A

ecct(cmm©

K

1

toitbaalhi

qt

B

0d

Available online at www.sciencedirect.com

Molecular & Biochemical Parasitology 158 (2008) 131–138

Crystal structures of Toxoplasma gondii pterin-4a-carbinolaminedehydratase and comparisons with mammalian

and parasite orthologues�

Scott Cameron, Stewart A. Fyffe, Simon Goldie, William N. Hunter ∗Division of Biological Chemistry and Drug Discovery, College of Life Sciences, University of Dundee, Dundee DD1 5EH, UK

Received 20 August 2007; received in revised form 4 December 2007; accepted 6 December 2007Available online 15 December 2007

bstract

The enzyme pterin-4a-carbinolamine dehydratase (PCD) is important for the recycling of pterins within eukaryotic cells. A recombinantxpression system for PCD from the apicomplexan parasite Toxoplasma gondii has been prepared, the protein purified and crystallised. Singlerystal X-ray diffraction methods have produced a high-resolution structure (1.6 A) of the apo-enzyme and a low-resolution structure (3.1 A) of aomplex with a substrate-like ligand dihydrobiopterin (BH2). Analysis of the hydrogen bonding interactions that contribute to binding BH2 suggesthat the ligand is present in an enol tautomeric state, which makes it more similar to the physiological substrate. The enzyme can process (R)- and

S)-forms of pterin-4a-carbinolamine and the ligand complex suggests that His61 and His79 are placed to act independently as general bases foratalysis of the individual enantiomers. Comparisons with orthologues from other protozoan parasites (Plasmodium falciparum and Leishmaniaajor) and with rat PCD, for which the structure is known, indicate a high degree of sequence and structure conservation of this enzyme. Theolecular determinants of ligand recognition and PCD reactivity are therefore highly conserved across species.2007 Elsevier B.V. All rights reserved.

n; Tau

viat

ppp[e

eywords: Apicomplexa; Crystal structure; Dehydratase; Kinetoplastida; Pteri

. Introduction

Tetrahydrobiopterin (BH4) is an important cofactor in essen-ial metabolic pathways of both eukaryotic and prokaryoticrganisms. The chemical properties of this cofactor are exploitedn the hydroxylase reactions that convert phenylalanine orryptophan to tyrosine, by nitric oxide synthase, and in theiosynthesis of neurotransmitters such as serotonin, dopaminend noradrenaline [1]. Little is known about the pterin contentnd metabolism of parasitic protozoa although putative ortho-

ogues of several enzymes involved in pterin metabolism inigher eukaryotes have been identified [1]. Research in this area,n large part driven by access to genomic data, is likely to pro-

Abbreviations: BH4, tetrahydrobiopterin; BH2, dihydrobiopterin; q-BH2,uinoid-dihydrobiopterin; PCD, pterin-4a-carbinolamine dehydratase; Rn, Rat-us norvegicus; Tg, Toxoplasma gondii.� Coordinates and structure factor data are deposited with the Protein Dataank under accession codes 2V6S, 2V6T and 2V6U.∗ Corresponding author. Tel.: +44 1382 385745; fax: +44 1382 385764.

E-mail address: w.n.hunter@dundee.ac.uk (W.N. Hunter).

tdcdp4ArbgE

166-6851/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.molbiopara.2007.12.002

tomer; Toxoplasma

ide a rich vein of new biological data in particular on organismsn the phylum Apicomplexa, including Plasmodium falciparumnd Toxoplasma gondii, the causative agents of malaria andoxoplasmosis, respectively.

In higher eukaryotes BH4 is synthesised de novo, but manyarasitic organisms are pterin auxotrophs. These include thearasitic protozoa of the class Kinetoplastida, which acquireterins through a specific transporter (e.g. BT1 in Leishmania)2]. Recycling is important in these organisms for the regen-ration of BH4 from the pterin-4a-carbinolamine formed whenhe molecule acts as a cofactor [3–5]. Pterin-4a-carbinolamineehydratase (PCD, EC 4.2.1.96) is key to this process andatalyses the conversion of pterin-4a-carbinolamine to quinoid-ihydrobiopterin (q-BH2, Fig. 1a). This PCD catalysed steprevents the spontaneous rearrangement of the unstable pterin-a-carbinolamine to less useful 7-substituted pterins [4].n NADH-dependent quinoid-dihydrobiopterin reductase then

educes q-BH2 to BH4. The function of the protein encodedy the pcd gene (GenBank accession number: DQ223777) in T.ondii has been verified by a complementation assay [5] using anscherichia coli tyrosine auxotroph lacking PCD, but carrying

132 S. Cameron et al. / Molecular & Biochemical Parasitology 158 (2008) 131–138

F ed BH( r of B

alco

iismaeasfnmlofp

t

gewRtmtofni

2

2

s

ig. 1. Pterin salvage and ligand structures. (a) Pterin recycling from metabolisb) Chemical structure and atomic numbering of the major and a minor tautome

quinoid-dihydrobiopterin reductase (DHPR). When a pheny-alanine hydroxylase is co-introduced, pterin recycling (Fig. 1a)an occur, and so tyrosine production is enabled in the presencef the introduced PCD.

