Exploring substrate binding and discrimination in fructose1,6-bisphosphate and tagatose 1,6-bisphosphate aldolases

Shaza M. Zgiby, Graeme J. Thomson, Seema Qamar* and Alan Berry

School of Biochemistry and Molecular Biology, University of Leeds, Leeds, UK

Fructose 1,6-bisphosphate aldolase catalyses the reversible condensation of glycerone-P and glyceraldehyde

3-phosphate into fructose 1,6-bisphosphate. A recent structure of the Escherichia coli Class II fructose

1,6-bisphosphate aldolase [Hall, D.R., Leonard, G.A., Reed, C.D., Watt, C.I., Berry, A. & Hunter, W.N.

(1999) J. Mol. Biol. 287, 383±394] in the presence of the transition state analogue phosphoglycolohydroxamate

delineated the roles of individual amino acids in binding glycerone-P and in the initial proton abstraction steps of

the mechanism. The X-ray structure has now been used, together with sequence alignments, site-directed

mutagenesis and steady-state enzyme kinetics to extend these studies to map important residues in the binding of

glyceraldehyde 3-phosphate. From these studies three residues (Asn35, Ser61 and Lys325) have been

identified as important in catalysis. We show that mutation of Ser61 to alanine increases the Km value for

fructose 1,6-bisphosphate 16-fold and product inhibition studies indicate that this effect is manifested most

strongly in the glyceraldehyde 3-phosphate binding pocket of the active site, demonstrating that Ser61 is involved

in binding glyceraldehyde 3-phosphate. In contrast a S61T mutant had no effect on catalysis emphasizing the

importance of an hydroxyl group for this role. Mutation of Asn35 (N35A) resulted in an enzyme with only 1.5%

of the activity of the wild-type enzyme and different partial reactions indicate that this residue effects the binding

of both triose substrates. Finally, mutation of Lys325 has a greater effect on catalysis than on binding, however,

given the magnitude of the effects it is likely that it plays an indirect role in maintaining other critical residues in

a catalytically competent conformation. Interestingly, despite its proximity to the active site and high sequence

conservation, replacement of a fourth residue, Gln59 (Q59A) had no significant effect on the function of the

enzyme. In a separate study to characterize the molecular basis of aldolase specificity, the agaY-encoded tagatose

1,6-bisphosphate aldolase of E. coli was cloned, expressed and kinetically characterized. Our studies showed that

the two aldolases are highly discriminating between the diastereoisomers fructose bisphosphate and tagatose

bisphosphate, each enzyme preferring its cognate substrate by a factor of 300±1500-fold. This produces an

overall discrimination factor of almost 5 � 105 between the two enzymes. Using the X-ray structure of the

fructose 1,6-bisphosphate aldolase and multiple sequence alignments, several residues were identified, which are

highly conserved and are in the vicinity of the active site. These residues might potentially be important in

substrate recognition. As a consequence, nine mutations were made in attempts to switch the specificity of the

fructose 1,6-bisphosphate aldolase to that of the tagatose 1,6-bisphosphate aldolase and the effect on substrate

discrimination was evaluated. Surprisingly, despite making multiple changes in the active site, many of which

abolished fructose 1,6-bisphosphate aldolase activity, no switch in specificity was observed. This highlights the

complexity of enzyme catalysis in this family of enzymes, and points to the need for further structural studies

before we fully understand the subtleties of the shaping of the active site for complementarity to the cognate

substrate.

Keywords: aldolase; fructose 1,6-bisphosphate; protein engineering; site-directed mutagenesis; tagatose 1,

6-bisphosphate.

The difficulty of stereochemical control during carbon±carbonbond formation using conventional chemical approaches [1] has

prompted the development of catalytic antibodies for aldolcondensations [1,2] and the use of naturally occurring aldolasesin synthetic chemistry [3,4]. The best studied enzymes fromthe aldolase family, capable of such enzyme chemistry, arethe fructose 1,6-bisphosphate aldolases. These catalyse thereversible cleavage of fructose 1,6-bisphosphate into twotriose sugars, dihydroxyacetone phosphate (glycerone-P) andglyceraldehyde 3-phosphate. Aldolases can be broadlydivided into two groups, designated Class I and Class II[5]. Whereas the Class I enzymes utilize an active-site lysine tostabilize a reaction intermediate via Schiff-base formation, theClass II enzymes have an absolute requirement for a divalentmetal ion, usually zinc [6,7], which polarizes the substratecarbonyl [8].

Eur. J. Biochem. 267, 1858±1868 (2000) q FEBS 2000

Correspondence to A. Berry, School of Biochemistry and Molecular

Biology, University of Leeds, Leeds LS2 9JT, UK. Fax: 1 44 113 2333167,

Tel: 1 44 113 2333158, E-mail: a.berry@leeds.ac.uk

Abbreviations: glycerone-P, dihydroxyacetone phosphate.

Enzymes: fructose 1,6-bisphosphate aldolase (EC 4.1.2.13);

glyceraldehyde 3-phosphate dehydrogenase (EC 1.2.1.12); glycerol

3-phosphate dehydrogenase (EC 1.1.1.8); tagatose 1,6-bisphosphate

aldolase (EC 4.1.2.40); triose phosphate isomerase (EC 5.3.1.1).

*Present address: Memorial Sloan Kettering Cancer Center, New York,

USA.

(Received 9 December 1999, accepted 28 January 2000)

q FEBS 2000 The active site of Class II fructose 1,6-bisphosphate aldolases (Eur. J. Biochem. 267) 1859

The structure of the Class II fructose 1,6-bisphosphatealdolase from Escherichia coli [9,10] showed the enzyme tobe a member of the (a/b)8 family of enzymes, and locatedthe active site and the bound zinc ion at the C-terminal end ofthe barrel. Subsequent solution of the crystal structure of theenzyme in the presence of the transition state analoguephosphoglycolohydroxamate [11], together with site-directedmutagenesis and enzyme kinetics [12±14], provided furtherinsights into the reaction mechanism of the enzyme anddelineated the roles of several critical residues in catalysis(Fig. 1). In the direction of aldol condensation, glycerone-Pfirst binds either with or following monovalent cation binding.A number of structural changes then occur: there are rearrange-ments and ordering of three loops and rotation of the imidazolerings of His110 and His264 is postulated to move the zinc ionfrom a fully coordinated, buried site to a more exposed site

suitable for catalysis [9,11]. The zinc can then act as a Lewisacid to polarize the carbonyl group of the glycerone-P anddeprotonation at C1 then occurs. The developing charge on theresulting ene-diolate is stabilized by the proximity of Asn286[11,13]. Aldehyde binding to the enzyme in the next step bringsthe two planes of the ene-diolate nucleophile and the carbonylacceptor into the correct alignment for carbon±carbon bondformation and also positions the aldehyde so that its carbonylgroup can be polarized by Asp109 [13]. Carbon±carbon bondformation then occurs either in concert with or immediatelybefore Asp109 transfers a proton to the bound substrate toproduce the C4-OH. Finally, product release regenerates thefree enzyme.

Despite this growing understanding of the catalytic mechan-ism of the enzyme, a number of questions remain. One of themost important of these relates to the binding of glyceraldehyde

Fig. 1. General reaction mechanism for the Class II fructose 1,6-bisphosphate aldolase. The reaction mechanism for the condensation of glycerone-P

(DHAP) and glyceraldehyde 3-phosphate to fructose 1,6-bisphosphate is shown. After glycerone-P binding to the enzyme (I ! II) rearrangement of the zinc

ligands occurs to bring the zinc into position to polarize the glycerone-P carbonyl group. Abstraction of the 1-proS proton from glycerone-P by residue X

then yields the aldolase-bound enediolate carbanion intermediate (III) which can be trapped chemically using hexacyanoferrate (III). This carbanion is

stabilized by Asn286 [11,13] whose position is not shown in this figure for clarity. Binding of glyceraldehyde 3-phosphate (III 1 IV ! V), formation of the

new carbon±carbon bond and protonation at C4 of the new hexose by Asp109 (V ! VI) [13] yields the enzyme-fructose 1,6-bisphosphate complex (VI),

which then releases fructose 1,6-bisphosphate (VII). All steps are fully reversible. The aldolase-bound enediolate carbanion can be formed from glycerone-P

as shown in the figure, or from fructose 1,6-bisphosphate by reversal of the mechanism shown in steps (VII ! III).

