STRUCTURES OF Staphylococcus aureus D-TAGATOSE-6-PHOSPHATE KINASEIMPLICATE DOMAIN MOTIONS IN SPECIFICITY AND MECHANISM

Linda Miallau1, 2, William N. Hunter2, Sean M. McSweeney1 and Gordon A. Leonard1*,§.From the 1Macromolecular Crystallography Group, European Synchrotron Radiation Facility, 38043Grenoble, France.2 Division of Biological Chemistry and Molecular Microbiology, College of Life

Sciences, University of Dundee, Dundee DD1 5EH, Scotland, U.K

Running Title: Crystal Structures of Staphylococcus aureus D-tagatose-6-phosphate kinaseAddress correspondence to: Dr. Gordon A. Leonard, Macomolecular Crystallography Group,European Synchrotron Radiation Facility, 6 rue Jules Horowitz, 38043 Grenoble Cedex, France. Tel:+33-4-76-88-23-94; Fax: +33-4-76-88-29-04; E-Mail: leonard@esrf.fr

High-resolution structures ofStaphylococcus aureus D-tagatose-6-phosphate kinase (LacC) in two crystalforms are reported. The structures defineLacC in apo form, in binary complexes withADP or the co-factor analogue AMP-PNPand in a ternary complex with AMP-PNPand D-tagatose-6-phosphate. The tertiarystructure of the LacC monomer, which isclosely related to other members of the pfkBsubfamily of carbohydrate kinases, iscomposed of a large α/β core domain and asmaller, largely β, 'lid'. Four extendedpolypeptide segments connect these twodomains. Dimerization of LacC occurs viainteractions between lid domains, whichcome together to form a β-clasp structure.Residues from both subunits contribute tosubstrate binding. LacC adopts a closedstructure required for phosphoryl transferonly when both substrate and cofactor arebound. A reaction mechanism similar tothat used by other phosphoryl transferasesis proposed although, unusually, when bothsubstrate and co-factor are bound to theenzyme two Mg2+ ions are observed in theactive site. A new motif of amino acidsequence conservation common to the pfkBsubfamily of carbohydrate kinases isidentified.

INTRODUCTIONIn those bacteria that use lactose as a

carbohydrate source, the disaccharide isusually either converted to glucose through theLeloir pathway (1) then metabolized byglycolysis (2) or catabolized directly throughthe D-tagatose-6-phosphate pathway (3).Staphylococcus aureus, S. epidermidis, and S.hominis are the only organisms known toexclusively use enzymes of the D-tagatose-6-

phosphate pathway to metabolize lactose andD-galactose (4).

In S. aureus, D-galactose and lactoseare imported and metabolised by proteinsencoded by the lactose operon, lacABCDFEG -the gene products of which have been shown tobe inducible by the addition of either D-galactose or lactose (5) or by D-galactose-6-phosphate (6). In fact, the Lac operon was thefirst discovered example of a group of genesunder the control of an operator region towhich a repressor (lacR) binds (7). Lactose isconverted by transglycosylation intoallolactose. This binds to the repressor, inhibitsbinding to the operator and allows thetranscription of mRNA for enzymes involvedin D-galactose metabolism and transport acrossthe membrane (8).

In the D-tagatose-6-phosphate pathway(encoded by lacABCD), the first step, carriedout by a D-galactose-6-phosphate isomeraseencoded by lacAB (9), converts D-galactose-6-phosphate into D-tagatose-6-phosphate. Thesecond step is an ATP-dependentphosphorylation of D-tagatose-6-phosphate byD-tagatose-6-phosphate kinase (LacC) thatyields D-tagatose-1,6-bisphosphate (Figure 1).A divalent cation is required for this reactionwith Mg2+ giving the optimum reaction rate(10). Finally D-tagatose-1,6-bisphosphatealdolase (LacD) cleaves D-tagatose-1,6-bisphosphate to produce glyceraldehyde 3-phosphate and dihydroxyacetone phosphate (5,11). The pathway is essential for the survivalof S. aureus in lactose rich media such as milksince if it does not function properly D-galactose accumulates and bacterial growth isinhibited (12). Interfering with this pathwaymay therefore be a useful strategy foreradicating infection by S. aureus. Such

1

http://www.jbc.org/cgi/doi/10.1074/jbc.M701480200The latest version is at JBC Papers in Press. Published on April 25, 2007 as Manuscript M701480200

Copyright 2007 by The American Society for Biochemistry and Molecular Biology, Inc.

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

infections can be important in breast feedingmothers (13) and are common in livestockwhere, if not treated, they cause damage tomilk secreting tissues affecting both qualityand quantity of dairy products (14). Structuralinformation concerning the enzymes in the D-tagatose-6-phosphate pathway will play a keyrole in elucidating a mechanism for itsinhibition. Currently, such information is onlyavailable for LacD (11).

On the basis of amino acid sequencecomparisons LacC belongs to the pfkB familyof carbohydrate kinases, which form a subsetof the ribokinase superfamily. Members of thissuperfamily include ribokinase (RbsK), theminor isoform D-fructose-6-phosphate kinase(PfkB), 1-phosphofructokinase (FruK) andinosine-guanosine kinase (15). A multiplesequence alignment of five members of theribokinase superfamily including LacC from S.aureus (Figure 2) shows regions of sequenceconservation. Two of these (Motifs I and II) areconserved even in the sequence of humanadenosine kinase (16) and represent theribokinase superfamily signature. The crystalstructure of Escherichia coli ribokinaseindicates that these motifs are involved insubstrate recognition, the catalytic mechanismand transition state stabilization (17).