Rat liver PCD (Rattus norvegicus, RnPCD), which is identicaln sequence to human PCD, has been biochemically character-zed [6,7] and crystal structures determined [8–11]. A secondtructure is derived from the Gram-negative eubacterium Ther-us thermophilus (Protein Data Bank (PDB) codes 1USO

nd 1USM) however no publications are associated with thesentries. Intriguingly, mammalian PCD has a second functionssigned to it. It is thought to activate transcription by forming atable complex with Hepatocyte Nuclear Factor 1 (HNF1) thatacilitates DNA binding [11]. This gives PCD the alternativeame of DCoH (Dimerisation Cofactor of HNF1). The preciseechanism by which transcription is activated is not yet estab-

ished. The absence of a detectable HNF1 homologue in T. gondiir indeed any other protozoan suggests that either such a second

unction is unlikely in these organisms or that a different partnerrotein is involved.

Here we report the preparation of a recombinant pro-ein expression system, purification and crystallisation of T.

RfimM

4, which generates pterin-4a-carbinolamine through q-BH2 and back to BH4.H2.

ondii PCD (TgPCD), X-ray crystal structures of the apo-nzyme and of the complex with 7,8-dihydrobiopterin (BH2),hich is a product analogue. Structural comparisons withnPCD and sequence comparisons with orthologues from

wo other protozoan parasites, P. falciparum and Leishmaniaajor are presented. These comparisons allow us to assess

he potential of this enzyme as a target for the developmentf inhibitors to assist the development of reagents useful tourther dissect pterin metabolism and that might underpinovel therapeutic approaches for diseases caused by protozoannfections.

. Materials and methods

.1. Cloning, expression and purification

The TimeSaver cDNA synthesis kit (Amersham Bio-ciences) was used to generate a cDNA library from total

NA extracted from T. gondii (RH strain) parasites. Briefly,rst strand synthesis, to make a full-length cDNA copy fromRNA, was performed using an oligo(dT)12–18 primer andoloney murine leukaemia virus reverse transcriptase. This

hemic

wcatw5sTs(c(t�goc(ocT7TsmmcoscTpliw1

2

doTdupwctistsac2o

2r

fRdtdSQdi

u(bwdowra[wm

Twatwwa[a

3

3s

mgmssnace

S. Cameron et al. / Molecular & Bioc

as followed by a second strand synthesis reaction using E.oli DNA polymerase I, which was in turn ligated to EcoRIdaptors and the total library cloned into pUC18 for amplifica-ion. The gene (pcd) was amplified from this library by PCRith the primers 5′-catatgGCACCACTTGCCCGCCT and′-ggatccCTACTTTTCGAAGTTCTTCGCCGC (lower caseequences correspond to the restriction sites used for cloning).he final construct was a pET15b (Novagen) expressionystem that had been modified to encode a Tobacco Etch VirusTEV) protease cleavage site in place of the thrombin proteaseleavage site. Protein expression was done in BL21 (DE3) cellsStratagene) by growth at 37 ◦C to an OD600 = 0.6–0.8 andemperature reduction to 22 ◦C after which 0.5 mM isopropyl-d-1-thiogalactopyranoside (Melford) was added. Cells wererown for a further 18 h, harvested by centrifugation and brokenpen with a French press. Purification was achieved by affinityhromatography over a HisTrap nickel chelating columnGE Healthcare), followed by TEV cleavage/dialysis at 4 ◦Cvernight and then further passage over the a HisTrap nickelolumn to remove the tag, contaminants and the His-taggedEV protease. A final purification step using a Superdex5 size-exclusion column (GE Healthcare) was performed.his column had been calibrated with proteins of differentizes (Amersham gel filtration calibration kits—high and lowolecular weight) to generate a standard curve from whicholecular mass can be estimated (proteins used were ferritin,

onalbumin, carbonic anhydrase, ribonuclease A, aldolase,valbumin and aprotinin). Protein concentration was determinedpectrophotometrically at A280 nm using a theoretical extinctionoefficient 22,460 M−1 cm−1, in a final buffer of 20 mMris–HCl, pH 7.6, 150 mM NaCl. The high degree of proteinurity was confirmed by sodium dodecyl sulphate polyacry-amide gel electrophoresis and matrix-assisted laser desorption/onization-time-of-flight mass-spectrometry. The sampleas concentrated using centrifugal force (Vivascience) to4 mg ml−1.

