Structure and Function of a Novel Purine Specific Nucleoside ...of the IAG-NH, absent from the IU-NH, imposes the purine speciﬁcity. Asp10 is the proposed general base in the reaction

Vrije Universiteit Brussel

Structure and function of a novel purine specific nucleoside hydrolase from TrypanosomavivaxVersées, W; Decanniere, K; Pellé, R; Depoorter, J; Brosens, E; Parkin, D W; Steyaert, J

Published in:Journal of Molecular Biology

DOI:10.1006/jmbi.2001.4548

Publication date:2001

Document Version:Final published version

Link to publication

Citation for published version (APA):Versées, W., Decanniere, K., Pellé, R., Depoorter, J., Brosens, E., Parkin, D. W., & Steyaert, J. (2001). Structureand function of a novel purine specific nucleoside hydrolase from Trypanosoma vivax. Journal of MolecularBiology, 307(5), 1363-1379. https://doi.org/10.1006/jmbi.2001.4548

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portalTake down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Download date: 16. Jul. 2021

https://doi.org/10.1006/jmbi.2001.4548

https://cris.vub.be/portal/en/publications/structure-and-function-of-a-novel-purine-specific-nucleoside-hydrolase-from-trypanosoma-vivax(19b062ae-f730-4dc1-9c75-52d84e90e6a3).html

https://doi.org/10.1006/jmbi.2001.4548

doi:10.1006/jmbi.2001.4548 available online at http://www.idealibrary.com on J. Mol. Biol. (2001) 307, 1363±1379

Structure and Function of a Novel Purine SpecificNucleoside Hydrolase from Trypanosoma vivax

W. VerseÂes1, K. Decanniere1, R. PelleÂ 2, J. Depoorter1, E. Brosens1

D. W. Parkin3 and J. Steyaert1*

1Dienst Ultrastructuur, VlaamsInteruniversitair instituut voorBiotechnologie, VrijeUniversiteit Brussel,Paardenstraat 65, B-1640Sint-Genesius-Rode, Belgium2International LivestockResearch Institute, P.O. Box30709, Nairobi, Kenya3Division of BiologicalChemistry & MolecularMicrobiology, School of LifeSciences, Wellcome TrustBiocentre, University ofDundee, Dundee, DD1 5EHScotland, UK

E-mail address of the [email protected]

Abbreviations used: PRTase,phosphoribosyltransferase; IU-, inospreferring; IG-, inosine-guanosine-pinosine-adenosine-guanosine-preferrhydrolase(s); Ni-NTA, nickel-nitrilomultiple wavelength anomalous disp-nitrophenylriboside or pNPR, parb-D-ribofuranoside.

0022-2836/01/051363±17 $35.00/0

The purine salvage pathway of parasitic protozoa is currently consideredas a target for drug development because these organisms cannot syn-thesize purines de novo. Insight into the structure and mechanism of theinvolved enzymes can aid in the development of potent inhibitors, lead-ing to new curative drugs. Nucleoside hydrolases are key enzymes in thepurine salvage pathway of Trypanosomatidae, and they are especiallyattractive because they have no equivalent in mammalian cells. Wecloned, expressed and puri®ed a nucleoside hydrolase from Trypanosomavivax. The substrate activity pro®le establishes the enzyme to be a mem-ber of the inosine-adenosine-guanosine-preferring nucleoside hydrolases(IAG-NH). We solved the crystal structure of the enzyme at 1.6 AÊ resol-ution using MAD techniques. The complex of the enzyme with thesubstrate analogue 3-deaza-adenosine is presented. These are the ®rststructures of an IAG-NH reported in the literature. The T. vivax IAG-NHis a homodimer, with each subunit consisting of ten b-strands, 12a-helices and three small 310-helices. Six of the eight strands of the centralb-sheet form a motif resembling the Rossmann fold. Superposition of theactive sites of this IAG-NH and the inosine-uridine-preferring nucleosidehydrolase (IU-NH) of Crithidia fasciculata shows the molecular basis ofthe different substrate speci®city distinguishing these two classes ofnucleoside hydrolases. An ``aromatic stacking network'' in the active siteof the IAG-NH, absent from the IU-NH, imposes the purine speci®city.Asp10 is the proposed general base in the reaction mechanism, abstract-ing a proton from a nucleophilic water molecule. Asp40 (replaced byAsn39 in the IU-NH) is positioned appropriately to act as a general acidand to protonate the purine leaving group. The second general acid,needed for full enzymatic activity, is probably part of a ¯exible looplocated in the vicinity of the active site.

# 2001 Academic Press

Keywords: nucleoside hydrolase; purine salvage pathway; target for drugdevelopment; Trypanosoma; multiple wavelength anomalous dispersion
*Corresponding author
Introduction

Parasitic diseases remain a huge health care pro-blem, especially in developing countries. Trypano-

ing author:

ine-uridine-referring; IAG-,ing; NH, nucleoside

triacetic acid; MAD,persion;

a-nitrophenyl

somiasis, or sleeping sickness, is caused by aunicellular protozoon of the genus Trypanosoma.Annually, about 25,000 to 50,000 new cases ofsleeping sickness are reported in the sub-Sahararegion.1-3 Trypanosomiasis is, moreover, one of themost important illnesses of domestic livestock inAfrica. Trypanosoma brucei gambiense and Trypanoso-ma brucei rhodesiense are the causative agents ofsleeping sickness in human beings, while Trypano-soma vivax, Trypanosoma congolense and Trypanoso-ma brucei brucei cause the disease in domesticanimals and wildlife. It is clear that there is anurgent need for new ef®cient drugs to treat trypa-nosomiasis, especially since the drugs used at

# 2001 Academic Press

1364 Structure and Function of IAG Nucleoside Hydrolase

present show very high host toxicity. Moreover,the wide-spreading drug resistance makes treat-ment very dif®cult in large areas of Africa.4,5 Inthis regard, a number of trypanosomal pathwaysare being investigated as possible targets for drugdevelopment.6,7

One of these pathways is the purine salvagepathway.8 All parasitic protozoa (e.g. Plasmodium,Toxoplasma, Leishmania, Trypanosoma) lack the abil-ity to synthesize purine nucleotides de novo.Instead, they use salvage enzymes to obtain purinebases and nucleosides from their hosts and convertthem to the corresponding nucleotides.9-11 Aftertheir uptake by speci®c transporters,12,13 the purinenucleosides are ®rst hydrolyzed to ribose and thebase by nucleoside hydrolases. The bases are thenconverted to the concomitant nucleotides by pho-phoribosyltransferases (PRTases). Interfering withpurine salvage by blocking the key enzymes couldbe an effective way of killing these organisms. ThePRTases have already been studied extensively astargets for speci®c inhibitor design.14-16 However,due to the many parallel routes that exist in thetrypanosomal purine salvage pathways, blockingof this enzyme alone probably will not be enoughto kill the parasite.7 The nucleoside hydrolasesform another important and ubiquitous class ofpurine salvage enzymes. Considering drug devel-opment, they have the advantage that, in contrastto the PRTases, no nucleoside hydrolase activityhas been reported thus far in mammalian cells.10

The nucleoside hydrolases catalyze the hydroly-sis of the N-glycosidic bond between the anomericcarbon atom of ribose and the purine or pyrimi-dine base. A nucleoside hydrolase activity hasbeen detected in crude lysates of different speciesof the Trypanosomatidae.11,17-20 Only four nucleosidehydrolases (two from Crithidia fasciculata.21,22 onefrom T. brucei brucei23 and one from Leishmaniamajor24) have thus far been puri®ed and character-ized. These have been divided into three subclassesaccording to their substrate speci®city. The ®rstclass is the base-aspeci®c inosine-uridine-preferringnucleoside hydrolases (IU-NH) found in C. fas-ciculata21 and L. major.24 The other two classes, aninosine-adenosine-guanosine-preferring nucleosidehydrolase (IAG-NH) from T. brucei brucei23 and aninosine-guanosine-preferring nucleoside hydrolase(IG-NH) from C. fasciculata,22 show speci®citytoward purine nucleosides.

The IU-NHs have been studied in depth. Usingkinetic isotope effects, it was shown that theenzyme-catalyzed reaction occurs via an SN1 mech-anism with an oxocarbenium ion transitionstate.25,26 The IU-NHs obtain most of their catalyticpower via stabilization of this intrinsically veryunstable oxocarbenium ion. Only minor inter-actions occur with the purine or pyrimidine leav-ing group.27 Three IU-NH crystal structures (up toa resolution of 2.3 AÊ ) have been reported.24,28,29

The structural properties of the purine-speci®cnucleoside hydrolases (IAG-NH and IG-NH) havenot been characterized. However, the substrate

speci®city and the pH-dependence of the kineticconstants indicate that the catalytic mechanismsand the transition state structures are fundamen-tally different for the IU and IAG isozymes.23

In comparison with the IU-NHs, the IAG-NHsrely much more on leaving group activation forcatalysis.30

Here, we report the cloning of a nucleosidehydrolase from Trypanosoma vivax, the causativeagent of the Souma disease in African cattle.31 Theprotein is expressed in Escherichia coli as a His-tagged fusion protein that allows easy puri®cation.Determination of the substrate speci®city estab-lishes the enzyme to be an IAG-NH. Kinetic dataobtained with a substrate analogue con®rm theimportance of leaving group activation in the cata-lytic mechanism. A 1.6 AÊ resolution X-ray structureof the unliganded protein and a 2.1 AÊ resolutionX-ray structure of the protein complexed with thesubstrate analogue 3-deaza-adenosine are reported.These structures reveal insights into the differencesbetween IU and IAG-NHs and are an indispensa-ble tool for inhibitor design.

