NAD-binding Site Research Articles

Stopped-flow kinetics of hydride transfer between nucleotides by recombinant domains of proton-translocating transhydrogenase.

Transhydrogenase catalyses the transfer of reducing equivalents between NAD(H) and NADP(H) coupled to proton translocation across the membranes of bacteria and mitochondria. The protein has a tridomain structure. Domains I and III protrude from the membrane (e.g. on the cytoplasmic side in bacteria) and domain II spans the membrane. Domain I has the binding site for NAD+/NADH, and domain III for NADP+/NADPH. We have separately purified recombinant forms of domains I and III from Rhodospirillum rubrum transhydrogenase. When the two recombinant proteins were mixed with substrates in the stopped-flow spectrophotometer, there was a biphasic burst of hydride transfer from NADPH to the NAD+ analogue, acetylpyridine adenine dinucleotide (AcPdAD+). The burst, corresponding to a single turnover of domain III, precedes the onset of steady state, which is limited by very slow release of product NADP+ (k approximately 0.03 s(-1)). Phase A of the burst (k approximately 600 s(-1)) probably arises from fast hydride transfer in complexes of domains I and III. Phase B (k approximately 10-50 s(-1)), which predominates when the concentration of domain I is less than that of domain III, probably results from dissociation of the domain I:III complexes and further association and turnover of domain I. Phases A and B were only weakly dependent on pH, and it is therefore unlikely that either the hydride transfer reaction, or conformational changes accompanying dissociation of the I:III complex, are directly coupled to proton binding or release. A comparison of the temperature dependences of AcPdAD+ reduction by [4B-2H]NADPH, and by [4B-1H]NADPH, during phase A shows that there may be a contribution from quantum mechanical tunnelling to the process of hydride transfer. Given that hydride transfer between the nucleotides is direct [Venning, J. D., Grimley, R. L., Bizouarn, T., Cotton, N. P. J. & Jackson, J. B. (1997) J. Biol. Chem. 272, 27535-27538], this suggests very close proximity of the nicotinamide rings of the two nucleotides in the I:III complex.

Interactions of reduced and oxidized nicotinamide mononucleotide with wild-type and αD195E mutant proton-pumping nicotinamide nucleotide transhydrogenases from Escherichia coli

The interaction of reduced nicotinamide mononucleotide (NMNH), constituting one half of NADH, with the wild-type and αD195E proton-pumping nicotinamide nucleotide transhydrogenase from Escherichia coli was investigated. Reduction of thio-NADP + by NMNH was catalysed at approximately 30% of the rate with NADH. Other activities including proton pumping and the cyclic reduction of 3′-acetyl-pyridine-NAD + by NMNH in the presence of NADP+ were more strongly inhibited. The αD195 residue is assumed to interact with the 2′-OH moiety of the adenosine-5′-phosphate, i.e., the second nucleotide of NADH. Mutation of this residue to αD195E resulted in a 90% decrease in activity with NMNH as well as NADH as substrate, suggesting that it produced global structural changes of the NAD(H) binding site. The results suggest that the NMN moiety of NADH is a substrate of transhydrogenase, and that the adenine nucleotide is not required for catalysis or proton pumping.

The mechanism of the elongation and branching reaction of Poly(ADP-ribose) polymerase as derived from crystal structures and mutagenesis

The binding site for the acceptor substrate poly(ADP-ribose) in the elongation reaction of the ADP-ribosyl transferase poly(ADP-ribose) polymerase (PARP) was detected by cocrystallizing the enzyme with an NAD + analogue. The site was confirmed by mutagenesis studies. In conjunction with the binding site of the donor NAD +, the bound acceptor reveals the geometry of the elongation reaction. It shows in particular that the strictly conserved glutamate residue of all ADP-ribosylating enzymes (Glu988 of PARP) facilitates the reaction by polarizing both, donor and acceptor. Moreover, the binding properties of the acceptor site suggest a mechanism for the branching reaction, that also explains the dual specificity of this transferase for elongation and branching, which is unique among polymer-forming enzymes.

Crystal structure of the NAD complex of human deoxyhypusine synthase: an enzyme with a ball-and-chain mechanism for blocking the active site.