.2. Crystallisation

Crystallisation screening was performed using the sittingrop vapour diffusion method in 96-well format with a rangef commercial screens. Crystals could only be obtained fromgPCD pre-incubated for 30 min on ice with 2 mM 7,8-ihydrobiopterin (BH2, Schircks Laboratories). Two conditionsltimately gave useful crystals. These involved mixing 1 �l ofrotein:BH2 solution with 1 �l of reservoir. When the reservoiras (a) 1.4 M ammonium sulphate and 1 M lithium sulphate,

rystals appeared after 24 h and (b) 0.4 M potassium–sodiumartrate, after 4 months. Crystals were cryo-protected by soak-ng for 1 min in 2 M sodium tartrate, cooled to 100 K in atream of nitrogen gas for diffraction experiments. The crys-als are tetragonal blocks and the diffraction data indicated Laueymmetry 4mmm with unit cell dimensions of approximately

= b = 120 A, c = 50 A, α = β = γ = 90◦. The asymmetric unitontains two subunits of total molecular mass approximately4.2 kDa, 60% solvent content and a Matthews coefficient [12]f 3.2 A3 Da−1.

ttil

al Parasitology 158 (2008) 131–138 133

.3. Data collection, processing, structure solution,efinement and model analysis

Data were first collected to medium resolution (2.2 A)rom crystals grown under condition (a) in-house using aigaku rotating anode-Raxis IV++ image plate system. Theata were processed and scaled using XDS [13] and usedo first solve and refine the structure. Subsequently a 1.6 Aataset was collected on beam-line ID23-1 at the Europeanynchrotron Radiation Facility (λ = 1.0064 A) with a Quantum315r charge coupled device detector. A third lower-resolutionataset from crystallisation condition (b) was later measuredn-house.

Molecular replacement was performed in PHASER [14]sing a poly-alanine monomer model derived from RnPCDPDB code 1DCP). Two solutions were then subjected to rigid-ody refinement. Molecular packing suggested the space groupas P43212 and this was confirmed by analysis of electron andifference density maps and successful refinement. Inspectionf maps, model alterations and incorporation of water moleculesere performed using COOT [15], interspersed with restrained

efinement using TLS (translation, libration, screw analysis)nd maximum-likelihood restrained refinement in REFMAC516,17]. Tight non-crystallographic symmetry (NCS) restraintsere applied at first then released towards the end of the refine-ent.The other datasets were processed in a similar way and the

gPCD dimer model was used to initiate refinement. These dataere processed to 1.6 A for the high-resolution synchrotron data,

nd to 3.1 A for the in-house dataset. One difference betweenhe refinements was that in the low-resolution BH2 complex,eak NCS restraints were retained. The quality of the modelsas assessed with PROCHECK [18]. Crystallographic statistics

re presented in Table 1. Figures were prepared with PyMOL19] and ALINE (C.S. Bond, personal communication). Surfacerea calculations used PISA [20].

. Results and discussion

.1. Protein purification, quaternary structure andequence alignments

Recombinant TgPCD was purified with a yield of approxi-ately 30 mg l−1 of bacterial culture. High-resolution analytical

el filtration indicated the presence of a single species ofolecular mass 46 kDa, which corresponds to a tetramer in

olution (data not shown). Fig. 2 shows an alignment of PCDequences derived from T. gondi, P. falciparum, L. major and R.orvegicus with secondary structure assigned from TgPCD. Themino acid sequence identity between TgPCD and R. norvegi-us PCD is 43%, 27% and 33% with the P. falciparum, L. majornzymes, respectively. The alignment indicates conservation of

he active site histidines (His61, His62 and His79) and aspar-ates/glutamates (Asp/Glu57, Asp/Glu80 and Asp88) implicatedn the function of RnPCD [5,7]. These will be discussedater.

