Complete The Aldolase Reaction Of Glycolysis By Drawing The Product Or Products.

Abstract

Background Aldolases catalyze a variety of condensation and cleavage reactions, with exquisite control on the stereochemistry. These enzymes, therefore, are attractive catalysts for synthetic chemistry. There are two classes of aldolase: class I aldolases utilize Schiff base formation with an active-site lysine whilst class II enzymes require a divalent metal ion, in particular zinc. Fructose-1,6-bisphosphate aldolase (FBP-aldolase) is used in gluconeogenesis and glycolysis; the enzyme controls the condensation of dihydroxyacetone phosphate with glyceraldehyde-3-phosphate to yield fructose-1,6-bisphosphate. Structures are available for class I FBP-aldolases but there is a paucity of detail on the class II enzymes. Characterization is sought to enable a dissection of structure/activity relationships which may assist the construction of designed aldolases for use as biocatalysts in synthetic chemistry.

Results The structure of the dimeric class II FBP-aldolase from Escherichia coli has been determined using data to 2.5 å resolution. The asymmetric unit is one subunit which presents a familiar fold, the (α/β)₈ barrel. The active centre, at the C-terminal end of the barrel, contains a novel bimetallic-binding site with two metal ions 6.2 å apart. One ion, the identity of which is not certain, is buried and may play a structural or activating role. The other metal ion is zinc and is positioned at the surface of the barrel to participate in catalysis.

Conclusions Comparison of the structure with a class II fuculose aldolase suggests that these enzymes may share a common mechanism. Nevertheless, the class II enzymes should be subdivided into two categories on consideration of subunit size and fold, quaternary structure and metal-ion binding sites.

Keywords

class II fructose-1,6-bisphosphate aldolase
crystal structure
metalloenzyme
zinc enzyme

Introduction

Fructose-1,6-bisphosphate aldolase (FBP-aldolases; EC4.1.2.13) catalyzes the cleavage of Fructose-1,6-bisphosphate (FBP) to a ketose, dihydroxyacetone phosphate (DHAP), and an aldose, glyceraldehyde-3-phosphate G3P), in glycolysis. FBP-aldolase also catalyzes the reverse condensation reaction in gluconeogenesis [

] (Figure 1). In each pathway this reaction represents a distinctive stage where a switch from six to three carbon units, or vice versa, occurs.

The aldol condensation is an important reaction in synthetic chemistry, although the more complicated the desired product the less efficient the reaction due to poor control over stereochemistry. Aldolases carry out specific conden-sations with exquisite control on the stereochemistry. In addition, because aldolases can be regulated depending on the concentrations of reactants and products, these enzymes are attractive as catalysts for synthetic chemistry. Indeed, the utilization of aldolases is already proving useful in a number of areas, in particular the synthesis of compounds of relevance to the pharmaceutical industry [

Wong C.-H.
Whitesides G.M.

Enzymes in synthetic organic chemistry.

Google Scholar

von der Osten C.H.
Wong C.-H.
et al.

Use of a recombinant bacterial fructose-1,6-bisphosphate aldolase in aldol reactions: preparative synthesis of 1-deoxynojirimycin, 1-deoxymannojirimycin, 1,4-dideoxy-1,4-imino-D-arabinitol, and fagomine.

Crossref
Scopus (193)
Google Scholar

Aldolases have been classified into two groups, class I and class II, on the basis of their reaction mechanism. The class I enzymes use an active-site lysine which stabilizes a reaction intermediate via Schiff base formation [

Wong C.-H.
Whitesides G.M.

Enzymes in synthetic organic chemistry.

Google Scholar

]. The class I aldolases are predominantly found in higher order organisms and are usually homotetramers with a molecular weight of about 160 kDa for the complete assembly. A number of crystal structures of class I enzymes have been determined including the FBP-aldolases from rabbit [

], Drosophila [

] and human [

]. Very recently the dimeric Escherichia coli transaldolase B structure was published [

]. In addition, the structures of a tetrameric N-acetylneuraminate lyase [

] and the trimeric bacterial phosphogluconate aldolase [

[9]

Lebioda L.
Hatada M.H.
Tulinsky A.
Mavaridis I.M.

Comparison of the folding of 2-keto-3-deoxy-6-phosphogluconate aldolase, triosephosphate isomerase and pyruvate kinase.

Crossref
PubMed
Scopus (27)
Google Scholar

] have been characterized. Although there is variation in the subunit orientation and interactions in the formation of the oligomeric assembly of the different aldolases, each subunit displays the same (α/β)₈ barrel fold. This fold was first observed in triosephosphate isomerase [

[10]

Banner D.W.
Waley S.G.
et al.

Structure of chicken muscle triosephosphate isomerase determined by crystallography at 2.5 å resolution using amino acid sequence data.

Crossref
PubMed
Scopus (585)
Google Scholar

] and is now recognized as the most frequently occurring protein fold [

]. Biochemical investigations designed to help understand class I aldolase substrate recognition and catalysis have complemented the structural studies [

[12]

Morris A.J.
Davenport R.C.
Tolan D.R.

A lysine to arginine substitution at position 146 of rabbit aldolase A changes the rate-determining step to Schiff base formation.

Crossref
PubMed
Scopus (18)
Google Scholar

By way of contrast, very little is known about the class II aldolases. The class II FBP-aldolases function as homodimers with a total molecular weight of around 78 kDa; they show only about 15 % sequence identity with the class I enzymes [

]. There is an absolute requirement for a divalent metal ion, usually zinc, and in addition the enzymes are activated by monovalent cations such as potassium [

]. The sequences of several class II FBP-aldolases have recently become available [

[14]

Qamar S.
Marshal K.
Berry A.

Identification of arginine 331 as an important active site residue in the class II fructose-1,6-bisphosphate aldolase of Escherichia coli.

Crossref
PubMed
Scopus (32)
Google Scholar

]. There has been speculation on the evolution of aldolases in particular given the two distinct types of FBP-aldolase that carry out the same reaction [

There is only one crystal structure of a class II enzyme, that of the fuculose-1-phosphate aldolase (FucA) from E. coli [

]. There is no appreciable sequence identity between FucA and the FBP-aldolases.

We have set out to characterize the structure, specificity and mechanism of the class II FBP-aldolase from E. coli. A full understanding of the structural features involved in substrate recognition and processing may guide site-directed mutagenesis experiments to create enzymes of altered specificity for use in synthetic chemistry. We report here a crystal structure analysis and detail aspects of the protein structure. In addition, we present comparisons of this new structure with other metalloenzymes, in particular the FucA system. We discuss the results from previous studies which described residues implicated in metal binding and substrate recognition.

Results and discussion

Comments on the model

The refinement protocol has produced a model of acceptable stereochemistry, as judged by standard criteria in protein crystallography, and relevant indicators are presented in the Materials and methods section. The model consists of three polypeptide segments, residues 1–176, 194–228 and 230–356. There is no convincing electron density for residues 177–193, 229 or the last two residues in the sequence. Two cations (each treated as zinc, one of full occupancy, one of 0.6 occupancy) and 70 solvent positions, treated as water, have been included in the model that we present.

The protomer

The subunit is an (α/β)₈ barrel fold commonly observed in many different types of proteins [

]. The barrel has a diameter of around 45 å and a height of 25 å (Figure 2). The core of the barrel is circular in cross-section, and is more reminiscent of the class I aldolase and N-acetylneuraminate lyase barrels [

] than of the more elliptical types of (α/β)₈ barrel structures. Secondary structure assignments have been made with a combination of facilities in the programs PROCHECK [

[18]

Laskowski R.A.
MacArthur M.W.
Moss D.S.
Thornton J.M.

