Báo cáo khoa học: Leishmania donovani phosphofructokinase Gene characterization, biochemical properties and structure-modelling studies doc

PFKs can be divided into ATP-dependent and PPi-dependent enzymes; the former use ATP as phospho donor in a reaction that is essentiallyirreversible under physiological conditions, wherea

Leishmania donovani phosphofructokinase

Gene characterization, biochemical properties and structure-modelling studies

Claudia Lo´pez1,2, Nathalie Chevalier2, Ve´ronique Hannaert2, Daniel J Rigden3, Paul A M Michels2 and Jose Luis Ramirez1,4

1

Instituto de Biologı´a Experimental, Universidad Central de Venezuela, Caracas, Venezuela;2Research Unit for Tropical Diseases, Christian de Duve Institute of Cellular Pathology and Laboratory of Biochemistry, Universite´ Catholique de Louvain, Brussels, Belgium;3CENARGEN/EMBRAPA, Brası´lia, Brazil;4Instituto de Estudios Avanzados-Ministerio de Ciencia y Tecnologı´a, Caracas, Venezuela

The characterization of the gene encoding Leishmania

donovaniphosphofructokinase (PFK) and the biochemical

properties of the expressed enzyme are reported L donovani

has a single PFK gene copyper haploid genome that encodes

a polypeptide with a deduced molecular mass of 53 988 and

a pI of 9.26 The predicted amino acid sequence contains a

C-terminal tripeptide that conforms to an established signal

for glycosome targeting L donovani PFK showed most

sequence similarityto inorganic pyrophosphate (PPi

)-dependent PFKs, despite being ATP-)-dependent It thereby

resembles PFKs from other Kinetoplastida such as

Trypanosoma brucei, Trypanoplasma borreli (characterized

in this study), and a PFK found in Entamoeba histolytica It

exhibited hyperbolic kinetics with respect to ATP whereas

the binding of the other substrate, fructose 6-phosphate, showed slight positive cooperativity PPi, even at high con-centrations, did not have anyeffect AMP acted as an acti-vator of PFK, shifting its kinetics for fructose 6-phosphate from slightlysigmoid to hyperbolic, and increasing consid-erablythe affinityfor this substrate, whereas GDP did not have anyeffect Modelling studies and site-directed muta-genesis were employed to shed light on the structural basis for the AMP effector specificityand on ATP/PPispecificity among PFKs

Keywords: phosphofructokinase; Kinetoplastida; allosteric regulation; mutagenesis; structure modelling

Glycolysis is a central metabolic pathway in all organisms

A keyenzyme of this pathwayis 6-phospho-1-fructokinase

(PFK) or ATP:D-fructose-6-phosphate

1-phosphotransfer-ase The activityof this enzyme is, in almost all organisms,

regulated by multiple mechanisms Whereas most glycolytic

enzymes have been remarkably conserved during evolution,

considerable sequence variabilityis found among PFKs of

different taxonomic groups [1]

In Kinetoplastida, a taxonomic order of protozoan

organisms that includes important pathogens (species of

Trypanosoma, Leishmania, Phytomonas) to man, animals

and plants, the first seven enzymes of the glycolytic pathway

are confined to an organelle called the glycosome [2,3] PFKs

can be divided into ATP-dependent and PPi-dependent enzymes; the former use ATP as phospho donor in a reaction that is essentiallyirreversible under physiological conditions, whereas the latter use PPiin a reversible reaction that can be near equilibrium in vivo [4] We have previouslyreported that Trypanosoma bruceiPFK, despite being an ATP-dependent enzyme, has an amino-acid sequence typical of PPi-PFKs [5] Furthermore, the activityof the T brucei enzyme appears onlyregulated to a limited extent: effectors that modulate PFK activityin other organisms (vertebrates, yeast) such as ATP, citrate, fructose 2,6-bisphosphate, ADP and Pi, have

no effect on the Trypanosoma enzyme [6,7] Only activation byAMP was observed

We hypothesized that an ancestor of the trypanosomes must have possessed a PPi-dependent PFK that changed its specificityfor phospho donor from PPi to ATP during evolution [5] The manystructural differences between the active site of the two classes of PFKs, and the striking differences in ligand-binding properties between the human and parasite enzymes suggest great potential for structure-based design of drugs [5,8]

For comparative purposes we decided to studythe PFK of Leishmania donovani another kinetoplastid organism that, contraryto the bloodstream-dwelling

T brucei, lives as an intracellular parasite in humans The results of this work are presented in this paper, together with a report of the cloning and analysis of the PFK gene of the fish parasite Trypanoplasma borreli (a representative of the Bodonina, a different sub-order of the Kinetoplastida)

Correspondence to J R Ramirez, Instituto de Biologı´a Experimental –

UCV, Calle Suapure, Colinas de Bello Monte, Caracas, Venezuela,

Caracas 1041-A, Venezuela.

Fax: + 58 221 962 1120, Tel.: + 58 221 751 0111,

E-mail: jramirez@reacciun.ve

Abbreviations: PFK, 6-phosphofructokinase;

ATP-PFK, ATP-dependent phosphofructokinase; PP i -PFK, PP i -dependent

phosphofructokinase; PTS, peroxisome-targeting signal.

Enzymes: 6-phosphofructokinase (EC 2.7.1.11); pyrophosphate–

fructose-6-phosphate 1-phosphotransferase (EC 2.7.1.90).