134 S. Cameron et al. / Molecular & Biochemical Parasitology 158 (2008) 131–138

Table 1Data collection, refinement and model geometry statistics

Medium resolution High resolution TgPCD:BH2 complex

Unit cell length (A): a = b, c 119.75, 49.33 120.13, 49.77 119.51, 49.93Resolution (A) 19.7–2.2 53.7–1.6 43.0–3.1Unique reflections 17,850 47,707 6969Redundancy/completeness 6.4/98.6 9.2/100 9.2/100〈I/σ(I)〉 11.1 (2.6) 27.8 (5.8) 11.5 (1.6)Rsym (%) 4.9 (29.1) 7.0 (47.7) 5.4 (46.8)R/Rfree (%) 17.8/21.8 17.3/19.4 19.0/27.1Protein residues/atoms 200/1615 202/1701 119/1620Residues subunit A/B 1-102/5-102 1-103/5-103 5-103/4-103Water molecules 219 287 11Wilson Ba (A2) 20.9 19.7 119.0

Average Ba (A2)Protein 19.6 14.5 78.2Waters/Na+ 30.9/NAb 33.1/21.5 66.8/NAb

BH2 123.0

RMSD bond lengths/angles 0.008/1.136 0.007/0.948 0.022/2.254Cruikshank’s DPIc (A) 0.15 0.07 0.41

Ramachandran plot (%)Most favoured region 94.6 95.2 84.3Add. allowed regions 4.3 3.7 12.4Gen. allowed regions 0.0 0.0 2.2Unfavoured regions 1.1 Tyr69-A/B 1.1 Tyr69-A/B 1.1 Tyr69-A/B

3

gadpfat

(trtd

FfSa

a Isotropic thermal parameter.b Not applicable.c Diffraction-component precision index [25].

.2. Overall structure

The crystals of TgPCD are isomorphous and display spaceroup P43212 with two monomers, labelled A and B, in thesymmetric unit. Three structures were determined and in theiscussion that follows we detail the 1.6 A resolution model and

resent information on BH2 binding in the active site derivedrom the low-resolution complex. The geometry of the models,nd for the sake of completeness we include the 2.2 A resolu-ion structure, which was the first one determined, is acceptable

frca

ig. 2. Alignment of four pterin-4a-carbinolamine dehydratase sequences. The specirom Rattus norvegicus (Rn, CAA06587); Toxoplasma gondii (Tg, ABB53416); Leistars indicate six conserved histidine, aspartate or glutamate active site residues discusnd arrows (�-strands).

Table 1). In all the structures a single residue of each polypep-ide chain, Tyr69, displays a �/� combination in the disallowedegion of the Ramachandran plot. This residue is placed in a tighturn linking strands �2–�3 and is well defined by the electronensity.

Several datasets from samples obtained under slightly dif-

erent crystallisation conditions, extending to about 2.3 Aesolution, were measured and taken through the refinement pro-ess but since they did not reveal a ligand in the active site theyre not described.

es is given then the UniProt/TrEMBL code. The four sequences compared arehmania major (Lm, CAJ02422) and Plasmodium falciparum (Pf, ABB53415).sed in text. Elements of secondary structure are depicted as cylinders (�-helices)

hemical Parasitology 158 (2008) 131–138 135

CtbtstaCrlra

soRc(cpesi��muraAoisbosw3atc

3

trtttftTwaa

Fig. 3. Ribbon diagrams of TgPCD. (a) The asymmetric unit dimer. Helices ofsubunit A are coloured red, �-strands pink, and helices of subunit B are bluewith �-strand cyan. The N-terminus and elements of secondary structure of eachpolypeptide are labelled. Black spheres represent BH2 and mark the position ofthe active sites. (b) The TgPCD tetramer. Subunits generated by the applicationoa

fcsg(sNcpitbt

S. Cameron et al. / Molecular & Bioc

Chain A was modelled with all residues present apart from the-terminal Lys104, for which there was no interpretable elec-

ron density. Residues 5–103 from chain B could be modelled,ut Met1, Ala2, Pro3, Leu4 and again Lys104 are not present inhe model due to the lack of interpretable electron density. Theubunit structure is well conserved between the two moleculeshat constitute the asymmetric unit. The root-mean-square devi-tion (RMSD) resulting from least-squares superposition of 99� positions is 0.43 A. Not surprisingly, given the use of NCS

estraints throughout the refinement, the RMSD obtained foreast-squares superposition of subunits A and B of the low-esolution TgPCD:BH2 complex is lower, 0.21 A for 99 C�toms.

