of nucleoside- and deoxynucleoside-triphosphate (NTP) substrates by nucleophilic substitution at Pbeta, to yield a nucleotide (NMP) and PPi. The best substrate of MutT is believed to be the mutagenic nucleotide 8-oxo-dGTP, on the basis of its 10(3.4)-fold lower K(m) than that of dGTP (Maki, H., and Sekiguchi, M. (1992) Nature 355, 273-275). To determine the true affinity of MutT for an 8-oxo-nucleotide and to elucidate the kinetic scheme, product inhibition by 8-oxo-dGMP and dGMP and direct binding of these nucleotides to MutT were studied. With Mg(2+)-activated dGTP hydrolysis, 8-oxo-dGMP is a noncompetitive inhibitor with K(I)(sl)(o)(pe) = 49 nM, which is 10(4.6)-fold lower than the K(I)(sl)(o)(pe)of dGMP (1.7 mM). Similarly, the K(I)(intercept) of 8-oxo-dGMP is 10(4.0)-fold lower than that of dGMP. PPi is a linear uncompetitive inhibitor, suggesting that it dissociates first from the product complex, followed by the nucleotide. Noncompetitive inhibition by dGMP and 8-oxo-dGMP indicates an "iso" mechanism in which the nucleotide product leaves an altered form of the enzyme which slowly reverts to the form which binds substrate. Consistent with this kinetic scheme, (1)H-(15)N HSQC titration of MutT with dGMP reveals weak binding and fast exchange from one site with a K(D) = 1.8 mM, in agreement with its K(I)(sl)(o)(pe). With 8-oxo-dGMP, tight binding and slow exchange (n = 1.0 +/- 0.1, K(D) < 0.25 mM) are found. Isothermal calorimetric titration of MutT with 8-oxo-dGMP yields a K(D) of 52 nM, in agreement with its K(I)(sl)(o)(pe). Changing the metal activator from Mg(2+) to Mn(2+) had little effect on the K(I)(sl)(o)(pe) of dGMP or of 8-oxo-dGMP, consistent with the second-sphere enzyme-M(2+)-H(2)O-NTP-M(2+) complex found by NMR (Lin, J., Abeygunawardana, C., Frick, D. N., Bessman, M. J., and Mildvan, A. S. (1997) Biochemistry 36, 1199-1211), but it decreased the K(I) of PPi 12-fold, suggesting direct coordination of the PPi product by the enzyme-bound divalent cation. The tight binding of 8-oxo-dGMP to MutT (DeltaG degrees = -9.8 kcal/mol) is driven by a highly favorable enthalpy ( = -32 +/- 7 kcal/mol), with an unfavorable entropy (<-TDeltaS(o)(binding)> = +22 +/- 7 kcal/mol), as determined by van't Hoff analysis of the effect of temperature on the K(I)(sl)(o)(pe) and by isothermal titration calorimetry in two buffer systems. The binding of 8-oxo-dGMP to MutT induces changes in backbone (15)N and NH chemical shifts of 62 residues widely distributed throughout the protein, while dGMP binding induces smaller changes in only 22 residues surrounding the nucleotide binding site, suggesting that the unusually high affinity of MutT for 8-oxo-nucleotides is due not only to interactions with the altered 8-oxo or 7-NH positions on guanine, but results primarily from diffuse structural changes which tighten the protein structure around the 8-oxo-nucleotide.

Solution structure and NH exchange studies of the MutT pyrophosphohydrolase complexed with Mg(2+) and 8-oxo-dGMP, a tightly bound product.

To learn the structural basis for the unusually tight binding of 8-oxo-nucleotides to the MutT pyrophosphohydrolase of Escherichia coli (129 residues), the solution structure of the MutT-Mg(2+)-8-oxo-dGMP product complex (K(D) = 52 nM) was determined by standard 3-D heteronuclear NMR methods. Using 1746 NOEs (13.5 NOEs/residue) and 186 phi and psi values derived from backbone (15)N, Calpha, Halpha, and Cbeta chemical shifts, 20 converged structures were computed with NOE violations

Kinetic and magnetic resonance studies of the role of metal ions in the mechanism of Escherichia coli GDP-mannose mannosyl hydrolase, an unusual nudix enzyme.