Results and Discussion

Cloning and sequence comparison

The cloned cDNA encoding the T. vivax IAG-NH is 1245 nt long and contains a mini-exon orspliced leader sequence from bases 1 to 39. Thenucleoside hydrolase ORF codes for a protein of327 amino acid residues, including the N-terminalmethionine residue. For cloning and puri®cationpurposes, the start codon was removed andreplaced by a sequence coding for 13 amino acidresidues, including the six histidine residues fromthe His-tag. Figure 1(a) compares the amino acidsequence of the nucleoside hydrolase from T. vivaxwith the IAG-NH from T. brucei brucei32 and theIU-NHs from C. fasciculata27 and L. major.24 TheT. vivax nucleoside hydrolase shares a highsequence identity with the IAG-NH from T. brucei.brucei (66 %). The sequence identity with bothIU-NHs is much lower (24 % with the IU-NH fromL. major and 26 % with that from C. fasciculata). TheIAG-NHs and IU-NHs align best in the regionsthat have been shown to interact with the ribosepart of the substrate in the IU-NH. The residues ofthe IU-NHs involved in binding of a calcium ion inthe active site are conserved in the IAG-NHs.29

Sequence alignment with the IU-NH is much moredif®cult in those regions that interact with the baseof the nucleoside.

Molecular mass and subunit structure

The nucleoside hydrolase subunit structure isquite variable. The IU-NH from C. fasciculata andL. major are homotetramers.21,24 The IG-NH fromC. fasciculata is a homotrimer.22 The IAG-NH fromT. brucei brucei is a homodimer.23

Figure 1. The primary and secondary structure of the T. vivax IAG-NH. (a) Amino acid sequence alignment ofT. vivax IAG-NH with T. brucei brucei IAG-NH, C. fasciculata IU-NH and L. major IU-NH. The secondary-structureelements corresponding with the T. vivax IAG-NH sequence are indicated. Amino acid residues 245 to 256 are miss-ing from the uncomplexed structure. (b) Topology diagram of the T. vivax IAG-NH. The numbering of the secondarystructure elements and color codes are the same as in (a). b1-b6 together with a1-a7 form a motif resembling theRossmann fold.

Structure and Function of IAG Nucleoside Hydrolase 1365

Table 1. Kinetic constants for the nucleoside hydrolase from T. vivax

Substrate or inhibitor kcat (sÿ1) KM (mM) kcat/KM (Mÿ1 sÿ1) KI (mM)

Inosine 5.19 � 0.08 5.37 � 0.42 (9.7 � 0.7) � 105

without His-taga 5.49 � 0.04 5.20 � 0.30 (10.5 � 0.6) � 105

Adenosine 1.46 � 0.06 8.54 � 1.29 (1.71 � 0.27) � 105

Guanosineb 1.90 3.82 (4.97 � 0.19) � 105

Cytidine 0.338 � 0.005 925 � 39 (3.66 � 0.16) � 102

Uridine 0.022 � 0.002 586 � 150 (3.75 � 1.02) � 101

p-nitrophenylriboside 0.206 � 0.005 257 � 18 (8.02 � 0.59) � 102

3-Deaza-adenosine 0.2 � 0.027-Deaza-adenosine 356.5 � 30.3

Kinetic constants calculated per active site as described in Materials and Methods.a Kinetic data for the nucleoside hydrolase with cleaved His-tag.b Errors on kcat and KM could not be determined separately, since kcat/KM was determined via a progress curve, see the text.


We estimated the apparent molecular mass ofthe active nucleoside hydrolase from T. vivax usinggel chromatography on a 10/30 Superdex-200 HRcolumn. The enzyme elutes with a distributioncoef®cient (Kd in equation (1)) of 0.42 correspond-ing to a calculated molecular mass of 59,600 Da.Denaturing polyacrylamide gel electrophoresisgave a single band with a subunit molecular massof approximately 35,000 Da, consistent with thepredicted subunit molecular mass of 37,584 Da.The crystal structure (see below) shows the enzymeas a dimer with a large and hydrophobic interface.All data indicate the protein to be a dimer. Thismeans that the enzyme is retarded on the gel-®l-tration column, causing an underestimation of theapparent molecular mass. Alternatively, theenzyme is in a fast equilibrium between the mono-meric and the dimeric form in solution.

Substrate specificity

The substrate activity pro®le of the T. vivaxnucleoside hydrolase shown in Table 1 is exemp-lary for an IAG-NH (see also Figure 2). Theenzyme is a factor of 1000 to 10,000 more speci®ctowards the naturally occurring purine nucleosidesas compared to the pyrimidine nucleosides, due toboth a faster turnover and a higher substrateaf®nity (lower KM) for the former. The substrateanalogue para-nitrophenyl b-D-ribofuranoside

Table 2. Comparison of kinetic properties of IAG-NHs fromand L. major, and IG-NH from C. fasciculata

IAG-NH Tv IAG-NH Tbba IU

kcat (sÿ1) KM (mM) kcat (sÿ1) KM (mM) kcat (sÿ

Inosine 5.2 5.4 34 18 28Adenosine 1.5 8.5 18 15 4.3Guanosine 1.9 3.8 38 46 1.5Cytidine 0.3 925 0.3 106 20Uridine 0.02 586 0.01 102 143p-NPRe 0.2 257 0.8 562 239

a Data from Parkin.23

b Data from Parkin et al.21 and Gopaul et al.27

c Data from Shi et al.24

d Data from EstupinÄ aÂn & Sahramm.22

e p-Nitrophenyl-b-D-ribofuranoside.

(p-nitrophenylriboside) is a relatively poor sub-strate for the T. vivax nucleoside hydrolase. Theturnover number equals 0.2 sÿ1 as compared to239 sÿ1 for the IU-NH from C. fasciculata27 (seeTable 2). This difference is informative for the waythe transition state of the enzyme-catalysed reac-tion is reached. para-Nitrophenylriboside cannot beprotonated on the leaving group, but is susceptibleto decomposition when the ribosyl group is con-verted to the oxocarbenium ion.33 Because theIU-NH obtain most of their catalytic power fromstabilisation of the oxocarbenium ion, p-nitrophe-nylriboside is hydrolysed ef®ciently by this class ofenzymes.27 For the IAG-NHs, leaving groupactivation is believed to be more important for cat-alysis, making p-nitrophenylriboside less suitableas a substrate.30

The compounds 3-deaza-adenosine and 7-deaza-adenosine are analogues of adenosine with thenitrogen atom of the purine ring replaced by a car-bon atom on position 3 and 7, respectively(Figure 2). Neither of those compounds is hydro-lysed at a detectable rate but they bind to theT. vivax IAG-NH with inhibition constants of0.2 mM and 356 mM, respectively. The observationthat 3-deaza-adenosine binds with such high af®-nity indicates that N-3 of the natural purine is notinvolved in any binding interactions. Nitrogen 7 ismost probably involved in one or more bindinginteractions since the loss of this nitrogen atom

T. vivax and T. brucei brucei, IU-NHs from C. fasciculata

-NH Cfb IU-NH Lmc IG-NH Cfd

1) KM (mM) kcat (sÿ1) KM (mM) kcat (sÿ1) KM (mM)

380 119 445 88 16460 0.6 185 0.001 106420 0.6 140 216 774700 0.4 422 0.001 37001220 32 234 <10ÿ5

110 220 185 0.07

Figure 2. Substrates and inhibi-tors for IAG-NHs and IU-NHs.IU-NHs catalyze the hydrolysis ofpurine and pyrimidine nucleosidesequally well, while IAG-NHs preferpurine nucleosides. Abbreviationsused are: Tv-IAGNH, IAG-NHfrom T. vivax ; Cf-IUNH, IU-NHfrom C. fasciculata. Data for theCf-IUNH are from Parkin et al.21,30

and Gopaul et al.27


results in a decrease of 4.6 kcal/mol in bindingenergy as compared to 3-deaza-adenosine.

Influence of the His-tag

The kinetic studies on the T. vivax nucleosidehydrolase were carried out with an N-terminallyHis-tagged enzyme. One could imagine that such afusion peptide would in¯uence the kinetic charac-teristics of the enzyme. In order to exclude thispossibility, a construct was made with a thrombincleavage site between the His-tag and the nucleo-side hydrolase protein itself. After puri®cation theHis-tag was cleaved off and total removal was con-®rmed on a Western blot. Table 1 shows that thekinetic parameters for inosine are comparable forthe His-tagged and non-His-tagged enzyme, estab-lishing that the N-terminal His-tag does not in¯u-ence the kinetic properties of the IAG-NH.

Description of the overall structure

The 1.6 AÊ resolution structure of the unligandedT. vivax IAG-NH has been solved using the MADphasing method on a samarium derivative, as

described in Materials and Methods (see Tables 3and 4 for data collection and re®nement statistics).Although the density for most of the structure waswell de®ned (Figure 3), the 12 amino acid residuesbetween methionine 244 and tyrosine 257 are miss-ing from the electron density of both subunits inthe asymmetric unit.