1860 S. M. Zgiby et al. (Eur. J. Biochem. 267) q FEBS 2000

3-phosphate or the C4-C6 of the hexose substrate in the reversereaction. The transition state mimicked by phosphoglycolo-hydroxamate involves only carbon atoms 1±3 of the substratehexose (Fig. 1) and no detailed information on the interactionsbetween C4-C6 of the substrate and the enzyme are thuspossible. Furthermore, no structure has been solved for the

Class II fructose 1,6-bisphosphate aldolases with a C6 substrateor analogue in the active site. Attempts at modellingglyceraldehyde 3-phosphate into the structure of the enzyme±phosphoglycolohydroxamate complex [11] have provided somesuggestions of how glyceraldehyde 3-phosphate might bebound. However, given the significant rearrangement of loopsand active site residues on substrate binding (see above), theexact molecular details of glyceraldehyde 3-phosphate bindingare left open.

It is also interesting to note that some aspects of substratestereoselectivity are controlled by binding in this region of thesubstrate. Indeed, attack of the glycerone-P enediolate inter-mediate on the si-face of the glyceraldehyde 3-phosphatecarbonyl group in the fructose 1,6-bisphosphate aldolasesgenerates 3S,4R fructose bisphosphate, whereas attack by thesame intermediate on the re-face generates 3S, 4S tagatosebisphosphate in the case of the tagatose bisphosphate aldolases.The Class II fructose 1,6-bisphosphate aldolase is highlyspecific for fructose 1,6-bisphosphate and uses thissubstrate almost 5 � 105 better than its diastereoisomer,tagatose 1,6-bisphosphate (Fig. 2 and below). Escherichiacoli contains two Class II aldolases (GatY and AgaY)responsible for aldol cleavage of tagatose 1,6-bisphosphate toform glycerone-P and glyceraldehyde 3-phosphate [15±17].One, encoded by the gatY gene, plays a role in the metabolismof galactitol [16,17], while the other, encoded by agaY, has itsmajor role in the metabolism of N-acetylgalactosamine [15].

Fig. 2. The structures of fructose 1,6-bisphosphate and tagatose

1,6-bisphosphate.

Fig. 3. Sequence alignment of Class II

tagatose 1,6-bisphosphate- and fructose

1,6-bisphosphate aldolases. Part of the multiple

sequence alignment of the 26 sequences known

for Class II fructose 1,6-bisphosphate aldolases

and the three sequences known for Class II

tagatose 1,6-bisphosphate aldolases. Only the

fructose 1,6-bisphosphate aldolase from E. coli

(FBPA E. coli ) [33], the GatY tagatose

1,6-bisphosphate aldolase from E. coli

(GatY_E. coli ) [16,17], the E. coli AgaY

tagatose 1,6-bisphosphate aldolase

(AgaY_E. coli ) [15] and the putative tagatose

1,6-bisphosphate aldolase from V. furnissii

(TagA_V. furnissii ) [34] are shown. The

alignment was produced using the GCG program

pileup [35]. Residues lying within 5 AÊ of

phosphoglycolohydroxamate in its complex with

the Class II fructose 1,6-bisphosphate aldolase of

E. coli [11] are underlined and emboldened.

Glu182 and the section of polypeptide chain

missing in this structure are double underlined.

Phe67 is marked with an asterix.

q FEBS 2000 The active site of Class II fructose 1,6-bisphosphate aldolases (Eur. J. Biochem. 267) 1861

These two tagatose 1,6-bisphosphate aldolases share 54%sequence identity with each other (72% conservation) and have< 28% identity (51% conservation) with the E. coli Class IIfructose 1,6-bisphosphate aldolase. This level of conservationmakes it likely that GatY and AgaY will adopt the (a/b)8 barrelfold found in the E. coli Class II fructose 1,6-bisphosphatealdolase [9,10]. The sequence alignment of the family ofClass II fructose 1,6-bisphosphate aldolases with the twotagatose 1,6-bisphosphate aldolases also shows that almostall of the residues identified as important in the fructose1,6-bisphosphate aldolase reaction mechanism are conserved inthe tagatose 1,6-bisphosphate aldolases (Fig. 3). This furthersuggests a conservation of the mechanism and highlights afascinating question regarding the determinants of stereo-specificity of the reaction.

Here we used site-directed mutagenesis and enzyme kinetics,together with a knowledge of the structure of the enzyme±phosphoglycolohydroxamate complex [11] to investigate therole of four previously uncharacterized residues (Asn35, Gln59,Ser61 and Lys325) in the binding of the C4-C6 end of thehexose substrate in the Class II fructose 1,6-bisphosphatealdolase. We also investigated the origin of substrate specificitybetween the tagatose 1,6-bisphosphate aldolases and thefructose 1,6-bisphosphate aldolases by cloning and expressionof one of the tagatose 1,6-bisphosphate aldolases, combinedwith a systematic study using site-directed mutagenesis.

M A T E R I A L S A N D M E T H O D S

Materials

Calf intestinal alkaline phosphatase, the restriction endo-nucleases EcoRI and HindIII, and glycerol 3-phosphatedehydrogenase/triose phosphate isomerase were from Boeh-ringer-Mannheim. d-Fructose 1,6-bisphosphate trisodium salt,glycerone-P lithium salt, NADH and NAD1 were from SigmaChemicals Ltd (Poole, UK). Tagatose bisphosphate was thegenerous gift of W. D. Fessner (University of Darmstadt,Germany).

T4 DNA ligase was supplied by Pharmacia and the Pfu DNApolymerase and reaction buffer were supplied by Stratagene.The WizardTM Minipreps DNA purification system andlDNA/HindIII markers were purchased from Promega. TheNucleon Easiclean DNA purification kit was obtained fromScotlab Bioscience. Diethylaminoethyl-cellulose (DE-52) wasfrom Whatman Biosystems Ltd. The SuperdexTM 200 prepgrade HiLoadTM 16/60column was supplied by Pharmacia(Milton Keynes, UK) and the Poros HQ20 and Poros 20HP2columns were from Perseptive Biosystems. Dialysis tubing wasfrom Medicell International, Ltd. All other chemicals were ofanalytical grade.

Bacterial strains and plasmids

E. coli strain KM3 [D(his gnd), Dlac, araD, fda, ptsF, ptsM,rpsL, pro, NAR/F 0, proA1B1 lacIq lacZDM15] was asdescribed previously [12]. E. coli Crookes strain (NCIMB8545) was obtained from the National Collection of Industrialand Marine Bacteria (Aberdeen, UK). The expression vectorpKK223-3 was from Pharmacia.

Purification of Class II fructose 1,6-bisphosphate aldolase

Both the wild-type and mutant enzymes were expressed andpurified to homogeneity as described previously [12]. Protein

concentrations were assayed by the bicinchoninic acid method[18] using BSA as standard.

Cloning of the AgaY tagatose bisphosphate aldolase

The oligonucleotides 5 0-GAGGATGAATTCATGAGCATTAT-CTCCACT-3 0 (agaY forward primer) and 5 0-AAAAACA-AGCTTTTATGCTGAAATTCGATT-3 0 (agaY reverse primer)were designed to amplify the full-length agaY gene from theE. coli chromosome [19] and to introduce EcoRI and HindIIIsites at the 5 0-end and 3 0-end, respectively (sites shownunderlined). A total genomic DNA preparation from E. coliCrookes strain was prepared according to the method of Kirbyet al. [20] for use as the template in the PCR reaction. PCRproducts of approximately the expected size, namely 860 bp,were generated using the PCR conditions outlined for site-directed mutagenesis (below). The PCR products were purifiedfrom an agarose gel and then digested with EcoRI and HindIIIrestriction enzymes. The gene was ligated into EcoRI/HindIII-treated pKK223-3. The resulting plasmid was termed pKagaYand was transformed into E. coli strain KM3. A high degree ofover-production was achieved from this plasmid without theneed for isopropyl thio-b-d-galactose induction.