Note: The novel nucleotide sequences reported in this paper have been

deposited in the EMBL, GenBank and DDBJ databases and are

available under the accession numbers AY029213 (L donovani PFK)

and AJ310928 (T borreli PFK).

(Received 11 June 2002, accepted 1 July2002)

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

Cloning and characterization of theL donovani

andT borreli PFK genes

A genomic libraryof L donovani (kindlydonated by

T deVos, Seattle Biomedical Research Institute, WA,

USA) constructed in Supercos7 vector was screened with

a 500 bp fragment of the PFK gene of T brucei as a probe

[5] One positive colonywas chosen and the bacteria were

grown for purification of the recombinant cosmid DNA

Southern analysis using the T brucei probe showed

hybridization with a 500 bp Sau3AI fragment and a 12 kb

NotI fragment The Sau3AI fragment was cloned in

pBluescript II KS– (Stratagene) and subsequently

sequenced The DNA sequence obtained had 55% identity

with the corresponding portion of the T brucei PFK gene

This sequence was used as a probe, and to design primers

for further experiments The NotI fragment was gel purified

and digested with EcoRI, and a 6.5 kb EcoRI fragment

recognized bythe Sau3AI probe was subcloned in

pBlue-script II KS– (PFK6.5-PBSKS) Finally, from the

recom-binant PFK6.5-PBSKS, a 2.2 kb EcoRI–PvuII fragment

containing the entire L donovani PFK gene (Fig 1) was

subcloned in pBluescript II KS– (PFK2.2-PBSKS) and

sequenced

DNA sequences of both strands of recombinant plasmid

inserts were determined byusing the T7 DNA polymerase

kit (Amersham Pharmacia Biotech) and [35S]dATP (NEN

Life Science Products), or with the Thermo Sequenase

fluorescent labelled primer cycle sequencing kit (Amersham

Pharmacia Biotech), and LI-COR automated DNA

sequence equipment

A genomic libraryof T borreli constructed in kGEM11

(Promega) [9] was screened with a probe consisting of the

entire T brucei PFK gene [5] The probe was hybridized

under moderatelystringent conditions: 3· NaCl/Cit/0.1%

SDS, in the presence of 10% dextran sulphate, at 60C

(1· NaCl/Cit ¼ 150 mM NaCl/15 mM sodium citrate,

pH 7.0) Post-hybridization washes were carried out at

60C for 2 h with 5 · NaCl/Cit/0.1% SDS, followed for

1 h with 3· NaCl/Cit/0.1% SDS Ten positive

recom-binants were purified and rescreened High-titre phage

lysates were prepared and DNA was purified from

phages as described previously[10] From one

recombi-nant phage a 4 kb EcoRI fragment recognized bythe

T brucei probe was subcloned in plasmid pZErO-2

(Invitrogen) and sequenced

A multiple alignment of the amino-acid sequences of

ATP-PFKs and PPi-PFKs was made as described

previously[5]

Structural analysis

MODELLER [11] was used to construct a model of

L donovani PFK based on the known structure of Escherichia coli PFK (PDB code 4pfk [12]) This was the most suitable template, given our interest in both catalytic and regulatory sites, among the various available structures of E coli and Bacillus stearothermophilus PFK,

as this 4pfk structure contains fructose 6-phosphate and ADP in the active site, and ADP in the effector site The

E coli and L donovani enzymes share 23% sequence identityoverall, although functional considerations have led to much greater conservation of the catalytic site and effector site Without these sites, the degree of sequence identityof  20% leads to uncertaintyin alignment of some regions However, the conservation of the catalytic and effector sites allows reliable alignment and corres-pondinglyaccurate modelling of the Leishmania enzyme in these regions of particular importance The program O

[13] was used for visualizing structures and for its library

of commonlyobserved, rotameric side-chain conforma-tions [14] PROCHECK [15] was used for stereochemical analysis of models and for identifying the most likely position of an important one residue insertion, in the

L donovani enzyme relative to that from E coli, at the effector site

Expression and purification of recombinant

L donovani PFK The following specific primers were synthesized to amplify the gene byPCR: (1) 5¢-CGAATCTCCATATGGAGA CTCG-3¢, containing a NdeI site (underlined) adjacent to the 5¢ end of the coding region; (2) 5¢-TAGGATCCTTACACCTTAGACGCCAG-3¢, comple-mentary to a 3¢ noncoding region followed bya BamHI site (underlined) The total volume of the amplification mixture was 100 lL containing 20 ng of DNA, 0.4 lMof each primer, 4 mM MgSO4, 200 lM each of the four deoxynucleotides, and 1 unit of Vent DNA polymerase with the corresponding 1· ThermoPol Reaction Buffer (New England Biolabs) PCR was performed in a Hybaid Thermal Reactor (Hybaid, UK) using the following program: 5 min 95C; 30 cycles consisting of 1 min denaturation at 95C, 45 s annealing at 65 C and 1.5 min extension at 72C; and a final 5 min incubation

at 72C The amplified gene was purified and ligated to the pCR2.1-TOPO vector (Invitrogen) The resulting recombin-ant plasmid (PFKLd-TOPO) was used to transform E coli strain XL-1 Blue, and the sequence of the insert was checked