Eukaryotic initiation factor 5A (elF-5A) contains an unusual amino acid, hypusine [N epsilon-(4-aminobutyl-2-hydroxy)lysine]. The first step in the post-translational formation of hypusine is catalysed by the enzyme deoxyhypusine synthase (DHS). The modified version of elF-5A, and DHS, are required for eukaryotic cell proliferation. Knowledge of the three-dimensional structure of this key enzyme should permit the design of specific inhibitors that may be useful as anti-proliferative agents. The crystal structure of human DHS with bound NAD cofactor has been determined and refined at 2.2 A resolution. The enzyme is a tetramer of four identical subunits arranged with 222 symmetry; each subunit contains a nucleotide-binding (or Rossmann) fold. The tetramer comprises two tightly associated dimers and contains four active sites, two in each dimer interface. The catalytic portion of each active site is located in one subunit while the NAD-binding site is located in the other. The entrance to the active-site cavity is blocked by a two-turn alpha helix, part of a third subunit, to which it is joined by an extended loop. The active site of DHS is a cavity buried below the surface of the enzyme at the interface between two subunits. In the conformation observed here, the substrate-binding site is inaccessible and we propose that the reaction steps carried out by the enzyme must be accompanied by significant conformational changes, the least of which would be the displacement of the two-turn alpha helix.

Inactivation of glyceraldehyde-3-phosphate dehydrogenase by ferrylmyoglobin

Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was rapidly inactivated by ferrylmyoglobin (ferrylMb). FerrylMb rapidly reacts with the sulfhydryl group of protein. We therefore surmised that the cysteine residues of GAPDH react with ferrylMb. However, the amount of ferrylMb required to inactivate the enzyme was in excess of the equivalent amount of cysteine in the enzyme. FerrylMb was reduced not only by cysteine, but also by tyrosine and tryptophane. Adding cysteine strongly blocked the inactivation of GAPDH induced by ferrylMb, but adding tyrosine and tryptophane did not prevent the enzyme inactivation. However, adding cysteine, but not tryptophane and tyrosine, produced a maximum absorption at 580 nm, suggesting the formation of sulfmyoglobin through the reaction of ferrylMb with cysteine. Furthermore, three new bands of molecular weights 50, 55 and 100 kDa occurred on the sodium dodecyl sulfate (SDS)–polyacrylamide gel during the exposure of GAPDH to ferrylMb. Cysteine, but not tryptophane and tyrosine, inhibited the formation of the bands. Kinetic data indicated that the binding site of NAD, but not glyceraldehyde-3-phosphate (G3P), was damaged by ferrylMb. These results suggest that inactivation of GAPDH induced by ferrylMb is predominantly due to oxidation of the essential cysteine 149, and that NAD protects the active site from oxidative attack of ferrylMb.

The reduction of acetylpyridine adenine dinucleotide by NADH: is it a significant reaction of proton-translocating transhydrogenase, or an artefact?