TgPCD displays, as expected given the level of sequence con-ervation, a similar fold to RnPCD [8–11]. A superimpositionf the 1.6 A resolution TgPCD structure with RnPCD reveals anMSD of 1.0 A across 94 C� atoms, a high degree of structuralonservation in agreement with the level of sequence identity43%). The subunit displays a characteristic �–� arrangementonsisting of three �-helices packed against a four-stranded anti-arallel �-sheet (Fig. 3a). The order of the secondary structurelements is �1–�1–�2–�2–�3–�4–�3. Dimerisation occurs byelf-association of �3-strands in anti-parallel fashion, allow-ng the �-helices to form a saddle-like structure across the-sheet (Fig. 3b). Between the �2-strands and the adjacent2 helices there is a pronounced hydrophobic core, formedainly by aromatic side chains, and interactions between sub-

nits here also contribute to dimer stability (not shown). The endesult is a crescent-shaped dimer with a twisted eight-strandednti-parallel �-sheet forming the concave surface (Fig. 3).pproximately 710 A2, or 11% of a subunit surface area isccluded by dimerisation. The tetramer observed in solutions generated in the crystal by a crystallographic twofold axis ofymmetry. Interactions between two sets of �2 helices of neigh-ouring dimers stabilise this quaternary structure by formationf a four-helix bundle (Fig. 3). There are three types of inter-ubunit interactions and the total protein–protein interface areaithin the tetramer is approximately 1940 A2, equal to about0% of the surface area of an individual subunit. This indicatesstrongly associated assembly in agreement with the gel filtra-

ion data. These interface area values are comparable to thosealculated for RnPCD (data not shown).

.3. The TgPCD active site

The major contributions to create the active site are fromhe N-terminal and C-terminal segments of strands �2 and �3,espectively, of one subunit with additional contributions fromhe �3–�4 turn and the �2–�2 turn from the partner. Althoughhe subunit structure is well conserved there are differences inhe occupancy of the two active sites. Despite the requirementor BH2 to be present in order to get crystals, we do not observehis ligand in either the medium- or high-resolution structures.

he high ionic strength under which the most ordered crystalsere obtained may have prevented ligand binding. Instead there

re 13 well-ordered water molecules (average B-factor 24.1 A2)nd a Na+ ion (B-factor 21.5 A2) in the active site that is mainly

loa

f twofold symmetry have the helices coloured yellow and cyan. Atoms of BH2

re shown as black spheres.

ormed by residues of subunit A. The Na+ displays octahedraloordination with five waters and the carbonyl of Ser77 (nothown). The mean Na+ O coordination distance is 2.39 A, inood agreement with the average distance for such coordination2.42 A) obtained in a survey of high-resolution protein crystaltructures [21]. The Na+ ion is about 1.5 A distant from the BH22 group when the structure and the low-resolution TgPCD:BH2

omplex are superposed and the presence of this metal ion likelyrecludes ligand binding here. In the other active site containedn the asymmetric unit, created mainly by residues of subunit B,here is diffuse electron density (not shown) but this could note interpreted satisfactorily as BH2 or any other component ofhe crystallisation mixture.

Under different, lower ionic strength conditions and a much

onger period of crystal growth, less ordered samples werebtained but which clearly had BH2 bound (Fig. 4) in eachctive site in a similar manner to that observed in RnPCD (PDB

136 S. Cameron et al. / Molecular & Biochemical Parasitology 158 (2008) 131–138

F accoB , αcalc

a 2 atomb a, N5

cdwc

stcTfasrHp

cidohcdO

pdgoNpauA

3himORs2aossSutaid[

lcoiibtclb

ig. 4. BH2 in the TgPCD active site. Amino acids are shown as sticks colouredH2 green, N blue, O red. The omit difference density map for the ligand (Fo–Fc

nd αcalc is the calculated phases without the scattering contributions of the BHonding interactions are shown as dashed lines and BH2 atoms N1, N3, O4, C4

ode 1DCP), and in a conformation consistent with the pre-icted mode of substrate recognition [9]. A feature also sharedith the structure of RnPCD:BH2 is that there are no major

onformational changes following ligand binding.The side chain of His61 creates the floor of the TgPCD active

ite (Fig. 4). Here, BH2 lies on top of the imidazole and is posi-ioned by a number of hydrogen bonds with main chain and sidehain groups which will be detailed shortly. The side chain ofyr69, from the partner subunit in the dimer, is 3.8 A distantrom the pterin C7 and together with the side chain of Phe42,lso from the partner subunit, helps to form part of the activeite cleft. By also packing against the side chain of His61 theseesidues, in conjunction with a hydrogen bond donated fromis61 NE2 to Asp88 OD1, serve to hold this critical histidine inosition.