The asymmetric unit of the crystal structure con-tains a homodimer, with each monomer being asingle-domain, globular protein (Figure 4). Eachsubunit consists of ten b-strands, 12 a-helices andthree short 310 helices (Figure 5). The overalltopology diagram of the monomer is shown inFigure 1(b). Eight of the ten b-strands form acentral mixed b-sheet (b1-b7 and b10), with sevenparallel strands and one antiparallel strand (b7).The ®rst six strands of this sheet (b1-b6), and thea-helices 1-7, form a motif that resembles theRossmann fold.34 The ®rst babab motif of this clas-sical fold is known to be involved in nucleotidebinding. Exceptionally, the phosphate-binding con-sensus sequence GXGXXG,35 normally located inthe loop connecting b1 and a1, is not conserved inthe nucleoside hydrolases. In the nucleoside hydro-lases, this consensus sequence is replaced by an

Table 3. Data collection, scaling and phase re®nement statistics for high-resolution native data, MAD data and dataof the 3-deaza-adenosine complexed structure

Native structure 3-Deaza-ado

Native MAD remote MAD peak MAD inflection

Space group P21 P21 P21 P21 P21

a (AÊ ) 52.09 52.26 52.26 52.26 51.67b (AÊ ) 74.00 74.97 74.97 74.97 74.29c (AÊ ) 81.00 82.14 82.14 82.14 81.51b (deg.) 104.56 104.587 104.587 104.587 103.54Volume (AÊ 3) 305,926 311,449 311,449 311,449 304,194No. subunits per a.u. 2 2 2 2 2l (AÊ ) 0.8445 0.94972 1.69518 1.69611 0.9786Resolution range (AÊ ) 30.0 - 1.6 30.0 - 2.0 20.0 - 2.3 20.0 - 2.3 30.0 -2.1No. observations 629,834 1,130,188 1,117,001 1,107,260 400,764No. unique reflections 79,522 41,563 27,369 27,379 34,667Overall

Rsym (%) 5.5 6.7 7.6 7.9 6.2I/sI 10.4 6.6 6.2 5.9 15.7Data completeness (%) 99.8 99.9 99.8 99.8 98.7Anomalous compl. (%) 99.4 99.2 99.4Mean multiplicity 4.4 8.3 6.2 6.2 3.39

Highest-shellResolution range (AÊ ) 1.69-1.6 2.11-2.0 2.42-2.3 2.42-2.3 2.18-2.1Rsym (%) 38.5 9.9 14.3 15.0 37.7I/sI 2.0 4.2 4.0 3.6 2.33Data completeness (%) 99.5 99.4 100.0 100.0 96.1Anomalous compl. (%) 98.4 99.4 99.5Mean multiplicity 3.4 6.5 5.9 5.9 3.17

No. Sm sites - 7 7 7 -Phasing power centric - 1.96 1.87 1.75 -Phasing power acentric - 3.89 4.52 3.98 -Rcullis centric - 0.67 0.70 0.73 -Rcullis acentric - 0.44 0.39 0.44 -Rcullis anomalous - 0.71 0.64 0.61 -Figure of merit Centric Acentric Overall

0.86 0.73 0.74

Rsym � �jIi (hkl) ÿ hIi (hkl)ij/�Ii (hkl), where Ii (hkl) are the intensities of multiple measurements and hIi (hkl)i is the average of themeasured intensities for the ith re¯ection.

Rcullis � �jjjFPH (hkl)j � jFP (hkl)jj ÿ jFH (hkl)jj/�jjFPH (hkl)j � jFP (hkl)jj, where FPH, FP and FH are the derivative, native and calcu-lated heavy-atom structure factors, respectively. The sum is extended over all centric re¯ections.


aspartate-rich region involved in calcium binding,explaining their speci®city for nucleosides as com-pared to nucleotides.

The dimer interface is formed by the helices a3,a9 and a10 of both subunits, with a3 of one mono-mer interacting with a9 and a10 of the other mono-mer (Figure 6(a)). The accessible surface areaburied in the interface is about 885 AÊ 2, represent-ing about 7 % of the total accessible surface area ofa monomer. The subunit-subunit interface is 63 %non-polar, without intermolecular salt-bridges andwith only two intermolecular hydrogen bondsbetween the side-chains of Asp92 and Ser220 ofboth monomers. This subunit interface is totallydifferent (Figure 6) from that reported in theIU-NHs. The IU-NHs are tetramers and the sub-unit-subunit interfaces are formed mainly by b-strands and coils, with the subunits pointing in adifferent relative orientation than they do in theIAG-NH.24

A 2.1 AÊ resolution crystal structure of the IAG-NH in complex with 3-deaza-adenosine was alsosolved. Here, ®ve amino acid residues (Cys248-Arg252) were excluded from both subunits of the

model, due to poor electron density in this region.No large conformational differences can beobserved between the uncomplexed and the3-deaza-adenosine-complexed T. vivax nucleosidehydrolase. This is in contrast with the IU-NH fromC. fasciculata, where large conformational changeswere observed in the loops surrounding the activesite upon binding of the transition-state analoguep-aminophenyliminoribitol.29 This different beha-viour probably re¯ects a fundamental differencebetween the two inhibitors. 3-Deaza-adenosine is asubstrate analogue and the enzyme-bound speciesmight not have the properties of the transitionstate necessary to invoke the catalytically import-ant conformational changes.

The active site

The T. vivax IAG-NH dimer contains two activesites that are structurally indistinguishable. Sinceno cooperativity was observed in any of the kineticstudies, we conclude that both active sites functionindependently. The active sites are located at theC-terminal ends of the core eight-stranded b-sheet

Figure 3. Representative section of the electron den-sity map (b-strand 10; Leu304-Arg309) of uncomplexedT. vivax IAG-NH, contoured at 1 s. (a) Experimentalmap at 1.6 AÊ resolution after density modi®cation withDM. (b) The 2Fo ÿ Fc map after re®nement at a resol-ution of 1.6 AÊ . The Figure was made with CON-SCRIPT69 and MOLSCRIPT.70

Figure 4. The T. vivax IAG-NH dimer. The location ofthe active sites is indicated by the presence of a calciumion, which is represented as a blue sphere. TheFigure was made with MOLSCRIPT.70


of both subunits (Figure 5). The active-site cleft isformed by strands b1, b2, b4 and b5, and helicesa1, a3, a10 and a11. A bound octa-coordinatedcalcium ion is situated in the active site, makingcontacts with the carboxyl groups of Asp15 (bothoxygen atoms), Asp10 and Asp 261, and the main-chain carbonyl group of Thr137 (Figure 7). In theunliganded structure, the other three coordinationpositions are occupied by water molecules (in oneof the subunits, one water molecule is replaced bya glycerol molecule from the cryo buffer).

3-Deaza-adenosine is bound in the active sitewith the ribose interacting with the calcium ionand the base pointing toward the exit of the activesite (Figure 7). The ribose adopts an O4

0 endoenvelope conformation.36,37 This conformation israrely observed in solution, and the energetic costto ®x a furanose ring in this state is about 3.8 kcal/mol.38 The conformation adopted by the exocyclicC40C50 bond is antiperiplanar (ap) with torsion

angles, fOO � 70 � and fOC � 173 �.39,40 In thecomplex, the purine adopts a syn conformation

Table 4. Re®nement statistics for uncomplexed and 3-deaza-a

Nativ

Resolution (AÊ ) 30.0-1No. reflections used in Rcryst 75,34No. reflections used in Rfree 3989Rcryst (%) 18.61Rfree (%) 20.96rmsd for bond lengths (AÊ ) 0.009rmsd for bond angles (deg.) 1.56Ramachandran plot (% most favoured,allowed, generously allowed, disallowedresidues) 89.7, 9.9, 0No. atoms (protein (including alternateconformations), ligand, water, ions) per a.u. 5453Average B-factors (AÊ .2) (protein, water) 36.25, 4

Rcryst � 100 � (�hjFobs, h ÿ Fcalc, hj/�h Fobs, h ), where Fobs and Fcalc aRfree � Rcryst calculated for the test set of re¯ections not used in re®

toward the ribose (torsion angle, w (O40-C10-N9-

C4) � ÿ 1 �),41 whereas in solution, purine nucleo-sides are in equilibrium between the syn and theanti conformations. Guanosine has a higher ten-dency to adopt the syn conformation in solution.42

This conformational preference of the substratemay contribute to the speci®city of the T. vivaxnucleoside hydrolase for guanosine as compared toadenosine (Table 1).

In the 3-deaza-adenosine complexed structure,two of the Ca2�-bound water molecules arereplaced by the 20OH and 30OH groups of theribose (see Figure 2 for atom numbering of nucleo-sides). The third water molecule, still bound toCa2�, is probably the nucleophile in the SN1 reac-tion, since it is positioned ideally to attack theanomeric carbon atom of the ribose (3.34 AÊ ).Asp10 is at hydrogen bond distance from thiswater molecule (2.58 AÊ ) and could act as an activa-tor of this water nucleophile by abstracting a pro-ton. The calcium ion plays an important role inthis regard, probably by lowering the pKa of the

denosine complexed structures

e 3-deaza-adenosine

.6 30.0-2.10 32,960

170719.1025.39

6 0.00881.55

.4, 0.0 88.1, 11.5, 0.4, 0.0

53606.47 33.79, 42.28

re observed and calculated structure amplitudes, respectively.nement.

Figure 5. Two orientations of a T. vivax IAG-NH sub-unit. 3-Deaza-adenosine is represented as ball-and-stickmodel, the calcium ion and the catalytic water moleculeare depicted as dark and light blue spheres, respectively.The substrate analogue 3-deaza-adenosine is bound atthe C terminus of the eight-stranded central b-sheet. Theamino acid residues between His247 and Asp253 aremissing from the model. The Figure was prepared withMOLSCRIPT.70

Figure 6. Comparison of the quaternary structures of(a) the T. vivax IAG-NH and (b) the C. fasciculataIU-NH. Totally different interfaces are formed in theIAG-NH dimer and the IU-NH tetramer. The secondarystructure elements involved in subunit interactions inthe IAG-NH are indicated in blue. Those involved insubunit interactions in the IU-NH are indicated ingreen.


catalytic water molecule prior to proton transfer toAsp10.29 The catalytic water is further coordinatedby the amide group of Asn186.