Transhydrogenase is a proton pump. It has separate binding sites for NAD +/NADH (on domain I of the protein) and for NADP +/NADPH (on domain III). Purified, detergent-dispersed transhydrogenase from Escherichia coli catalyses the reduction of the NAD + analogue, acetylpyridine adenine dinucleotide (AcPdAD +), by NADH at a slow rate in the absence of added NADP + or NADPH. Although it is slow, this reaction is surprizing, since transhydrogenase is generally thought to catalyse hydride transfer between NAD(H) – or its analogues and NADP(H) – or its analogues, by a ternary complex mechanism. It is shown that hydride transfer occurs between the 4 A position on the nicotinamide ring of NADH and the 4 A position of AcPdAD +. On the basis of the known stereospecificity of the enzyme, this eliminates the possibilities of transhydrogenation (a) from NADH in domain I to AcPdAD + wrongly located in domain III; and (b) from NADH wrongly located in domain III to AcPdAD + in domain I. In the presence of low concentrations of added NADP + or NADPH, detergent-dispersed E. coli transhydrogenase catalyses the very rapid reduction of AcPdAD + by NADH. This reaction is cyclic; it takes place via the alternate oxidation of NADPH by AcPdAD + and the reduction of NADP + by NADH, while the NADPH and NADP + remain tightly bound to the enzyme. In the present work, it is shown that the rate of the cyclic reaction and the rate of reduction of AcPdAD + by NADH in the absence of added NADP +/NADPH, have similar dependences on pH and on MgSO 4 concentration and that they have a similar kinetic character. It is therefore suggested that the reduction of AcPdAD + by NADH is actually a cyclic reaction operating, either with tightly bound NADP +/NADPH on a small fraction (<5%) of the enzyme, or with NAD +/NADH (or AcPdAD +/AcPdADH) unnaturally occluded within the domain III site. Transhydrogenase associated with membrane vesicles (chromatophores) of Rhodospirillum rubrum also catalyses the reduction of AcPdAD + by NADH in the absence of added NADP +/NADPH. When the chromatophores were stripped of transhydrogenase domain I, that reaction was lost in parallel with `normal reverse' transhydrogenation (e.g., the reduction of AcPdAD + by NADPH). The two reactions were fully recovered upon reconstitution with recombinant domain I protein. However, after repeated washing of the domain I-depleted chromatophores, reverse transhydrogenation activity (when assayed in the presence of domain I) was retained, whereas the reduction of AcPdAD + by NADH declined in activity. Addition of low concentrations of NADP + or NADPH always supported the same high rate of the NADH→AcPdAD + reaction independently of how often the membranes were washed. It is concluded that, as with the purified E. coli enzyme, the reduction of AcPdAD + by NADH in chromatophores is a cyclic reaction involving nucleotides that are tightly bound in the domain III site of transhydrogenase. However, in the case of R. rubrum membranes it can be shown with some certainty that the bound nucleotides are NADP + or NADPH. The data are thus adequately explained without recourse to suggestions of multiple nucleotide-binding sites on transhydrogenase.

Binding of thyroxine analogs to human liver aldehyde dehydrogenases.

A fragment of a cytosolic thyroid-hormone-binding protein from Xenopus liver was reported to be 92-100% identical to residues 236-258 in several cytosolic aldehyde dehydrogenases [Yamauchi, K. & Tata, J. R. (1994) Eur. J. Biochem. 225, 1105-1112], which have been proposed to form part of the hinge region necessary to bind the adenosine moiety of NAD. Here we investigated the effects of two thyroxine analogs, 3,3',5-triiodo-L-thyronine and 3,3',5-triiodothyroacetic acid, on purified human liver mitochondrial and cytosolic aldehyde dehydrogenases. The compounds were found to be competitive inhibitors against NAD and uncompetitive inhibitors with respect to aldehyde. At pH 7.4, the apparent Ki values were in the micromolar range when the concentration of NAD was varied. The inhibition against recombinantly expressed mutant forms of aldehyde dehydrogenase, which possessed diminished NAD binding, was determined. Essentially no differences were found between the native enzyme and the mutants, showing that the analog binding was not affected by altering the NAD-binding site. Furthermore, the analogs could displace NAD but not NADH from the enzyme. These findings indicated that the binding of NAD differed from that of NADH, and that aldehyde dehydrogenases, like other dehydrogenases, can be inhibited by thyroxine analogs.

Reversible folding of UDP-galactose 4-epimerase from Escherichia coli.

UDP-galactose 4-epimerase from Escherichia coli is a homodimer of 39-kDa subunits having 1 or 2 molecules of NAD bound non-covalently/dimer. The enzyme can be dissociated and denatured by 8 M urea at pH 7.0 to a state having only 15% of residual secondary structure. Dilution of the denaturant by 20 mM potassium phosphate, pH 8.5, leads to functional reconstitution of the enzyme. No addition of extraneous NAD is required for reactivation, indicating a strong affinity of the cofactor for refolded molecule. The reactivation follows a second-order kinetics (k = 1.2 +/- 0.07 X 10(3) M(-1) s(-1) at 25 degrees C) with an energy of activation of 23.79 +/- 0.33 kJ/mol. The native, denatured and renatured states of the enzyme were characterized by far-ultraviolet CD spectra for secondary structure: protein fluorescence, interaction with extrinsic fluorescence probe ANS (1-anilino 8-naphthalene sulfonic acid) and ultraviolet absorption spectra for tertiary structure and size-exclusion HPLC, gel-filtration chromatography and light-scattering for quaternary structure. The folding process could be broadly divided into two distinct steps: (a) regain of secondary structure and dimerization were fast and were complete within 2 min and 9 min, respectively, and (b) regain of catalytic activity was slow and was complete by 45 min. No active holoenzyme could be identified. It appears that generation of the NAD-binding site and subsequent assembly of NAD is the rate-limiting step expressing catalytic activity.