Escherichia coli GDP-mannose mannosyl hydrolase (GDPMH), a homodimer, catalyzes the hydrolysis of GDP-alpha-D-sugars to yield the beta-D-sugar and GDP by nucleophilic substitution with inversion at the C1' carbon of the sugar [Legler, P. M., Massiah, M. A., Bessman, M. J., and Mildvan, A. S. (2000) Biochemistry 39, 8603-8608]. GDPMH requires a divalent cation for activity such as Mn2+ or Mg2+, which yield similar kcat values of 0.15 and 0.13 s(-1), respectively, at 22 degrees C and pH 7.5. Kinetic analysis of the Mn2+-activated enzyme yielded a K(m) of free Mn2+ of 3.9 +/- 1.3 mM when extrapolated to zero substrate concentration (K(a)Mn2+), which tightened to 0.32 +/- 0.18 mM when extrapolated to infinite substrate concentration (K(m)Mn2+). Similarly, the K(m) of the substrate extrapolated to zero Mn2+ concentration (K(S)(GDPmann) = 1.9 +/- 0.5 mM) and to infinite Mn2+ concentration (K(m)(GDPmann) = 0.16 +/- 0.09 mM) showed an order of magnitude decrease at saturating Mn2+. Such mutual tightening of metal and substrate binding suggests the formation of an enzyme-metal-substrate bridge complex. Direct Mn2+ binding studies, monitoring the concentration of free Mn2+ by EPR and of bound Mn2+ by its enhanced paramagnetic effect on the longitudinal relaxation rate of water protons (PRR), detected three Mn2+ binding sites per enzyme monomer with an average dissociation constant (K(D)) of 3.2 +/- 1.0 mM, in agreement with the kinetically determined K(a)Mn2+. The enhancement factor (epsilon(b)) of 11.5 +/- 1.2 indicates solvent access to the enzyme-bound Mn2+ ions. No cross relaxation was detected among the three bound Mn2+ ions, suggesting them to be separated by at least 10 A. Such studies also yielded a weak dissociation constant for the binary Mn2+-GDP-mannose complex (K1 = 6.5 +/- 1.0 mM) which significantly exceeded the kinetically determined K(m) values of Mn2+, indicating the true substrate to be GDP-mannose rather than its Mn2+ complex. Substrate binding monitored by changes in 1H-15N HSQC spectra yielded a dissociation constant for the binary E-GDP-mannose complex (K(S)(GDPmann)) of 4.0 +/- 0.5 mM, comparable to the kinetically determined K(S) value (1.9 +/- 0.5 mM). To clarify the metal stoichiometry at the active site, product inhibition by GDP, a potent competitive inhibitor (K(I) = 46 +/- 27 microM), was studied. Binding studies revealed a weak, binary E-GDP complex (K(D)(GDP) = 9.4 +/- 3.2 mM) which tightened approximately 500-fold in the presence of Mn2+ to yield a ternary E-Mn2+-GDP complex with a dissociation constant, K3(GDP) = 18 +/- 9 microM, which overlaps with the K(I)(GDP). The tight binding of Mn2+ to 0.7 +/- 0.2 site per enzyme subunit in the ternary E-Mn2+-GDP complex (K(A)' = 15 microM) and the tight binding of GDP to 0.8 +/- 0.1 site per enzyme subunit in the ternary E-Mg2+-GDP complex (K3 < 0.5 mM) indicate a stoichiometry close to 1:1:1 at the active site. The decrease in the enhancement factor of the ternary E-Mn2+-GDP complex (epsilon(T) = 4.9 +/- 0.4) indicates decreased solvent access to the active site Mn2+, consistent with an E-Mn2+-GDP bridge complex. Fermi contact splitting (4.3 +/- 0.2 MHz) of the phosphorus signal in the ESEEM spectrum established the formation of an inner sphere E-Mn2+-GDP complex. The number of water molecules coordinated to Mn2+ in this ternary complex was determined by ESEEM studies in D2O to be two fewer than on the average Mn2+ in the binary E-Mn2+ complexes, consistent with bidentate coordination of enzyme-bound Mn2+ by GDP. Kinetic, metal binding, and GDP binding studies with Mg2+ yielded dissociation constants similar to those found with Mn2+. Hence, GDPMH requires one divalent cation per active site to promote catalysis by facilitating the departure of the GDP leaving group, unlike its homologues the MutT pyrophosphohydrolase, which requires two, or Ap4A pyrophosphatase, which requires three.