PROCHECK: a program to check the stereochemical quality of protein structures.

Crossref
Google Scholar

], O [

], DSSP [

] and by visual inspection. Secondary structure elements are mapped onto the amino acid sequence in Figure 3. Approximately 47 % of the residues in the model are in α helical conformations and 18 % are in β strands. In addition to the assigned elements of secondary structure, there are five sections of sequence which present an extended conformation. These sections involve the residues at positions 2–5, 12–14, 234–238, 310–312 and 323–326. The core of the structure consists of an eight-stranded parallel β-strand assembly, with strands labelled β1 to β8. When viewed from the C-terminal end of the β barrel, the twist of strand–helix–strand winds in an anticlockwise direction. The helices (termed α1–α11) that accompany each β strand in creating the barrel are helices α2, α4–α9 and α11. Helix α1 caps the N-terminal end of the barrel, helix α3 is a short segment linking strand β2 with helix α4 whilst helix α10 runs away from strand β8 to a loop region. Finally, the polypeptide changes direction to form the last helix, α11. Helices α10 and α11 are antiparallel with respect to each other and create an arm from the barrel which is important in dimer formation. Helix α11 contains residue Arg331, a residue which plays a critical role in substrate binding [

[14]

Qamar S.
Marshal K.
Berry A.

Identification of arginine 331 as an important active site residue in the class II fructose-1,6-bisphosphate aldolase of Escherichia coli.

Crossref
PubMed
Scopus (32)
Google Scholar

]. As with other (α/β)₈ barrel structures the active site is located in the depression at the C-terminal end of the barrel and will be detailed shortly.

Figure thumbnail gr2 — Figure 2 The overall fold of FBP-aldolase. **(a)** Stereo view Cα trace of the protomer viewed from the C-terminal end of the (α/β)₈ barrel. Every 20th residue in the model is labelled, as are the N and C termini. β Strands and loops are shown in black, α helices are in red; the two metal ion positions are depicted as spheres labelled 1 (blue) and 2 (white). **(b)** A schematic drawing showing the secondary structure of FBP-aldolase with the same view as (a); the colour scheme is the same as in (a). (Figure produced with MOLSCRIPT [54].)

View Large Image

Figure Viewer

Download Hi-res image

Download (PPT)

Figure thumbnail gr3 — Figure 3 The amino acid sequence of *E. coli* class II FBP-aldolase [55]. Secondary structure elements have been assigned as in Figure 2: ← → indicates β strands and ≈≈≈ indicates α helices. Every tenth residue is underlined and the metal ligands are shown in bold (Asp109, His110, Glu172, Glu174, His226, His264 and Lys284).

View Large Image

Figure Viewer

Download Hi-res image

Download (PPT)

Dimer assembly and the location of Arg331

Class II FBP-aldolase functions as a homodimer. The molecular twofold axis coincides with a crystallographic axis and the symmetry operation +x, x−y, 1/6−z generates the partner subunit. The dimer is approximately 100 å in length with the β barrels positioned almost at right angles to each other (Figure 4). The dimer interface is extensive and solvent accessibility calculations, using the algorithm of Lee and Richards [

] as coded in X-PLOR [

], indicate that dimerization buries around 2700 å² from the protein surface of each subunit.

Figure thumbnail gr4 — Figure 4 Schematic diagram of the FBP-aldolase dimer. The view is similar to that shown in Figure 2, perpendicular to the molecular twofold axis. Helices of one subunit are depicted as red ribbons, and helices of the partner subunit are depicted as yellow ribbons; β strands and loops are shown in black. Arg331 is shown in dark blue CPK representation; the two metal ion positions, M1 and M2, are labelled 1 and 2 and shown in blue and white, respectively. (Figure produced using MOLSCRIPT [54].)

View Large Image

Figure Viewer

Download Hi-res image

Download (PPT)

The areas of intersubunit contacts that are important in stabilizing the dimer all involve helix–helix interactions. The most extensive interactions involve the last two α helices. These two helices (α10 and α11) run antiparallel to each other interacting through van der Waals contacts and a number of hydrogen bonds plus a salt bridge formed between Arg344 and Asp290. This two helix structure then self-associates with the equivalent structure on the partner subunit through the C- and N-terminal regions of α10 and α11, respectively. The interactions produce a substantial hydrophobic core at the dimer interface; helix α11 also interacts with the short α3 helix of the adjacent subunit. Another area involved in dimerization comprises contacts between residues on the α4 helices of each subunit. These helices, which are a component of the (α/β)₈ barrels, are aligned antiparallel to each other.

The use of site-directed mutagenesis in combination with arginine specific α-dicarbonyl reagents has identified Arg331 as an important residue in terms of substrate recognition [

[14]

Qamar S.
Marshal K.
Berry A.

Identification of arginine 331 as an important active site residue in the class II fructose-1,6-bisphosphate aldolase of Escherichia coli.

Crossref
PubMed
Scopus (32)
Google Scholar

]. This residue is absolutely conserved in the class II FBP-aldolases and is implicated in binding the C-6 phosphate moiety of FBP, the G3P component (see Figure 1). The crystal structure shows that this residue is located at the N terminus of helix α11 of one subunit, directed into the active site and approximately 13 å from the catalytic site of the other subunit. The formation of the dimer positions Arg331 in such a way that it is able to bind G3P in the active site of the partner subunit. The DHAP component will therefore be positioned at the other end of the active site near metal site 2, which is the catalytic zinc ion.

During the course of the refinement we identified a strong electron-density peak near Arg331. This has been included in the final model as a solvent molecule, and is located 4.1 å and 3.8 å from Arg331 NH1 and NH2, respectively. This peak probably represents a partially ordered sulphate ion bound into the C-6 phosphate-binding site sequestered during an ammonium sulphate precipitation step in the enzyme purification procedure.

A novel bimetallic active site

The active sites in (α/β)₈ barrels are used in a diverse range of biochemical reactions [

] and there are numerous proteins which create binding sites for and utilize metal ions, either single or multiple ions, in these active centres. However, although there are similarities with some structural features of other metalloenzymes, the E. coli class II aldolase presents a novel bimetallic-binding site. One metal ion is buried, termed M1, the other is on the surface of the active site and is referred to in the text as M2. We cannot unambiguously identify the chemical nature of both metals although we are confident about the assignment of the metal M2 as the catalytic zinc. M1 may be potassium or zinc, some other cation or even a mixture of cations and discussion on this point is warranted.

Previous studies, (e.g. [

]) indicate that the ratio of zinc to monomer is 1:1 and we had expected to find a single zinc site. In addition, this aldolase is activated by monovalent cations, in particular by potassium [

], so the location of an activation site was also sought. Interestingly, two strong peaks of electron density were located in the active centre. Towards the end of the refinement procedure, omit difference Fouriers were calculated with either all sulphur atoms (from six methionines and three cysteines) or the two metal ions removed from the calculated structure factors. The average height of a sulphur atom in an F_o–F_c map is 4.2 eå³; for the metals, M1 has a height of 5.7 and M2 a height of 6.3 eå³. We have made the assignment of one zinc atom, Zn2, of full occupancy. M2 has been refined as zinc with 0.6 occupancy based on a comparison of electron-density peak height in a difference Fourier (Figure 5) in combination with the chemical considerations detailed below and taking into account results from refinements (see Materials and methods). The thermal parameters of M1 and M2 are 32 and 40 å², respectively. Both thermal parameters are slightly less than that observed for the liganding nitrogens on the chelating His264.