A bacteriophage T7 RNA polymerase system [16] was used to express L donovani PFK in E coli The PFK gene was excised from the PFKLd-TOPO recombinant plasmid and spliced into expression vector pET28b using the NdeI and BamHI sites The new recombinant plasmid named pLdPFK directs, under the control of the T7 promoter, the production of a fusion protein bearing a N-terminal extension of 20 residues including a (His)6-tag

E coli strain BL21(DE3)pLysS transformed with pLdPFK was grown at 30C in 50 mL Luria–Bertani medium plus 1 sorbitol and 2.5 m betaine [17]

Fig 1 Restriction map of recombinant plasmid PFK6.5-pBSKS The

hashed box marks the open-reading frame of the L donovani PFK

gene The ATG of the initiator methionine is indicated T7 and T3

indicate the orientation of the insert with respect to the promoter

sequences of the pBSKS vector The 2.2 kb EcoRI–PvuII fragment

was used as hybridization probe and for sequence analysis.

supplemented with 30 lgÆmL)1 kanamycin and

25 lgÆmL)1 chloramphenicol Expression was induced at

an A600of 0.6–0.8 bythe addition of 1 mMisopropyl

thio-b-D-galactoside and growth was continued overnight Cells

were collected bycentrifugation (12 000 g, 10 min at

4C) The cell pellet was resuspended in 20 mL of lysis

buffer (50 mM triethanolamine/HCl, pH 8.0, 300 mM

NaCl, 200 mM KCl, 1 mM KH2PO4, 5 mM MgCl2, 10%

glycerol, 0.1 mM fructose 6-phosphate, 0.3 mM glucose

6-phosphate and the protease inhibitors E64, leupeptin

and pepstatin, each at a concentration of 1 lM) Cells

were lysed by two passages through a SML-Aminco

French pressure cell (SML Instruments, USA) at 90 MPa

Nucleic acids were eliminated first byincubation with 500

units Benzonase (Merck, Germany) for 15 min at 37C,

and then with 10 mg of protamine sulphate for 15 min at

room temperature The lysate was centrifuged (20 000 g

15 min at 4C) and the supernatant used to further purify

the expressed enzyme using immobilized metal affinity

chromatography(TALON resin, Clontech, USA)

exploit-ing the (His)6-tag at the N-terminus of the PFK The

charged resin was washed with lysis buffer plus 10 mM

imidazole The enzyme was then eluted with 100 mM

imidazole in lysis buffer One millilter fractions were

collected to measure enzyme activity and to determine the

protein profile bySDS/PAGE

Site-directed mutagenesis of the L donovani PFK gene

was performed byPCR techniques as described by

Mikaelian & Sergeant [18] and using Vent DNA polymerase

The Leishmania PFK Lys224 codon AAG was changed into

the Glycodon GGG The mutated protein was expressed

and purified according to the same protocols as the

wild-type enzyme

Enzyme assays and kinetic analysis

The activityof PFK was determined bymeasuring the

decrease of NADH absorbance at 340 nm using a Beckman

DU7 spectrophotometer To follow PFK during

purifica-tion, a standard enzymatic assay was performed at 25C in

a 1 mL reaction mixture containing: 100 mM

triethanol-amine/HCl buffer, pH 8.0, 2.5 mM MgSO4, 10 mM KCl,

2 mM fructose 6-phosphate, 0.5 mM ATP, 2.2 mM PEP,

1.6 mMAMP, 0.42 mMNADH, 2 U lactate dehydrogenase

and 2 U pyruvate kinase One activity unit is defined as the

conversion of 1 lmol substrate per min under these

standard conditions

For kinetic analyses an assay was used in which the

PFK activitywas not coupled to a NAD-dependent

reaction through its product ADP, as in the standard

assay, but through its product fructose 1,6-bisphosphate

The assaywas performed at 25C in a 1 mL

reac-tion mixture containing 100 mM triethanolamine/HCl,

pH 8.0, 2.5 mM MgCl2, 0.42 mM NADH, 0.4 U

aldo-lase, 0.8 U glycerol-3-phosphate dehydrogenase and

20 U triosephosphate isomerase The reaction was

initi-ated bythe addition of 5 lL of enzyme diluted in buffer

(0.1 M triethanolamine/HCl, pH 7.4, BSA 0.1 mgÆmL)1,

EDTA 0.2 mM and dithiothreitol 0.5 mM) The effect of

the fructose 6-phosphate concentration was determined

byfixing the ATP concentration at 1 mM, in the

presence and absence of AMP (1.5 mM) and GDP

(1.0 m )