The pterin N2 accepts hydrogen bonds donated from the mainhain carbonyl groups of His62 and Ser77. His62 participatesn two additional hydrogen bonds with the ligand, the amideonates a hydrogen bond to N1 and the side chain ND1 acceptsne from N8. The His62 imidazole is positioned to accept thisydrogen bond due to a hydrogen bond between NE2 and thearboxylate group of Asp57. N5 is solvent accessible and 3.8 Aistant from Asp80 OD2 in subunit A, 4.1 A in subunit B. The9 hydroxyl group donates a hydrogen bond to Asp60 OD2.The possible hydrogen bonding interactions formed by the

rotein with BH2 N3 and O4 are of particular interest. Theistances of 3.1 and 3.3 A in subunits A and B, respectively, sug-ests a hydrogen bond is formed between the main chain amidef His79 and N3 though the major tautomer of BH2, which has3 as a hydrogen bond donor not an acceptor (Fig. 1b). The

terin O4 accepts a hydrogen bond donated by the main chainmide of Asp80 (separation of 3.3 and 3.6 A in the two sub-nits per asymmetric unit) and the distances of the side chainsp80 OD1 are also suggestive of a hydrogen bond, 2.9 and

wreT

rding to atom type: C for subunit A residues are grey, for subunit B pink, C forwhere Fo and Fc are the observed and calculated structure factors, respectively,s) is shown as cyan-coloured chicken wire at a 2.0σ level. Selected hydrogen

and N8 are numbered.

.5 A. Asp80 may be protonated however, it is likely that BH2as been stabilised in the minor enol tautomeric state, that whichs most similar to the substrate (Fig. 1) to provide the appropriateatch of hydrogen bond donor and acceptor groups at N3 and4. Although tautomerisation was not discussed in the study ofnPCD binding BH2 [9] our evaluation suggests that the sameituation arises also in the rat enzyme structure determined at.3 A resolution. The proximity of BH2 N3 to the main chainmide of RnPCD His80 clearly suggests that a tautomeric formf the ligand has bound. The average N· · ·N separation based oneven of the eight active sites in the asymmetric unit is 3.1 A. Theingle outlier is a subunit where a distance of 3.9 A is observed.tabilisation of a tautomeric state by binding to a protein is notnique and has been observed previously for ligands similar tohat studied here. For example in the structure of a ricin:pteroiccid complex [22]. We also note that such enolisation of pterinss exploited in enzyme mechanisms of important anti-parasiticrug targets dihydrofolate reductase [23] and pteridine reductase24].

The TgPCD:BH2 complex has been determined at low reso-ution and caution must be exercised in the assignment of mainhain protein ligand interactions in particular where on the basisf such interactions the presence of a minor tautomer is beingnvoked. A referee of our study has sensibly raised the possibil-ty that a peptide flip may occur to provide the correct hydrogenonding pattern to interact with the major tautomer of BH2 andhat such an alteration may also be present in the RnPCD:BH2omplex. We are unable to further investigate the higher reso-ution RnPCD:BH2 complex since the diffraction data have noteen placed in the public domain. For the TgPCD:BH2 complex

e deleted residues Thr78-His79-Asp80 and carried out several

ounds of refinement and generated omit electron and differ-nce density maps. We also flipped the peptide linkage betweenhr78 and His79, refined that model and generated maps. In both

hemic

ctpapnhco

oteTgH4pbg

dtTBtett

3

pucPooattnyAaaAtabs

oiar

LTcboIgrghf

4

Peaicezbistsspnom

A

oaPc

R

S. Cameron et al. / Molecular & Bioc

ases (data not shown) the best fit to electron density maps washe main chain conformation of the model we present here. Inarticular when analysing the flipped Thr78-His79 model, neg-tive difference density appeared on the carbonyl oxygen andositive difference density appeared to suggest that the origi-al model is correct. The diffraction data were available for theigh-resolution ricin:pteroic acid study [22] and in similar cal-ulations we were able to confirm that the minor tautomeric formf the ligand is indeed present there.

PCD catalyses dehydration of the (R)- and (S)-enantiomersf pterin-4a-carbinolamine equally [7] and different possibili-ies for the enzyme mechanism have been discussed by Cronkt al. [9]. There is adequate space in the active site of thegPCD:BH2 complex to accommodate the additional hydroxylroup attached at C4a for either the (R)- or (S)-forms (Fig. 4).is61 and His79 are positioned either side of C4a, at distances of.9 and 5.3 A, respectively. The imidazole groups are thereforelaced to participate in catalysis. His61 could act as a generalase in the dehydration of the (R)-conformation and His79 theeneral base to deal with the (S)-enantiomer.