We present the structure of LacC intwo crystal forms. In order to solve thestructure it was necessary to incorporate twoadditional selenomethionine (SeMet) residuesby site-directed mutagenesis. Fortuitously, thecrystals present a number of different states ofthe enzyme and the structures provide detailsof apo-LacC, binary complexes with ADP orthe co-factor analogue adenosine 5’-(β,γ-imido) triphosphate (AMP-PNP, selected sinceit is a relatively stable triphosphate) and aternary complex with AMP-PNP and D-tagatose-6-phosphate. The structures illustrateaspects of co-factor and substrate recognitionand suggest a mechanism for phosphoryltransfer dependent on conformational changes.We also identify a new region of amino acidsequence conservation specific to the pfkBsubfamily of carbohydrate kinases.

EXPERIMENTAL PROCEDURESCloning of LacC from S. aureusThe gene encoding tagatose-6-phosphatekinase (lacC) from S. aureus (genbankaccession number: X14827) was amplified byPCR using the specific primers 5’-CAT ATG

ATT TTG ACT TTG ACA TTA AAC C-3’(forward) and 5’-GGA TCC TTA CAC CTCTAA AAC TTC-3’ (reverse). The forwardprimer carries an NdeI restriction site and thereverse primer a BamHI restriction site. Theresulting PCR product was cloned into apET15b expression vector (Novagen) and theresulting plasmid heat-shock transformed intoE. coli strain BL21 (DE3) (Stratagene).

For the Leu-124-Met; Leu-125-Metdouble mutant of S. aureus LacC, PCR primerscontaining the mutations (forward primer:5’-TTT ATT AAA CAT TTT GAA CAAATG ATG GAA AAA GTT GAA GCA GTTGCT-3’, reverse primer: 5’-AGC AAC TGCTTC AAC TTT TTC CAT CAT TTG TTCAAA ATG TTT AAT AAA-3’ - mutationsshown in bold) were prepared using aQuickChange® site-directed mutagenesis kit(Stratagene), the PCR product cloned into apET15b expression vector and the resultingplasmid purified and sequenced to check thepresence of the desired mutation. The plasmidwas then heat-shock transformed into E. colistrain BL21 (DE3).

Recombinant LacC was producedfollowing similar protocols to those employedfor 4-diphosphocytidyl-2C-methyl-D-erythritolkinase (18). After cultivation, cells wereharvested and resuspended in 50 mM Tris-HCl,pH 7.7 containing 50 mM NaCl (buffer A).Cell lysis was performed with a cell disruptorat 4 ºC and the cell suspension was centrifugedat 20,000 x g for 30 minutes. The filteredsupernatant was applied to a Hitrap metal-chelating column (Amersham Biosciences).The pure protein, eluted in buffer A and 150mM imidazole, was dialyzed overnight at 4 ºCagainst buffer A and then concentrated to 34mg/ml.

An aliquot of the pure apo form of theprotein was injected onto a high resolutionSuperdex 200 HR 10/30 size exclusion column(Amersham Biosciences). The elution profile(data not shown) was compared to that ofvarious protein size markers (obtained fromSigma Aldrich) and indicated that LacC formsa dimer in solution.

To prepare the selenomethionylderivatives of LacC and the Leu-124-Met; Leu-125-Met double mutant, BL21 (DE3) cells,heat-shock transformed with the pET15b/lacCS. aureus, were grown in 3 ml of LBsupplemented with 100 μg/ml of ampicillin andgrown at 37 ºC for 3 hours. The culture was

2

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

harvested and cells resuspended in 25 ml ofM9 media supplemented with 100 μg/ml ofampicillin and allowed to grow for another 15hours at 37 ºC. These cells were used toinoculate 1 L of M9 media. When theabsorbance at 600 nm reached 0.56 at 37 ºC,solid amino-acid supplements were added tothe culture as follows: 100 mg/l of L-lysine, L-phenylalanine, L-threonine, 50 mg/l of L-isoleucine, L-leucine, L-valine and 60 mg/l ofSeMet. Following a delay of 15 minutes toallow inhibition of methionine synthesis, geneexpression was induced with 1 mM IPTG andthe culture left overnight. The samepurification protocol described for the nativeenzyme was then followed. The finalconcentration of the stock solution in buffer Awas 17.0 mg/ml.

Crystallization and data collectionThe wild-type protein gave crystals

that diffracted only to low resolution, about3.3Å. We proceeded to make selenomethionyl-LacC for structural analyses and discoveredthat, for reasons not understood, the SeMetprotein produced much more ordered crystalsso we worked only with them. Block-shapedmonoclinic crystals (crystal form I) of SeMet-LacC pre-incubated with 5 mM of ATP wereobtained from hanging drops consisting of 1.0μl of the protein stock solution (8.5 mg/ml) and1.0 μl of a reservoir solution comprising 13 %PEG 20 000 and 0.1 M MES pH 6.8 diffusedagainst 0.5 ml of the reservoir solution towhich 0.2 μl of 20 % w/v benzamidinehydrochloride was added. Cryo-protection wasachieved by soaking crystals in thecrystallisation mother liquid supplementedwith 15% MPD and data collected on ID14-EH2 of the ESRF (Table 1). Diffraction imageswere integrated with MOSFLM, intensitiesscaled and merged using SCALA. These werethen converted to structure factors andanomalous differences derived using theprogram TRUNCATE.

Data collected from crystal form I didnot allow de novo structure solution. Toachieve this we produced crystals of the SeMetderivative of the Leu-124-Met; Leu-125-Metdouble mutant to enhance the anomalousdispersion effects due to seleniumincorporation. Prior to crystallisation, thisprotein was incubated with 5 mM of substrate(D-tagatose-6-phosphate) and 5 mM AMP-PNP. Diffraction quality orthorhombic crystals(crystal form II) grew in the optimized

crystallization condition 0.2 M magnesiumacetate, 0.1 M sodium cacodylate pH 6.5, 10%PEG 8000, 15% PEG 550, 10 mM ATP. Cryo-protection was achieved by soaking crystals in0.2 M magnesium acetate tetrahydrate, 0.1 Msodium cacodylate pH 6.5, 10% PEG 8000supplemented with 10% glycerol and 10% PEG550. Diffraction data, measured on ESRFbeam-line ID14-EH2, were integrated andscaled using XDS (19) and intensities mergedusing SCALA. Intensities were then convertedto structure factors and anomalous differencesderived using the program TRUNCATE. SeeTable 1 for a summary of statistics.