R E S U L T S A N D D I S C U S S I O N

Analysis of kinetoplastid PFK genes

In the 30 kb insert of the cosmid obtained byscreening a

L donovani genomic library, only a single gene copy of PFK was detected Figure 1 shows a restriction map of the insert of recombinant plasmid PFK6.5-pBKS subcloned from that cosmid The coincidence between the restriction pattern of this cosmid and that obtained bySouthern analysis of whole Leishmania DNA, and the signal inten-sities of the bands (not shown), were consistent with the presence of one gene copyper haploid genome When blots

of L donovani chromosomal bands separated bypulsed-field gel electrophoresis were hybridized with an EcoRI– PvuIIfragment from recombinant PFK6.5-pBSKS (Fig 1),

a unique hybridization signal of 1.3 Mb was observed (Fig 2, lanes 1 and 2) This 1.3 Mb band maycorrespond

to chromosomes 27b, 28 or 29 [19] In L amazonensis, included for comparative purposes, the probe hybridized weaklyto a band of approximately1.7 Mb (Fig 2, lanes 3 and 4), the size of the proposed fusion product of chromosomes 8 and 29 in the L mexicana group [20] The amino-acid sequences encoded bythe ORFs found

in the L donovani and T borreli recombinants are shown

in Fig 3 The ORF in L donovani is 1461 bp, coding for a polypeptide of 486 amino acids (excluding the initiator methionine) with a molecular mass of 53 988 and a calculated isoelectric point (pI) of 9.26 The C-terminus has the tripeptide -SKV, a type 1 peroxisome-targeting signal (PTS-1) with an acceptable degeneracyof the canonical motif -SKL [21,22] The same tripeptide was previously found in another glycosomal enzyme of this organism, namely hypoxanthine-guanidine phosphoribosyltransferase

Fig 2 Chromosomal assignment of the PFK genes in L donovani and

L amazonensis Southern blot of Leishmania chromosomal bands separated bypulsed-field gel electrophoresis after hybridization with a probe consisting of a radioactivelylabelled EcoRI–PvuII restriction fragment (see Fig 1) containing the whole L donovani PFK gene Lanes 1 and 2, L donovani, lanes 3 and 4, L amazonensis The posi-tions of yeast chromosomes that were used as molecular size markers are indicated at the right-hand side.

[23] The sequence predicted from the T borreli ORF codes

for a polypeptide of 489 amino acids (excluding the first

methionine) with a molecular mass of 53 211 and a pI 8.9

The typical PTS-1 motif -SKL was found as the C-terminal

tripeptide An excess of positivelycharged residues and

resulting high pI are features often associated with

glycosomal enzymes, particularly in T brucei [24,25]

Figure 3 also shows the alignment of L donovani and

T borreli PFK sequences with those of some other

organisms Except for two N-terminal insertions in

T borreliPFK, the new sequences share the characteristics

of T brucei PFK as described previously[5] The

percentage identityin a pairwise comparison (Table 1)

of the amino acid sequences of L donovani and T brucei

is high (70%), whereas the value obtained bycomparing

T borreli PFK with the L donovani and T brucei

enzymes is 54% in both cases The extent of sequence

similarityamong the Kinetoplastida PFKs corresponds

with the values found for some other glycolytic enzymes

[9,26] and with the proposed phylogeny of this order

[27,28] The percentage identities of the kinetoplastid

PFKs with those from all other major taxonomic groups

are in the range 15–30% As alreadypreviouslyobserved

for the T brucei PFK [5], the L donovani and T borreli

enzymes show signatures typical of PPi-PFKs (see Fig 3)

and, in a phylogenetic analysis, theyappear firmlyplaced

within the cluster of the PPi-dependent enzymes, well

separated from the nonkinetoplastid ATP-PFKs (not

shown) Interestingly, the kinetoplastid PFKs showed the

highest percentage identity(37–38%) with the minor

48 kDa PFK from another protist, Entamoeba histolytica

Despite the higher overall similarityand its phylogenetic

relationship with the subset of PPi-PFKs [5,29,30], it was

recentlyreported that this E histolytica PFK uses ATP as

phospho donor (in contrast to the verydifferent 60 kDa

PPi-dependent PFK of this organism, see Fig 3) [31]

similar to the observation previouslyreported for the

T brucei enzyme [5] The PFK activity in Leishmania

species [32–34] and in T borreli (J Van Roy , F

Opper-does, N Chevalier & P A M Michels, unpublished

results) is also known to be ATP-dependent

Kinetics ofLeishmania PFK

The L donovani PFK was expressed in E coli with an

N-terminal His-tag and purified for kinetic analysis The

activityof the enzyme was ATP dependent No activity(less

than 1%) was observed when PPi(at concentrations up to

5 mM) was used as alternative phospho donor As reported

previouslyfor PFK of T brucei [7] and other Leishmania

species [32], the activity of the enzyme depends

hyperbol-icallyon the concentration of ATP The kinetic behaviour

of the expressed PFK was determined as a function of

fructose 6-phosphate at fixed, saturated ATP concentration

(1.0 mM), and in the presence or absence of AMP and GDP

(Fig 4) AMP, ADP and GDP are well-known effectors of

bacterial PFKs; in assays, GDP rather than ADP is often

used as effector, because ADP, being a reaction product,

mayact as a competitive inhibitor with respect to ATP In

the absence of AMP and at low fructose 6-phosphate

concentrations (less than approximately0.2 mM) the

enzyme showed slightly cooperative binding of the

substrate, with a Hill coefficient of 1.41 At higher substrate

concentrations, the enzyme displayed hyperbolic kinetics; the S0.5¼ 3.60 ± 0.48 mM In the presence of AMP, the curve is hyperbolic over the entire range of substrate concentrations; AMP has a clear stimulatoryeffect by increasing the affinityfor fructose 6-phosphate till a

Km¼ 0.157 ± 0.028 mM In contrast, GDP has no effect whatsoever on the activityof the enzyme