Figure thumbnail gr5 — Figure 5 Stereo views depicting the active site and metal ion positions. **(a)** Active site and metals with an associated F_o–F_c omit map (blue chicken wire), where the metal ions have been removed from the F_c terms. The map is contoured at a 7.5σ level; metal coordination is indicated by green dashed lines. **(b)** In a similar orientation to (a), the 2F_o–F_c electron-density map contoured at a 1.6σ level. **(c)** Schematic of metal sites plus ligands; distances are given in å and dashed lines indicate probable metal–ligand coordination. The charges of metals and ligands in M1 and M2 balance if we assume the presence of divalent metals, a neutral lysine and the imidazolate.

View Large Image

Figure Viewer

Download Hi-res image

Download (PPT)

The chemistry of K⁺ and Zn²⁺

Distances for K⁺ and Zn²⁺ ions to N and O ligands have been surveyed by searching the Cambridge Crystallographic Database [

[24]

Allen F.H.
Kennard O.

3D search and research using the Cambridge Crystallographic Database.

Google Scholar

]. The results of these searches are presented in Table 1 and provide a useful reference when considering the chemistry of the cations and the environment presented by the protein.

Table 1 A survey of the Cambridge Crystallographic Data Centre indicating metal–ligand distances for potassium and zinc cations coordinating N and O atoms.

Interaction		Number of observations	Range (å)	Mean^∗ (å)
K⁺	N^†	160	2.48–3.45	2.921(12)
K⁺	O^†	1040	2.36–3.20	2.821(3)
Zn²⁺	N^†	1612	1.81–2.55	2.092(2)
Zn²⁺	O^†	909	1.72–2.90	2.091(4)

K⁺	N^‡	12	2.75–2.95	2.870(18)
K⁺	O^‡	12	2.61–2.78	2.689(13)
Zn²⁺	N^‡	689	1.89–2.41	2.048(2)
Zn²⁺	O^‡	453	1.86–2.39	1.991(4)

^∗Standard deviation of the mean is given in parenthesis. ^†Metal–ligand with no restriction on the geometry subtended at the cation.

^‡Metal–ligand distances only considering tetrahedral coordination.

Open table in a new tab

The K⁺ ion, with an ionic radii of 1.33 å, generally adopts six to eightfold coordination. There are very few exam-ples of tetrahedral coordination. It is a class (a) hard acid cation preferring to interact with hard basic groups such as carboxylates. There are fewer crystal structures with K⁺ to N interactions than K⁺ to O. The Zn²⁺ ion (ionic radii of 0.69 å) is an intermediate hard/soft acid, frequently adopts tetrahedral geometry and coordinates hard or soft basic groups, O or N ligands.

Metal 1

M1 is effectively buried (Figure 2) and coordinated with a distorted tetrahedral geometry. The ligands are contributed from separate β strands and comprise Asp109 (β3), bidentate Glu172 (β5) which like the metal is buried, His264 (β7) and Lys284 (β8). The metal ligand distances range from 2.5 to 2.9 å. Asn286 is nearby with distances of 3.7 å and 3.3 å separating the Oδ1 and Nδ2, respectively, from the ion. The M1 ligand distances are closer to the mean values observed for K⁺ than to Zn²⁺, see Table 1. However, given the limited resolution of this structure, caution should be exercised in attributing too much significance to these distances; the distances represent a good indicator of the limitations of the molecular model.

That M1 might represent the activating K⁺ is an attractive proposition. It would mean that K⁺ is playing a structural role helping to create and/or stabilize the catalytic M2. However, the protein environment does not appear to be a particularly good K⁺-binding site. On the basis of the chemistry the site appears more suited for binding zinc or NH₄ ⁺. M1 may in fact represent a mixture of cations including some NH₄ ⁺, given that ammonium sulphate precipitation was used in the enzyme preparation.

One complicating factor is that our hexagonal crystal form is obtained in the presence of between 1.6 and 2.5 mM zinc chloride, which approximates to a tenfold excess of zinc to enzyme monomer, [

] whereas other crystal forms are observed without the addition of zinc [

]. We expect that with a tenfold excess of zinc ions the metal will be able to bind in any suitable environment presented by the enzyme. Studies using this aldolase in synthetic preparative methods [

[3]

von der Osten C.H.
Wong C.-H.
et al.

Crossref
Scopus (193)
Google Scholar

] noted that, in the absence of zinc, the enzyme was inactive after five days. However, in the presence of millimolar concentrations of zinc, enzyme activity was fully maintained for a period of at least ten days [

[3]

von der Osten C.H.
Wong C.-H.
et al.

Crossref
Scopus (193)
Google Scholar

]. Our structure may therefore represent an active and stable form of the enzyme; the form actually being used in preparative methods.

The close proximity (2.7 å) of Lys284 Nζ to M1 was a surprise. The only other enzyme which has a normal lysine coordinating a metal ion is the co-catalytic zinc enzyme, leucine aminopeptidase [

[26]

Burley S.K.
David P.R.
Sweet R.M.
Taylor A.
Lipscomb W.N.

Structure determination and refinement of bovine lens leucine aminopeptidase and its complex with bestatin.

Crossref
PubMed
Scopus (206)
Google Scholar

]. In Pseudomonas diminuta phosphotriesterase, a co-catalytic zinc site is formed in the (α/β)₈ barrel. In this site the binuclear metals are bridged by a carbamate functional group of a modified lysine [

[27]

Vanhooke J.L.
Benning M.M.
Raushel F.M.
Holden H.

Three-dimensional structure of the zinc-containing phosphotriesterase with the bound substrate analog diethyl 4-methylbenzylphosphonate.

Crossref
PubMed
Scopus (236)
Google Scholar

]. There is no evidence that Lys284 in FBP-aldolase has been modified in such a way and we conclude that it must be neutral, coordinating the metal through a lone pair. Although buried, this lysine is in a hydrophilic cavity which includes two solvent molecules and is lined by residues Gln59, His107, Ser140, Thr170, Glu172, Thr217 and Asn286. Gln59 and Asn286 are of note as they are both positioned to accept hydrogen bonds donated from Lys284. The hydrophobic component of the lysine side chain is positioned between Ile57 on one side and Val262 on the other.

Site-directed mutagenesis experiments had previously been carried out to identify possible zinc ligands [

] and His110 was correctly identified as such, see later. The fact that the mutation of His107 disrupted enzyme functionis intriguing. His107 is in the hydrophilic cavity, 4 å from both M1 and Lys284. Any perturbation of the structure at this position would be expected to influence coordination at M1, and in turn to perturb M2, thus influencing the enzyme kinetics.

Metal 2

M2 is at the surface of the β barrel and represents the catalytic metal, Zn²⁺. It also presents a distorted tetrahedral geometry being coordinated to three histidines (His110, His226 and His264) and Glu174 (bidentate). Metal–ligand distances are in the range 2.4 to 2.8 å. These values are longer than expected for Zn–O and Zn–N interactions and this is probably a consequence of the limited resolution of the structure. Three of these ligands are formed with residues located in β strands: His110 and His264 are associated with β3 and β7, respectively, and Glu174 with β5. His226 is located on an 18 residue loop linking β6 with α8; His264, the imidazolate, ligates both metals. Similar geometry to that observed at this M2 site has been observed in other enzymes, for example, the (α/β)₈ barrel adenosine deaminase [

[28]

Wilson K.
Rudolph F.B.
Quiocho F.A.

Atomic structure of adenosine deaminase complexed with a transition-state analog: understanding catalysis and immunodeficiency mutations.