Previous chemical studies have indicated that PCD toleratesifferent substituents at C6 [7,8]. This has erroneously beenaken to imply the presence of a flexible binding pocket [5].he structures of RnPCD [9] and now TgPCD in complex withH2 clearly show that active sites are well-ordered pockets and

hat no large-scale conformational changes need be inferred innzyme activity. The C6 substituent is simply directed out ofhe active site towards bulk solvent with no structural featureso impose steric restrictions.

.4. Comparisons with RnPCD, PfPCD and LmPCD

Conservation of key amino acids in the four PCD orthologuesresented in Fig. 2 suggests that the enzymes display similar sub-nit structures and quaternary assemblies. Of note is the strongonservation of hydrophobic residues (Met49, Trp24 and 65,he37, 39, 42 and 48) involved in creating the hydrophobic coref the subunit. Eleven residues in and around the active sitef TgPCD have been described (Fig. 4) and comparison of thective site structures of RnPCD and TgPCD reveal that here toohe sequence conservation is pronounced. They both contain thehree histidine residues (His61, His62 and His79 with TgPCDumbering), identified as important for ligand binding and catal-sis [13]. In addition the three aspartates (Asp57, Asp80 andsp88) in TgPCD that form hydrogen bonds with His62, His79

nd His61, respectively, and help to position them within thective site cleft are conserved in RnPCD as Glu58, Glu81 andsp89. Phe42 and Tyr69, which form a hydrophobic part of

he active site, are strictly conserved. The final three residuesre Asp60, Ser77 and Thr78, all three of which form hydrogenonds to the ligand and which are strictly conserved in the twotructures.

The amino acid sequences of the P. falciparum and L. major

rthologues also indicate a high level of sequence conservationn the active site with the three histidines, Phe42, Tyr69, Thr78nd Asp88 being strictly conserved (Fig. 2). The remaining fouresidues, Asp57, Asp60, Ser77 and Asp80 in TgPCD align to

al Parasitology 158 (2008) 131–138 137

ys52, Asp55, Tyr72, and Thr75 in PfPCD and to Glu62, Gln65,hr82 and Asp85 in LmPCD. The substitutions in LmPCD areonservative. The Asp57 to Lys52 and Ser77 to Tyr72 changesetween TgPCD and PfPCD are worth further comment. Neitherf the side chains is involved in direct interaction with the ligand.n the case of the serine it is the main chain that forms a hydro-en bond with BH2. The inclusion of a basic and large residue toeplace Asp57 in the PCD active site would remove the hydro-en bonding association with one of the important active siteistidine residues. Small structural changes might compensateor this change.

. Conclusions

We have determined a high-resolution crystal structure ofCD from T. gondii, and a low-resolution structure in the pres-nce of the product analogue, BH2. Our analysis of the structuresnd comparisons with RnPCD suggest that the ligand BH2 bindsn a less common tautomeric form. Part of our motivation toharacterize TgPCD was to investigate the potential of thisnzyme as a target for therapeutic intervention against proto-oan pathogens. There is a high degree of structural similarityetween TgPCD and the mammalian orthologue, RnPCD, whichs identical to the human enzyme. Consideration of the PCDequences derived from other parasites indicate conservation ofhe amino acids that contribute to the fold, that create the activeite and that participate in catalysis. Such a high degree of con-ervation in the active sites would render it difficult to identify aotent species-specific inhibitor of value in the development ofew therapeutic agents. There remains however the possibilityf discovering inhibitors to assist in delineating aspects of pterinetabolism in protozoan parasites.

cknowledgements

This work was funded by the BBSRC Structural Proteomicsf Rational Targets initiative and by the Wellcome Trust. Were grateful to C. Hunter and J.S. Strumhofer (University ofennsylvania) for the gift of T. gondii RNA used to generate theDNA library.

eferences

[1] Thony B, Auerbach G, Blau N. Tetrahydrobiopterin biosynthesis, regener-ation and functions. Biochem J 2000;347:1–16.

[2] Cunningham ML, Beverley SM. Pteridine salvage throughout the Leish-mania infectious cycle: implications for antifolate chemotherapy. MolBiochem Parasitol 2001;113:199–213.

[3] Lye LF, Cunningham ML, Beverley SM. Characterization of quinonoid-dihydropteridine reductase (QDPR) from the lower eukaryote Leishmaniamajor. J Biol Chem 2002;277:38245–53.