Expression and purification of AgaY tagatose bisphosphatealdolase

Cells expressing the agaY-encoded tagatose 1,6-bisphosphatealdolase (E. coli KM3/pKagaY) were grown overnight at 37 8Cin 5 L of tryptone-yeast (2 � TY) media supplemented with50 mg´mL21 ampicillin and 0.3 mm ZnCl2. The cells wereharvested by centrifugation in a Heraeus Contifuge 17RScontinual action centrifuge at 19 000 g and then suspended in50 mm potassium phosphate buffer pH 7.0, before beingdisrupted in a French Press at 4 8C and cell pressure of140 MPa. The extract was clarified by centrifugation at14 500 g for 1 h and was then fractionated with solid(NH4)2SO4. The 0±40% (NH4)2SO4 pellet, containing thetagatose 1,6-bisphosphate aldolase activity, was dialysedextensively against 50 mm Tris/HCl buffer pH 7.5. Thedialysed enzyme was then applied to a DEAE-cellulose column(13 cm � 2.6 cm diameter), equilibrated in the same buffer.After loading the enzyme sample, the column was washed withthe same buffer until the absorbance at 280 nm (A280) of theeluate decreased to zero. The column was then washed with80 mm NaCl in the same buffer until the A280 was again zero.The enzyme was then eluted with step gradients of 120 mm and200 mm NaCl in 50 mm Tris/HCl buffer pH 8.0. The peakfractions containing the enzyme activity were pooled.(NH4)2SO4 was added to 80% saturation and the enzyme-containing precipitate was resuspended in a small volume of100 mm Tris/HCl buffer pH 8.0 and dialysed against the samebuffer. The concentrated enzyme was injected in 2-mL aliquotsonto a Superdex 200 gel filtration column equilibrated with100 mm Tris/HCl buffer pH 8.0. Fractions containing tagatose1,6-bisphosphate aldolase activity were pooled and ammoniumsulfate was added to a final concentration of 1.5 m. The enzymewas further purified using a Poros 20HP2 column equilibratedin 50 mm Tris/HCl buffer pH 7.5 containing 1.5 m (NH4)2SO4

mounted on a BioCAD Sprint chromatography system. Theelution of the aldolase was carried out using a linear 1.5 mto 0 m ammonium sulfate gradient over 25 column volumesat 5 mL´min21. Further purification was then achieved usinga Poros HQ20 column, equilibrated in 50 mm Tris/HCl,pH 7.5 buffer. Elution of the tagatose 1,6-bisphosphatealdolase was carried out using a linear gradient (0±500 mm

1862 S. M. Zgiby et al. (Eur. J. Biochem. 267) q FEBS 2000

NaCl) over 25 column volumes at 5 mL´min21. The activepeak fractions were pooled and homogeneity was checkedby SDS/PAGE.

Enzyme assays

Standard fructose 1,6-bisphosphate cleavage assay. A coupledenzyme assay was used to determine fructose 1,6-bisphosphatealdolase activity as described previously [12]. The assays wereperformed at 30 8C and the decrease in A340 was recorded. Amodification of this assay was used for assays in the presenceof the products, glycerone-P or glyceraldehyde 3-phosphate, asdescribed previously [14]. In order to determine steady-statekinetic parameters, data were fitted to the appropriate rateequations using either the computer program kaleidagraph(Abelbeck Software) or the suite of programs described byCleland [21].

Tagatose bisphosphate aldolase activity. Tagatose 1,6-bisphos-phate aldolase activity was measured using a modification ofthe coupled enzyme assay described above, substitutingtagatose 1,6-bisphosphate for fructose 1,6-bisphosphate in anotherwise identical assay.

Hexacyanoferrate (III) oxidation of the carbanion intermediate.The oxidation of the carbanionic ene-diolate reaction inter-mediate by hexacyanoferrate (III) was followed spectro-photometrically at 420 nm [14,22]. A suitable aliquot ofaldolase was incubated at 30 8C with 2 mm potassiumhexacyanoferrate (III) in 50 mm Tris/HCl buffer containing0.1 m potassium acetate, pH 8.0, and various concentrations ofglycerone-P and fructose 1,6-bisphosphate. The initial rate ofdecrease in absorbance at 420 nm was recorded.

CD. Far-UV (190±260 nm) CD spectra of the wild-type andmutant aldolases were determined, at a protein concentration of0.5 mg´mL21, in 20 mm potassium phosphate buffer pH 7.6using a Jasco J720 spectropolarimeter with a pathlength of1 mm. Two spectra were accumulated per sample.

Site-directed mutagenesis. Site-directed mutagenesis of thealdolase gene was carried out using megaprimer PCR [23]. Twoflanking primers were designed to lie outside the fda genecoding region of the pKfda2 template. The forward primer was5 0-AGGACAGAATTCATGTCTAAGATTTTTGAT (encodingan EcoRI site, underlined) and the reverse primer, 5 0-GAAAG-GAAGCTTTTACAGAACGTCGATCGC (encoding a HindIIIsite, underlined). Mutants were generated in two consecutiverounds of PCR. The first stage involved using one of thefollowing mutagenic primers (N35A, 5 0-CCAGCAGTAGC-TTGCGTCGGTACT-3 0; Q59A, 5 0-GTTATCGTTGCTTTCTC-CAAC-3 0; S61A, 5 0-GTTCAGTTCGCTAACGGTGGT-3 0;S61T, 5 0-GTTCAGTTCACCAACGGTGGTGCT-3 0; K325A,5 0-CAGCCGAACGCGAATTACTACGATCCGGGC-3 0; G64T,5 0-TCCAACGGTACCGCTTCCTTTATC-3 0; F67H, 5 0-GGTG-CTTCCCACATCGCTGGTAAA-3 0; Q59A/S61T, 5 0-GTTATC-GTTGCTTTCACTAATGGT-3 0; S61T/G64T/F67H, 5 0-GTTCA-GTTCACCAACGGTAACGCTTCCCACATCGCTGGTAAA-30)in conjunction with the reverse primer to generate themegaprimer. The megaprimer was purified and used in thesecond round of PCR in conjunction with the forward primer toproduce a full-length mutant aldolase gene. The mutant geneswere digested with EcoRI and HindIII before being ligated intothe EcoRI/HindIII digested, dephosphorylated expression

vector pKK223-3. The resulting plasmids were used totransform the fda-deficient strain of E. coli KM3.

PCR reactions were carried out in a 50-mL reaction mixtureconsisting of 20 mm Tris/HCl pH 8.8, 2 mm MgSO4, 10 mmKCl, 10 mm (NH4)2SO4, 0.1% Triton X-100, 100 mg´mL21

nuclease free BSA, 0.2 mm each dNTP, 100 pmol of eachN-terminal and C-terminal primer, 0.5 mg of template DNAand 2.5 units of Pfu DNA polymerase. The reaction mixturewas overlaid with 50 mL of mineral oil. Amplification involvedan initial denaturation step at 95 8C for 5 min followed bycycling at 95 8C for 1 min, 55 8C for 1 min and 72 8C for2.5 min for 40 cycles then a final extension for 2 min at 72 8C.PCR products were purified from agarose gels using a NucleonEasiclean kit. DNA sequencing confirmed that no spuriousmutations were generated by the megaprimer PCR technique.