Structure solution, model building andrefinement

Crystal form IIThe structure of crystal form II was

solved using the Single wavelength AnomalousDispersion technique. The programs XPREP(20) and SHELXD (21) were used to determinea substructure comprising 16 Se sites from atetramer in the asymmetric unit. The solventcontent and Matthews coefficient are 50% and2.5 Å3/Da respectively. SHELXE (22)determined the correct enantiomorph of thesubstructure and the program SHARP (23)used to calculate phase probabilitydistributions. This resulted in experimentalphases with an average figure of merit (FOM)of 0.245 to dmin = 2.1 Å resolution.

The experimental phases wereimproved using the program DM by densitymodification using a combination of non-crystallographic symmetry (NCS) averaging,histogram matching and solvent flattening.Initial NCS operators were obtained using theprogram PROFESS. The resulting phasescoupled with the observed structure factorswere then input into ARP/wARP (24) whichplaced 1144 glycine, valine, and serineresidues in the asymmetric unit. The GuiSIDEmodule of CCP4i (25) then docked the correctside chains to these residues. Visual inspectionof the electron density using O (26) identifiedmodel fragments belonging to each of the fourmonomers. Rounds of refinement usingREFMAC5 interspersed with map inspectionsin O allowed the model of all four monomersin the asymmetric unit to be constructed. Thecareful placement of solvent molecules andactive site ligands concluded the analysis withthe refinement process being monitored usingRwork and Rfree (27). It was clear early in the map

3

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

interpretation that there was significantdeviation from the expected non-crystallographic symmetry and so all subunitswere treated independently. The subunits arelabelled A-D.

Crystal form ISubunit A from crystal form II,

stripped of ligands and solvent moleculesprovided the search model for molecularreplacement calculations to solve the structureof crystal form I. Molecular replacement wascarried out using the program PHASER. Theasymmetric unit of form I also presents foursubunits, again labeled A-D, and each waspositioned then refined as a rigid body usingREFMAC5 with data between 20.0 and 2.0 Å.The resulting electron density map provided abasis for automatic model building(ARP/wARP) and a model consisting of 992glycine residues was built. Similar protocols tothose applied in the analysis of form IIconcluded the structure determination.

Summaries of the composition andgeometry of the final models for both crystalforms along with refinement statistics arepresented in Table 1. The stereochemistry ofthe final models was assessed usingPROCHECK. Ramachandran plots (not shown)indicate excellent main chain geometry.However, in all subunits Asn-233 has Φ and Ψangles, which place it in a disallowed region.This residue is well defined by the electrondensity and the strained chain conformation isstabilized by a hydrogen bond between themain chain amide and the carbonyl of Glu-217.

The programs DM, MOSFLM,PHASER, PROFESS, PROCHECK,REFMAC5, SCALA, TRUNCATE andGuiSIDE are part of the CCP4 softwarepackage (28,http://www.ccp4.ac.uk/main.html).

RESULTS AND DISCUSSIONCrystallographic analyses

In order to solve the structure of LacCwe produced the SeMet variant of a L124M,L125M double mutant. This sample (crystalform II) provided an anomalous signal largeenough to be exploited for structuredetermination. The choice of residues tomutate was guided by knowledge that mutationof Leu to Met has little effect on proteinstructure (29) coupled with secondary structurepredictions for LacC (data not shown) thatindicated the residues concerned are located in

α-helix. Our crystal structure analyses provedthis prediction to be correct.

Subsequently molecular replacementmethods allowed us to determine the structureof crystal form I. Both crystal forms have beenrefined to high resolution (Table 1). Theasymmetric unit of each form contains foursubunits organised as a pair of homodimers.

A new motif for the pfkB subfamilyAmong the sequences aligned in Figure

2, there is clearly a previously unreported thirdregion of amino acid conservation common tofructose phosphate kinases and LacC. Thisconserved region, Motif III, is common to thepfkB subfamily of carbohydrate kinases but notto members of the ribokinase superfamily ofproteins in general and has the sequence Ser-Gly-Ser-Leu-Pro-X-Gly. Motif III may thus beassigned as being the characteristic signature ofthe pfkB family carbohydrate kinases and, aswill be described, this motif is specificallyinvolved in the binding of the tagatosesubstrate.

The overall structure of LacCThe primary structure of S. aureus

LacC consists of 310 amino acids. The tertiarystructure of a LacC monomer is composed oftwo domains (Figure 3(A)). A large α/β coredomain comprises residues 1-7, 37-83, 110-310and presents a Rossmann-type fold in whichthe central β-sheet has been extended.Additional α-helices abutt this central β sheetalong its entire length. A smaller 'lid' domainprotrudes from the α/β domain. It isconstructed from residues 10-33 and 89-104,and comprises five β-strands, four of which areorganised in a mixed β-sheet with strand orderβ2-β4-β7-β8. The fifth strand in this domain(β3) is part of the turn linking β2 and β4. Fourpolypeptide segments comprising residues 7-9,34-37, 83-88 and 105-110 connect the twodomains. The lack of secondary structure forthese connecting elements is suggestive of apotential flexibility in the relative orientationof the two domains.