Crossref
PubMed
Scopus (425)
Google Scholar

], and the class II fuculose-1-phosphate aldolase [

]. The latter enzyme is directly related to our study and a separate section comparing these two class II aldolases will be presented. In advance of that comparison it is appropriate to compare the bimetallic site with other zinc containing enzymes.

Zinc containing enzymes can be placed into two groups based on the number of metal ions: the mono-zinc systems [

] and the more recently defined co-catalytic, or coactive, sites [

]. This latter group of enzymes utilize two or more metals, always have a bridging acidic residue and often involve a water or hydroxide ion [

]. In general the distance separating the metals is less than 5.0 å. The class II FBP-aldolase has a two-metal site (Figure 5), the metals are bridged by an imidazolate (His264) and are 6.2 å apart. This arrangement bears a strong similarity to a copper, zinc superoxide dismutase structure [

], where the bridging histidine interacts with a copper and a zinc atom at a similar separation. The similarity extends to the coordination of the zinc which is achieved through interactions with three histidines and the carboxylate group of an acidic residue (aspartate in the dismutase, glutamate in the aldolase).

Most catalytic zinc sites have four ligands which includea water molecule [

]. Our structure does not have a water molecule coordinated to the catalytic zinc and if an activated solvent molecule is involved in catalysis then some rearrangement of ligands would have to occur. This rearrangement could involve an increase in the coordination number of the metal, which for zinc would be quite acceptable [

Overlap of class I and class II FBP-aldolases

Superpositions of the class II aldolase subunit structure with a subunit of a class I aldolase have been carried out using the program LSQMAN (GJ Kleywegt and TA Jones, personal communication). As the representative of the class I family we selected the Drosophila structure [

] as being the most reliable available in the Brookhaven Protein Databank. Alignment using the central portions of the β strands 1–8 of the class II enzyme with the corresponding strands of the class I aldolase, gave a root mean square (rms) deviation of 2.3 å for 124 Cα atoms; a value similar to that observed in other comparisons of (α/β)₈ proteins [

]. Key residues in each structure are located close to each other when the enzymes are superimposed. Two lysines residues (Lys146 and Lys229) are important for the class I system: Lys146 is essential for catalysis and is positioned to interact with the substrate; Lys 229 is the Schiff base-forming residue. The Nζ atoms of Lys146 and Lys229 are within 1 å of the Nϵ2 atoms of two of the catalytic zinc liganding residues, His110 and His264, respectively. Alternative overlaps, varying the strand alignments, decrease the number of Cα atoms that overlap within 2.5 å and do not align residues implicated in the enzyme mechanism. This suggests that the circular permutation likely to have occurred within the class I aldolase family [

] is not applicable to the class II system, assuming of course that they have evolved from a common ancestor.

Comparisons with FucA: comments on the class II mechanism

The only other class II aldolase, for which a structure has so far been published, is FucA from E. coli [

]. Indeed, this is the only aldolase, class I or II, for which the structural detail of a complex with inhibitor or substrate analogue is published [

]. This enzyme is a homotetramer with a subunit molecular weight of approximately 28 kDa. The subunits are arranged with C4 symmetry and the protein fold consists of a central nine-stranded β-pleated sheet flanked by two helices on either side of the sheet. On the basis of different folds it is evident that the class II aldolases could be subdivided into two groups: the tetrameric type with a single zinc-binding site per subunit and the dimeric (α/β)₈ barrel type encompassing a bimetallic site on each subunit. These aldolases are thus analogous to superoxide dismutases which are divided into the bimetallic copper/zinc type and the single metal (iron or manganese) containing enzymes; each type of dismutase presents a distinct fold [

Each FucA subunit active site binds a zinc atom using three histidines and a glutamic acid in the same way that M2 of the FBP-aldolase is liganded. Phosphoglycolohydroxamate simulates the enediolate transition state of the substrate DHAP and has been shown to be a potent inhibitor of zinc aldolases [

[36]

Collins K.D.

An activated intermediate analogue. The use of phosphoglycolohydroxamate as a stable analogue of a transiently occurring dihydroxyacetone phosphate-derived enolate in enzymatic catalysis.

Abstract
Full Text PDF
PubMed
Google Scholar

]. The structure of the complex of this inhibitor with FucA indicates that DHAP binds directly to the catalytic zinc replacing the glutamic acid from the coordination sphere [

]. Given the similarities in structural and reaction chemistry the obvious conclusion is that the same mechanism could be utilized by both of these class II aldolases, although aspects of substrate specificity would vary. This infers that, during the catalytic cycle of the class II FBP-aldolase, the liganding Glu174 moves away from the zinc to allow the DHAP component to coordinate the metal ion; there is no need to implicate an activated water molecule.

In both enzymes, zinc would function as a Lewis acid electron sink and polarize the carbonyl bond of ketose substrates ready for the cleavage or condensation reaction. This would be in agreement with spectroscopic studies on manganese (II) substituted yeast class II FBP-aldolase [

]. Direct coordination of the substrate to the metal could assist the precise alignment of reactants for catalysis. The presence of the imidazolate ligand, His264, with the ability for delocalization of charge, may also assist catalysis by further increasing the polarization capabilities of the metal-binding site.

It is of note that, based on the location of the G3P phosphate-binding residue Arg331, for the DHAP component to reach the catalytic zinc in the FBP-aldolase it would pass over the top of the active centre of the (α/β)₈ barrel. This would bring functional groups into range to interact with two basic residues, namely His110 and His226, two of the zinc ligands. His110, is spatially in a similar position to the class I aldolase Lys146.

Previous studies on the interaction of the yeast class II FBP-aldolase with substrates used spectroscopic methods [

]. Earlier studies implicated that the zinc directly binds to substrate, as mentioned previously. This view was subsequently changed so that an imidazole zinc ligand was put forward as the enzyme component that interacted directly with the polarized carbonyl group of the substrate [

]. The FucA–inhibitor structure supports the earlier conclusions for the enzyme reaction mechanism with the caveat that although the catalytic zinc coordination is conserved between the enzymes, FucA is not the aldolase that was studied. The superposition of the class I and class II enzymes positions the basic residues in each active site in such a way as to suggest a similarity in substrate recognition. The superposition also supports His110 as the intervening imidazole ligand used by the class II FBP-aldolase. However, the detailed mechanism as to how the class II FBP-aldolase recognizes and acts upon the substrate remains as yet unresolved. We cannot, for example, rule out some conformational change that alters the position of the catalytic zinc. Experiments designed to position substrate components and inhibitors in the FBP-aldolase active site, so as to address aspects of enzyme-specificity and mechanism, are now imperative. In this way it is hoped to determine which mechanism applies to the class II FBP-aldolases.

Heavy atom binding: is this a clue for the cation activation site?

Three heavy-atom derivatives were used to provide initial phase information leading to the structure determination (see Materials and methods). As one of our validation criteria, we ensured that these derivative binding sites are in chemically reasonable positions. As the class II FBP-aldolase is activated by monovalent cations so the identification of metal-binding sites may provide clues about cation activation.

The platinum soak produced three binding sites: one is located within 3 å of both His94 and His129; the second is 2 å from His252; and the third is located in a crevice between α9, α11 and β8 (3 å from the Sδ of Met285 and 5 å from the carboxylate of Glu351). The lead derivative produced two sites: a minor site which is 2.4 å from Glu155 and which like the three platinum sites is remote from the active site; and a major site near the active centre.