R E S U LT S A N D D I S C U S S I O N

Crystallographic studies of the Class II fructose 1,6-bisphos-phate aldolase revealed the molecular details of the interactionof the ene-diolate transition state analogue phospho-glycolohydroxamate with the enzyme, and hence, by analogy,between glycerone-P and the enzyme. Details of the bindingof glyceraldehyde 3-phosphate to the enzyme have, to date,been lacking. Modelling of glyceraldehyde 3-phosphate andglycerone-P into the enzyme structure [11] suggested possibleroles for Asn35, Ser61, Asp109, Asp288, Lys325 and Arg331 inbinding the glyceraldehyde 3-phosphate moiety. A combinationof site-directed mutagenesis, enzyme kinetics and biophysicalcharacterization has already demonstrated that Arg331 iscritically involved in binding of the phosphate group ofglyceraldehyde 3-phosphate [14], that Asp109 polarizes thealdehydic carbonyl group of glyceraldehyde 3-phosphate [13]and that Asp288 appears to have a minor role in glyceraldehyde3-phosphate binding [13]. The remaining residues (Asn35,Ser61 and Lys325) have now been investigated by mutagenesis.The mutants N35A, S61A, S61T and K325A were generated bymegaprimer PCR [23]. The genes were sequenced to ensurethat no spurious mutations had been incorporated, and weresubcloned into the expression vector pKK223-3. The mutantproteins were then overexpressed in soluble form in the fructose1,6-bisphosphate aldolase-deficient strain of E. coli (strainKM3) and purified to homogeneity as described previously[12]. The CD spectra for all the mutants were similar to that ofwild-type (data not shown), indicating that there were no majorstructural perturbations caused by the mutations.

The kinetic properties of each mutant were determined usingboth a fructose 1,6-bisphosphate-cleavage assay [12] (Table 1)and a hexacyanoferrate (III) assay for enediolate oxidation[14,22] (Table 2). This latter assay allows us to delineate moreprecisely the role of individual residues in binding and/orcatalysis, because the enzyme±enediolate intermediate can begenerated either from fructose 1,6-bisphosphate (by C4-OHproton abstraction and cleavage of the C3-C4 bond) or fromglycerone-P (by the abstraction of the 1-proS proton) [24](Fig. 1). We also used the two triose-phosphate products of thereaction as inhibitors of the fructose 1,6-bisphosphate-cleavagereaction. Glycerone-P binds to the aldolase at the C-1phosphate-binding site and acts as a competitive inhibitor ofthe cleavage reaction [14,25]. The Ki value for glycerone-P isthus a direct measure of its dissociation constant. In contrast,glyceraldehyde 3-phosphate (which binds at the C-6 phosphatesite of the enzyme) acts as a noncompetitive inhibitor of thecleavage reaction [14,25] and its Ki is, therefore, not a directmeasure of the dissociation constant. However, any mutation

q FEBS 2000 The active site of Class II fructose 1,6-bisphosphate aldolases (Eur. J. Biochem. 267) 1863

that affects glyceraldehyde 3-phosphate binding would beexpected to increase its Ki. The properties of each mutantenzyme are now discussed in turn.

Asparagine 35

The replacement of Asn35 with an alanine residue hadsignificant effects on the steady-state kinetics in the standardfructose 1,6-bisphosphate cleavage assay (Table 1). Theenzyme showed a 65-fold decrease in kcat and had a Km forfructose 1,6-bisphosphate fivefold higher than that of the wild-type enzyme, demonstrating that Asn35 plays importantroles in both binding and catalysis. Steady-state kineticmeasurements using the hexacyanoferrate (III) assay withfructose 1,6-bisphosphate as substrate revealed that the k 0cat forthis reaction was reduced by a factor of < 18, whereas withglycerone-P as substrate there was only a threefold changein the rate of carbanion oxidation (Table 2). Interestingly,however, there was a large increase in the Km

0 withglycerone-P as substrate (15-fold), whereas this parameterremained essentially the same as wild-type with fructose1,6-bisphosphate as substrate. Although Km is not a directmeasure of substrate binding for this kinetic mechanism, theresult suggests that replacement of Asn35 by alanine bringsabout changes in the active site that weaken the enzymeinteractions with glycerone-P.

The N35A mutant shows only 1.5% of the activity of thewild-type enzyme. Despite this, the inhibition constants for thetriose phosphates for this mutant could be determined usinglarge amounts of enzyme (< 1 mg) per assay (Table 1). Witheither glycerone-P or glyceraldehyde 3-phosphate as substrate,an 8±15-fold increase (relative to the wild-type enzyme) in theKi values was observed (Table 1). These data thus suggest thatAsn35 is not only involved in glycerone-P binding, but that itplays important roles in both triose binding sites.

These results all accord with the postulated roles for Asn35because removal of the important interaction between Asn35and the O2 atom of glyceraldehyde 3-phosphate in the N35Amutant would account for the increased Ki for glyceraldehyde3-phosphate observed in the N35A protein. In addition, thestructure of the enzyme in complex with phospho-glycolohydroxamate [11] shows that Asn35 also hydrogenbonds to another conserved residue, Asp288, which isitself involved in hydrogen bonding with both theglyceraldehyde 3-phosphate and glycerone-P components

of fructose 1,6-bisphosphate (Fig. 4). Disruption of thisnetwork of hydrogen bonds by removal of the amide sidechain of Asn35 would, therefore, be likely to affect binding atboth the C-1 and C-6 regions of the substrate molecule,consistent with all the observed kinetic behaviour of the N35Amutant. We thus conclude that Asn35 makes both direct andindirect interactions with both the triose substrates and iscrucial in their binding and orientation for catalysis.

Serine 61

Serine residues are one of the most common types of aminoacid found in enzyme phosphate binding sites [26] and thelocation of Ser61 close to the modelled position of glycer-aldehyde 3-phosphate [11], suggests that Ser61 might play arole in the binding of that triose phosphate group. Alternatively,Ser61 might interact directly with one of the substrate hydroxylgroups as found (for Ser176) in transaldolase [27] or (forSer271) in the Class I fructose 1,6-bisphosphate aldolases [28].Consistent with such an important role, Ser61 is highlyconserved across the 26 known Class II fructose 1,6-bisphos-phate aldolase sequences (conserved in 25 of the 26 knownClass II fructose 1,6-bisphosphate aldolase sequences) with

Table 2. Hexacyanoferrate (III) oxidation of the intermediate carba-

nion. Steady-state kinetics of the hexacyanoferrate (III) oxidation of the

enzyme bound enediolate intermediate for the wild-type and mutant

enzymes were measured using either glycerone-P or fructose 1,6-bisphos-

phate as substrates [14,22]. The kinetic parameters were estimated by fitting

the data to the Michaelis±Menten equation using the computer software

Kaleidagraph (Abelbeck Software). Values shown are the value of the fit ^

standard error of the fit.

Enzyme

Km0 DHAP

(mm)

k 0cat DHAP

(min21)

Km0 FBP

(mm)

k 0cat FBP

(min21)

Wild-type 0.03 �^ 0.002 166 �^ 3 0.021 �^ 0.006 108 �^ 8

N35A 0.46 �^ 0.025 51 �^ 0.71 0.032 �^ 0.004 6 �^ 0.15

S61A 0.39 �^ 0.07 76 �^ 3 1.4 �^ 0.04 22 �^ 3

S61T 0.055 �^ 0.01 121 �^ 3 0.082 �^ 0.02 95 �^ 6

K325A 0.04 �^ 0.001 23 �^ 0.1 0.04 �^ 0.01 9 �^ 0.2

Q59A 0.21 �^ 0.002 118 �^ 3 0.043 �^ 0.007 45 �^ 2

Table 1. Standard fructose 1,6-bisphosphate-cleavage assay kinetic parameters. The steady-state kinetics of the wild-type and mutant enzymes were

measured for the cleavage of fructose 1,6-bisphosphate using glycerol-3-phosphate dehydrogenase and triose phosphate isomerase in a coupled enzyme assay

[12]. The kinetic parameters were estimated by fitting the data to the Michaelis±Menten equation using the computer software kaleidagraph (Abelbeck

Software). Product inhibition studies were carried out using modifications of this assay. When glycerone-P was present the coupling enzyme was

glyceraldhyde-3-phosphate dehydrogenase, and when glyceraldehyde 3-phosphate was the inhibitor, the coupling enzyme was glycerol-3-phosphate

dehydrogenase. Values shown are the value of the fit ^ standard error of the fit. NM, not measurable. FBP, fructose 1,6-bisphosphate; G3P,

glyceraldehyde 3-phosphate.