The LacC monomer structure is closelyrelated to other members of the ribokinasesuperfamily. The common structural feature inthis superfamily is the large α/β domain, whilstthe lid domain is less well conserved. Thusribokinase (17), adenosine kinase (30), andglucokinase (31) have lid domains but 4-amino-5-hydromethyl-2-methylpyrimidine

4

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

phosphate kinase (32) and 4-methyl-5-β-hydroxy-ethylthiazole kinase (33) lack thisfeature. When both core and lid domains arepresent their folds are conserved but theirrelative orientations may differ. Asuperposition of the LacC monomers in thecrystal forms described here (data not shown)confirms this conformational flexibility withsubunits being found in either open (monomersAI BI, CI, AII, BII, DII), semi-closed (DI) or fully-closed (CII) conformations (the subscriptdenotes crystal form).

In common with other members of theribokinase superfamily LacC forms a dimersimilar to that seen for E. coli ribokinase (17,Figure 3(B)). Two monomers associate viainteractions between their lid domains, whichcome together to form a β-barrel structureknown as a β-clasp. The two β-sheets of the liddomains pack face to face and are rotatedapproximately 90° with respect to each other.Each face of the β-barrel is made up of a five-stranded β-sheet; four strands from onemonomer, the fifth from the second. β-claspstructures display closed and splayed corners.In the LacC dimers, the distances between thediagonally opposite closed and splayed cornersare around 17 Å and 25 Å respectively.Monomer-monomer interactions within theinterior of the barrel are plentiful and are bothhydrophobic and polar in nature.

The active site of LacCCrystal form I was obtained by co-

crystallization with ATP. However, only twomolecules of ADP, bound to monomers BI andDI, were identified. Monomers AI and CI

represent the structure of the apo form of theenzyme. That ADP, and not ATP, is foundbound to monomers BI and DI should not besurprising. In the crystal structure of E. coliribokinase (17), ADP is found in the co-factorbinding site even though the protein wasincubated with AMP-PNP (often used becauseof its stability) before crystallization. In thatcase, the authors speculated that AMP-PNPhad hydrolyzed before binding to the protein.Given the much greater propensity of ATP tohydrolyze it is likely that the same situation hasoccurred here.

The ADP binding site is on the surfaceof the α/β domain constructed using residuesfrom β11, the C-terminus of β12 and the turnlinking it to β13, the long loop linking β14 toα8, the N-terminus of the latter and the C-terminus of α9. Aliphatic side chains form a

binding pocket in which the adenine base isheld in an anti–conformation stabilized only byhydrophobic interactions. The ribose O2*interacts with the main chain carbonyl andND2 of Asn-278. A well-ordered watermolecule mediates hydrogen bonds betweenthe ribose O3* and the amide and carbonylgroups of Gly-224 and Gly-227. The riboseO3* atom also accepts a hydrogen bond fromAsn-278 ND2. The α-phosphate group interactswith Ser-222 OG and, via a water molecule,with Lys-38 NZ. The O3A link between α- andβ- phosphate groups accepts a hydrogen bondfrom the main chain amide of Gly-253. The β-phosphate forms hydrogen bonds with thecarbonyl group of Pro-249 (we assume that aphosphate O is protonated), the amide group ofGly-253 and (via water molecules) the amidegroups of Gly-251 and Asp-254.

Crystal form II, from which the firstexperimental phases were actually determined,was obtained from protein incubated with 5mM of substrate (D-tagatose-6-phosphate) andthe cofactor analogue (AMP-PNP). Here,positive difference electron density (mFo-DFc,αcalc) was observed in the active sites of all fourmonomers of the asymmetric unit. In the activesites of subunits AII, BII and DII this wasassigned as AMP-PNP with a hydrated Mg2+

ion binding between the β- and γ-phosphate.In monomer CII two different moieties

are bound in the active site. One is clearlytagatose-6-phosphate (Figure 4(A), right). TheOMIT (mFo-DFc, αcalc) difference density(Figure 4(A), left) indicated that an initialmodel for the second moiety should be AMP-PNP with two hydrated cations bound to itsphosphate tail. However, such a model resultedin significant negative difference densitycentered on the γ-phosphate group of the AMP-PNP, an indication that this model was notcorrect. Clean difference density maps wereeventually obtained in the active site of subunitCII by modeling AMP-PNP and ADP, eachwith half occupancy, coupled with the presenceof two fully occupied hydrated cations. One ofthe two cations binds between the α- and β-phosphates of AMP-PNP and (when present)ADP. The second cation is bound between theβ- and γ-phosphates of AMP-PNP or to the β-phosphate group of ADP. Further discussion ofthe structure of crystal form II of LacCassumes the presence of a fully occupied AMP-PNP. Figure 4(B) shows where the ligandsbind to the LacC subunit and Figure 5 shows

5

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

the key interactions in the active site that willbe discussed.

Based on coordination geometry, thecation bridging the α- and β-phosphates ofAMP-PNP is assigned as Mg2+. The nature ofthe cation bridging the β- and γ-phosphates ofthe AMP-PNP moiety in the active site onmolecule C of crystal form II is less obvious.Given its square pyramidal coordination itcould be assigned as either Na+ or K+ with theion - oxygen atom distances observed (2.0 -2.3Å) favoring assignment as Na+ (34). However,Mg2+ can also exhibit square pyramidalcoordination with Mg2+ - oxygen distancessimilar to those observed here (35). Given theassignment as Mg2+ of a five-coordinate cationbridging the β- and γ-phosphates of the AMP-PNP moiety in active site B of this crystal formof LacC this cation was also assigned as Mg2+.The temperature factors of the cations (andtheir associated water molecules) suggest thatthey are fully occupied. Thus, when bothsubstrate and ATP are bound in the active siteof LacC, hydrated Mg2+ ions bridge both the α-and β-phosphate groups and the β- and γ-phosphate groups of the co-factor. When ADPis bound in the presence of substrate, ahydrated Mg2+ ion bridges the α- and β-phosphate groups while a second forms abridge between its terminal β-phosphate andthe co-factor binding site residues of LacC.