Many of the derivative soaks produced an identical single binding site and were therefore not suitable for phasing. This binding site is located near Asp144 and Ser146, two residues which are conserved in other class II FBP-aldolases; the major lead site and the single samarium site are located here. In particular, the lead ion is 2.3 å from Asp144 and 3 å from Ser146. This metal-binding site is only about 4 å from the catalytic centre (Figure 5) and offers an alternative to M1 as the cation-activation site. If this is correct, then the role of the monovalent cation could be to assist in substrate binding and may help to achieve precise orientation of the C-1 phosphate. Such use of a monovalent cation as an activator has a precedent in fructose bisphosphatase [

[39]

Zhang R.
Villeret V.
Lipscomb W.N.
Fromm H.J.

Kinetics and mechanism of activation and inhibition of porcine liver fructose-1,6-bisphosphatase by monovalent cations.

Crossref
PubMed
Scopus (47)
Google Scholar

], the enzyme down stream in gluconeogenesis. Apart from this site, in the area around the active centre and Asp144, there is only one positively charged residue, Lys113, which may assist substrate binding.

Biological implications

Aldolases catalyze a variety of aldol condensation or cleavage reactions and are divided into two groups, class I and II. The class I aldolases use an active-site lysine residue in Schiff base formation whilst the class II enzymes are metal-dependent. The tetrameric class I enzymes all have an (α/β)₈ barrel subunit structure. We have determined the structure of the dimeric, class II fructose-1,6-bisphosphate (FBP-aldolase) which, although displaying the familiar (α/β)₈ barrel, presents a novel bimetallic active centre. One metal ion, which may represent an activating cation site, is buried. The other metal ion, the catalytic zinc, at the surface of the barrel is coordinated by three histidine residues (His110, His226 and His264) and Glu174. His264 acts as a bridging ligand between the two cations.

Only one other class II aldolase structure has been determined, that of a fuculose-1-phosphate aldolase [

]. This enzyme is a tetramer with a fold distinct from the (α/β)₈ barrel. Each subunit binds a single zinc ion with the same coordination as noted for the catalytic zinc in the FBP-aldolase bimetallic site. A comparison suggests that although the protein structures are different the same reaction mechanism may apply.

The potential of aldolases, in particular the more stable class II enzymes, in synthetic chemistry is widely recognized and there are numerous examples where they have proven effective in bio-transformations and synthetic organic chemistry. Aldolases have been used to catalyze the formation of sugars (¹³C labelled, N or F containing, deoxysugars, and novel high carbon sugars) and glycosidase inhibitors, such as deoxynojirmycin [

Wong C.-H.
Whitesides G.M.

Enzymes in synthetic organic chemistry.

Google Scholar

von der Osten C.H.
Wong C.-H.
et al.

Crossref
Scopus (193)
Google Scholar

]. Armed with a knowledge of the structural features that determine aldolase specificity and mechanism, in combination with site-directed mutagenesis it may be viable to engineer aldolases with defined specificity to catalyze selected reactions. The structure determination presented here of the class II FBP-aldolase has produced the template for use in just such a design process.

Materials and methods

Crystallization and derivitization

Enzyme samples were prepared as detailed previously [

]. Three crystal forms are available [

] and the structure determination was carried out on a hexagonal form in space group P6₁22, with a = b = 78.4, c = 292.3 å. The asymmetric unit contains a single subunit of molecular weight approximately 39 kDa with about 65 % solvent content. Lack of isomorphism, even between native crystals, presented a serious obstacle to structure solution and in addition many derivatives gave single but same site binding. We made use of a selenomethionine substituted aldolase sample (Semet–aldolase) (unpublished results). Useful derivatives were eventually obtained from three heavy-atom soaks of Semet–aldolase crystals with potassium tetrachloroplatinate (0.6 mM, three days), samarium acetate (6 mM, three days) and lead acetate (5 mM, 14 days).

Data collection and processing

Single wavelength X-ray data were recorded from crystals mounted in capillaries, as we have been unable to identify suitable cryoconditions. However, the high symmetry allowed us to record fairly complete data with good redundancy using only one crystal in each case. Data were processed, reduced and scaled together using MOSFLM (AGW Leslie (1993) MOSFLM manual, personal communication), DENZO [

] or Rigaku software [

] and the CCP4 programs [

]. The native data set, a Semet–aldolase data set, and data sets from the platinum and samarium derivatives were recorded on PX9.5 at Daresbury laboratory (MAR image plate, various wavelengths). The lead derivative data set was measured in-house (RAXIS-IIC, rotating anode Rigaku RU200, CuK_α); details are given in Table 2 and Table 3. Data beyond 3.0 å are weak in comparison to the lower resolution reflections and the intensity dropped off quickly. Wilson plots [

] invariably displayed thermal parameters in excess of 50 å².

Table 2 Data collection and processing statistics.

Data set	Number of unique measurements	Multiplicity	d_min(å)	R_sym ^∗ (å)	R_anom ^†(%)	R_iso ^‡ (%)	Completeness (%)
Native	18 922	6.7	2.5	7.1		28.1	98.0
Semet^§	8024	2.7	3.3	5.5	5.8		94.5
PtSe^∗∗	7316	7.8	3.5	6.7	6.5	31.0	99.6
SmSe^‡‡	7128	3.0	3.5	5.8	6.8	16.5	99.1
PbSe^#	4345	2.8	4.0	8.1		15.5	87.3

^∗R_sym=Σ|I−<I>|/ΣI, where the sum is over all symmetry equivalent reflections. ^†R_anom=Σ|I(+)−I(−)|/Σ(I(+)−I(−)). ^‡R_iso=Σ|F_PH−F_P|/ΣF_p,i with respect to data set 2 as the reference structure factor, F_p. §Semet = selenomethionine derivative; ^∗∗PtSe, ^††SmSe and ^#PbSe refer to the platinum, samarium and lead derivatives of the selenomethionine containing aldolase.

Open table in a new tab

Table 3 Heavy-atom refinement and phasing statistics.

Derivative	Number of sites	d_min (å)	Phasing power^∗		R_cullis ^†
			acentric	centric	acentric	centric
Pt	3	3.5	1.04	0.75	0.86	0.83
Sm	1	3.5	1.16	0.80	0.83	0.80
Pb	2	4.0	2.26	1.51	0.58	0.55

^∗Phasing power is the root mean square (|F_H|/E), where F_H is the calculated heavy-atom structure factor, and E is the residual lack of closure, ^†R_cullis=Σ|F_PH±F_P−F_H|/Σ|F_PH−F_P|, where F_PH is the structure factor of the derivative and F_P is the structure factor of the selenomethionine data.

Open table in a new tab

MAD data collection was carried out on BL19 at the European Synchrotron Radiation Facility (Grenoble) with a MAR image plate as the detector. A selenium K edge XANES (X-ray absorption near-edge structure) scan on a clump of Semet–aldolase crystals was used to guide the choice of three wavelengths. The first two (λ ₁ = 0.9792 å, λ ₂ = 0.9794 å) were selected to maximize f′′ (the absorptive component) and minimize f′ (the dispersive component), respectively. The third wavelength, λ ₃ = 0.9863 å, lies on the low energy side of the absorption edge. More usually the remote wavelength is taken on the high energy side of the edge but the rationale in our experiment was to minimize potential beam shifts that could result from larger alterations in the monochromator angle and to reduce the effects of radiation damage in crystals that we knew to be moderately radiation sensitive.

As it is difficult to align these crystals an inverse beam geometry was used for data collection on a number of samples. One crystal was aligned with Bijvoet mates being recorded on the same or adjacent images. If a crystal did not provide the same data at each of the three wavelengths or appeared to suffer badly from radiation damage then the data from that crystal were discarded. Eventually a 3.0 å, three wavelength data set was obtained from four crystals (Table 4). Individual wavelength data sets were merged in advance of scaling with λ2 being taken as the reference data set.