Enzyme

Km FBP

(mm)

kcat

(min21)

Ki glycerone-P

(mm)

Kis G3P

(mm)

Kii G3P

(mm)

Wild-Type 0.17 �̂ 0.003 630 �^ 40 130 �^ 11 30 �^ 5 230 �^ 42

N35A 0.9 �^ 0.1 9.7 �^ 0.2 1090 �^ 540 450 �^ 230 1900 �^ 920

S61A 2.7 �^ 0.33 51 �^ 2 210 �^ 30 NM NM

S61T 0.43 �^ 0.052 380 �^ 14 200 �^ 30 31 �^ 9 220 �^ 28

K325A 0.38 �^ 0.04 35 �^ 0.92 82 �^ 14 110 �^ 21 1880 �^ 620

Q59A 0.22 �^ 0.03 510 �^ 18 350 �^ 40 44 �^ 14 240 �^ 92

1864 S. M. Zgiby et al. (Eur. J. Biochem. 267) q FEBS 2000

only the enzyme from Aquifex aeolicus having a conservativechange (to threonine) in this position [29].

Replacement of Ser61 with an alanine resulted in an enzymethat was compromised catalytically with respect to thewild-type enzyme (Tables 1 and 2). The results from the fructose1,6-bisphosphate cleavage assay show a large increase in Km

(16-fold) and a decrease in kcat, clearly implicating it as animportant active-site residue. The results obtained using thehexacyanoferrate (III) assays show a similar pattern, especiallywhen fructose 1,6-bisphosphate was used as substrate (a 66-foldincrease in the Km

0 value was observed in this assay; Table 2)with a less dramatic effect when glycerone-P was the substrate.In a separate series of experiments, Ser61 was also changed tothreonine. This mutant showed only minor changes in thekinetic parameters compared with those of the wild-typeenzyme (Tables 1 and 2). This demonstrates that Ser61 playsan important role in substrate binding and, furthermore,indicates that an hydroxyl group is absolutely required at thisposition.

In order to identify whether Ser61 interacts with theglycerone-P or glyceraldehyde 3-phosphate moieties of thefructose 1,6-bisphosphate molecule, product inhibition kineticswere carried out on the S61A and S61T mutants. The results(Table 1) show that the Ki value for glycerone-P was notaffected significantly in the S61A mutant compared with thewild-type enzyme. In contrast, no inhibition was observed withglyceraldehyde 3-phosphate even at concentrations . 500 mm.Thus, the Ki for glyceraldehyde 3-phosphate (30 mm for wild-type) was increased dramatically by the mutation of Ser61 toalanine. No significant differences from the wild-type enzymewere observed for the inhibition constants for the S61T mutant,again demonstrating the importance of a hydroxyl group atresidue 61 for glyceraldehyde 3-phosphate binding. Thesimplest explanation for these findings is that the Ser61hydroxyl group plays a major role in binding glyceralde-hyde 3-phosphate and the C6 end of the substrate fructose1,6-bisphosphate. These results accord with the suggestion thatthe hydroxyl group of Ser61 participates in the catalyticreaction by forming a hydrogen bond with the phosphate groupof glyceraldehyde 3-phosphate [11].

Lysine 325

The final residue highlighted by Hall et al. [11] as potentiallyfulfilling a role in the glyceraldehyde 3-phosphate binding siteis Lys325. It lies close to the enzyme active site of the partner

Fig. 4. Schematic representation of the active site of the E. coli Class II

fructose 1,6-bisphosphate aldolase. Active site residues of the Class II

fructose 1,6-bisphosphate aldolase of E. coli that interact with substrate

(bold) are shown. The position of glycerone-P is inferred from the location

of phosphoglycolohydroxamate in the crystal structure of its complex with

the enzyme and glyceraldehyde 3-phosphate is placed in the position

modelled in Hall et al. [11]. Dashed lines indicate possible hydrogen bonds.

Fig. 5. Stereoview of the active site of the E. coli Class II fructose 1,6-bisphosphate aldolase. Molecular graphics representation of a model of

glyceraldehyde 3-phosphate and the transition state analogue phosphoglycolohydroxamate in the active site of the Class II fructose 1,6-bisphosphate aldolase

of E. coli [11]. Broken lines indicate possible interactions between substrates and the enzyme. The carbon atoms of Arg331 and Lys325 from the partner

subunit are in green.

q FEBS 2000 The active site of Class II fructose 1,6-bisphosphate aldolases (Eur. J. Biochem. 267) 1865

subunit in the dimer of the enzyme and a lysine residue isstrongly conserved (70%) in this position. Mutation of Lys325to alanine resulted (Tables 1 and 2) in a slight increase(twofold) in the Km for fructose 1,6-bisphosphate in thestandard fructose 1,6-bisphosphate cleavage assay, togetherwith a substantial decrease in kcat (18-fold), compared with thewild-type. Similar results were determined for this mutant inthe hexacyanoferrate (III) assay (no significant change in K 0mfor either glycerone-P or fructose 1,6-bisphosphate andsevenfold and 12-fold decreases in k 0cat for glycerone-P andfructose 1,6-bisphosphate, respectively). Inhibition studies withglycerone-P or glyceraldehyde 3-phosphate show that the Ki forglycerone-P for the K325A mutant is not significantly differentto the wild-type enzyme. However, the inhibition constantsmeasured for glyceraldehyde 3-phosphate are approximatelyfivefold higher than the wild-type. Taken together, these resultsdemonstrate an important, but complex, role for Lys325 incatalysis. The K325A mutation clearly has a larger effect at theglyceraldehyde 3-phosphate end of the enzyme active site thanat the glycerone-P subsite, thus mapping its interaction to thepreviously unmapped end of the active site. The changes inkcat for the K325A enzyme indicate that Lys325 may play adirect role in catalysis or more probably given the magnitude ofthe effects on kcat upon mutation, that it is important inmaintaining other critical residues in a catalytically competentconformation.

Glutamine 59

In addition to the residues highlighted by Hall et al. [11] ashaving a potential role in glyceraldehyde 3-phosphate binding,we noticed that Gln59 might also be important in glyceralde-hyde 3-phosphate binding (Fig. 5). Gln59 is located 6.2 AÊ fromthe active site zinc and is positioned to accept a hydrogen bondfrom Lys284 [10]. In turn, Lys284 is a ligand for either amonovalent cation or a tightly bound water molecule that isclosely associated with the active-site zinc and may perform astructural role in ensuring proper active site alignment. As

would be expected for an important residue, Gln59 is highlyconserved in the Class II fructose 1,6-bisphosphate aldolasefamily, being present in 22 of the 26 sequences now known. Wetherefore explored the effects of mutating Gln59 to an alanine.

This change resulted in only minor changes in the kineticparameters (Tables 1 and 2), although the Km

0 value forglycerone-P in the carbanion oxidation assay was slightlyhigher (sixfold) than the value for the wild-type protein.Comparison of the Ki values for glycerone-P and glyceralde-hyde 3-phosphate with those for the wild-type enzyme alsoshow only minor variations from the wild-type parameters(Table 1). These results rule out the possibility that Gln59 isdirectly involved in binding of the substrate.

The results presented above identified two new interactions(with Asn35 and Ser61) at the previously unmapped glyceralde-hyde 3-phosphate-binding end of the enzyme active site andsupport the model of glyceraldehyde 3-phosphate built into theactive site of the E. coli Class II fructose 1,6-bisphosphatealdolase [11]. Furthermore, Lys325 has been shown to play animportant role in enzyme catalysis. Taken together withprevious results [9±14] we now have a detailed understandingat the molecular level of the residues involved in binding andcatalysis in Class II fructose 1,6-bisphosphate aldolases (Figs 4and 5).