Apart from the interaction between the Ca2�

cofactor and the 20OH and 30OH groups of theribose, several enzyme residue side-chains are incontact with 3-deaza-adenosine (Figures 7 and 8).The enzyme forms hydrogen bonds with all threehydroxyl groups of the ribose. The 50OH groupinteracts with Asn173 and Glu184, while Asn186and Asp261 are hydrogen bonded with the 30OHgroup. Asp14 occurs in multiple conformations inthe unliganded structure. Binding of 3-deaza-

adenosine ®xes its side-chain in an ideal positionto hydrogen bond with the 20OH group. One non-polar residue in the active site, Met164, is forminga van der Waals interaction with the C5

0 . It waspreviously shown for the IAG-NH from T. bruceibrucei that interactions with the 50OH group mainlycontribute to substrate binding, while interactionswith the 20OH and the 30OH groups are importantfor the stabilisation of the oxocarbenium ion in thetransition state.23 Extrapolating these results to theT. vivax IAG-NH would mean that the interactions

Figure 7. Close-up stereoview of the T. vivax IAG-NH active site in complex with 3-deaza-adenosine. The 2Fo ÿ Fc

map around 3-deaza-adenosine contoured at 1s is also shown. The calcium ion and the catalytic water molecule aredepicted as dark and light blue spheres, respectively. The residues interacting with ribose or the calcium ion arecolored light gray. The residues interacting with the 3-deaza-adenine base are colored dark gray. Asp10 (in yellow) isthe proposed catalytic base. Asp40 (in yellow) could act as leaving group activator. The Figure was made withCONSCRIPT69 and MOLSCRIPT.70


between the 50OH group and Asn173 and Glu184are fully developed in the observed ground state,while the interactions of the enzyme with the 20OHand 30OH groups would be at full strength only inthe transition state complex.

Kinetic studies with the substrate analoguep-nitrophenylriboside30 (see also Table 1) estab-lished that a large fraction of the catalytic powerfor the IAG-NHs is obtained through interactionswith the purine leaving group. Protonation of a

Figure 8. A schematic of inter-actions in the active site of theT. vivax IAG-NH. The catalyticwater molecule is situated at3.34 AÊ from the anomeric carbonatom and is indicated in red.Asp10 abstracts a proton from thiswater molecule. Asp40 probablydonates a proton to N-9 of the pur-ine. Trp83 and Trp260 are involvedin an aromatic stacking interactionwith the substrate base. Interactionsare indicated by blue dotted lines.Distances are given in AÊ .

Figure 9. Superposition of the active sites of the T. vivax IAG-NH in complex with 3-deaza-adenosine and theC. fasciculata IU-NH in complex with p-aminophenyliminoribitol (pdb 2mas). The IU-NH residues and inhibitor arecolored gray. The IAG-NH residues are colored yellow (the 3-deaza-adenosine carbon atoms are colored red). The cal-cium ion and catalytic water molecule are depicted as a dark and light blue sphere, respectively (transparent for theIU-NH). (a) The residues interacting with the ribose and with the calcium ion align perfectly between the IAG-NHand the IU-NH. (b) A much greater divergence is observed in the residues surrounding the base (see the text)


purine lowers its pKa, thus promoting its depar-ture. Protonation of N-7 has been shown to beinvolved in the acid-catalysed solvolysis of nucleo-sides and is a key step in the mechanisms ofIU-NHs, nucleoside phosphorylases and AMPnucleosidases.43,44 Hydrogen bonds between pro-teic proton donors and acceptors on the purinemay also activate the leaving group by loweringthe electron density in the aromatic system. Asp40is located at 3.5 AÊ from the leaving nitrogen atom(N-9) and is located appropriately to protonate, orhydrogen bond with, N-9 of the purine leavinggroup. Since the crystals have been soaked with 3-deaza-adenosine at neutral pH, Asp40 is probablyin the unprotonated state in this structure. Protona-tion of this carboxylate group at lower pH wouldlikely induce the Asp40 side-chain to approach thepurine N-9 (note that the pH optimum of thisenzyme is around pH 5; unpublished results).From the inhibition studies with 7-deaza-adeno-sine, it appears that the enzyme interacts with thepurine N-7 as well. Asn12 is located at 3.89 AÊ fromN-7, but given some induced ®t a hydrogen bondmight be formed.

The 3-deaza-adenine leaving group is involvedin extensive aromatic stacking interactions. Thepurine is ``sandwiched'' between the side-chains ofTrp83 and Trp260 (Figure 7). Upon binding of3-deaza-adenosine, the region between Glu82 andLeu86 undergoes a slight conformational change,aligning Trp83 perfectly parallel with the purinebase. Heterocyclic aromatic ring systems like pur-ines and tryptophan are frequently involved inthese face-to-face stacking interactions. This con-

tributes to the purine-speci®city of the IAG-NHs,since non-heterocyclic pyrimidines can only estab-lish edge-to-face stacking interactions.45 Edge-to-face stacking cannot be accommodated in theactive-site pocket. The aromatic network aroundthe leaving group is further completed by Phe79,Trp185 and Tyr257 (see Figure 10(b)).

Comparison between the IAG-NH and theIU-NH

The different quaternary structure between theT. vivax IAG-NH and the IU-NHs is striking.Totally different secondary structure elements areinvolved in the subunit interactions in the T. vivaxdimer as compared to the IU-NH tetramer fromC. fasciculata (pdb1mas) (Figure 6). The T. vivaxand C. fasciculata monomers differ mostly in theloops around the active site and in the regionsinvolved in the subunit interactions.

We also compared the active sites of theliganded IAG-NH and the IU-NH of C. fasciculatain complex with p-aminophenyliminoribitol (pdb2mas). Figure 9 illustrates the fact that the inhibi-tors in both structures are bound in the sameposition in the active sites. The 20OH, 30OH and50OH groups of both inhibitors virtually superim-pose. The leaving groups (3-deaza-adenine andp-aminophenyl) are bound in an orientation per-pendicular to each other.

All residues interacting with the ribose (or theiminoribitol in the case of p-aminophenyliminoribi-tol) are conserved in the amino acid sequence ofIU-NHs and IAG-NHs (Figure 1(a)) and superim-pose very well in the structure (Figure 9(a)). There

Figure 10. Molecular surface of the T. vivax IAG-NH around the active site, calculated without 3-deaza-adenosine,which is shown in the active site. In (a) the molecular surface is colour coded according to its electrostatic potential.Blue refers to positive and red to negative potential. 3-Deaza-adenosine is clearly bound with its ribose in a highlynegatively charged pocket. A number of charged residues, not involved in interactions with 3-deaza-adenosine, sur-round the active site. These might be employed for inhibitor design. In (b) the aromatic residues (Tyr, Phe and Trp)are colored green and the apolar residues are colored yellow. The calcium ion in the bottom of the active site isshown in blue. The outer side of the cleft clearly has a highly aromatic character. Trp83 and Trp260 form aromaticstacking interactions with the purine. The Figure was generated with GRASP.71


is only one exception, Asn39, which interacts withthe 20OH group in the IU-NH is replaced by Asp40in the IAG-NH. In the IAG-NH structure, however,the Asp40 side-chain is shifted away from the20OH group toward N-9. It thus appears thatAsn39 and Asp40 serve different functions in theserelated enzymes. Whereas Asn39 stabilises the oxo-carbenium ion, Asp40 may contribute to the acti-vation of the leaving group.

The different substrate speci®city betweenIU-NHs and IAG-NHs is re¯ected by a muchgreater divergence in the structure around the baseof the nucleoside substrate (Figure 9(b)). Ile81 inthe IU-NH is replaced by Trp83, which constitutedone half of the parallel aromatic stacking sandwich

around the purine base in the IAG-NH. The otherhalf of the sandwich is Trp260, which replacesHis241 as compared to the IU-NH. The nearlyparallel aromatic stacking interactions46 in theIAG-NH active site, absent from the active site ofthe IU-NH, contribute to the high speci®city forpurine bases of the IAG-NHs. His82 of the IU-NHis not conserved in the IAG-NH, neither. His82 hasbeen proposed to be involved in a cation-p inter-action with the leaving group.29 In the T. vivaxIAG-NH, the His82 imidazole group is structurallyreplaced by the amide group of Asn12 (Figure 9(b)).Mutagenesis studies are in progress to establish thefunction of Asn12.


The quest for proton donors

The pH pro®les and pH optima of the IAG-NH(from T. brucei brucei and T. vivax) differ consider-ably from those of the IU-NHs. It has beenreported that the IAG-NH from T. brucei bruceirequires three ionizable residues, one acid with apKa of 6.5 and two with pKa values of 8.8, allneeded in a protonated state for maximal enzy-matic turnover (kcat). Leaving group activation byproton donation to N-3 and N-7 of the purine ringis the proposed role for these residues.23 For theT. vivax nucleoside hydrolase, two groups, with apKa of 5.5 and 8.5, are found to be required in aprotonated state (unpublished results). The pHpro®le of kcat for the C. fasciculata IU-NH follows aquite different, bell-shaped curve, with a singleacid (pKa � 9.1) needed in a protonated state and abase (pKa � 7.1) required in its deprotonatedstate.47 Comparing these pH pro®les, we concludethat both classes of hydrolases require at least oneprotonated acid with a high pKa that is requiredfor the protonation of the leaving group. Appar-ently, the IAG-NHs engaged additional catalyticacids (with lower pKa) to activate the leavinggroup.