Table 4 MAD data collection and processing statistics.

Data set	Number of unique measurements	Multiplicity	d_min (å)	R_sym∗ (%)	R_anom ^† (%)	R_iso ^‡ (%)	Completeness (%)
λ ₁	9991	5.0	3.0	5.5	6.4	6.1	88.6
λ ₂	10 155	5.1	3.0	6.7	6.2		90.2
λ ₃	9882	4.8	3.0	9.7	5.4	7.0	86.3

Open table in a new tab

Phasing

Heavy-atom soaking experiments were carried out on both native and Semet–aldolase and produced a number of possible derivatives. Only the Semet–aldolase derivitized samples proved useful. Heavy-atom positions were located using difference Pattersons and difference Fourier syntheses. Sites were refined and multiple isomorphous replacement with anomalous dispersion (MIRAS) phases calculated using MLPHARE [

[44]

Otwinowski Z.

Maximum likelihood refinement of heavy atom parameters.

Google Scholar

] to give an overall figure of merit (FOM) of 0.53, for 6832 reflections between 10 and 3.5 å resolution. Phasing the anomalous data from the platinum derivative, and later, as an independent, test an Semet–aldolase data set clearly identified the space group enantiomorph as P6₁22. Density modification (DM) [

[45]

Cowtan K.

DM, an automated procedure for phase improvement by density modification.

Google Scholar

] followed. These low resolution phases, although indicating the protein location in the unit cell and a couple of segments of α helix, were not good enough to allow reliable model construction. However, these phases allowed us to identify six selenomethionine positions out of a possible seven in a low resolution Semet–aldolase data set. Inclusion of the selenium positions in phasing calculations made no improvement to the electron-density maps. Knowing the location of six selenium atoms provided confidence for a multiwavelength anomalous diffraction (MAD) phasing exercise [

]. We estimated the MAD signal according to Smith [

]. Assuming seven seleniums per monomer, the best case, the dispersive signal would be 3.7 %, and the anomalous signal 4.2 %. We were never able to locate the remaining selenium, Met190, as it is located on a flexible loop.

MAD phasing was accomplished using an MIR type approach (MLPHARE) treating λ2 as the native data set (Table 5). This provided phases for 9957 reflections in the resolution range 10.0 to 3.0 å with a figure of merit to 0.31. The use of density modification (DM) by both solvent flattening and histogram matching improved the figure of merit to 0.71 and produced an interpretable electron-density map.

Table 5 Phasing statistics for variable wavelength data using an MIR approach.

Data set	Phasing power		R_cullis ^†
	acentric	centric	acentric	centric	anomalous
λ ₁	0.40	0.30	0.98	0.94	0.84
λ ₂					0.81
λ ₃	0.30	0.23	0.99	0.96

^†R_cullis=Σ|F_PH±F_P−F_H|/Σ|F_PH−F_P|, where F_PH is the structure factor of the derivative and F_P is the structure of the selenomethionine data.

Open table in a new tab

Model building and refinement

The MAD phased map was skeletonized using MAPMAN and a Cα trace achieved using 'bones' options in program O [

]. The first model was a conservative polyalanine trace consisting of four polypeptide segments of 245 residues in total. Most of the side chains were clearly visible in the map and the selenium positions provided markers for the assignment of sequence to the model. However, we used the polyalanine model for a round of phase recombination with the MAD phases [

]; side chains were then included. Later a map consisting of MAD/MIRAS phases was generated and although not necessary for the initial model building this map did show improved continuity in some parts. During the refinement stage we always had this experimentally phased map accessible for inspection on the graphics and made frequent reference to it (Figure 6).

Figure thumbnail gr6 — Figure 6 Electron density. Examples of **(a)** the MAD/MIR map (green chicken wire) contoured at the 1σ level and **(b)** the 2F_o–F_c map (blue chicken wire) also at the 1σ level. The maps shown are for the sequence Ile-Val-Gln-Phe, residue numbers 57–60. The average real space residual for these four residues is 25 %, a value close to the average for the whole structure (see Materials and methods section).

View Large Image

Figure Viewer

Download Hi-res image

Download (PPT)

Rounds of model building and graphics fitting (O) and refinement (X-PLOR, [

]) were carried out at 3.0 å using the Semet–aldolase λ2 data. Although two strong peaks in electron-density maps identified metal-binding sites quite early in the analysis we left them out for several rounds until we had the protein mostly complete. For each round of X-PLOR calculations, a range of weighting schemes were tested and the R_free value, using 10 % of the data, was monitored to assist decision making during the refinement [

]. Simulated-annealing protocols [

] and conjugate gradient minimization inconjunction with conservative thermal parameter adjustment reduced the R and R_free values to 36 % and 43 % for all data, respectively. We then transfered to a native data set; this started with rigid-body refinement at 3.0 å, then rounds of graphics inspection, model building and X-PLOR refinement with gradual extension of phases to 2.5 å. Solvent positions were included using very conservative criteria. We dispensed with the R_free once it had dropped below 33 %, and incorporated this data into the calculations. The R factor at this point had dropped below 24 %. The R_free had indicated that our refinement protocols were acceptable and we wanted to include as much of the reliable data into the calculations as possible. This stage of refinement was concluded with final rounds of graphics fitting/checking and X-PLOR. Although aware of further options to refine the metal ion at M1 we were confident that the electron count, as presented earlier, from the difference omit map gave a good indication as to the nature of M1, and that it was not justified to refine occupancy given the limited resolution to which we have data. Nevertheless, we carried out further refinement along four paths. Thermal parameters of metals were reset to 30 å² and treated as: (a) both Zn²⁺s with full occupancy; (b) M1 as a 0.6 occupancy Zn²⁺, M2 as full occupancy Zn²⁺; (c) M1 as K⁺, M2 as Zn²⁺; and (d) M1 as a water, M2 as full occupancy Zn²⁺. This latter case provides a useful test as to how reliable these calculations are given that the height of the difference electron-density peak at M1 exceeds that of any sulphur atoms in the structure and is far greater than any of the solvent positions located elsewhere. A 10 % R_free data set was generated (though perhaps contaminated this R_free subset is still useful) and used as a guide on each refinement; all data were included for further refinement cycles until convergence was achieved. The same restraints and weighting schemes were employed in each case. The lowest R_free was obtained from case (d), 27.81 % then case (b) 28.53 %, then (c) 28.54 % and finally (a) 28.66 %. In ascending order, the R factors are (b) 22.79 %, (c) 22.80 %, (a) 22.86 % and finally (d) 23.04 %. The thermal parameters for M1, and these do not have the benefit of being restrained when refined at limited resolution, are in case (a) 53.5 å², (b) 31.8 å², (c) 32.7 å² and (d) 7.3 å². We suggest that there are a variety of possibilities for the nature of the metal of M1 and a definitive assignment will only be possible with further experimentation. The differences between case (b) and (c) are marginal and we have selected the model derived from protocol (b) for deposition. All details given pertain to that model.