Discrimination between fructose bisphosphate and tagatosebisphosphate

A second important objective is to understand the origin ofsubstrate specificity in the aldolase family of enzymes. In thisrespect, the aldol cleavage of tagatose 1,6-bisphosphate bytagatose-1,6-bisphosphate aldolase offers an ideal opportunity.Tagatose 1,6-bisphosphate and fructose 1,6-bisphosphate arediastereoisomers at C4 (Fig. 2) and the Class II enzymesresponsible for their cleavage in E. coli are clearly related tothe Class II fructose 1,6-bisphosphate aldolase. Despite this,the enzymes are highly specific for their cognate substrates.The major interaction of the enzyme with the C4 position in

Table 3. Kinetics of wild-type and mutant aldolases with fructose 1,6-bisphosphate or tagatose 1,6-bisphosphate. Steady-state kinetics of the wild-type

and mutant enzymes were measured for the cleavage of fructose 1,6-bisphosphate by a coupled enzyme assay [12] or for the cleavage of tagatose 1,6-

bisphosphate by a modification of this assay in which tagatose 1,6-bisphosphate replaced fructose 1,6-bisphosphate (see Materials and methods). The kinetic

parameters were estimated by fitting the data to the Michaelis±Menten equation using the computer software kaleidagraph (Abelbeck Software). FBP,

fructose 1,6-bisphosphate; TBP, tagatose 1,6-bisphosphate. Values shown are the value of the fit ^ standard error of the fit. NM, not measurable even with

1 mg of enzyme and tagatose 1,6-bisphosphate concentrations up to 4 mm.

FBP TBP

Enzyme

kcat

(min21)

Km

(mm)

kcat /Km

(min21´mm21)

kcat

(min21)

Km

(mm)

kcat /Km

(min21´mm21)

kcat /Km(FBP)/

kcat /Km(TBP)

Wild-type FBP aldolase 630 �̂ 4 0.17 �^ 0.003 3700 0.9 �^ 0.08 0.35 �^ 0.06 2�.6 1423

Wild-type AgaY TBP aldolase 4.1 �^ 0.4 1.3 �^ 0.3 3�.2 280 �^ 10 0.26 �^ 0.03 1080 0�.003

S61T 380 �̂ 14 0.43 �^ 0.052 880 0.8 �^ 0.02 0.40 �^ 0.26 2 440

Q59A 510 �̂ 18 0.22 �^ 0.03 2320 0.61 �^ 0.03 0.38 �^ 0.04 1�.6 1450

G64T 160 �̂ 6 0.20 �^ 0.03 800 0.77 �^ 0.02 0.14 �^ 0.02 5�.5 145

D288A 130 �̂ 7 1.0 �^ 0.02 130 0.41 �^ 0.02 0.36 �^ 0.07 1�.1 118

F67H 800 �̂ 0.03 0.21 �^ 0.02 3810 0.43 �^ 0.02 0.32 �^ 0.03 1�.3 2930

K325A 35 �^ 0.92 0.38 �^ 0.04 92 N.M. N.M. N.M. ±

Q59A/F67H 440 �^ 14 0.55 �^ 0.06 800 0.79 �^ 0.03 0.16 �^ 0.03 4�.9 163

Q59A/G64T/F67H 11 �^ 0.6 5.9 �^ 0.7 1�.9 0.38 �^ 0.03 0.19 �^ 0.06 2 0�.95

S61T/G64T/F67H 13 �^ 0.2 0.2 �^ 0.01 65 1.3 �^ 0.03 0.18 �^ 0.02 7�.2 9

Q59A/S61T/D288A 2.9 �^ 0.1 1.7 �^ 0.2 1�.7 0.55 �^ 0.02 0.23 �^ 0.04 2�.4 0�.71

1866 S. M. Zgiby et al. (Eur. J. Biochem. 267) q FEBS 2000

fructose 1,6-bisphosphate is made by Asp109 and this residuehas been shown to be responsible for deprotonation of theC4-OH group during fructose 1,6-bisphosphate cleavage [13]. Itis interesting to note that an aspartic acid residue (Asp81 inAgaY) is conserved in the equivalent positions in the AgaY andGatY tagatose 1,6-bisphosphate aldolases. It is likely thereforethat Asp109 and its equivalent (Asp81) play the same role in thetwo enzymes and that it is the subtleties of substrate bindingthat position the relevant groups for substrate discrimination.We, therefore, decided to investigate the molecular basis of thisspecificity.

In order to provide comparative details, one of the tagatose1,6-bisphosphate aldolases from E. coli (AgaY) was cloned andoverexpressed. Oligonucleotides were designed to prime at the5 0-end and 3 0-end of the agaY gene in the E. coli chromosomeand to introduce EcoRI and HindIII sites at the 5 0-end and3 0-end, respectively, for subsequent subcloning. The full-lengthgene was amplified from chromosomal DNA by PCR and wassubcloned into the expression vector pKK223-3. High levels ofexpression of a protein of m < 30 kDa were detected in E. colicells harbouring this construct, which was designated pKagaY.Cells containing this vector were also shown to have high levelsof tagatose bisphosphate aldolase activity. The enzyme waspurified to homogeneity by a combination of ion exchange,hydrophobic and gel-exclusion chromatography. Electrosprayionization mass spectrometry showed the subunit molecularmass of the protein to be 31165.9 ^ 2.95 Da, in excellentagreement with the mass expected from the published genesequence minus the N-terminal methionine (31162.7 Da). Theprotein was judged to be tetrameric by gel-filtration chromato-graphy (data not shown) and steady-state kinetic parametersmeasured for the enzyme were in agreement with apreviously purified but otherwise unidentified tagatose 1,6-bisphosphate aldolase from E. coli [30].

Table 3 shows the kinetic parameters for the wild-typeClass II fructose 1,6-bisphosphate aldolase and the wild-typeAgaY tagatose-1,6-bisphosphate aldolase measured for activitywith either fructose 1,6-bisphosphate or tagatose 1,6-bisphos-phate. The fructose 1,6-bisphosphate aldolase is highly specificfor fructose 1,6-bisphosphate, preferring this substrate almost1500-fold over tagatose 1,6-bisphosphate as judged by the ratioof the kcat /Km values. In contrast, the tagatose 1,6-bisphosphatealdolase shows a strong preference for tagatose 1,6-bisphos-phate as substrate: (kcat /Km(FBP)/kcat /Km(TBP) � 0.003; seeTable 3). The overall discrimination between the two substratesis therefore almost 5 � 105. This exquisite discriminationbetween substrates must be brought about by the residues liningthe active site of the enzymes, despite the fact that the E. colifructose 1,6-bisphosphate aldolase and AgaY show 28% overallsequence identity, and . 50% identity (. 70% conserved)when only those residues that lie within 5 AÊ of phospho-glycolohydroxamate or the modelled glyceraldehyde 3-phos-phate [11] are considered (Fig. 3).

In order to pinpoint the roles of individual amino acids in thissubstrate discrimination, residues were targeted for site-directed mutagenesis based on an analysis of the crystalstructure of the fructose 1,6-bisphosphate aldolase both in thepresence [11] and absence of phosphoglycolohydroxamate [10],combined with multiple sequence alignment of the knownClass II fructose 1,6-bisphosphate and tagatose 1,6-bispho-sphate aldolases (Fig. 3). We assume here that residues that liewithin 5 AÊ of the substrate modelled in the active site offructose 1,6-bisphosphate aldolase are largely responsible forsubstrate discrimination. We thus looked for residues thatare highly conserved within each group of enzymes

(fructose 1,6-bisphosphate aldolases or tagatose 1,6-bisphos-phate aldolases), but are not conserved between the groups.Gln59, Asn62, Gly64, Val225, Asp-288, Gln292 and Lys325were thus identified as being potentially involved in substratediscrimination. The equivalent residues in the AgaY tagatose1,6-bisphosphate aldolase are Ala, Pro, Thr, Ala, Ala, Lys andPro, respectively. Of these residues only Gln59, Asp288 andLys325 appear to be in a position in the enzyme where theycould directly influence substrate discrimination at C4 of thehexose sugar. These residues were therefore chosen as initialtargets for site-directed mutagenesis. We also investigatedthe effect of mutation of Gly64 to determine whether stericconstraint in the active site might influence choice ofsubstrate by the enzyme. In addition to the residuesdescribed above, Ser61, Gly266, Ile287 and Asp290 in thefructose 1,6-bisphosphate aldolase lie within 5 AÊ of theproposed substrate binding sites and are conservativelyreplaced by Thr, Ala, Val and Glu, respectively, in AgaYtagatose 1,6-bisphosphate aldolase. Ser61 has been shownabove to be involved in substrate binding and the effect of itsmutation on substrate discrimination was thus also studied.