Looking at the crystal structure of the IAG-NHin complex with 3-deaza-adenosine, Asp40 is thebest candidate for the low pKa proton donor(pKa � 5.5 and 6.5 for the T. vivax and T. bruceibrucei nucleoside hydrolase, respectively). Asp40 isin an appropriate position to protonate N-9 of theleaving purine (Figures 7 and 8). Moreover, Asp40aligns with Asn39 of the IU-NH (Figure 9(b)),explaining the absence of a second (low pKa)proton donor in the pH pro®le of the IU-NH.

The identi®cation of the high pKa proton donorthat protonates the leaving group is less obviousfrom the structure of the IAG-NHs. For theIU-NHs, His241 has been proposed for this role,despite its high pKa of 9.1.27 In the T. vivax IAG-NH, this histidine residue is replaced by Trp260(Figure 9(b)). This tryptophane residue does notcorrespond to the ``high pKa`` proton donorbecause it does not ionise in the physiological pHrange. Tyr257 is a possible candidate, but the dis-tance between the ionizable hydroxyl group ofTyr257 and the exocyclic amino group of 3-deaza-adenosine is large and amounts to 3.6 AÊ . However,it should be kept in mind that ring nitrogen andketo oxygen atoms of purines and pyrimidines arepreferably protonated as compared to exocyclicamino groups.40 Most pH pro®les have been deter-mined using inosine as the substrate with a ketooxygen atom at ring position 6.

It is possible that the missing proton donor(s),required for chemical turnover, is situated in thatpart of the structure (between Met244 and Tyr257)that is poorly de®ned in the electron density. Poss-ibly, this loop undergoes a conformational changeduring the formation of the transition state tobring these groups into position for catalysis. The¯exible loop hosts a number of ionizable groups

(His247, Glu249, Arg252, Asp253 and Asp255) thatmay serve catalytic functions. Considering the highpKa and the ¯exibility of its side-chain, Arg252 isthe candidate of choice. Flexible loops containingcatalytic residues have been described. In tyrosyl-tRNA synthase, three catalytic lysine residues anda catalytic arginine residue are part of two ¯exibleloops. An induced-®t mechanism allows access ofthe substrates to the active site and enables thetransition state to be completely surrounded bygroups on the protein that would otherwise blockentry.48 Note that the IU-NH also has a similar¯exible loop that contains an arginine residue(Arg233). Mutagenesis studies will be needed topositively identify the missing proton donor and tocharacterise the functional role of the ¯exible loop.

Implications for drug design

Nucleoside hydrolases are key enzymes in thepurine salvage pathway of trypanosomes and theyhave no homologous counterparts in mammalianorganisms. Blocking these enzymes could providea means for inhibiting (or slowing) the growth ofthese parasites. For this purpose, highly speci®cand potent inhibitors are needed. The theory of``absolute reaction rates'' implies that enzymesenhance the rate of a reaction to the extent thatthey bind the altered substrate in the transitionstate (S{) more tightly than they bind the substratein the ground state (S).49 The catalytic rateenhancement for nucleoside hydrolases isestimated at 6 � 1012.50 This means that the ES{

dissociation constant is of the order of 10ÿ18 M.51

Considering this attomolar af®nity, stable ana-logues of the substrate in the transition state mayprovide very powerful non-covalent inhibitors.44

Because IU-NHs obtain most of their catalyticpower from oxocarbenium ion stabilisation, inhibi-tor design against these enzymes has focused onmimicry of the oxocarbenium ion. For example,some iminiribitols have been shown to be sub-nanomolar inhibitors of IU-NHs and IAG-NHs. Incomparison with the IU-NHs, the IAG-NHs relymuch more on interactions with the purine forturnover of the substrate. Capturing part of thisactivating energy in pre-existing, stable oxocarbe-nium analogues may be the strategy of choice fordesigning new potent inhibitors.

The present crystal structures are valuable toolsin the process of inhibitor design. However, oneshould bear in mind that a crystal structure incomplex with a substrate analogue like 3-deaza-adenosine might differ considerably from the genu-ine ES{ complex in the transition state, due to con-formational changes. For example, the ¯exible loop(linking Met244 and Tyr257) may shield more gen-uine transition-state analogues completely fromsolvent by an induced-®t mechanism. Several othergroups surrounding the active site, e.g. Lys81,Glu82, and Arg84 (Figure 10), not involved ininteractions with 3-deaza-adenosine, may interactwith the transition state. Protein engineering is the


method of choice for evaluating their role in tran-sition-state binding. Despite the shortcomings ofour structures for drug design, one improvementto the inhibitor can be thought of. Since the nucleo-sides bind to the IAG-NHs in a syn conformation,the introduction of a bulky group (e.g. bromide or¯uoride) at the C-8 position may increase the af®-nity of any analogue by favouring this confor-mation in solution.40 Bromide or ¯uoride at thisposition may interact with Asn12 via a hydrogenbond. The high cooperativity between theinteractions in the ES{ complex implies that a newfeature in an inhibitor that leads to an extra inter-action could contribute to a much greater extent tobinding af®nity than would be expected for thesame isolated interaction.49

Conclusions

T. vivax is a causative agent of trypanosomiasisin African cattle. The crystal structures of a T. vivaxIAG-NH reported here provide the ®rst structuraldescription of a purine-speci®c nucleoside hydro-lase. The structure in complex with the substrateanalogue 3-deaza-adenosine sheds light on themechanism of the enzyme-catalysed reaction. Allhydroxyl groups of the ribose are involved in inter-actions with (mainly acidic) residues of the enzymeand with a calcium ion in the bottom of the cataly-tic pocket. The speci®city toward purine bases isprovided by multiple aromatic stacking inter-actions between the purine and enzyme residues.While Asp10 is the general base in the reactionmechanism, Asp40 is proposed as a general acid.Comparison of this structure with an IU-NHshows the molecular basis of the different substratespeci®city between both enzyme classes. Mutagen-esis studies will be required to identify a proposedsecond general acid in the enzyme, and to map theindividual interactions in the enzyme/substrate orenzyme/transition state complex.

Materials and Methods

Cloning of the complete IAG-NH cDNA

The T. vivax IAG-NH homologue was cloned by PCRampli®cation of ss-cDNA generated from total poly(A) �

RNA of the bloodstream form of T. vivax, IL2160. Theprimers were designed from T. brucei brucei and T. congo-lense conserved amino acid sequences ADCFVE (forwardprimer 50-GCG-GAT-TGC-TTC-GTT-GA-30) and KVEEC(reverse primer 50-GCA-TTC-CTC-CAC-CTT-30). Theforward primer was used with an oligo(dT) primer togenerate the 30 fragment. The reverse primer was usedwith the mini-exon-derived forward primer (50-AGAA-CAGTTTCTGTACTATATTG-30) to generate the 50 frag-ment. This mini-exon is a sequence common to the50 end of all African trypanosomes.53 The two overlap-ping 50 and 30 PCR fragments were cloned into pGEM-Tvector (Promega) and sequenced.

Construction of an overexpressing T. vivaxIAG-NH clone

A single DNA fragment of 1 kb corresponding to theIAG-NH ORF was generated by PCR ampli®cation oftotal ss-cDNA from T. vivax. The 50 (amino-terminal)primer sequence was 50-GGGATCCGCAAAGAATGTCGTGCTG-30 and the 30 (carboxyl terminal) primer, about100 bases downstream to the TGA stop codon, was50-CCTGCAGTGGCGGGCAATAGATATG-30, with the®rst seven nucleotides being extrinsic to the nucleosidehydrolase gene. These introduce the underlined uniqueBamHI and PstI sites at the 50 and 30 ends of the ampli-®ed DNA fragment, respectively, and delete the ATGinitiation codon. The PCR product was puri®ed fromagarose and cloned into a pGEM-T vector. The clonedfragment was excised with BamHI and PstI from therecombinant plasmid and subcloned into a pQE-30 (Qia-gen) expression plasmid vector. After its start codon, thepQE-30 expression vector possesses six consecutive histi-dine codons (coding for a 6 � His-tag). The resultingrecombinant plasmid was transformed into Escherichiacoli JM109 or E. coli WK6 for overexpression.

In order to obtain a system with a removable His-tag,the sequence for a thrombin cleavage site was intro-duced into this construct between the His-tag and thenucleoside hydrolase coding sequence. This was doneusing primer-extension-overlap-PCR techniques.54 Theforward primer 50-CACGGATCCCTGGTGCCGCGCGG-CAGCATGGCAAAGAATGTCGTG-30 and reverseprimer 50-GACATTCTTTGCCATGCTGCCGCGCGGCACCAGGGATCCGTGATGGTG-30, which contain thethrombin cleavage site coding sequence (underlined),were used with a general reverse and forward primer forthe pQE-30 plasmid to generate the 30 and 50 fragments,respectively. A PCR with the general primers was usedto link both fragments together. This last PCR productwas recloned in the pQE-30 plasmid using the BamHIand HindIII restriction sites.