This model has an R factor of 23.2 % for 14 290 reflections with I ≥ 2σI in the range 6.0 to 2.5 å resolution; this is for 82.3 % of the theoretically available data. The model comprises 338 residues (2 566 non-hydrogen atoms), two Zn²⁺ ions, one refined with a fixed occupancy of 0.6, and 70 waters. The refinement has produced a model with an rms in bond lengths, bond angles, dihedral and improper angles of 0.010 å, 2.4°, 23.9° and 2.0°, respectively, in comparison with the Engh and Huber protein parameter dictionary [

]. Average thermal parameters are 36 å² for all protein atoms, 39 å² for the solvent molecules. For M1 and M2 the thermal parameters are 32 å² and 40 å², respectively. A number of solvents may well be partially ordered Zn²⁺ ions given that this was used in the crystallization mixture but we cannot be certain and have preferred to leave these as solvent positions. There are ten residues (Lys166, Lys230, Lys250, Glu147, Thr197, Val228, Arg242, Thr239, Tyr309 and Asp356) for which we do not see good side-chain density and these have been left as alanine residues. All of these residues are on the surface of the protein, are in loop regions or are large flexible amino acids. A Ramachandran plot [

], (Figure 7) has 83 % of residues in most favoured regions and a further 14 % in allowed regions. Residues Lys2 and Phe4 at the N terminus are the only outliers. The real space residual values, as calculated in program O [

], are 14–67 % with an average of 27 % for the protein (Figure 6).

Accession numbers

Structure factors of the native and MAD data, together with their coordinates, have been deposited in the Protein Data Bank, access code ID 1ZEN.

Acknowledgements

We thank A Deacon, S Harrop, J Helliwell, Y Kitagawa, R Nuttall and M Peterson for discussions, encouragement and contributions to this project, and J Raftery for computational support. Particular thanks are to G Schulz, for communicating results in advance of publication, and J Sygusch for discussions on their aldolase studies. Staff at Daresbury Laboratory, especially E Duke are thanked for excellent support and a referee for constructive comments. Funded by the Biotechnology and Biochemistry Science Research Council, the Engineering and Physical Sciences Research Council, the Royal Society, the Leverhulme Trust and the Wellcome Trust. WNH acknowledges a Nuffield Science Research Fellowship 1994-95 and Leverhulme Trust Fellowship 1995–96.

References

- Horecker B.L.
- Tsolas O.
- Lai C.Y.
Aldolases.
in: Boyer P.D. The Enzymes. Academic Press, New York, USA 1972: 213-258
- Wong C.-H.
- Whitesides G.M.
Enzymes in synthetic organic chemistry.
in: Tetrahedron Organic Chemistry Series vol 12. Pergamon Press, New York, USA 1994
- von der Osten C.H.
- Wong C.-H.
- et al.
Use of a recombinant bacterial fructose-1,6-bisphosphate aldolase in aldol reactions: preparative synthesis of 1-deoxynojirimycin, 1-deoxymannojirimycin, 1,4-dideoxy-1,4-imino-D-arabinitol, and fagomine.
J. Am. Chem. Soc Vol 111. 1989; : 3924-3927
- Sygusch J.
- Beaudry D.
- Allaire M.
Molecular architecture of rabbit skeletal muscle aldolase at 2.7 å resolution.
Proc. Natl. Acad. Sci. USA. 1987; 84 (): 7846-7850
- Hester G.
- Piontek K.
- et al.
The crystal structure of fructose-1,6-bisphosphate aldolase from Drosophila melanogaster at 2.5 å.
FEBS Lett. 1991; 292 (): 237-242
- Gamblin S.J.
- Davies G.J.
- Grimes J.M.
- Jackson R.M.
- Littlechild J.A.
- Watson H.C.
Activity and specificity of human aldolases.
J. Mol. Biol. 1991; 219 (): 573-576
- Jia J.
- Schneider G.
- et al.
Crystal structure of transaldolase B from Escherichia coli suggests a circular permutation of the α/β barrel within the class I aldolase family.
Structure. 1996; 4 : 715-724
- Izard T.
- Lawrence M.C.
- Malby R.L.
- Lilly G.G.
- Colman P.
The three-dimensional structure of N-acetylneuraminate lyase from Escherichia coli.
Structure. 1994; 2 (): 361-369
- Lebioda L.
- Hatada M.H.
- Tulinsky A.
- Mavaridis I.M.
Comparison of the folding of 2-keto-3-deoxy-6-phosphogluconate aldolase, triosephosphate isomerase and pyruvate kinase.
J. Mol. Biol. 1982; 162 (): 445-458
- Banner D.W.
- Waley S.G.
- et al.
Structure of chicken muscle triosephosphate isomerase determined by crystallography at 2.5 å resolution using amino acid sequence data.
Nature. 1975; 255 : 609-614
- Brändén C.-I.
The TIM barrel-the most frequently occurring folding motif in proteins.
Curr. Opin. Struct. Biol. 1991; 1 : 978-983
- Morris A.J.
- Davenport R.C.
- Tolan D.R.
A lysine to arginine substitution at position 146 of rabbit aldolase A changes the rate-determining step to Schiff base formation.
Protein Eng. 1996; 9 : 61-67
- Naismith J.H.
- Hunter W.N.
- et al.
Initiating a crystallographic study of a class II aldolase.
J. Mol. Biol. 1992; 225 (): 1137-1141
- Qamar S.
- Marshal K.
- Berry A.
Identification of arginine 331 as an important active site residue in the class II fructose-1,6-bisphosphate aldolase of Escherichia coli.
Protein Sci. 1996; 5 : 154-161
- Marsh J.J.
- Lebherz H.G.
Fructose-bisphosphate aldolases: an evolutionary history.
Trends Biochem. Sci. 1992; 17 (): 110-113
- Dreyer M.K.
- Schulz G.E.
The spatial structure of the class II L-fuculose-1-phosphate aldolase from Escherichia coli.
J. Mol .Biol. 1993; 231 (): 549-553
- Dreyer M.K.
- Schulz G.E.
Catalytic mechanism of the metal-dependent fuculose aldolase from Escherichia coli as derived from the structure.
J. Mol. Biol. 1996; 259 : 458-466
- Laskowski R.A.
- MacArthur M.W.
- Moss D.S.
- Thornton J.M.
PROCHECK: a program to check the stereochemical quality of protein structures.
J. Appl .Cryst. 1993; 26 : 283-291
- Jones T.A.
- Zou J.Y.
- Cowan S.W.
- Kjeldgaard M.
Methods for building protein models in electron density maps and the location of errors in these models.
Acta Cryst. A. 1991; 47 : 110-119
- Kabsch W.J.
- Sander C.
Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features.
Biopolymers. 1983; 22 (): 2577-2637
- Lee B.
- Richards F.M.
The interpretation of protein structures: estimation of static accessibility.
J. Mol. Biol. 1971; 55 : 379-400
- Brünger A.T.
X-PLOR Version 2.2 manual. Howard Hughes Medical Institute, Yale University, USA 1990
- Berry A.
- Marshal K.E.
Identification of zinc-binding ligands in the class II fructose-1,6-bisphosphate aldolase of Escherichia coli.
FEBS Lett. 1993; 318 (): 11-16
- Allen F.H.
- Kennard O.
3D search and research using the Cambridge Crystallographic Database.
Chemical Design Automation News. 1993; 8 : 130-137
- Kitagawa Y.
- Berry A.
- et al.
Additional crystal forms of the E. coli Class II aldolase.
Acta Cryst. D. 1995; 51 : 833-834
- Burley S.K.
- David P.R.
- Sweet R.M.
- Taylor A.
- Lipscomb W.N.
Structure determination and refinement of bovine lens leucine aminopeptidase and its complex with bestatin.
J. Mol. Biol. 1992; 224 (): 113-140
- Vanhooke J.L.
- Benning M.M.
- Raushel F.M.
- Holden H.
Three-dimensional structure of the zinc-containing phosphotriesterase with the bound substrate analog diethyl 4-methylbenzylphosphonate.
Biochemistry. 1996; 35 : 6020-6025
- Wilson K.
- Rudolph F.B.
- Quiocho F.A.
Atomic structure of adenosine deaminase complexed with a transition-state analog: understanding catalysis and immunodeficiency mutations.
Science. 1991; 252 (): 1278-1284
- Vallee B.L.
- Auld D.S.
Active-site ligands and activated H₂O of zinc enzymes.
Proc. Natl. Acad. Sci. USA. 1990; 87 (): 220-224
- Christianson D.W.
Structural biology of zinc.
Adv. Protein Chem. 1991; 42 (): 281-355
- Vallee B.L.
- Auld D.S.
New perspective on zinc biochemistry: co-catalytic sites in multi-zinc enzymes.
Biochemistry. 1993; 32 (): 6493-6500
- Chevrier B.
- D'orchymont H.
- Schalk C.
- Tarnus C.
- Moras D.
The structure of the Aeromonas proteolytica aminopeptidase complexed with a hydroxamate inhibitor.
Eur. J. Biochem. 1996; 237 : 393-398
- Tainer J.A.
- Getzoff E.D.
- Beem K.M.
- Richardson J.S.
- Richardson D.C.
Determination and analysis of the 2 å structure of copper, zinc superoxide dismutase.
J. Mol. Biol. 1982; 160 (): 181-217
- Bock C.W.
- Katz A.K.
- Glusker J.P.
Hydration of zinc ions: a comparison with magnesium and beryllium ions.
J. Am. Chem. Soc. 1995; 117 : 3754-3765
- Borgstahl G.E.O.
- Parge H.E.
- Hickey M.J.
- Beyer W.F.
- Hallewell R.A.
- Tainer J.A.
The structure of human mitochondrial manganese superoxide dismutase reveals a novel tetrameric interface of two four-helix bundles.
Cell. 1992; 71 (): 107-118
- Collins K.D.
An activated intermediate analogue. The use of phosphoglycolohydroxamate as a stable analogue of a transiently occurring dihydroxyacetone phosphate-derived enolate in enzymatic catalysis.
J. Biol. Chem. 1974; 249 : 136-142
- Mildvan A.S.
- Kobes R.D.
- Rutter W.J.
Magnetic resonance studies of the divalent cation in the mechanism of yeast aldolase.
Biochemistry. 1971; 10 : 1191-1204
- Smith G.M.
- Mildvan A.S.
- Harper E.T.
Nuclear relaxation studies of the interaction of substrates with a metalloaldolase from yeast.
Biochemistry. 1980; 19 (): 1248-1255
- Zhang R.
- Villeret V.
- Lipscomb W.N.
- Fromm H.J.
Kinetics and mechanism of activation and inhibition of porcine liver fructose-1,6-bisphosphatase by monovalent cations.
Biochemistry. 1996; 35 (): 3038-3043
- Otwinowski Z.
Oscillation data reduction program.
in: Sawyer L. Isaacs N. Bailey S. Data Collection and Processing. SERC Daresbury Laboratory, Warrington, UK 1993: 56-62
- Higashi T.
Process: a program for indexing and processing R-AXIS II image plate data. Rigaku Corporation, Japan 1990
- Collaborative Computational Project No. 4
The CCP4 suite: programs for protein crystallography.
Acta Cryst. D. 1994; 50 : 760-763
- Wilson A.J.C.
The probability distribution of X-ray intensities.
Acta Cryst. 1949; 2 : 318-321
- Otwinowski Z.
Maximum likelihood refinement of heavy atom parameters.
in: Wolf W. Evans P.R Leslie A.G.W Isomorphous Replacement and Anomalous Scattering. SERC Daresbury Laboratory, Warrington, UK 1991: 80-85
- Cowtan K.
DM, an automated procedure for phase improvement by density modification.
Joint CCP4 and ESF-EACBM Newsletter on Protein Crystallography. 1994; 31 : 24-28
- Hendrickson W.A.
Determination of macromolecular structures from anomalous diffraction of synchrotron radiation.
Science. 1991; 254 (): 51-58
- Smith J.L.
Determination of three-dimensional structure by multiwavelength anomalous diffraction.
Curr. Opin. Struct. Biol. 1991; 1 : 1002-1011
- Jones T.A.
Making the first trace with O.
in: Bailey S. Hubbard R. Waller D. From First Map to Final Model. SERC Daresbury Laboratory, Warrington, UK 1994: 1-13
- Read R.J.
Improved Fourier coefficients for maps using phases from partial structures with errors.
Acta Cryst. A. 1986; 42 : 140-149
- Brünger A.T.
Assessment of phase accuracy by cross validation: the free R value.Methods and applications.
Acta Cryst. D. 1993; 49 : 24-36
- Brünger A.T.
- Krukowski A.
- Erickson J.W.
Slow-cooling protocols for crystallographic refinement by simulated annealing.
Acta Cryst. A. 1990; 46 (): 585-593
- Engh R.A.
- Huber R.
Accurate bond and angle parameters for X-ray protein-structure refinement.
Acta Cryst. A. 1991; 47 : 392-400
- Ramachandran G.N.
- Sasiskharan V.
Conformation of polypeptides and proteins.
Adv. Protein Chem. 1968; 23 : 283-437
- Kraulis P.J.
MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures.
J. Appl. Cryst. 1991; 24 : 946-950
- Alefounder P.R.
- Baldwin S.A.
- Perham R.N.
- Short N.J.
Cloning, sequence analysis and over-expression of the gene for the class II fructose-1,6-bisphosphate aldolase of Escherichia coli.
Biochem. J. 1989; 257 (): 529-534