In most cases, the residue in fructose 1,6-bisphosphatealdolase was changed to the equivalent residue in the tagatose1,6-bisphosphate aldolases, although we were concerned thatsubstitution of Lys325 with a proline might adversely affect theenzyme fold, and hence used the K325A mutant describedabove. All the mutant proteins (Q59A, S61T, G64T, D288A andK325A) were expressed in E. coli as soluble enzymes and werepurified as described above. CD spectroscopy showed that therewere no gross structural changes from the wild-type enzyme(data not shown). The mutants were then characterized byenzyme kinetics (Table 3).

The Q59A mutation was slightly less active and had aslightly raised Km for the hexose substrate than the wild-type fructose 1,6-bisphosphate aldolase in both the fructose1,6-bisphosphate and tagatose 1,6-bisphosphate aldolase assaysand the ratio of (kcat /Km)FBP/(kcat /Km)TBP was unaffected by themutation. This result rules out the hypothesis that Gln59 aloneis responsible for substrate discrimination between fructose1,6-bisphosphate and tagatose 1,6-bisphosphate. In contrast, theS61T mutant is less specific for fructose 1,6-bisphosphate thanthe wild-type enzyme (fructose 1,6-bisphosphate used < 450times better than tagatose 1,6-bisphosphate in the mutant,compared with 1420 times better for the wild-type; Table 3).However, this difference reflects the decrease in kcat withfructose 1,6-bisphosphate as substrate in the S61T mutantenzyme and there is no increase in activity with tagatose1,6-bisphosphate as substrate. This result shows that althoughSer61 provides part of the binding site for the substrate (seeabove), the single replacement of this residue by a threoninecannot explain the discrimination between fructose 1,6-bis-phosphate and tagatose 1,6-bisphosphate. This further suggeststhat Ser61 binds to part of the substrate common to bothfructose 1,6-bisphosphate and tagatose 1,6-bisphosphate and, inaccordance with the results described above, it is likely thatSer61 forms a hydrogen bond with the C6 phosphate group.

The mutation of Gly64 to threonine results in an enzyme thatdoes not function well with fructose 1,6-bisphosphate assubstrate (kcat reduced by fourfold compared with the wild-type enzyme; Table 3). With tagatose 1,6-bisphosphate assubstrate, the G64T mutant shows a marginal increase (twofold)in the (kcat /Km)TBP value because of a slight decrease inthe measured Km for tagatose 1,6-bisphosphate. However,this enzyme is relatively inactive with either substrate andstill utilizes fructose 1,6-bisphosphate better than tagatose

q FEBS 2000 The active site of Class II fructose 1,6-bisphosphate aldolases (Eur. J. Biochem. 267) 1867

1,6-bisphosphate [(kcat/Km)FBP/(kcat/Km)TBP � 145]. The D288Amutant has already been shown to affect the binding of fructose1,6-bisphosphate, causing a significant increase in the Km forthat substrate [13]. However, this change is not accompanied byany change in the kinetic parameters measured with tagatose1,6-bisphosphate as substrate (Table 3). Finally, the K325Amutant showed no activity with tagatose 1,6-bisphosphate assubstrate even when < 1 mg of enzyme was used per assaywith tagatose 1,6-bisphosphate concentrations up to 4 mm.

None of the single amino-acid changes investigated show anysignificant alterations in the substrate specificities shown by theE. coli Class II fructose 1,6-bisphosphate aldolase. We there-fore investigated the role of Phe67 in substrate discrimination.Phe67 is conserved as either a phenylalanine or a tyrosine in allthe fructose 1,6-bisphosphate aldolases, whereas in the AgaYand GatY tagatose 1,6-bisphosphate aldolases a histidineresidue is found in this position. Phe67 lies < 8 AÊ from theputative binding site for the phosphate group of glyceraldehyde3-phosphate and it is difficult to predict how an amino acidsubstitution at this distance from the substrate could affectdiscrimination. Nevertheless, mutations this far from a substratecan have a significant impact on specificity. For example, themutation of Gly179 in glutathione reductase dramaticallychanges the coenzyme specificity despite the fact that itlies < 9.5 AÊ from the 2 0-phosphate of NADPH [31,32].Phe67 of fructose 1,6-bisphosphate aldolase was thereforemutated to histidine and the mutant protein was produced andexpressed as for the other mutants. The kinetic parameters forfructose 1,6-bisphosphate were then measured. There was aslight increase in kcat (800 min21 compared with 630 min21 forwild-type) and no significant change in Km. There were also nosignificant changes in kcat or Km for tagatose 1,6-bisphosphateas substrate. No single mutation has therefore been found toaccount for the differences in specificity between the twoenzymes.

Determination of the specificity may be spread over anumber of active site residues. We therefore constructed anumber of double and triple mutants of the residues describedabove (Q59A/F67H; Q59A/G64T/F67H; S61T/G64T/F67H andQ59A/S61T/D288A). All these mutant enzymes were muchless active with fructose 1,6-bisphosphate than the wild-typeenzyme (Table 3) with both the Q59A/G64T/F67H and Q59A/S61T/D288A mutants having Km values for fructose 1,6-bis-phosphate . 1.5 mm. None of the mutants, however, showedany significant changes in the kinetic parameters measured fortagatose 1,6-bisphosphate. The overall result of these experi-ments is therefore that the Q59A/G64T/F67H and Q59A/S61T/D288A mutants both utilize tagatose 1,6-bisphosphate inpreference to fructose 1,6-bisphosphate, but this is caused bythe loss of activity with fructose 1,6-bisphosphate rather thanany gain in activity with tagatose 1,6-bisphosphate as substrate.

The mechanism of substrate discrimination between fructose1,6-bisphosphate and tagatose 1,6-bisphosphate thus remainsunclear. It has been proposed that the loop carrying Glu182undergoes a conformational change on substrate binding andthis closure over the substrates in the active site may beresponsible for the correct positioning of the cognate substratein the active site, together with rejection of the `wrong'substrate. It is also noteworthy that the tagatose 1,6-bisphos-phate aldolases have several deletions in their primarysequences compared with the fructose 1,6-bisphosphate aldo-lase (Fig. 3) and the determinants of specificity may lie inthese loops. Future X-ray crystallographic work on bothfructose 1,6-bisphosphate aldolase and tagatose 1,6-bisphos-phate aldolase, including the structures in complex with

substrates or substrate analogues, combined with loop-engineering experiments will thus be required in order toprovide a definitive answer to the fascinating question ofsubstrate discrimination in these enzymes and to permit newaldolases to be generated at will.

A C K N O W L E D G E M E N T S

This work was supported by The Wellcome Trust and forms a contribution

from The Astbury Centre for Structural Molecular Biology at The

University of Leeds which is funded by the BBSRC. We thank Prof

Woody Fessner for the gift of tagatose bisphosphate. We are grateful to Drs

Bill Hunter and Dave Hall (University of Dundee) for useful discussions.

We thank Dr Alison Ashcroft for the electrospray mass spectrometry of the

AgaY enzyme.

R E F E R E N C E S

1. Franklin, A.S. & Paterson, I. (1994) Recent developments in

asymmetric aldol methodology. Contemp. Org. Synth. 1, 317.

2. Barbas, C.F., Heine, A., Zhong, G.F., Hoffmann, T., Gramatikova, S.,

Bjornestedt, R., List, B., Anderson, J., Stura, E.A., Wilson, I.A. &

Lerner, R.A. (1997) Immune versus natural selection: antibody

aldolases with enzymic rates but broader scope. Science 278,

2085±2092.

3. Fessner, W. (1998) Enzyme mediated C±C bond formation. Curr. Opin.

Chem. Biol. 2, 85±97.