Expression and purification

Overnight cultures of E. coli WK6 cells,55 containingthe iagnh ORF in pQE-30 were used to inoculate 15baf¯ed ¯asks of 300 ml of Terri®c Broth56 containing50 mg/l ampicillin. These were incubated at 37 �C withshaking until the cells reached an A600 between 0.6 and1. The medium was then cooled to 28 �C and IPTG wasadded to a ®nal concentration of 0.5 mM to initiate theexpression of the recombinant protein. The next morn-ing, the cells were harvested by centrifugation. The cellpellets were resuspended in 50 mM phosphate buffer(pH 7.0), 1 M NaCl, 1 mM CaCl2 containing the proteaseinhibitors leupeptine (1 mg/ml) and 4-(2-aminoethyl)-benzenesulfonyl ¯uoride hydrochloride (AEBSF; 0.1 mg/ml). The cells were lysed via three consecutive passagesthrough a French pressure cell (French1 Pressure CellPress, SLM-Aminco1 Spectronic instruments) and thedisrupted cell suspension was centrifuged for 30 minutesat 10,000 rpm in a Beckman JA-20 rotor. The presence ofsix histidine residues at the N terminus of the recombi-nant protein allowed an easy puri®cation using Ni-NTAaf®nity chromatography (QIAexpress NI-NTA ProteinPuri®cation System, Qiagen). The intracellular fractionwas applied to a Pharmacia 10/2 HR column packedwith Ni-NTA, equilibrated with 50 mM phosphate buffer(pH 7.0), 1 M NaCl using a Pharmacia FPLC system.After extensive washing of the column, the His-taggednucleoside hydrolase was eluted with 50 mM acetate


buffer, pH 5.0 containing 1 M NaCl. The nucleosidehydrolase-containing fractions were neutralized and con-centrated using a Vivaspin concentrator (Vivascience).This fraction was then loaded onto a HiLoad 26/60Superdex-200 column (Pharmacia Biotech Inc.) equili-brated with 20 mM Tris-HCl (pH 8.0), 150 mM NaCland eluted at 1 ml/minute using the same buffer. Thepurity of the enzyme after this step was estimated bySDS-PAGE.

The expression and the ®rst step of the puri®cation ofthe nucleoside hydrolase with removable his-tag werethe same as stated above. After the Ni-NTA puri®cationstep, the His-tag was cut off by thrombin. The biotiny-lated thrombin was captured afterwards by passagethrough a streptavidin column as described in theThrombin Cleavage Capture Kit (Novagen). Theresidual, uncleaved, enzyme was removed by a secondpassage through a Ni-NTA column. Removal of the His-tag was con®rmed via Western blotting using mouseanti-his-tag antibodies (Dianova) in combination with ananti-mouse IgG-alkaline phosphatase conjugate (Sigma)for detection.

Molecular mass of the IAG-NH

The elution position of active IAG-NH relative to pro-teins of known molecular mass (Gel Filtration Chroma-tography Standard, Bio Rad) was determined by 10/30Superdex-200 HR gel chromatography. Blue dextran2000 was used to determine the void volume (V0).The column was eluted with 50 mM Hepes (pH 7.2),0.3 M KCl at a ¯ow-rate of 1 ml/minute. Thedistribution coef®cients Kd:

Kd � Ve ÿ V0

Vt ÿ V0�1�

where Ve is elution volume, V0 is void volume and Vt istotal volume of the column, of the standard proteins andthe IAG-NH were determined graphically from the chro-matogram, and plotted against the logarithm of the mol-ecular mass of the standard proteins. This standardcurve was used to calculate the apparent molecular massof the IAG-NH.

Kinetic analysis of substrates and inhibitors

The kinetic properties of the nucleoside hydrolase forthe substrates inosine, guanosine and adenosine weredetermined spectrophotometrically using the differencein absorption between the nucleoside and the purinebase.23 All measurements were carried out at 35 �Cin 50 mM phosphate buffer (pH 7.2). The �e values(mMÿ1 cmÿ1) used were: inosine, ÿ0.92 at 280 nm; ade-nosine, ÿ1.4 at 276 nm; guanosine, 0.16 at 308 nm. Theexperimental data were ®tted to the Michaelis-Mentenequation using the program Origin (Microcal). For thesubstrate guanosine, kcat/KM was determined from asingle progress curve. The conversion of a subsaturatingconcentration (5 � 10ÿ7 M; S < KM) of guanosine toguanine was followed on a C18 HPLC column(100 � 4.6 mm ODS HYPERSIL RP-C18, 5 mm). Substrateand product were eluted with a linear gradient of aceto-nitrile in water, plus 0.1 % (w/v) trichloroacetic acid andmonitored spectrophotometrically at 253 nm. The datawere ®tted to the equation:

S � S0 � eÿ kcat

KM�E0�t

� ��2�

where: S is substrate concentration, t is time, S0 is sub-strate concentration at t0 and E0 is total enzyme concen-tration. The hydrolysis of cytidine and uridine wasmonitored by the release of reducing sugar at 35 �C in50 mM phosphate buffer (pH 7.2).23 Hydrolysis of p-nitrophenyl b-D-ribofuranoside was followed at 35 �C inphosphate buffer (pH 7.5) by release of the p-nitropheno-late anion, which has strong absorbance at 400 nm withan extinction coef®cient of 15 mMÿ1 cmÿ1 under theassay conditions. All kinetic parameters were calculatedper active site, making them independent of the multi-merisation state of the enzyme.

The inhibition constants for 3-deaza-adenosine and7-deaza-adenosine were determined by measuring therate of hydrolysis of a ®xed concentration of p-nitro-phenyl b-D-ribofuranoside at variable inhibitor concen-trations. The results were ®tted to the equation:

v � kcat � E0 � S

S� KM � 1� I

KI

� � �3�

where S is substrate concentration, I is inhibitor concen-tration, E0 is total enzyme concentration, kcat and KM aresteady-state constants for substrate and KI is dissociationconstant for the enzyme-inhibitor complex.

This analysis requires the assumption that the inhibi-tor competes with the substrate for the catalytic site.24

The validity of this assumption is supported by the highstructural similarity between both inhibitors and thenatural substrates (Figure 2) and by the crystal structuresreported here.

Crystallization and data collection

All IAG-NH crystals were obtained using the hangingdrop vapor diffusion method: 5 ml of protein solution (at8 mg/ml) was mixed with 5 ml of reservoir solution con-taining 100 mM Tris (pH 8.5) with 1.6 M ammonium sul-fate as precipitant. Crystals with average dimensions of0.3 mm � 0.3 mm � 0.05 mm grew after one to twoweeks. The space group and the cell parameters of thedifferent crystals used are given in Table 3. The crystalused for the high-resolution native dataset was trans-ferred to a cryo solution, which contained 100 mM Tris(pH 8.5), 1.6 M ammonium sulfate and 35 % (w/v) gly-cerol. Data were collected at 100 K on beamline BW7B ofthe EMBL outstation at DESY (Hamburg, Germany)using a MAR345 (Hamburg, Germany) image plate.

A samarium derivative was obtained by soaking acrystal for two days in mother liquor containing 5 mMsamarium(III) acetate hydrate (Sm(O2C2H3)3). The crystalwas transferred to the same cryo solution as mentionedabove and a three wavelength MAD57 was collected on asingle crystal at 100 K on beamline BW7A (EMBL, Ham-burg) using a MAR (Hamburg, Germany) CCD detector.The presence of samarium was con®rmed with a ¯uor-escence scan and the wavelengths were set to 0.9497 AÊ

(remote, l1), 1.6952 AÊ (peak, l2) and 1.6961 AÊ (in¯exionpoint, l3).

The complex with the substrate analogue 3-deaza-ade-nosine was obtained by soaking a crystal for two days in100 mM Hepes (pH 7.5) containing 40 % (w/v) PEG6000 and 3 mM 3-deaza-adenosine. This solution wasalso used as cryo solution for the data collection.


Data was collected at 100 K on beamline BW7A (EMBL,Hamburg).

Data processing

The high-resolution native data and the MAD datawere indexed and integrated using DENZO/HKL,58 andscaled and merged with SCALA and other routines fromthe CCP459 program package. The data for the 3-deaza-adenosine complex were processed and scaled with theHKL package. Intensities were converted to structurefactors with TRUNCATE. Relevant statistics for the datasets are summarized in Table 3.

Structure determination

The high-resolution native data and the differentMAD data were scaled relative to the l1 data set withSCALEIT (CCP4). Difference and anomalous differencePatterson maps were calculated. Heavy-atom positionswere localized using PEAKMAX (CCP4) and con®rmedwith SHELX (60). Two samarium positions were found.These sites were re®ned and used for phasing inMLPHARE (CCP4). Five additional Sm positions wereidenti®ed in difference Fourier maps. Phases wereimproved with the density modi®cation procedure asimplemented in DM,61 using the solvent-¯attening andhistogram-matching options. Phases were extended from2.3 to 1.6 AÊ . The density modi®cation procedure washampered by an overestimate of the ®gures of merit(FOMs) by MLPHARE. Therefore, the FOMs were scaledby 0.7 manually before running DM. Visual inspection ofthe modi®ed electron density map did not allowstraightforward building of the protein model. However,part of the structure could be built automatically withthe warpNtrace mode in ARP/WARP.62 The warpmodel represented mainly one of the two protein chainsof the asymmetric unit. To speed up the process, thesecond protein chain was located using a molecularreplacement step using AMoRe.63 The NCS operatorrelating the two protein chains was identi®ed, and amask was calculated with NCSMASK (CCP4). At thispoint, the density modi®cation step was repeated, com-bining the solvent-¯attening and histogram-matchingprocedures with NCS averaging. ARP/WARP automaticstructure building was resumed using the improved elec-tron density map. Structure building was ®nished manu-ally using TURBO.64

Structure refinement

The warp model was subjected to the simulatedannealing procedure as implemented in CNS, using thenative data between 30.0 and 1.6 AÊ .65-67 The resultingmodel was corrected manually in TURBO. After severalcycles of positional and temperature-factor re®nementcombined with manual corrections, solvent moleculesand alternative conformations were included in themodel. Structure re®nement was considered completeafter positional and thermal parameters had converged,and the difference density was without interpretable fea-tures. The ®nal model includes 424 water molecules,eight glycerol molecules and two calcium ions. Residues245 to 256 are missing from both protein chains in theasymmetric unit. The ®nal model has an R-factor of18.6 % and an Rfree of 21.0 %.