Biography

SJ Cooper, GA Leonard, JH Naismith and WN Hunter, Department of Chemistry, University of Manchester, Oxford Road, Manchester, M13 9PL, UK.

Present address for GA Leonard: ESRF, F38043, Grenoble, CEDEX, France.

SM McSweeney and AW Thompson, EMBL, F38043, Grenoble, CEDEX, France.

Present address for JH Naismith: Department of Chemistry, University of St. Andrews, St. Andrews, UK.

S Qamar, A Plater and A Berry, Department of Biochemistry and Molecular Biology, University of Leeds, Leeds, LS2 9JT, UK.

Present address for WN Hunter: Department of Biochemistry, University of Dundee, Dundee, DD1 4HN, UK.

E-mail address for WN Hunter (corresponding author): [email protected]

SJ Cooper and GA Leonard contributed equally to this work.

Article Info

Publication History

Accepted: September 23, 1996

Received in revised form: August 22, 1996

Received: June 28, 1996

Identification

DOI: https://doi.org/10.1016/S0969-2126(96)00138-4

Copyright

User License

Elsevier user license |

How you can reuse Information Icon

Permitted

For non-commercial purposes:

Read, print & download
Text & data mine
Translate the article

Not Permitted

Reuse portions or extracts from the article in other works
Redistribute or republish the final article
Sell or re-use for commercial purposes

Elsevier's open access license policy

ScienceDirect

Access this article on ScienceDirect

View Large Image
Download Hi-res image
Download .PPT

Cell Press commenting guidelines

To submit a comment for a journal article, please use the space above and note the following:

We will review submitted comments within 2 business days.
This forum is intended for constructive dialog. Comments that are commercial or promotional in nature, pertain to specific medical cases, are not relevant to the article for which they have been submitted, or are otherwise inappropriate will not be posted.
We recommend that commenters identify themselves with full names and affiliations.
Comments must be in compliance with our Terms & Conditions.
Comments will not be peer-reviewed.