Our data thus showed that the expressed L donovani PFK binds its substrate fructose 6-phosphate in a cooper-ative manner, similar to manyother PFKs, such as the enzymes from mammals [35] and bacteria (reviewed in [36]) This behaviour, and the increased affinityfor the substrate

in the presence of AMP, have also been reported for the enzymes partially purified from cultured L donovani and

L braziliensis[32] and for T brucei PFK [6,7] The observed cooperative substrate binding and allosteric activation by AMP suggest a multimeric structure for the enzyme Indeed,

T brucei PFK was shown to be tetrameric [25], like the ATP-PFKs of most other organisms [1] Hyperbolic kinetics have been reported for the Trypanosoma cruzi enzyme, but the relevance of this finding maybe questioned, because the authors described an enzyme with a 17 kDa subunit mass [37] Despite the relativelyhigh sequence identityof Kinetoplastida PFK and PPi-dependent enzymes, AMP stimulation has onlybeen described for one PPi-dependent PFK, that from Naegleria fowleri [38] In this case the AMP effect was attributed to promoting a more active enzyme aggregate

Active site of kinetoplastid PFKs Table 2 presents a summaryof the active-site residues of

E coliPFK involved in the binding of ADP and fructose 6-phosphate as observed in its crystal structure [12], and the corresponding residues in the PFKs of human, T brucei,

L donovani, T borreli and in the minor 48 kDa enzyme of

E histolytica A comparison of the kinetoplastid and human PFKs shows four differences in the residues involved

in nucleotide binding and three differences in the residues involved in the binding of fructose 6-phosphate The same

Fig 3 Alignment of L donovani and T borreli PFK amino acid sequences with other ATP and PP i dependent enzymes Sequences include the ATP-dependent enzymes of T brucei, E coli, B stearothermo-philus, E histolytica, the N-catalytic domain of the human muscle enzyme and the catalytic subunit of S cerevisiae, and the PP i -depen-dent enzymes of E histolytica, A methanolica and P freundenreichii Sequences of yeast and human expanding beyond the N- or C-termini

of Kinetoplastida sequences are not shown The E coli and L dono-vani enzymes are numbered above and below the alignment, respec-tively Sy mbols are: black arrows, b strands; black cylinders, a helices; open circles, substrate ATP-binding residues; black circles, fructose 6-phosphate binding residues; black triangles, effector-binding resi-dues Boxes mark regions sharing identical residues between either the set of kinetoplastid and E histolytica ATP-PFKs and the set of typical ATP-PFKs, or the kinetoplastid and E histolytica ATP-PFKs and the set of PP i -PFKs Residues common to all sequences are in bold + italic font; bold onlyis used where there is one disagreement among the entire sequence set Residue 224 of L donovani PFK that was studied bymutagenesis is indicated bya black triangle underneath the align-ment The figure was made using ALSCRIPT [53].

substitutions occur in the 48 kDa ATP-PFK of E histolytica

(Fig 3) [39], as well as in one of the PFKs of the prokaryotic

Spirochaetes Treponema pallidum and Borrelia burgdorferi

(data not shown; GenBank accession numbers AE001195

and AE001172) Indeed, the identityof active-site residues

in all these organisms is in agreement with the branching order in a phylogenetic analysis based on full-length PFK sequences [5,29,30] The PFKs of these organisms form a well-supported monophyletic cluster within the PPi-PFK subset, well separated from the typical ATP-PFKs [5,29,30]

The identityof the phospho donor of the PFKs from

T pallidum and B burgdorferi has not been established

yet

The E coli residues involved in fructose 6-phosphate binding that have been substituted in the kinetoplastid and related PFKs are Arg162, Arg243 and His249 The

Fig 3 (Continued.)

corresponding residues found in all these PFKs are Gly, Lys

and Tyr, respectively Whereas Arg162 seems conserved in

all typical ATP-PFKs, a positively charged residue at the

corresponding position seems not essential for substrate

binding in the PPi-PFKs and the atypical ATP-PFKs The

Arg243 of ATP-PFKs is replaced byLys in all PFKs of the

subset comprising the Kinetoplastida; apparently, a

positivelycharged residue at this position is essential in all

ATP-PFKs It is possiblyequivalent to Lys315 of the

PPi-dependent PFK of Propionibacterium freundenreichii

(E coli position 241 in Fig 3), because Xu et al [40] have

shown that an alteration of this residue bysite-directed

mutagenesis causes a 389-fold increase of the Km for

fructose 6-phosphate The substitution of His249 (E coli

numbering) byTyr in the Kinetoplastida is also found in

some PPi-PFKs such as the enzymes of P freudenreichii and

E histolytica(Fig 3), and is possiblywithout much effect

Out of 10 residues involved in ADP binding in E coli PFK, four are not conserved in Kinetoplastida (Table 2) Strikingly, the ATP-dependent E histolytica PFK and the putative ATP-PFKs of two Spirochaetes discussed above have the same residues as the ATP-dependent kinetoplastid enzymes [39,41] Therefore, these residues may be important determinants for the phospho-donor specificityif, in future research, the isoenzymes of the Spirochaetes turn out to be ATP-dependent as well

In addition to the substrate-binding residues in Table 2, another active-site residue is of particular interest: the residue corresponding to E coli Gly124 is a Lys in the kinetoplastid ATP-PFKs, in the ATP-dependent isoenzyme

of E histolytica (see Fig 3) and in the PFKs of the two Spirochaetes (not shown) A Glyis typical of ATP-PFKs,

Table 1 Percentage identity of the PFK amino-acid sequences given in Fig 3.