Photodependent Inhibition of Bovine Spleen NAD+Glycohydrolase by 8-Azido Carbocyclic Analogs of NAD+

Carba-NAD+and pseudocarba-NAD+, and their 8-azidoadenosyl derivatives, were found to be good competitive inhibitors of calf spleen NAD+glycohydrolase. The 8-azido compounds, tested as photoaffinity labels, inhibited the enzyme in a light- and time-dependent manner; this inhibition could be prevented by 3-aminopyridine adenine dinucleotide (n3PdAD+), a competitive inhibitor of NAD+glycohydrolase. Irradiation in the presence of the [3H]-labeled 8-azido-carba-NAD+derivative resulted in an irreversible incorporation of the radioactivity into the enzyme that could be largely prevented by addition of n3PdAD+. These results indicate that carbocyclic-analogs of NAD+will be useful in identifying the substrate binding site of NAD+glycohydrolase.

Oxygen-controlled regulation of the flavohemoglobin gene in Bacillus subtilis

A gene, hmp, which encodes a ubiquitous protein homologous to hemoglobin was isolated among genes from Bacillus subtilis that are induced under anaerobic conditions. The hmp protein belongs to the family of two-domain flavohemoproteins, homologs of which have been isolated from various organisms such as Escherichia coli, Alcaligenes eutrophus, and Saccharomyces cerevisiae. These proteins consist of an amino-terminal hemoglobin domain and a carboxy-terminal redox active site domain with potential binding sites for NAD(P)H and flavin adenine dinucleotide. The expression of hmp is strongly induced upon oxygen limitation, and the induction is dependent on a two-component regulatory pair, ResD and ResE, an anaerobic regulator, FNR, and respiratory nitrate reductase, NarGHJI. The requirement of FNR and NarGHJI for hmp expression is completely bypassed by the addition of nitrite in the culture medium, indicating that fnr is required for transcriptional activation of narGHJI, which produces nitrite, leading to induction of hmp expression. In contrast, induction of hmp was still dependent on resDE in the presence of nitrite. A defect in hmp in B. subtilis has no significant effect on anaerobic growth.

Association of the cytoplasmic domain of intercellular-adhesion molecule-1 with glyceraldehyde-3-phosphate dehydrogenase and beta-tubulin.

To elucidate the molecular mechanisms of the transendothelial migration of leukocytes, we attempted to identify the cellular proteins capable of interaction with the cytoplasmic domain of the intercellular adhesion molecule-1 (ICAM-1) in a rat brain microvessel endothelial cell line (RBE4 cells). A 27-amino-acid synthetic peptide, corresponding to the cytoplasmic domain of rat ICAM-1, was covalently linked to a Sepharose matrix. Upon affinity chromatography of RBE4 cell cytosol, several ICAM-1-interacting proteins were specifically eluted by the soluble peptide. Two of these proteins have been identified by microsequencing as the cytoskeletal protein beta-tubulin and the glycolytic enzyme glyceraldehyde-3-phosphate dehydrogenase (GraP-DH). Experiments carried out with purified GraP-DH or CNBr fragments of GraP-DH indicated that binding to the ICAM-1 matrix was mediated by the C-terminal domain of GraP-DH, containing the binding site of the cofactor NAD+, and that NAD+ could compete with this binding. Using a series of ICAM-1 C-terminal truncated peptides, we could demonstrate that (a) the nitric-oxide-induced covalent linkage of NAD+ to GraP-DH was impaired by these peptides, (b) the glycolytic activity of GraP-DH was drastically inhibited by a truncated peptide containing the 15 C-terminal residues, (c) nitric oxide appeared to prevent this inhibition. Together, our results demonstrate that GraP-DH specifically associates with the isolated ICAM-1 cytoplasmic domain. Since GraP-DH is known as a microtubule bundling protein, these findings suggest that, in a cellular environment, GraP-DH may behave as an adaptor molecule by linking ICAM-1 to the microtubule network. The role of nitric oxide in the modulation of this interaction deserves further investigation.