The roles of active-site residues in the catalytic mechanism of trans-3-chloroacrylic acid dehalogenase: a kinetic, NMR, and mutational analysis.

trans-3-Chloroacrylic acid dehalogenase (CaaD) converts trans-3-chloroacrylic acid to malonate semialdehyde by the addition of H(2)O to the C-2, C-3 double bond, followed by the loss of HCl from the C-3 position. Sequence similarity between CaaD, an (alphabeta)(3) heterohexamer (molecular weight 47,547), and 4-oxalocrotonate tautomerase (4-OT), an (alpha)(6) homohexamer, distinguishes CaaD from those hydrolytic dehalogenases that form alkyl-enzyme intermediates. The recently solved X-ray structure of CaaD demonstrates that betaPro-1 (i.e., Pro-1 of the beta subunit), alphaArg-8, alphaArg-11, and alphaGlu-52 are at or near the active site, and the >or=10(3.4)-fold decreases in k(cat) on mutating these residues implicate them as mechanistically important. The effect of pH on k(cat)/K(m) indicates a catalytic base with a pK(a) of 7.6 and an acid with a pK(a) of 9.2. NMR titration of (15)N-labeled wild-type CaaD yielded pK(a) values of 9.3 and 11.1 for the N-terminal prolines, while the fully active but unstable alphaP1A mutant showed a pK(a) of 9.7 (for the betaPro-1), implicating betaPro-1 as the acid catalyst, which may protonate C-2 of the substrate. These results provide the first evidence for an amino-terminal proline, conserved in all known tautomerase superfamily members, functioning as a general acid, rather than as a general base as in 4-OT. Hence, a reasonable candidate for the general base in CaaD is the active site residue alphaGlu-52. CaaD has 10 arginine residues, six in the alpha-subunit (Arg-8, Arg-11, Arg-17, Arg-25, Arg-35, and Arg-43), and four in the beta-subunit (Arg-15, Arg-21, Arg-55, and Arg-65). (1)H-(15)N-heteronuclear single quantum coherence (HSQC) spectra of CaaD showed seven to nine Arg-NepsilonH resonances (denoted R(A) to R(I)) depending on the protein concentration and pH. One of these signals (R(D)) disappeared in the spectrum of the largely inactive alphaR11A mutant (deltaH = 7.11 ppm, deltaN = 89.5 ppm), and another one (R(G)) disappeared in the spectrum of the inactive alphaR8A mutant (deltaH = 7.48 ppm, deltaN = 89.6 ppm), thereby assigning these resonances to alphaArg-11NepsilonH, and alphaArg-8NepsilonH, respectively. (1)H-(15)N-HSQC titration of the enzyme with the substrate analogue 3-chloro-2-butenoic acid (3-CBA), a competitive inhibitor (K(I)(slope) = 0.35 +/- 0.06 mM), resulted in progressive downfield shifts of the alphaArg-8Nepsilon resonance yielding a K(D) = 0.77 +/- 0.44 mM, comparable to the (K(I)(slope), suggestive of active site binding. Increasing the pH of free CaaD to 8.9 at 5 degrees C resulted in the disappearance of all nine Arg-NepsilonH resonances due to base-catalyzed NepsilonH exchange. Saturating the enzyme with 3-CBA (16 mM) induced the reappearance of two NepsilonH signals, those of alphaArg-8 and alphaArg-11, indicating that the binding of the substrate analogue 3-CBA selectively slows the NepsilonH exchange rates of these two arginine residues. The kinetic and NMR data thus indicate that betaPro-1 is the acid catalyst, alphaGlu-52 is a reasonable candidate for the general base, and alphaArg-8 and alphaArg-11 participate in substrate binding and in stabilizing the aci-carboxylate intermediate in a Michael addition mechanism.