B stearo.

Human muscle-N S cere.-N T bruce i L donovani T borreli

E histo.

ATP

E histo.

PP i A meth.

B stearothermophilus 54

Fig 4 Kinetics of recombinant L donovani PFK with respect to

fruc-tose 6-phosphate Activitywas measured at a fixed ATP concentration

of 1.0 m M Symbols are: j, no additions; d, + 1.5 m M AMP;

m, + 1.0 m M GDP Values of kinetic parameters (see text) were

cal-culated after optimal curve fitting of the experimentallydetermined

data using the SIGMAPLOT program The values given in the text

are ± SD for the fit.

Table 2 Amino acid residues involved in binding fructose 6-phosphate and ADP in E coli PFK, and the corresponding residues in the organisms L donovani, T brucei, T borreli, E histolytica-ATP,

T pallidum and B burgdorferi Differences are highlighted in bold.

E coli Other organisms Fructose 6-phosphate Thr125 Thr

whereas a Ly s is present in all PPi-dependent enzymes Xu

et al [40] have shown that a change of this lysine into a

methionine in P freudenreichii PFK caused a 132-fold

increase in the Kmfor PPiand a 490-fold decrease in kcat,

providing strong indication for a direct involvement of this

Lys residue in PPibinding From a phylogenetic analysis, we

previouslyconcluded that the kinetoplastid PFKs must

have been derived from a PPi-dependent ancestral PFK,

which changed its phospho-donor specificityearlyin the

evolution of the lineage [5] However, the Lys is no longer

involved in PPi binding, so whyhas it been retained in

the present-daykinetoplastid PFKs which are all ATP

dependent, and in the ATP-dependent isoenzyme of

E histolytica? Has it obtained a function in ATP binding,

implying that the mode of nucleotide binding is different in

kinetoplastid PFKs from that in other ATP-PFKs? Or has it

been retained for structural reasons? Relevant to these

questions is the observation that the Lys is also present in

the ATP-dependent enzyme from the actinomycete

Strep-tomyces coelicolor that, in a phylogenetic analysis, also

clusters with the PPi-dependent PFKs and that must have

had a common ancestor with the PPi-dependent PFKs of

other Actinomycetes such as Amycolatopsis methanolica

[42]

We therefore investigated whether substituting the Lys

would have anyeffect on the kinetoplastid enzyme To this

end, we replaced it byGlyin L donovani PFK The

resulting LdPFK Lys224fiGlymutant did not displayany

activity In contrast, the corresponding LysfiGlymutant of

T brucei(TbPFK Lys226fiGly) appeared to be active [43]

Strikingly, this T brucei mutant showed onlya slight

decrease in its affinityfor ATP, but the mutation had a

major effect on the enzyme’s behaviour with respect to

fructose 6-phosphate In the absence of AMP, the S0.5for

fructose 6-phosphate was 5 mMcompared to 0.59 mMin

the wild-type enzyme However, the higher substrate affinity

can still be induced bythe addition of AMP: Kmfor fructose

6-phosphate¼ 0.82 mMcompared to 0.15 mMin the

wild-type PFK [43] It seems that the mutation leads to a local

disruption of the active site with accompanying lowering of

fructose 6-phosphate affinity This is independent of the

allosteric changes in fructose 6-phosphate affinityas AMP is

capable of similar enhancements of fructose 6-phosphate

affinity in both wild-type and mutant enzymes In the case

of the L donovani enzyme, a greater degree of disruption

leads to abolition of fructose 6-phosphate binding, either

through local or global effects

As discussed previously, the comparative analysis of PFK

sequences suggests that onlysubtle changes maybe required

for a change of phospho donor specificity[5] This notion

was reinforced bythe recent publication [44] describing

mutations in the ATP-PFK of E coli and the PPi-PFK of

E histolytica, similar to those described above for

trypan-osomatid PFKs The Gly124fiLys substitution in E coli

effectivelyeliminated activitywith ATP as a substrate, but

no PPi-dependent activitywas observed However, the

reverse Lys201fiGlymutation in the PPi-dependent, major

PFK of E histolytica reduced the kcat with PPi as the

phospho donor byfour orders of magnitude, while having

onlya limited effect on the apparent PPi affinityof the

residual enzyme activity Importantly, the performance of

the enzyme with ATP as a phospho donor increased about

eightfold (although this is still 105 times less than the

performance of the wild-type enzyme with PPi) essentially byan increase in kcat

Understanding of the role of Lys224 in L donovani PFK, and corresponding residues in other PFKs, is hampered by the lack of a crystal structure with a lysine at this position Modelling of the Glyto Lys mutation shows that, without significant local structural changes, the lysine side chain becomes entirelyburied in the protein interior, with no possibilityof electrostatic interaction with an acidic residue,

a situation essentiallyunknown in protein structure One hypothesis is that, in enzymes containing a lysine at this position, the peptide bond with the preceding proline adopts

a cis configuration [45] It is suggestive that proline is entirelyconserved at position 123 (E coli numbering) when

a lysine is present at position 124, while valine is also tolerated in other PFKs Modelling of possibilities for a