Native, but Not Genetically Inactivated, Pertussis Toxin Protects Mice against Experimental Allergic Encephalomyelitis

Treatment of SJL mice with 400 ngBordetella pertussistoxin (PT) either in saline or emulsified in incomplete Freund's adjuvant protected the mice against experimental autoimmune encephalomyelitis (EAE) induced 28 days later by a synthetic peptide of myelin proteolipid protein (PLP139-151) in complete Freund's adjuvant. However, treatment with a genetically inactivated pertussis toxin in which the catalytic and NAD-binding sites of the ADP-ribosyltransferase subunit were modified by site-directed mutagenesis was without effect.In vitro,lymphocyte proliferation was considerably enhanced by both the native and the inactivated toxin, at concentrations of 0.1–1 μg/ml. However, strong inhibition of proliferation was also observed with the native toxin only, at concentrations that were two to three orders of magnitude lower than that required for the mitogenic effect (0.1–1 ng/ml). The inhibition of proliferation was detectable in the case of high-background proliferation, after stimulation with antigen (PLP139-151 or purified protein derivative ofMycobacterium tuberculosis), or with anti-CD3 monoclonal antibody, but not after stimulation with concanavalin A or phorbol esters and Ca2+ionophore. These results suggest that the inhibitory effect of PT operates by interfering selectively with a T cell receptor-dependent signaling pathway. The biological significance of thein vitroinhibitory effect of PT was demonstrated by a considerable decrease and/or delay in the ability of lymphocytes grown with PLP139-151 and low concentrations of PT to transfer EAE to naive recipients.

Histidine-450 is the catalytic residue of L-3-hydroxyacyl coenzyme A dehydrogenase associated with the large alpha-subunit of the multienzyme complex of fatty acid oxidation from Escherichia coli.

Multienzyme complexes of fatty acid oxidation from Escherichia coli with Gln or Ala substituting for His450 or with Ala in place of Gly322 in the large alpha-subunit have been purified and characterized. The alpha/Gly322-->Ala mutation did not significantly affect the catalytic efficiencies (kcat/k(m)) of different component enzymes except for a 6.1-fold decrease in the kcat/k(m) of L-3-hydroxyacyl-CoA dehydrogenase and a 10-fold increase in the k(m) for NADH. This observation confirms the prediction [Yang, X.-Y. H., Schulz, H., Elzinga, M., & Yang, S.-Y. (1991) Biochemistry 30, 6788-6795] that the E. coli dehydrogenase has an NAD-binding site at its amino-terminal domain and structurally resembles the pig heart dehydrogenase. The pH dependence of the kcat/k(m) of the E. coli dehydrogenase suggested the catalytic involvement of an amino acid residue with a pKa of 6, which is presumably a histidine residue as proposed previously on the basis of chemical modifications. Since His450 of the E. coli multifunctional protein is the only histidine conserved in all known L-3-hydroxyacyl-CoA dehydrogenases, and since its counterpart in pig heart enzyme appeared to be close to the 3-keto group of the fatty acyl moiety of the substrate, His450 was replaced by either Gln or Ala. The catalytic properties of 3-ketoacyl-CoA thiolase, enoyl-CoA hydratase, and delta 3-cis-delta 2-trans-enoyl-CoA isomerase of the alpha/His450-->Gln mutant complex were virtually unchanged except for a small decrease in the kcat values of the latter two enzymes. In contrast, the dehydrogenase of this mutant complex was almost inactive due to a greater than 3000-fold decrease in its kcat and a 6-fold increase in the k(m) for NADH. The alpha/His450-->Ala mutant complex showed similar catalytic behaviors. Taken together, several lines of evidence lead to the conclusion that His450 is the catalytic residue of L-3-hydroxyacyl-CoA dehydrogenase of the E. coli multifunctional fatty acid oxidation protein.