The crystal structure of the 3-deaza-adenosine com-plex is isomorphous to the unliganded form. Therefore,

re®nement was started from the native structure (with-out solvent), taking care to use the same set of re¯ectionsfor cross-validation. After an initial round of simulatedannealing using CNS, the position of 3-deaza-adenosinecould be clearly identi®ed in the 2Fo ÿ Fc and Fo ÿ Fc

difference maps. The ®nal model of this structure con-tains one 3-deaza-adenosine molecule per protein chainand 396 water molecules in the asymmetric unit. Resi-dues 248 to 252 are missing from both protein chains.This model has an R-factor of 19.1 % and an Rfree of25.4 %. Re®nement of the 3-deaza-adenosine componentrequired a custom restraints ®le. A copy of this ®le willbe submitted to the pdb databank complementing thecoordinate ®le. Both models were checked with the pro-gram PROCHECK.68 Re®nement statistics are summar-ized in Table 4.

Accession numbers

The nucleotide sequence of the T vivax IAG-NH ORFhas been submitted to the GenbankTM/EBI with acces-sion number AF311701. The coordinates have been sub-mitted to the RSCB Protein Data Bank with accessionnumber 1HOZ for the unliganded structure and 1HP0for the 3-deaza-adenosine complexed structure.

Acknowledgements

This work was supported by the Vlaams Interuniversi-tair Instituut voor Biotechnologie and by the NationaalFonds voor Wetenschappelijk Onderzoek - Vlaanderen.W.V. is a recipient of a grant from the FWO-Vlaanderen.The authors thank Maia De Kerpel for excellent technicalassistance. We are also grateful to Yves Geunes for com-puter support. The substrate analogue p-nitrophenylb-D-ribofuranoside was a generous gift from Dr VernSchramm from the Albert Einstein College of Medicine,New York. We acknowledge the use of beamlines BW7Aand BW7B at the EMBL-Hamburg. Travel to and accom-modation in Hamburg was supported by the EU TMR/LSF, grant number ERBFMG3CT 980134.

References

1. Molyneux, D. H. (1997). Current public health statusof the trypanosomiases and leishmaniases. In Trypa-nosomiasis and Leishmaniasis (Hide, G., Mottram, J. C.,Coombs, G. H. & Holmes, P. H., eds), pp. 39-50,CAB INTERNATIONAL, Oxon.

2. Barrett, M. P. (1999). The fall and rise of sleepingsickness. Lancet, 353, 1113-1114.

3. Smith, D. H., Pepin, J. & Stich, A. H. R. (1998).Human African trypanosomiasis: an emerging pub-lic health crisis. Brit. Med. Bull. 54, 341-355.

4. Ross, C. A. & Sutherland, D. V. (1997). Drug resist-ance in trypanosomatids. In Trypanosomiasis andLeishmaniasis (Hide, G., Mottram, J. C., Coombs,G. H. & Holmes, P. H., eds), pp. 259-270, CABINTERNATIONAL, Oxon.

5. Denise, H., Matthews, K., LindergaÊ rd, G., Croft, S.& Barrett, M. P. (1999). Trypanosomiasis andleishmaniasis: between the idea and the reality ofcontrol. Parasitol. Today, 15, 43-45.

6. Fairlamb, A. H. (1989). Novel biochemical pathwaysin parasitic protozoa. Parasitology, 99, S93-S112.


7. Barrett, M. P., Mottram, J. C. & Coombs, G. H.(1999). Recent advances in identifying and validat-ing drug targets in trypanosomes and leishmanias.Trends Microbiol. 7, 82-88.

8. Berens, R. L., Krug, E. D. & Marr, J. J. (1995). Purineand pyrimidine metabolism. In Biochemistry and Mol-ecular Biology of Parasites (Marr, J. J. & MuÈ ller, M.,eds), pp. 89-117, Academic Press.

9. Ogbunude, P. O. J., Ikediobi, C. O. & Ukoha, A. I.(1985). Adenosine cycle in African trypanosomes.Ann. Trop. Med. Parasitol. 79, 7-11.

10. Hammond, D. J. & Gutteridge, W. E. (1984). Purineand pyrimidine metabolism in the Trypanosomatidae.Mol. Biochem. Parasitol. 13, 243-261.

11. Davies, M. J., Ross, A. M. & Gutteridge, W. E.(1983). The enzymes of purine salvage in Trypano-soma cruzi, Trypanosoma brucei and Leishmania mexi-cana. Parasitology, 87, 211-217.

12. James, D. M. & Born, G. V. R. (1980). Uptake of pur-ine bases and nucleosides in African trypanosomes.Parasitology, 81, 383-393.

13. de Koning, H. P., Watson, C. J., Sutcliffe, L. &Jarvis, S. M. (2000). Differential regulation of nucleo-side and nucleobase transporters in Crithidia fascicu-lata and Trypanosoma brucei brucei. Mol. Biochem.Parasitol. 106, 93-107.

14. Craig, S. P. & Eakin, A. E. (1997). Purine salvageenzymes of parasites as targets for structure-basedinhibitor design. Parasitol. Today, 13, 238-241.

15. Li, C. M., Tyler, P. C., Furneaux, R. H., Kicska, G.,Xu, Y., Grubmeyer, C., Girvin, M. E. & Schramm,V. L. (1999). Transition-state analogs as inhibitors ofhuman and malarial hypoxanthine-guanine phos-phoribosyltransferases. Nature Struct. Biol. 6, 582-587.

16. Aronov, A. M., Munagala, N. R., de Montellano,P. R. O., Kuntz, I. D. & Wang, C. C. (2000). Rationaldesign of selective submicromolar inhibitors ofTrichichomonas foetus hypoxanthine-guanine-xanthinephosphoribosyltransferase. Biochemistry, 39, 4684-4691.

17. Schmidt, G., Walter, R. D. & Konigk, E. (1975).A purine nucleoside hydrolase from Trypanosomagambiense, puri®cation and properties. Tropenmed.Parasitol. 26, 19-26.

18. Miller, R. L., Sabourin, C. L. K., Krenitsky, T. A.,Berens, R. L. & Marr, J. J. (1984). Nucleoside hydro-lases from Trypanosoma cruzi. J. Biol. Chem. 259,5073-5077.

19. Koszalka, G. W. & Krenitsky, T. A. (1979). Nucleo-sidases from Leishmania donovani. Pyrimidine ribo-nucleosidase, purine ribonucleosidase, and a novelpurine 20-deoxyribonucleosidase. J. Biol. Chem. 254,8185-8193.

20. Dewey, V. C. & Kidder, G. W. (1973). Partial puri®-cation and properties of a nucleoside hydrolasefrom Crithidia. Arch. Biochem. Biophys. 157, 380-387.

21. Parkin, D. W., Horenstein, B. A., Abdulah, D. R.,EstupinÄ aÂn, B. & Schramm, V. L. (1991). Nucleosidehydrolase from Crithidia fasciculata. Metabolic role,puri®cation, speci®city, and kinetic mechanism.J. Biol. Chem. 266, 20658-20665.

22. EstupinÄaÂn, B. & Schramm, V. L. (1994). Guanosine-Inosine-preferring nucleoside N-glycohydrolase fromCrithidia fasciculata. J. Biol. Chem. 269, 23068-23073.

23. Parkin, D. W. (1996). Purine-speci®c nucleosideN-ribohydrolase from Trypanosoma brucei brucei.Puri®cation, speci®city, and kinetic mechanism.J. Biol. Chem. 271, 21713-21719.

24. Shi, W., Schramm, V. L. & Almo, S. C. (1999).Nucleoside hydrolase from Leishmania major. Clon-ing, expression, catalytic properties, transition stateinhibitors, and the 2.5-AÊ crystal structure. J. Biol.Chem. 274, 21114-21120.

25. Horenstein, B. A., Parkin, D. W., EstupinÄ aÂn, B. &Schramm, V. L. (1991). Transition-state analysis ofnucleoside hydrolase from Crithidia fasciculata.Biochemistry, 30, 10788-10795.

26. Horenstein, B. A. & Schramm, V. L. (1993). Elec-tronic nature of the transition state for nucleosidehydrolase. A blueprint for inhibitor design. Biochem-istry, 32, 7089-7097.

27. Gopaul, D. N., Meyer, S. L., Degano, M., Sacchettini,J. C. & Schramm, V. L. (1996). Inosine-uridinenucleoside hydrolase from Crithidia fasciculata.Genetic characterization, crystallization, and identi®-cation of histidine 241 as a catalytic site residue.Biochemistry, 35, 5963-5970.

28. Degano, M., Gopaul, D. N., Scapin, G., Schramm,V. L. & Sacchetini, J. C. (1996). Three-dimensionalstructure of the inosine-uridine nucleoside N-ribo-hydrolase from Crithidia fasciculata. Biochemistry, 35,5971-5980.