L donovanimodel containing such a cis peptide bond yields structures in which the lysine side chain is solvent-exposed such as that illustrated in Fig 5 In this structure the lysine side chain is placed at the heart of the catalytic site A direct interaction with substrate ATP seems unlikelyfrom the kinetic results presented here, although it is unfortunate that structural inferences mayonlybe drawn from a product-bound PFK structure (Fig 5) However, a limited descrip-tion of a PFK-AMPPNP-fructose 6-phosphate substrate analogue complex [46] (coordinates not deposited) supports the notion of a close resemblance between substrate- and product-bound protein structures Whythen, in contrast, should PPibinding be dramaticallyaffected when this lysine

is mutated in E histolytica and P freudenreichii PFKs [40,44]? The explanation maylie in another residue, clearly implicated in substrate specificity[44] Position 104 (E coli numbering) is always a Gly in ATP-PFKs and an Asp in

PPi-PFKs Structural examination (Fig 5) shows that the presence of anynon-Glyresidue leads to steric clashes with the bound nucleotide in its crystallographically observed

Fig 5 Positions of phospho-donor specificity-determining residues rel-ative to the catalytic site of E coli PFK bound to products Numbering is according to the E coli enzyme The figure was produced using

[54].

position In the structure of a PPi-dependent PFK bound to

substrates, it would be reasonable to expect that the PPi

substrate binds in the corresponding position as the b- and

c-phospho groups of bound ATP in ATP-dependent

enzymes However, analysis shows that any rotameric

conformation of an Asp104 side chain leads to positioning

of its negative charge near to the phospho group occupying

the a position, also negativelycharged Minimum

oxygen-oxygen interatomic distances range from 1.1 to 3.9 A˚,

depending on Asp104 rotamer This electrostatic repulsion

maytherefore force the PPi into a slightlydifferent

conformation, further from Asp104 and hence nearer to

Lys124 This hypothesis allows an explanation of the

apparent involvement of this lysine in PPibinding [40,44]

but not in ATP binding A more prosaic explanation may

underlie the almost complete lack of PPi- or ATP-dependent

PFK activityseen for the Gly124fiLys E coli mutant [44]

Without a preceding cis peptide bond onlyside chain

conformations that unfavourablyburythe positive charge

of the new Lys are attainable and the protein would

therefore be destabilized The lack of confirmation of native

fold for the mutant, byCD experiments for example,

suggests that the mutant mayhave undergone gross

structural changes resulting in loss of activity The modelled

lysine in the L donovani model is well packed and not

apparentlywell positioned to interact directlywith fructose

6-phosphate These considerations support the previously

advanced explanation of the effects of the Lys224fiGly

mutation in terms of destabilizing local structural changes

Effector-binding site of trypanosomid PFKs

Table 3 presents a comparison of the residues in B

stearo-thermophilusand E coli PFKs involved in the binding of the

allosteric activator ADP with the corresponding residues in the L donovani and T brucei enzymes (according to the alignment in Fig 3) A structural comparison of the residues binding the activator ADP in B stearothermophilus PFK and a putative AMP binding mode for the L donovani, suggested bymodelling, is shown in Fig 6 This comparison suggests that the kinetoplastid enzymes mayemploythe same region for binding their allosteric activator AMP The b-phospho group binding residues of the bacterial enzymes show the most changes, with the most striking substitution being the replacement of the Mg2+-ligating Glu187 with Asn The loss of Mg2+and the replacements of Arg25 and Arg154, both of which electrostaticallyinteract with the b-phospho group of ADP (Fig 6A), effectivelyeliminate the b-phospho-binding pocket in the trypanosomatid enzymes The residues at positions 211 and 213 of the

B stearothermophilus enzyme that bind the a-phospho

Fig 6 Comparison of (A) the crystallographically observed effector site of E coli PFK with bound ADP and (B) the modelled structure of L donovani PFK effector site bound to AMP Ligand and protein are shown in ball-and-stick representation with the exception of the protein backbone, in the vicinityof the one residue insertion, which is drawn as a tube Possible hydrogen bonds are shown with dotted lines The figure was produced using

[54].

Table 3 Amino acid residues involved in binding the allosteric activator ADP in B stearothermophilus and E coli PFK, and corresponding residues in T brucei and L donovani PFKs Differences are highlighted

in bold.

B stearothermophilus E.coli T brucei L donovani

group of ADP (Fig 6A) are better conserved and maybe

involved in binding the phospho group of AMP (Fig 6B)