[4] Curtius HC, Adler C, Rebrin I, Heizmann C, Ghisla S. 7-Substitutedpterins: formation during phenylalanine hydroxylation in the absence of

dehydratase. Biochem Biophys Res Commun 1990;172:1060–6.

[5] Wang Q, Hauser V, Read M, et al. Functional identification oforthologous genes encoding pterin recycling activity in Plasmodium fal-ciparum and Toxoplasma gondii. Mol Biochem Parasitol 2006;146:109–12.

1 hemic

[

[

[

[

[

[

[

[

[

[

[

[

[

[

38 S. Cameron et al. / Molecular & Bioc

[6] Rebrin I, Bailey SW, Boerth SR, Ardell MD, Ayling JE. Catalyticcharacterization of 4a-hydroxytetrahydropterin dehydratase. Biochemistry1995;34:5801–10.

[7] Rebrin I, Thony B, Bailey SW, Ayling JE. Stereospecificity and catalyticfunction of histidine residues in 4a-hydroxy-tetrahydropterin dehydrataseDCoH. Biochemistry 1998;37:11246–54.

[8] Endrizzi JA, Cronk JD, Wang W, Crabtree GR, Alber T. Crystal structure ofDCoH, a bifunctional, protein-binding transcriptional coactivator. Science1995;268:556–9.

[9] Cronk JD, Endrizzi JA, Alber T. High-resolution structures of the bifunc-tional enzyme and transcriptional coactivator DCoH and its complex witha product analogue. Protein Sci 1996;5:1963–72.

10] Rose RB, Bayle JH, Endrizzi JA, Cronk JD, Crabtree GR, Alber T. Struc-tural basis of dimerization, coactivator recognition and MODY3 mutationsin HNF-1alpha. Nat Struct Biol 2000;7:744–8.

11] Ficner R, Sauer UH, Stier G, Suck D. 3-Dimensional structure of thebifunctional protein Pcd/Dcoh, a cytoplasmic enzyme interacting withtranscription factor Hnf1. EMBO J 1995;14:2034–42.

12] Matthews BW. Determination of molecular weight from protein crystals. JMol Biol 1974;82:513–26.

13] Kabsch W. Automatic processing of rotation diffraction data from crys-tals of initially unknown symmetry and cell constants. J Appl Crystallogr

1993;26:795–800.

14] Storoni LC, McCoy AJ, Read RJ. Likelihood-enhanced fast rotation func-tions. Acta Crystallogr Sect D Biol Crystallogr 2004;60:432–8.

15] Emsley P, Cowtan K. Coot: model-building tools for molecular graphics.Acta Crystallogr Sect D Biol Crystallogr 2004;60:2126–32.

[

[

al Parasitology 158 (2008) 131–138

16] Murshudov GN, Vagin AA, Dodson EJ. Refinement of macromolecularstructures by the maximum-likelihood method. Acta Crystallogr Sect DBiol Crystallogr 1997;53:240–55.

17] Winn MD, Isupov MN, Murshudov GN. Use of TLS parameters to modelanisotropic displacements in macromolecular refinement. Acta CrystallogrSect D Biol Crystallogr 2001;57:122–33.

18] Laskowski RA, MacArthur MW, Moss DS, Thornton JM. PROCHECK: aprogram to check the stereochemical quality of protein structures. J ApplCrystallogr 1993;26:283–91.

19] DeLano WL. The PyMOL molecular graphics system. San Carlos, CA,USA: DeLano Scientific; 2002. http://pymol.sourceforge.net.

20] Krissinel E, Henrick K. Inference of macromolecular assemblies fromcrystalline state. J Mol Biol 2007;372:774–97.

21] Harding MM. Small revisions to predicted distances around metal sitesin proteins. Acta Crystallogr Sect D Biol Crystallogr 2006;62:678–82.

22] Yan X, Hollis T, Svinth M, et al. Structure-based identification of a ricininhibitor. J Mol Biol 1997;266:1043–9.

23] Casarotto MG, Basran J, Badii R, Sze KH, Roberts GC. Direct measurementof the pKa of aspartic acid 26 in Lactobacillus casei dihydrofo-late reductase: implications for the catalytic mechanism. Biochemistry1999;38:8038–44.

24] Gourley DG, Schuttelkopf AW, Leonard GA, et al. Pteridine reductasemechanism correlates pterin metabolism with drug resistance in trypanoso-matid parasites. Nat Struct Biol 2001;8:521–5.

25] Cruickshank DWJ. Remarks about protein structure precision. Acta Crys-tallogr Sect D Biol Crystallogr 1999;55:583–601.