As described for ADP binding, thecofactor binding sites are located on thesurface of the core domain (Figure 4(B)) andinteractions between the enzyme and theadenine base are conserved. However,interactions between the sugar/phosphate tailof the AMP-PNP/ATP and the protein aredifferent from those seen for ADP binding toLacC (not shown).

When substrate is absent (monomersAII, BII, DII), the ribose O2* of AMP-PNPinteracts with the protein in the same way aswhen ADP is bound. However, the AMP-PNPribose O3* now accepts a hydrogen bonddirectly from Asn-278 ND2 and the previouslyobserved solvent-mediated hydrogen bondsbetween the ribose O3* and the protein arereplaced by a direct interaction of this groupwith the carbonyl oxygen of Gly-227. Thisdifference is due to a change in the position ofthe loop involving residues Ser-222 to Gly-227which, when AMP-PNP rather than ADP isbound, is shifted toward the cofactor by around1.6 Å. The α-phosphate accepts hydrogen

bonds from the amide of Gly-224 and thehydroxyl of Ser-222. When substrate is notpresent, the α-phosphate is also involved inwater-mediated hydrogen bonds with thecarbonyl group of Pro-249 and the linking N3Batom. The β-phosphate accepts hydrogen bondsfrom ND2 of Asn-185 and NZ of Lys-183. Theγ-phosphate is held in place tightly with O1Gaccepting a hydrogen bond from the Gly-251amide while O2G accepts two hydrogen bondsdonated by the amide groups of Gly-253 andAsp-254 and in addition interacts with a watermolecule, which coordinates the Mg2+ ion. ThisMg2+ is coordinated by β- and γ-phosphategroups of the nucleotide via two non-bridgingoxygen atoms. The coordination sphere of thecation is completed by three water molecules inthe active site of monomer D and four watermolecules in those of monomers A and B.These water molecules form hydrogen bondswith a number of amino acid side chainsincluding those of Lys-38 and Asp-254.

The majority of the hydrogen bondsand metal ion-cofactor interactions aremaintained when AMP-PNP/ATP is bound inthe presence of substrate (Figure 5). However,in this case a major difference is that a secondMg2+ ion bridges the α- and β-phosphategroups. Octahedral coordination of this cationis completed by water molecules, whichmediate interactions of the α- and β-phosphategroups with the side chain of Gln-99 and themain chain carbonyl of Pro-249. As well asinteracting with this Mg2+ ion, the β-phosphateO1B also accepts a hydrogen bond donated byAsn-185 ND2. Lys-183 NZ donates a hydrogenbond to the β-phosphate O2B, which alsointeracts with the second Mg2+ ion. This ionbridges the β-phosphate O2B and the γ-phosphate O2G and square pyramidalcoordination is completed by three watermolecules, the positions of which areconserved in all four active sites of crystalform II of LacC. The linking N3B interactswith the amide group of Gly-253. The γ-phosphate O1G accepts hydrogen bondsdonated by the amide group and, the likelyprotonated, OD2 of Asp-254, and participatesin a solvent-mediated interaction with thecation bridging the β- and γ-phosphate groups.The γ-phosphate O3G accepts a hydrogen bonddonated by the amide group of Gly-251.

In monomer CII the D-tagatose-6-phosphate substrate binds to residues from bothdomains of LacC and to residues in the linker

6

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

region (Figure 4(B), 5). The substrate O1group, the phosphorylation site, formshydrogen bonds with Asp-254 OD2 and Lys-38NZ with the latter also donating a hydrogenbond to the D-tagatose-6-phosphate O5hydroxyl and interacting with the substratephosphate via a water molecule (not shown).The substrate O2 interacts with Asn-41 ND2while its O3 and O4 groups form hydrogenbonds with Asp-12 OD2 and OD1 respectively.As well as the solvent-mediated interactionwith Lys-38, the substrate phosphate alsoforms direct hydrogen bonds with the sidechains of Arg-88 and (from the partner subunit)Arg-27, the side chain hydroxyl of Ser-136 andthe amide of Gly-135. The dimeric nature ofLacC is thus important to construct the activesite, in particular to position the basic Arg-27so that it can participate in substrate binding.

The catalytic mechanism of LacCAs described above, monomer CII

represents the structure of LacC with bothsubstrate and co-factor simultaneously bound.The active site comprises residues from the liddomain, the linking fragment and the α/βdomain. Amino acids in all three conservedmotifs (Figure 2) play a role in the binding ofsubstrate and/or co-factor. In particular, and asalso seen in the structure of ribokinase (17), theamino acids in Motif I form an α-helix ofwhich the N-terminus and the side-chain ofLys-38 point toward the substratephosphorylation site. From Motif II the mainchain amides of Gly-251 and Gly-253 formhydrogen bonds with the γ-phosphate of the co-factor and the side-chain of Asp-254 pointstowards the substrate phosphorylation site. Ser-136 from Motif III donates a hydrogen bond tothe substrate's phosphate group. Thisphosphate is present only in substratesprocessed by 1-phosphofructokinase, D-fructose-6-phosphate kinase and D-tagatose-6-phosphate kinase and this explains theconservation of this serine in these sequencesonly.

The catalytic centre (Figure 5) islocated at the interface between co-factor andsubstrate. The γ-phosphate of AMP-PNP is 2.9Ǻ distant from the substrate phosphorylationsite, which is sandwiched between the sidechains of Asp-254 and Lys-38. Thisconfiguration suggests a catalytic mechanismfor LacC in which the side chain of Asp-254abstracts the proton from the substrate C1hydroxyl group to generate a nucleophile,

which then attacks the γ-phosphate group ofthe co-factor to yield the phosphorylatedproduct (Figure 6). The optimal positioning ofthe side chain of Lys-38 likely stabilizes thenegative charge of the transition state.