29. Degano, M., Almo, S. C., Sacchetini, J. C. &Schramm, V. L. (1998). Trypanosomal nucleosidehydrolase. A novel mechanism from the structurewith a transition-state inhibitor. Biochemistry, 37,6277-6285.

30. Parkin, D. W., Limberg, G., Tyler, P. C., Furneaux,R. H., Chen, X.-Y. & Schramm, V. L. (1997).Isozyme-speci®c transition state inhibitors for thetrypanosomal nucleoside hydrolases. Biochemistry,36, 3528-3534.

31. Gardiner, P. R. & Wilson, A. J. (1987). Trypanosoma(Dutonella) vivax. Parasitol. Today, 2, 255-257.

32. PelleÂ, R., Schramm, V. L. & Parkin, D. W. (1998).Molecular cloning and expression of a purine-speci®c N-ribohydrolase from Trypanosoma bruceibrucei. Sequence, expression, and molecular analysis.J. Biol. Chem. 273, 2118-2126.

33. Mazzella, L. J., Parkin, D. W., Tyler, P. C., Furneaux,R. H. & Schramm, V. L. (1996). Mechanistic diag-noses of N-ribohydrolases and purine nucleosidephosphorylase. J. Am. Chem. Soc. 118, 2111-2112.

34. Rossmann, M. G., Liljas, A., BraÈndeÂn, C.-I. &Banaszak, L. J. (1975). Evolutionary and structuralrelationship among dehydrogenases. In The Enzymes(Boyer, P. D., ed.), vol. 11, part A edit., pp. 61-102,Academic Press, New York.

35. Bellamacina, C. R. (1996). The nicotinamide dinu-cleotide binding motif: a comparison of nucleotidebinding proteins. FASEB J. 10, 1257-1269.

36. Gelbin, A., Schneider, B., Clowney, L., Hsieh, S.-U.,Olson, W. K. & Berman, H. M. (1996). Geometricparameters in nucleic acids: sugar and phosphateconstituents. J. Am. Chem. Soc. 118, 519-529.

37. Altona, C. & Sundaralingam, M. (1972). Confor-mational analysis of the sugar ring in nucleosidesand nucleotides. J. Am. Chem. Soc. 94, 8205-8212.

38. Olson, W. K. & Sussman, J. L. (1982). How ¯exibleis the furanose ring? A comparison of experimentaland theoretical studies. J. Am. Chem. Soc. 104, 270-278.

39. IUPAC-IUB Joint Commission on BiochemicalNomenclature (1983). Abbreviations and symbolsfor the description of conformations of polynucleo-tide chains. Eur. J. Biochem. 131, 9-15.


40. Saenger, W. (1984). Principles of Nucleic Acid Struc-ture, Springer-Verlag, New York.

41. Haschemeyer, A. E. V. & Rich, A. (1967). Nucleosideconformation: an analysis of steric barriers torotation about the glycosidic bond. J. Mol. Biol. 27,369-384.

42. Berthod, H. & Pullman, B. (1973). Complementarystudies on the rigidity-¯exibility of nucleotides.FEBS Letters, 33, 147-150.

43. Zoltewicz, J. A., Clark, D. F., Sharpless, T. W. &Grahe, G. (1970). Kinetics and mechanism of theacid-catalyzed hydrolysis of some purine nucleo-sides. J. Am. Chem. Soc. 92, 1741-1750.

44. Schramm, V. L. (1998). Enzymatic transition statesand transition state analog design. Annu. Rev.Biochem. 67, 693-720.

45. Burley, S. K. & Petsko, G. A. (1988). Weakly polarinteractions in proteins. Advan. Protein Chem. 39,125-189.

46. McGaughey, G. B., GagneÂ, M. & RappeÂ, A. K.(1998). p-Stacking interactions. Alive and well inproteins. J. Biol. Chem. 273, 15458-15463.

47. Parkin, D. W. & Schramm, V. L. (1995). Bindingmodes for substrate and a proposed transition-stateanalogue of protozoan nucleoside hydrolase.Biochemistry, 34, 13961-13966.

48. Fersht, A. R., Knill-Jones, J. W., Bedouelle, H. &Winter, G. (1988). Reconstruction by site-directedmutagenesis of the transition state of tyrosine by thetyrosyl-tRNA synthetase: a mobile loop envelopesthe transition state in an induced-®t mechanism.Biochemistry, 27, 1581-1587.

49. Wolfenden, R. (1999). Conformational aspects ofinhibitor design: enzyme-substrate interactions inthe transition state. Bioorg. Med. Chem. 7, 647-652.

50. DeWolf, W. E. J., Fullin, F. A. & Schramm, V. L.(1979). The catalytic site of AMP nucleosidase.Substrate speci®city and pH effects with AMP andformycin 50-PO4. J. Biol. Chem. 254, 10868-10875.

51. Schramm, V. L., Horenstein, B. A. & Kline, P. C.(1994). Transition state analysis and inhibitor designfor enzymatic reactions. J. Biol. Chem. 269, 18259-18262.

52. Miles, R. W., Tyler, P. C., Evans, G. B., Furneaux,R. H., Parkin, D. W. & Schramm, V. L. (1999).Iminoribitol transition state analogue inhibitors ofprotozoan nucleoside hydrolases. Biochemistry, 38,13147-13154.

53. Sutton, R. E. & Boothroyd, J. C. (1988). The cap ofboth miniexon-derived RNA and mRNA of trypano-somes is 7-methylguanosine. Mol. Cell Biol. 8, 494-496.

54. Ho, S. N., Hunt, H. D., Horton, R. M., Pullen, J. K.& Pease, L. R. (1989). Site-directed mutagenesis byoverlap extension using the polymerase chain reac-tion. Gene, 77, 51-59.

55. Zell, R. & Fritz, H.-J. (1987). DNA mismatch repairin Escherichia coli counteracting the hydrolyticdeamination of 5-methyl-cytosine residues. EMBO J.6, 1809-1815.

56. Tartof, K. D. & Hobbs, C. A. (1987). Improvedmedia for growing plasmid and cosmid clones.Bethesda Res. Lab. Focus, 9, 12.

57. Hendrickson, W. A. (1991). Determination of macro-molecular structures from anomalous diffraction ofsynchrotron radiation. Science, 254, 51-58.

58. Otwinowski, Z. & Minor, W. (1997). Processing ofX-ray diffraction data collected in oscillation mode.In Methods in Enzymology (Carter, C. W. & Sweet,R. M., eds), vol. 276, part A, pp. 307-326, AcademicPress, New York.

59. Collaborative Computational Project Number 4(1994). The CCP4 suite: programs for protein crystal-lography. Acta Crystallog. sect. D, 50, 760-763.

60. Sheldrick, G. M. (1997). Patterson superposition andab initio phasing. In Methods in Enzymology (Carter,C. W. & Sweet, R. M., eds), vol. 276, part A edit.,pp. 628-641, Academic Press, New York.

61. Cowtan, K. (1994). DM, an automated procedure forphase improvement by density modi®cation. JointCCP4 ESF-EACBM Newsletter Protein Crystallog. 31,34-38.

62. Perrakis, A., Morris, R. M. & Lamzin, V. S. (1999).Automated protein building combined with iterativestructure re®nement. Nature Struct. Biol. 6, 458-463.

63. Navaza, J. (1994). AMoRe: an automated packagefor molecular replacement. Acta Crystallog. sect. A,50, 157-163.

64. Roussel, A., Fontecilla-Camps, J. C. & Cambillau, C.(1990). TURBO-FRODO: a new program for proteincrystallography and modeling. In XV IUCr CongressAbstracts, pp. 66-67, Bordeaux, France.

65. BruÈ nger, A. T., Krukowski, A. & Erickson, J. W.(1990). Slow-cooling protocols for crystallographicre®nement by simulated annealing. Acta Crystallog.sect. A, 46, 585-593.

66. BruÈ nger, A. T., Adams, P. D., Clore, G. M., DeLano,W. L., Gros, P., Grose-Kunstleve, R. W., Jiang, J.-S.,Kuszewski, J., Nilges, N., Pannu, N. S., Read, R. J.,Rice, L. M., Simonson, T. & Warren, G. L. (1998).Crystallography and NMR system (CNS): a newsoftware system for macromolecular structure deter-mination. Acta Crystallog. sect. D, 54, 905-921.

67. BruÈ nger, A. T. (1992). Free R-value: a novel statisti-cal quantity for assessing the accuracy of crystalstructures. Nature, 355, 472-475.

68. Laskowski, R. A., MacArthur, M. W., Moss, D. S. &Thornton, J. M. (1993). PROCHECK: a program tocheck the stereochemical quality of proteins. J. Appl.Crystallog. 26, 283-291.

69. Lawrence, M. C. & Bourke, P. (2000). CONSCRIPT:a program for generating electron density isosur-faces for presentation in protein crystallography.J. Appl. Crystallog. 33, 990-991.

70. Kraulis, P. J. (1991). MOLSCRIPT: a program to pro-duce both detailed and schematic plots of proteinstructures. J. Appl. Crystallog. 24, 946-950.

71. Nicholls, A., Sharp, K. A. & Honig, B. (1991). Proteinfolding and association: insights from the interfacialand thermodynamic properties of hydrocarbons.Proteins: Struct. Funct. Genet. 11, 281-296.

Edited by R. Huber

(Received 13 December 2000; received in revised form 20 February 2001; accepted 5 February 2001)

Documents

Structure and Function of a Novel Purine Specific Nucleoside ...of the IAG-NH, absent from the IU-NH, imposes the purine speciﬁcity. Asp10 is the proposed general base in the reaction