Xanthine oxidase and xanthine dehydrogenase

Xanthine oxidase and xanthine dehydrogenase are enzymes involved in the metabolism of purines and pyrimidines in various organisms. Their relationship to one another has been the subject of considerable debate, primarily because of their proposed roles in ischemia/reperfusion damage in tissues. Differences in the kinetics and oxidation-reduction behavior of the two forms are accounted for by the presence in the dehydrogenase of a binding site for NAD+, as well as a substantially lower reduction potential for the flavin FADH./FADH2 couple of the dehydrogenase relative to the oxidase. This review presents recent advances of our understanding of the biochemistry and molecular biology of these systems, including a model for the overall morphology of xanthine oxidizing enzymes. The evidence that the two enzymes represent alternate forms of the same gene product, in some cases reversibly interconvertible between one another, is discussed.

His273 of 3-Isopropylmalate Dehydrogenase from Thermus thermophilus HB8 Is Involved in the Coenzyme Binding

The coenzyme binding site of NAD+-dependent 3-isopropylmalate dehydrogenase from Thermus thermophilus was analyzed by chemical modification and site-directed mutagenesis, and His273 of the enzyme was identified to be involved in the coenzyme binding.

Three‐dimensional structure prediction of the NAD binding site of proton‐pumping transhydrogenase from Escherichia coli

A three-dimensional structure of the NAD site of Escherichia coli transhydrogenase has been predicted. The model is based on analysis of conserved residues among the transhydrogenases from five different sources, homologies with enzymes using NAD as cofactors or substrates, hydrophilicity profiles, and secondary structure predictions. The present model supports the hypothesis that there is one binding site, located relatively close to the N-terminus of the alpha-subunit. The proposed structure spans residues alpha 145 to alpha 287, and it includes five beta-strands and five alpha-helices oriented in a typical open twisted alpha/beta conformation. The amino acid sequence following the GXGXXG dinucleotide binding consensus sequence (residues alpha 172 to alpha 177) correlates exactly to a typical fingerprint region for ADP binding beta alpha beta folds in dinucleotide binding enzymes. In the model, aspartic acid alpha 195 forms hydrogen bonds to one or both hydroxyl groups on the adenosine ribose sugar moiety. Threonine alpha 196 and alanine alpha 256, located at the end of beta B and beta D, respectively, create a hydrophobic sandwich with the adenine part of NAD buried inside. The nicotinamide part is located in a hydrophobic cleft between alpha A and beta E. Mutagenesis work has been carried out in order to test the predicted model and to determine whether residues within this domain are important for proton pumping directly. All data support the predicted structure, and no residue crucial for proton pumping was detected. Since no three-dimensional structure of transhydrogenase has been solved, a well based tertiary structure prediction is of great value for further experimental design in trying to elucidate the mechanism of the energy-linked proton pump.

Rho-ADP-ribosylating exoenzyme from Bacillus cereus. Purification, characterization, and identification of the NAD-binding site.

The ADP-ribosyltransferase produced by a pathogenic strain of Bacillus cereus was purified to near homogeneity. The transferase is a 28,000 Da molecular mass enzyme with a pI of 10.3. The specific enzyme activity is 7.0 nmol of ADP-ribose min-1 mg-1 with a Km for NAD of 0.3 microM. Partial amino acid sequence analysis of the exoenzyme reveals no significant homology to Clostridium botulinum C3 nor to Clostridium limosum exoenzyme. The novel exoenzyme selectively modifies the small GTP-binding proteins of the Rho family presumably at the same acceptor amino acid (Asn-41) as determined for C3. Besides cellular Rho, recombinant RhoA and -B are substrates for the exoenzyme. However, recombinant Rac1 and CDC42, although belonging to the Rho family, are not modified. B. cereus exoenzyme was photolabeled with [carbonyl-14C]NAD resulting in inhibition of ADP-ribosyltransferase and NAD-glycohydrolase activity. A glutamic acid residue was identified as part of the NAD-binding site which corresponds to Glu-174 of C3. This glutamic acid is located in a domain which shows high homology with the C-terminal part of C3 exoenzyme, C. limosum exoenzyme, and Staphylococcus aureus EDIN and which probably represents the catalytic site of the transferases. The data indicate that B. cereus exoenzyme is a novel member of the family of C3-like ADP-ribosyltransferases which share the same substrate protein Rho and which have an identical highly conserved catalytic domain.