This mechanism is similar to thatproposed for other phosphoryl transferases.However, unlike other enzymes that carry outthe reaction in an Mg2+–dependent manner (36-41), the crystal structures described hereindicate that the active site of LacC containstwo, rather than one, Mg2+. The cation thatbridges the β- and γ- phosphates of the co-factor serves to correctly orient both the γ-phosphate and the side chain of the catalyticAsp-254. This cation may also serve toneutralize any negative charge on the transitionstate and may favor an associative reactionmechanism by polarizing the co-factor γ-phosphate thus making it more susceptible tonucleophilic attack (42). The role of the Mg2+

ion that bridges the α- and β- phosphates isless clear although it may enhance theeffectiveness of ADP as a leaving group (43).

Conformational changes in LacCLacC monomers adopt distinct

conformations in which the relativeorientations of the lid and core domains vary.Conformational changes upon substratebinding are a well known strategy fornucleotide-dependant transferases to positionappropriate amino acids around the reactingpartners (44). In the crystal structuresdescribed here monomer CII is found in a fullyclosed conformation. But which domainmoves when this conformation is formed?

The conservation of the distances(around 25 Å) between the splayed corners ofthe β-clasps formed in the AB and CD dimersin crystal form II, suggest that theconformational changes seen in LacC involve amovement of the α/β domain toward the liddomain resulting in a closure of the active site.The fully closed conformation seen for the CII

monomer of LacC is essential for transfer ofthe phosphoryl group to the substrate as only inthis configuration are the catalytic partnersclose to each other and the catalytic Lys-38 andAsp-254 in the correct position to facilitatephosphoryl transfer. Moreover, in LacC, fulldomain closure is also required to provide afully formed substrate-binding pocket (Figure7).

7

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

The phenomenon of domain closurehas been observed in many multi-domainenzymes. In one such enzyme,phosphoglycerate kinase, complete domainclosure may only be achieved with binding ofboth substrate and AMP-cofactor (45). Such asituation also seems to be the case for LacC.However, domain closure is not common to allenzymes that possess a lid domain. Thus, thecrystal structures of 2-keto-3-deoxygluconatekinase from Thermus thermophilus in complexwith ATP or ATP and substrate revealed thatonly local structural – and not large scale -conformational changes – are required toenable catalysis (46).

The conformational changes observedin monomer CII of LacC are also critical inallowing the recruitment of the Mg2+ ion thatbridges the α and β- phosphates of the cofactorwhen substrate is bound. Only when domainclosure is complete can the side chain of Gln-99 help stabilize binding of a second Mg2+ ionby the ATP co-factor (Figure 5, 7(B)). That thefully closed conformation of LacC is alsocrucial to allow proper substrate bindingsuggests that the presence of this second Mg2+

ion may be important to phosphoryl transfer ascatalyzed by LacC.

CONCLUSIONSThe structure of LacC from S. aureus

has been determined in two different crystalforms. The fold is characteristic of the

ribokinase superfamily, and consists of twodomains, a core α/β domain and a lid domain,comprised of β-strands, which is involved indimer formation. These two domains are linkedby unstructured polypeptides, which conferflexibility in the relative orientation of the twodomains. As the dimer interface forms a rigidβ-clasp structure, the conformational changesobserved involve a movement of the coredomain towards the lid domain. Domainclosure has to be complete to bring substrateand cofactor close enough to react, to provide afully formed substrate-binding pocket and toallow the recruitment of a second Mg2+ ion thatmay be important in catalysis.

Our study provides understanding ofhow LacC recognizes and binds its substratesand allows us to propose a mechanism for thecatalysis of phosphoryl transfer. The structuralinformation may, in addition, provide the basisfor the design of a specific LacC inhibitor,which might have implications for human andanimal health.

REFERENCES1. Frey, P. A. (1996). Faseb J. 10, 461-470.2. Bissett, D. L., and Anderson, R. L. (1974). J. Bacteriol. 117, 318-320.3. van Rooijen, R. J., van Schalkwijk, S., and de Vos, W. M. (1991). J. Biol. Chem. 266 , 7176-7181.4. Schleifer, K., H., Hartinger, A., and Gotz, F. (1978). FEMS Microbiology Letts. 3, 9-11.5. Rosey, E. L., Oskouian, B., and Stewart, G. C. (1991). J. Bacteriol. 173, 5992-5998.6. Morse, M. L., Hill, K. L., Egan, J. B., and Hengstenberg, W. (1968). J. Bacteriol. 95, 2270-2274.7. Oskouian, B. and Stewart, G. C. (1990). J. Bacteriol. 172, 3804-3812.8. Jacob, F., and Monod, J. (1961). J. Mol. Biol. 3, 318-56.9. Bissett, D. L., and Anderson, R. L. (1980). J. Biol. Chem. 255, 8745-8749.10. Kotlarz, D., and Buc, H. (1982). Methods Enzymol. 90, 60-70.11. Hall, D.R., Bond, C.S., Leonard, G.A., Watt, I, Berry, A., and Hunter, W.N. (2002) J. Biol. Chem.277, 22018-22024.12. Bettenbrock, K., Siebers, U., Ehrenreich, P., and Alpert, C. A. (1999). J. Bacteriol. 181, 225-230.13. Stagno, S., Reynolds, D. W., Pass, R. F., and Alford, C. A. (1980). New Eng. J. Med. 302, 1073-1076.14. Grove, T. M., and Jones, G. M. (1992). J. Dairy Sci. 75, 423-434.15. Bork, P., Sander, C., and Valencia, A. (1993). Protein Sci. 2, 31-40.16. Spychala, J., Datta, N. S., Takabayashi, K., Datta, M., Fox, I. H., Gribbin, T., and Mitchell, B. S.(1996). Proc. Natl. Acad. Sci. USA 93, 1232-1237.