4. Wong, C.-H. & Whitesides, G.M. (1994) Enzymes in Synthetic Organic

Chemistry. Pergamon, Oxford, UK.

5. Rutter, W.J. (1964) Evolution of aldolase. Fed. Proc. 23, 1248±1257.

6. Kobes, R.D., Simpson, R.T., Vallee, B.L. & Rutter, W.J. (1969) A

functional role of metal ions in a Class II aldolase. Biochemistry 8,

585±588.

7. Harris, C.E., Kobes, R.D., Teller, D.C. & Rutter, W.J. (1969) The

molecular characteristics of yeast aldolase. Biochemistry 8,

2442±2454.

8. Mildvan, A.S., Kobes, R.D. & Rutter, W.J. (1971) Magnetic resonance

studies of the divalent cation in the mechanism of yeast aldolase.

Biochemistry 10, 1191±1204.

9. Blom, N.S., Tetreault, S., Coulombe, R. & Sygusch, J. (1996) Novel

active site in Escherichia coli fructose-1,6-bisphosphate aldolase.

Nat. Struct. Biol. 3, 856±862.

10. Cooper, S.J., Leonard, G.A., McSweeney, S.M., Thompson, A.W.,

Naismith, J.H., Qamar, S., Plater, A., Berry, A. & Hunter, W.N.

(1996) The crystal structure of a Class II fructose-1,6-bisphosphate

aldolase shows a novel binuclear metal-binding site embedded in a

familiar fold. Structure 4, 1303±1315.

11. Hall, D.R., Leonard, G.A., Reed, C.D., Watt, C.I., Berry, A. & Hunter,

W.N. (1999) The crystal structure of Escherichia coli Class II

fructose-1,6-bisphosphate aldolase in complex with phospho-

glycolohydroxamate reveals details of the mechanism and specificity.

J. Mol. Biol. 287, 383±394.

12. Berry, A. & Marshall, K.E. (1993) Identification of zinc-binding

ligands in the Class II fructose-1,6-bisphosphate aldolase of E. coli.

FEBS Lett. 318, 11±16.

13. Plater, A.R., Zgiby, S.M., Thomson, G.J., Qamar, S., Wharton, C.W. &

Berry, A. (1999) Conserved residues in the mechanism of the E. coli

Class II FBP-aldolase. J. Mol. Biol. 285, 843±855.

14. Qamar, S., Marsh, K.L. & Berry, A. (1996) Identification of arginine

331 as an important active site residue in the Class II fructose-

1,6-bisphosphate aldolase of E. coli. Protein Sci. 5, 154±161.

15. Reizer, J., Ramseier, T.M., Reizer, A., Charbit, A. & Saier, M.H.J.

(1996) Novel phosphotransferase genes revealed by bacterial genome

sequencing: a gene cluster encoding a putative N-acetylgalactosamine

metabolic pathway in Escherichia coli. Microbiology 142, 231±250.

16. Nobelmann, B. & Lengeler, J.W. (1995) Sequence of the gat operon for

galactitol utilization from a wild-type strain EC3132 of Escherichia

coli. Biochim. Biophys. Acta 1262, 69±72.

17. Nobelmann, B. & Lengeler, J.W. (1996) Molecular analysis of the gat

1868 S. M. Zgiby et al. (Eur. J. Biochem. 267) q FEBS 2000

genes from Escherichia coli and their roles in galactitol transport and

metabolism. J. Bacteriol. 178, 6790±6795.

18. Smith, P.K., Krohn, R.I., Hermanson, G.T., Mallia, A.K., Gartner, F.H.,

Provenzano, M.D., Fujimoto, E.K., Goeke, N.M., Olson, B.J. &

Klenk, D.C. (1985) Measurement of protein using bicinchoninic acid.

Anal. Biochem. 150, 76±85.

19. Blattner, F.R., Plunkett, G., Bloch, C.A., Perna, N.T., Burland, V.,

Riley, M., ColladoVides, J., Glasner, J.D., Rode, C.K., Mayhew, G.F.,

Gregor, J., Davis, N.W., Kirkpatrick, H.A., Goeden, M.A., Rose, D.J.,

Mau, B. & Shao, Y. (1997) The complete genome sequence of

Escherichia coli K-12. Science 277, 1453±1474.

20. Kirby, K.S., Fox-Carter, E. & Guest, M. (1967) Isolation of

deoxyribonucleic acid and ribonucleic acid from bacteria. Biochem.

J. 104, 258±262.

21. Cleland, W.W. (1979) Statistical analysis of enzyme kinetic data.

Methods Enzymol. 63, 103±138.

22. Healy, M.J. & Christen, P. (1973) Mechanistic probes for enzymic

reactions. Oxidation±reduction indicators as oxidants of intermediary

carbanions (studies with aldolase, aspartate aminotransferase,

pyruvate decarboxylase and 6-phosphogluconate dehydrogenase).

Biochemistry 12, 35±40.

23. Reikofski, Y. & Tao, B.Y. (1992) Polymerase chain reaction (PCR)

techniques for site-directed mutagenesis. Biotech. Adv. 10, 535±547.

24. Rose, I.A. (1958) The absolute configuration of dihydroxyacetone

phosphate tritiated by aldolase reaction. J. Am. Chem. Soc. 80,

5835±5836.

25. Hill, H.A., Lobb, R.R., Sharp, S.L., Stokes, A.M., Harris, J.I. & Jack,

R.S. (1976) Metal-replacement studies in Bacillus stearothermo-

philus aldolase and a comparison of the mechanisms of Class I and

Class II aldolases. Biochem. J. 153, 551±560.

26. Copley, R.R. & Barton, G.J. (1994) A structural analysis of phosphate

and sulphate bindings sites in proteins: estimations of propensities for

binding and conservation of phosphate binding sites. J. Mol. Biol.

242, 321±329.

27. Jia, J., Schorken, U., Lindqvist, Y., Sprenger, G.A. & Schneider, G.

(1997) Crystal structure of the reduced Schiff-base intermediate

complex of transaldolase B from Escherichia coli: mechanistic

implications for Class I aldolases. Protein Sci. 6, 119±124.

28. Blom, N.S. & Sygusch, J. (1997) Product binding and role of the

C-terminal region in Class I d-fructose 1,6-bisphosphate aldolase.

Nat. Struct. Biol. 4, 36±39.

29. Deckert, G., Warren, P.V., Gaasterland, T., Young, W.G., Lenox, A.L.,

Graham, D.E., Overbeek, R., Snead, M.A., Keller, M., Aujay, M.,

Huber, R., Feldman, R.A., Short, J.M., Olson, G.J. & Swanson, R.V.

(1998) The complete genome of the hyperthermophilic bacterium

Aquifex aeolicus. Nature 392, 353±358.

30. Eyrisch, O., Sinerius, G. & Fessner, W.D. (1993) Facile enzymatic de

novo synthesis and NMR characterisation of d-tagatose 1,6-bisphos-

phate. Carbohydrate Res. 238, 287±306.

31. Scrutton, N.S., Berry, A. & Perham, R.N. (1990) Redesign of the

coenzyme specificity of a dehydrogenase by protein engineering.

Nature 343, 38±43.

32. Mittl, P.R.E., Berry, A., Scrutton, N.S., Perham, R.N. & Schulz, G.E.

(1993) Structural differences between wild-type NADP-dependent

glutathione reductase from Escherichia coli and a redesigned

NAD-dependent mutant. J. Mol. Biol. 231, 191±195.

33. Alefounder, P.R., Baldwin, S.A., Perham, R.N. & Short, N.J. (1989)

Cloning, sequence analysis and over-expression of the gene for the

Class II fructose 1,6-bisphosphate aldolase of Escherichia coli.

Biochem. J. 257, 529±534.

34. Bouma, C.L. & Roseman, S. (1996) Sugar transport by the marine

chitinolytic bacterium Vibrio furnissii. J. Biol. Chem. 271,

33468±33475.

35. Devereux, J., Haeberli, P. & Smithies, O. (1984) A comprehensive set

of sequence analysis programs for the VAX. Nucleic Acids Res. 12,

387±395.