The modelling reveals that, in addition to the residue

differences identified bysequence comparisons, other

significant structural differences mayexist The L donovani

enzyme, along with others from kinetoplastids and the

E histolytica ATP-dependent PFK, has a one residue

insertion, relative to bacterial enzymes, at around position

287 (L donovani enzyme numbering) Two possible

posi-tions for this insertion were analysed and one, as shown in

Fig 6B, found to be favoured for avoiding the positioning

of model residues in unusual areas of the Ramachandran

plot The altered main chain conformation in the vicinity

causes the Gln287 side chain to intrude into the area

corresponding to the b-phospho-binding site of the bacterial

enzymes Furthermore, it may form a hydrogen bond with

the phospho group of effector AMP (Fig 6B) Another

interesting structural difference revealed bymodelling

relates to the side chain position of Arg157 in the

L donovani enzyme which replaces a Ser or Gly in the

bacterial enzymes Adopting a rotameric conformation, this

residue mayfill the space caused bythe replacements of

B stearothermophilus Arg211 with Gln and Arg25 with

Leu In this position it mayhydrogen bond to the phospho

of effector AMP in the L donovani model structure and

contribute to the positive electrostatic potential of the

effector binding pocket

C O N C L U S I O N S

The PFK genes of L donovani and T borreli have been

cloned and sequenced The encoded enzymes show most

similarityto the subset of PPi-PFKs, as did the previously

analysed T brucei PFK Nevertheless, ATP is the

phos-pho substrate of all these kinetoplastid PFKs It is

possible that a common ancestral organism changed its

phospho donor specificityduring evolution The currently

available data do not allow us to draw anyconclusion as

to how and whythe Kinetoplastida and other protists

such as Entamoeba obtained their ATP-dependent, PPi

-like PFKs Did theyevolve from a PPi-PFK in both

lineages independently, or did they originate in a common

ancestor of these protists? Were theyacquired from

Spirochaetes bylateral gene transfer? In this respect, it

maybe relevant that phylogenetic studies based on

sequences of other glycolytic enzymes,

glyceraldehyde-3-phosphate dehydrogenase and enolase, showed grouping

of Kinetoplastida (and/or the related Euglenoida) and

Spirochaetes [47–49]

Strikingly, all kinetoplastid PFKs, as well as the

Entamoeba PFK contain a Lys on position 124 (E coli

numbering), whereas all other ATP-PFKs contain a Gly

Previous mutagenesis studies have provided strong evidence

that this Lys residue is involved in PPibinding Structure

modelling suggests that the Lys may have been retained in

the kinetoplastid PFKs to maintain the stabilityof the

active-site structure These results are supported

bymuta-genesis studies No active L donovani Lys224fiGlymutant

could be obtained, whereas the kinetic properties of a

corresponding Lys226fiGlymutant of T brucei PFK

could be interpreted in terms of a destabilized active site

The L donovani PFK shows slightlycooperative binding

of fructose 6-phosphate at low concentrations of this

substrate The enzyme was allosterically activated by AMP bya significant increase in the affinityfor the substrate However, trypanosomatid PFKs are not activa-ted byADP, in contrast to their counterparts in the bacteria

E coli and B stearothermophilus Modelling studies have provided a possible structural basis for the AMP specificity

We have provided evidence for significant structural differences between trypanosomatid PFK and other ATP-PFKs including the human enzyme Such differences were found in both the active site and the region of the enzyme presumablyinvolved in effector binding Indeed, the differences in the effector-binding site tallywith the apparentlylow level of activityregulation of trypanosoma-tid PFK as compared to that of the human enzyme This limited regulation of trypanosomatid PFK seems physio-logicallyrelevant in view of the intraglycosomal localization

of the enzyme and the low permeability of the organelle’s membrane for manymetabolic intermediates that in other cells act as PFK effectors [3,50] The structural differences observed offer great potential for the design or selection of drugs Although our computer analysis using a kinetic model of glycolysis suggested that PFK in bloodstream-form T brucei is present in excess [51], we have argued elsewhere [8,52] that this does not necessarilyexclude the enzyme as a target for selective inhibitors that bind with high affinity, particularlyirreversiblybinding inhibitors The most important aspects to consider in drug target selection are that an enzyme should have an essential (or at least veryimportant) metabolic role and that its structure should be sufficientlydifferent from that of the correspond-ing host enzyme Moreover, metabolism in bloodstream-form T brucei is highlyspecialized, and in manyrespects not representative for the infective stages of other trypan-osomatid parasites such as the trypomastigotes and amastigotes of Leishmania species and T cruzi [3] Therefore, we consider the trypanosomatid PFK as a highly promising drug target

A C K N O W L E D G E M E N T S This research was financiallysupported bythe European Commission (programmes STD3 and INCO-DC) Financial support for C L for a

1 year stayat the ICP in Brussels was provided bythe Fundacio´n Gran Mariscal de Ayacucho and CONICIT Venezuela (grant S1-9500524).

We are grateful to Dr Theo deVos (SBBI, Seattle) for providing the genomic L donovani library, and to Drs Linda Fothergill-Gilmore (Universityof Edinburgh) and Fred Opperdoes (ICP, Brussels) for critical reading of the manuscript.

R E F E R E N C E S

1 Fothergill-Gilmore, L.A & Michels, P.A.M (1993) Evolution of glycolysis Prog Biophys Mol Biol 59, 105–235.

2 Opperdoes, F.R & Borst, P (1977) Localization of nine glycolytic enzymes in a microbody-like organelle in Trypanosoma brucei: the glycosome FEBS Lett 80, 360–364.

3 Michels, P.A.M., Hannaert, V & Bringaud, F (2000) Metabolic aspects of glycosomes in Trypanosomatidae – new data and views Parasitol Today 16, 482–489.

4 Mertens, E (1991) Pyrophosphate-dependent phosphofructokin-ase, an anaerobic glycolytic enzyme FEBS Lett 285, 1–5.

5 Michels, P.A.M., Chevalier, N., Opperdoes, F.R., Rider, M.H & Rigden, D.J (1997) The glycosomal ATP-dependent phospho-fructokinase of Trypanosoma brucei must have evolved from an