8

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

17. Sigrell, J. A., Cameron, A. D., Jones, T. A., and Mowbray, S. L. (1998). Structure 6, 183-193.18. Miallau, L., Alphey, M.S., Kemp, L.E., Leonard, G.A., McSweeney, S.M., Hecht, S., Bacher, S.,Eisenreich, W., Rohdich, F., and Hunter, W.N. (2003) Proc. Natl. Acad. Sci. USA 100, 9173-9178.19. Kabsch, W. (1993). J. Appl. Cryst. 26, 795-800.20. Bruker (2000). SMART, SAINT, SADABS and XPREP Software Reference Manual. Bruker AXSInc., Madison, Wisconsin, USA.21. Schneider, T.R., and Sheldrick, G.M. (2002). Acta Crystallogr. D58, 1772-1779. 22. Sheldrick, G.M. (2002). Z. Kristallogr. 217, 644-650.23. Bricogne G, Vonrhein C, Flensburg C.M.S., and Paciorek, W. (2003). Acta Crystallogr. D 59,2023-2030.24. Perrakis, A., Morris, R. and Lamzin, V., S. (1999). Nat Struct Biol. 6, 458-463.25. Potterton, E., Briggs, P., Turkenburg, M., and Dodson, E.J. (2003). Acta Crystallogr. D59, 1131-1137.26. Jones, T., A., Zou, J., Y., Cowan, S., W., and Kjeldgaard (1991). Acta Crystallogr. A47, 110-9.27. Kleywegt, G. J., and Brünger, A. T. (1996). Structure 4, 897-904.28. The Collaborative Computational Project and Number 4 (1994). Acta. Crystallogr. D50, 760-763.29. Gassner, N.C. and Matthews, B.W. (1999). Acta Crystallogr. D57, 401-411.30. Cook, W. J., DeLucas, L. J., and Chattopadhyay, D. (2000). Protein Sci. 9, 704-712.31. Ito, S., Fushinibu, S., Yoshioka, I., Koga, S., Matsuzawa, H., and Wakagi, T. (2001). Structure 9,205-214.32. Cheng, G., Bennett, E. M., Begley, T. P., and Ealick, S. E. (2002). Structure 10, 225-235.33. Campobasso, N., Mathews, II, Begley, T. P., and Ealick, S. E. (2000). Biochemistry 39, 7868-7877.34. Harding, M. M. (2002). Acta Crystallogr. D58, 872-874.35. Harding, M. M. (2001). Acta Crystallogr. D57, 401-411.36. Lui, J-Y., Timm, D.E., and Hurley, T.D. (2006) J. Mol. Biol. 281, 6601-6607.37. Fioravanti, E., Haouz, A., Ursby, T., Munier-Lehmann, H., Delarue, M., and Bourgeois, D. (2003).J. Mol. Biol. 327, 1077-1092.38. Sabini, E., Ort, S., Monnerjahn, C., Conrad, M., and Lavie, A. (2003) Nature Struct. Biol. 10, 513-519.39. Ostermann, N., Schlichting, I., Brundiers, R., Konrad, M., Reinstein, J., Veit, T., Goody, R.S., andLavie, A. (2000). Structure 8, 629-642.40. Schlicting, I., and Reinstein, J. (1996) Biochemistry 39, 9290-9296.41. Abele, U., And Schulz, G.E. (1995) Protein Science 4, 1262-1271.42. Pappu, K.M., Gregory, J.D., and Serpersu, E.H. (1994). Arch. Biochem. Biophys. 311, 503-508.43. Knowles, J. R. (1980). Annu. Rev. Biochem. 49, 877-919.44. Matte, A., Tari, L. W., and Delbaere, L. T. (1998). Structure 6, 413-419.45. Auerbach, G., Huber, R., Grattinger, M., Zaiss, K., Schurig, H., Jaenicke, R. and Jacob, U. (1997). Structure 5, 1475-1483.46. Ohshima, N., Inagaki, E., Yasuike, K., Takio, K., and Tahirov, T. H. (2004). J. Mol. Biol. 340,477-89.47. Notredame, C., Higgins, D.G. Heringa J. (2000). J.Mol. Biol. 302, 205-17

FOOTNOTES

*W.N.H., thanks the Wellcome Trust and the Biotechnology and Biological Sciences ResearchCouncil (Structural Proteomics of Rational Targets) for support.

§The atomic coordinates and structure factors (codes: 2JGV and 2JG1) have been deposited in the ProteinData Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ(http://www.rcsb.org/).

FIGURE LEGENDS

9

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

Figure 1. The reaction catalysed by LacC.

Figure 2. Amino acid sequence comparisons. A multiple sequence alignment of 1-phosphofructokinase from E. coli (K1PF), D-fructose-6-phosphate kinase from E. coli (K6PF2), LacCfrom S. aureus (LACC), ribokinase from L. lactis (RBSK) and human adenosine kinase (ADK).Residues blocked in red are identical in the sequences aligned while yellow blocks indicate regions ofsequence homology. Motifs I and II previously identified for the ribokinase superfamily are indicatedas is Motif III common to only the pfkB family of carbohydrate kinases (see main text). This sequencealignment was made with T-Coffee (47).

Figure 3. Tertiary and quaternary structure of LacC. (A) Ribbon diagram of a subunit. Helices arecolored cyan and labeled, β-strands purple and numbered, loops are light brown and the segmentslinking the lid domain to the α/β domain are red. (B) The dimeric arrangement that forms the β-clasp.