Common features of the NAD‐binding and catalytic site of ADP‐ribosylating toxins

Computer analysis of the three-dimensional structure of ADP-ribosylating toxins showed that in all toxins the NAD-binding site is located in a cavity. This cavity consists of 18 contiguous amino acids that form an alpha-helix bent over a beta-strand. The tertiary folding of this structure is strictly conserved despite the differences in the amino acid sequence. Catalysis is supported by two spatially conserved amino acids, each flanking the NAD-binding site. These are: a glutamic acid that is conserved in all toxins, and a nucleophilic residue, which is a histidine in the diphtheria toxin and Pseudomonas exotoxin A, and an arginine in the cholera toxin, the Escherichia coli heat-labile enterotoxins, the pertussis toxin and the mosquitocidal toxin of Bacillus sphaericus. The latter group of toxins presents an additional histidine that appears important for catalysis. This structure suggests a general mechanism of ADP-ribosylation evolved to work on different target proteins.

Anatomy of an engineered NAD‐binding site

The coenzyme specificity of Escherichia coli glutathione reductase was switched from NADP to NAD by modifying the environment of the 2'-phosphate binding site through a set of point mutations: A179G, A183G, V197E, R198M, K199F, H200D, and R204P (Scrutton NS, Berry A, Perham RN, 1990, Nature 343:38-43). In order to analyze the structural changes involved, we have determined 4 high-resolution crystal structures, i.e., the structures of the wild-type enzyme (1.86 A resolution, R-factor of 16.8%), of the wild-type enzyme ligated with NADP (2.0 A, 20.8%), of the NAD-dependent mutant (1.74 A, 16.8%), and of the NAD-dependent mutant ligated with NAD (2.2 A, 16.9%). A comparison of these structures reveals subtle differences that explain details of the specificity change. In particular, a peptide rotation occurs close to the adenosine ribose, with a concomitant change of the ribose pucker. The mutations cause a contraction of the local chain fold. Furthermore, the engineered NAD-binding site assumes a less rigid structure than the NADP site of the wild-type enzyme. A superposition of the ligated structures shows a displacement of NAD versus NADP such that the electron pathway from the nicotinamide ring to FAD is elongated, which may explain the lower catalytic efficiency of the mutant. Because the nicotinamide is as much as 15 A from the sites of the mutations, this observation reminds us that mutations may have important long-range consequences that are difficult to anticipate.

Crystal structure of a NAD-dependent d-glycerate dehydrogenase at 2·4 Å resolution

d-Glycerate dehydrogenase (GDH) catalyzes the NADH-linked reduction of hydroxypyruvate to d-glycerate. GDH is a member of a family of NAD-dependent dehydrogenases that is characterized by a specificity for the d-isomer of the hydroxyacid substrate. The crystal structure of the apoenzyme form of GDH from Hyphomicrobium methylovorum has been determined by the method of isomorphous replacement and refined at 2·4 Å resolution using a restrained least-squares method. The crystallographic R-factor is 19·4% for all 24,553 measured reflections between 10·0 and 2·4 Å resolution. The GDH molecule is a symmetrical dimer composed of subunits of molecular mass 38,000, and shares significant structural homology with another NAD-dependent enzyme, formate dehydrogenases The GDH subunit consists of two structurally similar domains that are approximately related to each other by 2-fold symmetry. The domains are separated by a deep cleft that forms the putative NAD and substrate binding sites. One of the domains has been identified as the NAD-binding domain based on its close structural similarity to the NAD-binding domains of other NAD-dependent dehydrogenases. The topology of the second domain is different from that found in the various catalytic domains of other dehydrogenases. A model of a ternary complex of GDH has been built in which putative catalytic residues are identified based on sequence homology between the d-isomer specific dehydrogenases. A structural comparison between GDH and l-lactate dehydrogenase indicates a convergence of active site residues and geometries for these two enzymes. The reactionscatalyzed are chemically equivalent but of opposing stereospecificity. A hypothesis is presented to explain how the two enzymes may exploit the same coenzyme stereochemistry and a similar spatial arrangement of catalytic residues to carry out reactions that proceed to opposite enantiomers.