Figure 4. Electron density maps for ligands and location of the active site. (A) The mFo-DFc, αc

OMIT density contoured at 2.5σ in the active site of monomer CII. That shown on the left-hand siderepresents AMP-PNP and ADP each with half occupancy and two fully occupied hydrated Mg2+ ions(see text). Mg2+ ions are shown as spheres colored orange and green, water molecules are light bluespheres. C, N, O, P positions are black, blue, red and purple respectively. On the right-hand sidetagatose-6-phosphate and associated density. The atomic positions C, O, P are orange, red and purplerespectively. (B) The CII subunit is shown as in Figure 3 (A) with the addition of a semitransparent vander Waals surface (grey). The ligands are depicted as sticks, AMP-PNP is colored black and D-tagatose-6-phosphate orange.

Figure 5. The active site of LacC. A stereo-view of the active site of subunit CII. Protein C atoms aregrey except for Arg-27, which are green. This residue is contributed from the partner subunit. AMP-PNP and D-tagatose-6-phosphate C atoms are colored green and black respectively. N atoms arecolored blue, O red and P yellow. The two Mg2+ ions and seven water molecules are shown asspheres, colored orange, green and blue. The Se atom of SeMet282 is purple. Potential hydrogenbonds are depicted as black dashed lines. Metal ion coordination is depicted with orange and greendashed lines. For the purpose of clarity, only selected water molecules and hydrogen bonds have beenshown.

Figure 6. The proposed mechanism for the transfer of the ATP γ-phosphate to D-tagatose-6-phosphate O1 by LacC.

Figure 7. Conformational changes in LacC. (A) A stereo-view of a superimposition (based on Cα

atoms of the lid domains) of closed (red lines) and open (blue lines) forms of LacC from crystal formII. The AMP-PNP in the cofactor-binding site of the open form is shown as a mauve surface, thesubstrate and cofactor in the active site of the closed form as black and green surfaces respectively.(B) Stereo-view showing the that only in the closed conformation are the side chains of amino acidsimplicated in the binding of the substrate close enough to interact with it and is the side chain of Gln-99 in a position where it can help stabilize binding of a second Mg2+ ion by the cofactor. In this figurethe C atoms of Gln-99 are shown in cyan, the C atoms of the substrate in black, and those of the closedand open form positions of AMP-PMP in green and mauve respectively. Residues that can onlyinteract with the substrate when LacC is in the closed form are shown as red (closed form) or blue(open form) sticks.

10

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

Crystal form I Crystal form IIData collectionBeamline ID14-EH2 (ESRF)Wavelength (Å) 0.933Space group P21 P212121

Unit cell dimensions a = 77.6 Åb = 96.4 Åc = 94.8 Åβ = 94.3º

a = 95.2 Åb = 97.1 Åc = 154.5 Å

Resolution range (Å) 32.8-2.0 50.0-2.0Rsym (%) 7.4 (48.1) 8.6 (62.4)Completeness (%) 98.5 (98.5) 99.8 (99.8)Multiplicity 4.2 (4.3) 10.9 (9.6)<I>/<σ(I)> 12.9 (3.0) 7.6 (1.2)Anomalous completeness (%) 97.3 (96.3) 99.6 (97.7)Refined model compositionMonomers / asymmetric unitProtein residuesWater moleculesAMP-PNPADPTagatose-6-phosphateMg2+ ions

412517050200

412648753.50.515

Average isotropic thermal parameters (Å2)Wilson B-valueOverallMain chain / Side chain

27.032.0

26.8 / 29.9 (A)27.4 / 30.8 (B)30.4 / 33.1 (C)36.6 / 39.3 (D)

31.031.1

30.0 / 31.4 (A)31.4 / 32.6 (B)31.4 / 33.1 (C)26.8 / 28.04 (D)

Model quality indicatorsRwork / Rfree (%)Rmsd for bond lengths (Å)/bond angles (°)Estimated coordinate errors (Å)

19.8 / 24.70.009 / 1.129

0.17

19.0/23.40.009 / 1.23

0.17Ramachandran analysis% Favoured regions% Additionally allowed regions% Generously allowed regions% Disallowed regions

93.06.60.00.4

93.56.20.00.4

Table 1. Crystallographic statistics. For crystal form I the model includes five residues fromthe purification tag for monomer A, four residues from the tag for monomer B and 1 residuefor monomers C and D. For crystal form II the model includes eight residues from thepurification tag for monomers A and B, five residues from the tag for monomer C and threeresidues for monomer D. Coordinate errors are estimated from Rfree. For simplicity details ofthe Ramachandran analysis are given only for molecule A in each crystal form. The details forall other molecules are extremely similar.

11

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

Figure 1

P

O

O

OO

O

OHHO

tagatose-6-phosphate + Mg2+ + ATP

HO

OH

P

O

O

OO

O

OHHO

tagatose-1,6-bisphosphate + Mg2+ + ADP

HO

P

O

O

OO

61

LacCEC 2.7.1.144

12

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

Figure 2

13

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

Figure 3

(a)

(b)

14

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

Figure 4

(a)

(b)

15

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

Figure 5

16

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

Figure 6

P

O

O

OOO

OHHO

tagatose-6-phosphate

HO

OH

61

O

O

Asp-254

NH3

Mg2+

Mg2+

Lys-38

ATPβ γ

N

N N

N

NH2

O

OH OH

O P O

O

O

P O

O

O

P O

O

O

α

17

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

Figure 7

(a)

(b)

18

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from

Linda Miallau, William N. Hunter, Sean M. McSweeney and Gordon A. Leonarddomain motions in specificity and mechanism

Structures of Staphylococcus aureus D-tagatose-6-phosphate kinase implicate

published online April 25, 2007J. Biol. Chem. 

  10.1074/jbc.M701480200Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

by guest on June 14, 2020http://w

ww

.jbc.org/D

ownloaded from